Thursday, January 17, 2002

8:00 a.m.












Holiday Inn Gaithersburg

Gaithersburg, Maryland


Mark S. Dykewicz, M.D., Chairman

Kimberly Topper, Executive Secretary


James K. Stoller, Ph.D.

Robert J. Fink, M.D.

Andrea J. Apter, M.D.

Jesse Joad, M.D.

Thomas P. Atkinson, M.D., Ph.D.

Polly E. Parsons, M.D.

Karen Schell, RRT, Consumer Representative


Henry G. Bone, M.D.

Robert A. Wise, M.D.

Saul Malozowski, M.D., Ph.D.


Robert Meyer, M.D.

Mary Purucker, M.D.

Lydia Gilbert-McClain, M.D.

Charles Lee, M.D.



Welcome, Mark S. Dykewicz, M.D. 4

Conflict of Interest Statement, Kimberly Topper 6

Welcome and Topic Introduction,

Robert Meyer, M.D., FDA 9

GlaxoSmithKline Presentation:

Introduction, David Wheadon, M.D. 15

Scientific and Clinical Rationale,

Malcolm Johnson, M.D. 20

Clinician's Perspective, James Donohue, M.D. 35

Clinical Efficacy and Safety, Tushar Shah, M.D. 48

Conclusions, David Wheadon, M.D. 89

FDA Presentations:

Flovent Diskus for COPD, Charles Lee, M.D. 119

Advair Diskus for COPD,

Lydia Gilbert-McClain, M.D. 137

Summary and Issues for PADAC,

Mary Purucker, M.D. 158

Discussion 212

Questions 282



DR. DYKEWICZ: Good morning. Welcome to

the meeting of the Pulmonary and Allergy Drugs

Advisory Committee. I am Mark Dykewicz, the chair,

associate professor of internal medicine in the

Division of Allergy and Immunology at Saint Louis


First of all, a few ground rules to

maintain order. Members of the committee, when you

are going to speak, first I would like you to raise

your hand so I can recognize you. Then, when you

do speak we want you to push down on the button to

activate your microphone. Perhaps even more

important, when you are done speaking, push the

button again to deactivate the microphone so we

don't hear all sorts of side bars that will confuse

us. So with those sort of ground rules, let's

begin with introductions, starting with Bob Meyer.

DR. MEYER: I am Dr. Robert Meyer. I am

the director of the Division of Pulmonary and

Allergy Drug Products in the Center for Drug

Evaluation and Research.

DR. PURUCKER: I am Dr. Mary Purucker, in

the Division of Pulmonary Drug Products.


Gilbert-McClain, medical reviewer in the Division

of Pulmonary and Allergy Drug Products.

DR. LEE: I am Charles Lee, medical

reviewer in the Division of Pulmonary and Allergy

Drug Products.

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

consumer rep.

DR. JOAD: I am Jesse Joad. I am a

pediatric pulmonologist and allergist at UC Davis.

DR. APTER: I am Andrea Apter. I am an

allergist and immunologist from the Pulmonary,

Allergy and Critical Care Division of the

University of Pennsylvania.

DR. ATKINSON: I am Preston Atkinson. I

am an allergist/immunologist from Children's

Hospital at University of Alabama at Birmingham.

DR. FINK: Bob Fink, pediatric

pulmonologist at Children's Hospital in Washington,


DR. STOLLER: I am James Stoller, I am a

lung doctor at the Cleveland Clinic.

DR. BONE: I am Henry Bone. I am an

endocrinologist, specializing in bone and internal

disorders, in Detroit.

DR. PARSONS: I am Dr. Polly Parsons. I

deal in pulmonary and critical care medicine at the

University of Vermont.

DR. WISE: Robert Wise, pulmonologist,

from Johns Hopkins University, in Baltimore.

DR. MALOZOWSKI: I am Saul Malozowski. I

am a pediatric endocrinologist at NIDDK, NIH.

DR. DYKEWICZ: Thank you. Now we will

begin with the statement of conflicts of interest.

Conflict of Interest Statement

MS. TOPPER: The following announcement

addresses the issue of conflict of interest with

regard to this meeting, and is made part of the

record to preclude even the appearance of such at

this meeting. Based on the submitted agenda for

the meeting and all potential interests reported by

the committee participants, it has been determined

that all interests in firms regulated by the Center

for Drug Evaluation and Research present no

potential for an appearance of a conflict of

interest at this meeting, with the following


Dr. Andrea Apter has been granted waivers

under 18 USC, Section 208(b)(3) and Section

505(n)(4) of the FDA Modernization Act for her

ownership of stock in two competitors. The first

stock is valued at between $50,001 to $100,000, and

the second between $5,001 and $25,000. The waivers

permit Dr. Apter to participate in the committee's

deliberations and vote concerning the new drug

applications, NDA 20-833 and 21-077, sponsored by


A copy of these 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.

We would also like to disclose that

because of their reported interests, Dr. Nicholas

J. Gross and Dr. Michael S. Niederman, who are

committee members, are excluded from participating

in all official matters concerning NDA 20-833 and

21-077, sponsored by GlaxoSmithKline.

Further, with respect to FDA's invited

guest, Dr. Robert Wise's employer, Johns Hopkins

Bayview Medical Center, has the following contracts

and/or grants: Research grant negotiations are in

progress with the American Lung Association-Asthma

Clinical Research Centers, ALA-ACRC, and

GlaxoSmithKline and AstraZeneca. Dr. Wise serves

as the PI of the ALA-ACRC data coordinating center.

He is PI of the Clinical Center for Lung

Health Study, funded by NIH-NHLBI with drug

contribution, ipratropium, from


He is PI of the Clinical Center for Lung

Health Study II, funded by NIH-NHLBI with drug

contribution (triamcinolone, from Rhone-Poulec

Rohrer, now Aventis.

He is PI of a pending research grant for a

clinical trial of tiotropium, sponsored by

Boehringer Ingelheim.

He is co-sponsor for a Childhood Asthma

Management Program, funded by NIH-NHLBI with drug

contribution (budesonide) by AstraZeneca, and

nedocromil, contributed by Aventis.

He is the PI of Clinical Center for a

study of COPD evaluation instruments, sponsored by


He receives less than $5,000 in consulting

fees from Aventis, Boehringer Ingelheim, Novartis,

Bristol-Myers Squibb, McNeil and AstraZeneca.

Her receives between $5000 to $10,000 in

consulting fees from Johnson & Johnson, He

consults on a non-pulmonary drug.

He receives between $10,000 and $15,000 in

consulting fees from Pfizer. He consults on a

non-pulmonary drug.

Finally, he receives less than $5000 from

GlaxoSmithKline in support of a local thoracic

society and pulmonary grand rounds.

In the event that the discussions involve

any other products or firms not already on the

agenda for which an FDA participant has a financial

interest, the participants are aware of the need to

exclude themselves from such involvement and their

exclusion will be noted for the record.

With respect to all other participants, we

ask in the interest of fairness, that they address

any current or previous financial involvement with

any firm whose products they may wish to comment

upon. Thank you.

DR. DYKEWICZ: Thank you. We will now

begin with the topic introduction by Dr. Bob Meyer,

Director of the Division of Pulmonary and Allergy

Drugs of the FDA.

Welcome and Topic Introduction

DR. MEYER: Thank you. First off, I

should mention that Dr. Sandra Kweder, who is our

Office Director, sends her apologies for not being

able to be here today. She had prior commitments.

On behalf of the Center for Drug

Evaluation and Research and the Pulmonary and

Allergy Division, I would like to welcome the

advisory committee members and guests, and to thank

you in advance for your participation in what I

consider to be an important meeting. We very much

appreciate your willingness to lend your expertise

to the advice which we hope to take away today.

We are here today to discuss two

applications; Advair Diskus for the maintenance

treatment of COPD and Flovent Diskus for the

maintenance treatment of COPD. While we certainly

understand that corticosteroids are commonly used

in the treatment of patients with COPD, both in the

acute setting where the treatment is mainly

systemic, and in the maintenance setting where

treatment is commonly either inhaled or systemic.

The FDA has not to date approved such use.

We are also very much aware of guidelines,

including those published by the National

Institutes of health, that make recommendations for

the use of inhaled corticosteroids in COPD for a

limited subset of such patients, based, by their

own evidentiary standards, on less than substantial

evidence. So, these two applications we are

discussing today, one for a corticosteroid in

combination with a long-acting bronchodilator and

one for a corticosteroid alone, represent

groundbreaking and important issues for the FDA.

I would like to make clear that

GlaxoSmithKline has done a very elegant clinical

development program for Advair Diskus for the

maintenance treatment of COPD that also has allowed

the FDA to separately address the efficacy of the

two components of that combination product, both

fluticasone and salmeterol for the treatment of

COPD as well. Since the salmeterol metered-dose

inhaler is already approved for COPD, we do not fee

that the application for the Serevent Diskus for

COPD warranted advisory committee discussion.

In designing this program that

GlaxoSmithKline conducted, including the choice of

endpoints, GlaxoSmithKline met and worked with the

Pulmonary Division. Therefore, FDA agreed

beforehand on the choice of primary endpoints.

However, as in any development program but

particularly for a novel groundbreaking program, we

stated at the time of these discussions, and feel

now, that a full assessment of efficacy and safety

needs to be considered in assessing the

advisability of approving these drugs for this

indication, not just the effect on the primary


Further, I would point out that our

knowledge of the potential risks and of the

potential benefits of inhaled corticosteroids for

COPD has evolved since we had these discussions

with the sponsor at the inception of the program.

I think we need to put the sponsor's specific data

which we will hear today into the perspective of

what we now know and what we need to know about any

such treatment in the year 2002 and beyond.

The Advair and Flovent Diskus products, as

you know, are approved for the maintenance

treatment of asthma and are available for us in the

practice of medicine. In fact, I would suspect a

rather large number of COPD patients are already

receiving one or the other of these products since

both are rather brisk sellers in the marketplace

and since we know such treatment of COPD is common.

However, FDA neither wants to nor can it restrict

the practice of medicine.

What you advise us today will not lead to

nor lift any restriction of the practice of

medicine. Rather, if your recommendation is for

approval, you are indicating that you believe the

sponsor has provided substantial evidence of the

efficacy and safety of each of these two products

for the maintenance treatment of COPD, including

chronic bronchitis and emphysema, and that they

should be labeled and promoted as such.

I would like to remind you of the

evidentiary standard of the Food, Drug and Cosmetic

Act under which the FDA has its authority. The

FD&C Act calls for, and I quote, evidence from

adequate and well-controlled investigations [are

done] evaluate the effectiveness of the drug

involved, on the basis of which it could be fairly

and responsibly concluded by experts that the drug

will have the effects it is purported to have.

Further, the sponsor, and again I am

quoting, has included all tests reasonably

applicable to show the drug is safe under the

conditions of use suggested in the proposed

labeling thereof, end quote. This is, necessarily,

a higher standard than what a practitioner would

use to make a judgment that an individual drug is

right for an individual patient. What we are

talking about today is not such a choice in the

practice of medicine but whether the U.S. Food and

Drug Administration should specifically label these

drugs for this use as providing a clear and defined

benefit, given the safety risks.

Finally, let me again be clear that we are

here today to talk about two separate applications

for two separate products. Each of these

applications must be separately thought about and

discussed, and recommendation on one drug should

not, in and of itself, force a recommendation on

the other. Neither should we be focusing in this

meeting about class issues with inhaled

corticosteroids, because your recommendation should

be focused on the data that you have read and will

see presented today, though, admittedly, you must

consider these data in the milieu of what we know

or don't know about COPD and what we know or don't

know about inhaled corticosteroids in the treatment

of COPD.

Again, I would like to thank you for your

time and effort, and welcome you to what I am sure

will be a very interesting and important

discussion. Thank you.

DR. DYKEWICZ: Thank you, Bob. We will

now begin with the GlaxoSmithKline presentations,

beginning with an introduction by Dr. David


GlaxoSmithKline Presentations


DR. WHEADON: Good morning.


I am David Wheadon, Senior Vice President

of Regulatory Affairs at GlaxoSmithKline.


On behalf of GlaxoSmithKline, I would like

to thank the agency and the committee for this

opportunity to review our applications for Flovent

Diskus and the combination product, Advair Diskus,

for long-term, twice-daily maintenance treatment of

chronic obstructive pulmonary disease, including

emphysema and chronic bronchitis.


As you have already heard, one medication

already approved for the treatment of bronchospasm

associated with COPD is Serevent inhalation

aerosol. Serevent Diskus for COPD was studied as

part of the development program for Flovent Diskus

and Advair Diskus, and a supplemental new drug

application for Serevent Diskus is also under

review by the FDA. However, as the active

ingredient of Serevent Diskus, salmeterol, is

already approved for COPD it is not the subject for

today's meeting.


Flovent contains fluticasone propionate, a

synthetic glucocorticoid with high topical

anti-inflammatory activity and negligible oral

bioavailability. Flovent is indicated for the

maintenance treatment of asthma, and is approved in

the U.S. as a metered-dose inhaler and in two

powder formulations.

No inhaled corticosteroid, including

Flovent, is currently approved for the treatment of

COPD in the U.S. To date, Flovent has been

approved for the treatment of COPD in 67 countries

outside the U.S. Worldwide, as of August 31, 2001,

the exposure to Flovent was estimated to be 14.4

million patient years for the treatment of asthma

and COPD. For the treatment of COPD, marketing

approval is sought for doses of 250 mcg and 500 mcg

administered twice daily.


Advair, the combination of fluticasone

propionate and salmeterol in a single inhaler, is

indicated for the maintenance treatment of asthma,

and is available in the U.S. as a powder

formulation via Diskus. Advair is not currently

approved for the treatment of COPD in any country,

and the U.S. application was the first submission

globally. Worldwide, as of April 30, 2001, the

exposure to Advair was estimated to be 1.4 million

patient years for both asthma and COPD. Marketing

approval is sought for both the 250/50 mcg and the

500/50 mcg strengths of Advair Diskus administered

twice daily for the treatment of COPD.


COPD continues to be a significant public

health challenge. It remains a major cause of

morbidity and mortality in the U.S. and worldwide

and, sadly, the rates are increasing, in contrast

to many other diseases. COPD affects an estimated

21.7 million Americans and is currently the fourth

leading cause of death in the U.S., with

expectations for it to become the third leading

cause by 2020. The burden of this disease on

society is enormous. In 1997, direct and indirect

costs associated with COPD were estimated to be

over 30 billion dollars in the U.S. alone, and it

is likely these costs will continue to increase.


Despite the enormous burden of COPD, this

disorder often fails to receive adequate attention

from the medical community. Recognition of this

oversight led to the Global Initiative for Chronic

Obstructive Lung Disease, or GOLD. An output of

this initiative was the development of

evidence-based guidelines for the management of

COPD. These guidelines were developed through

collaboration with the National Heart, Lung and

Blood Institute and the World Health Organization.


In GOLD, the recommendation for the use of

inhaled corticosteroids in the management of COPD

was based on a considerable body of evidence, which

you can see on this slide. These studies

demonstrated the beneficial effects of inhaled

corticosteroids on a number of clinical parameters.

Dr. Malcolm Johnson, who will follow me, will

present further details on these studies.


This growing body of evidence has also led

to the use of inhaled corticosteroids as common

practice for the treatment of COPD in the U.S. As

shown here, prescription data from the NDC health

patient database of U.S. patients diagnosed with

COPD, show that 40 percent of patients are already

using inhaled corticosteroids. Furthermore, this

data indicates that 46 percent of patients were

prescribed two more COPD maintenance medications.

Of these patients, 72 percent were prescribed

inhaled corticosteroids as part of their treatment

regimen. Additionally, more than half, 57 percent

were being treated with an inhaled corticosteroid

in combination with an inhaled maintenance


Thus, we can see inhaled corticosteroids

are already being used extensively by physicians

for the management of this chronic debilitating

disease. In order to ensure that they are used

appropriately, we need to provide guidance to

physicians on how best to use these agents for the

treatment of COPD.


In summary, COPD is a serious public

health issue for the U.S. with considerable unmet

needs. Approval of new medicines is important for

the appropriate treatment of this debilitating

disease. The data we will share with you this

morning will show that Flovent and Advair provide

valuable treatment options for physicians in the

management of their patients with COPD.


To outline the order of the presentations

today, Dr. Malcolm Johnson will review the

scientific and clinical rationale for the use of

Flovent Diskus and Advair Diskus for the

maintenance treatment of COPD. Following that, Dr.

James Donohue, of the University of North Carolina,

will present a clinician's perspective on the

diagnosis and management of this difficult

condition. Dr. Tushar Shah will then review the

efficacy and safety data from our clinical

development programs for Flovent and Advair.

Finally, I will return to present concluding

statements and presenters will then be available to

respond to questions from the committee. I would

now like to introduce Dr. Malcolm Johnson.

Scientific and Clinical Rationale

DR. JOHNSON: Thank you, David.


Good morning, ladies and gentlemen. I am

Malcolm Johnson. I am the global director of

Respiratory Science for GlaxoSmithKline.


COPD is a disease characterized by a

multi-component pathology -- inflammation,

structural changes in the airways and airway



The underlying pathophysiologic processes

involved in this disease are shown in this slide.

With airway inflammation we see increased numbers

of inflammatory cells in airway tissue, cells such

as neutrophils, macrophages and the CD8 positive

subgroup of T-lymphocytes. There is evidence of

increased pro-inflammatory mediators, such as

interleukin-8 and tumor necrosis factor alpha. In

addition, there is an imbalance between protease

and anti-protease enzymes.

The structural changes involve alveolar

destruction and an increase in alveolar air space.

There is deposition of collagen, hypertrophy of

glandular tissue and, in some cases, airway

fibrosis has been detected.

With airway obstruction, this is a result

of smooth muscle contraction and

bronchoconstriction, increased cholinergic tone and

loss of elastic recoil due to a destruction in

parenchymal tethering. It is this complex

underlying pathophysiology that leads to the

clinical characteristics of the disease symptoms,

fall in lung function and exacerbations. It is

against this pathophysiological construct and the

clinical characteristics involved that we need to

assess the effectiveness of drug intervention.


So, beginning first with the inhaled

steroids, the first question we need to address is,

is there evidence that inhaled corticosteroids have

an anti-inflammatory effect in the disease COPD?


Well, ten studies have been reported that

have addressed this issue. Seven of these studies

have concluded that there is evidence of an

anti-inflammatory effect of inhaled corticosteroids

in COPD. These studies range from six weeks in

duration to 24 weeks in duration and, importantly,

four of the studies involved fluticasone


These studies concluded that there was a

reduction in neutrophils not only in the

bronchoalveolar lavage fluid but in the sputum of

these patients. There was a reduction in some of

the key inflammatory mediators involved in this

pathophysiology and, importantly, two of the

studies detected an important anti-inflammatory

effect of inhaled steroids at the level of airway

tissue, in particular, a reduction in this

important CD8 positive subgroup of T-lymphocytes

and in one case a change in the ratio between the

CD8 and CD4 positive cells.

Three studies failed to find an

anti-inflammatory effect of inhaled steroids. In

the main, they tended to be of shorter duration and

involved a smaller number of patients.


What I would like to do is to focus in a

little bit more detail on one of the studies taken

from this table. This is a study involving

fluticasone propionate at a dose of 1500 mcg a day

for a treatment period of 8 weeks. During the

course of this treatment the numbers of

inflammatory cells in the sputum of these patients

was the endpoint assessed.

There was, indeed, a significant reduction

in the total number of inflammatory cells in these

patients receiving fluticasone, and this was

largely accounted for by reduction in the numbers

of neutrophils in the sputum. Evidence that this

is, indeed, a treatment effect is afforded by a

washout phase during this study. As you can see,

the numbers of inflammatory cells then increased

back to pre-baseline values.


So, I think we can conclude, based on the

ten published studies, that there is overwhelming

evidence of an anti-inflammatory effect in this

disease, a reduction in some of the key

inflammatory cells and inflammatory mediators. In

addition, there is at least experimental evidence

that inhaled steroids, like fluticasone propionate,

may have a beneficial effect on some of the

structural changes associated with COPD, in

particular, to reduce the degree of hypertrophy of

glandular tissue.


The next important question then is to

assess the evidence for inhaled corticosteroids

having a clinical effect in this disease.


Here, we have 19 studies from which to

draw the evidence. Thirteen of these studies --

and you saw this table in Dr. Wheadon's

presentation, concluded that there was a clinical

efficacy effect of inhaled corticosteroids in COPD.

These studies ranged from four weeks duration up to

three years of treatment with the inhaled steroid.

They involved a total of four different types of

inhales steroids administered on a daily basis,

with the doses shown in this column of the table.

The clinical outcomes assessed here were

to show an increase in either pre- or

post-bronchodilator FEV1. Six of the studies

showed a reduction in symptoms in COPD and four of

the studies showed a reduction in exacerbations.

The longer-term studies, although not showing an

impact on decline in lung function over time, did,

indeed, show a clinical effect on other outcomes

and I will come back to this later in the


Six studies concluded there was no

clinical benefit for inhaled steroids in this

disease. Two of these studies were also those that

failed to find an anti-inflammatory effect of

inhales steroids. In a further three studies, the

patients involved in these studies had lung

function equal to or greater than 80 percent



As before, what I would like to do is to

take a number of these studies now and look in a

little bit more detail. I would like to begin with

this study, which is a large Canadian epidemiology

study that was conducted in Ontario between the

years of 1992 and 1997. It involves more than

22,000 patients that had been hospitalized as a

result of an exacerbation of COPD.

The study focused on the outcome of the 12

months after discharge and looked at all-cause

mortality in these patients or the risk of repeat

hospitalization. The analysis showed, in fact,

that those patients that were previously taking

inhaled corticosteroids had a 26 percent lower

relative risk of either all-cause mortality or

repeat hospitalization as a result of an

exacerbation of their disease.


What I would like to do now is to go on to

look at two studies that have specifically looked

at fluticasone propionate. The first study is the

Paggiaro study. Patients included in this study,

more than 280 in number, were classified as

non-reversible in that they did not achieve more

than a 15 percent increase in FEV1 following


The study involved fluticasone propionate

at a dose of 500 mcg twice daily, and the study was

conducted over a 24-week period. Those patients

receiving the inhaled steroid showed a progressive

increase in pre-dose FEV1, whereas those on placebo

showed a progressive decline in pre-dose FEV1. At

24 weeks there was a 160 ml difference between the

steroid treated arm and the placebo treated arm.


The key significant factor about the

Paggiaro study was that this was the first study

that was designed to look at exacerbations of this

disease. The patients included in this study -- as

I said, there were more than 280 -- had a history

of exacerbations. They had at least one per year

for the previous three years.

For the purposes of this study,

exacerbations were defined as worsening of COPD

symptoms requiring changes to normal treatment.

Severity of exacerbations were further defined if

they were mild, if they were self-managed by the

patient at home; if they were moderate they were

treated by a physician; and if there were severe

exacerbations, they required the patient to be

hospitalized. The last point is that multiple

exacerbations requiring oral corticosteroids were

allowed to be included in the analysis of this



There were about 140 patients in the

placebo and 140 patients in the fluticasone

propionate treated arm of the study. There was a

numerical reduction in the total number of

exacerbations in the steroid arm but this did not

reach statistical significance. When the number of

patients that experienced one or more exacerbations

were further analyzed, again there was no change in

the total number but there was a significant

decrease in the numbers of patients experiencing

either a moderate or a severe exacerbation and,

interestingly, a significant increase in the number

of patients experiencing a mild exacerbation,

suggesting that the steroid treatment was changing

the spectrum of exacerbations from moderate to

severe towards the mild end of the spectrum.


The next study that has involved

fluticasone that I would like to discuss is the

ISOLDE study. This was the largest and longest

study conducted with fluticasone propionate in

COPD. It involved more than 750 patients who,

again, were non-reversible, showing a less than 10

percent change in predicted values, and the mean

FEV1 was 50 percent at baseline.

The study involved first of all a

two-month period where corticosteroids were

withdrawn, this run-in period. Interestingly, a

subanalysis carried out by Gerard and colleagues,

and published in Respiratory Medicine in 1999,

showed that in the patients that were withdrawn

from steroids there was a six-fold higher incidence

of exacerbations compared to those patients not

previously treated with this class of drug.

After the run-in period the patients were

randomized to either receive fluticasone propionate

at a dose of 500 mcg a day from a metered-dose

inhaler or a corresponding placebo for a period of

three years.


In this slide I am looking at the

post-bronchodilator FEV1 data taken from the ISOLDE

study. On the vertical axis is the

post-bronchodilator FEV1; on the horizontal axis,

the time points up to three years of treatment.

Those patients receiving the inhaled steroid in the

first three months of treatment showed a

significant increase in the post-bronchodilator

FEV1 volume. This value then remained

statistically higher than those patients in the

placebo arm of the study, and there was no evidence

that these two lines were converging over the three

years of treatment.


The second important data from the ISOLDE

study was that the effect of the steroid on

exacerbation rate was assessed, and for the

purposes of this study exacerbations were defined

as requiring oral corticosteroid and/or antibiotic

intervention. The patients receiving fluticasone

propionate showed an approximately 25 percent

reduction in exacerbation rate in this study,

exacerbations per patient per year.


The final element of the ISOLDE study I

would like to review is the impact of the steroid

on quality of life or health status. This was

assessed in this study using the St. George's

Respiratory Questionnaire. Using this

questionnaire, the data is expressed that an

increase in score from this questionnaire reflects

a decline in quality of life. The data from the

ISOLDE study showed that fluticasone propionate

treatment reduced the decline in quality of life in

these patients. That decline was reduced to an

extent where there was approximately a two-fold

increase in the time required before the decline in

quality of life passed through a level of clinical



So, I think we can conclude that the

weight of evidence is in favor of a clinical effect

of inhaled corticosteroids in COPD. That effect is

to reduce symptoms, to increase both pre- and

post-bronchodilator FEV1 and to reduce



Turning now to salmeterol, and as we heard

from Dr. Wheadon, this is a drug already approved

for the use of COPD here, in the United States.


Salmeterol is quite clearly a long-acting

bronchodilator in this disease condition. This is

change in FEV1 on day one, shown in green, and

after 12 weeks of treatment shown in yellow.

Quite clearly the drug is effective in

increasing FEV1 and, importantly, this effect show

no effect of tolerance during this prolonged period

of treatment. In addition, an important increase

in baseline lung function was detected in these

patients as a result of exposure to the long-acting



So, I think we can conclude that the major

impact of a long-acting beta-agonist is to address

the component of airway obstruction and to reduce

the element of broncho-constriction associated with

this disease.


Now, if we consider salmeterol and

fluticasone in combination, what we have here are

two drugs that influence different aspects of the

underlying pathophysiological process, salmeterol,

as I have just said, largely addressing airway

obstruction and fluticasone propionate addressing

some of the key elements of the underlying

inflammatory process in this disease. When

salmeterol and fluticasone are brought together in

the context of Advair as a combination therapy,

there is an opportunity then to capitalize on the

fact that these drugs do influence different

elements of the underlying disease process and

therefore, there is an opportunity to have an

additive nature to the combined treatment. The

additive nature would increase, then, the reduction

in symptoms, the increase in bronchodilator

activity and a reduction in exacerbations.


However, finally, there is increasing

evidence that there is a positive molecular

interaction between corticosteroids and long-acting

beta-agonists. The first example of that is shown

on this slide. It comes from a study from Dr.

Braniuk and his colleagues at Georgetown

University. The study showed that the

administration of clinical doses of corticosteroids

-- here beclomethasone dipropionate was

administered intranasally for a period of three

days, and this led to an increase in the density of

beta-2 receptors in the respiratory mucosa of the


This effect is the result of activation of

the gene and coding for the beta-2 receptor. It is

a classic effect shown by all corticosteroids.

And, a result of this increased density of beta-2

receptors will be to promote the activity of



In turn, there is increasing evidence that

long-acting beta-agonists like salmeterol will

potentiate the inhibitory effect of

corticosteroids, like fluticasone propionate, on

the release of the inflammatory mediators from key

cells. In this study, which was conducted in human

airway smooth muscle cells, human necrosis factor

alpha was used to induce the release of the

neutrophil chemoattractant interleukin-8 and

TNF-alpha and IL8 are two key mediators in the

pathophysiology of COPD. The corticosteroid alone

has an inhibitory effect on IL8 release and, as you

can see from the slide, this effect is further

increased by the addition of the long-acting

beta-agonist salmeterol. This effect is due to

salmeterol priming the glucocorticoid receptor, the

target receptor for corticosteroids, and rendering

then that receptor more sensitive to

steroid-dependent activation. The outcome of this

then would be to promote the anti-inflammatory

effects of fluticasone propionate.


So, I can conclude then for the scientific

and clinical rationale for combining fluticasone

and salmeterol in the treatment of COPD in the

following way: fluticasone propionate is an

effective inhaled anti-inflammatory corticosteroid

and the evidence is overwhelmingly in support of a

clinical benefit in COPD. Salmeterol is a

long-acting beta-2 agonist with, again,

demonstrated efficacy in this disease. Each

molecule is, indeed, influenced by a different

aspect of the underlying pathophysiology of COPD

and, when brought together, they provide a broader

cover than either drug alone.

Finally, there is increasing evidence of a

positive interaction between these two classes of

drugs, which may be important in improving their

overall efficacy when used in combination in COPD.

I will now hand over to Dr. James Donohue,

who will give you a clinician's perspective on the

treatment of COPD. Jim?

Clinical's Perspective

DR. DONOHUE: Thank you, Malcolm.


Good morning. My name is Jim Donohue. I

am presently chief of the Division of Pulmonary and

Critical Care Medicine at the University of North

Carolina, in Chapel Hill. I am happy to be here

today to speak to the advisory committee on behalf

of GlaxoSmithKline. I will speak to you as a

clinician with many years experience in the

management of patients with COPD and the use of

COPD therapies. I will also speak to you as a

clinical investigator who has conducted numerous

clinical trials on COPD in the course of my career

for many different companies, including

GlaxoSmithKline. Therefore, I hope I can be of

some help and provide some insight into therapeutic

management of COPD.


Today I would like to focus on COPD from a

clinician's perspective, including how we diagnose

this condition and how treatment is evaluated. I

would like to also mention the Global Initiative

for Obstructive Lung Disease, or GOLD, guidelines

which are useful for COPD. Finally, I will discuss

how I use the available medications in my own

clinical practice.


COPD is a clinical diagnosis. It is based

on a patient's medical history, especially their

smoking history; their age; their symptoms; and

persistent airflow obstruction on spirometry. A

post-bronchodilator FEV1 of less than 80 percent of

predicted, in conjunction with an FEV1/FVC ratio of

less than 70 percent confirms the presence of

airflow limitation. The key words -- it is not

fully reversible. This is the definition of COPD

from the GOLD guidelines.


While the airway response to short-acting

bronchodilators, as expressed in the percent

increase or change from baseline FEV1, is very

important in the diagnosis of asthma.

Reversibility to albuterol does not exclude the

diagnosis of COPD and, in fact, is more the rule

than the exception. Recent treatment guidelines

for COPD do not include reversibility testing as a

criterion for the diagnosis of COPD, and we will

review some of the data to support this in the next

few slides.


One of the pivotal studies in North

American, known as the Intermittent Positive

Pressure Breathing Trial, the IPPB Trial, for COPD

was conducted in the late '70s and early '80s, and

published by Nick Anthonisen, in Winnipeg, and

colleagues. The study was supported by a grant

from the National Institutes of Health and involved

multiple sites.

The investigators evaluated the response

to isoproterenol, the short-acting bronchodilator.

It was a huge study, 985 patients, the largest

study at its time. It was conducted over a

three-year period. Pre- and post-bronchodilator

spirometry was evaluated at screening and

subsequently every three months over the duration

of the study. The entry criteria for COPD included

an FEV1 of less than 60 percent of predicted and an

FEV1/FVC ratio of less than 60 percent -- so very

reasonable criteria.

Looking at the demographic data, they are

very typical of the patients that we see in a

modern pulmonary practice and also in our clinical

trials program. The patients have a mean age of

61; at this stage predominantly male, although that

will change; the smoking status, 54 pack years; 40

percent were current smokers; and the lung function

as reflected in the FEV1 was 36 percent of



One of the most interesting things about

this most important study was the point that

reversibility is common in patients with COPD,

particularly in clinical trials. Using a 12

percent increase in FEV1 over baseline as a

threshold for defining reversibility, fully half of

these patients were reversible at baseline or at

screening. Furthermore, for those patients who

were non-reversible at the screening visit, 30

percent of those would be reversible on a

subsequent clinic visit. Overall, nearly 68

percent of these subjects had a 15 percent increase

in post-bronchodilator spirometry at least once

over the seven visits during the trial.

These data clearly demonstrate that

bronchodilator response at a single visit does not

necessarily correlate at subsequent evaluations,

and based on this study, reversibility really is

typical of COPD and doesn't have a whole lot of

relevance to the diagnosis. For this reason,

reversibility has not been included as a diagnostic

criterion in either the ATS guidelines nor the

recent GOLD guidelines.


To follow-up on this, data from several

large clinical trials conducted in COPD are

consistent with the IPPB trial. Shown on this

slide are the key demographic data from these

trials. The patients included are similar, of

course, to the IPPB. The mean age of the patients

is in their 60s; male predominant, heavy pack

smoking; FEV1 40 percent, 36 percent and 45.

Please draw your attention here to the

percent of patients who are considered reversible.

Now, the criteria for reversibility testing varies

amongst these trials, ranging from 12 to 15

percent, albuterol either two to four puffs,

ipratopium or the combination would be used. We

see 62 percent, 68-73 percent in the Durinsky

article for the Combivent data, and 42 percent for

formoterol. So, reversibility is very common in

COPD and, of course, it is not fully reversible.

Let's change gears now and see how we

assess treatment responses in COPD, both in

clinical trials and in our practices. It is

important to recognize up front that the magnitude

of changes in COPD are much less than asthma.


Spirometry is regarded as the gold

standard in evaluating COPD. It provides objective

and reproducible results. We use spirometry to

establish the diagnosis in our patients; to tell

them how severe their disease is or for prognostic

information; and we monitor the response to


Additional measures that we use to

evaluate treatment responses include assessment of

the health status or quality of life of our

patients, their symptoms and, most importantly,

their exacerbations. These other measures are more

subjective and usually studies are not optimally

powered or designed to detect treatment

differences. However, even small differences

across groups may translate to very important

benefits for our individual patients.


Currently very few drugs are approved for

the management of COPD. I would like to review two

relevant examples of the types of responses that we

typically observe. Here are the results from the

first combination product evaluated in subjects

with COPD, Combivent. As you can see, there is a

nice, brisk response and there is a reasonable

difference between the combination of ipratropium

and albuterol and the individual components. That

change is 70 ml. So, those of us who are used to

thinking in terms of asthma, that may not seem like

a great deal. Nonetheless, from my own personal

experience, this small change corresponds to large

benefits to the patients that we see in our

practices, and this is the first-line therapy for

the treatment of patients with COPD.


Furthermore, let's review some of the

other efficacy measurements that we look at in

clinical trials. Despite the improvement noted in

the FEV1 with the combination of albuterol and

ipratropium over its components, quality of life,

physician's global evaluation, COPD symptom scores

and the peak expiratory flow rate do not reach a

significant difference amongst treatment groups and

didn't change over time. Nonetheless from my

clinical experience and that of most

pulmonologists, this is a very, very valuable agent

and is widely used in patients who have symptoms.


Let's switch now to other clinical trials

that we have participated in. You have already

seen the FEV1 data from Dr. Johnson for the

salmeterol clinical trials program in North

America. Let's look at the additional efficacy

measures evaluated in those trials.

We note consistent improvement in peak

flow and Ventolin use when compared to placebo.

However, for many other measures the treatment

effects favored salmeterol but did not reach

statistical significance. This does not mean that

this drug is not beneficial in the treatment of

COPD, rather, it suggests that small changes

between treatment groups may underestimate the

benefit to individual patients.

As you can see from these two examples,

the types of treatment responses that we can expect

in COPD are quite modest. The effects from

Combivent and salmeterol, although modest in the

clinical trials, have proved to be extremely

valuable treatment options for our patients with



Let's address just briefly the GOLD

guidelines. These recently developed

evidence-based, Global Initiative for Chronic

Obstructive Lung Disease or GOLD guidelines have

evaluated the appropriate use of

pharmacotherapeutics in the therapeutic management

of COPD. It was my pleasure to serve as a

consultant reviewer for these guidelines. These

guidelines are endorsed by the American Thoracic

Society and the American College of Chest

Physicians. These guidelines clearly recognize the

role of bronchodilators in the treatment of COPD.

More importantly, the guidelines recognize that

bronchodilators alone or in combination are not

adequate to treat all the symptoms associated with

this disease. Many patients require a therapy with

other classes of medications including inhaled



In the GOLD guidelines four stages of

disease severity were established: at risk, mild,

moderate and severe. Based on these guidelines

almost all the patients that we see in the clinical

trials that I have presented would be classified as

having moderate to severe disease. For patients

with moderate to severe disease, maintenance

bronchodilator treatment is recommended. However,

for many patients this is not enough.

As discussed by Dr. Johnson, numerous

clinical trials have been conducted evaluating the

efficacy and safety of inhaled corticosteroids in

patients with COPD. The GOLD reviewing committee

evaluated all the peer reviewed published clinical

trials assessing the overall benefit-risk of

inhales steroids in COPD. Based on the totality of

the data, there was a consensus to recommend

inhaled steroids for symptomatic patients who

demonstrate a response to inhaled corticosteroids

or for those patients with an FEV1 of less than 50

percent of predicted who experience repeated


Thus, current evidence-based treatment

guidelines acknowledge the value of inhaled

corticosteroids in the therapeutic management of

moderate and severe COPD. While effective, oral

corticosteroids are associated with significant

side effects, as every one knows. For this reason,

the GOLD guidelines state that chronic maintenance

therapy with oral corticosteroids should be avoided

regardless of the severity of the disease.


COPD, make no mistake about it, is a

devastating disease. First and foremost, we want

to prevent worsening of this condition. In my own

practice I emphasize smoking cessation,

immunization, prevention of further lung injury

such as avoiding crowds in an influenza epidemic,

avoiding outdoor air pollution when the ozone

levels are high, or what-have-you. We also

strongly recommend an exercise program, some type

of pulmonary rehabilitation whether or formal or

just an exercise program at home.

The choice of pharmacotherapy is based on

the severity of our patient's symptoms, defined

both by lung function and symptoms. We usually

begin with bronchodilators, anticholinergics,

long-acting bronchodilators either alone or in

combination. However, for many patients, despite

optimal use of bronchodilators, control of their

disease remains quite poor. For these patients I

would institute a trial of inhaled corticosteroids

per the GOLD guidelines recommendations. My

experience working in a university practice is

similar to that of SININ2 of Toronto. I have found

that inhaled steroids reduce exacerbations and

reduce the exposure of our patients to oral

corticosteroids which are part and parcel of the

treatment of an exacerbation.


In conclusion, the diagnosis of COPD is

based on several clinical parameters. As I have

illustrated, a high proportion of the patients with

COPD do demonstrate reversibility to

bronchodilators. For this reason, the presence of

reversibility does not exclude the diagnosis of

COPD. In evaluating treatment response, spirometry

still remains the gold standard. We have had

considerable success and experience with use of

this measure in assessing treatment responses

compared to other more subjective measures.

In general, treatment responses in COPD

are quite small, very modest and often are quite

variable. However, these small changes between

treatment groups may underestimate the benefits

seen in individual patients. The use of inhaled

corticosteroids in COPD is advocated by

evidence-based international guidelines. My

personal experience and practice are consistent

with this view. Even more importantly, I have

found that use of inhaled steroids has reduced

reliance on oral corticosteroids with their much

greater safety concerns.

I would like to thank you for this

opportunity to address the committee and to speak

on these issues. I would next like to introduce

Dr. Tushar Shah who will be presenting the clinical

results on the Advair and Flovent Diskus program.

Thank you very much.

Clinical Efficacy and Safety

DR. SHAH: Thanks, Dr. Donohue, and good

morning everyone.


My name is Tushar Shah, and I am the vice

president of Respiratory Clinical Development for



In the next 50 minutes, I am pleased to

review results of Flovent and Advair Diskus program

which was designed in consultation with the FDA. I

will share with you results which will demonstrate

that we achieved the primary objectives for this


For Flovent, the results will show that we

demonstrated greater efficacy on the primary

efficacy measure compared to placebo, with no

significant safety concerns. For Advair, the

results will show that we demonstrated greater

efficacy than the primary efficacy measures

compared to each component, without additional

safety concerns.

We realize that long-term safety

information will be an important consideration in

assessing the benefit-risk ratio for the use of

inhaled corticosteroids in the treatment of COPD.

I will also summarize some of the relevant

long-term safety data from the use of fluticasone

propionate in asthma and COPD which will support

the safety information from our clinical program.


Before I begin, it may help for me to

define some of the abbreviations I will be using in

my presentation. FSC refers to fluticasone

propionate and salmeterol combination product.

Whereas the slides will display FSC 500/50 and FSC

250/50, I will refer to Advair 500 and Advair 250

in my text for purposes of clarity and ease of



We performed three large multicenter,

randomized, double-blind, placebo-controlled trials

for the develop of Flovent and Advair in COPD. All

three studies were conducted in the U.S. and had

identical inclusion/exclusion criteria. For

Flovent Diskus, FLTA3025 compared two dosages of FP

versus placebo, whereas SFCA3006 and SFCA3007

compared a single dose of FP versus placebo.

Hence, three independent studies were performed to

assess the effects of Flovent in COPD. For Advair

Diskus, SFCA3006 and SFCA3007, compared Advair to

each individual component at the corresponding FP

dose and to placebo. In all three studies

treatments were administered twice daily for 24

weeks duration.


The design of FLTA3025 included a two-week

placebo run-in period during which time patients

discontinued all COPD medications other than PRN

albuterol. The use of concurrent methylxanthines

was permitted as long as the dose remained

relatively constant during the trial.

The purpose of the run-in period was to

assess if patients met enrollment criteria for

randomization and to ensure their adherence to

study procedures. Eligible patients were then

randomized to either FP 500, FP 250 or placebo for

24 weeks of treatment.

Patients were evaluated at regular

scheduled visits during the course of the trial and

we had slightly over 200 patients in each treatment

group in this trial. Pulmonary function tests and

symptom and quality of life questionnaires were

administered at clinic visits, and patients also

completely diary cards for collection of some

efficacy and safety information.


The study design for SFCA3006 and SFCA3007

were similar to FLTA3025, with the exception of the

treatment groups. In SFCA3006 Advair 500 was

compared to Flovent Diskus 500, Serevent Diskus and

placebo. The number of patients enrolled in this

trial ranged from 164 to 185 per treatment group.


In SFCA3007 Advair 250 was compared to

Flovent Diskus 250, Serevent Diskus and placebo.

The number of patients enrolled in this trial

ranged from 177 to 185 per treatment group.

One advantage of this study design is that

within each study we actually have two

opportunities to assess the effect of the 250 mcg

dose of FP. One is comparing FP alone versus

placebo. The second is comparing Advair 250 versus



The inclusion/exclusion criteria were

identical for the three trials and were comparable

to the criteria used in previous clinical trials

conducted for COPD. Patients had to be 40 years of

age or older and have COPD as defined by ATS in

order to enter these trials. Patients could be

current or former smokers, with a 20 pack a year or

greater smoking history. They had to have FEV1

less than 65 percent predicted and an FEV1/FVC

ratio less than or equal to 70 percent.

Additionally, patients had to achieve a score of

greater than or equal to 2, which is regarded as

moderate dyspnea on the Modified Medical Research

Council, or MMRC, dyspnea scale at screening, and

also have symptoms of chronic bronchitis, morning

cough and sputum at baseline in order to enter

these trials.


Patients were excluded from the trials if

they had a current diagnosis of asthma. Patients

were also excluded if they needed to use systemic

corticosteroids or high dose inhaled

corticosteroids, defined as a dose of greater than

1000 mcg a day of fluticasone propionate or an

equivalent or other inhaled corticosteroids during

the six weeks prior outcome the screening visit.

Patients were also excluded if they needed

long-term oxygen therapy or experienced COPD

exacerbation during the run-in period.


As reviewed by Dr. Donohue, FEV1 was

selected as the primary efficacy measure since it

is clinically relevant, objective and, most

importantly, has been useful in discriminating

treatment effects in COPD. These studies were

optimized to evaluate this measure. Since

salmeterol and FP exert their pharmacological

action by different mechanisms, two measures of

lung function were prespecified as the primary

efficacy measure for assessing treatment effects.

Pre-dose FEV1 was used to compare FP

versus placebo and to evaluate the contribution of

FP and Advair when compared to salmeterol.

Two-hour post-dose FEV1 was used to evaluate the

contribution of salmeterol and Advair when compared

to FP. The two-hour post-dose FEV1 was selected

because it corresponded to the peak bronchodilation

period for salmeterol and correlated well with the

post-dose 12-hour FEV1 AUC results. This approach

for selection of primary efficacy measures was

reviewed and agree with the FDA prior to initiating

our trials.


The secondary efficacy measures we

discussed and agreed with the FDA for inclusion in

this program included the transition dyspnea index,

or TDI, for assessment of dyspnea; the chronic

respiratory disease questionnaire, or CRDQ, for

assessment of quality of life, and the chronic

bronchitis symptoms questionnaire, or CBSQ, for

assessment of symptoms of cough and sputum

production. These three measures were prespecified

as key secondary efficacy measures. The TDI and

CRDQ are validated instruments and have defined a

change from baseline which is regarded as

clinically significant. The CBSQ was a new

instrument which had not been previously validated

or had been evaluated in a clinical trial.

Additional secondary efficacy measures included

daily diary card information, such as morning peak

flow, Ventolin use and nighttime awakenings.

It is important to note that Ventolin use

in this program was PRN. Hence, the use of this

product during the course of the trial represents a

marker of symptoms. This is also true for

nighttime awakenings since only information on

awakenings requiring Ventolin were collected.

Exacerbations based on physician discretion were

also recorded, and were defined by the need for

treatment with antibiotics and/or oral

corticosteroids. This is similar to the definition

that has been used in other COPD trials.


Since these trials were of six months

duration and placebo-controlled, withdrawals from

double-blind treatment were anticipated. In order

to allow for a potential bias caused by patient

withdrawal in the analysis of the results, endpoint

defined a priori was used. The endpoint

represented the last baseline observation. This

allowed us to include nearly all patients who

received study drug in our efficacy analysis.


Patient demography and baseline

characteristics integrated for all three studies

are presented on this slide. Results for the

individual studies were similar to these and are

included in your briefing document. Patient

demography and baseline characteristics were

similar across treatment groups for the integrated

data, as well as for the individual studies.

Patients enrolled in these trials had a mean age of

approximately 63 years. About 65 percent were

male. About 94 percent were Caucasian. Half were

current smokers and had a greater than 60 pack year

smoking history.

In this program, slightly more than 25

percent used inhaled corticosteroids previously.

This is lower than the level of inhaled

corticosteroid use in current practice. They had

moderate to severe alveolar obstruction of

approximately 41 percent predicted, and slightly

more than half were reversible to albuterol,

defined as greater than 12 percent, and 200 ml

increase in FEV1 following four puffs of albuterol.

As reviewed by Dr. Donohue, this is consistent with

the level of reversibility seen in previous

clinical trials conducted in COPD. Approximately

73 percent of these patients were reported as

having emphysema. There can be no question that

these patients are representative of the types of

patients who are likely to be diagnosed and managed

as having COPD in the U.S.


I will now share the efficacy results from

these trials. Due to time constraints, I will

focus my presentation on the primary measures of

efficacy and briefly summarize the findings of the

secondary efficacy measures. I will first review

the results for Flovent, followed by Advair.


As previously mentioned, the primary

measure of efficacy for assessing the treatment

effects of FP was morning pre-dose FEV1.


The results of pre-dose FEV1 from FLTA3025

are displayed on this slide. Before reviewing

these results, let me quickly orient you to the

information on this slide. The Y axis represents

the change in FEV1 in milliliters and the X axis is

the study week. Additionally, on the right side of

the slide are presented the endpoint results for

the treatment groups. We have also provided the

percent change from baseline in FEV1 at endpoint,

and results of statistical analysis will only be

presented for endpoint. I will be using this

format during the next few slides which will be

reviewing the FEV1 results.

On this slide the FP 500 treatment group

is depicted in orange, on the top; the FP 250

treatment group in yellow, in the middle; and the

placebo in blue, on the bottom. Results from this

trial indicate that treatment with FP was

associated with dose-related improvements in FEV1.

However, we were surprised by the smaller

improvement seen in this trial compared to previous

results reviewed by Dr. Johnson.

As the figure indicates, the greatest

separation from placebo occurred near the end of

the treatment period. At endpoint the improvements

in FEV1 were significantly greater for the FP 500

compared to placebo, with a model estimated

treatment difference of 57 ml. In this trial no

significant differences between FP 250 and placebo

were observed for FEV1, the model estimated

treatment difference being 32 ml. As you will see

in the next two trials, we had more robust

treatment effects with both doses of FP in COPD.


Results for pre-dose FEV1 for the FP 500

treatment group from SFCA3006 are shown on this

slide. The other treatment groups have been

omitted for purpose of clarity and will be

presented later.

The FP 500 treatment group is depicted in

orange and the placebo group in blue. Results from

this trial indicate that treatment with FP 500 was

associated with more robust and significantly

greater improvements in pre-dose FEV1 compared to

placebo. Once again, the greatest separation from

placebo occurs near the end of the trial,

indicating we may have not reached a plateau for

treatment response. At endpoint the improvements

in FEV1 were significantly greater for FP 500

compared to placebo, with a model estimated

treatment difference of 105 ml.


Results for pre-dose FEV1 for the FP 250

treatment group from SFCA3007 are shown on this

slide. In contrast to what we saw on FLTA3025,

results from this trial indicate that treatment

with FP 250 was associated with a more robust and

significantly greater improvement in pre-dose FEV1

compared to placebo. Even in this trial we do not

appear to have reached a plateau in the treatment

response. At endpoint the improvements in FEV1

were significantly greater for FP 250 compared to

placebo, with a model estimated treatment

difference of 112 ml.


On this slide we have provided results for

all four treatment groups from SFCA3007. In

addition to the comparisons with the FP 250 and

placebo group which I just reviewed, we have

provided the results with Advair and salmeterol

treatment groups in purple and green, respectively.

This study provides us a second

independent opportunity to assess the treatment

effects of the FP 250 mcg dose comparing Advair to

salmeterol. Greater improvements in FEV1 were seen

with Advair versus salmeterol, with a model

estimated treatment difference of 69 ml at

endpoint, which was statistically significant.

Hence, we have three opportunities to

assess if the FP 250 mcg twice daily dose provides

clinical benefits in COPD. In two of the three

instances we demonstrated robust treatment effects

with this dose of FP.


This slide summarizes the pre-dose FEV1

results for patients who were defined as reversible

or non-reversible at baseline for the FP treatment

groups across the three trials. As expected, a

greater magnitude of response was observed with FP

treatment in the reversible patients since, by

definition, these patients had greater room for

improvement. However, in SFCA3006 and SFCA3007,

where we had more robust treatment effects, even in

the non-reversible patients fairly large treatment

effects were observed in this population. As

reviewed by Dr. Johnson, results from studies

conducted in less reversible patients indicate that

FP provides benefits beyond improvements in lung

function. In these trials, reduction in

exacerbations and improvements in health status

were also seen with FP treatment.


This table summarizes the results of the

statistical analysis for the secondary efficacy

measures between FP and placebo across the three

trials. A check represents where p values were

less than 0.05, and dash where p values were

greater than 0.05. It is important to note that in

SFCA3006 and SFCA3007 we amended the protocols a

priori to adjust for multiple comparisons for the

three key secondary efficacy measures, denoted by a


Before I review these results, I would

like to emphasize that we designed and optimized

these studies for the primary, not secondary,

endpoints. So, our expectations for secondary

measures are that they should be supportive of the

findings we see on our primary efficacy measures.

This is what we observed in these trials.

For comparisons of FP versus placebo

greater improvements were seen for nearly all

secondary efficacy measures with both doses of FP

across the three trials, with most differences

achieving p values less than 0.05. In FLTA3025

many of the differences from placebo for secondary

measures achieved p values less than 0.05 for the

FP 250 compared to the placebo comparison, as is

shown in the first column. This indicates that

even in this trial FP 250 provided clinical

benefits in the treatment of COPD. For most

measures similar improvements were seen between the

two doses of FP with the exception of TDI, which is

shown in the first line. For this measure, the FP

500 twice daily dose was consistently and

significantly better than placebo, and in SFCA3006

achieved a clinically significant difference of 1

from placebo. This was the only instance where we

achieved a predefined, clinically significant

difference between placebo and FP for the three key

secondary efficacy measures.

In general, the treatment effects were not

significant for the CBSQ compared to placebo,

suggesting that this new questionnaire may not be

sensitive at discerning treatment effects.

Unlike previous trials which demonstrated

inhaled corticosteroid treatment, including FP, was

associated with reductions in exacerbations, in

this clinical program none of the treatments

significantly reduced the time to COPD

exacerbations compared to placebo. The most likely

reason for this discrepancy is that we did not

require patients to have a history of COPD

exacerbations for entry into these studies, and we

withdrew patients who needed oral corticosteroid

bursts for treatment of an exacerbation.

We did this because we wanted to ensure

that the use of concomitant oral corticosteroids

did not hinder our ability to assess treatment

effects on the primary efficacy measure, which is

FEV1. If our primary endpoint would have been to

examine exacerbations, we would have designed a

clinical trial very differently than what we have

and more like what Dr. Johnson reviewed where

significant improvements with FP treatments on

exacerbations were seen. These results do support

the efficacy of FP in treatment of COPD.


The efficacy results for Flovent from

these trials can, hence, be summarized as follows:

In all three studies greater improvements in the

primary efficacy measure, pre-dose FEV1, were seen

with FP treatment compared to placebo. In FLTA3025

a dose-related improvement was seen with a response

to FP 500 achieving statistical significance. In

SFCA3006 and SFCA3007 more robust improvements in

FEV1 were observed which were significantly

different from placebo. Comparison of the response

with Advair versus salmeterol in SFCA3007 provides

additional evidence for the benefits of the FP 250

twice daily dose in COPD. As expected, a greater

magnitude of response in FEV1 was observed in

reversible versus non-reversible patients following

FP treatment.

Results for the secondary efficacy

measures support the primary analysis. Greater

improvements were demonstrated for most secondary

efficacy measures with FP compared to placebo, with

many differences achieving p values less than 0.05.

Overall, both doses of FP provided comparable

benefits with some suggestion of a dose effect. In

FLTA3025 dose-related improvements in FEV1 were

seen and were consistent and greater improvements

in dyspnea as measured by the TDI were seen with

the FP 500 versus the FP 250 dose.

The magnitude of improvements for most

efficacy measures seen with both doses of FP were

similar to that seen with currently available

treatments, as reviewed by Dr. Donohue. These

treatments that are currently available are

regarded as clinically useful in the management of

COPD. This indicates that the benefits we see with

FP treatment in COPD will also be clinically

important for these patients.


I will now share the efficacy results for

Advair from these trials. Due to time constraints,

once again I will focus my presentation on the

primary efficacy measures and briefly summarize the

findings of the secondary efficacy measures.


As previously noted, to assess the

contribution of FP pre-dose FEV1 was defined as the

primary efficacy measure to compare Advair versus

salmeterol. This approach had been agreed with the

FDA during the design of the program.


Results for pre-dose FEV1 from SFCA3006

are shown on this slide. I would like to draw your

attention to the purple line on top which

represents Advair 500, and the green line below it

which represents the salmeterol treatment group.

As before, the Y axis represents the mean change in

FEV1 in milliliters and the X axis is the study

week of treatment. On the right side we have

included the endpoint results, and have highlighted

statistically significant differences only for the

comparisons being discussed.


We have now included the FP, depicted in

orange, and the placebo group, in blue, for

completeness. Results from this trial indicate

that treatment with Advair 500 resulted in

significantly greater improvements in pre-dose FEV1

compared to salmeterol. Improvements were noted as

early as the first week, with maintenance of

improvement during the treatment period. At

endpoint the improvements in FEV1 were

significantly greater for Advair 500 compared to

salmeterol, with a model estimated treatment

difference of 67 ml. The model estimated treatment

difference between Advair and placebo was 159 ml.


Results for the pre-dose FEV1 from

SFCA3007 with Advair 250 are shown on this slide.

The results indicated that treatment with Advair

250 was associated with significantly greater

improvements in pre-dose FEV1 compared to

salmeterol. At endpoint the improvements in FEV1

were significantly greater for Advair 250 compared

to salmeterol, with a model estimated treatment

difference of 69 ml. The model estimated treatment

difference between Advair 250 and placebo in this

trial was 161 ml. The magnitude of improvements

with both doses of Advair on pre-dose FEV1

represent an advance in the treatment of COPD.


As reviewed previously, to assess the

contribution of salmeterol two-hour post-dose FEV1

was used to compare Advair versus FP.


Results for the two-hour post-dose FEV1

from SFCA3006 with Advair 500 are shown on this

slide. The purple line on top represents Advair

500 and the orange line below it represents FP.


We have now included the remaining

treatment groups, salmeterol in green and placebo

in blue, for completeness. Results from this trial

indicate that treatment with Advair 500 was

associated with significantly greater improvements

in the two-hour post-dose FEV1 compared to FP. At

endpoint improvements in FEV1 were significantly

greater for Advair 500 compared to FP, with a model

estimated treatment difference of 129 ml. Compared

to placebo, this improvement was also significant,

with a model estimated treatment difference of 232



Results for the two-hour post-dose FEV1

from SFCA3007 with Advair 250 are shown on this



Results from this trial indicate that

treatment with Advair 250 was associated with

significantly greater improvement in two-hour

post-dose FEV1 compared to FP 250. Improvements

were noted as early as the first week, with

maintenance of improvement over the study interval.

There was no evidence that the benefits waned with

continued treatment. At endpoint improvements in

FEV1 were significantly greater for Advair 250

compared to FP, with a model estimated treatment

difference of 124 ml. Compared to placebo, this

improvement was also significant with a model

estimated treatment difference of 214 ml.


This slide summarizes the pre-dose and

post-dose FEV1 results for patients defined as

reversible and non-reversible at baseline for the

four treatment groups in the two Advair trials. As

expected, a greater magnitude of response was

observed with all treatments in the reversible

patients since, by definition, these patients have

greater room for improvement. However, for the

pre-dose FEV1 responses in even the non-reversible

patients were over 100 ml compared to placebo for

both doses of Advair. This indicates robust

treatment effects for Advair even in this



This table summarizes results of the

statistical analysis for the secondary efficacy

measures between FP and salmeterol versus placebo

for the two Advair trials. A check represents

where p values were less than 0.05, and a dash

where p values were greater than 0.05. As

previously noted, in SFCA3006 and SFCA3007 we

amended the protocols a priori to adjust for

multiple comparisons for the three key secondary

efficacy measures, denoted by a star.

Results for salmeterol and FP on secondary

efficacy measures in these two trials were

comparable, with several comparisons achieving p

values less than 0.05. However, only treatment

with FP was associated with significant differences

from placebo for any of the key secondary efficacy

measures. These data support the efficacy of the

individual agents in COPD.


For Advair, we have used the same

presentation format for the statistical analysis

compared with placebo as shown on the previous

slide. Greater improvements were seen for nearly

all secondary efficacy measures with both doses of

Advair versus placebo. Most of these differences

resulted in p values less than 0.05. This is

better than what was seen with FP and salmeterol

alone, shown in the previous slide, indicating that

both components are contributing to the effects we

see with Advair.

Additionally, for the TDI and the CRDQ,

only treatment with Advair consistently achieved a

clinically important change from baseline as

specified by the developer or this instrument.

However, none of the differences between treatment

groups achieved this magnitude of change with the

exception of TDI in SFCA3006. As already reviewed,

none of the treatments in this program were

associated with reductions in time to COPD

exacerbations, most likely due to differences in

study design and duration.

For most measures similar improvements

were seen between the two doses of Advair with the

exception of TDI, shown on the first line. For

this measure, Advair 500/50 provided greater

magnitude of improvements, which were significantly

better compared to placebo and salmeterol, as I

will review shortly.


In addition, when numerical trends for

greater effect with Advair versus the individual

components were seen for most secondary efficacy

measures, p values less than 0.05 for these

comparisons were demonstrated in some instances

only. This is shown by a green check for Advair

versus salmeterol and an orange check for Advair

versus FP, on this slide. The most likely reason

for the lack of significance between Advair and

components for some of these secondary measures is

that these trials were not optimally designed to

assess treatment effects on the secondary efficacy


I will now share with you results for TDI

from SFCA3006 with Advair 500 and morning peak flow

from SFCA3007 with Advair 250 to illustrate the

types of effects we observed on these secondary



Shown on this slide are the results of the

transition dyspnea index, or TDI, for Advair 500.

The Y axis represents the TDI result, whereas the X

axis represents the study week. On the right side

of the slide are the endpoint results for the four

treatment groups.

The TDI measures the change in patient's

level of dyspnea from baseline. A value of 1 has

been defined as a clinically significant treatment

effect by the developer of the instrument. These

results demonstrate that treatment with Advair 500

was associated with greater improvements in dyspnea

as assessed by the TDI score compared to each of

the individual agents. Improvements were noted as

early as the first week, with further improvements

noted during the trial. At endpoint the

improvements were significantly different for

Advair compared to placebo, with a model estimated

treatment difference of 1.7. It was also

significantly greater compared to salmeterol, with

an estimated treatment difference of 1.2. Both of

these differences represent a clinically important


It is important to note that while the

results with FP and salmeterol for TDI were

similar, only the FP 500 treatment group achieved

statistical significance and a clinically

meaningful difference from placebo. The magnitude

of improvements in TDI seen with Advair 500

represent one of the best treatment effects of any

medication which was evaluated with this



Shown on this slide are the results of the

mean change on morning peak flow from SFCA3007,

which was the Advair 250 trial. The Y axis

represents the change in peak flow in liters per

minute, and the X axis represents the day of


Treatment with Advair 250 was associated

with a greater increase in morning peak flow

beginning one day after initiating treatment, which

increased further during the course of the trial.

The magnitude of improvement in peak flow between

Advair versus each component achieved p values less

than 0.001. Similar findings were observed with FP

and salmeterol versus placebo, with the magnitude

of improvement seen with FP and salmeterol being

comparable. Results from SFCA3006 with Advair 500

were similar to these results. Hence, these

results for the secondary measures do support the

primary efficacy measures we see with Advair in the

treatment of COPD.


The efficacy conclusions for the results

with Advair can be summarized as follows:

Significantly greater improvements with Advair were

seen on the primary efficacy measures at each

strength. This was shown for Advair versus

salmeterol for pre-dose FEV1, and for Advair versus

FP for the two-hour post-dose FEV1. Results of the

secondary efficacy measures support the primary

analysis. Greater improvements with Advair versus

placebo were seen for nearly all measures,

indicating that both components were contributing

to the benefit seen with Advair.

Trends for greater improvements versus the

individual components were also seen for most

efficacy measures, with some differences achieving

p values less than 0.05. As expected, greater

treatment effect was seen in more reversible

patients. However, the improvements in FEV1 with

Advair relative to placebo at both doses were

robust even in the patients regarded as

non-reversible. As agreed with the agency, a

formal dose response assessment was not performed

for Advair. However, comparing the responses to

Advair in the two trials, it appears that Advair

250 and Advair 500 provided similar benefits for

most efficacy measures, with the exception of

dyspnea as assessed by TDI. A significant and

clinically meaningful greater improvement in

dyspnea was only seen with Advair 500 compared to

placebo or salmeterol. The magnitude of

improvements seen with Advair represent a real

advance in the treatment of COPD.


I will now share with you some of the

safety results from the Flovent and Advair clinical

program. I will present results of the integrated

data since it includes all patients enrolled in the

clinical program and represents the best method for

determining treatment effects. I will focus my

presentation on addressing the two main issues

which are relevant with regards to assessing the

safety of inhaled corticosteroids in the treatment

of COPD.

The first is the issue of topical effects,

with specific review of the pneumonia cases. The

second is the evidence we have regarding systemic

safety of administering FP in COPD. I will not be

reviewing results of laboratory data,

cardiovascular data or assessment of safety in

different populations. Overall, we did not see any

evidence of a safety concern in these measures and

groups. This information is provided in your

briefing documents and is available for review

during the Q&A if needed.


The safety database as part of this

program was comprised of 2054 patients with COPD,

790 of whom were treated with FP and 347 were

treated with Advair. This safety data was

supported by safety data from 1298 additional COPD

patients from non-U.S. trials evaluating FP and by

the extensive safety data in trials conducted in



Displayed on this slide are the percent of

patients with adverse events, withdrawn due to

adverse events, and experiencing serious adverse

events from the three U.S. trials. Four patients

died in the trial in the placebo group. The mean

duration of exposure was similar to slightly more

in the active treatment groups versus placebo. The

percent of patients who experienced adverse events

was slightly higher in the FP-containing groups.

This was primarily due to a higher incidence of

expected topical adverse events associated with the

use of inhaled corticosteroids. A slightly higher

percentage of patients was withdrawn due to adverse

events in the FP 500 group, however, most of these

events were not attributed to the drug treatment by

the investigators. A similar percent of patients

experienced serious adverse events across the

treatment groups. There was no evidence that

treatment with Advair was associated with a higher

incidence of adverse events compared to the

individual agents or placebo in this program.


This slide summarizes the adverse events

of special interests when inhaled corticosteroids

are administered. A slightly higher incidence of

expected topical adverse events such as

candidiasis, throat irritation and

hoarseness/dysphonia was seen in the treatment

groups containing FP. These were generally

considered mild to moderate in severity and rarely

led to patient discontinuing treatment.

The incidence of AEs which are attributed

to systemic corticosteroids, such as fractures,

cataracts or ocular pressure disorder, occurred in

a similar rate across the treatment groups.


During the review, the FDA raised the

concern that a higher incidence of pneumonia

occurred in treatment groups containing FP. In

order to better understand these concerns, we have

summarized the adverse events and serious adverse

event of pneumonia which occurred during the

trials. These numbers may be slightly different

from those in your briefing document but are the

most accurate reflection of the events. These

changes have been reviewed with the agency.

The incidence of pneumonia overall was

low, with variable distribution across the

treatment groups. While there appears to be a

slightly higher incidence of these events in the

FP-containing groups, we do not see this in the

groups containing Advair, hence, attribution to

drug treatment is questionable.


HPA axis assessments were performed by the

measurements of 12-hour cortisol profile in

FLTA3025 in 86 patients, and assessment of morning

cortisol concentrations and short ACTH stimulation

test in SFCA3006 and SFCA3007 in 359 patients.


As expected, we did see a dose-related

decrease in the 12-hour unstimulated plasma

cortisol profile at these doses of FP. This was

significantly different from placebo with FP 500

but not the FP 250 twice daily dose. Now, we have

to remember that this test is regarded as a very

sensitive test for assessing the presence of

exogenous corticosteroids. From our experience in

asthma, at these doses this magnitude of HPA axis

effects has not been associated with clinically

significant changes in other more clinically

relevant measures, as I will review shortly.

The incidence of abnormal morning cortisol

and short ACTH stimulation test was low and similar

between the treatment groups. This further

supports that the small changes in unstimulated

plasma cortisol profiles are unlikely to be

associated with clinically significant HPA axis



The safety results from our clinical

program can be summarized as follows: Treatment of

COPD patients with v Diskus was well tolerated.

Other than a slightly higher incidence of expected

topical adverse events, no new clinically relevant

safety concerns were identified compared to


Treatment of COPD patients with Advair

Diskus was also well tolerated. There was no

evidence of a greater safety risk compared to the

use of the individual agents or placebo.


I will now review some of the long-term

safety data that we have available with FP. I will

focus my presentation on the systemic safety data.

It is well established that the systemic side

effects of inhaled corticosteroids are due to the

amount of the drug absorbed into the body. Less

systemic absorption of FP in COPD than in asthma

allows us to extrapolate the systemic safety data

from asthma to patients with COPD. This approach

was agreed with the FDA during the design of this

clinical program.

Since we observed similar or less systemic

exposure to FP in COPD compared to some of the data

in asthma, I will review results from clinical

trials in asthma which examined the effects of FP

treatment on bone mineral density and ocular

effects. In addition to our extensive long-term

safety data from asthma, I will review the findings

from relevant safety information from a three-year

trial conducted in patients with COPD. This was

the ISOLDE trial was reviewed previously by Dr.

Johnson in his presentation.


Shown on this slide are the systemic

exposures seen with FP in patients with asthma on

the left and COPD on the right, as assessed by the

area under the curve, or AUC, of the plasma FP

concentrations versus time measurements. The Y

axis on this slide represents the FP AUC in

pgm/hour/ml and the square boxes are the individual

patient systemic exposure to FP obtained from the

FLTA3001 trial, using the CFC MDI formulation of FP

which is currently on the market. I will be

reviewing the results of this trial shortly.

Similar results for the Diskus are presented in

circles for patients with COPD, on the right, from

our FLTA3025 trial. The 440 mcg dose from the MDI

corresponds to the 500 mcg dose of the Diskus.

These results indicate that the range of

exposure to FP in patients with COPD is similar or

less than what has been seen in asthma with the FP

CFC MDI. This is consistent with the relationship

noted in asthma where the systemic exposure from

the FP CFC MDI is approximately double that of the

Diskus. These analyses indicate that the results

from studies in patients with asthma conducted with

CFC MDI specifically can be used to assess the

potential for systemic side effects in COPD with

the FP Diskus.


This slide summarizes the clinical trials

which examined bone mineral density, or BMD, and/or

ophthalmic effects of FP treatment in adult

patients with asthma. We acknowledge that a unit

change in bone mineral density is of a greater

clinical concern in COPD patients since they have

lower bone mass compared to patients with asthma.

However, the evidence with oral corticosteroids

does not indicate that older, postmenopausal

patients are more sensitive to the BMD effects or

loss compared to younger patients with

corticosteroids. Hence, we believe that these data

from patients with asthma still are useful in

assessing the risk of bone mineral density effects

following FP treatment in patients with COPD. Our

data from asthma includes two two-year

placebo-controlled trials and five comparative

trials where assessments of bone mineral density

were performed.

The two placebo-controlled trials were

similar in design and patient inclusion and

exclusion criteria. One trial, FLTA3001, examined

two doses of FP, 88 mcg and 440 mcg twice daily,

and the other trial was FLTA3017, which examined a

single dose of FP 500 twice daily via the Rotadisk

versus placebo.

The comparative trials were all conducted

with the CFC MDI as well, and include three trials

which compared FP versus beclomethasone

dipropionate, or BDP, and two trials which compared

FP versus budesonide.


I will now review results for the two-year

randomized, placebo-controlled trial conducted with

the CFC MDI. Once again, as you recall, the

exposure we see with CFC MDI in this study was

similar or greater than what we observed with the

Diskus in COPD patients, as relevant to this study

to this application.

Asthma patients in this study had limited

systemic exposure to corticosteroids; were 18 to

50-year olds for males and 18 to 40-year olds for

women; and had to have had normal bone mineral

density or eye exams at baseline. The percent

change in lumbar spine bone mineral density in this

trial is displayed on this slide. The Y axis

represents the percent change in lumbar spine bone

mineral density in grams per centimeter squared at

the various times they were performed for the 25,

52, 76 and 104 weeks of treatment.

The placebo group is depicted in blue; the

FP 88 mcg BID group in yellow; and the FP 440 mcg

BID group in orange. These data show no

significant differences in lumbar spine bone

mineral density seen during the two years of FP

versus placebo treatment. These results are

reassuring and suggest that small changes in HPA

axis that we see at these doses are unlikely to be

associated with clinically significant systemic


Now, in these studies we also evaluated

other regions of the bones, in particular the

femoral region and the total body. However, these

regions were not prospectively QA and, hence, the

conclusions from these other results cannot be

made. However, even in those results there were

other confounders that affected the interpretation

of those data and overall the results are

consistent with the lumbar spine, that there is no

significant effect as seen with FP treatment at

these doses in regards to bone density.


Results of eye examples demonstrated no

evidence of posterior subcapsular cataracts or

diagnosis of glaucoma during the two-year treatment

with FP. The results of the other two-year study

with Rotadisk versus placebo were similar, and are

summarized in your briefing documents.


The results of bone mineral density from

comparative trials with FP provide further

reassurance that long-term systemic side effects

with FP treatment are unlikely. These results are

summarized on this slide and are provided in your

briefing document.

Additionally, in the three randomized,

double-blind trials which compared FP to BDP, there

was evidence in each trial that FP treatment was

significantly better compared to BDP therapy on

several bone mineral density measures. These

results suggest that not all inhaled

corticosteroids may have the same propensity to

affect bone mineral density.


In addition to data from trials conducted

in patients with asthma, safety results from the

three-year ISOLDE trial in patients with COPD were

also reassuring. This trial was conducted with the

CFC MDI using a spacer, which we know can further

enhance the systemic exposure to FP. This needs to

be considered when extrapolating the systemic

safety data from this trial to the FP Diskus where

lower exposure has been observed.

This slide summarizes the adverse events

and serious adverse events of fractures, cataracts

and ocular pressure disorders seen in this trial,

for the placebo group on the left and the FP group

on the right. In interpreting these data we have

to consider that the FP group remained in the study

for a longer duration and, hence, had more of a

chance for experiencing adverse events. Despite

the greater duration of treatment in the FP group,

there is no clear evidence that FP was associated

with increased risk of these events. These data

provide further evidence that long-term systemic

side effects in patients with COPD are unlikely at

these doses of FP.


The long-term safety data, hence, can be

summarized as follows: The range of systemic

exposure with FP Diskus in COPD is similar or less

than the CFC MDI in asthma where considerable

long-term safety data are available.

Results from two two-year

placebo-controlled FP studies were reassuring. No

clinically relevant bone mineral density or ocular

effects were noted in these studies. These results

indicate that small changes in HPA axis observed

with FP at these doses are unlikely to be

associated with clinical side effects.

Studies comparing FP to BDP provide

additional reassurance on the long-term safety of

FP administration. These studies further suggest

that not all inhaled corticosteroids may have the

same predisposition to affect bone mineral density.

Results from a three-year trial in patients with

COPD demonstrated no evidence of increased fracture

or ophthalmic adverse events with FP treatment.


The results from this clinical program,

hence, support the following dosage and

administration recommendations for Flovent and

Advair Diskus. For Flovent Diskus, the recommended

starting dose is 250 mcg twice daily. For Advair

Diskus the recommended starting dose is 250/50 mcg

twice daily.

While the responses in most measures were

similar between the two strengths, there were

suggestions of greater improvements in lung

function and dyspnea at the higher dose of FP in

FLTA35, and dyspnea with the higher dose of Advair

in SFCA3006. Due to these findings, we recommend

that patients who do not respond adequately to the

starting doses increase their dose to 500 mcg twice

daily for Flovent Diskus and to 550 mcg twice daily

for Advair Diskus which may provide additional



In summary, I have shared with you results

from our clinical program which fulfilled its

regulatory objectives. With Flovent we

demonstrated greater improvement on the primary

efficacy measure compared to placebo, with no new

safety issues noted. For Advair we demonstrated

superior efficacy compared to the individual

components for the primary efficacy measures.

These clinical benefits with Advair were not

associated with any evidence of a greater safety

risk. Results from our long-term safety data

provide further reassurance on the use of FP in


Thank you for your attention. I would

like to now reintroduce Dr. David Wheadon, who will

provide some concluding remarks.


DR. WHEADON: I will be brief.


The information we have presented this

morning provides compelling evidence for the

approval of both Flovent Diskus and Advair Diskus

for the treatment of COPD.


As you have seen and heard, COPD remains a

significant public health issue with increasing

morbidity and mortality in the U.S. population, in

contrast to many other diseases. Despite the

increasing burden of this disease, COPD remains

both under-diagnosed and under-treated. The only

treatment options currently approved in the U.S.

are bronchodilators. Despite optimal use of these

agents, many patients require additional therapy.


As was demonstrated by Dr. Malcolm

Johnson, inhaled corticosteroids, including

fluticasone, reduce inflammation associated with

COPD. Furthermore, the majority of clinical

studies with inhaled corticosteroids, including

fluticasone, illustrated clinically important

benefits in the treatment of this disorder. Due to

the complex pathophysiology associated with COPD,

the combination of salmeterol and FP allows

treatment of different aspects of the disease,

leading to greater clinical benefits than the

individual components alone. The use of inhaled

corticosteroids in COPD are further supported by

current clinical practice and the new

evidence-based international GOLD guidelines.


The objectives of the clinical program for

both Flovent and Advair Diskus were achieved. For

Flovent, demonstrably greater improvements in the

primary efficacy parameters were seen compared to

placebo. Results for the secondary efficacy

measures support the primary analyses. The

magnitude of improvements observed with Flovent are

similar to those seen with currently available

treatments which are regarded as clinically

meaningful in the management of COPD. No

clinically significant safety concerns were noted

with the use of Flovent in COPD.

For Advair, demonstrably greater

improvements in the primary efficacy parameters

were seen compared to the individual agents alone

at each strength. Results for the secondary

efficacy parameters support the primary analyses.

The magnitude of improvements seen with Advair

compared to placebo represent a clear advancement

in treatment of COPD. This greater improvement in

efficacy was not associated with an increased

safety risk compared to the individual agents or



I would like to conclude by once again

thanking the agency and the committee for allowing

us this opportunity to present the findings from

our pivotal clinical studies. COPD is a treatable

disease. Because there are limited approved

medications available for the management of this

chronic, debilitating condition, new therapeutic

options are very much needed. We believe that we

have presented compelling evidence that Flovent

Diskus and Advair Diskus are such new therapeutic

options. The benefits of either medication

outweigh the risk of treatment. The approval of

these treatments will allow physicians to make

informed decisions about the appropriate use of

these agents for the management of COPD in their



In closing, I would also like to introduce

two additional experts who have joined us here

today. Prof. Romain Pauwels is head of the

Department of Respiratory Diseases at the

University of Ghent, in Belgium. He is also the

Chairman of GOLD, the Global Initiative for Chronic

Obstructive Lung Disease, and has been involved in

many of the major clinical trials in COPD.

Dr. Jonathon Adachi is professor of

medicine at McMaster University, in Hamilton,

Ontario. He is a member of the scientific advisory

committee for the International Osteoporosis

Foundation, and is a past president of the

Osteoporosis Society of Canada. His major research

interest is steroid-induced osteoporosis.

We, along with our experts, will be happy

to address any points of clarification and

questions that you may have. Thank you.

DR. DYKEWICZ: Thank you. We will now

begin the segment where we have questions posed to

the sponsors. I would like to actually begin with

one question, perhaps more directed to Dr. Shah,

about secondary efficacy measurements. Of course,

one of the things that I think may be disappointing

about the data presented here is the relative lack

of effect on the secondary efficacy endpoints, with

the possible exception of dyspnea. This, of

course, contrasts with some of the data you

presented about studies with other inhaled

corticosteroids where some secondary efficacy

endpoints were found to be benefited by the trials.

You mentioned that the study design of the

trials that were conducted was such that you may

not have been able to capture improvements in

secondary efficacy endpoints. You mentioned, for

instance, that if you changed the population so

that subjects had a history of previous

exacerbations of COPD and then were studied you

might be able to capture some improvement in terms

of time to first onset of an exacerbation.

What other sorts of changes in protocol

could you conceive of that would help perhaps

better capture changes or impact on the secondary

efficacy endpoints? And, does Glaxo have any

intention of doing such studies?

DR. SHAH: Those are very valid questions.

I think, once again, as Dr. Donohue reviewed, we

have to lower our expectations when it comes to the

types of effects we can expect in COPD. We don't

have as much experience in conducting clinical

trials in COPD relative to the amount of experience

we have in asthma, where we do show very consistent

treatment effects on multiple endpoints, as you are

all aware of.

However, I think when you look at the data

that we presented, the secondary results we see

with our own program are actually very supportive

of our primary. We did see evidence of effects on

almost all measures that we looked at for the

secondary. Hence, you know, if you compare to how

the benefits that have been seen with currently

available treatments that are regarding as the gold

standard, the results we show in secondary efficacy

measures are actually quite robust.

In the context of the exacerbation

question, I think we are learning, in terms of

looking at exacerbations in COPD, that actually the

greatest benefit of inhaled corticosteroid therapy

is the reduction in repeat exacerbations. Indeed,

if you look at where we see the greatest benefits

in studies that have been done in Europe to date,

it is the rate of exacerbations, where you have

opportunities for multiple exacerbations, where we

are seeing the greater treatment effects. In our

study, as I indicated, we withdrew patients if they

required one burst of oral corticosteroids because

we didn't want that to confound our primary

efficacy measure analysis and our ability to

discriminate on the primary efficacy measure

because, obviously, concurrent oral corticosteroids

could have an effect on FEV1 which would

potentially limit our ability to show treatment

effects on that primary endpoint.

So, as I indicated, you know, our

secondary efficacy measures are actually some of

the best that we have seen in this disease, as has

been reviewed by Dr. Donohue. However, we accept

that the magnitude of changes we are seeing are

lower than our experience that we might be

accustomed to in asthma. But I think we also have

to understand that this is a different disease and

the expectations may need to be lower.

Maybe I can ask one of our experts, Prof.

Pauwels, to also comment from his own experience,

given his vast wealth of experience in doing

clinical trials in COPD.

DR. PAUWELS: Indeed, there is a

methodological issue that explains why you don't

see an effect on exacerbations. From other

studies, I can communicate the following

information and the following experience. That is,

first of all, it is probably better to select

beforehand the people who have repeated

exacerbations in order to demonstrate an effect on

exacerbations with any treatment intervention in


Secondly, what is absolutely needed is a

duration of at least six months with a large group

or, preferentially, observation of a one-year

treatment period.

Of course, the third issue in these

studies was that they selected as a secondary

outcome measure the time to first exacerbation but,

at the same time, actually withdrew their subjects

once they had an exacerbation and the majority of

the effects that have been seen, and are repeatedly

seen with inhaled corticosteroids is on the

exacerbation rate, which is the number of

exacerbations over a fixed time period. So, the

design that was used in these studies didn't allow

for study of that.

DR. DYKEWICZ: Thank you. Questions from

other members of the committee? Dr. Bone?

DR. BONE: Thank you. I have a couple of

questions. I suppose Dr. Donohue would be the one

to answer them and I would, obviously, be very

interested in the comments of the committee members

or from the experts on pulmonary disease. I am

just an interloper here for special reasons.

The first question I wanted to ask is what

is the life expectancy of the patients that would

be candidates for treatment here with moderate to

severe disease?

DR. DONOHUE: Of course, that is a very

difficult question. We have some data that the

life expectancy of a patient with COPD -- again, it

depends on what part of the country you are in, but

with 39 percent FEV1, 50 percent of those people

live five years. That is in the literature. So,

the exposure would be, you know, perhaps less.

These folks are older. We saw the mortality,

110,000 Americans die each year with COPD; 668,000

hospitalizations. So, we are talking probably

about a lot shorter exposure to medications than we

would be, of course, in the asthma population.

DR. BONE: Thank you. That was kind of

what I was wondering about generally. The second

question has to do with the exacerbation issue. I

guess I have a different reason for asking about

it, but is there a dominant cause for

exacerbations, such as infection, or do

exacerbations just sort of occur spontaneously?

DR. DONOHUE: Yes, that is also a very

important question. Exacerbations are much more

frequent in the more severe group. So, when we

stratify disease, like in level three of GOLD, we

might see as many as two to three exacerbations.

At level one it is about 0.8. So, there are issues

there. In general we estimate that one-third of

exacerbations are due to bacterial infections, H.

influenzae, strep., pneumonia and Moraxella

catarrhalis being the leading cause. One-third

would be environmental irritants, air pollution or

what-have-you, and one-third is everything else.

So some are, indeed, seemingly just a loss of

control, gradually going down hill. It is very,

very hard to identify an offending agent. We can't

tell clinically. That is why we just really

empirically treat the exacerbations, usually with a

short burst of antibiotics and perhaps oral


DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: Thank you. My question too

regards the COPD exacerbation issue and perhaps Dr.

Shah could respond. Recognizing the confounding

effect of exacerbations on the primary outcome

measure, my question is a descriptive one. In

fact, looking at the briefing document and the

frequency of dropouts for exacerbations, it is

actually rather low across all groups. That is the

first observation. I would appreciate a comment

about that.

The second is I can't tell how many

patients actually dropped out for an exacerbation

during the run-in period. Third, at baseline, what

is the baseline frequency historically of

exacerbations in the study cohort? In other words,

there are data about the baseline frequency and I

don't see that represented here.

DR. SHAH: Let me see if I can remember

all those questions. In terms of the last question

about what was the baseline level of exacerbations,

we actually didn't collect the historical

information in the study which, in retrospect, I

think we would do differently in the future. So,

we don't really know the level of exacerbations

these patients were experiencing prior to enrolling

into the studies.

We actually did have people withdraw for

exacerbations during the run-in period. As you

know, we did withdraw inhaled corticosteroids. In

order to get into the study they had to stop them

at the screening visit if they were, you know, on

lower doses. It varied across the three trials and

it was anywhere from about 15 percent, 30 percent

of patients who withdrew for exacerbations during

the run-in period in the clinical trials.

I can't remember the last question. Is

there a further question on that?

DR. STOLLER: The baseline frequency, the

dropout rate for exacerbations.

DR. SHAH: Right. I mean, the other thing

on retrospect is that I think we may have affected

the overall withdrawal rate for exacerbation

differences between these and other studies that

have been done. We actually had criteria for lack

of efficacy, which was up to physician discretion.

We didn't really have good control on what the

physicians may have considered was appropriate for

lack of efficacy. However, it was, unfortunately,

a remnant from our asthma studies and it was there,

and we did have physicians withdraw patients for

those reasons and there was an imbalance across the

treatment groups, with more people in the placebo

group being withdrawn for that compared to

certainly the high dose FP and Advair groups.

We actually prespecified as part of this

program a term that we called COPD-related

conditions. So, withdrawals due to COPD-related

conditions, which included patients who would have

withdrawn for exacerbations; people who withdrew

for lack of efficacy; people who withdrew for these

other types of reasons. When we actually looked at

the integrated data -- you know, in individual

studies we just didn't have enough numbers of

patients to look at exacerbations with any kind of

statistical comparisons. And, these are all,

obviously, post hoc; they are exploratory types of

analyses. But we clearly showed when you look at

the COPD-related conditions withdrawals, which

includes exacerbations and lack of efficacy and

adverse events that were related to respiratory

conditions, clear evidence of treatment effect with

almost all groups, except for the FP 250, achieving

p values less than 0.05.

So, there were clear suggestions of

effects on the respiratory-related withdrawals in

this program but again, as I indicated, we didn't

optimize these studies to look at that. Hence, I

think the results reflect that.

DR. STOLLER: May I just ask a follow-up

just to clarify? I understand that you didn't

query these individuals about the baseline

frequency of exacerbations. Was there any

retrospective attempt to ask individuals as the

study was progressing if they could recall,

recognizing the limitations of recall bias in those


DR. SHAH: Actually no, because, to be

frank, we obviously expected to see some effects on

all endpoints. So, we didn't have any reason to

believe that there was a need to do that. Clearly,

on retrospect, I think we would do it differently.

DR. STOLLER: Then just in response to

your comment, I understood your comment that the

prevalence of dropouts for exacerbations during the

run-in period was 15 percent to 20 percent.

DR. SHAH: Fifteen to 30 across the three


DR. STOLLER: Was that mal-distributed

among the placebo versus the individual drug


DR. SHAH: Well, remember, these were at

screening so they were not on any treatment. These

were basically people coming into the screening


DR. STOLLER: But I presume randomization

was also obtained prior to that --

DR. SHAH: No, the randomization would

have been after screening.

DR. STOLLER: Oh, after it? Fine.


DR. FINK: Yes, this is probably for Dr.

Shah. Since you were dealing with a group of

patients with severely impaired pulmonary function

at baseline, what would the average FEV1 percent

change be if expressed as a change in percent

predicted for the patient rather than a change from


DR. SHAH: In terms of a treatment


DR. FINK: Yes, the treatment response.

DR. SHAH: In terms of our program, the FP

response, as I said, we averaged in two of those

studies about 100-plus milliliter difference as

change in FEV1 over baseline. The percent

predicted was about 40 percent at baseline. So,

that degree of change I think would correspond to

about a five percent predicted improvement in FEV1,

and with Advair it represents about eight to ten

percent predicted improvement over baseline.

DR. FINK: And if you, in your analysis,

stratified the change in FEV1 for those patients

whose baseline was above 50 percent was the data

driven predominantly by those patients who had very

low values at baseline, where a two to three

percent change in FEV1 would be a ten percent


DR. SHAH: No. Actually, we did look at

the results by different severity in terms of FEV1

percent predicted at baseline, and what we did see

was that across that spectrum of severity, like

from 30 to 50 percent because -- recall, the

cut-off we had was 65 percent or less in the study

-- what we see across all those subtypes of

patients is that the response is fairly constant in

each of those groups as a magnitude. Obviously, as

a percent that would be different because if the

magnitude is similar you would expect percent to be


DR. DYKEWICZ: Dr. Atkinson?

DR. ATKINSON: When you were noting the

adverse events -- you may have said this, but there

were four deaths in the placebo group. I know it

wasn't an outcome measure but was there any effort

to ascertain the cause of death? Was it

respiratory in any of the cases?

DR. SHAH: No, they were all related to

other causes. Three of them were, I believe,

malignancy and one, I believe, was a cardiac event.

DR. DYKEWICZ: Dr. Apter?

DR. APTER: Dr. Shah, I have three

questions. One is just to clarify. You said that

converting the 100 ml change in FEV1 corresponds to

about a five percent change in predicted?

DR. SHAH: Yes, in that range. Again, I

am estimating this and my math skills have clearly

gone down-hill since the advent of the computer but

I think that is roughly right. We can get those

data for you if that would be useful.

DR. APTER: It would. Secondly, you gave

the mean age of the patients as about 64. Do you

have any information about the range of standard

deviation frequencies? In other words, with

respect to Dr. Bone's comment, how many were

actually very elderly?

DR. SHAH: Very elderly? Well, as I said,

the mean age was about 65 and so about half the

patients were over that age compared to, you know,

younger. I would say that about 25, 30 percent

were about 75, in that order. Again, I don't have

those data at the top of my head but we did have a

fairly large proportion of patients who were over

65 in this program, and we did look at both the

efficacy and safety in that population and the

results were comparable to the overall results.

DR. APTER: Actually, a question for Dr.

Wise, are asthma patients who are minority affected

more adversely? Do we have information on that in

COPD? Because this population was virtually all

white. A second question is, of course, that more

and more women are smoking and this population has

a minority of women. Do you have anything to say

about the choice of the population to study?

DR. WISE: The issue of COPD in

African-Americans is one that is of interest.

There seems to be a protective effect, if anything,

at least in terms of the diagnosis of

COPD among African-Americans. This has sometimes

been called the COPD paradox because

African-Americans who smoke have higher rates of

lung cancer than Caucasians but lower rates of

COPD. Emphysema deaths, at least from death

certificate data -- only eight percent of those

deaths are African-American. So, there seems to be

a paucity of COPD in African-Americans, and this

has been borne out in some of the large

NIH-sponsored clinical trials, for example, Lung

Health Study I, and I believe it was around six

percent African-Americans. In Baltimore, where we

recruited in the inner city with a majority of

African-Americans, screening for early COPD, we had

approximately 12 percent African-Americans; similar

numbers in Detroit. So, there does seem to be a

protective effect, if anything, in contrast to

asthma, as you pointed out, where there is an

increased risk.

Gender -- I think Dr. Donohue implied that

each year, each study that comes out with COPD

shows increasing prevalence of COPD among women,

and there is an increasing, although not

incontrovertible, body of evidence suggesting that

women may be more susceptible to the adverse events

of cigarette smoke in terms of developing

obstructive lung disease. The prevalence of COPD

in the general population, men versus women, is

probably around 60-40 or 65-35, in that order.


DR. JOAD: Yes, my question is for Dr.

Shah. I want you to address the statistically

significant differences that you got in your

questionnaires, the symptom and quality of life

questionnaires, versus the lack of clinically

significant differences for all of them except the

Advair 550. Do you have any arguments with those

questionnaires that you chose, or any arguments

with what was said as a clinically difference that

you were looking for?

DR. SHAH: Yes, I think if we look at the

two questionnaires -- well, we had three, we had

the one for dyspnea, the TDI, and I will say that

we have more experience in general in COPD with

that instrument because there are numerous clinical

trials that have looked at that instrument in terms

of evaluating pharmacological agents. As we showed

you, the results on that instrument were actually

some of the most robust effects we have seen up to

date with any currently available drug that has

evaluated that instrument with treatment with both

FP as well as Advair.

As you know, there are two quality of life

health questionnaires that are currently used in

assessing COPD. There is the SGRQ, the St.

George's Respiratory Questionnaire which was

reviewed by Dr. Johnson. That was a study that was

done in Europe. At the time we designed this

program it had not been validated in the U.S.,

hence, was not available for us to use in the U.S.

at that time. Clearly, now we do use that

questionnaire in assessing treatment response.

We had CRDQ with which we did have some

experience. As you saw, we had used that

questionnaire in a previous clinical program with

Serevent MDI inhalation aerosol, and in that study

we saw small trends for effects with the MDI, which

is currently approved, but we did not see any

evidence of a statistically significant difference.

The numbers of patients were very similar in those

studies to what we have in our own studies with FP.

Whereas, in the FP studies we actually showed

evidence of statistical significance or p values

less than 0.05 on those measures.

What we didn't achieve is the threshold of

what the instrument has defined to be clinically

significant. But we have to remember that those

thresholds were defined as a change from baseline,

not as a between treatment group difference. We

have no knowledge currently of what degree of

change between two treatment groups would

constitute a clinically significant difference. I

think, you know, the developers of those

instruments would confirm that view. However, we

had no other basis to make some determinations of

what would be a clinically significant difference

between these treatment groups. So, we used the

same value that has been defined as a change from

baseline as a value or threshold for showing

differences between treatment groups.

I think we have to realize that those

thresholds for between treatment groups may be

unattainable. You know, we don't have a lot of

experience and certainly we have not seen any

evidence up to now, at least in the published

literature and in our own experience, where those

degrees of differences have been achieved between

treatment comparisons in studies in COPD. So, we

did see statistical differences favoring FP on that


Again, with Advair the mean change from

baseline on the CRDQ at both doses achieved a value

over 10, which the developer of the instrument

would regard to be a clinically important change

for these patients. That is how clinical

significance was defined in the development of this


We have to be careful because the

clinically significant difference values for these

instruments were not defined to be assessed in

comparing treatment responses between two

treatments. They were defined as individual

patient and group responses, how did they change

from baseline and is that change from baseline

clinically meaningful for those patients.

So, you know, without having much other

evidence, we used the same threshold but I think we

have to be careful because that may be a threshold

that is unattainable in COPD based on the current

evidence that we have.

DR. DYKEWICZ: A last question from Ms.


MS. SCHELL: I am not quite sure who to

direct this question to but I was wondering about

airway remodeling and the effects of the inhaled

corticosteroids on that. There has been some

research that reverses airway remodeling and I just

wondered if anyone has looked at that, or if you

have any studies on COPD effects.

DR. SHAH: Yes, I think I would have to

defer to maybe Dr. Johnson to speak to that and

maybe Dr. Pauwels could add some comments as well.

DR. JOHNSON: That is a very good

question, and it reflects I think the issue that

very few studies have been carried out in COPD

comparing to asthma, and we are only now beginning

to understand what we even mean by the term

remodeling in asthma. When you compare that to

remodeling in COPD it is a very much more difficult


These studies take a long time to carry

out, and we are, again, missing what we could take

as acceptable surrogate endpoints for remodeling.

Clearly, it is difficult to do biopsy studies over

10, 15 years in any patient population.

I think what we are seeing though is, as I

have presented, particularly in terms of the

evidence at the biopsy level, is at least a

potential for inhaled steroids to influence some of

the inflammatory cells and some of the processes

that we know could contribute to a remodeling

process. So, the potential is there; the proof is

not as yet. And, it represents, I think, a key

area for further study of any intervention of this

disease. You know, has it the capacity to slow

down the remodeling process or even, probably more

importantly, to try and reverse the remodeling

process? I think you raise a very important

question, I just wish we had some better data.

That also applies to asthma.

DR. DYKEWICZ: One last question by Dr.


DR. MALOZOWSKI: I have a series of

questions. Can I ask them or just one?

DR. DYKEWICZ: Let's try to play it by


DR. MALOZOWSKI: I will try to focus on

your slide A106, please. This is the slide that

shows the mean percent change in lumbar spine bone

mineral density and I would like to make some

comments. I cannot really comment about COPD but I

can comment on issues related to safety and how to

look at safety information.

In this particular slide, if you don't pay

attention to the first three columns at week 24,

and also if you don't pay attention to the column

that is in yellow that represents a dose that

doesn't have anything to do with this particular

study, what you see is that patients on placebo

tended to have positive bone mineral density, while

patients that received this medication -- at least

the estimates, the mean and the standard deviation

are going negative. Okay?

When you look at safety information you

cannot look at data in this manner because this

really hides what is going on in the patient

population. If this were a drug that was developed

to treat osteoporosis or a condition such as this,

this depiction of the data would be adequate, but

here we are looking at signals for adverse events

and what we need to look at is outliers. And, we

don't have this because the only information we

have is depicted as mean and standard deviation.

But the directions in which these columns are going

suggest that probably there are some patients that

are losing bone mineral density to a larger extent

in those receiving fluticasone or the combination

therapy versus those patients on placebo.

DR. SHAH: Can I comment on that?


DR. SHAH: First I will walk you through

the response we see. In any clinical study you

have variability around the mean. That is the

nature of clinical research. So, you look at the

response with the high dose FP, which is in orange.

Let's just follow what happens to these patients

over the course of the two years of treatment.

You see essentially at six months not much

of a difference. Now you go to one year, a slight

suggestion of a decrease but, again, given the

state of deviation, not much of an effect. Now

look at what happens at a year and a half. It has

now flipped the other way. You have now what

clearly suggests a treatment benefit. If I believe

what you were saying, if a low value means a

significant negative effect, obviously a plus value

would mean there is a positive effect. Now it

flips back down at the two-year time point.

I will also say that prospectively in

these studies we defined a value of a five percent

change from baseline to be clinically significant

over the two years, and we have two patients in the

placebo group and two to three patients in the FP

500 group who achieved that degree of change over

the two years of treatment.

So, I appreciate the comment that we are

looking for signals here, but I think we have to be

very careful that we look at the data relatively

objectively and make conclusions based on data and

not perception.

DR. MALOZOWSKI: I am saying that you

cannot look at safety data as mean and standard

deviation. This is my point. And, this particular

slide does not show what really happened with

patients. You don't show the outliers. You just

take that unifying as five percent.

Also, you define, for example,

hyperglycemia as 175. The American Diabetes

Association defines diabetes as equal to or above

126. Therefore, you know, this particular

definition -- I don't know what weight it has, but

the point I am trying to make is that it is very

difficult to capture outliers that somehow depict

what really happens in a clinical study when you

try to compress everything to the mean and the

standard deviation.

DR. SHAH: Sure and, as I say, we did look

for outliers as part of this program and we didn't

see a difference. We had about the same proportion

of patients who achieved a prespecified change of

five percent in bone marrow density in both groups.


DR. SHAH: We actually picked 175 because

that is what we have traditionally been doing which

previously, by the FDA, has been acceptable. We

reanalyzed the data and, actually, we can show you

the data for the glucose by using a 120 value and,

again, I think the data will speak for themselves.


At these doses of FP, here is the result

on mean glucose in our program which we just

reviewed. This is for the integrated data for the

treatment groups. Here are the screening values

for the mean results. Here are the values at week

12. This is using a cut-off now of 120 in change

in glucose. What you see is that at screening we

had a reasonable proportion of patients, slightly

higher in this group on Advair who achieved greater

than 120 or had a 120 change in glucose compared to

the other groups, but fairly similar. At week 12

you see that clearly there isn't a suggestion of a

signal here in terms of the effects on glucose as

the proportion of patients who had a value over 120

change. Here is week 24. Again, there certainly

doesn't appear to be any significant signal

suggestive of a treatment effect on these measures.

So, clearly, we have looked at these data.

We are more than willing to look at the data any

way that, you know, you would feel would be best to

make an assessment but the results are very

reassuring. We don't see concerns on glucose with

these doses of FP, which is what we have

traditionally shown in our experience in asthma.

DR. DYKEWICZ: We will now take a break.

In view of the time, let's resume at 10:45.

[Brief recess]

DR. DYKEWICZ: Please take your seats. We

will next begin the segment of our presentations by

the FDA. The first presentation will be by Dr.

Charles Lee.

FDA Presentations

Flovent Diskus for COPD

DR. LEE: I am Charles Lee, medical

reviewer for the Division of Allergy and Pulmonary

Drug Products for the FDA.


This morning we will be discussing the new

drug application for Flovent Diskus, NDA 20-833 for

a COPD indication.


The proposed labeling states that Flovent

Diskus is indicated for the long-term, twice daily

maintenance treatment of COPD, including emphysema,

and chronic bronchitis.


This is the dosage and administration

section of the proposed COPD label. The starting

dosage for adults is one inhalation, 250 mcg, twice

daily. For patients who do not respond adequately

to the starting dose, increasing the dose to 500

mcg twice daily may provide additional control.


In this presentation I will cover a review

of efficacy and a review of safety for this

application. The sponsor submitted three pivotal

studies in support of efficacy. Safety is

supported by data from these three pivotal studies

and from additional supportive studies.


In our view, there were key efficacy and

safety issues in this application. The treatment

effects noted for the primary efficacy endpoint

were small and did not appear to be dose related

across studies. There were small differences from

placebo for secondary endpoints in patient-reported

outcomes. The majority of COPD patients in the

pivotal studies had reversibility with


In addition, there were concerns raised in

the pivotal studies and in other studies submitted

in support of safety. These included respiratory

infections, upper and lower, and systemic effects

such as adrenal effects and effects on bone.


We would like to have you consider these

data in light of the following questions: How

clinically relevant is the change observed in the

primary efficacy endpoint where there is a small

amount of support from secondary endpoints and

patient-reported outcomes?

How typical is the COPD population studied

in these trials? And, how would this impact the

ability to generalize the results from these

studies to all COPD patients, specifically with

regards to reversibility with bronchodilator, in

the presence of chronic bronchitis and emphysema?

How sufficient is the safety database to

support the use of the product for long-term

maintenance treatment of COPD? Finally, is the

risk-benefit profile suitable for approval of this

product for this indication?


As you have heard, there were three

pivotal studies. Fluticasone 500 twice a day was

studied in this study and in this study.

Fluticasone 250 twice a day was studied in this

study and in this study. Increasing the dose in

patients who do not respond to fluticasone 250, as

proposed in the labeling, was not studied.

These studies, as you have heard, had

similar design; were randomized, double-blind,

placebo-controlled, parallel group studies of 24

weeks in duration in patients with an established

history of COPD. All patients had chronic

bronchitis and patients could have self- or

physician-reported emphysema.


The primary efficacy variable for

fluticasone was the pre-dose FEV1. The primary

efficacy endpoint was change from baseline in FEV1

at study endpoint. Secondary efficacy variables

included measures of symptoms of chronic

bronchitis; measures of symptoms of dyspnea; peak

flows; measures of albuterol use; and COPD

exacerbations. Patient-reported outcomes or,

so-called, health related quality of life, was

measured by the Chronic Respiratory Disease

Questionnaire, an instrument developed by Guyatt.


Safety variable included adverse events,

serious adverse events, withdrawals, vital signs,

physical exam and oropharyngeal exam. Physical

exams and oropharyngeal exams were performed and

any abnormalities were recorded as adverse events.

ECGs, hematology and chemistry studies were also

performed. Serum cortisols were measured in one

study and standard dose cosyntropin stimulation

testing with 250 mcg of cosyntropin was performed

per package insert in the other two pivotal

studies. Bone mineral density or ophthalmologic

examinations were not performed. These studies

were performed over a six-month period and were not

likely to be have been sufficiently long to detect

any bone or ocular effects.


We will look at demographics and baseline

characteristics of the population of the pivotal

studies next. Demographics and baseline

characteristics were similar in the three pivotal

studies. Approximately 65 percent of patients were

of male gender, with a mean age of approximately 63

years. Ninety-four percent of patients were of

Caucasian race. Non-Caucasian races were not well

represented in the pivotal studies, with about five

percent of patients being Black race and about two

percent of "other" race. Approximately 25-30

percent of patients were using inhaled

corticosteroids at the time of screening, and about

47 percent of patients were smokers at the time of



Randomization was stratified based on

patient response to bronchodilator. Reversible

patients were those who had a 12 or more percent

increase in FEV1 with bronchodilator or an absolute

change in FEV1 of 200 ml or greater with

bronchodilator. There was a high percentage of

patients, a majority, who were reversible, ranging

from between 54 percent and 59 percent. The ATS

standards for the diagnosis and treatment of

patients with COPD mention that up to 30 percent

may have an increase of 15 percent or more in FEV1

with use of beta-agonists.


Here we are looking at the mean response

to bronchodilator or the degree of reversibility of

the population -- the reversible group, the

non-reversible group and overall. They were

similar among the studies. The amount of

reversibility in the reversible group ranged from

30 percent to 32 percent increase in FEV1 with

bronchodilator. The non-reversible group had a 9

percent increase in FEV1 with bronchodilator.

Overall the degree of reversibility is shown here,

with between a 20 and 23 percent increase in FEV1

with bronchodilator.


Next we will look at measures of efficacy

in the pivotal studies.


Here we are looking at the primary

efficacy endpoint, mean change from baseline in the

pre-dose FEV1. Values are in liters. Differences

from placebo are displayed. Baseline values are

displayed in parentheses, and statistically

significant values have an asterisk.

For fluticasone 250 an increase in FEV1

was noted of 27 ml in this study and 108 ml in this

study. For fluticasone 500 there was a 50 percent

increase in FEV1 at study endpoint and 113 ml

increase in this study. Statistical significance

was replicated for fluticasone 500 only.

Statistical significance was not replicated for

fluticasone 250. Across studies there was not

strong evidence of a dose-related effect.


In general relatively small differences

from the placebo group were noted for secondary

endpoints in patient-reported outcomes. We will

cover COPD exacerbations, total daily albuterol use

and CRDQ, the Chronic Respiratory Disease

Questionnaire, the patient-reported outcome

instrument. Total daily albuterol use and CRDQ are

representative of the small changes noted for the

other secondary endpoints. COPD exacerbations

showed no effect.


Here we are looking at percentage of

patients with one or more exacerbations of COPD in

each of the three studies. There were fewer COPD

exacerbations in fluticasone-treated patients in

this study, with an appearance of a dose-related

effect. However, in the other two studies the

numbers went the other way. In the fluticasone

group a higher percentage of patients had COPD

exacerbations. These are actually the studies that

had the largest change in the primary efficacy

endpoint. Overall, there seems to be no evidence

of a treatment effect of fluticasone on COPD

exacerbations. There were similar results in the

percentage of patients who had moderate to severe

COPD exacerbations.


Here we are looking at the change from

baseline in daily albuterol use. Again, mean

change from baseline and the difference from plasma

is shown. Decreases in daily albuterol use range

from a decrease of 0.3 puffs per day to 0.9 puffs

per day over a baseline of about five puffs per



Here we are looking at the Chronic

Respiratory Disease Questionnaire, the health

reported outcomes instrument. We are looking at

change from baseline in the overall score. Again,

the difference from placebo is displayed. The

minimum clinically important change is 10. The

baseline ranges from 84 to 89. The amount of

change attributed to active drug ranged from a

negative 0.2 in one study to a high of 8.1 in this

study. There was very little difference from

placebo in the study that had the largest effect

size in the primary efficacy endpoint. In summary,

improvements were noted for active treatment but

the differences between the active treatment group

and the placebo group were small, and less than the

minimal clinically important change.


A subgroup analysis of efficacy was

provided for the non-reversible group. As we

mentioned, the non-reversible group were those

patients with an increase in FEV1 with

bronchodilator of less than 12 percent or those

patients who had less than 0.2 liter increase in

absolute volume with bronchodilator.


Here we are looking at subgroup analysis

for the non-reversible group of the primary

efficacy endpoint, mean change from baseline in

pre-dose FEV1. Again, the difference from placebo

is displayed. Baseline measurements are in

parentheses. Improvement or increase in FEV1 at

baseline ranged from 2 ml to 101 ml in the

non-reversible group. Overall there was a smaller

amount of change noted for the non-reversible group

than was seen in the overall group.


In summary of efficacy, we see an effect

on the primary efficacy endpoint, change from

baseline in FEV1, that is statistically

significant, and replicated for fluticasone 500 but

is not replicated for fluticasone 250. A small

effect was noted in the non-reversible group.

Secondary endpoints and patient-reported outcomes

showed small differences from the placebo group.


Next we will be looking at the safety data

from the pivotal studies.


In general, the types of adverse events

reported in the pivotal studies were similar to

those noted in current labeling for Flovent

products. These will be integrated data and

display adverse events that occurred more commonly

in fluticasone-treated patients than in

placebo-treated patients.

There was a higher percentage of patients

treated with fluticasone who had adverse events and

this appeared to be dose related. There was also a

higher percentage of patients treated with

fluticasone who had upper respiratory infections

and viral respiratory infections.


The rate of candidiasis was high, 13

percent in the fluticasone 500 group, 7.3 percent

in the fluticasone 250 group. Obviously, it would

appear to be a dose-related effect. There was a

higher percentage of fluticasone-treated patients

who reported dysphonia and a slightly higher

percentage of fluticasone-treated patients who had



Adrenal effects were measured in the

pivotal clinical studies. Serum cortisols were

measured at the end of week one in this study, and

cosyntropin stimulation testing, per package

insert, was performed in the other two pivotal



Here we are looking at serum cortisol data

after treatment compared with the placebo group.

AUC 12 represents an integrated measure of serum

cortisol or serum sampling over a 12-hour period.

There appears to be a dose-related suppression of

serum cortisol. For the fluticasone 250 group

values were 10 percent less than those of the

placebo group. For the fluticasone 500 group

values were 21 percent less than those of the

placebo group. A similar pattern was seen with

lowest cortisol concentration, or the Cmin. The

value in the fluticasone 250 group was 5.2 percent

lower than the placebo group, and in the 500 group

30.7 percent lower than the placebo group. In

summary, we see a treatment effect that appears to

be dose related.


Cosyntropin stimulation testing was

performed in the other two pivotal studies. There

was no evidence of adrenal insufficiency that was

observed. However, this test may not be

sufficiently sensitive to conclude that there were

no adrenal effects at all.


Next I will present pertinent safety data,

submitted in support of this application, from

other studies.


This is a Phase I pharmacokinetics and

pharmacodynamic study. It was a single center,

open-label, randomized, four-way crossover design

in which 1000 mcg of fluticasone was administered

with different dosage strengths of the Diskus

device. Dose proportionality of the different

dosage strength devices was the objective of this

study. There was a five-day washout period between

the study periods.


Here we are looking at mean 24-hour

urinary cortisol excretion. Values are the

pre-dose measurements, post-dose measurements and

percent change from the pre-dose measurement. Each

of these groups received 1000 mcg of fluticasone as

a single dose, with each of the different dosage

strength devices noted there. The post-dose

24-hour urinary cortisol excretion ranged from 35

percent less than the pre-dose to 59 percent of the

pre-dose measurement after these single doses of

1000 mcg of fluticasone.

It should be noted that this study was

performed in normal volunteers, and inhaled

fluticasone appears to have a lower degree of

bioavailability in patients with COPD. Despite

this, the results do show clear evidence of adrenal

effects. We should note that the higher of the two

proposed daily doses for fluticasone, 500 mcg twice

a day, is the same total daily dose that is

administered in this study.


This is a multicenter, double-blind,

randomized, placebo-controlled study of Flovent

metered dose inhaler, 500 mcg twice a day for three

years in patients with COPD, also known as the

ISOLDE study. It was published in the British

Medical Journal in 2000.


Here we are looking at notable adverse

events that occurred during the inhaled treatment

phase of the study. Nearly all of the patients

reported adverse events in both treatment groups.

There was a higher percentage of

fluticasone-treated patients who reported

respiratory adverse events as seen in the pivotal

studies. These included lower respiratory

infection, upper respiratory infection, viral

respiratory infection, as well as pneumonia. There

was also a higher rate of serious adverse events

due to pneumonia. The numbers aren't displayed

here but they are five percent versus two percent,

and that data was not included in the paper.


Adverse events that could be contributed

to systemic activity of inhaled fluticasone are

displayed in this slide. There was a higher

percentage of fluticasone-treated patients who had

decreased cortisols compared with placebo. These

are patients who had laboratory abnormalities that

were considered to be clinically significant and,

therefore, were reported as adverse events.

The adrenal effects were reported, in a

somewhat different fashion than the paper, as mean

serum cortisol levels listed and presented in the

table. There was a higher percentage of

fluticasone-treated patients who were reported as

having diabetes. There was also one

fluticasone-treated patient who was reported as

having Cushing's syndrome, and one

fluticasone-treated patient who was reported as

having adrenal hypofunction. Although it is

unclear whether this is an adverse event due to

adrenal hypofunction or a laboratory abnormality,

the paper does state that no decreases in cortisol

were associated with symptoms of hypoadrenalism.

It should also be noted that the MDI formulation

which was used in this study is more bioavailable

than the Diskus formulation, but one notes similar

concerning patterns.


This was a multicenter, randomized,

double-blind, placebo-controlled, parallel group

study of 1000 mcg twice a day of fluticasone by the

metered-dose inhaler for a four-week period in

patients who had an acute COPD exacerbation. In

this group of 126 fluticasone-treated patients

there was one fluticasone-treated patient who was

reported as having a serious adverse event due to a

decreased cortisol level, although I have no other

details about that patient. The study used the MDI

formulation and the dose in the study is twice the

proposed dose in this application.


Data on bone mineral density was submitted

in support of this application. Data was from two

two-year studies of asthma patients. Patients

ranged from 18-50 years, and females were

premenopausal. The study population may be at

lower risk for osteoporosis than the population

proposed in this NDA.


In the first study a slight numerical

decrease in bone mineral density for the

fluticasone 440 mcg dose was noted at the lumbar

spine, but an increase in bone mineral density for

the 88 mcg dose and placebo make it difficult to

interpret these data. The sponsor reported no

changes for the proximal femur or total body. In

the next study, decreased bone mineral density was

noted at the femoral neck, although this data was

retrospectively QA'd.

As noted previously, these studies were

performed in younger asthma patients, a study

population who may be at lower risk for

osteoporosis than those proposed in this

application. These studies raise some concerns,

and I would like to point out that bone mineral

density has not been studied in the COPD population

with the proposed drug product.


In conclusion, a statistically significant

treatment effect for the primary efficacy endpoint

was replicated only for fluticasone 500; was not

replicated for fluticasone 250. There were small

differences from placebo for the secondary

endpoints and patient-reported outcomes. These

findings were in a study population in which a

majority of the patients were reversibly and may

not be representative of the COPD population at

large. Non-Caucasian patients were also


Safety concerns noted in the pivotal

studies and supporting studies included respiratory

infections, upper, lower and pneumonia; adrenal

effects; and bone density, which has not been

studied in the COPD population for this product.

We question if the degree of benefit justifies the

potential risk in light of these safety concerns,

particularly with long-term use in an older, more

fragile population.

I will conclude my presentation and then

Dr. McClain will be presenting, and I think we are

going to entertain questions after the three of us

all present. Thank you.

Advair Diskus for COPD

DR. GILBERT-MCCLAIN: Good morning.


I am Lydia Gilbert-McClain, a medical

reviewer for the Advair Diskus product. My

objective during this talk is to present to you the

Pulmonary Division's perspective on the safety and

the efficacy of the Advair Diskus product as it

relates to the indication for COPD. During this

presentation I will bring out all the issues that

raised some concern within the Division as they

relate to the applicability of the Advair Diskus

product for COPD indication.


One of our concerns is with respect to the

clinical relevance of the efficacy data; secondly,

the applicability of the data from these trials to

the general COPD population. In other words, if

these products are deemed to be approvable, should

this approval be generalized to the COPD

population, or should it be for a subpopulation of

COPD populations? Thirdly, are the safety data

adequate to support approval?


As you are aware, Advair is a combination

drug product of fluticasone propionate and

salmeterol. Advair Diskus was approved in August

of 2000 for long-term maintenance treatment of

asthma. Salmeterol, as an inhalation aerosol, was

approved in 1998 for the relief of bronchospasm

associated with COPD. Fluticasone propionate has

not been approved for use in COPD in the United

States. Therefore, with respect to Advair for a

COPD indication the critical issue is the addition

of fluticasone propionate, an inhaled



The sponsor has already gone through their

develop program and I will not do that in this

talk. But just to set the background for my

presentation, I would just like to highlight the

two trials that I will be discussion, the SFCA3006

and 3007, which evaluated the two strengths of

Advair, Advair 500/50 and Advair 250/50. During my

talk I will be referring to these products simply

as Advair 500 and Advair 250. The corresponding

treatment arms are shown here, fluticasone 500 and

250 and the salmeterol and placebo arms. The

dosing administration was one inhalation twice



The sponsor-stated objectives were to

evaluate the efficacy and safety of these Advair

products compared to the individual components,

fluticasone and salmeterol, and placebo in COPD

patients treated over 24 weeks. Additionally, the

sponsor's third objective was actually to compare,

to use the sponsor's own words, the quality of life

in COPD subjects using these Advair products

compared to subjects using the individual

components, fluticasone and salmeterol and placebo,

over 24 weeks of treatment. More recently, the

agency has been using the term patient-reported

outcomes instead of quality of life. During my

talk I will also use the term patient-reported



Given that Advair is a fixed combination

drug, the studies were designed to make the fixed

combination drug's policy. This policy, stipulated

in the Code of Federal Regulations 21 CFR 300.50,

states that two or more drugs may be combined in a

single dosage form when each component makes a

contribution to the claimed effects of the

combination, and the combination is safe and

effective for the patient population requiring such

therapy. In this regard, the two Advair trials,

3006 and 3007, were adequately designed to fulfill

the efficacy requirements of the combination drug



Just to highlight some key entry criteria,

all subjects had to fulfill all the inclusion

criteria to be eligible for these studies. They

had to have a diagnosis of COPD as defined by ATS.

They must have a history of cough, productive of

sputum on most days, for at least three months of

the year for at least two years, that was not

attributable to another disease process; baseline

FEV1 of less than 65 percent and FEV1/FVC ratio of

less than 70 percent.


The biometric indices indicate that the

subjects enrolled in these studies did have airflow

limitation. Their mean FEV1 ranged from 40 percent

to 42 percent, and the mean FEV1/FVC ratio ranged

from 47 percent to 51 percent. The percentage of

subjects across studies with a 12 percent

improvement in FEV1 and a greater than 200 ml

absolute change in FEV1 post-bronchodilator therapy

was 54 percent to 55 percent. The demographics of

this patient population mirrored the demographics

that we see in the COPD population in general in

that most of the patients were of Caucasian origin,

and the majority of the patients were male. This

is pretty typical of the COPD population in general

and the FDA acknowledges that.


One of the concerns we have with the

patient population in these trials is that the

patient population was made up of only persons who

met the stringent clinical symptomatic definition

of chronic bronchitis. While it is well understood

that chronic bronchitis and emphysema can occur

together, the entry criteria eliminated patients

who did not have chronic bronchitis who would have

otherwise met the definition of COPD. The sponsor

did report that 75 percent of patients had

emphysema, but this was based on patient

self-reporting without objective criteria.

The COPD symptoms of cough frequency,

cough severity, sputum production and chest

symptoms were evaluated on a Chronic Bronchitis

Symptom Questionnaire, and the baseline scores

ranged from 6.9 to 7.5 out of a maximum possible

score of 16. At baseline most patients had a

dyspnea score of about 2, a moderate dyspnea, on

the 5-point scale of the Modified Medical Research

Council Dyspnea Scale.


This bargraph depicts the percentage of

patients discontinuing from the study for any

reason. Shown in purple is Advair; green, placebo;

teal, salmeterol; and gold, fluticasone. I will

use this color code in subsequent bargraphs.

The percentage of discontinuations in both

studies was relatively high. In the Advair 250

study 30 percent of subjects discontinued from the

study. In the Advair 500 study the discontinuation

was 35 percent. Looking at the Advair group

compared to the placebo group, the percentage of

discontinuation in the Advair and placebo groups is

quite similar. Thirty percent of subjects in the

Advair group discontinued compared to 32 percent in

the Advair 250 study, and 32 percent of subjects in

the Advair group compared to 38 percent of subjects

who discontinued in the Advair 500 study.

There are two concerns with these data.

The discontinuation rate for Advair is quite

similar to the discontinuation rate for placebo in

both studies. One might expect in a clinical trial

with an active treatment that the discontinuation

rate in the active treatment would be much less

than the discontinuation rate in the placebo group.

Secondly, such a high dropout rate complicates the

interpretation of the effect size.


As Dr. Meyer pointed out in his

introductory remarks, the agency agreed with the

prespecified primary endpoints chosen by the

sponsor to evaluate Advair. Again just to refresh,

pre-dose FEV1 was the endpoint chosen to evaluate

the contribution of fluticasone in the combination,

and for this evaluation the comparison of interest

is Advair versus salmeterol. The two-hour

post-dose FEV1 was selected to evaluate the

contribution of salmeterol in the combination, and

for this evaluation the comparison of Advair versus

fluticasone is the comparison of interest.


Depicted on this table are the results for

the pre-dose FEV1, in other words, the evaluation

of fluticasone in the combination. Shown here are

the results for the Advair 250 study, and here are

the results for the Advair 500 study. This first

line depicts the mean FEV1 baseline values for

Advair -- quite similar in both studies. The

second line depicts the mean FEV1 at baseline for

salmeterol -- again, quite similar results. The

mean change from baseline at endpoint between

Advair and salmeterol, in the Advair 250 group was

an adjusted mean difference of 69 cc. In the

Advair 500 group the adjusted mean difference was

67 cc. These numbers had statistically significant

p values.

Not, however, that in the Advair 500 group

the result is almost identical for the Advair 250

group. This is noteworthy given that the dose of

fluticasone being evaluated here is twice the dose

evaluated here.


Looking at the two-hour post-dose FEV1, or

in other words, the contribution of salmeterol to

the combination, again, shown in the first row is

the mean baseline FEV1 for Advair which was seen

before in the previous table. The mean FEV1 at

baseline for fluticasone is shown here in this

table, and they are fairly similar. At endpoint

the mean change from baseline in two-hour post-dose

FEV1 between Advair and fluticasone is 124 cc

adjusted mean difference for the Advair 250 product

and 129 cc adjusted mean difference for the Advair

500 product. Again, the results are quite similar

and have statistical significance but, as opposed

to the previous study, these similarities are not

unexpected because in this situation we are

evaluating the same dose of salmeterol.


Looking at the overall efficacy of the

Advair product, that is, the comparison of Advair

versus placebo mean change from baseline at

endpoint, looking at the primary endpoint, pre-dose

FEV1, the overall ITT population, both for the

Advair 250 product and the Advair 500 product, had

a similar treatment effect, 164 cc for the Advair

250 product compared to 160 cc for the Advair 500


When these results are broken down by the

reversible and non-reversible population we see

that in the reversible population the treatment

effect is greater than in the non-reversible

population. We did not perform inferential

statistics on these data, however, looking at the

Advair 250 product, the effect in the reversible

population is numerically about two-fold the effect

in the non-reversible population. Looking at the

Advair 500 product, the effect is about one and a

half times in the reversible population compared to

the non-reversible population. Again, these are

numerical differences.


Looking at the overall efficacy, Advair

versus placebo for the two-hour post-dose FEV1,

again the results in the overall population

indicate that both the Advair 250 and the Advair

500 products had a similar treatment effect, 223 cc

for the Advair 250 product and 233 cc for the

Advair 500 product. Again, the reversible

population had a greater treatment effect than the

non-reversible population, and these are numerical

differences. Inferential statistics were not done

on these data.


As stated earlier, one of the sponsor's

stated objectives of this program was to compare

patient-reported outcomes or quality of life in

COPD patients receiving Advair compared to patients

receiving fluticasone, salmeterol or placebo.

We do agree that evaluation of

patient-related outcomes may be helpful in

assessing the clinical relevance of FEV1 changes

and assessing whether pharmacotherapy is of

benefit. The sponsor used the Chronic Respiratory

Disease Questionnaire in both studies to evaluate

this. The minimally important clinical change was

defined as improvement of ten or greater in the

overall score. This was based on the 0.5 point per

item improvement in minimally clinically important

change that has been previously described in the

literature. So, in our assessment this definition

for the minimal clinically important change was

appropriate. For treatment comparisons, a

difference in the mean change from baseline at

endpoint between treatment groups of at least ten

in the overall score was considered clinically



This table gives the results for the

overall score for the disease questionnaire. What

we are looking at is the treatment difference in

change from baseline at endpoint between treatment

groups. Compared to placebo and compared to its

individual components, neither the Advair 250

product nor the Advair 500 product achieved a

difference that was clinically meaningful. For the

Advair 250 product and the Advair 500 product,

compared to placebo, the amount of change

attributable to Advair was 5. Similarly, compared

to salmeterol and fluticasone, the amount of change

was less than the minimal clinically important



Additionally, in neither of the four

domains -- dyspnea, fatigue, emotional function and

mastery -- did Advair achieve a clinically

meaningful important change at endpoint or at any

other time point between its comparators, placebo

or other individual components. For example, in

the dyspnea domain, where the minimal clinically

important change was defined as 2.5, again based on

the 0.5 point per item improvement criterion,

compared to placebo the amount of change

attributable to Advair 250 was 1.2, and for Advair

500 2.1. Compared to its individual components,

the amount of change attributable to Advair was

even smaller.


One of the purported benefits of inhaled

corticosteroids in the literature is reduction in

COPD exacerbations. As you have heard before, one

of the recommendations in the recently published

NIH GOLD document for the use of inhaled

corticosteroids in COPD patients is for patients

who have repeated exacerbations. Therefore, we

feel that it is important to look at these data.

The sponsor evaluated four secondary

endpoints related to COPD exacerbations. They were

severity of exacerbations, time to first

exacerbation, time to first moderate or severe

exacerbation, and number of withdrawals due to COPD



This bargraph depicts the percentage of

subjects with COPD exacerbations of any severity.

Again for the color code purple represents Advair;

green, placebo; teal, salmeterol; and gold,

fluticasone. The percentage of exacerbations was

relatively similar across treatment groups for the

Advair 250 product and for the Advair 500 product.

In the Advair 250 study, Advair had 40 percent of

subjects with COPD exacerbations compared to 39

percent in the placebo group. In the Advair 500

group, 41 percent of subjects on Advair reported

exacerbations compared to 44 percent in the placebo



Looking at the percentage of subjects with

moderate of severe exacerbations -- and you have

heard the definition of moderate and severe

exacerbations before. It was based on treatment.

Patients treated with antibiotics were defined as

having moderate exacerbations. Patients treated

with corticosteroids or patients who were

hospitalized for an exacerbation were defined as

having a severe exacerbation.

Again, the results are quite similar for

the Advair 250 product and the Advair 500 product.

In fact, in the Advair 250 study they were

identical. Thirty-four percent of subjects in the

Advair group and in the placebo group reported

exacerbations. In the Advair 500 study 37 percent

of subjects receiving Advair, compared to 35

percent of subjects receiving placebo, reported

moderate exacerbations.


Looking at the percentage of withdrawals

due to COPD exacerbations -- and subjects were

withdrawn from the study for a COPD exacerbation if

they had a severe exacerbation or if they had more

than two exacerbations requiring antibiotic

therapy, in other words, if they had severe

exacerbations or if they had more than two moderate


The results were low across both studies

and they were quite similar for the Advair 250

study and the Advair 500 study. In both studies

the percentage of subjects withdrawing due to COPD

exacerbations was about eight percent in the Advair

250 and the Advair 500 groups, with similar

percentages in the placebo group.


The sponsor assessed COPD symptoms using a

modified bronchitis symptoms questionnaire, which

is a modified version of the Thomas Pettit

questionnaire. The sponsor evaluated the symptoms

of cough frequency; cough severity; chest

discomfort and sputum production. Each symptom was

graded on a scale of 0-4, and 0 denotes no symptoms

and 4 denotes worst symptoms. Individual scores

were added to give what was called a global

assessment score, or GS.

To define the minimal clinically important

change for this questionnaire, the sponsor matched

changes from the baseline GS with a separate

measure of change in chronic bronchitis symptoms

called the global rate of change. The global rate

of change, as you are aware, is described in the

literature and scoring goes from minus 7 to plus 7,

where 0 denotes no change, negative numbers denote

deterioration and positive numbers denote

improvement. With this assessment, the sponsor

defined a minimal clinically important change for

this questionnaire as 1.4 or greater. With this

evaluation, we feel that this was a reasonable



This table shows the results for the

Chronic Bronchitis Questionnaire, the differences

from baseline endpoint, treatment comparisons for

the two studies, Advair 250 and Advair 500.

Compared to placebo neither in the Advair 250 study

nor the Advair 500 study did Advair achieve a

minimal clinically important change of 1.4. In

other words, with respect to symptom improvement

there was no difference when the patients in the

Advair group were compared to the placebo group.

Similarly, compared to the individual components

Advair did not appear to have a treatment advantage

for chronic bronchitis symptoms either in the

Advair 250 product nor the Advair 500 product.


The sponsor also evaluated the impact of

Advair on dyspnea using the Transitional Dyspnea

Index. In this instrument the minimal clinically

important change is defined as 1 or greater. For

the Advair 250 study compared to placebo,

salmeterol and fluticasone, the Advair 250 product

did not achieve the minimal clinically important

change. However, for the Advair 500 product

compared to placebo, Advair 500 had a change of 1.7

which is greater than the minimal clinically

important change of 1.0. Also, compared to

salmeterol, the Advair 500 product had a change of

1.2, greater than the minimal clinically important

change. Compared to fluticasone there was really

no change.


Moving on to talk about safety, the

sponsor conducted an extensive assessment of

cardiovascular safety using ECGs and Holter

monitoring. The pattern of adverse events did not

suggest that COPD patients taking the combination

of salmeterol and fluticasone were at increased

risk for cardiovascular events. The incidence of

cardiovascular events was similar across treatment

groups. There was no clinically significant change

in heart rate. There were no drug-related QTc

changes. On Holter monitoring there was one case

of heart block identified with Advair 500, but

there were other Holter monitoring changes in other

treatment groups.


A relatively high percentage of patients

reported adverse events during these two studies.

However, this finding is not unusual in studies of

this duration. In these studies a higher

percentage of subjects in the Advair groups

reported adverse events compared to placebo. For

the Advair 250 product, 70 percent compared to 64

percent in the placebo group, and for the Advair

500 product, 78 percent compared to 69 percent in

the placebo group.


The profile of adverse events noted that

were at a higher frequency in the Advair group

compared to the placebo group was similar to the

profile of adverse events seen with inhaled

corticosteroids. For example, for the Advair 250

product ten percent of patients reported

candidiasis of the throat and mouth compared to one

percent in the placebo group. Five percent

reported hoarseness and dysphonia compared to no

reporting in the placebo group. Three percent

reported viral respiratory infections.


With the Advair 500 product 17 percent

reported upper respiratory tract infections

compared to 10 percent in the placebo group; 8

percent reported viral respiratory infections

compared to 3 percent in the placebo group; and

candidiases reporting here was 7 percent for Advair

500 compared to 1 percent in the placebo group.

There were less differences for placebo and Advair

for hoarseness and dysphonia.


Looking at other adverse events, across

the studies fractures were rarely reported and

there was no clear signal. No cataracts were

reported in these two studies. There were two

reports of ocular pressure disorders in the Advair

500 group and one in the placebo group. Elevated

blood glucose was reported as being similar in the

Advair and placebo groups, but the cut-off for that

was fasting blood glucose over 175 mg/dl.


In looking at HPA axis effects, the mean

AM cortisol levels were comparable in the Advair

and placebo groups on treatment day one and

endpoint. No adrenal insufficiency was observed

with the ACTH stimulation testing but, as you heard

before in Dr. Lee's talk, ACTH stimulation is less

than a sensitive method to evaluate for less than

complete adrenal insufficiency.


Summarizing, Advair 250 and Advair 500

both meet the efficacy criteria for combination

drugs and the primary endpoints. The efficacy for

Advair 250 and Advair 500 was very similar, and

almost identical in some evaluations. Numerically

the effect size in reversible subjects was greater

than the effect size of the non-reversible



Of clinical importance is the observation

that no clear treatment advantage with Advair was

noted for COPD-related quality of life or

patient-reported outcomes, COPD symptoms or COPD

exacerbations. It is also not clear whether there

is a treatment advantage for improvement in

dyspnea. There was a clinical significant

improvement at endpoint with the TDI instrument for

the Advair 500 product, however, there was no

clinically significant improvement in dyspnea

compared to Advair 500 and its components in the

dyspnea domain of the Chronic Disease

Questionnaire, a well validated instrument.

Taken together, these overall efficacy

findings form the basis of our concern regarding

the clinical relevance of the FEV1 findings since

the efficacy of Advair on airflow limitation did

not translate into a clear clinical benefit.


With respect to safety, the adverse events

that were seen that were higher in the Advair group

compared to the placebo group were similar events

that have been previously noted with inhaled

corticosteroids -- candidiasis, viral strain

infections, hoarseness and dysphonia with both

Advair products, and in the case of Advair 500 a

higher incidence of upper respiratory tract


Again, no adrenal insufficiency was

observed in these two studies but bear in mind that

this method of testing for adrenal insufficiency

might not be able to determine subtle changes in

adrenal function. Finally, the studies were not

designed, nor were they of significant duration, to

evaluate bone mineral density or ocular effects.

This concludes my talk. We will now have Dr. Mary

Purucker who will summarize and then we will have


Summary and Issues for PADAC

DR. PURUCKER: Good morning, everyone -- I

guess it is almost noon.


I am Mary Purucker, a medical team leader

in the Division of Pulmonary and Allergy Drug



I would like to present a brief summary of

our review of the two applications submitted for

the indication of maintenance treatment of COPD,

starting with efficacy. This will be followed by a

safety summary, primary from the perspective of the

corticosteroid moiety common to the two products,

fluticasone propionate. I will cover the

information submitted with the two applications but

will also briefly discuss some relevant

non-application safety data. I will then proceed

with a wrap-up and discussion points I would like

to have the advisory committee consider.


With regard to efficacy, statistical

significance was not replicated for the primary

endpoint, change from baseline in pre-dose FEV1 for

Flovent 250 mcg BID. It was replicated for the 500

mcg BID dose, with an effect size of 50 cc and 113

cc. This effect size, seen at 24 weeks in trial

3025 was similar in magnitude to that seen in the

ISOLDE study at three months, that is, 70 cc and

100 cc.

The combination product, Advair, also

replicated the finding of efficacy for both primary

endpoints. I show only pre-dose FEV1 versus

placebo comparison because this endpoint measures

the contribution of the fluticasone moiety to the

drug product, and it is this moiety that is novel

in COPD. Also, it is the fluticasone that varies

with the strength of this product, not the

salmeterol. Therefore, it is important to repeat

the finding that was discussed earlier by Dr.

McClain, that is, there is no dose response evident

for the two doses of Advair in this analysis, 165

cc and 160 cc.

Also, as you have heard, we have raised

several concerns related to the robustness of the

finding of efficacy. In particular, there is a

failure to demonstrate clinically significant

differences from placebo with the quality of life

or patient-reported outcome instrument. There is

also a failure to demonstrate clinically

significant differences in COPD exacerbation

between active treatment and placebo. We also have

concerns about the generalizability of this finding

to the overall COPD population.


With regard to safety, our primary concern

is with the corticosteroid moiety that is common to

the two products, fluticasone. The safety of the

moiety salmeterol at the proposed doses has been

previously established in this population and FDA's

review disclosed no new or unique toxicities that

could be attributed to the long-acting beta-agonist

component in the combination product.

With regard to fluticasone,

steroid-related adverse events were observed in a

dose-related manner in the three pivotal trials --

oral candidiasis and dysphonia, for example,, as

you have just heard.

Fluticasone is systemically available in

the relevant population in a dose-dependent manner,

as demonstrated by stead state PK sampling

conducted during pivotal trial 3025.

Moreover, there was a dose-related effect

on the HPA axis, as shown by a 10 and a 21 percent

reduction in serum cortisol AUC relative to

placebo. The potential for the corticosteroid

system effects should, therefore, be assume, in

particular on bone, eyes, connective tissue and

metabolism. If approval is granted, then the

products ought to be labeled for these effects as

accurately as possible.

Unfortunately, the pivotal and supportive

studies submitted with the package were not

designed or powered to detect a difference in many

endpoints that correlate with corticosteroid

systemic safety in the population of interest. I

will return to this issue momentarily.

Let me add that the long-term safety is

important in this application. Contrary to the

five-year 50 percent mortality cited earlier today

for severe COPD, the patients in these three

studies had an annualized mortality rate of 0.4

percent. Also, I think that while the observation

that FP levels in patients with COPD may be less

than in patients with asthma is irrelevant since

the PK/PD relationship is not necessarily the same.


This slide shows the results of a search

by indication of the FDA adverse event database for

all reports submitted for any inhaled

corticosteroid, including fluticasone, for the

indications of COPD, emphysema and chronic

bronchitis. The search was performed by Dr. Joyce

Leber, of the Office of Drug Safety, who used a

cut-off of November 15 of last year.

A total of 206 cases were retrieved, all

but 14 from the past three years, accounting for a

total of 213 adverse events. Patients were in

general elderly; about half were women; and the

doses of ICS ranged from 80 to about 8000 mcg per

day and varied with moiety. About half of all

adverse events were reports of lack of efficacy or

worsening of COPD. Several of the remainder

adverse events are notable for systemic

corticosteroid events, as shown on this slide. At

least one of the cataracts was reported as a

posterior subcapsular cataract. The bone events

included pathological fractures, osteoporosis and

aseptic necrosis. Adrenal events were equally

divided between insufficiency and hypocorticism,

and skin adverse events included bruising and easy



Let me now return to the issue of specific

systemic effects or corticosteroids starting with

bone. Chronic systemic corticosteroids may lead to

osteoporosis through a variety of mechanisms,

including inhibition of osteoblasts, inhibition of

GI calcium absorption and its effect on collagen

synthesis. There is also individual susceptibility

related to activity level, gender, menopausal

status, genetics and smoking history.

On a population basis, therefore, bone

effects may occur with chronic ICS, particularly at

high doses. Ideally, these bone effects should be

quantified by a proper risk-benefit assessment.

Although bone mineral density was not specifically

studied in the three pivotal trials of this

supplemental NDA submission for the population in

question, summary data from two two-year supportive

trials of asthmatics, 3001 and 3017, was provided.

I might add that we were not provided with the

primary data from 3001 to review, only with data


Other considerations with regard to this

data is that this is a different population, and

that they were generally younger. They were

asthmatic and the women were all premenopausal.

Given these caveats by sponsor report, trial 3001

did find a decrement in bone mineral density in the

lumbar spine at the high dose, and trial 3017

reported decreased bone mineral density in

measurements of the femur. The latter site was not

prospectively validated however.


I might add that the published bone

density trials involving fluticasone cited by Dr.

Shah in his presentation earlier today, in

particular in slide 108A, were very small, with the

Ns per treatment arm typically less than 30. The

patients were generally young asthmatics and

treatment duration was one year or less. This

provides no reassurance of the safety of

fluticasone on bone in the COPD population. With

this in mind, we should turn to additional evidence

in the published literature.

This slide provides additional information

regarding the long-term effects of ICS on bone.

Important caveats include the fact that a different

moiety and ICS formulation was used for the Lung

Health Study, and multiple different ICS moieties

were used by patients in the two asthma trials.

The latter two trials also studied a different

patient population than COPD.

Nevertheless, I believe it is important to

recall the results of the Lung Health Study II,

reported a little over a year ago, which showed

that treatment of a population of COPD patients

with 1200 mcg per day of the ICS triamcinolone over

the three-year period was associated with a

statistically significant decrement in bone mineral

density at both the femur and the lumbar spine.

The first of the two asthma studies was

published in The New England Journal last year, and

was authored by Eliot Israel and his colleagues.

This study was a three-year prospective cohort

study of 109 premenopausal women with generally

mild to moderate asthma. A statistically

significant dose-related decline in bone mineral

density at the total hip and trochanter was found,

which persisted even after the exclusion of women

who had received oral or parenteral


The second study, by Wong and colleagues,

was published in Lancet and was a cross-sectional

study of 196 young asthmatics between the ages of

20 and 40 years. Of this group, a little over half

were women. The mean duration of ICS use was six

hears, and BDP and fluticasone were the ICS

moieties used. The study showed a statistically

significant cumulative dose-related decrement in

bone mineral density at the hip, the trochanter,

Ward's triangle and the lumbar spine.

An accompanying editorial by Philip

Sanbrook used data from this study to estimate that

seven years of treatment with a dose of ICS that is

equivalent to 2000 mcg per day of BDP would result

in a decrement of one standard deviation in bone

mineral density or one T-score. This approximately

doubles the risk of fracture.

While we need to be cautious about

applying the results of these studies to the

products under consideration today, given the

caveats that I have identified earlier, we must

also be cautious in the other direction in that

these data imply a class effect of ICS on bone. It

is, therefore, important to quantify this effect

for a given ICS for a given population whenever



This slide summarizes the HPA axis

information on fluticasone provided in the

submission from the three pivotal trials and

supporting studies. To review, in study 3025 there

is a dose-related effect on the HPA axis as

demonstrated by a 10 percent and a 25 percent

reduction in serum cortisol AUC for Flovent 250 and

500 mcg BID respectively.

The other two pivotal trials conducted

cosyntropin stimulation testing in a subset of

about 20-25 percent of the participants. No renal

insufficiency was reported but, as noted earlier,

the test is not designed or validated to quantify

levels of adrenal suppression.

The ISOLDE study measured AM cortisol at

baseline, then at three-month intervals for the

duration of the three-year study. As reported by

the sponsor in the submission, there was a 10-15

percent reduction in mean AM cortisol for the

fluticasone group in comparison to placebo at all

post-baseline time points. Further analysis by

shift tables disclosed that 20 percent of the

fluticasone group had a shift from normal cortisol

values to low cortisol values compared to nine

percent of the placebo group.

Finally, the clinical pharmacology study,

1003, was a single dose PK/PD crossover study of

1000 mcg of fluticasone, proposed total daily dose,

administered to normal volunteers. There was a

35-59 percent reduction from baseline in 24-hour

urinary cortisol that was observed in these



This slide covers the epidemiological

evidence that draws an association between the dose

and duration of ICS use in the occurrence of

cataracts or posterior subcapsular cataracts in one

of the studies in a middle aged and elderly


Again, to be fair, I want to point out at

the start that these studies are not randomized

controlled trials. Several different ICS moieties

were used in the populations in question and, in

fact, fluticasone may not even have been approved

in these two countries at the time that the studies

were conducted, which was the early and mid-90s.

Nevertheless, I believe that we have

established that fluticasone is systemically

available at the doses proposed and in the

population of interest, and has measurable systemic

effects and, therefore, an association of posterior

subcapsular cataracts with chronic use of

fluticasone-containing drug products should not be


The first study, by Cumming and

colleagues, was a cross-sectional study of about

3700 subjects in Australia. Among the 370 ICS

users identified, PSC was found at a two-fold

greater prevalence among the ICS users than

non-users, and prevalence was higher among subjects

with a higher cumulative lifetime dose.

The second, or the Canadian study, was a

case control study. They selected cases based on a

history of surgical cataract extraction using the

Provincial Insurance Health Database. The study

determined that the use of ICS for greater than

three years was significantly associated with

undergoing cataract extraction, for an odds ratio

of slightly greater than 3. For high average daily

doses the risk was elevated after only two years.

In conclusion, given the limitations of

this analysis based upon the above caveats, the

possibility of ocular adverse events should be

considered in the overall risk-benefit assessment

of Advair and Flovent proposed for the indication

of maintenance treatment of COPD.


In conclusion, efficacy has been very well

studied in these applications. There is

substantial data that is open to clinical

interpretation. If approval is recommended for one

or both of these products there would be labeling

issues remaining with regard to efficacy, but they

would not be insurmountable.

In contrast, the safety database for this

population is limited in describing long-term

risks. One of the questions that we posed for the

advisory committee is whether there are adequate

data from which to construct a label for the

potential long-term effects in the COPD population,

particularly with regard to bone.

The potential for other systemic

corticosteroid effects must be assumed, and we must

ask ourselves whether there are sufficient data to

write an informative label so that the practitioner

may make a reasoned choice as to safely and

effectively using these drugs in the COPD

population if they are, indeed, to be recommended.


Which brings us to the specific issues for

consideration by the committee to which I have

already alluded. First as it relates to product

efficacy, we would like you to discuss the patient

population with regard to the generalizability of

the findings to the COPD population as a whole.

Factors to consider may include the degree of

reversibility and the presence of chronic

bronchitis. Bear in mind that the proposed

indication is for long-term twice daily maintenance

treatment of COPD, including chronic bronchitis and


Second, also as it relates to product

efficacy, we would like you to discuss the primary

endpoint, change from baseline in FEV1, with regard

to its clinical relevance to the treatment of COPD.


Finally with regard to safety, we would

like to ask the committee to consider whether the

data are sufficient with regard to the potential

long-term impact on bone or other relevant systemic

corticosteroid safety endpoints.


Thank you for your attention. I would

like to acknowledge my colleagues who contributed

professionally to the thorough and timely review

and presentation of these applications. Thank you.

DR. DYKEWICZ: Thank you. I am going to

allow only about five minutes for questions at this

point because I think this afternoon there will be

plenty of time for discussion and for posing any

questions. So, for members of the committee, do we

have any questions at this point that are kind of

burning issues that you would like to get off your

chest? Dr. Fink?

DR. FINK: Just a question for FDA in

general as regards interpreting this data, is it

the FDA's concern that inhaled corticosteroids have

a worse safety profile, both in COPD and asthma,

than initially was apparent, or is the concern

solely that the risk-benefit ratio in COPD is worse

than in asthma?

DR. PURUCKER: I don't mean to imply that

inhaled corticosteroids are not safe or effective

for asthma. At this point what we are trying to

ascertain is what the safety profile is; what the

long-term risks are so that if they are, indeed,

recommended for approval how to construct the

label, or how, indeed, to weigh the risk-benefit.

Some of what I presented is the fact that there is

an absence of data for the long-term use of

fluticasone in the population of interest, the COPD

population. That is one of the issues that we

have. There are probably safety issues and we

can't quantify them.

DR. MEYER: Let me just follow on to that.

I agree with what Dr. Purucker said. I think we

have, as an agency and as a scientific community or

medical community, better appreciated some of the

long-term systemic effects of corticosteroids in

recent years, although I think the agency has known

about such issues for a long time, particularly

with regard to our experience with the spontaneous

adverse event reports.

But I think the main question here is not

that issue so much as it is specific to COPD,

whether we even know enough to say whether this

population is particularly sensitive to some of

these safety issues, number one and, number two,

given the differences in the efficacy seen in these

studies and perhaps other studies compared to the

kind of efficacy you see in asthma, are the safety

issues that we know about or don't know about --

how do they factor into the risk-benefit equation?

DR. DYKEWICZ: Thank you. Maybe just one

more question before we break for lunch.

DR. STOLLER: My question is also

procedural. Bob can address this. In assessing

the efficacy outcome measure, obviously the

pre-dose FEV1 for fluticasone and the two-hour post

for the salmeterol dimensions have been selected

and agreed upon. Clearly, there are at least four

randomized trials that have assessed delta FEV1

over time about which we have, obviously, heard

nothing as an efficacy measure. The question is in

assessing efficacy versus safety outside the

parameters reflected here, how relevant is that to

the conversation of the committee to an assessment

of efficacy?

DR. MEYER: Are you asking about

essentially whether any improvement in baseline

FEV1 continues over time?

DR. STOLLER: I am asking about the change

in slope of FEV1 --

DR. MEYER: Right.

DR. STOLLER: -- not being assessed in

these studies, but available from antecedent

literature but not being negotiated a priori as a

primary outcome measure. So, is it off the table

as a relevance issue, or is it a consideration? It

is a procedural question.

DR. MEYER: We certainly had this kind of

discussion with the sponsor beforehand, and in our

mind it is somewhat a different issue as to whether

there is shorter-term benefit that you might see in

a six-month trial versus preservation of lung

function over time, which would require much larger

and longer trials. Even with asthma where it is

quite clear that the inhaled corticosteroids have a

short-term effect that is durable, it is still not

entirely clear to me that there is a lung

preservation effect. If you look at the CAMP data,

for instance, that is not entirely clear, and I

don't think it is entirely clear or, in fact, very

well supported by the data in the literature to

date for COPD either. But I think what you need to

focus on today is the sort of shorter-term but

durable response that was studied in these studies.

DR. DYKEWICZ: Thank you. Let's now

adjourn for lunch and reconvene at 1:10 p.m.

[Whereupon, the proceedings were recessed

for lunch, to be resumed at 1:20 p.m.]


DR. DYKEWICZ: What I am first going to

do, because we broke off for lunch just shortly

after the FDA presentations, is to give an

opportunity to members of the committee to pose

questions to the FDA presenters. Would anyone like

to be recognized at this time? Dr. Joad?

DR. JOAD: I am curious for the FDA to

answer the same question that I asked to the

sponsor, which is to go over why the statistically

significant differences in the three questionnaires

didn't impress you, and how confident you are about

these numbers that have been given as clinically


DR. MEYER: I think the easiest one to

speak to is the Chronic Respiratory Disease

Questionnaire that Guyatt and Juniper developed. I

think the bronchitis questionnaire, as the sponsor

said, was a relatively new modification of an

instrument that is not as well validated. So, I

think we can sort of put that one aside.

The CRDQ, as with perhaps all instruments

that Juniper and Guyatt have developed, they

defined a meaningful clinical difference by an

actually non-interventional setting, looking at a

cohort of patients over time, finding patients who

change in a global scale of "how do you feel," or

sort of a global, broad one question quality of

life assessment, and correlate that to changes in

their particular instrument over time. Based on

the results of how the patient has fared in the

global question, then determine what would be a

clinically important difference, something

meaningful to the patient that they might be able

to detect, or might mean a change in therapy for

their particular instrument.

So, it is strictly true that these are not

to look at between treatment differences, but it is

also true that it is not really meant to assess --

it wasn't developed and validated specifically to

assess treatment effects at all. This important

minimal difference was looked at and derived from

spontaneous change over time, not change in

response to intervention.

All those caveats aside, I think we go

with what seems to be the best assessment of what

might be a clinically meaningful difference to a

patient, which is what Juniper and Guyatt have

determined and what the sponsor prespecified. So,

I think that on the CRDQ you would want to see not

only that change from baseline in a particular

therapy, but you would want to see that the

attributable effect in reference to placebo was

meaningful as well.

I think the TDI is also a very well

validated instrument. I think a meaningful

difference in that is less well validated. I think

the sponsor did a reasonable job of identifying

what they thought it would be based on speaking to

the developer of the instrument, and I think we

felt it was reasonable a priori as well.

DR. BONE: Could I just pursue this a

little bit? I guess sometimes we need to make sure

we are applying the same scientific rigor to the

selection of our tools as we are to what we are

trying to measure. I am trying to understand here,

have these instruments, specifically these minimal

significant differences, been shown to be strongly

correlated with, let's say, other harder outcomes

in clinical trials, such as survival or other

indices of morbidity, or other measures where we

can say, okay, a difference of so much in this

scale predicts a better outcome over a period of


DR. MEYER: I think most of such

instruments do have some level of correlation with

those kinds of endpoints. That is generally done

as part of the validation of the instruments. Of

course, if you take FEV1, for example, that

correlation is not particularly high but you

wouldn't expect it to be because it is not

measuring the same thing. It is really measuring

the patient's perception of their disease which is

multifactorial, and lung function is only a part of

that. I don't know the specifics actually of the

CRDQ as far as follow to morbidity and mortality,

or at least major morbidity, but I think that those

kinds of looks are generally done with these

instruments, and CRDQ is a reasonably well

validated instrument.


DR. WISE: I think I would like to

follow-up on that a little bit, and the notion of

clinically meaningful changes, particularly in

looking at the mean of a population since these

have been validated in terms of what is important

to an individual or perceptible to an individual in

terms of a change in status. Very commonly we see

mean changes in populations that seem small but

have important clinical effects at the ends of the

population. It is kind of a multiplier effect

where people out at the ends, if it is a broad

distribution or a skewed distribution, can show

remarkable benefits. Sometimes people have looked

at percentage of people in one group versus another

group who have a clinically meaningful response. I

wonder what your views are on that, and whether

that data has been available in this.

DR. MEYER: Let me make two observations

about that. I think that is certainly true and I

think that neither the agency nor necessarily other

parties have fully settled on the best way to

assess the clinical interpretation of these

results. One thing that that raised in my mind --

this sort of gets back to Dr. Joad's question a

little bit -- is that it is important to understand

that the statistical experience or the experience

we have with these instruments, particularly the

instruments of Juniper and Guyatt, is that the

numbers that you would need to enroll in a trial

should the difference between the treatments that

you are comparing reach that clinically significant

difference, or clinically important difference, is

only about 30-35 patients. It is not very many.

So, in fact, just as the sponsors caveated

some of the observations about secondary endpoints,

that the trials were not designed to specifically

do differential testing on all those, I think, that

one of the design issues for interpreting these

statistically significant results that don't meet

the prespecified clinically important difference

between groups is the fact that these studies are,

in fact, very much overpowered for looking at the

statistics of these instruments.


DR. FINK: From an FDA standpoint, could

you give us some perspective, particularly for

quality of life or patient-reported outcomes, how

this data compares with previously approved drugs,

such as salmeterol and ipratropium where they

clearly showed a pulmonary function effect, were

those drugs capable of showing patient-reported

outcome effects?

DR. MEYER: I really hesitate to do that

based on cross-study comparisons. I can say that

if you look at the labeling for ipratropium it

specifically mentions the use of a patient-reported

outcome instrument and specifically states that

there were not significant differences found. So,

I can say that.

I would also emphasize, however, that

those drugs have a very specific indication for the

relief of bronchospasm associated with COPD. They

are not for the maintenance treatment of COPD,

which is a rather different kettle of fish.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: I think Dr. Wise's question

raises for me a follow-up that I guess I would

appreciate your comment on, and it has to do with

looking at mean values in a dichotomous way. As I

understand the issues put on the table, and I

agree, if there is a mean delta of the

pre-bronchodilator FEV1 of 100 ml in aggregate, as

a group, it might be relevant, to my understanding

of efficacy, to have that stratified by different

subsets. Admittedly, not done a priori but even in

an ad hoc way after the fact, to have it stratified

by baseline FEV1 strata and then to look at the

percentage of individuals who experienced an

increment of a certain defined value stratified by

those subsets. So, dichotomously analyzing the

data by subsets, which is characteristically a

dangerous business after the fact but, nonetheless,

it would speak a little bit to the issue I think

you have put on the table with which I agree, which

is, is there a segment of the population, given the

relatively paltry overall FEV1 rise of 100 ml, 70

ml -- can one one ask to see data around the

dichotomous analysis in subsets? Have I made

myself clear? It is kind of a procedural question

I suppose. That is why I put it forward.

DR. PURUCKER: We looked at subsets post

hoc really only based on reversibility, and we

found that the patients who were highly reversible

contributed numerically more to the effect size

than those who were not reversible. Similarly with

cigarette smoking, although I believe the effect

size was more in clinical trial 3025 than it was in

the salmeterol trials, but we did look at those

particular variables.


DR. BONE: To pursue this, were responder

analyses done as secondary analyses by either the

sponsor or the agency, looking at the minimum

significant differences as the criterion for


DR. GILBERT-MCCLAIN: No, we didn't do


DR. BONE: Did the sponsor do that?

DR. SHAH: Yes.

DR. BONE: Were those data submitted to

the NDA?

DR. SHAH: They were for some.

DR. BONE: Well, they were or they


DR. SHAH: Again, we did submit as part of

our integrated summary of efficacy, which is part

of an NDA submission, analyses for subgroups by

responder analysis. Those data were there, but I

don't believe they were done at the individual

study level but we certainly have those data here

if it would be of any use for the committee to see.

DR. DYKEWICZ: I think they would be, and

maybe I will give you an opportunity later to

respond to that. Dr. Malozowski?

DR. MALOZOWSKI: I am not familiar with

this condition, therefore I don't know how common

it is to see in a 24-week study this 40 percent

patient withdrawal from the study. This is either

for the FDA or for the sponsor. How can you assess

data integrity applicability of the outcomes that

are measured when 40 percent of the patients did

not complete the study?

DR. MEYER: Actually, I would say, at

least from the Division of Pulmonary and Allergy

Drugs standpoint, that we don't have a lot of

experience with six-month trials. More commonly we

see three-month trials and I don't think it was

entirely unanticipated that there would be some

dropouts certainly and I think it does raise some

issues, particularly not whether there is an effect

or not but really nailing down what the effect size

is. But this was discussed with the company and I

think that we chose the endpoint analysis basically

as a way to try to deal with that.


DR. BONE: This is a relatively specific

question for Dr. Lee. You referred to a patient

who had a low cortisol value -- if I understood

correctly; it went by pretty quickly -- who was

reported to have suffered a serious adverse event.

I was a little surprised to have a serious adverse

event without any clinical information because the

criteria for severity are clinical. So, could you

explain that further? I am sure there was an SAE

report to be reviewed.

DR. LEE: No, there was no case report

form for that patient. It was supportive data, not

in the pivotal clinical trials and there was no

case report form.

DR. PURUCKER: We didn't have the primary

data from that trial; we just had summary data and

a patient was reported as having had a serious

adverse outcome. We don't have any other details.

DR. DYKEWICZ: Dr. Parsons?

DR. PARSONS: There has been a lot of data

presented on the group of patients that had a

reversible process versus the non-reversible group.

On my reading through, the study was not designed

to look at those two groups independently. Is that

correct? It was initially all-comers and it was a

post hoc analysis to look at those two groups.

DR. MEYER: Clearly, the overall analysis

was going to be the primary analysis. I think it

was understood that we would have an interest in

looking at the results in those two separate

populations but that was not the primary interest.

DR. PARSONS: Are the two populations

large enough to draw conclusions versus one versus

the other?

DR. MEYER: As I think Dr. McClain

mentioned, we were really not paying attention to

the inferential statistics there because we weren't

trying to draw inferential conclusions on this



DR. JOAD: Since asthmatics appear to

respond well to these drugs, how confident are you

that this group that has COPD and does not have a

fair number of asthmatics also included in this

study group -- since asthma can now be defined as

non-reversible -- have a non-reversible component?

It would just confound the data if there were a

group that were highly responding that really maybe

should be called asthma instead of COPD.

DR. LEE: Well, patients with a diagnosis

of asthma were a priori excluded. Could there be

some overlap? It is probably true, there may be

some patients whom some people might define as

being asthma in the population but, you know, I

feel relatively confident with the figures that

were presented.

DR. GILBERT-MCCLAIN: Just to add one

thing to follow-up on Dr. Lee, also the mean

FEV1/FVC ratio that we saw across those studies was

47-51 percent, which is much lower than what we

have seen in the all the asthma studies. So, we

felt that overall the population was representative

of obstruction.


DR. BONE: I am sorry to belabor this

point, but how can you not have an SAE report? If

it was from another trial, you would still have the


DR. LEE: This was supportive data in a

study reported and, you know, I did not have the

entire case report forms for all the withdrawals.

It was not a pivotal study.

DR. BONE: But SAEs are still reported

from any clinical trial. You have to report them

in ten days.

DR. PURUCKER: This was an old trial that

was submitted as supportive data, and perhaps we

should direct the question to the sponsor. Perhaps

they could tell us what the SAE was.

DR. SHAH: We usually provide case

narratives. We may not have provided a case report

form because it was a study that was done several

years ago and it was done for Europe, not U.S. The

case narrative should have most of the information

that I think you may be looking for. We are trying

to see if we can dig it up and if we find it, we

will be happy to share that with this committee.


DR. FINK: Was there anything in the

design of the trials, since we didn't get the total

design of the inclusion/exclusion criteria, or in

the conduct of these trials that would with any

probability have led to a bias toward responders

versus non-responders at enrollment, since we have

heard that clinically 50-60 percent responders from

the COPD population is not surprising? I believe

Dr. Donohue, from North Carolina, said that he was

actually quoting 60 percent of patients who have


DR. GILBERT-MCCLAIN: Just to respond to

that, in the papers that Dr. Donohue referred to,

reversibility in all of those studies was defined

as 12 percent or 15 percent response with

bronchodilator. It was not 12 percent and 200 or

more cc change in FEV1. So, I think that needs to

be taken in the context of those percentages.

DR. FINK: Weren't these studies either 12

percent or 200 ml?

DR. PURUCKER: No, 12 percent and 200 cc.

DR. GILBERT-MCCLAIN: Just to clarify, the

studies that Dr. Donohue referred to were 12

percent or 15 percent. There was no absolute

change in FEV1 as part of the criteria as opposed

to the reversibility here.

DR. FINK: But if FEV1 was below 0.7 L the

200 ml requirement was dropped?



DR. SHAH: Can I just clarify? Well,

maybe Dr. Donohue can speak to that.

DR. DONOHUE: In the IPB trial, just for

everyone's benefit, if you stratify COPD into,

let's say, level three where the FEV1 is 1000 ml or

700 ml, you know, 100 cc can be 20 percent or a

very, very large response. So, in people with very

high lung volumes, they respond primarily with


The IPB -- what originally was reported in

The Annals of Internal Medicine was 15 percent. In

all the recent studies we really tried to include

the 12 percent and the 200 ml so we can take in

those two extremes. I presented at the American

College of Chest Physicians and also at ERS in an

evidence-based symposium and we tried to get at

this question, because it is so key, about

reversibility and the preponderance of the evidence

is that it falls out anywhere around from 50 to 60,

65 percent that do respond with about 12 percent

and 200 ml.

The European studies use a different

standard, and they use, of course, percent of

predicted FEV1. So, we have been interested in

looking at a lot of the United States studies

against that standard from Europe. In fact, the

majority of our patients really would meet the

European standard of non-reversibility.

So, again, those of us who work in the

field, we appreciate the difficulty in trying to

deal with this question, separating asthma from

COPD, but our believe is that most patients do

exhibit reversibility and we teach all the doctors

that if you are giving a bronchodilator to a

patient, give it for a month, two months or three

months. You can't go by this acute reversibility.

Many of the patients that we saw, let's say, in the

salmeterol study that seemed to be irreversible at

baseline, over the course of a month, two months,

you will see a response.

So, that is sort of the caveat but I

appreciate the difficulty we are all having with

that question. It is a tough one, but it does seem

that if you use 12 and 200 a little over half seem

to respond at baseline to albuterol two puffs. The

difficulty in these studies and the salmeterol

study, of which I am a co-author with Mahler, we

used two to four puffs of albuterol. In other

studies we have used 15 percent. So, again, it is

a very, very difficult subject to tackle. Did I

answer the question?

DR. DYKEWICZ: Thank you. First Dr. Wise

and then Dr. Stoller.

DR. WISE: I wanted to clarify was the

reversibility measured after two or four puffs of

albuterol, and if it was four, did that have

influenced the percent?

DR. DONOHUE: Yes, it is up to four in

this study. Maybe Tushar should answer this -- I

was involved in the Mahler studies and in that one

the majority of the people responded to two. Only

a small extra increment needed four puffs to reach

the 12 percent and 200. Do you have the specifics

for this?

DR. SHAH: In this program it was

prespecified to be four puffs of albuterol. So, we

were really trying to get the reversible patients

fully defined and, clearly, we know that there is

some dose response between 200 and 400. We get the

maximal effect at the two puffs but there is an

additional effect that we do see with going up to

four. As Dr. Donohue alluded to, in studies that

we have done in the past and other sponsors have

done it is usually two puffs and occasionally they

do go up to four puffs.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: I would like to make a

comment just really regarding clarification of the

generalizability of reversibility from these data.

The issues, as they appear, have to do with the

dose -- I am glad Dr. Wise clarified that because

that was my follow-on question, but four puffs BID

is characteristically higher than most of the

baseline reversibility studies of which I am aware.

I am not sure what the conclusion of that is but it

is an observation.

The second would be that as I remember the

Anthonisen data which used the 15 percent

criterion, the presence or absence of reversibility

was measured over time with up to seven serial

spirometries. It was not a single baseline

assessment. So, in aggregate over time, as has

been pointed out, most of these individuals will

over time demonstrate reversibility. The question

then defaults, in my mind, to how generalizable is

the experience of reversibility regarding a single

baseline measurement in which 56 percent of

patients demonstrate 12 percent and 200 ml rise in

the post-bronchodilator FEV1, with a dose of

albuterol that is higher characteristically than

was used in any of the other studies that

characterized reversibility? Now, I am not sure

how to interpret that but I think, for the sake of

clarity, we should understand that phenomenon.

I would make one other point with regard

to the presence of reversibility in emphysema,

which is clearly a prevalent phenomenon, and

perhaps data from the alpha-1 antitrypsin deficient

subset of patients, with which I have some

familiarity, bears here. When we looked at this

experience using 12 percent and 200 ml with FEV1,

not with FVC and I presume FVC was not the

reversibility criterion here, again, over serial

examinations, up to three serial spirometries, and

a cohort of FEV1 is about 43 percent of predicted

and an N of 1129 individuals, up to 55 percent of

these individuals with two puffs BID of albuterol

or a comparable drug would satisfy the 12 percent

and 200 ml criterion. But, again, that was

serialized over at least three serial spirometries.

In any given test in a run-in period the prevalence

would not have been anywhere near as high as the

50-55 percent that is being reported here.

Again, I am not sure what the conclusion

of that is, but I hope that that lends some clarity

to the question posed to us about how

representative, with regard to the reversibility of

this population, is it of the COPD population at

large, with perhaps alpha-1 being the

quintessential example of emphysema, not asthma or

not chronic bronchitis.

DR. DYKEWICZ: Thank you. Any other

questions for the FDA at this point? Dr. Wise and

then Dr. Parsons.

DR. WISE: Whoever wants to take this, I

was impressed with the prevalence of thrush, oral

candidiasis at around 12 percent, which seems more

than a clinical experience would warrant. I

wondered how that compared to the trials with the

fluticasone MDI in asthma. Is that comparable?

DR. LEE: Yes, I think the label mentions

about five percent, five percent in both the


DR. MEYER: I think the thing that

complicates that answer, however, is that most of

the pivotal trials done for the fluticasone program

were 12 weeks, and I believe the

corticosteroid-sparing trials were 16 weeks; they

were not 24 weeks. So, this is a higher percentage

than we saw pretty much for the asthma trials. I

think the highest dose of the oral

corticosteroid-sparing trial had a significant

amount of thrush but, of course, many of those

patients were also on oral corticosteroids. I

think this is higher but it is also a longer time

period. Commonly adverse events get called

incidents but we don't commonly correct for the

amount of time that you are looking at. So, you

are not true incidents.

DR. DYKEWICZ: Dr. Parsons?

DR. PARSONS: I actually have the same

question but I want to take it one step further.

Some of the other issues independent of candidiasis

were increased incidence of viral respiratory

infections and upper respiratory tract infections.

Are those also similar percentages to what you saw

in the asthma trials? Even accounting for the

difference in time, is there any way to equate

those at all?

DR. MEYER: I think we would actually have

to look back at the data. I don't think that we,

in my mind, had that as an issue. I don't know if

the sponsor would have better recollection of those

data than I do. I was the primary medical review

officer on Flovent but I don't know the data



DR. SHAH: Thank you. Yes, in asthma,

again, remember that the studies we did with

fluticasone were very differently designed.

Because they were placebo-controlled and we were

withdrawing patients who were having worsening of

their condition, we did have very significant

differences in exposure between treatment groups,

meaning that people in the FP groups were treated

for a lot longer durations. So, if you were to

look at just the overall results, the trends in the

placebo versus the active groups were suggestive of

a higher number on the actives but, as Dr. Meyer

indicated, you know, when you have a study design

which was designed to withdraw patients in

potentially one group you do have to look at the

adjusted results. Though it is not easy to find a

good way to adjust for this phenomenon, I think

clearly the experience would be that in asthma we

don't see any concerns when you try to adjust

between the active and placebo for these various


DR. MEYER: Let me just follow-up on that

because I think that that is an important point.

There really was a very striking difference in most

of those fluticasone trials because of wanting to

protect the placebo patients in the amount of time

that the patients spent in the trials. So, the

placebo patients had a much shorter time as a mean

exposed to the study and, therefore, able to report

study-related adverse events than did the

fluticasone. The recollection, without citing the

specific data, is that we looked at the data and

did not feel that any trends seen were significant

given the differences in exposure time. So, that

is not an issue here in terms of the amount of time

that the groups were exposed. It looks relatively

well balanced across the groups.

DR. DYKEWICZ: Dr. Atkinson?

DR. ATKINSON: I think my question is for

Dr. Gilbert-McClain. Did I understand, as you were

speaking, that the patient selection criteria would

have excluded patients with emphysema primarily?

DR. GILBERT-MCCLAIN: There were no

objective criteria to define emphysema. I know it

is difficult to define emphysema, but emphysema was

by patient self-reporting. As I mentioned earlier,

they had strict criteria for chronic bronchitis but

not for emphysema.

DR. DYKEWICZ: Dr. Apter?

DR. APTER: I guess this is for Dr. Shah.

We were told that GlaxoSmithKline has an ongoing

three-year international study to evaluate the

effect of Advair Diskus 550 mcg BID and fluticasone

500 mcg BID on the survival of COPD patients, and

also there is an evaluation of bone mineral density

and ophthalmologic effects over a three-year

period. What we are hearing today is that the time

of these studies is long by FDA standards but not

long at all for such a chronic disease. Where are

we on those ongoing studies, and when will the

results be available?

DR. SHAH: Yes, this was a study we

initiated about a year ago because we recognized,

clearly, the high morbidity and mortality in this

population and the need for treatment that may

actually improve those outcomes. So, that was the

primary objective of these studies. In the

international study there are going to be over 5000

patients enrolled in the study, but we are planning

to do a subgroup of those patients looking at

various safety measures such as bone density, eye

exams, as well as HPA axis assessments. But that

study has just started so we are looking at almost

at least four more years before we will have

results from those studies.


DR. JOAD: This question can be for either

group. I am just curious. As a pediatrician, I am

used to everything being presented as predicted

because an absolute amount of change in FEV1 is

meaningless to us because it would depend on the

size of the child. Now, I would imagine that

adults differ enough in size, in height and also

effects difference so that it is odd to me that

everything is done in absolute milliliters as the

primary endpoint.

DR. DYKEWICZ: Any response from the FDA

on that?

DR. MEYER: I guess I would just say that

in pediatric trials we do look primarily at percent

predicted or change in percent predicted for the

reasons cited. I think, as you say, adults come in

different shapes and sizes and there, of course,

are differences but in trials of this size they

also tend to be balanced across the groups. So, I

would not traditionally look primarily at that, but

commonly the sponsors, not just GlaxoSmithKline but

other sponsors, do multiple metrics of FEV1 and

often the results are not very different from one

look versus the other in the adult population.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: This is a data and methods

question so it may default to the sponsor as well.

The data question gets to the comment about

objective evidence of emphysema. Although I am

aware that there was no baseline standard

measurement of other parameters that we

characteristically use to evaluate the presence of

emphysema, i.e., diffusing capacity measurements,

and I am aware that that was not part of the data

gathered, I am also aware that occasionally these

measurements are clinically available on members of

the cohort and I am wondering whether any such

descriptors are available for this cohort with

regard to diffusing capacity measurements or CT

assessments that would get to, if you will, other

objective measures of the presence of emphysema as

opposed to chronic bronchitis, and whether that is

allowable or requestable.

DR. GILBERT-MCCLAIN: We didn't have those

data in our submission. Whether or not the sponsor

has those data we do not know.

DR. SHAH: The answer is no.

DR. DYKEWICZ: Dr. Parsons?

DR. PARSONS: In the written review from

the FDA there was a lot of concern about the lack

of ethnic diversity in the study population. I

know that has been discussed to some extent. Are

there any concerns about relating the safety or

lack thereof to ethnic populations based on the

tiny numbers of patients studied? Specifically,

are different ethnic populations more or less

likely to be vulnerable to any of the safety

concerns? You know, hypoglycemia or something like

that that would need to be configured differently

for those populations?

DR. GILBERT-MCCLAIN: We don't really know

that but, again, I would just like to point out

that as far as the demographics are concerned

although, yes, in the study the majority of the

patients were of Caucasian race, the demographics

in the study itself pretty much mirror what we see

in the COPD population in general. Whether or not

minority patients respond differently to therapy,

that issue is not very well clarified. We don't

know that and I am not sure anyone actually knows

that at this time.

DR. MEYER: Again, just maybe to augment

that important point, I think we end up in a

situation, somewhat because of the demographics of

the disease, where we don't have a lot of data. We

neither really have priors to suspect a

differential response in safety or efficacy, nor do

we have data to say that there is not, and we are

not going to get that out of this data set because

of the realities of the disease and what the

sponsor entered. So, just to be very clear about

the point, we are not indicting the sponsor's

attempt to enroll in this instance. It is really

reflective of the population but it leaves us with

a paucity of data.

DR. DYKEWICZ: Dr. Stoller, do you have a



DR. DYKEWICZ: Ms. Schell?

MS. SCHELL: I am just curious if there

are any studies in progress that look at the oral

steroids. Many of our patients are on oral

steroids for their emphysema, and is there any

looking at substituting or decreasing the oral

while implementing inhaled so that you can decrease

the dose of the oral but still not give as much by

benefit of the inhaled? I just wonder if there is

anything out there.

DR. DYKEWICZ: No? I will allow Dr.

Donohue and then any further questions should be

directed to the FDA before moving to the open

public hearing.

DR. DONOHUE: We don't have a specific

answer to that but three of the authors of this

paper from Chest are in the audience, and if you

look at what people are taking as concurrent

medicine in a very, very large data set from

Boehringer Ingelheim of over 3000 patients from the

late 1980s to 1995, there was a reduction in oral

steroids from 30 percent concurrent to 16 percent.

Also, very important to our discussions, a

reduction of oral theophylline, one of the most

toxic drugs that we deal with, and concomitant to

that there was an increase in inhaled steroids from

15 to, I think, in the 40 percent range and, of

course, there was more widespread use of

bronchodilators such as anticholinergics in that

study. So, that is the kind of data. But

certainly the gestalt I have as a doctor in

treating these patients every day is that the

switch to the inhaled corticosteroids does, indeed,

enable us to lower the exposure to oral steroids.

Thank you.

DR. DYKEWICZ: Thank you. Any further

questions directed to the FDA? Dr. Bone?

DR. BONE: I have a question that could be

answered by either the agency or the company. That

has to do with some comments that were made back

and forth about the femoral densities in 3017 and

the spine densities. I don't know if the company

would have a slide they could show. What I would

really like to see is all the bone density data

from those studies displayed so we can really look

at it kind of all at one time.

DR. LEE: Yes, I have a slide. Let's see,

it should be slide 93 of my presentation. This is

not all the data as far as all the time points but

these are endpoint measurements.


DR. MEYER: These are the data for 3017.

To be very clear about this, the primary interest

and the way they validated their look was in the

lumbar spine where, actually, the effect

numerically is favorable for fluticasone. The

femoral neck data was not as well validated, and

was not a primary place of interest. That is

really what you can say about it. I mean that is

where there seems to be a numerical effect of more

concern, but that was not a priori where they were

intending to look and I think we need to be clear

on that.

DR. SHAH: As Dr. Meyer indicated, the

femoral neck was prospectively QA'd and so the

reliance on the validity of this data is

questionable. But I think there is another factor

that we have to consider in interpreting these

data, that there was a substantial dropout effect

that occurred over the course of the two years in

these studies.


Indeed, if you look at the results, what

you are presented there was really the completer

analysis, which is the patients who actually

completed two years of therapy. Certainly in FP

groups about half the patients at the high dose had

withdrawn from the study. Obviously, this raises

the question, well, how do you try to adjust for

that but if you try to do a regression analysis of

the slopes to look at the bone density change over

time for the different patients in these studies --


-- what you see in this slide is the

results from lumbar spine looking at that. The

yellow lines are showing the completers, which is

what we just reviewed for lumbar spine. On the

left is 3001 and on the right is 3017. As you see

in the yellow, for lumbar spine we didn't see any

evidence of a change with treatment on lumbar spine

for both the doses studied versus placebo.

In the green is shown the regression

analysis which is adjusting for the dropout effect.

When you look at the regression analysis, again, in

this study there does not appear to be a treatment

effect on this measure.


If you look at the femoral neck data from

those two studies, the yellow is again the

completer analysis, which is all patients who

completed only, both on the left and the right, and

clearly you would conclude that maybe there is a

suggestion of an effect of treatment. Again,

remembering the caveat that these are not

prospectively QA'd, but if you do adjust for

dropouts in these two studies, you clearly see that

on the green there are no essential differences on

these measures in these patients in this study.

So, I think the evidence from these

studies is fairly reassuring that even in the

femoral neck region clearly there wasn't a great

suggestion of an effect when you adjust for the

confounder of dropouts.

DR. BONE: I would like to ask one or two

more methodologic questions. I do a lot of

osteoporosis trials and have designed a lot of

them, and I am not sure what you mean by

prospectively QA'd exactly. First of all, who did

this and did you have a central --

DR. SHAH: We did. It was the Oregon

Osteoporosis Center.

DR. BONE: That is an excellent

laboratory. Now, did they do central reading?

DR. SHAH: Yes.

DR. BONE: On all the sites?

DR. SHAH: They did.

DR. BONE: What do you mean by prospective

QA? I am not quite sure what that means.

DR. SHAH: Personally, I am not an expert

in that either, but my understanding is that when

they do the comparisons of the results versus -- I

believe there is a standard involved that they are

making comparisons with. They only did that

prospectively for the sites of lumbar spine and not

for those other regions. Hence, those data are

actually retrospective. They then went back and

they looked at the results and tried to make those

comparisons. I don't have a good expertise in this

area either so I am not a hundred percent certain,

but that is how these data were designed and how

those analyses were reported.

DR. BONE: Just a comment for the

committee's benefit. It is getting a little

obscure. It is almost as obscure as FEV1's and

things like that.


There is a lot more to measuring bone

density than it looks like there ought to be when

you just look at people doing it. I guess there

may be some technique involved in spirometry. I

don't know. And, one of the things that is very

helpful in doing this kind of study is to have a

reliable central laboratory. The one that they

used is the best, or at least among the very best.

The issue with the femoral neck as a

specific site is the most susceptible of the major

sites that we follow to positional changes. The

first thing you need to know about bone density is

we are not measuring bone density; we are measuring

the amount of bone mineral over a projected area.

So, that is why the results are in milligrams per

squared centimeter. It is not a volumetric

density. It is the amount of bone mineral measured

over a projected picture of the bone. So, if there

is rotation in an out of the plane there will be a

tendency -- any foreshortening will cause an

apparent increase in density, and a reduction in

foreshortening will cause a decrease in the

measured density. So, if the femoral neck is not

exactly as flat each time, if there is rotation in

and out of the plane of measurement there is going

to be a change in the measurement. The total

proximal femur measurement is less susceptible to

this and that is why many trials use this. It is

not as sensitive in terms of the magnitude of the

change or the percentage of the change but it is

more precisely reproducible in most cases.

So, what I suspect if there was a quality

assurance issue here, it may have been as much

related to how the technicians were instructed to

position hips and whether this could have changed

over the study, or something like that. It may not

just be the instrument calibration kind of issue

with the phantoms that were referred to. But that

is very important. Each of these sites has its


The spine is usually the most sensitive

change because bone remodeling is a surface

phenomenon, for the most part. There is a lot of

surface in cancellous or trabecular bone so you see

bigger changes in bone remodeling when there is a

lot of surface on the trabecular plate that can

remodel, whereas cortical bone has a smaller

percentage of the bone actually involved in

remodeling so it tends to change less. The one

area that is essentially of no use at all is Ward's

triangle because it is not an anatomically defined

area in bone densitometry; it is just the lowest

place that the machine can find on the image at the

time of the study. So, it doesn't mean the

anatomical Ward's triangle. So, we don't pay too

much attention to that.

Again, it is unusual to see a discrepancy

between sites of measurement if we are using the

major sites. They generally are all going in the

same direction, even if they are not all going in

the same direction at the same rate. When you see

something like we just saw where there was a fair

sized difference in the femoral neck and no change

at all in the spine, that raises the question that

Dr. Meyer raised about the possibility of a

technical issue being involved. Again, it doesn't

sound like we are going to resolve that.


DR. DYKEWICZ: Thank you. Moving now to

the open public hearing section, this gives an

opportunity for members of the public in the

audience to approach the microphone and express

concerns or make comments, with the limitation of

three minutes at the microphone. I do not see any

movement in the audience.

Being none, let's move to open discussion.

I thought the most efficient way of dealing with

the myriad of issues that are being presented to us

is to go through the outline that had been given to

us by Dr. Purucker. Specifically, we will first

discuss the patient population of the studies that

have been submitted for review, and the questions

are how representative these are of the COPD

population; questions, of course, about

reversibility; perhaps about the history of

bronchitis, chronic bronchitis that as a

requirement for study entry; then, leading to

whether these studies would, therefore, be

supportive of a broader indication for the use of

these agents in treatment of COPD. Who would like

to begin? Dr. Wise?

DR. WISE: I think the demographics are

relatively similar to the COPD population in the

United States, as I think has been pointed out.

The major way that this study group differs is that

they all have chronic bronchitis, defined in a

typical epidemiologic fashion. To that extent, it

is not not representative of all COPD. The

prevalence of chronic bronchitis defined this way

in a COPD population varies widely from study to

study but might represent anywhere from 30-70

percent of a general COPD population.

Nonetheless, I think that this population

does reflect what we are coming to describe as

chronic obstructive bronchitis, that is to say,

people with chronic bronchitis who have significant

airway flow limitation. To that extent, it is an

identifiable subpopulation of the COPD population.

DR. DYKEWICZ: Thank you. Other

discussion at this point? We have discussed a

number of aspects of this already. If it came to

the point then that the products were approved,

either the Flovent or the Advair, what

recommendations might there be in terms of the

indication that you could feel comfortable with?

For instance, I would say that you would want to

have some sort of a statement that this, if it were

approved, would be indicated for treated of

emphysema with chronic bronchitis, or something

that would indicate that phrase of chronic

bronchitis in the labeling. Dr. Wise?

DR. WISE: Yes, I would agree with that.

I think I would either use the terminology chronic

obstructive pulmonary disease with chronic

bronchitis or chronic obstructive bronchitis.

DR. DYKEWICZ: All right. If there are no

more comments on that point for discussion, let's

move on to discussion about the issues about the

primary endpoint, that, of course, being change

from baseline in FEV1. Do we think that this is

clinically relevant, the degrees of the changes

that we saw? Comments on that, please. From.


DR. FINK: Actually more a question than a

comment, for those people who are experts in COPD,

does COPD have a fairly linear decline, like cystic

fibrosis does, in FEV1 that can be used almost as a

surrogate for survivability, or is FEV1 related to

survival much more variable in COPD?

DR. DYKEWICZ: I recognize you.

DR. PAUWELS: Maybe I can answer that

question. When we actually looked at the

longitudinal changes in FEV1 in individuals in the

EUROSCOP study, which was a two-year study, what

you can't do is, on an individual basis, determine

the decline in slope. So, when you total the

population and you have a nice linear decline over

time like has been shown in the Lung Health Study,

that reflects quite a lot of individual variations.

So, it is almost impossible to predict for an

individual patient the change over time. What you

can say is that there is usually progressive

decline in FEV1 over time, but what is going to be

next year and the year after that is impossible to

predict for an individual.

DR. FINK: On average, how many ml/year is


DR. PAUWELS: Well, it varies from study

to study but, for example, in the EUROSCOP study

the mean decline over time was 60 ml/year, and that

is comparable to some of the data from the Lung

Health Study. So, it is between 50-60 ml in most

of the studies, which compares to, for example, 20

ml in non-smoking, non-COPD patients. That really

puts into perspective the type of changes that you

see here.

DR. FINK: I guess my reason for asking

that is then could the 120-180 ml seen in these

studies correlate to a two- to three-year change in

decline in FEV1 over time.

DR. PAUWELS: Sure. And, the figures that

you have seen here, if you compare them to what is

seen in other large pharmacological studies, are

quite impressive.

DR. DYKEWICZ: Other comments on this

point? Dr. Joad?

DR. JOAD: This is in answer to your

question about how do I feel about FEV1 as a

primary endpoint or the main endpoint in these

studies. Since I don't think anyone is purporting

that this is a change in the underlying

pathophysiology or saying that this study shows

that, and since we are talking about the purpose

being for symptomatic control, to me, an FEV1 is

not sufficient as an indication of efficacy to show

that you have symptomatic control. You have to

show symptomatic control because that is why you

use it. It becomes an issue of whether they showed

symptomatic control; that is a different issue.

But FEV1 all by itself would not be sufficient in

my mind.

DR. DYKEWICZ: Other comments on that?

Dr. Fink?

DR. FINK: I guess I was just going to

comment that I would find FEV1 an adequate marker

that is commonly used in lung trials to show

response. In my mind, it clearly casts the issue

that we are going to end up discussing as to the

risk-benefit ratio because I think there is a

change in FEV1 that is reproducible among the

trials, and depending upon how you want to look at

it, at least is of some significance in COPD but

whether it justifies the risk of therapy becomes, I

think, a much grizzlier question.

DR. DYKEWICZ: Dr. Parsons?

DR. PARSONS: Although there is the

perception that Dr. Donohue mentioned of physicians

that are currently using inhaled corticosteroids in

their patients that they, quote, do allow that and

I think you inferred that they have fewer

exacerbations, I am concerned that with the change

in FEV1 we saw there wasn't an apparent really

stunning correlation toward improvement in

symptoms, although this is supposedly physician

perception. Usually when physicians perceive that

their patients are that much better, you would like

to think you could get a clear objective measure of

that. So, I too am concerned that the FEV1

increase exists but I am surprised that we didn't

see bigger clinical correlates.

DR. DYKEWICZ: One of my comments would be

that we are being charged, from a regulatory

standpoint, with giving some advice relative to

evidentiary standards that are asking for

substantial evidence of efficacy and safety. Of

course, on one hand, the use of the term

substantial would mean that we are seeing

relatively great changes in whatever measure we may

be using, but I think it also is appropriate to

consider this in the context of the disease state

and what would be plausible as a substantial

improvement in whatever parameter in the context of


I would almost defer to some of the

pulmonologists on the committee to try to address

that. It seems to me that, in contrast to asthma

with which I do have much more experience, we are

looking at far less dramatic changes here in the

FEV1 and if this were in the context of asthma

treatment these improvements in FEV1 I would

certainly not think would be very substantial. But

do any of the pulmonologists on our committee have

comments about that? Dr. Stoller?

DR. STOLLER: I will try to lend my

perspective on this. I have been trying to kind of

forge my thoughts around to what my way of thinking

is kind of the key relationship between the

magnitude of efficacy and the magnitude of safety

and the level of evidence which supports safety

which, to my way of thinking, is kind of where this

issue turns. But to your question, I would say, as

we heard and has already been said, certainly an

FEV1 rise of 100 ml, particularly if it were

present in patients on the lower end of the

obstructive spectrum -- you know, it matters less

in a patient whose FEV1 is 1.7 L than it would in a

patient whose FEV1 is 0.7, which gets to my prior

question about examining dichotomous outcomes in

strata which would be informative to my comments.

That not withstanding, 100 ml is, on the one hand,

an impressive absolute increment, at least

potentially, in some subset of those individuals

and certainly, as has been nicely stated,

comparable to data from prior studies, Lung Health

Study, ISOLDE, EUROSCOP, in terms of the magnitude

of change that one might see with inhaled steroids

of various kinds.

On the other hand, I think it is a leap of

faith to say that that rise in FEV1 translates --

even though we recognize that patients

characteristically whose FEV1's fall below a liter

have a 50 percent mortality risk. They die

usually, if you believe the support data of acute

respiratory data, often with ICU hospitalizations

and acute respiratory failure. Their physiologic

reserve whittles away and they reach a critical

point with the slightest insult that translates

into a major clinical event that you and I wouldn't

experience with better pulmonary reserves. So, it

is not as though they sort of dribble off the

court, if you will, they reach a physiologic

threshold at which slight insults translate into

major clinical events and then they are in this

terrible vicious cycle of recurrent exacerbations,

hospitalizations, acute respiratory failure, etc.,

which I think in most of our experience would be a

characteristic kind of description of the natural


So, it is a leap of faith in my view to

say that a 100 ml rise translates into benefit over

the longer term in terms of what impact it has on

the natural history. Of course, we are not

demanding that level of evidence here but, of

course, it should be said that there is no evidence

from the literature that inhaled steroids change

the rate of change of slope. It is not pertinent

to this data set, but there are now three or four

randomized trials that have examined this very

carefully in large numbers of individuals over a

longer period of time than this would show, with no

change in the slope. So, it is somewhat akin to

the experience with ipratropium which showed an

increment, a one-time increment and that benefit

pervades over time but there is no change in the

rate of decline of lung function.

That is a long-winded way of saying, yes,

100 ml is impressive but I am not sure how much

clinical impact to give to it. With that said, as

my colleagues have mentioned, I am concerned and

the question then defaults to does that translate

into other clinically meaningful events in these

patients' lives? One would hope that that would

translate into exacerbation frequency, as was

demonstrated in ISOLDE. There are always

methodologic problems about defining exacerbations.

In ISOLDE they were physician defined. As I

recall, there were no objective criteria, perhaps

the best of which come from Anthonisen and his data

from 1987, stratified -- mild, moderate and severe

by constellation of symptoms, dyspnea, purulence

and volume of phlegm. Even ISOLDE didn't have

those data, as I recall, and we don't have that

here in terms of strata of exacerbations. So, I am

concerned that there isn't that connectedness

between a physiologic increment and symptomatic

increment by the best validated

Guyatt score or the Paul Jones St. George's, which

was not looked at here which I think would be

regarded to be the best disease-specific quality of

life measures. It is a very valuable instrument

but it has not been looked at as carefully as the

others with regard to meaningfully significant

change. So, one has to view with some

circumspection what a transitional dyspnea score

means as the isolated secondary symptomatic benefit

in the absence of demonstrable changes in the

other, if you will, better validated quality of

life indices.

I am not sure I am answering your question

but I am having a turmoil explicitly around what it

means. I would feel much better and much more

comfortable -- and maybe it is outside the context

of this discussion -- if there were, for example,

evidence of survival benefit. That would be a home

run. That would be an absolute home run if this,

as one might imagine from the comments being made,

translated into a two-year survival benefit as you

might imagine it could given the rate of decline of

FEV1 of 50-60 ml/per year in a patient non-smoking

with COPD. That would be impressive. But as it

stands, with the data before us, it is a little bit

more difficult to kind of embrace; let me put it

that way.

DR. DYKEWICZ: Thank you. Dr. Meyer?

DR. MEYER: I just actually want to make a

point that follows up on Dr. Stoller's and Dr.

Dykewicz' question and comments. That gets back to

sort of the regulatory standard we work under at

the FDA. The other piece to having that

substantial evidence is for the purported effect.

There was a reference made by the sponsor earlier,

correctly so, that both Serevent and ipratropium

for treatment of COPD also did not show much effect

on some of these secondary endpoints. But I would

point out to the committee that they have an

indication for the relief of bronchospasm

associated with COPD.

That is really doing two things. It is

telling you it works as a bronchodilator -- that

those agents work as bronchodilators, and also

maybe a little bit in a less than overt manner

restricts the population in those patients with

COPD who experience bronchospasm who benefit then

from either salmeterol or ipratropium in the way

they are labeled.

The label claim that is being proposed

here is the maintenance treatment of COPD. So, in

comments, one of the things that I would be very

much wanting folks to advise us on, if you do come

to approval, is that the way you want to describe

the purported effect? Is that what the data is

giving you, maintenance treatment of COPD?

DR. DYKEWICZ: Or one might conceive of

maintenance improvement in lung function. That is

just one possible thought. I don't know. It is

open discussion so, Dr. Donohue, I will let you

make a comment.

DR. DONOHUE: I would like to make a

comment on the robustness of these changes of 100

ml and, I believe, 164 ml in the pre-dose. This

clinically is the trough. This is what patients

have when they get up in the morning. And, I am an

old enough lung doctor to have grown up in the

period when we used to do surgery in the afternoon

because our patients were so impaired in the

morning when they would awaken. Their lung

function would be very low. The six-hour

bronchodilators would be gone; the theophylline

would be gone, and what-have-you. So, any

improvement in pre-dose or trough value has really

a great benefit to our patients.

In terms of the clinical trials that I

have experience with and been involved with,

particularly the Advair, 161 or 164 is extremely

robust for an inhaled, other than a solution device

is pretty good compared to anything I know of.

Let me come back to the 100 ml. Again, we

really can't define what is a meaningful change in

terms of that FEV1, unfortunately. I wish we

could. But it is important to remember that in

COPD you need every improvement you can get.

Furthermore, as opposed to asthma -- this is a very

important distinction -- narrowness of the airway,

the geometric area of the airway correlates with

bronchial hyper-responsiveness. So, that

improvement of 100 ml -- maybe it is not the

greatest thing in the world; it is not the 400 or

500 that we saw with Advair in asthma, but we can't

really get that in COPD. And, 100 can correlate --

maybe Bob would clarify this; he has also studied

bronchial hyper-responsiveness, and that is an

important number. It may translate into some

benefits in our patient's life.

I would just like you to realize that

these magnitude of changes are always small in

COPD, yet, they can have some benefits to our

patients and sometimes it takes a long time to see

the effect of that. Thank you.

DR. DYKEWICZ: Thank you. Dr. Wise?

DR. WISE: I think I have a couple of

points to make about the magnitude of the effect

and how to interpret it. First of all, I think

part of the struggle is that there is a very linear

relationship between physiologic effects and

impairment or symptoms, if you will. If you have

someone on the low end, a very small increment in

FEV1 can make the difference between them being on

a ventilator or not. As the lung function gets

better, then small increments mean very little.

So, I think this speaks to Dr. Stoller's point that

it would be useful to know what these distributions

are like based on their baseline lung function.

The second thing is that in terms of

interpreting a substantial effect, it seems to me

that there is a subtlety there in substantial

evidence of benefit, if that is the right

terminology. You can have evidence that you are

very clear on statistically, that you are very sure

about, but which may not be substantial in

magnitude or, oppositely, you might have evidence

that you are not so sure about but which appears to

have a large effect. And, I think we need to keep

that in mind. I am not sure how that term of law

is interpreted, in other words, whether substantial

evidence of benefit is substantial evidence and

that you are very clear about it, or that there has

to be evidence of substantial benefit.

The third point -- and I made this with

respect to the CRQ or the quality of life

questionnaire, you really need to look, I think, at

the distribution of those changes to interpret it.

In other words, if you have a very narrow

distribution, let's say, of FEV1 and for a

particular treatment there is only a 50 ml change,

that might be trivial if everybody in the

population just shifted 50 ml. On the other hand,

if you have a whole wide range, perhaps with even

skewing toward the low end, and you shift that

population 50 ml you can have a tremendous benefit

in terms of the ends of that curve. The whole

magnitude of the effect of cigarette smoking, for

example, is only about 30 ml in a population but

nobody says, well, cigarette smoking has a trivial

effect. It has its effect out in the ends of the


So, one of the things that I think we want

to examine is that how distribution is shifted,

both for the FEV1 and some of the symptomatic


DR. DYKEWICZ: Dr. Shah, if you would like

to respond?

DR. SHAH: Yes, there are several

questions that are being asked and maybe we can

provide some data that can help you interpret these

in the context of the effects, first of all,

exacerbations. As I indicated, these studies

really weren't designed and optimized to look at

that endpoint. Hence, from our perspective, you

know, we were not that surprised that the

differences don't appear at first glance to be

suggestive of treatment effect.

But as I indicated, we did prospectively

define to look at what we would define as COPD

related conditions, which were worsening of COPD

related to multiple reasons, including, as I

indicated, some patients who were withdrawn for a

definitional lack of efficacy. In retrospect, I

think we should have been much more specific in our

instructions to investigators about what would have

constituted lack of efficacy, but that was not done

in this program.


But when you look at the data in

aggregate, let me show you the slide on response we

see over time, as I said, the COPD related

conditions, which is the withdrawal due to this

condition, included these reasons -- people who had

COPD exacerbations; people with adverse events

related to COPD exacerbations; people who withdrew

for lack of efficacy, which clearly would be a

marker of control of COPD; and then adverse events

of respiratory nature which required antibiotics or

corticosteroids. Essentially, people who were

withdrawn for what would be considered as an

exacerbation were recorded as potentially an

adverse event rather than an exacerbation. This

was prospectively defined so this was not a data

dredging type exercise. We had defined the

subgroup to look at this response in the studies.


What I am going to show you in this slide

are the actual results. This is the survival

Kaplan-Meier over time across the treatment groups.

Indeed, what you see here is that the placebo group

had the greatest proportion of patients who

withdrew for these various reasons, which we would

consider as respiratory related in this program,

versus the individual treatment groups. Indeed,

for all treatment comparisons with placebo in this

post hoc -- and I appreciate you can't draw

inferential conclusions from this but the p values

speak for themselves and they were less than 0.05,

indicating that there was evidence of treatment

effect on control of COPD comprised of

exacerbations and withdrawals for those reasons.

But I think in the study there is evidence

of effect but, again, I want to remind the

committee that we didn't design these studies to

look at these various reasons and I think it isn't

surprising that when you don't prospectively design

studies for a certain measure you don't always get

the results that you expect and hope top see.

The other endpoint and measure that people

are very interested in is quality of life. If the

committee will indulge me, let me invite our

pharmacoeconomic expert, who has done a lot of work

in quality of life, who has looked at this area a

great deal and who maybe can review some data from

the studies and give you some perspective on

actually the questions about response rates as


DR. EDIN: I am Heather Edin, and I am

with the Global Health Outcome Group at Glaxo.


These are the data integrated over all

three studies. What we are looking at is the

overall change from baseline in the CRDQ score. As

you can see, subjects who were treated with Advair

500 and Advair 250 reached the clinically

significant threshold of 10. So, they had

clinically meaningful improvements in their overall

COPD-related quality of life. We also see

improvements in the FP 250 and FP500 groups, as

well as the salmeterol groups related to the

combination of the two products to really result in

a clinically meaningful change.


We have all discussed the clinically

meaningful change, and Dr. Meyer addressed these

very well. The meaningful change was really

developed and anchored as a change from baseline

with an individual patient and that is what it has

been validated for, and has been published in the

literature. The ten-point change has never been

validated officially or published for a within

treatment group comparison. We did define it a

priori for absence or anything better to use or

different to use, and because we were required to

state something up front. However, the ten-point

change is valid, and we should consider the fact

that because differences between groups might not

reach a threshold or a particular level, even the

small benefits could be important, which Dr.

Donohue pointed out in his presentation.


A few of the reasons why we have a smaller

magnitude of difference from placebo, if you will

-- one of the reasons is that, again, this is COPD.

We are not going to see the magnitude of difference

between groups that we see in asthma. These

patients are older and they have a significant

smoking history.

Other reasons are that the placebo groups

in these trials received albuterol and that is

definitely going to affect their quality of life

scores. One of the reasons is that because two of

the domains in the CRDQ are mastery over disease

and emotional function. All patients received

additional education and information about

management of their disease, and that is going to

cause these two levels to increase, even in the

placebo group and, therefore, there is going to be

a smaller difference when you subtract out.

Another issues is that, again, these

trials were not designed around quality of life.

There were no inclusion or exclusion criteria on

quality of life scores. So, patients, regardless

of their quality of life score or quality of life

impairment, were included in this trial.


Again, coming back to the integrated data,

we are seeing significant improvements in both

overall score as well as individual domains. One

thing I want to highlight is that on dyspnea this

is one area where we see significantly greater

improvements in the combination Advair than

placebo, but also statistically significant

compared to the individual components. I think

that is an important thing to link back to our

improvements in FEV1 and symptomatology.

If we had designed these trials to look

specifically at quality of life, we might look at

patients who had lower scores. One way to do that,

we post hoc, just recently, looked at data with

patients who had a median score less than 84.


I want to point out on this slide, this is

a difference from placebo. So, not only were

patients demonstrating a significantly greater

improvement in their quality of life that was a

clinically meaningful change from baseline, but

when we subtract out the placebo group and this

group of patients with a CRDQ score of less than

87, which was the median value, we are coming close

to approaching that ten that we defined a priori.

So, depending on the population that we

look at we see greater or lesser changes, but the

point is that even small magnitude of change in a

COPD population can be clinically meaningful.

One other point that was mentioned about

looking at responder analyses, which I don't think

we have a slide for but we did look at responder

analyses in each of the individual studies, and I

believe that all of the active treatment groups

were significantly greater than placebo when we

compared the proportions of patients who achieved a

clinically meaningful change.

If my memory serves me correctly, it was

approximately 50 percent of patients on Advair who

had a clinically meaningful improvement, versus

approximately 40 percent on fluticasone or

salmeterol, versus approximately 30 percent on

placebo. We did put this integrated data into the

ISC which was presented to the agency in our

document, however, we did not do any differential

analyses on the data so we don't have statistical

values for those comparisons.

DR. DYKEWICZ: Thank you. Any other

comments on that point? If not, let's move on to

discussion of safety, and specifically the issues

about fluticasone and its systemic availability.

The charge that was given to us by the FDA was to

consider the impact that fluticasone would have on

the HPA axis in these preparations and other

systemic effects that might go along with that.

So, would anybody like to make some comments about


DR. MALOZOWSKI: I have a question for the

sponsor. What was the detection limit of the assay

where you measured cortisol?

DR. KUNKA: Hello. My name is Bob Kunka,

from the clinical pharmacology department. I

believe it was a value of 50 but I will

double-check that.

DR. MALOZOWSKI: If you are using the

lower level of 4.5, I think that you are probably

misquoting 50. This question is relevant to

understand the data that the company presented

stating that there was not an effect on basal

cortisol, etc. I saw the data. They may be

absolutely right but I don't know what the lower

limit of detection of the assay is.


DR. SHAH: We will have to look for that.

I am sorry, we don't have that at the tip of our

fingers. But I will say that these were done by

central laboratory so these were standard reference

ranges that were used and methods. It wasn't done

in-house or anything of that type. So, this is

very common analysis but we will see if we can find

those for you.

DR. MALOZOWSKI: I am asking this because

you mentioned that you measured with HPLC and

although I know that this method is extremely

accurate, I don't know really what the range of the

data is.


DR. JOAD: I have a question for you, Dr.

Malozowski. What would be a clinical correlate of

an impaired HP axis without it being totally

depressed? What would they look for clinically to

see if it were meaningful, this 21 percent decrease

or, in the other study, 50 percent decrease in


DR. MALOZOWSKI: I think this is a very

difficult question to answer but I will try to

elaborate by saying the following: I think if we

accept statistical differences as valid when we

look at efficacy endpoints, by the same token, when

we look at safety data, if they are statistically

different we cannot dismiss them. This is one

point that I would like to make. I think it is

very important.

Secondly, I don't think that this

particular test, although probably this was a test

that was used in the mid-90's or the late 90's, is

an adequate test to address adrenal insufficiency

because the dose, if I am not mistaken, was 250 mcg

for this test. There is a test that is using 1

mcg, just to give you an idea of the magnitude of

the dose that you can use. Therefore, a dose of

250 is a massive dose and it is very difficult to

assess adrenal insufficiency. You would have to

have a patient that really is impaired. This is

one point.

Second, I think that there is a problem in

the way we look at patients receiving exogenous

glucocorticoids, that we are looking at Addison's

disease when, in earnest, what these patients have

is hypercortisolemia and all the data seeing the

trends in lymphocytosis eosinophilia, etc., etc.,

suggest that these drugs are absorbed. I think the

PK and the PD data talk volumes as to the fact that

these drugs are absorbed. Probably there is a

difference between the patients with asthma and

COPD. I am not familiar with the rate but, if I am

not mistaken, in data presented in this advisory

committee in 1998 by this company in patients

treated for asthma, children, 50, 100 and 150, all

the doses showed some level of impairment in growth

velocity with the dose of 50 being less impairing

growth rates in children. Clearly, the dose of 100

and 150 twice a day had this effect. Therefore, I

think it would be fair to say that a dose of 500

mcg twice a day or 250 twice a day will have


This also has to be put into the

perspective of the condition that you are treating

because if there is no other way to treat these

patients, the agency may be willing to undertake

the risk or to label this properly to make people

aware that this is happening. So, I don't see this

is a great impediment. I see that there is

absorption. I think the company agrees that there

is absorption. I think the PK data show that there

is absorption. The question is how you have to

deal with the fact that this drug is absorbed, and

how long will it take for it to lead to other

complications, and whether we are willing to

tolerate these complications in order to take

advantage of the benefit that the drug is offering



DR. FINK: One thing that bothers me about

the safety profile, if my logic is correct and,

please, someone correct me if I am wrong, is that

we have heard today that in severity of COPD, the

more severe the pulmonary impairment, the less the

systemic bioavailability of the inhaled drug at any

given dose. We have heard that the more severe the

pulmonary function, the shorter the life

expectancy. So, the logical conclusion is the

better your lung function when exposed to these

agents, the longer you will be taking and the

greater your risk for suffering an adverse event,

i.e., the safety ratio is worse for the healthier

patient than it is for the sickest patient.

DR. DYKEWICZ: Dr. Meyer?

DR. MEYER: I think the one part of that

logic that I want to speak to is what we know, at

least what we as the agency knows and the sponsor

may have a comment on this, about the relationship

between lung function and systemic bioavailability.

We do know both in asthma patients and in COPD

patients that the systemic bioavailability of

fluticasone, which is almost entirely through the

lung, is lower than in normals. I don't know that

we know that there is such a tight relationship

between lung function, however. We know that in

aggregate in people with impaired lung function it

appears to be lower than in normals but I am not

sure we know anything about the tight relationship

below that.

DR. FINK: But it would still be then that

the healthier patients would face a greater risk of

adverse effects from a longer period of exposure to

the drug.

DR. MEYER: Yes, it was just that one

piece I wanted to clarify. I don't know whether

GlaxoSmithKline has better data that look at strata

of lung function in relationship to bioavailability

through the inhaled route.

DR. YATES: I am Peter Yates, from

clinical pharmacology. We have looked at systemic

exposure both in asthmatics and in COPD patients

and we compared asthmatics and COPD patients in a

formal way. So, we have compared healthy subjects

and asthmatics formally and we have compared COPD

subjects and healthy subjects formally. We found

that in both cases about half of the exposure in

healthy subjects either matched asthmatics or

matched COPD subjects. But there doesn't seem to

be a sort of continuous relationship between lung

function as measured by FEV1 and systemic exposure.


DR. BONE: Just to pursue this question

about the effects of the absorbed steroid on the

hypothalamic pituitary adrenal axis, and to build

on Dr. Malozowski's comments and some of the

comments that were made in the review, recall the

sequences that the hypothalamus releases

corticosteroid releasing factor which stimulates

the pituitary gland to secrete adrenocorticotropic

hormone which then stimulates the adrenal gland to

secrete cortisol. The reduction in the AUC for

cortisol implies that there has been a reduction in

the perceived need for steroid by the hypothalamus,

and that there can also be some feedback on the

pituitary gland so that if you take those two

together as a unit, there is the central perception

of a reduced requirement for glucocorticoid steroid

and decreased stimulation to the adrenal gland.

This has amounted to a 10 or 20 percent reduction

in the amount of cortisol secreted in the study

where the AUCs were measured. Just to be clear, I

don't think that any of us regard this as evidence

of adrenal suppression at all. Right?

This is not suppression of the adrenal

gland at all. This is a suppression of the

regulatory system by partial replacement of

glucocorticosteroids. This is also not a degree of

suppression that any endocrinologist would expect

to result in the kind of incapacitation of the

hypothalamic, pituitary and adrenal axis that we

would see in a patient who was on higher dose,

long-term glucocorticosteroids and then would be,

in effect, unable to respond without exogenous

steroids under stress. So, there is no evidence of

any of that kind of deficiency, just as Dr.

Malozowski has suggested.

The question here is only one, I think, of

excess glucocorticosteroid effect is a problem.

Now, remember too that we did see some reduction in

the amount of endogenous glucocorticosteroid

secreted. In other words, there is some internal

compensation. It is only when the

glucocortico-steroid administered is greater in

amount than the reduction in glucocorticosteroid

produced endogenously that you would get an

incremental change in total glucocorticosteroid

exposure. There may very well be some altered

biological effects in terms of the timing of

secretion. I would defer to Saul Malozowski on

this subject because he is more knowledgeable than

I am, I am sure.

But the point here is that the question to

be sharply focused upon is whether there is a net

increment of glucocorticosteroid exposure that has

adverse effects when you look at the combined

effect of the exogenous steroid and the reduced but

persistent endogenous production. That is what we

are talking about. We are not talking about

causing hypoadrenalism or anything like that.

Would you agree, Saul?

DR. MALOZOWSKI: I would. I have a caveat

only, this is also mean data. We have not seen the

distribution, and there is plenty of difference in

severity among patients. Some patients will have

much more than 20 percent and some patients may

have 50 percent impairment, and maybe these

patients are at risk but I didn't have a chance to

look at these data.

DR. BONE: This is one point. Another of

several points here where categorical data in terms

of response rate of possibly adverse experience

rate would be more informative than mean data, and

I would suggest that it may be worthwhile for both

the company and the agency to recapitulate their

analyses looking in that way. In other words, were

there an appreciable number of patients with

profound suppression of the pituitary adrenaline

axis even if the mean effect was very small, which

is what was recorded. Similarly, what were the

rates of response on other topics.

I don't have anything further to say about

that. When we get to bone I will have quite a bit

more to say.

DR. DYKEWICZ: Dr. Parsons?

DR. PARSONS: I have a question for maybe

Dr. Bone or Dr. Malozowski. Serum cortisol

measures were done once at four weeks. Is that a

measurement that you anticipate would be stable

over years if the patient stayed on the same dose,

or is this something that is going to continue to

drop? In other words, I am not sure what the time

frame is.

DR. MALOZOWSKI: There is the idea that

the more time you are exposed, the more probably

you will be affected but we are not sure. Also, I

don't know the PK/PD changes with time. I was not

privy to that information either. The fact that

the patients have stable respiratory function

during the 24 weeks of the study, I don't know

whether they do correlate and the level of

absorption is the same or not. I didn't see the

data; I cannot comment. But in general I think

that time is one of the main variables when you

look at induction of hypoadrenalism. You are

familiar with this in treating patients with either

oral or inhaled corticosteroids, the more the

chances are that a complication may happen.

DR. BONE: Agreeing with Dr. Malozowski

but just to add that probably four weeks is

reasonable for this type of study, and that the

issue about the duration of effect has at least as

much to do with how persistent the suppression

might be. Our experience with this question of how

long you have to be on exogenous

glucocorticosteroids to have a persistent

suppressive effect is entirely based on doses that

are more than, equal to or greater than normal

endogenous production, in other words fully

suppressive doses. If, in fact, a patient only has

a 10 or 20 percent reduction I wouldn't expect 8

weeks or 12 weeks to be any different from 4 weeks.

If you had a patient who had profound suppression,

then how long they were exposed would become more


DR. DYKEWICZ: Any further discussions

about HP axis? Glaxo comment?

DR. YATES: Yes, I want to show slides.


There is a question about the range of

values in the cortisol. This shows the individual

values. You can see that there isn't a great deal

of difference between the three groups here.


There is also a further point I wanted to

make. This really addresses the question about the

contribution to the endogenous corticosteroid pool

of the doses that we are considering here. That is

FP 250 and 500. So, the normal pool of endogenous

corticosteroid cortisol is about 10-20 mg/day. If

you convert, using the kinetic parameters that we

have for FP, those doses of fluticasone into

cortisol equivalent doses in terms of milligrams of

cortisol that is contributing to the endogenous

pool, only account for one or two milligrams of

additional cortisol. So, this would correspond

when you get to steady state to 12 and 22 percent

cortisol suppression, but this is calculated values

based on an established PK/PD model so it agrees

closely with the values that were found in the


So, to put that into context with, say,

oral prednisolone, you see that for various doses

you get correspondingly greater contributions to

the endogenous pool, such that if you are giving a

dose bigger than about 7.5 mg of prednisolone --

that should be per day, not BID actually -- you are

then contributing more than 50 percent to the

endogenous pool. So, clearly, when you get to much

higher doses, like 15 a day, you are getting the

majority of your corticosteroid requirements from

endogenous sources. Therefore, there is potential

to get adrenal insufficiency and fairly long-term

effects. But that is a big difference from the

sort of contributions we are getting with these

doses here.

DR. MALOZOWSKI: Let me comment on this.

If you look at the bottom left, with this dose of

prednisone you suppress a patient. Therefore,

despite the fact that you see green all over the

place, you should be aware that you can suppress a

patient with 5 mg of prednisone. Therefore, it is

very difficult to make inferences. I am not

arguing as to the validity of this data. With 10,

you have patients that if you treat with 5 mg, you

can render them Cushing's after two, three months

if the patients are sensitive enough.

On the other hand, this product -- and I

think that Dr. Purucker was very clear in the

presentation, she made comparisons with other

products in the class and this product, in

comparison with all the other products, is the one

that is the least damaging from the systemic point

of view. This product seems to be quite less

active, or rapidly metabolized, or have some kind

of characteristics that make it slightly different

from the others that were presented during Dr.

Purucker's presentation.


DR. YATES: I have just put another slide

up which has comparisons with other inhaled

corticosteroids, in addition to the comparisons

with prednisolone. You can see that for these

typical dose regimes, again, you get much bigger

contributions, not only because you have higher

bioavailability but also, for something like

triamcinolone, you have a lower clearance. You

also have a higher unbound fraction in circulation.

So, all these factors combine to give greater

systemic effects. In fact, that dose of

triamcinolone there is the top end of the dose that

was used in the Israel study in that report. So,

you can see these are much higher exposures than

you would expect to find with the doses that we are

considering today.

DR. DYKEWICZ: Thank you. Dr. Joad?

DR. JOAD: I just want to make sure I

understood what you said. Would it be fair to say

that your opinions are that although there is

documented systemic absorption, you do not

anticipate a clinically important consequence of

that systemic absorption?

DR. MALOZOWSKI: Not at all. I didn't

mean that. I mean that this will happen. Some

patients will have complications due to the

systemic effect of this drug.

DR. JOAD: Assuming it is a 20 percent

reduction in serum cortisol over 12 hours, how

would that hurt your health? Why is that bad for

your health?

DR. MALOZOWSKI: From looking at the data

that was available in the documents that I was

given, I saw only two patients that appeared to be

suppressed, and this is only a 24-week study and

these were two patients among 32, if I am not

mistaken. Whether these patients will have

complications, what kind of complications is very

difficult to predict. That is the main reason I

asked what was the detection limit of the assay.

The depiction you saw with the 50 or 100 limit is

not the same unit that you are presenting in the

document, therefore, I do not know how they


It is very difficult to predict I think,

and I think FDA presented one case of a patient

that developed Cushing's. I don't know how long

this patient was exposed to the medication.

Cushing's is a very rare syndrome. You don't

develop this in one week unless, you know, you are

susceptible. I don't know whether this patient was

also on oral steroids. I am not familiar with the


DR. BONE: Could I just add a couple of

comments to that? One, I am not sure how

convincing the evidence of suppression actually was

because the patients weren't actually rigorously

tested to see if they were suppressed.

DR. MALOZOWSKI: The patients had a value

of about 5 and when tested didn't change.

DR. BONE: Nobody rechecked the test and

made sure they got their dose. In other words,

these were not confirmed.

DR. MALOZOWSKI: Your point is well taken.

The same happened with the patients on placebo,

therefore, it is very difficult.

DR. BONE: I think we all know, as

endocrinologists, that none would accept a single

measurement as proof that somebody had adrenal

suppression. They would want to reproduce the test

and even do a more sophisticated test to make a

diagnosis. This is just information that is in the


I don't think either Dr. Malozowski or

myself would suggest that the patient who has an

average reduction in cortisol or average degree of

suppression, the 20 percent patient, would be

likely to have any toxic effect. I think the point

that he is making is that there may be individuals,

and that comes back to this question of going back

and looking at the data to see if there were a

significant number of individuals who exhibited

persuasive evidence that there was some harmful

effect. We don't have that information.

I think the reviewer's comment and Dr.

Lee's review was that the reduction in cortisol

AUCs, for example, was indicative that there was

absorption that was physiologically active, but

that it wasn't evidence in and of itself of a toxic

effect. Is that a correct statement?

DR. LEE: Yes.

DR. DYKEWICZ: Any further discussion on

HP axis? Glaxo?

DR. YATES: I just wanted to clarify this

point about the assay limit of detection. We have

no subjects where cortisol was undetectable --

DR. DYKEWICZ: Could you speak more into

the microphone?

DR. YATES: -- the assay limit was about

20 nm/l.

DR. DYKEWICZ: If there is no further

discussion on HPA axis issues, let's talk about

bone. Dr. Bone?

DR. BONE: Well, I think one of the very

important things that has occurred in the course of

this discussion was something that the FDA has

pointed out. I am not going to actually embarrass

any of my pulmonary colleagues by asking them how

frequently they obtain a bone mineral density

measurement on a patient just because they have

chronic obstructive pulmonary disease. Dr. Parsons

said all of them. This makes her unique amongst

all pulmonary physicians in my acquaintance.

The point is that essentially anyone with

chronic obstructive pulmonary disease that would

have met entrance criteria for this study or be

considered for taking any of these medications

should have a bone density test and meets criteria

for risk factors -- age, history of smoking, low

body mass, poor nutritional status and so on. We

know that the kinds of things that make people with

chronic obstructive pulmonary disease look old and

sick on the outside affect their skeletons as well.

So, the chances are that practically

everybody, as I said, with COPD or a very large

percentage of the patients will have at least

osteopenia, which means a low normal bone density,

from between 1 and 2 or 1 and 2.5 standard

deviations below the mean for healthy young adults,

whereas osteoporosis is variably defined as 2 or

2.5 standard deviations or more below the mean for

healthy young adults usually, but not always, using

race and sex specific controls. We also use

presence of a typical osteoporotic fracture as a

diagnostic criterion. So, if such a fracture has

occurred the bone density requirement would not be

essential to making a diagnosis. We would like to

make the diagnosis based on bone density.

The National Osteoporosis Foundation has

recommended that the intervention threshold, that

is, the level of bone density at which treatment

should be considered be raised, that is to say,

more people considered for treatment if additional

risk factors for osteoporosis are present, as would

be the case in many of the patients with COPD, if

not all.

So, we are looking at a population that is

probably significantly under-recognized and

under-treated for osteoporosis regardless of

whether or not there is a glucocorticosteroid

exposure. I guess the question here is in that

setting, how much difference would exposure to this

drug make in terms of their risk of osteoporosis or

progression of osteoporosis or their outcomes.

Now, one of the ironies is that just about the time

those of us in bone disease finally convinced

everyone who prescribes steroids to be extremely

careful about bones, we finally actually have a way

of testing for this problem and anticipating it,

and actually have drugs that have been approved by

the FDA for treating steroid-related osteoporosis,

as well as postmenopausal osteoporosis, as well as

osteoporosis in men.

So, we start off with a very high

prevalence of a condition that is easily diagnosed

and is treatable with available medications, never

mind whether somebody takes steroids or not. So,

we are starting off with that setting, and the big

issue that I think is an extension of Dr.

Purucker's remarks is that we ought to be paying a

lot more attention and doing a lot more about

osteoporosis in patients with this kind of

pulmonary disease, no matter what we do about

treating their lung disease. That is my little

commercial for skeletology.

The evidence that we have over a long

experience with adverse events with

glucocorticosteroids on bone is mainly based on

oral exposure, as you all know. There is

suppression of bone formation that is mainly due to

accelerated apoptosis of osteoblasts. There is

perhaps some preservation of osteoclasts and

increase in bone resorption for that reason. There

is certainly renal calcium wasting by the effect on

the kidney, and there is impaired intestinal

absorption of calcium due to an effect on the

calcium transport protein in the gut. There is an

effect on sex steroids in a number of patients.

So, we have a condition in which subclinical

secondary hypoparathyroidism can develop due to the

imbalance between absorption and renal

conservation. Then, we have all of these things

going on at the tissue level that can easily result

in a major negative balance in bone formation

versus resorption.

One of the things that has made many

people enthusiastic about the use of inhaled

glucocorticosteroids is that there is no doubt that

inhaled glucocorticosteroids represent a major

reduction in risk of this type of adverse effect

because systemic exposures in asthma, for example,

are a lot less than would be necessary with oral

treatment to achieve the same clinical end. So,

the first point is that all the effects we talked

about are diminished by reducing the amount of

systemic exposure.

I think the question here is what residual

effect is there? Since we have evidence that the

amount of systemic exposure is certainly not zero,

how do we estimate the magnitude of the risk of an

effect on the skeleton and how do we inform

patients and physicians so that they can take this

into account if that is necessary?

The first point here is that we don't have

direct data for the NDAs under review from the

major clinical trials. In other words, we don't

have the sort of data that the company is telling

us about from the long-term trial that they have

initiated. So, we have to make some judgment now,

which I think we should regard as temporary pending

getting more data.

Dr. Purucker and others have reviewed some

of the data that has been published mainly with

other glucocorticosteroids. Mostly triamcinolone I

think has been the one where the greatest attention

has been paid. The Lung Health Study, which was

reported in The New England Journal of Medicine,

did use triamcinolone, and that was quoted as

showing a decline in the femoral neck density in

comparison with the placebo group and a smaller

difference at the spine. This was a three-year

study with triamcinolone, and Dr. Malozowski and

others have already commented on the relative

magnitude of effect being expected, which may be

somewhat greater for triamcinolone.

I think this is certainly a piece of

information that we should take into account. We

should be a little careful since the methodology

was not actually included in the report at all,

only the data and just a table, and it did not

mention all of the anatomical sites which would

have been measured. So, I am not quite sure how

heavily to weigh that, but it did imply that there

was a difference of about 0.6 percent per year at

the femoral neck between the treatment and placebo


The study that was mentioned from Dr. Wang

is a little less clear about the steroid exposure.

This was a cross-sectional study from patients in

the U.K., and it does appear from this that there

was a statistically significant relationship

between the cumulative reported exposure to

glucocorticosteroids and bone density. While it

was highly statistically significant, the magnitude

was fairly small. It was just about the same as

the association with the history of calcium intake.

So, there were small effects, statistically

significant, in this cross-sectional study based on

patient reports; not a prospective study.

The other prospective study which was, at

least according to the report, more carefully done

with regard to the bone densitometry is the report

from Israel on a prospective trial of triamcinolone

in a moderate number of patients, which found a

very gradual effect. It was 0.0004 trams per

squared centimeter per puff per year, which comes

out to about 0.003 grams per squared centimeter if

you averaged 8 puffs of triamcinolone. They

predicted a one standard deviation change in 20

years in their analysis.

So, all of this information indicates that

at least for the other steroids that have been

discussed there may be a modest effect. The point

that the FDA has raised is how important is this

clinically in the setting of patients with a

considerable background risk, and what should we

say about that, if we are going to say something

about that, in labeling. I think that is a fair


This is just an opinion at this point, I

think we would have to say that the patient's

underlying situation is by far more important to

their osteoporosis risk than the risk that we can

infer from the indirect information that we have.

The information that we have about the compound

under discussion and the other data suggest that

the effect is small. If it exists, it is a small

effect. If there is a class effect, it is as I


Now, if patients have osteoporosis before

they get any such treatment, they ought to be

treated for it, and we know that the medications

that are appropriate for the treatment of

osteoporosis in postmenopausal women and

osteoporosis in men are efficacious and actually

increase bone mass in patients on oral

glucocorticosteroids at doses of 7.5 mg of

prednisone or higher on a continuing basis. They

actually will increase the bone density or maintain

it. So, it seems to me that if we identify the

patients with osteoporosis and take care of the

patients with osteoporosis, at least those patients

should be well protected against the much smaller

anticipated effect of the inhaled steroid.

So, the next question is what do we do

about people who don't have osteoporosis? Well, if

we measure the bone density in patients who are on

long-term inhaled steroids who are in the low

normal range, for example, and a decision is made

that they don't require treatment, periodic

monitoring of bone density by an extremely safe,

relatively inexpensive test should be very adequate

to intercept any problem that we would have on

those lines and institute appropriate therapy,

remembering that the reasons why such a patient

would develop osteoporosis would be multifactorial,

closely related to their underlying health problems

and only marginally influenced by the inhaled

glucocorticosteroid, perhaps some less than others.

Such an approach would be based on the

fact that these patients already have bone risk

factors and should be monitored anyway. So, I

guess what I am saying here is that I think if we

implement what we already know about risk factors

and management of osteoporosis we can I think

really marginalize the issue of inhaled

glucocorticosteroids, and in particular this one.

The recommendation I would make to the committee's

deliberations then is to make the decision based on

pulmonary issues and other safety concerns that the

committee might be concerned about, but that the

bone issue here, while it is extremely important

that it was raised by the agency, if it is

approached in the way I would think it should be in

general, the bone issue should not be dispositive

in the case of this particular NDA.

DR. DYKEWICZ: Dr. Meyer?

DR. MEYER: I wanted to just make a quick

point and then ask Dr. Bone a question. The first

point I just wanted to clarify is that I don't want

anybody to really take away from any of the

discussion that has gone on what the relative

effects or likely systemic effects for fluticasone

versus the various corticosteroids discussed here.

That has been well nailed down, and that

fluticasone is either good or bad in that regard to

other steroids. I am sure the sponsor has some

data and has shown some data. There are other

sponsors in this room that probably show other

data. And, that is highly dependent on whom you

are studying, when you are studying, what you are

studying. So, I think we really need to look at

this just for fluticasone alone. You know, I think

your points are well taken even if you just do


My question to you -- and we are very

helpful for your expertise here -- is do we know

enough about the effects of corticosteroids, be

they oral or however administered, to say whether

elderly patients are any more or less susceptible,

dose for dose, to the bone effects with

corticosteroids than younger patients? I

understand that the older patients, in particular

this population, is likely to have a lower baseline

bone mineral density, but do we know anything about

the pharmacokinetic response of elderly patients or

elderly cigarette smokers, and so on, relative to

the younger patients?

DR. BONE: First, thank you for your

comment about relative points here. I didn't mean

to draw a conclusion. What I was saying was if we

base our considerations on the largest effects that

have been reported with triamcinolone, then if

anything is better, that is all to the good. But I

think my remarks would apply to most of the data I

have reviewed. What I am saying, in short, is that

there may be a modest effect but it is in the

context of patients who ought to be managed for

this anyway.

The question of age-glucocorticosteroid

interaction is an interesting one. The problem is

we don't have any reason to think that elderly

patients would respond less well to intervention

for osteoporosis. We have done trials in patients

up into their 80s. Some of the patients who were

studied in glucocorticosteroid-induced osteoporosis

trials are older patients with polymyalgia

rheumatica, for example, and there didn't seem to

be any interaction with the condition or age in

that type of setting.

One of the murkier issues I think in the

pathogenesis of osteoporosis in unhealthy people is

the issue of growth factors, and specifically IgF1

for example, which are important for maintenance of

bone formation but, obviously, very susceptible to

effects of the patient's general health. I don't

know that we have looked in that specific context,

but I think that comes back to my original point

which was that the patient, as Dr. Purucker pointed

out and I think it is one of the essential points

of today's discussion -- the patient population

that we are talking about is at substantial risk

for osteoporosis irrespective of the

glucocorticosteroid exposure. So, the

glucocorticosteroid exposure by inhaled dosage may

have a marginal effect but we ought to be attending

to the problem and managing it irrespective of that


DR. DYKEWICZ: I believe we have two

comments from Glaxo. Will you identify yourself?

DR. ADACHI: Yes, I am Dr. Adachi. I have

run a number of the randomized clinical trials

looking at various therapies for steroid-induced

osteoporosis. Specifically with regard to the

question of do we see a difference in terms of bone

loss between premenopausal and postmenopausal

women, if we take a look at the placebo groups in

the study that we conducted with itidernate we saw

in fact, if anything, that premenopausal women lost

more bone than the postmenopausal women. If we

take a look at the studies with alendronate in

which we looked at prevention, again we saw that

premenopausal women lost equivalent amounts to what

we saw with postmenopausal women. I believe that

the same might be said of residenate trials.

The other issue about bone and bone

safety, when we take a look at the one randomized,

controlled trial that did report fractures we saw

that there were more fractures, in fact, in the

placebo group than there were in the fluticasone


Then, finally, when we take a look at

larger numbers looking at fractures, there is a

study by Van Staa and Cyrus Cooper looking at the

general practice research database, and when they

took a look at fractures in folks who were on

inhaled steroids they found that there was a slight

increase in fracture rate compared to normal

controls, however, when they compared them to

others who were on bronchodilators alone they found

that there wasn't any significant difference, and

they felt that the disease itself would be a

contributor to that risk for fracture.

Having said that, this is a cohort study.

It is not a randomized, controlled trial, but what

it does do is lend some further evidence to what

Dr. Bone was saying, that the patients that are

being described today tend to be sicker patients;

tend to be the ones that are at increased fracture


DR. MEYER: On the other hand, you also

might tend to not use inhaled corticosteroids or

oral corticosteroids in a patient you are very

worried about because, say, of a prior fracture. I

understand your points but I just want to say that

you can make arguments in those kinds of cohort

studies one way or the other. I think the data are

what they are, but they are hard to interpret.

DR. BONE: Could I just respond to the

question about what to do with a patient who has

had a prior fracture. I would manage their

osteoporosis and once that patient is on

appropriate osteoporotic therapy, with attention to

their calcium and vitamin D homeostasis, I think at

that point a judgment could be made about the risk

of adding a glucocorticosteroid. I think that most

of us would be much less concerned about the effect

of an inhaled glucocorticosteroid than an oral

glucocorticosteroid, but even the core of

glucocorticosteroids in that patient group, while

they would be expected to reduce the benefit of

treatment would be expected to still show some

increase in bone density on the average.

DR. DYKEWICZ: Dr. Parsons?

DR. PARSONS: I am still a little bit

confused in that if the average physician in the

United States still doesn't give aspirin or

beta-blockers post myocardial infarction, and those

are well established therapies, I think bone

densitometry as a measurement -- maybe we should be

doing it but it is not happening and we are not

identifying patients and treating them in the

general population. So, when a patient with COPD

gets put on an oral steroid for a long period of

time, I think with most physicians that clicks a

little bit and they think, oh, maybe I should worry

about bone densitometry and I should worry about

treatment of osteoporosis. But I am not sure right

now, based on our discussion, that I quite

understand if I am the same physician but I used

inhaled steroids in that patient is there an

additional risk of inhaled steroids no matter how I

am or am not treating the patient at baseline for

osteoporosis or complications from that therapy. I

am just trying to go into the real world and

generalizability. I think it is true that as

physicians as a whole we probably aren't dealing

very well with osteoporosis but I don't think that

this drug or the use of it is going to solve that

problem. I don't think we can cure that issue with

this discussion.

DR. DYKEWICZ: Dr. Pauwels for Glaxo.

DR. PAUWELS: For information for the

panel, actually I want to show you the published

data on the bone density in the EUROSCOP study.

The data on triamcinolone has been mentioned, but

the EUROSCOP study was a three-year study in people

who were continuing to smoke because that was an

inclusion criterion, and they were 55 years old and

had COPD. So, we treated with 800 mcg of

budesonide or with placebo.


There you see actually the changes over

the three years in the different sites of the

lumbar spine and the different femoral sites for

budesonide and placebo. Here you see the p values.

So, what that really shows is that in that study,

and I am talking about 190 subjects that were

followed over three years, there was absolutely no

difference between placebo and the active component

with regard to bone density.

We also looked in a larger sample, in more

than 600 subjects, with spine radiographs before

and at the end of the study. Indeed, as mentioned,

if you do a careful analysis of the spine, there

was the presence of vertebral fractures defined

with the computer in, in fact, around 12 percent of

the subjects. At the end of the three-year study

there were three subjects in the placebo group and

five subjects in the budesonide group who had an

additional computer-defined fracture.

In other studies in asthma we have also

seen, for example, differences in the effect on

bone density between beclomethasone and

fluticasone. I don't know if the slide is



Here you see over the six months -- this

was a crossover study in the same individuals. We

actually observed that for the beclomethasone

treatment there was a further decrease in bone

density, whereas there was an increase in the

subjects who were switched to fluticasone. So,

there are differences between inhaled

corticosteroids with regard to the effect on bone.

DR. MEYER: Would you please clarify the

doses and the sizes of this trial?

DR. PAUWELS: Yes, the dose in that study

was actually 1 mg or two times 500 mcg of

beclomethasone and two times 250 for fluticasone.

So, they were equipotent doses.

DR. MEYER: I will let you pass with the

equipotent doses without further comment. The N,

please? The size of the trial?

DR. PAUWELS: I don't remember exactly but

that was a fairly large study that we did in

different Belgian centers. I don't know if

somebody has the publication.

DR. SHAH: It was over 100.

DR. PAUWELS: Over 100, yes.

DR. DYKEWICZ: Dr. Malozowski?

DR. MALOZOWSKI: In a 24-week study you

will not see fractures related to glucocorticoids,

either oral or inhaled. Therefore, it is very

unlikely that in a study of this length you will

see anything. In fact, the fractures that are

reported -- I don't have the information, but these

are clinical fractures I imagine, the five that

were reported. We don't know whether these were

due to accidents or other things, or not the

typical fracture that you will somehow attribute to

glucocorticoids in the spine, maybe subclinical,

etc., etc.

We can continue to discuss this for three

or four more hours, but the point is like this from

my perspective, glucocorticoids lead to

osteoporosis. Whether this one does it to a larger

extent compared to others I don't know because we

don't have the data. A study that lasts 24 weeks

cannot show this kind of marker unless the doses

are astronomical. With time, patients on

glucocorticoids, either oral or inhaled, may be

prone to have osteoporosis and potentially


DR. BONE: I think I would like to say

that from the FDA comments I understood that the

major concerns were based on these two- and

three-year long studies which showed some effect on

bone density. So, I addressed my comments

primarily to those studies, and they did show, not

with this glucocorticosteroid but with other

glucocorticosteroids that were inhaled, a

relatively small decline in density that was

entirely consistent from site to site in some of

the studies.

I think that what we are talking about

here is how we might label the product in the

absence of product-specific data. That is the

problem the agency has. If we were to advise the

agency, I think we could say something like this,

patients with COPD are at substantially increased

risk for osteoporosis and probably ought to be

tested in general. This is an important piece of

information to have in hand when considering what

to do about steroid therapy of any kind. The sort

of worst case data based on the publications would

suggest that there may be a small adverse effect of

at least some inhaled glucocorticosteroid

preparations on bone density, based on published

trials, and that in patients who appear to have

increased risk for osteoporosis consideration

should be given to intervention and that continued

monitoring of patients who are at that kind of risk

is appropriate and one of the additional factors to

take into account would be such an exposure.

The rate of loss in these studies that

were referred to is measurable but not extremely

fast, for the most part. So, we are not talking

about something where people are going to develop

osteoporosis overnight. If they don't have it

today, they are not going to have it in six months

from this exposure. So, I think a measured

approach to the labeling, pending the availability

of product-specific information, would be one that

could give considerable comfort from the standpoint

of the agency. But discussions in the labeling

should be clear that this is subject to very

substantial amendment when product-specific data

from the prospective trial become available.

My recommendation to the committee is that

the magnitude of the published effect for most

patients, on the average, is small in comparison or

modest, I should say, in comparison to other risks

that these patients have for osteoporosis and

should be kept sort of in balance and in context.

I would strongly urge all of you to do bone density

testing on patients with COPD because you are going

to find an awful lot of patients who will benefit

from treatment whether you give them steroids or


DR. DYKEWICZ: Any other comments on bone?

If not, I think we are actually going to take a

15-minute break now. We will have a little bit of

discussion after that and then we will finally hit

the questions.

[Brief recess]

DR. DYKEWICZ: Let's resume please. What

I would like to do before considering the questions

is to still offer some time for discussion about

what may be still a few remaining issues, although

I think we have really dealt with a lot of issues

quite well.

Discussion about ocular issues --

cataracts, glaucoma, comments from the committee or

concerns that you have about this? Dr. Stoller?

DR. STOLLER: I just have one methodologic

question. I think I know the answer, but in the

ascertainment of cataracts or glaucoma, presumably

this was all self-reported in this study as opposed

to looked for explicitly by intraocular pressure

measurements, eye exams, etc. So, it is patient

self-reported data? Is that correct?

SHAH: Yes, it was.

DR. MEYER: For the record, the answer was

yes from the sponsor.

DR. DYKEWICZ: But in terms of slit lamp

exams looking for posterior cataracts, was that

something that was looked at?

DR. SHAH: No, it wasn't.


DR. WISE: Just for the record, the Lung

Health Study did, in fact, do slit lamp exams on a

subset of the population with triamcinolone 12 mcg

a day and did not find any evidence of increased

opacification of the lens, and no increase in


DR. DYKEWICZ: Any other comments on

ocular concerns? We will finally have anything

about miscellaneous, connective tissue, other

systemic events.

DR. MALOZOWSKI: I have a question to the

sponsor. Did you look at weight gain across

groups? If you did, was there any difference in

weight gain?


DR. MALOZOWSKI: You did not look or there

was not, or both?

DR. SHAH: The answer is we did not look

at the end of the treatment at weight gain. Again,

we have done weight measurements in asthma with

these doses of Flovent and don't see evidence at

these doses of weight changes due to treatment.

DR. DYKEWICZ: Any other general -- yes,

Dr. Bone?

DR. BONE: I am sure it wasn't done here,

or at least I didn't see anything about it, but I

was wondering if the three-year study is going to

be looking at gonadal steroids, particularly in


DR. WIRE: This is Patrick Wire, from

GlaxoSmithKline. The question is looking at

gonadal steroids --

DR. BONE: In the prospective three-year

study are you looking at gonadal steroids as well?

DR. WIRE: No, sir.

DR. BONE: But you have some frozen serum

some place, I bet.

DR. MALOZOWSKI: If I may follow-up, are

you measuring insulin in the studies?

DR. WIRE: No, sir.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: Again, one question of

clarification, I appreciate the presentation about

the stratified analysis by the Chronic Respiratory

Questionnaire but the question has come up several

times about looking at delta FEV1's by FEV1 strata.

I assume that because we have not seen it, it has

not been done? Or, has it been done and may we see


DR. SHAH: Sure, we have done it and we

can show you the data.


What we have here is model estimated

effects that we would observe in that study

according to various baseline percent predicted.

Essentially, you know, you can do this several

different ways. You can actually take cuts of the

proportion of patients who were below a certain

FEV1 and then look at the response. The problem

with that obviously, as we all know, is that you

are only looking at a very small proportion of the

overall data and there are different baseline

variables that obviously have to be considered in

interpreting the results.

What this does is adjust for a lot of

baseline differences, such as demographics and

such, and it helps you get a better estimate of the

results between treatment in the setting of these

different lung function tests or percent predicted.

These are adjusted for difference from

placebo so it is actually subtracting out the

response of placebo in here. So, what you are

seeing on the left is that the mean change -- this

is for the integrated data, again, keeping in mind

that we did have one study where we had, you know,

much less robust treatment effects and that is

going to drive the overall mean data -- but you

see, the left results are for people who had a 30

percent predicted and this is the type of effect

you would expect to see. On the right is the 50

percent predicted and this type of effect you would

expect to see. You do see treatment effect in the

same trend across all subsets, with greater

improvements on the combination versus the

individuals, and the individuals being clearly

better than placebo since each of them is above the


So, the differences are fairly small

across those three groups, and each of these

percent predicted actually turns out to be a

quartile of our patient population. So, the

numbers actually correspond very closely to the

distribution of these FEV1 results in our studies.

So, what you see is reasonably consistent and

robust responses across all subsets of patients,

with the responses being about 80 ml to 100 ml

depending on the dose of FP across the three strata

of percent predicted at baseline of FEV1.

DR. STOLLER: Let me make sure I

understand. So, this is a modeled baseline

stratification, not all-comers, number one. The

follow-up question would be, would you agree with

the notion, based on what you have shown us, that

the increments in FEV1 across strata are as

generous in the low end group as they are in the

high end group?

DR. SHAH: That is what these data would



DR. DYKEWICZ: Other comments or

questions? If not, before actually turning to the

questions I would like Dr. Meyer to make a few

comments and review issues about regulatory

language, and clarify that for us, and also, if you

could, make some statement about the level of

evidence that the FDA would like to have us

consider in making determinations.

DR. MEYER: I will try to do that. I

think there was a question from Dr. Wise earlier

about this substantiation and level of evidence,

and so on, and without getting into a very detailed

legal discussion of the Act, which I am not really

qualified to do anyway, the substantial evidence

generally refers to sort of the quantum of

evidence, not necessarily the quality of the

evidence but the quantum of evidence. In other

words, we interpret that generally as two or more

trials. So, to put that into perspective, one

might say that the fluticasone 250 data was not

replicated and, therefore, does not meet those

criteria, although you always have to put it in the

context of everything else you know so I am not

presupposing anything on the committee's part by

saying that; it is just an example.

The other thing in the statute though is

that it really does refer to data that is

convincing to a body of experts; that they feel it

reasonably convinces them that the drug will have

the effect that it is purported to have and that it

is safe, and that the sponsor has done all tests

reasonably applicable to demonstrate its safety.

So, that is where you folks come in. You are the

body of experts and that is the kind of advice we

are seeking from you folks.

Clearly, we have said for both strengths

of Advair and at least for the fluticasone 500 we

would not dispute that the sponsor met their

prespecified endpoint, and have done so in

replication for all those and for the 250 that they

did in one study. I think the question that we put

to you as a body of experts is, given what we know

about COPD and looking at these data in the milieu

of what we know about COPD, how do you interpret

those? Does that give you a lot of comfort that

there is substantial evidence that this drug has

its purported effects and that it is safe under the

conditions of use?


DR. DYKEWICZ: Any questions of

clarification on that from the committee? If not,

let's begin discussion of the questions. The

format of this is that I am going to, just to help

focus us, read the stated question that is given to

us. But then we will go around and really each of

us chime in with some thoughts about this.

Finally, at the end of that discussion I will again

read the question and we will take a formal vote,

recording the vote of each voting member of the


So, the first question that we were asked

was do the data provide sufficient evidence of

efficacy of Flovent Diskus, at a dose of 250 mcg

twice daily, 500 mcg twice daily or both, for the

indication of long-term, twice daily maintenance

treatment of COPD, including emphysema and chronic


To pen that up for discussion, who would

like to begin? Dr. Parsons, you are a voting

member of the committee.

DR. PARSONS: Well, I think Flovent 250 is

up for discussion; the 500, I am still a little

torn. I think they met their primary efficacy goal

with the dose of 500 and it did show a change in

FEV1 that they set out to show, but I am still

concerned about the lack of a clinical benefit as

measured by the secondary endpoints. What I can't

determine, and I know we are supposed to evaluate

these separately, is efficacy and safety. Part of

my hesitation is that I still am not sure that I

understand the safety consequences. So, in my own

mind it is a little difficult to dissociate

efficacy and safety.

DR. DYKEWICZ: There might be some value

to recognize that we are actually looking at three

types of decisions here. First of all, we are

being asked to make a statement solely focusing on

efficacy. Then we will be asked about safety

concerns. Then we will finally be asked for

recommendations about approval. So, it is quite

possible, just as a paradigm of thought, that one

might feel that there is efficacy that has been

demonstrated but then, in terms of safety concerns,

the safety concerns would outweigh the efficacy,

thereby leading to a recommendation for no

approval. So, we really should, at least

initially, focus on each of these issues

separately. Dr. Bone?

DR. BONE: I really think that the

efficacy assessment requires the expertise of the

regular members of the committee and not mine.

DR. MALOZOWSKI: I second that.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: I guess I would frame my

response to the first question in the following

way, I view the language of the indication as very

broad-reaching and so it calls upon my analysis of

the definition of COPD as it relates to this

population, which we have talked about. I think

that GlaxoSmithKline has done an impeccable job of

showing the FEV1 increment in this population. As

I have mentioned before, I think that given what we

know about the prevalence of reversibility, I am

struck by the high frequency of reversible airflow

obstruction in this population with a higher than

usual dose of a bronchodilator on a single baseline

assessment, as opposed to the more serialized

assessment of bronchodilator reversibility as we

understand it in the literature both from the

alpha-1 and the IPPB trial.

So, in that context, I would have to say

that part of the language, as it is put in the

first question, that gives me the greatest pause is

the notion of COPD, parentheses, including

emphysema and chronic bronchitis. To Dr. Wise's

comment before, recognizing that the FEV1 increment

of 100-160 ml I think has been clearly shown,

incontrovertibly shown and elegantly shown in this

study, I would have to say that I would be much

more comfortable framing that by defining this

population as having chronic obstructive bronchitis

than I would be saying anything about the actual

prevalence of emphysema in this group, or

generalizing it to the broader target population of

COPD patients. So, I would say yes, but with the

caveats as stated.

DR. DYKEWICZ: For both doses would you

say that?

DR. STOLLER: Yes, I would be comfortable

with that language for both doses.


DR. FINK: I would echo Dr. Stoller's

concern about the labeling and I would be much more

comfortable with chronic obstructive bronchitis

also. My other concern is long-term maintenance

therapy. I would have to say although they have

shown effectiveness or efficacy for these drugs in

a 24-week trial, I don't think I can endorse

long-term maintenance therapy without data on

either mortality outcome or change in rate of

decline of FEV1 because I think they have shown

efficacy in a 24-week trial; I do not think that

they have established a body of data that supports

long-term maintenance therapy.

DR. DYKEWICZ: Since this does seem to be

a recurring dilemma, let's say that the question

were phrased in terms of whether the data was

sufficient evidence for efficacy for, shall we say,

just treatment of chronic obstructive bronchitis.

Would that make you feel more comfortable?

DR. FINK: I would you feel comfortable

with that labeling or that indication.

DR. DYKEWICZ: Dr. Atkinson?

DR. ATKINSON: I agree. I have been

persuaded that the company has demonstrated

efficacy. The lack of reproducibility at the 250

mcg dose is a little bit concerning. As a

pediatric allergist/immunologist, I am a little bit

also unfamiliar with how adult pulmonologists view

these relatively small volumes but adult

pulmonologists seem to feel that in patients with

at least moderate to advanced COPD they can be


I wonder if the problem with the increased

reversibility in this population might have been

because it was skewed towards chronic bronchitis.

At any rate, I am convinced.

DR. DYKEWICZ: I do also think that the

sponsor has demonstrated efficacy in terms of the

primary variable, efficacy endpoint. I am mindful

though of the directive of the regulatory language

here about substantive, meaning demonstration in

two different trials and, therefore, I would make

some distinction between the 250 and the 500 dose

of the Flovent where we see one trial that shows

efficacy, the other that does not statistically

significantly show efficacy.

So, on that basis I would actually split

my vote and say that I believe that efficacy has

been clearly demonstrated for the fluticasone 500

for chronic obstructive bronchitis but has not been

convincingly demonstrated for the 250 dose.

DR. APTER: Well, most things have been

said before. I agree that the sponsor has shown

that the 500 dose has efficacy, and it is

questionable about the 250 dose but I don't see

that as a bad starting point. So, I have no

objection to either of the doses. I too am

concerned about long-term because I don't think we

have the data. So, I would be happier with just

chronic bronchitis. That is all.


DR. JOAD: I actually would not say that

substantial efficacy has been shown. I think we

are treating the patients for their symptoms, or

the purpose of this is to treat them for their

symptoms and unless we can show clinically

important change of symptoms, then that is the

substantial -- to me, that would be the substantial

improvement of efficacy I would want. This is

going to treat a lot of people and I think it is a

huge decision to make, and I think basing a

decision just on improvement in FEV1 without a

clinically important improvement in symptoms would

not be advisable.

DR. DYKEWICZ: Ms. Schell?

MS. SCHELL: I agree that the 500 did show

efficacy, but I have some concerns on the secondary

again with the symptoms. I agree with it but want

to see symptom-related improvement as well in

patients, and I think as a clinician that is what I

would like to see in treatment of a patient with

the drug.

DR. DYKEWICZ: All right. Let me just ask

Dr. Meyer, can I rephrase the question a bit as I

proposed and state do the data provide sufficient

evidence of efficacy of Flovent Diskus for

treatment of chronic obstructive bronchitis? Would

that be acceptable language?

DR. MEYER: I think you are certainly free

to do that. I would also suggest that you split

the voting to the particular doses. We can infer

the "both" after taking the separate doses.

Although it was phrased sort of as one or both, I

think you can split it out.

DR. DYKEWICZ: So, question 1(a) and

question 1(b). Let's take the formal vote then if

there is no further discussion about that. So, we

will say question 1(a), do the data provide

sufficient evidence of efficacy of Flovent Diskus

at a dose of 250 mcg twice daily for treatment of

chronic obstructive bronchitis? Dr. Parsons?



DR. BONE: I abstain.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: I would say yes.



DR. DYKEWICZ: Dr. Atkinson?


DR. DYKEWICZ: Dykewicz, no.




DR. DYKEWICZ: Now question 1(b), the same

question but for the 500 dose, do the data provide

sufficient evidence of efficacy of Flovent Diskus

at a dose of 500 mcg twice daily for treatment of

chronic obstructive bronchitis? Dr. Parsons?

DR. PARSONS: Yes, but ...

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: Yes, although I think

Polly's comment about yes, but -- there is always

that "but" but yes.


DR. FINK: Yes.

DR. DYKEWICZ: Dr. Atkinson?


DR. DYKEWICZ: Dykewicz, yes.




DR. DYKEWICZ: I am instructed to read the

summary of the vote tallies. The question about

Flovent 250 for treatment of chronic obstructive

bronchitis, there were six no, two yes and one

abstention. For the question about the 500 dose,

one no, seven yes, one abstention.

We will now move on to question two.

Again, I will read the general question but then we

will open things up for discussion before we

actually take a formal vote. The formal question

given to us is do the data provide sufficient

evidence of efficacy of Advair --

DR. APTER: Were you going to break it up

using the word chronic? For instance, not using

the word chronic but long-term rather.

DR. MEYER: If your vote is affirmative in

that, I think we can get to some of these issues

with the greater question about the approvability

and then we can talk after that question about

whether you have any recommendations about specific

changes to labeling.

DR. DYKEWICZ: All right, then moving on,

do the data provide sufficient evidence of efficacy

of Advair Diskus at a dose of 250/50 mcg twice

daily, 500/50 mcg twice daily or both for the

indication of long-term -- well, in this case we

will just say again chronic obstructive bronchitis.

We will begin discussion on this side so we don't

always pick on Dr. Parsons. Any thoughts, Ms.


MS. SCHELL: I agree that there is proof

that it works but I again have a question about


DR. DYKEWICZ: So, if we said that it was

demonstrating efficacy for treatment of chronic

obstructive -- if we just used that phrase,

treatment of chronic obstructive bronchitis, would

that make a difference in your view of that?

MS. SCHELL: Yes, it would.


DR. JOAD: I would have the same arguments

I had before about this one. I think the only

difference here is that there was a difference with

the Advair 500 on the TDI questionnaire but because

things are not internally consistent, for instance

there wasn't a dose response with other things and

it wasn't confirmed by the other measures, I would

have to disregard that at this time.

DR. DYKEWICZ: Dr. Apter?

DR. APTER: I want to exclude the

long-term. I want to include, as in the previous

discussion of Flovent the concept of chronic

bronchitis to be included because that is where the

studies were performed. I think efficacy has been

shown for both doses of Advair.

DR. DYKEWICZ: In terms of my view, I do

think efficacy has been demonstrated for both

strengths. Here, of course, we are looking not

only at the FEV1 increase, but I do think there is

with the TDI 500 study some additional confidence

derived from that. So, I just note that for the

record. Dr. Atkinson?

DR. ATKINSON: Yes, I feel that both

strengths have been shown to have efficacy.


DR. FINK: I would agree that both

strengths have shown efficacy with the altered


DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: Yes. I would say yes to the

language as proposed, but I would say at the same

time that when I endorse the statement of

sufficient evidence of efficacy I am playing by the

rules that pre-negotiated FEV1 criteria were

clearly shown.

As a clinician, I will tell you that I am

a little bit disappointed by the discordance

between those data, a little bit uncomfortable with

the subsets that we saw. They are difficult to

interpret in terms of my clinical practice as a

straight, up-front, not modeled but stratified

analysis of where the increments come. Recognizing

the difficulties of satisfying efficacy criteria

for multiple clinical endpoints, in particular COPD

exacerbations, I am a little bit surprised by the

baseline inattention to COPD exacerbation frequency

in the study which would, in my mind, clinically

help to characterize the labeling individual for

these individuals. For example, it would help me a

lot more in my clinical practice if I knew that

there was evidence that the drug achieved 100 ml

increment in patients whose FEV1's were at the low

end of the spectrum rather than at the upper end of

the spectrum, and who had substantial evidence of

frequent exacerbations, which is kind of the

take-home point in some ways from ISOLDE and which

I think, in my mind and reading of the literature,

helps to inform, although I obviously defer to Dr.

Pauwels and others who authored these documents but

in my mind must have helped inform the indications,

the guidelines with regard to the indications for

inhaled corticosteroids in those widely influential


So, I would say yes to both, framed in the

way that I have said but I am hoping that the text

outside the dichotomous yes/no answer is

informative to the agency in helping to think about

things that would be clinically important and

persuasive to those of us who actually administer

these drugs to our patients.

DR. DYKEWICZ: Thank you. Dr. Parsons?

DR. PARSONS: The answer to this one is

also yes, with a slightly lower "but" in that the

FEV1 changes were shown clearly. At least in this

one there was more evidence of what I would

consider clinical efficacy. So, I feel a little

bit more comfortable with this one than the

previous yes.

DR. DYKEWICZ: Any further general

comments before we take a formal vote?

DR. MALOZOWSKI: Am I correct in stating

that the two doses had similar effects on the

primary endpoint?

DR. DYKEWICZ: I believe that is a correct

representation. Let's begin the formal vote then.

Once again, we really haven't had any discordance

in thought about the 250 and the 500 dose here so

we will have it as a single question.

So, the formal question is doe the data

provide sufficient evidence of efficacy of Advair

Diskus at the 250 and 500 fluticasone component

doses for treatment of chronic obstructive

bronchitis? Ms. Schell?







DR. FINK: Yes.



DR. DYKEWICZ: Very well. We have to take

the tally. So, the secretary tells me seven yes,

one abstention and one no.

Now turning to two questions regarding

safety, the first discussion will be about the

safety of the Flovent Diskus for treatment of

chronic obstructive bronchitis. We will go back to

Dr. Parsons.

DR. PARSONS: I don't know. I think we

have had very limited data to evaluated related

directly to this product related to this patient

population, and I cannot with the data available

clearly that this is or is not safe.

DR. DYKEWICZ: What would you say would be

your biggest reservation about safety?

DR. PARSONS: My biggest reservations

about safety are the potential consequences of

long-term steroid effect in this patient

population, where many of the patients will be

exposed to this for several years and this is

already a relatively potentially debilitated

population. So, the data provided suggest, as I

think was clearly pointed out, that there is a

steroid effect. I think the magnitude of it in

this population is difficult for me to determine

and, therefore, I think longer-term data in this

population, using these drugs at these doses, would

be very helpful. But if you ask me outright do I

think it is not safe, I can't say yes to that

either. I would have to say I don't know.

DR. DYKEWICZ: Dr. Bone, did you want to

make a comment on this question?

DR. BONE: Briefly. I said most of what I

had to say earlier in the areas where I felt like I

had some competence and would certainly defer in

the overall picture to the people who use these

drugs as part of their practice. But I was just

going to reinforce the point here that in most of

the reservations that were expressed by the

Division and by people in commentary relate really

to the term "long-term" and I think there is a

tremendous difference here if you use the term

"maintenance treatment" and exclude use of

"long-term" because the issues that arose were ones

that would be expected to arise rather gradually.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: I would like to think my

comments about this abide by the same rules of

evidence. That is to say, I was willing to concede

efficacy framed by the negotiated outcomes on FEV1.

I would say that in this particular instance I am

not satisfied about safety over the relatively

short-term data from the data in this application.

Recognizing that there are ancillary data that have

been cited from the Lung Health Study about ocular

manifestations, I could not say that this 24-week

trial, monitored in the way it was for the outcomes

of interest, would assuage concerns that I would

have about the truly long-term implications given

that physicians will likely treat for far longer

durations than the 24 weeks that we see here. So,

I would have to say I am not satisfied. Not that I

am satisfied that it is unsafe, but I am not

satisfied that it is safe. So, it flips on the

confidence of the absence of an endpoint.


DR. FINK: I guess I would take a slightly

different approach in my deliberation and say that

it is not safe. My reasoning behind that is that

looking at the data, as it was presented, there is

no data that really identified that fluticasone 250

or 500, particularly the 500, had efficacy that was

better than salmeterol and the safety profile of

fluticasone 500 is not that of salmeterol. So, I

can't see saying that this is a safe drug when we

have an equally effective drug that is already

approved that is clearly safer.

DR. DYKEWICZ: Dr. Atkinson?

DR. ATKINSON: I am also undecided, but I

think I am torn in the direction of saying that I

know that there are probably going to be some

endocrine changes; there probably is going to be

some decrease in bone mass over time but it is

going to be very slow. So, the question is how

unsafe is it? There are side effects to most

medications. COPD is an inflammatory disease, and

what I know about inflammation tells me that

anti-inflammatory medicines are very helpful.

These patients are getting those orally, which are

likely to be a lot more unsafe for the skeleton but

I know that is not a consideration for deliberating

on this. I guess I would have to say that overall

for considering the lung health of the patient and

the skeletal health and what I view as a very slow

erosion of their bone density, I would have to say

the overall safety profile is good.

DR. DYKEWICZ: From my perspective on the

safety I am considering several things. One is we

do have an agent for which we do know what the

potential side effects would be. We know what the

signals would be in terms of adverse effects of

corticosteroids on the body. So, it is not a

question that we are dealing with a totally unknown

entity where we don't know what to look for.

In the case of the studies of fluticasone

in asthma which generally are reassuring, of

course, that certainly has relevance to a

consideration of the use of this agent in COPD.

But I am quite conscious that the population of

COPD patients may potentially be more vulnerable to

certain adverse effects than might occur in a

population of asthmatics which will tend to be,

among other things, younger for instance.

So, I would like to, of course, ideally

see much longer follow-up than just the 24-week

study. As Dr. Malozowski has pointed out, there

are certain potential side effects that will be

perhaps more apparent with longer-term follow-up.

However, in the main, with those qualifications, I

do believe that there is reasonable safety data

that would then have to be judged in a risk-benefit

assessment when we finally come to the

approvability question.

DR. APTER: Like many of my colleagues, I

cannot say that these drugs are safe long-term for

chronic obstructive bronchitis. I am concerned

also about the large number of dropouts which made

the follow-up even shorter. I am concerned, like

the others, that this is a different population

than the asthmatic population -- older, more



DR. JOAD: I am also concerned about this

product in this age group for this disease. I

think we need more data before we can decide on

safety. I just wanted to comment that changing the

package insert to saying for the treatment of

chronic obstructive bronchitis -- I think what that

will become is long-term therapy of chronic

obstructive bronchitis. I don't see how, in

practice, people will somehow give it for six

months and then stop it, and then give it again

sometime later. I think if it is approved for the

treatment it is going to be approved for the

long-term prevention of symptoms.

DR. DYKEWICZ: Ms. Schell?

MS. SCHELL: On this issue, I have some

concern and I can also see the potential benefit.

I am trying to weigh the benefit-risk in what I

have seen in patients who have been on this drug

already for treatment. I do have a problem with

the long-term wording because the studies weren't

geared towards the COPD patients that demonstrate

safety. I appreciate Dr. Bone's remarks and his

input on that, but looking at the benefit-risk,

again, it is difficult for me to say. I would

think as long as we continue to monitor, and maybe

that could be in the wording of the labeling, this

would be a benefit to the patient.

DR. DYKEWICZ: Let's then begin the formal

vote. Again, do the data provide sufficient

evidence of safety of Flovent Diskus for treatment

of chronic obstructive bronchitis?


DR. DYKEWICZ: Did you want to vote on

this question, Dr. Bone?

DR. BONE: I think the overall assessment

really belongs to people who work in this area so I

will pass on this.

DR. STOLLER: I will say no.







DR. DYKEWICZ: The vote on the question is

three yes, five no, one abstention.

Moving on to question four, do the data

provide sufficient evidence of safety of Advair

Diskus for the treatment of chronic obstructive

bronchitis? I think we will go back to that side

of the table. Some thoughts on that question, Ms.


MS. SCHELL: Much of my opinion is the

same as with the previous question as long as take

the "long-term" out and we just say treatment.

Also, again, looking at the population studied, I

have concerns about COPD patients compared to

asthma patients in this study.


DR. JOAD: With Advair, my concern is the

steroid components so my thoughts would be the same

as before.

DR. DYKEWICZ: Dr. Atkinson?

DR. ATKINSON: Yes, I am not that

concerned about the salmeterol component either so

I think my opinion wouldn't change either.

DR. DYKEWICZ: Thank you, Dr. Atkinson.

Dr. Apter?

DR. APTER: My concerns are the same as

previously for Flovent. I could say "but" and

change if the wording included "not for long-term


DR. DYKEWICZ: So, if I rephrased it

long-term, you would vote no, but if I stated that

it was just for treatment you might vote yes?

DR. APTER: That safety has only been

established for short-term treatment.

DR. DYKEWICZ: I guess my view is that the

data is paralleling that of the safety data for

Flovent Diskus because I don't really believe that

the addition of the salmeterol component raises any

significant safety issues. Dr. Fink?

DR. FINK: I would agree with that and I

think actually there is an additional safety factor

for the Advair Diskus in that clinical practice I

think it is general knowledge that Advair is only

dosed twice a day, and I think the potential that

physicians would escalate the dosage of Advair to

three or four inhalations a day is far less than

for the Flovent 500 Diskus being escalated. So, by

being a combination product -- I never would have

imagined myself saying this, I think it is probably

safer in clinical practice.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: My comments would be as

before, although I would say, again, in the

interest of being as helpful to the agency as I can

with regard to really framing what I think, I think

in the context of what we have been shown, as I

mentioned before, I don't have huge safety concerns

within the scope of the 24 weeks as shown. My

concerns are extrapolated to the clinical

implications of long-term use which is not

satisfied by the data at hand. So, it really

defaults to kind of a word-smithing issue. You

know, if you are going to write the label so it

says it is approved for the relatively short-term

management of chronic obstructive bronchitis and it

is safe in that regard, my concerns are less. If

it is going to be kind of open-endedly endorsed for

the in perpetuity treatment of patients, I have

concerns, as I have said before, about its safety.

So, I would say that, very simply put, I am not

satisfied that long-term safety benefit has been

shown by the data at hand. I think within the

framework of what we have been shown I have no

major immediate concerns about major adverse

clinical events right now.

DR. DYKEWICZ: Thank you. Dr. Parsons?

DR. PARSONS: I have the same concerns

with this one that I did with the Flovent, which is

that the long-term safety has not been established.

I think, no matter how we word that, whether it is

just simply for treatment of chronic obstructive

bronchitis or whatever, it will be used long-term.

I think to try to label it that it both has

efficacy and is safe for 24 weeks -- I guess that

is an option but I think every physician that then

treated their patient for 24 weeks, at 24 weeks

would have a therapeutic decision to make and I

think the patients would stay on the drug. So, I

think we need to look at it as something that is

going to be used as long-term persistent therapy,

and I think in that use there is not adequate

safety data.

DR. DYKEWICZ: Any other general comments

from the committee? If not, let's begin the formal

vote. Do the data provide sufficient evidence of

safety of Advair Diskus for the treatment of

chronic obstructive bronchitis? Ms. Schell?



DR. APTER: Again, no in terms of long




DR. FINK: Yes.



DR. DYKEWICZ: On the issue of safety of

Advair Diskus, the votes were four yes, four no,

one abstention. Dr. Meyer?

DR. MEYER: Just before we go on to the

next question, I just wanted to really set the

stage for this question because I think it is

important that here you don't change the wording;

that you use the wording that we have because that

is the wording that is proposed by the sponsor.

But, I think a yes vote, if the committee will look

down below, allows for several options. So, in

essence, what we are asking here is any level of

yes. In other words, if you are clear that your

answer is no, no matter what is done to the

labeling or what other kind of Phase IV studies

might be recommended, vote no. If you are in any

other category vote yes, and then you will have an

opportunity to give us advice as to what labeling

or other restrictions might be needed, whether only

one dose or two doses and what Phase IV studies you

might recommend if it is a yes.

DR. FINK: Just a clarification I guess,

do we have to consider both products? If we have

one product that has clearly higher efficacy with

no additional toxicity, I am not sure why we would

want to market or approve two different products

even if they showed efficacy and safety when you

have a product that clearly has better efficacy

with no additional safety concerns.

DR. MEYER: I understand your point but we

are asking these as separate questions about

separate applications and I think we need separate


DR. DYKEWICZ: A question by Dr. Parsons?

DR. PARSONS: I just have a question about

procedure. If the answer is yes but there are

Phase IV studies that are recommended, do those

studies get completed before the drug gets



DR. PARSONS: Is there a time frame in

which they need to be completed?

DR. MEYER: We do agree to a time frame

with the sponsors in instances of Phase IV studies.

Generally, the kind of studies that are often done

in Phase IV, it is two or three years before we get

the data in.

DR. DYKEWICZ: Dr. Apter?

DR. APTER: I was just going to comment

about Flovent. Some patients will probably use

Flovent and Serevent separately. Maybe their

insurance won't allow a combination, or things like


DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: Just a procedural question,

my understanding is that, obviously, as a

hypothetical were it to be approved and were there

to be recommendation for Phase IV studies, and we

were two and a half years into the Phase IV studies

with clear evidence of higher incidence of

fractures in a dose-related way, what implications

would that have retrospectively for the indication?

In other words, what are the teeth of a Phase IV

study from the agency's perspective?

DR. MEYER: In usual approvals -- because

there are approvals contingent on a Phase IV study

-- it is understood under the actual mechanism of

the statute that if it comes out negative you

withdraw approval. I think with most of this we

would be talking about perhaps very stringent

labeling changes.

Bear in mind that in this specific

instance these drugs are already on the market and

they are, I am sure, being used for COPD now and

they will continue to be used for COPD one way or

the other after our discussion today. So, I don't

think we would be talking about a Phase IV study

that would lead us to absolute withdrawal --

certainly the approval, but perhaps not even the


DR. DYKEWICZ: Comment from Glaxo?

DR. WHEADON: Just one or two points of

clarification. I fully respect the vote and the

commentary of the committee and we really

appreciate the input. I think it is important to

note that there are several precedents where drugs

have been approved for chronic illnesses. I am a

psychiatrist; depression being a prime example of

one, where the language can be such that you

indicate that the studies were of certain duration.

In the case of studies for depression they

typically are six to eight weeks duration. So, the

labeling clearly can reflect the duration of

treatment in the studies that we have presented

before you.

Additionally, a number of concerns have

been raised by the committee concerning the

potential for long-term use. From our perspective

certainly, labeling is perhaps the most informative

place for physicians to understand what we do know

and what we don't know about safety. A number of

committee members have sort of been reflecting on

just how safe or unsafe these things may be.

Clearly, labeling can be a very cogent repository

of that state of affairs. I think that is

important to keep in mind as we go through the next

level of discussion.

DR. DYKEWICZ: Thank you. I guess I would

make my personal response that I think oftentimes

physicians are remiss in looking at labels and the

details in labels. Some of my colleagues have

expressed concerns that although one can nuance

phrasing in labeling, there still is the concern

that when you do give an approval status you have

to think that, in fact, physicians won't be reading

the fine print. So, that is a consideration.

But just to redirect to Dr. Meyer in terms

of the decision about approvability, based upon the

statutory language that we are using in our

deliberations here, there is the statement that we

are making an assessment about substantial evidence

of efficacy and safety and, thereby, kind of an

implicit assessment of relative benefits versus

risks of the drug. So, that is what I am

personally going to use in my deliberation but

presumably other people will consider that as well.

DR. MEYER: Absolutely. This is the

question that really integrates what we know from

the efficacy and what we know from the safety, and

how you put that together in making your


DR. DYKEWICZ: This now is again just

discussion on what your thoughts are about the

approvability of Flovent Diskus for the indication

of long-term, twice daily maintenance treatment of


DR. PARSONS: Well, I voted that, yes, it

had shown efficacy at the 500 mg dose, but it was a

yes, "but" and I thought that there was not

adequate safety data. So, I think if I add those

together the answer would be no, I would not

recommend approval at this time, primarily because

of safety concerns, which are not necessarily all

that great but the level of efficacy shown also

wasn't that great. So, I would err on the side of

saying no.

DR. DYKEWICZ: Thank you. Dr. Stoller,

your thoughts?

DR. STOLLER: Again, I want to be very

explicit in my response to the process, which is to

respond to Dr. Meyer's lead that if there is any

dimension of yes one has the opportunity to qualify

the yes. So, I would say overall yes. I would say

that there would need to be very stringent

constraints on the labeling regarding the

difference between COPD and chronic obstructive

bronchitis. I would have to put very specific

language about duration of therapy in regard to

what the indication would say, and I do that

cognizant of the difference between what it says on

the label and how it is used clinically. I live in

that world and I understand that world very well,

but I think the rules of engagement, if you will,

are around the specific endpoints. We are not

turning the clock back and saying, you know, could

we design the study from first principles. I am

sympathetic to the significant amount of work and

energy that has gone into trying to evaluate it

along those lines.

So, I would say yes, but in terms of

long-term for both doses I would have those

labeling contingencies on both, and I would say

that in Phase IV studies I would absolutely be

interested in long-term monitoring, not just on

patient-reported data but with regard to the

explicit investigation for both bone and ocular

manifestations. Given what we know from some of

the asthma literature, admittedly what we don't

know from the Lung Health Study, and also on a more

prolonged examination of survivorship, which I

gather will ultimately be forthcoming and which

will clearly inform the clinical relevance with

delta FEV1 of 100 or 160 ml. I think all of us, if

we were presented with data which showed that there

was a 100 ml increment that was reproducible but

translated over longer term to no symptomatic

benefit, perhaps no survival benefit and a higher

frequency of fractures, or even one more cataract,

it would be very difficult to clinically embrace

the use of these drugs.

So, I would say yes in terms of the

overall possibility that there would be benefit,

but I would think that yes would have to be very

carefully crafted in the labeling around those

concerns, and I would have those recommendations on

Phase IV monitoring.

DR. DYKEWICZ: Thank you. Dr. Fink?

DR. FINK: I would lean towards saying no

with the fact that the relatively modest effect

carries with it all of the toxicity and safety

concerns, and it would be hard to approve the

steroid component alone when you think of the

additional benefit with the use of the combination


DR. DYKEWICZ: Dr. Atkinson?

DR. ATKINSON: I would recommend yes, but

I would also agree with the comments that have been

made previously about specifying that the treatment

period for which safety had been shown was only 24

weeks, and that the population that is was most

likely to be effective in was chronic bronchitis.

DR. DYKEWICZ: Thank you. I guess I would

view a qualified yes, echoing Dr. Stoller and Dr.

Atkinson, with some additional consideration about

labeling relative to Dr. Bone's discussion earlier

about the appropriateness of considering that

because the long-term adverse effects of

fluticasone and bone density are not well known in

COPD patients, are not well characterized in COPD

patients, consideration should be made to

assessment of periodic bone density measurements.

You know, the exact phrasing might be worked out

but I think there would be some caution statement

that I would put in that would reflect that


DR. APTER: I agree with Dr. Stoller and

Dr. Dykewicz, and I would say yes to both doses. I

think some patients would use Flovent and Serevent

separately. Again, I do think labeling

restrictions are needed. I am concerned about the

lack of long-term data. Phase IV studies, I agree,

should look at both side effects and efficacy

markers like survival, like exacerbations, like

prednisone requirements and side effects on bone

density findings and adrenal status.

DR. DYKEWICZ: Thank you. Dr. Joad?

DR. JOAD: I know it seems clear I am

going to say no but the caveat I think I would like

to say is that I think there is potential in both

of these products, and my concern is we haven't had

the demonstration of them, that they are effective

and that they are safe. This is such a large

number of patients that will receive it, and they

are elderly, and I think now is the time, before

you approve it, to show that it really is effective

and to show that it really is safe. To me, it

would be jumping the gun to approve it now when we

could require very carefully controlled studies to

satisfy ourselves that it is effective, really

changing symptoms, and it really is safe.

DR. DYKEWICZ: Thank you. Ms. Schell?

MS. SCHELL: As to the wording in question

five as it is, I still have problems with the

wording but I would say yes, and I think the only

dose I would approve would be the 500. I would

also like to see restriction on labeling,

including, as Dr. Bone said, some pretesting on

patients for their bone density and follow-up.

Also, I would like to see in the Phase IV studies

some pre-exacerbations after they get started on

the drug to see if there was a comparison in less

frequency, and dose.

DR. DYKEWICZ: Thank you. Now let's take

the formal vote on question five, with the provisos

that we have. So, do you recommend approval of

fluticasone Diskus for the indication of long-term,

twice daily maintenance treatment of COPD,

including emphysema and chronic bronchitis?

DR. PARSONS: Could I just ask a quick

clarification question? Dr. Stoller, I think Dr.

Fink and a couple of others, when you are saying

yes with specific labeling restrictions, are you

thinking of restrictions being yes for 24 weeks in

chronic obstructive bronchitis?

DR. DYKEWICZ: I personally was thinking


DR. PARSONS: I am sorry --

DR. DYKEWICZ: No, that is fine, I think

four of us may have been of a similar mind on this

-- but that there would be a labeling statement

that the studies conducted were of limited duration

of 24 weeks. I think, if I am correctly

summarizing, the thought was that we would

recommend that Phase IV studies be aggressively

pursued about looking at concerns of systemic side

effects and particularly bone issues. As I

understand it from the charge given to us by Dr.

Meyer, if those provisions or provisos or caveats

are articulated, if we would then feel we could

state yes, then we should vote yes.

DR. PARSONS: I vote no.

DR. DYKEWICZ: Dr. Stoller?

DR. STOLLER: Again, under the rules as I

understand them from Dr. Meyer's charge, I would

say yes contingent upon all the comments I made.


DR. ATKINSON: Yes, under the same


DR. DYKEWICZ: Yes, with restrictions.

DR. APTER: Yes, with restrictions.



DR. DYKEWICZ: The formal vote on the

Flovent recommendation for approval would be five

yes, three no and one abstention.

Last question, number six, do you

recommend approval of Advair Diskus for the

indication of long-term, twice daily maintenance

treatment of COPD, including emphysema and chronic

bronchitis? Let's begin discussion with Ms.


MS. SCHELL: I would agree to approve this

drug with the same reservations I had previously.

I would approve both doses with labeling

restrictions as well, and a continued Phase IV

study with those recommendations.


DR. JOAD: I don't really have anything to

add, but I would just like to repeat that I think

it is unlikely that people will only give it for 24

weeks, highly unlikely.

DR. DYKEWICZ: Dr. Apter?

DR. APTER: I would vote yes, with the

same restrictions and arguments as previously.

DR. DYKEWICZ: I am of a similar mind,


DR. ATKINSON: Yes, as before.

DR. FINK: I would vote yes on this drug,

but I would like to see a required Phase IV trial

both for Advair and Flovent, if it is approved. I

think there should be a required Phase IV dose

escalation study to actually provide data on how

many patients would have a suboptimal response at

the 250 dose of either drug and have a better

response at the 500 dose.

DR. DYKEWICZ: Thank you. Dr. Stoller?

DR. STOLLER: I would say yes, again

subject to the same contingencies. I guess I would

also perhaps use this as an opportunity to talk

about -- I don't see the language about the

doubling dose reflected in this commentary and that

was, as I remember, Dr. Lee's initial comment, that

there was language about doubling the dose for

failure to respond. I would say one of the other

labeling contingencies that I would create would be

to eliminate that as I am not satisfied with the

dose responsiveness data and I think that in order

to justify that comment it would require evidence

that within a single patient, who failed to respond

at a lower dose, that there was essentially an

inter-patient crossover experience rather than a

parallel controlled comparison of two cohorts to

show that doubling the dose was justified for

non-response to the lower dose. So, I would not be

comfortable with language about a higher dose may

help in the absence of benefit at the lower dose.

So, that is the other qualification of language

that I would apply. I know it is not on the table

here because it is not framed in the question, but

as I recall it was one of the language indication

and I guess I would comment on that probably around

both of these doses.

DR. DYKEWICZ: Actually, I am very glad

you mentioned that because that would be a concern

of mine as well, that the recommendation for dose

escalation in an individual patient has not been,

obviously, looked at with the data presented. Dr.


DR. PARSONS: I have the same concerns

regarding this one as I did for Flovent. I would

still answer no. I think if Phase IV studies were

done they would probably show that this drug is

safe. I think that is likely to occur. I think

with the limited efficacy that has been shown, it

is worth waiting to be sure the drug is safe. So,

that would be my recommendation.

DR. DYKEWICZ: Thank you. Now for the

formal vote with the rules of engagement that have

been articulated, do you recommend approval of

Advair Diskus for the indication of long-term,

twice daily maintenance treatment of COPD,

including emphysema and chronic bronchitis? Ms.


MS. SCHELL: Yes, with the stated




DR. DYKEWICZ: Dr. Apter?

DR. APTER: Yes, with the restrictions and

endorsements of Dr. Fink's suggestion for a dose

escalating study.



DR. FINK: Yes.

DR. STOLLER: Yes, again with the

contingencies as stated.


DR. DYKEWICZ: The final vote on question

six about recommending approval of Advair Diskus is

six yes, two no, one abstention. Are there any

final comments that any members of the committee

want to make, maybe about stipulations about

product labeling, additional safety studies that

were recommended, or have you all articulated your

concerns previously? Dr. Joad?

DR. JOAD: Was the final labeling going to

say chronic obstructive bronchitis? Did we say

that? I thought we had said that with the efficacy


DR. DYKEWICZ: Maybe what we should do is

get a consensus from the committee, but Dr. Meyer?

DR. MEYER: At the risk of offending

folks, I think we heard that, maybe not as a

consensus but as a very strong opinion and I think

we will take that under very strong advisement.

I did want to make a closing statement. I

think I heard some folks earlier talking about

framing linguistic "buts." I think we have framed

our physical ones here in these seats.


This has been a very useful discussion.

In all seriousness, I thank you very much for all

your advice and very careful thought. For our

guests, for Dr. Bone, Dr. Wise and for Dr.

Malozowski, I am specially thankful for your

expertise in these matters, and again thank the

committee for their time today.

DR. DYKEWICZ: As chair, I would again

like to thank everyone for their attentiveness and

for their participation. Have a good evening. We

are adjourned.

[Whereupon, at 5:10 p.m., the proceedings

were recessed, to resume on Friday, January 18,

2002 at 8:00 a.m.]

- - -