1

DEPARTMENT OF HEALTH AND HUMAN SERVICES

UNITED STATES FOOD AND DRUG ADMINISTRATION

CENTER FOR DRUG EVALUATION AND RESEARCH

Cardiovascular and Renal Drugs Advisory Committee

Meeting

THURSDAY, FEBRUARY 24, 2005

8:00 a.m.

Food and Drug Administration

CDER Advisory Committee Conference Room

Room 1066

5630 Fishers Lane

Rockville, Maryland 20005

P A R T I C I P A N T S

Steven E. Nissen, M.D., F.A.C.C., Chair

Lt. Cathy Groupe, RN, BSN, Executive Secretary

Committee Members:

Blase A. Carabello, M.D.

Susanna L. Cunningham, Ph.D., Consumer

Representative

William R. Hiatt, M.D.

Frederick J. Kaskel, M.D., Ph.D.

John F. Neylan, M.D., Industry Representative

Thomas Pickering, M.D., D.Phil.

Ronald Portman, M.D.

John R. Teerlink, M.D.

Special Government Employee Consultants (Voting):

Jonathan Sackner-Bernstein, M.D.

Ralph B. D'Agostino, Ph.D.

FDA Participants:

Robert Temple, M.D.

Norman Stockbridge, M.D.

C O N T E N T S

Call to Order and Introductions,

Steven E. Nissen, M.D., Chair 4

Conflict of Interest Statement,

Lt. Cathy Groupe, BSN, Executive Secretary 6

Welcome and Comments, Norman Stockbridge, M.D.,

Acting Director, Division of Cardiac and

Renal Drug Products 9

Sponsor Presentations:

Regulatory Overview, Cindy Lancaster, M.S.,

M.B.A., J.D., AstraZeneca, L.P. 10

Background and Rationale, James B. Young,

M.D., Cleveland Clinic Foundation 17

ACE Inhibitor Choice, Dose and Drug Utilization,

John J.V. McMurray, M.D.,

Glasgow University, Scotland 27

Efficacy, Mark A. Pfeffer, M.D., Ph.D.,

Brigham and Women's Hospital, Boston 40

Safety, James Hainer, M.D., M.P.H.,

AstraZeneca, LP 58

Benefit/Risk Summary, James B. Young, M.D.

Cleveland Clinic Foundation 67

Discussion, Marc A. Pfeffer, M.D., Ph.D.,

Brigham and Women's Hospital, Boston

Questions from the Committee 71

Committee Discussion and Questions 151

P R O C E E D I N G S

Call to Order and Introductions

DR. NISSEN: I think we have all our

committee members. My name is Steve Nissen. I am

a cardiologist in the Cleveland Clinic, and we are

going to do some introductions first so that you

all know who is on the committee. Let's start with

John, over there.

DR. NEYLAN: Yes, I am John Neylan. I am

the industry representative on the committee, from

Wyeth Pharmaceuticals.

DR. CARABELLO: Blase Carabello, a

cardiologist from Houston.

DR. HIATT: Bill Hiatt, University of

Colorado, vascular medicine.

DR. PICKERING: Tom Pickering,

hypertension, Columbia University Medical School.

DR. PORTMAN: Ron Portman, pediatric

nephrologist from the University of Texas in

Houston.

DR. TEERLINK: John Teerlink, heart

failure specialist from University of California

San Francisco and San Francisco VA.

LT. GROUPE: Cathy Groupe, the executive

secretary for the Cardiac and Renal Drugs Advisory

Committee.

DR. KASKEL: Rick Kaskel, pediatric

nephrologist, Albert Einstein College of Medicine.

DR. SACKNER-BERNSTEIN: Jonathan

Sackner-Bernstein, cardiologist from North Shore

University Hospital in New York.

DR. D'AGOSTINO: Ralph D'Agostino,

biostatistician from Boston University and the

Framingham study.

DR. STOCKBRIDGE: I am Norman Stockbridge.

I am the Acting Director of the Division of

Cardiorenal Drug Products. To my right would be

Dr. Temple, but it is completely unreasonable for

us to start on time and expect him to be here.

[Laughter.]

DR. NISSEN: Dr. Temple usually is awake

by ten o'clock in the morning so I expect him

later. Lt. Cathy Groupe is going to read the

conflict of interest statement.

Conflict of Interest Statement

LT. GROUPE: The following announcement

addresses the issue of conflict of interest with

respect 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 and

all financial 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 exceptions:

In accordance with 18 USC Section

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

following participants, Dr. Ralph D'Agostino for

consulting for two competitors on unrelated matters

for which he receives less than $10,001 per year

per firm; Dr. William Hiatt for consulting and

speaking for a competitor on unrelated matters for

which he receives between $10,001 to $50,000 per

year per firm; Dr. Steven Nissen for consulting for

the sponsor and for four competitors on unrelated

matters for which he receives less than $10,001 per

year per firm; Dr. Thomas Pickering for consulting

and speaking for two competitors on unrelated

issues for which he receives less than $10,001 per

year per firm; Dr. Ronald Portman for consulting

for two competitors on unrelated issues for which

he receives less than $10,001 per year from one

firm and between $10,001 to $50,000 per year from

the other firm; Dr. Sackner-Bernstein for

consulting for a competitor on a related matter

which was general in nature for which he receives

less than $10,001 per year.

In accordance with 18 USC Section

208(b)(1) a full waiver has been granted to Dr.

John Teerlink for his role as an independent and

blinded adjudicator, consulting and steering

committee member on unrelated matters for two

competitors. He receives from $10,001 to $50,000

per year from one firm and less than $10,001 per

year from the other; for his role as an endpoint

committee member on a related matter for a

competitor for which he receives from $10,001 to

$50,000 per year; for his role as a

sub-investigator on a related matter for a

competitor for which the contract was less than

$100,000 per year.

A copy of the waiver statements may be

obtained by submitting a written request to the

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

of the Parklawn Building.

In the event that the discussions involve

any other products or firms not already on the

agenda for which an FDA participants 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.

We would also like to note that Dr. John

Neylan has been invited to participate as an

industry representative acting on behalf of

regulated industry. Dr. Neylan is employed by

Wyeth Research.

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.

DR. NISSEN: Dr. Stockbridge, I believe

you have some opening comments.

Welcome and Comments

DR. STOCKBRIDGE: The first thing I wanted

to say was sort of in the form of a public service

announcement. Last week someone, using the name of

a Cardiorenal Advisory Committee member but

claiming to be from the Division of Cardiorenal

Drug Products, made calls to several parties, one

on an investigator side and another a

pharmaceutical company, clearly trying to get some

kind of information. If anyone else ever hears

about a case like that I would like to suggest that

you bring it to my attention so we can coordinate

the investigation of any new case with the current

one.

The other thing I wanted to say is that

two days ago the division took an action to approve

candesartan for use in heart failure and I have

made sure that everybody, this morning at least,

got the relevant parts of the labeling that

resulted largely from the CHARM-Alternative trial.

So, the question about whether candesartan works in

heart failure is not what you have been invited to

comment on. Instead, there is a fairly simple

question--it only takes three pages for me to ask

it--

[Laughter.]

--about use of candesartan together with

an ACE inhibitor. Thank you.

DR. NISSEN: Thanks, Norman. Let's then

just proceed to the sponsor presentation. If it

pleases the committee, I think what we would like

to do is let the sponsor go ahead and go through

their presentation and then maybe hold all the

questions together because it is going to be, I

think, easier to integrate everything. However, if

anybody has burning questions after any of the

individual presentations, please let me know and we

will try to make sure you get clarification.

Sponsor Presentation:

Regulatory Overview

MS. LANCASTER: Good morning, Mr.

Chairman, members of the committee, members of FDA

and ladies and gentlemen. I am Cindy Lancaster,

and on behalf of AstraZeneca I would like to thank

the division and the committee for giving us the

opportunity to present the results of our clinical

program for candesartan cilexetil in heart failure.

Atacand has been approved since 1997 for

the treatment of hypertension and, more

specifically, approved in the United States in

1998. Atacand is currently marketed in 92

countries and to date we have 20 million

patient-years of exposure available.

Let me begin by sharing a list of

individuals who are here today to participate in

these proceedings. These are the sponsor

representatives. We have also invited our expert

external advisers to share their experiences with

the heart failure clinical program. Dr. Pfeffer

served as a co-chair on the CHARM executive

committee. Dr. Young and Dr. Dunlap served as

CHARM U.S. national leaders. Dr. McMurray served

as he principal investigator for the CHARM-Added

trial. Dr. Granger served as the principal

investigator for the CHARM-Alternative trial. They

also served as members of the CHARM executive

committee.

In addition, Dr. Lewis, Dr. McLaughlin,

Dr. Kronmal and Dr. Hennekens are also available to

assist today. Dr. Hennekens is here in his role as

the chair of the CHARM data and safety monitoring

board.

To set the stage for the forthcoming

presentations, here is a brief history as of 1996

of the product's development and key previous

interactions with the FDA in regard to the heart

failure clinical program. Three pilot studies were

conducted to help identify the optimum dose and

evaluate neurohormonal effects, LV systolic volume

and tolerability of the 32 mg high dose under the

U.S. IND, prior to the initiation of the CHARM

program.

In 1998 AstraZeneca met with the Division

of Cardiorenal Drug Products to discuss the design

of the CHARM program, and gained agreement that the

program would support a claim for heart failure.

The CHARM program was initiated in 1999, and in

March, 2003 we completed the program. Later in

2003 a pre-sNDA conference was held with FDA to

discuss the content and format of the application.

The heart failure supplement was then submitted to

the FDA in June, 2004 and a priority review was

assigned for CHARM-Added.

An approvable letter was issue by the FDA

at the end of December for the CHARM-Added study.

As Dr. Stockbridge stated this morning, on Tuesday

of this week the division granted approval for the

use of candesartan in heart failure primarily based

on CHARM-Alternative. As such, today we are here

to specifically discuss CHARM-Added and approval

based on the results from this particular study.

To that point, let me first provide a little

background on the CHARM program.

CHARM-Alternative and CHARM-Added were

part of the most comprehensive trial program

completed to date with this class of drugs for

heart failure. The CHARM program consists of three

separate but complementary randomized,

double-blind, placebo-controlled, parallel group

studies including 7,601 patients.

Alternative was conducted in patients with

ejection fraction less than or equal to 40 percent

and not on an ACE inhibitor. This Tuesday's

approval was primarily based on this study. Added,

which is the focus of today's discussion, was

conducted in patients with ejection fraction less

than or equal to 40 percent and receiving an

optimized dose of ACE inhibitor. Preserved was

conducted in patients with preserved left

ventricular systolic function.

The primary endpoint for each trial was CV

death and heart failure hospitalizations. The data

demonstrated a statistically significant and

clinically important benefit for candesartan in the

low ejection fraction studies, Added and

Alternative. The primary endpoint for Preserved

was not statistically significant. These results

from Alternative, supported by the Added study,

formed the basis of Tuesday's approval by FDA for

candesartan in heart failure. Additionally, to

date candesartan has been approved in 18 countries

for the treatment as add-on therapy based on

CHARM-Added or without an ACE inhibitor based on

CHARM-Alternative.

Specifically, in the United States the

indication approved on Tuesday states Atacand is

indicated for the treatment of heart failure (New

York Heart Association class II-IV and ejection

fraction less than or equal to 40 percent) to

reduce the risk of death from cardiovascular causes

and reduce hospitalization for heart failure.

In addition, the clinical trial section

mentions CHARM-Added as a supportive study in the

first sentence of the text you see on the screen.

Also note there was a 15 percent lower risk of

cardiovascular mortality based on both

CHARM-Alternative and CHARM-Added together.

Furthermore, symptoms of heart failure, as assessed

by New York Heart Association functional class,

were also improved.

Based on CHARM-Added, AstraZeneca requests

approval for candesartan as add-on therapy when a

patient is already receiving an ACE inhibitor.

CHARM-Added was designed to allow an investigator

to optimize the dose of ACE inhibitor treatment on

an individual patient basis when either placebo or

candesartan is used for the treatment of heart

failure. Treatment resulted in a statistically

significant and clinically important benefit when

candesartan was added to an evidence-based dose of

an ACE inhibitor.

The FDA has posed the question does

CHARM-Added provide compelling evidence that

candesartan should under some circumstances be

recommended for use in patients on an ACE

inhibitor.

To help answer this and other questions

posed today, we have conducted supplemental

analyses, the results of which will be presented

here to assist with these proceedings. Next, Dr.

Young will present the rationale for use of ARBs in

heart failure. The ARBs and ACE inhibitors have

distinct and complementary mechanisms, and data

from pilot studies are supportive of the beneficial

effects demonstrated from treatment with

candesartan added to an ACE inhibitor.

Following that, Dr. McMurray will present

information on the selection of the recommended

dose of an ACE inhibitor in CHARM-Added. Dr.

Pfeffer will then provide a summary of efficacy for

CHARM-Added as well as the analyses for maximum ACE

inhibitor doses defined by the FDA. Dr. Hainer

will present safety information. Dr. Young will

then present the benefit/risk profile. That will

conclude our formal presentation. Now, Dr. Young?

DR. NISSEN: Any clarification issues for

anybody or can we go ahead and move on? If not,

let's do it.

Background and Rationale

DR. YOUNG: Thank you, Cindy. Dr. Nissen,

ladies and gentlemen of the panel, the FDA and the

audience, it is an honor for me to be here today so

we can all reconsider an extraordinarily important

healthcare challenge and review data which supports

a new pharmacotherapeutic strategy for chronic

heart failure.

I need not detail the devastating impact

of chronic heart failure's morbidity and mortality.

Particularly concerning is the high prevalence of

this syndrome and the number of hospitalizations

precipitated annually which is increasing, and in

those patients associated with even higher

mortality rates during follow-up.

This survival data from the Framingham

cohort study is important as it demonstrates that

though some progress has been made over time heart

failure mortality is still great. Even in the

so-called modern era of heart failure, the last

decade, which would have included ACE inhibitors

and to a lesser extent beta-blockers, the 5-year

survival rate for men with CHF is still only about

40 percent and women fare only slightly better.

Germane to today's CHARM program

presentation is question 1 from the FDA, and

specifically question 1.4, are ACE inhibitors and

ARBs sufficiently different that CHARM-Added can

support use of candesartan with ACE inhibitors?

To answer that question we need to

consider the pathophysiology of heart failure and

the relationship of ACE inhibitors and ARBs to the

renin-angiotensin-aldosterone system. It had been

gratifying to see the insight gained over the last

30 years into the pathophysiology of heart failure

and this has helped us design better therapies.

Particularly important is understanding

implications of the renin-angiotensin-aldosterone

system.

Indeed, the vast majority of drugs

beneficial in this system, including beta-blockers,

attenuate adverse effects of angiotensin-II.

Emphasizing that point is this RAAS cascade. I

know everyone here has their own favorite RAAS

cascade. This happens to be mine. Here we can see

the potentially detrimental effects of

angiotensin-II effected through the AT-I receptor,

as well as some putative beneficial effects of

angiotensin-II effected through the AT-II receptor,

specifically increasing kinin and nitric oxide

activity.

These observations have significant

implications when we consider ACE inhibitor and ARB

use in heart failure, and particularly their

combination. First, angiotensin-converting enzyme

is not the only molecule affecting production of

angiotensin-II. During long-term ACE inhibitor

prescription chymase activity, for example, can

increase levels of angiotensin-II even at doses of

ACE inhibitors which completely inhibit this

enzyme.

ACE inhibitors have another important

effect. They are bradykinin potentiating factors.

Indeed, when first isolated from the Brazilian pit

viper venom, the molecule was labeled BPF. It is

also important to remember that candesartan, the

agent of focus today, is a selective angiotensin-II

type I receptor blocker that is tightly bound and

long acting.

Again keeping in mind the last diagram, we

can illustrate how ACE inhibitors mediate benefit

in heart failure remembering the BPF and ACE escape

issues. Here we see the ARB effects which result

in more specific and complete blockade of the

angiotensin-II type I receptor. Here, the

rationale for combination ACE inhibitor and

candesartan therapy is the fact that angiotensin-II

produced by chymase activity will be attenuated

without abrogation of ACE inhibitor BPF effects

while allowing potentially beneficial effects of

AT-II receptor activity.

There is robust basic scientific evidence

that supports these concepts. For example, in

canine heart failure models ACE inhibitor and ARB

combination improved hemodynamics, collagen volume

fraction and mRNA for collagen 1 and 3 compared to

either agent alone.

In Pfeffer model rats with heart failure

the combination of valsartan and fosinopril was

more effective in suppressing myocardial remodeling

assessed by collagen production and decreased

infarct size, while valsartan and benazopril

improved more subsequent left ventricular

hypertrophy and lusitropic properties noted in

these pathophysiologic models. In obese and

hypertensive rats, blood pressure, left ventricular

hypertrophy and renal function were improved more

with the ACE inhibitor/ARB combination than with

use of either agent alone.

We also see clinical evidence that a

combination of an ACE inhibitor and an ARB could be

beneficial. For example, this now classic report

of the ACE inhibitor escape phenomenon demonstrates

the time-dependent increase of angiotensin-II

despite almost complete reduction of plasma ACE

activity over time.

This is one example of several very

elegant demonstrations of a complicated interaction

between ACE inhibition and AT-I receptor blockade

in heart failure patients. This experiment

specifically focused on the contribution of

bradykinin to vasodilation in patients on enalapril

compared to losartan. Specifically, all subjects

received an infusion of a bradykinin receptor

antagonist before an ACE inhibitor or ARB was

given.

This is a complicated diagram but focus on

the change in mean arterial pressure and change in

systemic vascular resistance. The top line is the

ACE inhibitor; the middle line the ARB. What this

study shows is that in patients with chronic heart

failure infusion of a bradykinin receptor

antagonist attenuates the blood pressure lowering

effects of long-term enalapril therapy when

compared with losartan treatment indicating loss of

the BPF activity of the ACE inhibitor.

Additional information has also become

available supporting the hypothesis that an ACE

inhibitor/ARB combination will produce incremental

benefit with respect to significant clinical

outcomes, albeit in a non-cardiac vascular bed.

The first three small clinical studies listed on

this slide explored in type 1 and 2 diabetics the

value of adding valsartan, candesartan or

irbesartan to substantive doses of an ACE inhibitor

and consistently demonstrated, when a crossover

trial design was used, significantly greater

reduction in proteinuria with the contribution of

an ACE inhibitor and ARB.

The COOPERATE trial was a small but

significant clinical outcome study in nondiabetic

renal insufficiency patients when a maximally

effective dose of trandolapril, and this was

determined as the dose above which there was no

further reduction in proteinuria, was combined with

100 mg of losartan. There was significantly

greater reduction in proteinuria with the drugs

combined, but most important, with the combination

there were significantly fewer primary endpoints of

combination of developing end-stage renal disease

or a doubling of creatinine.

With respect to clinical effects of

combination of ACE inhibitors and ARB in heart

failure, a ValHeFT pilot study demonstrated that

adding valsartan to 20 mg of lisinopril effected

more reduction in some hemodynamic parameters.

RSOLVe was a very important pilot study of

candesartan in heart failure patients. Its primary

purpose was to determine if this ARB in varying

doses could be added safely to 20 mg of enalapril

and then if long-acting metoprolol could be added

to the ACE inhibitor/ARB combination.

Exploratory efficacy endpoints were

included and this slide demonstrates the important

finding that BNP dropped significantly in the

combination group at the 43-week follow-up point.

The combination of candesartan and enalapril also

more favorably affected aldosterone and

angiotensin-II levels, not shown on this slide.

The combination ACE inhibitor/ARB

pharmacologic effects seemingly translated into

greater beneficial cardiac remodeling, demonstrated

by this data also from the RESOLVe pilot study.

Candesartan alone and enalapril alone had about the

same effect on left ventricular end diastolic and

end systolic volumes during the course of this

trial, whereas, a more substantial effect was

apparent with the combination.

Another small clinical study demonstrated

the additive effects of ACE inhibitor and ARB on

heart failure symptoms and exercise capacity. Here

we see a significant increase in peak exercise

oxygen uptake and improvement in New York Heart

Association symptomatic classification when 50 mg

of losartan was added to either lisinopril and

enalapril.

Setting the stage for the CHARM program,

and particularly the CHARM-Added study is this

clear imperative to develop better strategies for

heart failure treatment. Certainly, attenuating

the adverse effects of RAAS is important. There is

now substantial preclinical and clinical evidence

that the combination of an ACE inhibitor and ARB

will be effective interventions. This is supported

by clinical outcomes data in diabetes and chronic

renal insufficiency patients, as well as

hemodynamic, neurohormonal, cardiac remodeling,

symptomatic and exercise changes in heart failure

patients.

To discuss in more detail the rationale

for very important design characteristics of the

CHARM-Added study is Prof. John McMurray of the

University of Glasgow, in Scotland. John is the

global principal investigator for the CHARM-Added

trial. As we consider in more detail the

CHARM-Added program design, Dr. McMurray will

specifically address the issue of baseline ACE

inhibitor choice, dose and utilization in our

study. This will address several additional

questions posed by the FDA. Then Dr. Pfeffer will

subsequently present our outcomes data. So, if

there are no clarification questions, we can turn

to John to deal with the ACE inhibitor issue.

DR. NISSEN: Can we move on? Okay.

ACE Inhibitor Choice, Dose and Drug Utilization

DR. MCMURRAY: Mr. Chairman, ladies and

gentlemen, Dr. Young has explained to you that ARBs

and ACE inhibitors have pharmacologically distinct

mechanisms of action. He has explain to you the

scientific rationale for combining the two. He has

shown you the mechanistic data to show that there

may be benefit from using the two different types

of drugs together. But to show that there is an

important improvement in clinical outcome when you

combine the two drugs you obviously have to conduct

a trial like CHARM-Added, and what I want to

consider is the way we approached this question

when we designed CHARM-Added. In particular, I

want to show you the approach we took to ensuring

that the background dose of ACE inhibitor was

optimized because to test this hypothesis in an

outcomes study it was important that candesartan

was added to a good dose of an ACE inhibitor, to

optimum background ACE inhibitor therapy.

So, in line with the questions that we

received from the agency, I am going to speak to

how we did this in the CHARM protocol, and I am

going to tell you how we tried to optimize

background ACE inhibitor dose, and I am going to

show you what our investigators actually did. So,

I am going to talk about which drug and what dose.

I am going to show you the evidence-based trials on

which we based our recommendations and then also

address a question raised by the agency which is

about higher than evidence-based doses. I will

come back to that at the end of my presentation.

So, what did we do when we designed

CHARM-Added? What did we write in the protocol?

What did we tell our investigators at all the

meetings that we spoke at? Well, at the time that

we were designing the study there were five ACE

inhibitors that you could call evidence-based. In

other words, five ACE inhibitors that have been

used in large-scale clinical outcomes

studies--captopril, ramipril, trandolapril,

lisinopril and enalapril. These are the five ACE

inhibitors that we recommended to our investigators

that ideally they should use in their patients.

What about dose? What did we say about

dose? Well, here are some words from the protocol.

I am sorry, this is quite a long slide to read but

I will just draw your attention to the last

sentence. We say here the investigators are

reminded that these trials--so we referred to the

trials I just mentioned--had target ACE inhibitor

doses higher than those commonly used in clinical

practice. We have an appendix, which I will come

to, which showed the doses. We also said at that

time that the recently reported ATLAS trial, which

compared a very low dose of ACE inhibitor to a

higher dose, that trial suggested that there is

more morbidity benefit from using a higher dose of

ACE inhibitors. So, we were very strong. We felt

that to test the hypothesis it was very important

that our investigators used the target doses, if

possible, of the ACE inhibitors that had been shown

to be of benefit in the large randomized trials.

You can see here those trials and the target doses

that were recommended. These were what were put in

the protocol. These were what we spoke about at

the investigator meetings.

So, that is what we planned. What

actually happened? Well, in addition to those two

things we also asked, once the investigators had

individually optimized ACE inhibitor dosing in

their patients, that the patients should be on a

stable dose of an ACE inhibitor for at least 30

days before randomization.

So, I want to now look at what our

investigators actually did. Well, if you remember,

I said there were five ACE inhibitors proven to be

of benefit in large-scale randomized trials. We

were pleased to find that, in fact, in 80 percent

of the patients in CHARM-Added those five proven

ACE inhibitors were the ones that were used.

The agency also recently asked us to look

at all approved ACE inhibitors. In fact, there are

two additional ACE inhibitors. There are seven

FDA-approved ACE inhibitors for the treatment of

heart failure. In fact, it was 90 percent of

patients in CHARM-Added who received an

FDA-approved ACE inhibitor. So, that is something

about the drugs that were used.

What about the doses that were used by the

CHARM-Added investigators? Well, we asked our

investigators to tell us that they actually felt

that they had tried to individually optimize the

dose of ACE inhibitor. We did that by asking them

to check a box before randomization on the CRF. We

wish we had collected more information about this

but we didn't.

But I will show you what I believe is

evidence to support the view that our investigators

did a good job in trying to use evidence-based

doses of ACE inhibitor. On this slide you see the

mean dose of ACE inhibitor used in those landmark

trials. You also see the mean dose of the same ACE

inhibitors used in CHARM-Added. For example, in

the SOLVD treatment trial the mean dose achieved

was 16.6 mg. In CHARM-Added the mean dose of

enalapril used was 17 mg. Broadly, I think this

slide shows that our investigators generally did

achieve the sorts of doses of ACE inhibitor seen in

the forced titration trials.

I am just going to focus on enalapril a

little bit more, and the reason I am going to do

that is two-fold. Firstly, enalapril is by far the

most evidence-based ACE inhibitor in heart failure

and, secondly, it is the one where we have the most

information about doses achieved during forced

titration.

You see on this slide all the trials that

force titrated enalapril in heart failure. You see

the mean daily dose achieved which was generally

between 15-18 mg, and in CHARM-Added our patients

received 17 mg and enalapril was the most commonly

used ACE inhibitor in CHARM-Added.

Perhaps an even more important slide I

think is this one because it shows you the ACE

inhibitor doses used in other recent important

heart failure trials looking at treatments given in

addition to an ACE inhibitor. So, on this slide

you see two of the recent key beta-blocker trials

and you also see the RALES trial and you see the

baseline dose of ACE inhibitor used in these

trials. In every case for these key ACE inhibitors

the CHARM-Added investigators had their patients on

a larger dose of ACE inhibitor than in these other

trials. We think that that tells us that our

investigators did heed our advice; did follow the

instructions in the protocol; did listen to what we

said at the investigators meetings.

Here is another important slide and it

really goes to the heart of what we were trying to

do in CHARM-Added. Here you see all the evidence

that we can find about the use of ACE inhibitors in

ordinary clinical practice in the community and in

hospitals. You can see again that the patients in

CHARM-Added got much higher doses of ACE inhibitor

than were used in ordinary clinical practice.

I want to now turn to the interesting

question raised by the agency, what if we were to

go to even higher doses of ACE inhibitors than

those proven to be of benefit in the clinical

trials? That is actually quite a difficult thing

to look at because though there are many

dose-response study for ACE inhibitors, most of

these haven't addressed that question. What they

have looked at is actually very small doses or

medium doses compared to evidence-based doses.

They haven't looked at the question that we were

asked, which is what happens if you go above

evidence-based doses?

It is interesting to think about that

question because the first part of it is really is

it possible to do that? Can patients get to these

much higher doses? Secondly, even if they do, is

there additional benefit? Well, I am a heart

failure specialist and I know there are other

people here who are, and we know that in our

practice you can get some people to bigger doses

than have been used in the key landmark trials, but

I think individually it is very hard to get a

handle on how many patients, what proportion of

your patients can get above those doses.

It is interesting just to note that in the

SOLVD treatment trial only about half the patients

got 10 mg twice a day of enalapril. In the

CONSENSUS study it was only about a fifth of

patients who actually got up to 20 mg twice a day.

The one trial in the literature that has actually

tested this question is shown on this slide. That

is a study that compared an evidence-based dose of

enalapril, 20 mg a day, to a much larger dose, 60

mg a day. You can see the details of this trial

here. You can see that about a third of patients

could get this larger dose of enalapril. But what

is of interest is that there was no statistically

significant or clinically important difference in

blood pressure, heart rate, ejection fraction or

NYHA class in the group who got the larger dose of

enalapril than in the group who got the

evidence-based dose of enalapril. There was also

no significant difference in any of the clinical

outcomes measured, though this was a relatively

small trial but just so you can see what happened.

Here is the endpoint of death or admission to

hospital with worsening heart failure. You can see

the two treatment groups and I think you will agree

that in this small study there is no difference

between the two treatment groups.

To summarize, Mr. Chairman, ladies and

gentlemen, in CHARM-Added we believe that our

patients did receive an evidence-based ACE

inhibitor; 80 percent of them got a proven ACE

inhibitor. We believe that they did get doses

comparable to those obtained in the forced

titration studies, for example 17 mg of enalapril.

The doses patients in CHARM-Added got were much

higher than doses used in other recent add-on

trials, and clearly higher than doses used in

ordinary clinical practice. And, I have shown you

what little evidence there is about whether going

to higher dose of ACE inhibitor has any additional

benefit.

So, to conclude, in our protocol and at

our investigational meetings we advocated the use

of evidence-based ACE inhibitor treatment, and we

believe our investigators did do that. In other

words, we believe that CHARM-Added did test the

hypothesis of whether adding an ARB to an

evidence-based dose of ACE inhibitor would provide

further clinical benefit, and my colleague, Dr.

Pfeffer, will speak to the evidence that that is

the case when he presents the efficacy findings

from the CHARM-Added study. Thank you very much.

DR. NISSEN: Any clarification? Yes,

Bill?

DR. HIATT: Just a quick question, when

you presented the dose of ACE inhibitors how

different was the median from the mean?

DR. MCMURRAY: The medians were slightly

smaller for one or two ACE inhibitors but they were

generally similar.

DR. HIATT: So, the mean data were

representative of the distribution of use--

DR. MCMURRAY: They were.

DR. NISSEN: Before we go on, we have had

two people join us a little bit late so perhaps

they could introduce themselves. Dr. Temple?

DR. TEMPLE: Bob Temple, regularly late,

Office Director.

DR. CUNNINGHAM: Susanna Cunningham,

University of Washington.

DR. NISSEN: And you might tell them what

your role is here.

DR. CUNNINGHAM: I am the consumer

representative on the committee.

DR. NISSEN: Thank you very much. Let's

move on unless there are other questions of

clarification.

DR. TEMPLE: I have a question.

DR. NISSEN: Yes, sir?

DR. TEMPLE: The point was made that the

doses used in CHARM-Added were similar to doses

used in a variety of add-on studies. But our view

was that that isn't really relevant unless it is

another drug that works the renin-angiotensin

system. The question here is whether it is sort of

like giving another extra dose of your ACE

inhibitor. So, the fact that RALES used lower

doses really doesn't matter particularly.

DR. MCMURRAY: I understand that, Dr.

Temple. The dose of ACE inhibitor in CHARM-Added

was larger than in any of the other add-on trials.

We had the same view that you do. I mean, we tried

to design a study to test the question and I was

only showing that slide to try to emphasize that I

think our investigators did try and do better,

certainly have done better than in ordinary

clinical practice and actually did better than

other investigators in other clinical trials.

DR. TEMPLE: Yes, I take that point but

the immediate question is whether you are just

adding a little more of the same. So, it really

only matters in the ACE inhibitor trials.

DR. NISSEN: Other clarifications?

[No response.]

Fortunately, I visited Scotland so I

understood every word without English translation.

DR. MCMURRAY: Thank you very much.

[Laughter.]

Efficacy

DR. PFEFFER: Mr. Chairman, members of the

panel, ladies and gentlemen, I am glad to be

representing the CHARM investigators to present the

efficacy data, and I will be concentrating on

CHARM-Added. But I would first like just to remind

you that this was a program of research, and you

met Dr. McMurray who led the CHARM-Added, which I

will be talking about. Dr. Granger is here. He

led CHARM-Alternative. Dr. Slim Yusuf led the

CHARM-Preserved, and I co-chaired this with Dr.

Carl Swedberg.

The program of research had some

interesting aspects which relate to CHARM-Added

particularly. By definition, by protocol the

program was three individual projects, each asking

its own question in its own population; each with

its own sample size; and each was united under the

banner of the same investigator, same form, same

dose titration; same committees. But one of the

aspects of the protocol I call your attention to is

that by definition the protocol stated that we

would follow the last patient randomized for a

minimum of two years. That means the greatest

exposure we have is in CHARM-Added for the longest

observation of those on the experimental

medication.

For each of the projects--but we can

concentrate on CHARM-Added--it is the same; the

primary endpoint was cardiovascular mortality or

hospitalization, unplanned hospitalization for

management of heart failure, all adjudicated

centrally.

The secondary endpoints for each of the

projects was to look at all-cause mortality or

hospitalization for heart failure, and another

prespecified secondary endpoint was to add nonfatal

MI to our primary endpoint of CV mortality or

hospitalization for heart failure.

The dose titration regimen for all the

protocols was the same. The investigator had the

option, after assessing patient status, of starting

either at the first step or the second step. So,

effectively, they could have started either with 4

mg or 8 mg of candesartan or matching placebo in a

blinded fashion. Investigators were asked to

titrate at 2-week intervals according to clinical

standards and whether or not they wanted to

proceed. As you can see, 71 percent of our placebo

patients were able to be titrated to the full dose

and 61 percent of the candesartan, which is quite

comparable to other trials with forced titration.

The analyses that I will present within

our analysis plan--and if I leave our analysis plan

I will specify that--were all intention-to-treat.

It is all time to first event for the primary and

secondary endpoints. We will be using log rank

test for comparisons; the Cox proportional hazard

models to estimate the effect size. You will be

seeing effects over time as a Kaplan-Meier. For

the secondary endpoints we are using a hierarchical

closed test procedure.

Inclusion criteria for the whole program

were symptomatic heart failure patients above the

age of 18, and they had to be stable for at least 4

weeks, and II-IV. For the CHARM-Added we had the

additional criteria that if a patient was class II

they could be admitted but they had to have a

history of a cardiac hospitalization in the

previous 6 months.

For the program, patients were to be

excluded if their creatinine was greater than 3;

potassium greater than or equal to 5.5; and known

contraindications to inhibitors of the

renin-angiotensin system or use of an ARB.

I think Dr. U's report demonstrates that

we did achieve balance in the randomization process

so I just want to highlight that approximately 17,

18 percent of our patients were over 75 years of

age and 21 percent were female. The predominant

New York Heart Association class was III. The

background of co-morbid diseases is well-known to

this group, with about a third known diabetics;

hypertension in about a half; and atrial

fibrillation in just over a quarter; and a prior

myocardial infarction in about 55 percent.

Concomitant medications is an important

point for any study. Our enrollment started in

1999 and ended in 1999 for this trial. Around 1990

were very exciting times with the proof of

beta-blockers continuing to mount. As I mentioned,

Dr. Swedberg was one of the co-chairmen and he has

been on the vanguard of beta-blocker use. So, our

investigators were well on top of the wave at the

time so for a study randomizing in 1999 I think we

have the highest use of a beta-blocker at 55

percent. We did allow the use of spironolactone at

the physician's discretion, and our exposure will

be on 17 percent on patients.

Here are the results of the primary

endpoint. CV death or hospitalization for heart

failure is reduced by 15 percent, showing the

confidence interval here. This is a significant

reduction. This relative risk really represents

44/1000 events reduced, and that event is either a

CV death or a hospitalization for heart failure.

The number needed to treat over the time course

would be 23 to prevent either a CV death or first

hospitalization for heart failure.

I will just use this opportunity to say

that this is the first hospitalization for heart

failure and, as this group knows, this is a

revolving door. Once a person has that, they are

much more likely to come back again. Subsequent

total hospitalizations will be discussed.

Well, here are the components of the

endpoint. The endpoint was a composite of CV death

or hospitalization for heart failure. This is

basically what I was showing on the Kaplan-Meiers

but if we look at the contribution of both

components, they are a 16 percent reduction in risk

of CV death and a 17 percent reduction in the risk

of a hospitalization for heart failure. As

everyone knows, if you add the components, it

exceeds that because a person can have a

hospitalization for heart failure and subsequently

die, and that was a common finding more often in

the placebo group.

Here are the components looked at

individually. Here is the Kaplan-Meier for CV

death. We are also showing the non-CV death but

the impact on CV death over time--I have shown you

that data. Here is the impact on hospitalization

and this, of course, is skewed by the survivor

bias. Obviously, there were more placebo patients

at risk to have this but despite that fewer

candesartan patients were hospitalization for heart

failure, at least a first hospitalization.

Our secondary endpoints, prespecified,

were to look at all-cause mortality, not the

adjudicated but all-cause and add that to the

hospitalization for heart failure. As you can see,

this secondary endpoint was also achieved and the

components of this are also shown where both

contribute to this important secondary endpoint.

Another prespecified secondary endpoint

was to add nonfatal myocardial infarctions, and we

add an equal number. We add 13 and 19 to the

primary endpoint--I may have this wrong; I can't do

it from this one. We add very few--

[Laughter.]

--equal numbers, but the point is how few

it is relative to the primary endpoint.

Subgroups. We do this with caution and I

am showing 13. I could show many more. The

analysis plan had several others. These are the

ones we thought would be of interest to the

clinical audience. Thirteen are on this. There

were no interactions, which allows me to say that

the benefit we have been discussing was not

modified by these subgroups.

There was really at the time, when we

first analyzed our data and presented our data in

the year 2003, clinically a very major issue

addressed, and that was beta-blockade. A study

prior to ours had given an indication from a

subgroup analysis of the potential safety issue.

With that knowledge, our data monitoring board

chaired by Dr. Hennekens, and our investigators and

the world clearly wanted to know what was the

exposure with beta-blockers.

I will remind you that in CHARM-Added

everyone is on an ACE inhibitor, 100 percent. So,

when we talk about beta-blocker, it is ACE

inhibitor, beta-blocker, plus candesartan or

placebo. Here is the experience. There was no

signal of loss of efficacy so the effectiveness was

not modified by the presence or absence of a

beta-blocker.

This is a safety analysis--was there a

mortality signal of using this now triple

therapy--the so-called triple therapy, ACE

inhibitor, beta-blocker, candesartan--and no signal

of a safety issue. So, this was an important group

looked at, at the time.

Spironolactone was an opportunity for us

to query potential issues, with 17 percent of

patients on spironolactone. We had 436 and there

was no interaction here. This is a

non-prespecified sub-subgroup that I put here with

trepidation, just to say everyone is on an ACE

inhibitor, beta-blocker, spironolactone, placebo or

candesartan, and it is only 237 patients but there

is the data in that non-prespecified sub-subgroup.

If we do that, we must look at safety and the best

measure of safety would be all-cause mortality and

we are showing that here with no signal but,

certainly, the confidence is based on 237 people in

the sub-subgroup.

So, this part of my presentation is really

the standard CHARM-Added and we believe we have

addressed the hypothesis that we set out to test,

that for patients with symptomatic heart failure

already being treated with an ACE inhibitor and

other conventional therapies the addition of

candesartan improved clinical outcome, and

improving clinical outcome by our definition was

reducing the risk of CV death or a hospitalization

for heart failure, and we can confirm that with our

secondary endpoint of reducing all-cause mortality

and hospitalization for heart failure which was

also reduced.

In response to the agency's very pointed

and very stimulating questions, I will present some

other data. One is to put CHARM in external

perspective. There have been three major outcomes

trials with ARBs in patients with depressed

ejection fraction and symptomatic heart failure.

One was a head-to-head comparison and in that the

dose of the ARB was not found to provide clinical

benefit or to be even comparable.

Here is the closest study to CHARM-Added.

This is the ValHeFT experience which has been

presented to this group. In the ValHeFT it was

conventional therapy and an ARB. For the composite

outcome, one of their co-primaries of morbidity and

mortality, there was a significant reduction. In

the CHARM study there was a significant reduction.

So, I think the external validation of adding an

ARB, without looking at subgroups but looking at

the total group, gave very similar information.

The reason we have more events here is, again,

because of the longer exposure and longer

follow-up.

The other questions from the agency which

we will try to address the best we can--Dr.

McMurray told you how the study was conducted and

we did find that investigators were using a variety

of ACE inhibitors. So, if I look at those ACE

inhibitors, as Dr. McMurray showed you, there were

12 including enalapril and four of these did not

have an FDA approval so we couldn't find the dose

that would be used.

So, now just talking about the agents

themselves with the different use of the agents, we

used an analysis of was there a difference in the

outcome of those who received an ACE inhibitor that

had FDA approval or those that did not. That

analysis is a non-prespecified one that I am

showing here. Here are the patients that had the

FDA approval using an ACE inhibitor, and here are

agents that were not approved. Again, the best

estimate is the overall. So, as far as the agent,

we did not see any difference.

The real probing question that we have

seen through your questions is the dose issue. To

get at that, I have to say the first analysis that

the investigators and the sponsor did was the

prespecified one. Prior to unblinding, the

academic group made a list of the evidence-based

therapies and the doses. We had made that

definition called the recommended by the

evidence-based. When we did that, there were 1291

patients who at baseline were receiving that dose.

I will talk about that dose in a moment

but I think one of the questions about trial design

and trial conduct that has to be addressed right up

front was in order to test the addition of the new

medication, candesartan, the study medication, did

the investigators sustain the levels that Dr.

McMurray was so proud of, or did they just reduce

that to start the other inhibitor of the

renin-angiotensin system--a very important and

valid question.

To do that, I will just be talking about

the five most commonly used, which is approximately

80 percent of our patients and is representative,

and the dose, and look at the titration time

period. While patients were being titrated to

either placebo or candesartan there was no

down-titration of the ACE inhibitor. That was

something that was conveyed to investigators. If

your patient is stable on these doses of an ACE

inhibitor, that is what you should be sustaining.

If you have issues you should be down-titrating the

experimental medication.

I also have some additional data here on

the use of the ACE inhibitors over time, and I

think it is quite reflective of our baseline

numbers, that there was no attrition of the use of

ACE inhibitors. So, we are looking at the added

value of candesartan. It is on top of holding good

doses of ACE inhibitor over the time frame.

So, what was the analysis? This is the

prespecified one from the investigators. These are

the 1291 patients who at baseline were receiving

doses equivalent to those in the evidence-based

trials, and these are the patients who were not.

That does not mean these patients weren't receiving

optimal dose for them; it is individualized care.

But just making this definition, there was no

interaction here. The observation of the overall

benefit means that this benefit was not modified by

the baseline dose of the ACE inhibitor using this

definition.

In subsequent communication with the

agency, there were requests to create additional

subgroups. Since our forms were designed to know

the ACE inhibitor and the dose, we are able to

comply with those requirements. The agency asked

for different doses, a definition of maximum where

now the lisinopril dose is increased and some of

the other agents are increased. So, we go from

having 1291 who met our definition to now 721 who

met the new subgroup criteria.

If we look at the results of that, I think

you can see the consistency that there was no

modification of this benefit of candesartan that I

have been describing based on the ACE inhibitor

dose at baseline with these two definitions of ACE

inhibitor dose.

In subsequent communications with the

agency another subgroup was defined, and we were

pleased to be able to comply. This one raises the

captopril to 300 mg and we did have 2 percent of

our patients at baseline. More importantly, it

raised the enalapril dose to 40 mg and we did have

10 percent of our patients on enalapril at that

dose. So, overall now we are talking about 20

percent of the patients, 529, who met the new

definition.

Here are the results of this new subgroup.

The 529 and the remainder had the same efficacy so

this candesartan benefit on reducing risk of

cardiovascular death or hospitalization for heart

failure was not modified by any definition of ACE

inhibitor dose at baseline, our prespecified one

and the two definitions that the agency requested.

Because we are a program of research, we

can give one more, and that is the zero dose of an

ACE inhibitor. So, we have a whole trial that you

have evaluated and that trial is zero,

CHARM-Alternative, 2028 patients not receiving an

ACE inhibitor.

So, I think we have run the whole spectrum

here and you can see the results. Now if we pool

the two, the benefits that we are describing of

candesartan were not modified by the dose of the

ACE inhibitor from zero to predefined levels to

subsequently defined maximum levels at baseline.

That allows us to conclude that we really

have an additional opportunity to help patients who

are already on an ACE inhibitor and, more than 55

percent, on a beta-blocker. That really is the

clinical question. When CHARM was designed that

was the issue, can we make an improvement in the

practice of medicine? We didn't know the answer.

We now share that answer with you and we think we

do. We reduce the patient's risk of cardiovascular

death or hospitalization for heart failure on top

of other therapies, irrespective of the dose of the

ACE inhibitor, and we offer that opportunity to

reduce cardiovascular morbidity and mortality.

That opportunity does come with some

responsibilities, and Dr. Hainer will discuss the

risk of inhibiting the renin-angiotensin system in

doses that improve morbidity and mortality, and

then Dr. Young will come back and describe the

risk/benefit. Thank you.

DR. NISSEN: Thank you, Mark. Are there

questions right now? Yes?

DR. HIATT: Just a quick one on slide 28.

Is that a typo, the maximal FDA-revised for

lisinopril? Did the dose go down from 40 mg to 20

mg? Is that true?

DR. PFEFFER: That is not a typo. We were

responding to definitions provided to us.

DR. PICKERING: Could you give us a

breakdown of which beta-blockers the patients in

CHARM-Added were taking, in particular how many

were on carvedilol?

DR. PFEFFER: Yes, I could do that and I

would like to do that. I said 55 percent at the

start and obviously that number increased to the

mid-60s by the time it was over. If I can show the

beta-blockers that were used at baseline, the

predominant beta-blockers were metoprolol and

carvedilol, 81 percent. These doses were sustained

over time, but the number of patients alive on a

beta-blocker increased over time.

DR. SACKNER-BERNSTEIN: In light of that

slide, you did a nice job of showing the effect of

coronary heart disease on top of approved ACE

inhibitors, trying to make sure that we really were

evidence-based. Can you show us a similar analysis

for approved beta-blockers as background therapy?

DR. PFEFFER: I don't think I can,

Jonathan, but with 80 percent of the people on the

approved, I would think the numbers would be the

same--if I have this information, and I don't think

I have.

DR. NISSEN: We are going to have lots of

time for questions. If there ar clarifications,

let's do that.

DR. TEMPLE: Just one thought, I just

wanted to say that with all these after the fact

analyses, don't try these in your own home.

[Laughter.]

DR. NISSEN: We have some very solid

advice. So, we are kind of going to finish the

sponsor presentations and then we are going to have

lots and lots of time for questions.

Safety

DR. HAINER: Good morning, Dr. Nissen,

members of the advisory panel, FDA, public guests.

I am Jim Hainer from AstraZeneca, and I would like

to begin by stating that the candesartan safety

profile in the CHARM program relative to

placebo--the findings were really quite consistent

across all three CHARM studies. For the purposes

of this presentation I will, like my other

colleagues, review now the safety of candesartan in

chronic heart failure when added to evidence-based

doses of ACE inhibitors, the CHARM-Added trial.

Let's start then with two points that are

really important to safety monitoring. First, the

CHARM provided explicit monitoring directives for

the clinicians. Second, the CHARM protocol was

particularly specific about monitoring for

hypotension, renal dysfunction and hyperkalemia,

events expected for any drug which inhibits the

renin-angiotensin system when added to an ACE

inhibitor.

These directives included monitoring of

blood pressure, creatinine and potassium at

multiple intervals. These were baseline, within 2

weeks of dose adjustment, at the end of dose

titration, annually and, of course, at any time in

the judgment of the responsible clinician. These

monitoring directives are entirely consistent with

usual clinical practice in caring for heart failure

patients.

With that said, let's look then at

hypotension, renal dysfunction and hyperkalemia.

Hypotension was reported as an adverse event in

23.2 percent of the patients receiving candesartan

and evidence-based doses of ACE inhibitors and 14.5

percent among those receiving only ACE inhibitors.

Hypotension was reported as one reason for

treatment discontinuation for 5.4 versus 3.5; for

hospitalization, 4.3 versus 1.7; and for serious

fatal adverse events 0.2 versus 0.1 percent.

Note here, expressed as proportions of

patients, that discontinuations due to hypotension

in patients 75 years and older, those taking

spironolactone or beta-blockers, were similar to

the overall discontinuation rates. The rate for

candesartan was about 3.5 times higher though among

patients entering the trial with a baseline

systolic blood pressure less than 100 mmHg.

Renal dysfunction was reported for 15.4

percent of the patients receiving candesartan and

ACE inhibitors; 9.4 percent among those receiving

only ACE inhibitors. Renal dysfunction was

reported as one reason for discontinuation in 8.2

versus 4.2 percent; for hospitalization, 4.5 versus

3.0 percent; dialysis, 1.6 and 1.6; and for a

serious fatal adverse event, 0.9 versus 1.5

percent.

Discontinuations due to renal dysfunction

in patients 75 years and older and diabetics taking

spironolactone or with systolic blood pressure less

than 100 were similar to the overall

discontinuation rates in the trial.

For patients entering the trials with a

creatinine already greater than 2, the rates were

high in both groups but the rate for candesartan

was really no higher than for placebo.

Next, hyperkalemia was reported in 9.6

percent of the patients receiving candesartan and

3.6 percent receiving placebo. Hyperkalemia was

reported as one reason for discontinuation in 3.8

versus 0.9 percent; for hospitalization, 1.2 versus

0.7 percent; and for a serious fatal adverse event,

0.2 versus 0.0 percent.

Despite the potential for hyperkalemia to

increase rates of sudden death and fatal

ventricular fibrillation, both rates were somewhat

lower in the candesartan group, specifically 11.2

versus 13.7 and 0.7 versus 1.3 percent

respectively. Discontinuations due to hyperkalemia

in diabetics and patients taking spironolactone was

similar to the overall discontinuation rates in the

trial. The rates were higher in patients 75 years

and older and those with potassium greater than 5.

In patients entering the trial with a serum

creatinine of 2 or greater, the rates were high but

similar in both groups.

Now, having led with this data,

highlighting these three specific areas of

interest, let's examine whether they translate into

global adverse consequences. Any adverse event was

reported in 80.4 percent of the patients receiving

candesartan and evidence-based doses of ACE

inhibitors and 78 percent among those receiving ACE

inhibitors. Of particular interest, serious

adverse events were reported in 75.9 percent in

both groups, of which serious fatal events were

29.5 and 32.5 percent in the candesartan and

placebo groups respectively. Treatment

discontinuations due to adverse events were 24.3

and 17.6 percent. Dose reduction due to adverse

events were 17.2 and 9.7 percent respectively.

Listed here are the common serious fatal

adverse events by treatment. Sudden death occurred

in 11.2 percent of the patients receiving

candesartan and 13.7 percent amongst those

receiving placebo. For heart failure the

corresponding figures were 5.8 and 8.8 percent

respectively. Other causes of death were far less

common. Of note, there was no trend toward a

consistently higher risk in the candesartan group.

Now, safety concerns also surround the

concomitant use of other heart failure treatment

drugs, as already alluded to by Dr. Pfeffer. To

that end, Dr. Pfeffer presented this slide which

demonstrates the benefits of candesartan on the

primary prespecified endpoint of cardiovascular

mortality or heart failure hospitalization, both

overall as well as for subgroups of patients

receiving spironolactone or spironolactone plus a

beta-blocker.

One logical concern is that the reduction

in heart failure hospitalization may not be

reflected in all-cause hospitalizations. But, in

fact, these data show no significant increases in

all-cause hospitalizations either overall or in

these subgroups.

A second logical concern is that the

reduction in cardiovascular mortality might not be

reflected in all-cause mortality. But here, again,

these data show no significant increases in

all-cause mortality either overall or in any of

these subgroups.

These trends in hospitalizations are

further reinforced by the cumulative number of

hospital admissions for any cause shown here in the

candesartan and placebo groups and, as Dr. Pfeffer

pointed out, even though the risk remains larger

for the candesartan group. Importantly, there is

no increase in the non-cardiovascular rate for

hospitalization in the candesartan group.

Next, if you can recall the all-cause

mortality data for CHARM-Added, note how they are

reinforced by the cumulative number of deaths from

any cause in the candesartan compared to the

placebo groups.

Having now examined the safety of

candesartan in chronic heart failure when added to

evidence-based doses of ACE inhibitors, I want to

conclude with two final slides. First, let me

summarize the safety findings and conclusions. As

expected, due to greater renin-angiotensin

inhibition, rates of hypotension, abnormal renal

function and hyperkalemia were greater with

candesartan. But these predictable adverse events

did not translate into any increase in all-cause

hospitalization or mortality, sudden death, renal

failure or ventricular fibrillation. These data

show that candesartan is safe and generally well

tolerated by patients with heart failure receiving

evidence-based doses of ACE inhibitors.

Second, understand that AstraZeneca is

firmly committed to risk minimization. We also

wish to maximize opportunities for benefits. In

order to ensure proper use of candesartan with

heart failure receiving ACE inhibitors, AstraZeneca

will implement all of the following risk

minimization activities: Administration and dosing

instructions which are consistent with those that

guided the CHARM-Added investigators; labeling

which includes precautions and warnings regarding

these adverse events; collaboration with major

societies involved in the treatment of heart

failure patients; and educational activities to

ensure that healthcare providers understand the

risks as well as the benefits of using candesartan

in heart failure. This includes focused training

of sales force; and expert scientific liaison

groups; continuing medical education activities;

and prominently displaying information on all

promotional materials regarding the risk of using

candesartan in heart failure.

With these measures in place, candesartan

can be safely used as another important treatment

option to reduce cardiovascular events in patients

with heart failure who are receiving ACE

inhibitors. I will turn now to Dr. Young once

again who will elaborate on the issues of benefits

and risks of candesartan in the treatment of

chronic heart failure.

DR. NISSEN: If there are any burning

questions on this presentation let's have them,

otherwise I think we are ready to launch into full

questions after Dr. Young.

Risk/Benefit Summary

DR. YOUNG: Thank you, Jim. It is now to

overview our data and quickly consider the impact

we can make on ill patients with significant heart

failure.

Our CHARM program in its entirety, and

specifically the CHARM-Added study, the broad

patient population, comprehensively characterized

the risks associated with treatment, particularly

the combination of an ACE inhibitor and

candesartan. We believe that we have clearly

delineated net benefits for this therapeutic

strategy in CHF patients with depressed left

ventricular ejection fraction.

Particularly important, CHARM-Added

addressed the previously unresolved question of

whether adding an ARB to an ACE inhibitor in

patients with low EFV heart failure provided

incremental benefit by reducing risk of

cardiovascular death or heart failure

hospitalization. Interesting and also important is

the fact that we have demonstrated added benefit in

patients receiving evidence-based doses of ACE

inhibitors proven effective in previous clinical

trials, and we also believe we have demonstrated a

favorable benefit/risk profile.

This benefit/risk profile is best

summarized in this slide. Overall there was a

significant 15 percent relative risk reduction for

the primary endpoint, cardiovascular death or heart

failure hospitalization, over the 41-month median

follow-up. When analyzing the data per 1000

patient-years, this translates into an absolute

risk reduction of 25 patients having a primary

endpoint event over that period of time, as

summarized in the third column on this table.

Importantly, no increased risk for

all-cause mortality or all-cause hospitalization or

the combination was noted. These observations were

all less in the candesartan treatment group, again

noted in this table. This should assuage concern

about adverse events precipitated by this

therapeutic strategy.

Thus, candesartan, at a target dose of 32

mg daily, significantly reduces the risk of

cardiovascular death or heart failure

hospitalization when added to an ACE inhibitor,

irrespective of agent and irrespective of dose.

Given our understanding of heart failure, it is

prudent to look at the most common adverse events

in this population--hypotension, hyperkalemia,

abnormal renal function. Proposed instructions for

the use of this strategy are consistent with those

provided to the CHARM investigators and good

clinical management of any patient with heart

failure.

We will emphasize attention to volume

status, blood pressure, renal function and

potassium levels, and recommended monitoring of

these measures will be with initiation of

candesartan dose titration and periodically

thereafter the same as we manage all of our

patients with heart failure.

In conclusion, we believe that the

addition of candesartan to an ACE inhibitor

treatment of heart failure patients, as was done in

the CHARM-Added trial, will result in substantial

cardiovascular morbidity and mortality benefit.

The positive risk/benefit profile is further

supported by numerical reductions in both all-cause

hospitalization and all-cause mortality. We

believe these findings support the use of

candesartan with or without an ACE inhibitor at

varying doses for the routine management of heart

failure so that candesartan can be prescribed for

managing these patients with left ventricular

systolic dysfunction.

Dr. Nissen, ladies and gentlemen of the

panel, thank you very much. I will ask Dr. Mark

Pfeffer to come back to the podium so that we can

direct any questions to the group.

DR. NISSEN: Thank you very much. I must

compliment the sponsor. It is rare that we finish

ahead of time. We don't have a break scheduled

until ten o'clock so I think we can maybe start

taking some questions and we will take our break a

little bit later. Blase?

Questions from the Committee

DR. CARABELLO: Mark, based on ValHeFT I

have routinely avoided the use of an ARB in

patients already receiving a beta-blocker and an

ACE inhibitor. Now CHARM-Added seems to ameliorate

that. So, what is the difference? Is this the two

agents? Is this the kind of beta-blockers that

were used in the two different studies? Is this a

statistical glitch among the two studies? How can

we reconcile those two studies?

DR. PFEFFER: Well, Dr. Carabello, I can't

be definitive but I can give you my opinion on

that. I, like you and every clinician, wanted to

be adding an ARB on top of other therapies to

reduce adverse outcomes in patients and that

beta-blocker subgroup gave us pause. It really did

because what we do know is that beta-blockers have

a profound benefit and they do on top of an ACE

inhibitor. So, that was the conundrum in 1999.

Then, with the publication of our

experience, I think it really showed that maybe

that was a hazard of a subgroup. It turns out, if

we look at the numbers in our experience, there

were even more patients having events. if I could

show that, because we had more patients on a

beta-blocker and greater exposure time when we are

giving you our subgroup, prespecified subgroup, it

is based on more events. Just to give you an idea

of the two trials, the deaths, which is really what

we are concerned about, the total deaths were 226

in ValHeFT and really 370. So, I think there is

more confidence in our subgroup based on the

increased number of events.

You then asked about the agent. I think

there is an excellent answer to that because there

was a very large study, called VALIANT, which used

that agent in a large number of people on triple

therapy, actually more patients on triple therapy

than here, and did not show an adverse safety

interaction with beta-blocker, ACE inhibitor and

that agent.

So, I think there was a pause because

safety doesn't require the same boundaries of

statistics that efficacy does, and that pause I

think is now erased by what we showed you for

candesartan and that other study. So, I do think

the message for clinicians--and this is really the

important thing, the message for clinicians should

be ACE inhibitors at the optimized dose,

beta-blockers and then this addition of candesartan

in the strategy we have shown can reduce morbidity

and mortality.

DR. NISSEN: Go ahead, Tom.

DR. PICKERING: As a follow-up to that,

you said 31 percent of the beta-blockers were

carvedilol and I wasn't able to see what the

proportion was in ValHeFT and, you know, there is

the COMET study that suggests that there may be a

difference between different beta-blockers in heart

failure. I wonder could that be one possible

explanation.

DR. PFEFFER: I am here for the CHARM

data. I really don't have detailed knowledge about

ValHeFT and I would say, based on the small numbers

we are talking about, if we start dividing that up

by the agents it would be even more unreliable, but

I don't have that information.

DR. NISSEN: Ralph, you had a question?

DR. D'AGOSTINO: In Table 59 of the recent

material that you sent and our response to C-25 and

C-29, I am trying to understand--I know this is all

post hoc and I should not be excited about looking

at post hoc analyses, but I am trying to understand

what happens as you go from maximum dose no to yes.

If I look at slide 25, what seems to happen is when

you are dealing with the no--this is the

recommended and you are dealing with the no you

basically have the placebo and drug pretty much the

same. There is only something like a 12 events

difference. When you move to the yes you have a 43

events difference, and the change is all basically

in the candesartan. Its events drop down. The

placebo, whether no or yes, 165 in terms of the

events per 1000 follow-up years and the candesartan

goes from 151 to 131.

Then when you move to the next slide,

slide 29, here the no for analysis one has in terms

of the placebo rate 172 versus 152, when you go to

the yes where the candesartan has 145 to 133.

Again, when you go from the no to the yes it is the

candesartan that is showing the reduction. The

same with analysis two. In analysis two if you

look long enough you will find an analysis that

will produce statistical significance. So, my

question is it seems to be the action in the

candesartan. Does that say anything about the

added benefit to the ACE?

DR. PFEFFER: Well, Dr. D'Agostino, I know

enough not to discuss statistics with you on this--

DR. D'AGOSTINO: Granted, we shouldn't

have done this.

DR. PFEFFER: I think you are asking me is

there a pattern here, and I think there is no

pattern here and I think the interpretation--may I

have the slide, please? You are asking is there a

pattern in the no's. Obviously, by every

definition we are making a new definition of no.

But I think the way to handle this is in any

definition was there a hint of an interaction, and

the answer--

DR. D'AGOSTINO: The interaction test is

notoriously lacking in power, which is the problem.

DR. PFEFFER: But let's look for

consistency here, is there a consistent message?

If anything, we are not making the message that we

are even better on top of an ACE because we also

have this 2000 experience here of zero. That is

the definite no. So, I think we run the range of

no's from low doses, from zero doses to higher--as

we go here we have a higher and higher dose of no

really, the no group, because of the higher dose of

ACE inhibitor. So, I personally don't see any

consistency here and I don't see any pattern. But

if you do, then I would be worried--

DR. D'AGOSTINO: Well, I am just trying to

sort out why you would say that candesartan adds to

the ACE inhibitor. What is the revelation in the

data that would say that?

DR. PFEFFER: I think it is this point

right here that candesartan adds to an ACE

inhibitor. A 100 percent of these patients are on

ACE inhibitor. I will remind you that from the

clinician's perspective--I will go back to what Dr.

McMurray was saying, from the clinician's

perspective, 96 percent of our clinicians checked

the box that says I believe I have optimized their

care. Now, that is a box. We then upped the ante.

We made the evidence-based medicine definition.

The FDA made these definitions. So, really the

best way to look at our data is overall and I don't

see a pattern here with the different definitions

of doses.

DR. NISSEN: I wanted to ask a question

related to CS-12. You may not have this but I sure

would like to see it. This is a little unusual

Kaplan-Meier plot. It is cumulative number of

hospital admissions and I would like to see time to

first hospital admission for any cause because that

is a more traditional analysis.

DR. PFEFFER: Yes, and Dr. McMurray has

done a lot of analyses of pharmacoeconomics so for

that we needed cumulative numbers. For safety, and

this was presented in our safety presentation, we

think the burden is the cumulative. That is

something I was alluding to also although our

analysis plan didn't let me show you that because

we were timed to first. I think in the clinical

scenario we are really trying to keep the revolving

door. And, this is showing all admissions for any

cause and we thought this was the strongest safety

statement we could make about the population. I

don't know if I have hospitalization as time to

first event. I don't know that I have that.

DR. NISSEN: Let me tell you why I am

asking the question. I want to understand if there

is an early hazard. That is where time to first is

very helpful. That is, when you are titrating up

candesartan and you are getting these admissions,

there is a fair number of admissions for

hypotension and for hyperkalemia, and I want to see

whether the pattern shows an early hazard within a

more favorable effect later on because I think it

is very important for clinicians. I assume

somebody has done that analysis.

DR. PFEFFER: That is a very important

point. We can show early efficacy. We were

showing that. And time to first hospitalization

for any cause--let's see if I can get that for you.

DR. NISSEN: That would be really helpful.

DR. D'AGOSTINO: The graphs they do show

seem to have a consistent hazard. That is a good

question if you go to all-cause hospitalizations.

DR. NISSEN: I did a little Tom Fleming

type back of the envelope calculation and I want to

see if I am right about that, but there are a fair

number of those hypotension hospitalizations and I

am guessing that they are early, that when you are

trying to titrate up the drug you run into some

difficulty. So, I think to inform clinicians about

how to do this it is very important to understand

whether there is in fact and early hazard.

DR. PFEFFER: I totally agree. I don't

think that is the case and I would like--somebody

is showing me CV hospitalizations but I need all

hospitalizations to reassure. CHF hospitalizations

won't reassure you and I need all hospitalizations

to reassure you.

DR. PORTMAN: To turn from cardiorenal to

renal for a second, based on DOQI guidelines and

Framingham studies and so forth, we know that

microalbumenuria is an important cardiovascular

risk, independent risk. Do you have data on the

prevalence of microalbumenuria? Was there

improvement with the ACE/ARB or just the ACE alone

in microalbumenuria? In fact, did you even see

resolution in a portion of the population in

microalbumenuria?

DR. PFEFFER: I have to say that that is a

sub-study which is being run out of McMaster

University and that as of this moment I don't have

the results on the 600 people who were in what we

call micro-CHARM. My friend Dr. McMurray is closer

to that data. Do we have that?

DR. MCMURRAY: No, we don't.

DR. PFEFFER: We have yet to see that

data, sorry.

DR. KASKEL: With regard to kidney, those

patients with creatinines less than 3 and maybe

above 1.5 are still at risk for dysfunction and you

had hyperkalemia as one of the early changes. I am

just wondering if there are any other guidelines

that might be helpful to prevent hyperkalmeic

episode in patients with diminished renal function.

DR. PFEFFER: Definitely, the patients

with impaired renal function are much more

vulnerable. They are also the patients at highest

CV risk. Here is where cardiorenal really should

be cardiorenal; we should be getting together more.

So, we identified the same risk and now that we

have learned how to use the MDRD equation we are

suddenly realizing we have more patients at risk.

But that was true for placebo as well as for

candesartan. All the augmentations are related to

baseline renal function, more so on candesartan,

but you need the same monitoring for someone with

impaired renal function whether or not you add

candesartan because they are at high risk also.

Let me see if I can show you something

like that. I would like to show you the EGFR and

just to show the adverse experience, just to share

that with you. I believe I have a better

opportunity to show you that than all-cause

hospitalizations as a function of time. May I have

the EGFR? We do have that information and it is

concerning for both placebo and candesartan. I

think the message we have to get out there for

education is that we should be looking at renal

function and we should be alerting ourselves to

vulnerable patients. I will have that for you a

little later.

DR. SACKNER-BERNSTEIN: Getting back to

Steve's point about how we can create a way for

clinicians to understand how to utilize the drug

and manage the patients who are getting the drug,

as well as the point you just made about renal

function, I am wondering if you could provide us

with some insight as to what happens to patients

who develop worsening renal function specifically

during the titration. I look back to the SAVE

trial where you did such a nice job of talking

about the prognostic importance of heart failure

hospitalization and subsequent course. What can

you tell us about worsening renal function?

DR. PFEFFER: I am going to ask Dr. Lewis

but I do want to show the slide that I was just

alluding to. Let me just show this first. I will

get back to the EGFR and then we will continue the

thread of what happens to people.

So, here cardiologists have learned how to

do EGFR, and it is a risk for discontinuation of

any causes and candesartan augments that risk. But

this also tells us how carefully we have to monitor

the placebo patients with impaired renal function.

Your specific question about discontinuation due to

renal function and outcome, I am going to ask Dr.

Lewis, our renal consultant.

DR. LEWIS: I am Dr. Lewis, a Vanderbilt

nephrologist. I would first like to remind the

panel that there is a great body of data in renal

literature that inhibition of the renin-angiotensin

system benefits people in terms of preserving renal

function across a wide range of kidney disease and

across a wide range of GFR, including CKD for the

lowest GFR groups, which has now been reported from

several of the major clinical trials.

There are two settings in which inhibition

of the renin-angiotensin system can cause renal

dysfunction. One is that patients have ischemic

renal disease or fixed renal artery stenosis. The

second, more relevant to the CHARM study, is if a

patient has decreased effective arterial blood

volume. That occurs in two settings, decreased

cardiac output which, of course, these patients

were at risk for, and decreased intravascular

volume, which they were at risk for because of the

use of diuretics.

In both those settings the kidney becomes

critically dependent on efferent arterial

resistance to maintain GFR. It is a hemodynamic

effect. One would predict when a patient has

decreased effective arterial blood volume and

develops renal dysfunction that the stopping of the

agent, the inhibition of the renin-angiotensin

system, would repair that renal hemodynamic and the

patient should recover. It should be a reversible

event.

Evidence to support that--first I will

remind you that Dr. Hainer showed you that the

number of patients requiring dialysis was

equivalent in the two groups, on his safety slide.

Also, if I could have slide 48, looking at the

ultimate outcomes for people who had renal

dysfunction?

So, these are the patients who had any

kind of renal dysfunction event during the course

of the trial and what happened to them. I have

already told you that they had an equivalent amount

of dialysis. As you can see, 38 percent of the

placebo group was alive at the end of the trial and

55 percent of the candesartan group was alive at

the end of the trial. So, I think the signals we

have from the CHARM-Added is what you would expect

from the physiology, that this was a reversible

event.

DR. SACKNER-BERNSTEIN: Just to clarify,

what was the definition of renal dysfunction in

that analysis?

DR. LEWIS: The definition of renal

dysfunction in this analysis was if an investigator

indicated in a narrative form that the patient had

renal dysfunction of any sort. The narratives were

scanned very closely. There was an appendix about

renal dysfunction attached to the protocol that had

precise instructions for a given change in renal

function. So, for more than 1 mg/dL increase to a

level greater than 2, the investigator was

instructed to respond to that. But for the

purposes of the safety analysis we used any change

of renal dysfunction that the investigators noted.

DR. SACKNER-BERNSTEIN: Part of the reason

I am bringing this up is because of a little bit of

discomfort that I have about how to know the

optimal way to interpret changes in creatinine.

Certainly, if you take a heart failure patient and

you treat them with an inhibitor of the

renin-angiotensin system you would almost hope to

see an increase in creatinine, consistent with the

hemodynamic mechanism you defined, as reflecting

the fact that you are achieving a pharmacologically

relevant level of inhibition. That is the way most

people, I believe most people think about the use

of these agents in a chronic setting such as this

trial. In the acute setting there is a growing

body of literature that increases in creatinine

during treatment of acutely decompensated heart

failure in a hospitalized setting portends a worse

long-term prognosis.

In trying to bring those two observations

together I found a relative paucity of data to look

at what happens to people in a chronic setting

where serum creatinine goes up by 0.3 mg/dL, 0.5

mg/dL during initiation of therapy. Should

clinicians be looking for that physiologic effect

on efferent arterial as something that is a good

sign or is it potentially a bad sign?

DR. LEWIS: I think this is a great issue.

I am actually giving cardiology grand rounds at

Vanderbilt next week so I am going to address this

issue.

DR. SACKNER-BERNSTEIN: What day? What

time?

DR. LEWIS: I think this is so good

because I think we really are learning more because

I think what your paradox is--first let me say that

in renal trials, as well as in cardiology

literature, you are exactly right. The patients

who most benefit from inhibition of the

renin-angiotensin system in the first three

months--in terms of, you know, don't go into

end-stage renal disease or hard outcome--in the

first three months of exposure to the inhibition of

the renin-angiotensin system do two things. They

drop their proteinuria and they drop their GFR by a

hemodynamic mechanism because we have shown

reversibility. It is 3-5 mL. It is not clinically

significant but it is a signal, like you said, in

heart failure patients that they are responding to

the inhibition of the renin-angiotensin system.

I think the reason why you have the

paradox is that the patient in the hospital who,

despite you doing all you can do for them in a

hospital setting has a very poor cardiac output, is

the patient who has decreased effective arterial

blood volume and you can't make it any better

because they have reached a point where, short of a

heart transplant, you can't make their cardiac

output any better. When you give that patient an

ACE inhibitor or an ARB you can't get their heart

to be better. Nothing is going to get that heart

to be better. In that setting the kidney is giving

you the message that the patient has reached an

end-stage heart situation.

DR. MCMURRAY: Jonathan, I can actually

answer your question directly because we are all

interested in this in heart failure at the moment.

I will show you a slide that shows you the change

in GFR over time, but it is in a slightly different

way than my own personal slide of this issue in

CHARM-Added because what you see in CHARM-Added is

you see a sort of steady decline in GFR over the

three and a half years of follow-up. The placebo

group and the candesartan group run in parallel.

But if you plot those two lines together what you

see is this initial little drop in the candesartan

group and thereafter they run parallel with the

placebo group.

So, it is interesting to me because I

think, unlike the nephrology issue, we don't see

protection or preservation of GFR over time with an

ACE inhibitor or with an ARB or with the

combination. We see this initial little decline in

GFR but then the two lines run absolutely parallel.

It intrigues me why the kidney in heart failure

seems to be a bit different than the kidney in,

say, diabetic nephropathy.

DR. NISSEN: I would be very interested in

seeing the U.S.-non-U.S. analysis. There are some

obvious differences there. I presume you have a

slide that drills down on that, or maybe by region

if that would be possible. Do we have that?

DR. PFEFFER: Yes, I think this is the

observation that you are discussing. This is one

of multiple subgroups.

DR. NISSEN: Of course, and obviously I

recognize the hazards of this but, to me, it is a

rather striking difference. We have seen this now

in a fair number of drug development programs where

the effect is seen outside the U.S. but not in the

U.S. and I want to understand it.

DR. PFEFFER: Well, first you would have

to believe that that is a truism. So, if you just

take the countries it bounces around like crazy.

You would expect that. One of the real strengths

of CHARM is that we have 7599 patients with

long-term follow-up, and if there is something

about carrying a U.S. passport you would expect to

see a consistent message. So, we really are coming

to you with three trials.

I would like to show you this slide. This

is the point, and it was just over the line at

1.019. On this scale it looks like it is on line,

just over. But there was no inconsistency here.

But let's look at the total program. If

there is something about being a U.S. citizen that

means you are not going to see the benefit of

candesartan, let's look at all patients. When we

get down to the 7,500 patients U.S.-non-U.S., I

think you would agree with me there is nothing

here. More importantly, I think when you look at

studies was the U.S. represented? The U.S. was the

major contributor to the CHARM program.

DR. NISSEN: Were the overall event rates

different in the U.S. and other countries?

DR. PFEFFER: I am going to represent Dr.

Granger because he has done a complex analysis that

only the Duke group can do of the CHARM data,

looking for the modifiers and predictors of

outcome. Despite hundreds of man and women hours,

the things you know about--ejection fraction,

diabetes, age--I asked Chris what else have you

done; put in re-vascularization? No. Race? If

anything, we don't have enough African Americans to

talk about but the point estimate goes the right

way. The other issue in the model, if you now

force the U.S. into the model it does not come out

as a predictor.

DR. TEMPLE: We are sort of watching this.

It keeps showing up or at least you notice it when

it does show up, which is probably more to the

point. Sometimes there are oddities to it. In

both RENAL and IDNT the action was all in the Asian

population, Asian including Israel and a variety of

places you don't usually think of as Asian. But

when we actually looked at the end-stage renal

disease endpoints it didn't look that way anymore.

So, the long-term follow-up no longer was as

conspicuous in the U.S. population. So, I don't

know what you make of something like that but these

things are jarring when they show up.

DR. NISSEN: Let me tell you why these

things catch my attention and bother me.

Obviously, the FDA is charged with regulating drugs

in the United States and we are presented with a

certain number of trials where the U.S.

contribution was a minority of the population and

where sometimes the point estimates like this are

quite variable. One of the things I always worry

about is, you know, are these patients somehow

different? Is the underlying care, particularly if

there are a lot of Eastern European and other

countries involved different? I am just trying to

get an understanding of this because I know this

must come up for you a lot. It always gives us

pause for thought considering the fact that this is

a drug that we are considering for use in the

United States. So, any advice, Bob?

DR. TEMPLE: No, it is just hard to know

what to make of it. My bias is that if people are

treated badly they probably benefit more from a

drug that they are actually getting. So, maybe the

U.S. is too well--you know, you could say, well, in

the U.S. they really all got their ACE inhibitor

and in the other places they all lied.

DR. PFEFFER: That is a very U.S.-centric

view--

DR. TEMPLE: I am not alleging that it is

true. I am just saying what is the worst thing you

could imagine.

DR. PFEFFER: I am not speaking about

CHARM now but in almost every database the

presumption was that U.S. are better treated,

better outcomes. I have many friends in Canada and

every time we have sliced it Canadians do a little

bit better, so less procedures and do a little bit

better so it is hard to even support the

hypothesis.

DR. TEMPLE: I am in no way saying it is

true. I am just saying, you know, what is the

worst thing you can imagine?

DR. NISSEN: One way to test this which

would be very helpful to me just to get comfortable

here is what the actual event rate was in the U.S.

versus the non-U.S.

DR. MCMURRAY: To answer your question

directly, if you look at the two low ejection

fraction groups pooled, and I am only saying that

because I think that is the type of heart failure

we all know most of all, if you look at the placebo

groups, if you compare U.S. to non-U.S. the event

rates are almost identical. One is 41.7 percent,

the other is about 42 percent. So, the event rates

in the conventional type of heart failure that we

are all familiar with are virtually identical.

DR. NISSEN: What are they in the

CHARM-Added?

DR. MCMURRAY: Someone is going to have to

do the mathematics very rapidly for me. I put the

two low ejection together simply because it was

large numbers but, again, you can see they are

almost exactly the same.

DR. D'AGOSTINO: Yes, they are almost

identical. It is a smaller group. The confidence

bands are large; lots of multiple comparisons.

DR. NISSEN: And I do recognize that. You

know, this is not by any means definitive. It is

an observation that pops out and you want to try

and understand it. I mean, if we saw an event rate

in the non-U.S. that was radically different from

the U.S. that would be a signal to me that this is

meaningful, and we don't see that here.

DR. MCMURRAY: I was going to comment that

on so many trials showing this with drugs and drugs

being different--I mean, carvedilol was brought up

earlier and that is an interesting example. In the

large trials done outside the U.S. the effect size

of carvedilol was smaller than in the U.S.

carvedilol trials.

DR. TEMPLE: Well, you tend to notice it

when the U.S. doesn't do well--

[Laughter.]

--so there is probably some selection. We

have actually done an internal analysis and there

is some suggestion of it but it is mostly driven, I

think and I don't know if Norm agrees, by the two

studies that formed the hypothesis, RENAL and IDNT.

Those didn't look so conspicuous. You know, you

are not supposed to use the ones that form the

hypothesis, but it is certainly an interesting

question.

I have one other question. If you look at

hyperkalemia can you show any relationship to what

dose of diuretic people were on? Should the dose

of diuretic be higher in people who are getting

both of these drugs?

DR. PFEFFER: I was bragging about our

case report forms. We had doses of the ACE

inhibitor, doses of the beta-blocker. We did not

have doses of diuretics which changes during time,

so I could not tell you that.

DR. TEERLINK: Mark, was there entry

criteria for blood pressure in this trial?

DR. PFEFFER: I mentioned that Dr. Yusuf

was part of the executive committee so let's make

this broad; let's make this inclusive; let's not

have a blood pressure level as long as people are

talking to you and are not symptomatically

hypotensive. So, we did not have a cut-off for a

low blood pressure.

DR. TEERLINK: The reason I ask is because,

obviously, given that we are only considering

additive therapy here and clinicians only have so

many millimeters of mercury to spend, and in slide

CS-4 there is a conspicuous increase, as one would

expect, in terms of the increase in hypotension in

patients who start out with a blood pressure that

is already borderline low. Then we also recognize

that many adverse events can spin off that

hypotension so you can have hypotension that then

leads to renal failure and then leads to other

aspects. Is there a blood pressure--and we can

choose 100--at which the risk to benefit of

candesartan in addition to other therapies is no

longer favorable?

DR. PFEFFER: John, it is a tough question

because one person's blood pressure of 98 and

another person's blood pressure of 98 are totally

different, as you know. So, by opening the door

and allowing these patients in we have a total

experience of about 120 patients. They are

vulnerable patients. A patient who walks around

with symptomatic heart failure and blood pressure

less than 100 is more likely to have an adverse

event, and more likely to discontinue due to

hypotension. So, it is the person you want to put

on the medications and are unable to.

100

So, everything I have been showing you is

intent-to-treat but I will show you, John, in

direct answer to your question that for

hypotension, if you came into this trial with a

blood pressure less than 100 systolic, and only 54

of the placebo patients did and they not

infrequently had to be discontinued, but then

trying to add the active therapy, we discontinued

their medication. Now, that is not a demerit.

Investigators tried. This is a blinded study

medication. They discontinued and everything I

have shown you has been intent-to-treat.

DR. NISSEN: Another way to look at it is

that in spite of allowing these patients in the

trial it didn't undermine the results. So, I

presume those people didn't end up on much

candesartan.

DR. PFEFFER: They didn't. That is why I

was bringing back the intent-to-treat not the per

protocol.

DR. HIATT: I have a slightly different

question. I tried to resolve the results of this

101

development program with the other ones,

particularly the valsartan. I think a number of

questions can be raised in that regard but I am

struck by the interaction in ValHeFT between ACE

inhibitors, beta-blockers and the addition of an

ARB showing a worse outcome in contrast to your

data. Could you speak to that?

Then I have a follow-up question related

to that, and that has more to do with the

pharmacokinetics of these different agents.

Valsartan has a very long half-life; candesartan

has less. I am worried about the receptor

interactions and how they might differ because are

all these ARBs created equal is sort of where I am

going with this.

DR. PFEFFER: These are key clinical

questions and you can imagine the question in the

year 2003 when we came up with these results. I

have no more insight than the distinguished panel

but I will give you my personal views. The

question was of the agent, and I would have to say,

no based on the VALIANT experience where a good

102

number of patients were on so-called triple therapy

and harm was not seen.

We are showing no harm and benefit. I

think that is the message. If you look at overall

the entire ValHeFT experience there is consistency.

It is just when you get to that particular

subgroup. And that is where I gave you the

numbers. You have to look at the robustness of one

subgroup and another. We happen to have more

events because we had a higher use of beta-blocker

and longer follow-up. But beyond that I would be

speculating.

DR. HIATT: Can anyone from the company

sponsor distinguish some of the PK potential

differences--dwell time on the receptor, those

kinds of things, between these different agents?

DR. PFEFFER: I am sure somebody from the

company can tell you about the PK differences.

DR. NISSEN: What do you say we do that

after the break so you, guys, can kind of gather

your thoughts together? I am actually give you

some thoughts; I was on that ValHeFT panel and also

103

on a panel that reviewed candesartan compared to

losartan, and I will give you some thoughts about

that that might help you understand this. Let's

break for about 15 minutes. We are doing very

well, everybody.

[Brief recess.]

DR. NISSEN: If everybody can take their

seats we will try to get started again.

Bill, before the break you asked about

differences in any ARBs, and I can offer a little

bit of perspective. Sometimes there is a little

institutional memory around here and I served on

the advisory panel for ValHeFT and we also looked

for comparative data between losartan and

candesartan. I think both were helpful to me in

understanding some of this. At the time the

ValHeFT data were presented there were a number of

us on the committee that were very suspicious that

the result, the beta-blocker hazard--you know, the

triple therapy hazard observation was spurious.

One of the reasons is that that particular

analysis, as I recall, was not really prespecified

104

so it was an exploratory analysis. You know, I

opine that you really couldn't--shouldn't make any

regulatory decisions on that basis; that it was

hypothesis generating at best and that, again, if

you look at enough trials and enough people and

enough subgroups you are going to see something

like that happen once in a while.

I must say, it was very intense. The

final vote was 4-4, which meant that we actually

had an even number so we didn't actually make a

decision on the primary indication for valsartan.

Even though the nominal p value looked very good

and the data looked very good for the overall

study, at the time I felt like people were being

unduly influenced by the observational data on the

subgroup. I think now, in retrospect, that

probably was spurious. That is my own personal

interpretation that it was just simply an unusual

result.

DR. HIATT: Where I was sort of going with

this, is there really a difference in dosing

between ARBs, or are there different pharmacologic

105

differences that we should be recognizing between

ARBs?

DR. NISSEN: There is some subtlety here.

Again, there are obviously things that are class

effects and there are things that are not class

effects. We looked at two trials comparing

losartan and candesartan, and this committee voted

I think unanimously that there was evidence that

the blood pressure lowering effect was greater with

candesartan than with losartan, both given in their

full therapeutic doses.

DR. TEMPLE: Well, the labeled full

therapeutic dose.

DR. NISSEN: Yes.

DR. TEMPLE: I think we all had the

impression that losartan probably should be higher

but wasn't pushed.

DR. NISSEN: Yes.

DR. TEMPLE: Whatever the reason, they

beat them.

DR. NISSEN: Yes. But what we can say is

that 32 mg of candesartan had a very big effect on

106

blood pressure, bigger than the full doses of

another ARB. So, it is like any other therapeutic

class, there are sometimes agents that are somewhat

more potent than others, that perhaps have more

affinity for the receptor. So, if you want to test

the hypothesis that blocking at the AT-I receptor

produces an added benefit you want to probably do

it where you are really blocking the receptor as

well as you can block it, and I think that is one

of the things that CHARM did. They got to a really

very robust dose of a very potent angiotensin

receptor blocker so it really does test the

hypothesis.

DR. TEMPLE: Of course, our concern has

been you can only test it if you really are on

whatever the full dose is, but a full dose of the

ACE inhibitor. That is what has been addressed

here. In the case of ValHeFT, that was sort of a

very Bayesian episode. We actually approved the

use on what was not a primary analysis at all. I

mean, that was just an accidental 7 percent of the

people that weren't on any other drug. That is

107

what we approved, even though that was only

300-some odd patients in a 5000 patient trial

because the result was so conspicuously large. The

beta-blocker thing, we were skeptical about it too

but it was the mortality outcome and we just didn't

feel we could say anything about it.

DR. SACKNER-BERNSTEIN: Also, in terms of

the mortality in ValHeFT, I wasn't part of the

committee but the randomization in that trial was

stratified based on background beta-blocker therapy

which does add some robustness to that analysis,

just to clarify that.

DR. HIATT: Before I leave this question,

is there anyone from the sponsor who can talk about

the differences in the pharmacokinetics and

dynamics of these different ARBs? I mean, I was

struck that valsartan has a longer half-life. It

is certainly a less potent drug but then it is just

a matter of milligrams. If you can get them to the

same equivalent dose you should overcome that but,

if anything, candesartan maybe should be dosed more

frequently. So, I am just questioning whether there

108

are any other pharmacologic properties between

these agents we should be discussing today.

DR. NISSEN: That is a fair question so

can somebody just tell us about PK and PD data?

DR. YOUNG: I can give you a clinician's

perspective because this is important when we are

setting up a lot of these clinical trials and we

are looking at this, and all these are different

molecules and it gets at this issue of variability

from an ACE inhibitor to an ACE inhibitor, from a

beta-blocker to a beta-blocker, an ARB to an ARB,

and there are differences, some of them subtle and

some of them may translate into outcomes data that

are important.

But with respect to the ARBs, candesartan

is the most tightly bound of the ARBs. It has an

insurmountable binding property, sort of a

non-competitive type of binding property that lasts

well over 24 hours. You can detect effects in

binding activity after 24 hours, and what happens

is that the PK levels will go up and drop and the

half-life will appear to be less when you are

109

looking at it from a PK or a drug exclusionary

phenomenon, but if it is still tightly bound to the

receptor you won't have candesartan coming off the

receptor and causing it to go back up.

DR. HIATT: And that was my understanding.

That is where I was going with this. I wanted to

say that because I think valsartan does not have

that same kind of receptor affinity. Am I correct?

DR. YOUNG: It does not; you are correct.

DR. HIATT: So, if it really is bound

across the 24-hour dosing cycle and has a very high

affinity there could be a pharmacologic basis for a

slightly different clinical result.

DR. YOUNG: And I stress the word "could

be."

DR. NISSEN: Yes, Tom?

DR. PICKERING: I would like to discuss

further the issue of hyperkalemia and

spironolactone use. I think the issue here is

really one of labeling and whether it should

specifically say anything about whether patients

should be also taking spironolactone or not. If

110

you look at slide CE-17, there doesn't seem to be

any advantage of being on spironolactone in terms

of the primary outcome variables. Although there

is a trend in the lower panels for improved

mortality, I guess it is not significant.

The other one was CS-8 which shows that

the hyperkalemia occurrence is increased in

patients taking spironolactone in diabetics, and so

forth. I guess the reason for the concern was the

publication in The New England Journal about what

happened after the RALES trial was published, that

the hospitalization rate for hyperkalemia rose from

2.4 per 1000 to 11 per 1000 with an increase in

mortality. You know, obviously, in this trial

everything was very nicely controlled and people

were doing what they were supposed to be doing, but

what will the consequences be when it sort of gets

out into the real world? So, perhaps we could

discuss that.

DR. PFEFFER: Certainly, inhibiting

renin-angiotensin system does have its issues and,

fortunately, one of my colleagues wrote the

111

editorial that accompanied that New England Journal

article. So, let me ask Dr. McMurray to talk about

that.

DR. MCMURRAY: We share your concern. In

fact, I think the only reason I wrote that

editorial was that we had already published our own

experience with the misuse of spironolactone which

became widespread after the publication of the

RALES trial and I think that was a lesson from the

Ontario experience and, indeed, from 13 other case

series that have been published reporting the same

thing in smaller numbers of individuals. The

striking thing about that was that essentially it

boiled down to two problems, the use of the wrong

dose of spironolactone, much higher than the small

dose used in RALES which was 25 mg a day, and also

misuse in the wrong patients. So, RALES was a

study targeted at a carefully defined group of

patients and the Ontario experience with

spironolactone was used in a completely different

patient population, much older; many patients with

preserved rather than low ejection fraction; more

112

diabetics, and so on.

So, one of the points I tried to make in

that editorial was that RALES was a very unusual

trial in one respect, and that was that it was a

trial done with a generic drug that had no sponsor

and the usual care in terms of risk management, in

terms of educational programs, in terms of meetings

and so on, to emphasize how you must use the drug;

you must monitor what happens to patients. I think

perhaps I would look more to the experience with

ACE inhibitors where they were used in a more

responsible way because there was a sponsor acting

behind them to ensure that the program education

was carried out. Unfortunately, that didn't happen

after RALES. Certainly my personal interpretation

would be that the reason there have been major

problems is because people didn't go through the

usual process of introducing a new treatment and

ensuring it was used as carefully as possible.

DR. TEMPLE: There is, of course, no

labeling of any spironolactone product reflected in

RALES despite my attempts to embarrass people into

113

producing one.

[Laughter.]

For fairly obvious reasons, it was

unsuccessful.

DR. PICKERING: But if this gets approved

there is going to be labeling that could or could

not say something about concomitant spironolactone

use.

DR. TEMPLE: Indeed, it could. Actually,

my question from before goes. I mean, everybody

has moved down to low doses of diuretics because

they are worried about hypokalemia. Maybe they

should make a comeback in the face of all this

potassium retention. Higher doses do work slightly

better than 12.5. That seems worth exploring too.

DR. NISSEN: Actually, it does reflect a

real problem for clinicians in managing heart

failure. The computer term for it is combinatorial

explosion, which is you have four or five therapies

and how do you combine, what kind of combinations

and permutations of them can be used in individual

patients. It is not so easy. I often don't know

114

what to do so I am looking for guidance from FDA.

DR. TEMPLE: So, you don't think anybody

can just make a single pill that will just do it?

DR. NISSEN: Yes, I don't think so. I

wanted to explore something else with you, guys.

Obviously, one of the reasons we are here is

because the agency reviewer and the agency has some

concerns about has the hypothesis been proven that

on top of maximal doses of ACE inhibitors

candesartan produces an incremental benefit. This

all could have been resolved if you just picked

enalapril, you know, pushed it to the heart failure

doses and then everybody would have gotten the same

ACE inhibitor and we would know exactly what they

got.

I know what your answer is going to be.

Your answer is going to be you wanted to make this

a real-life trial with the real-life drugs that

people use, but it does, in fact, undermine a

little bit our ability to interpret the experiment.

Did you, guys, consider actually just specifying

the ACE inhibitor, pushing it up in the way they

115

did in the ACE inhibitor trials and then, once you

got to the maximum tolerated dose, randomize?

DR. PFEFFER: Well, I gave you the names

of the people involved in the planning so you can

imagine we did consider it. The other issue would

be I could imagine if we came back here with the

same findings on the most commonly used medication,

which was enalapril, somebody--I am not saying

who--

[Laughter.]

--would say what about the other ACE

inhibitors, the other approved ACE inhibitors?

Then we realized that to dictate the use of an ACE

inhibitor, with the VA system telling us what ACE

inhibitor you have to use, my healthcare system

telling us what ACE inhibitor you have to use, we

really did make the decision to optimize the

individual dose and see if adding on improves

outcome.

DR. NISSEN: Although you did allow use of

ACE inhibitors that were not approved for heart

failure. Was the assumption that everybody would

116

feel okay about that, that even though the drug

wasn't actually approved--

DR. PFEFFER: Well, this was FDA approved.

We are talking about 26 countries. I was reminded

in this international trial that the U.S. is one

country.

[Laughter.]

DR. NISSEN: Yes, we are getting reminded

of that all the time now. Other questions? Yes,

Jonathan?

DR. SACKNER-BERNSTEIN: Just to get back

to the U.S.-non-U.S. finding, there are a couple of

different ways that I was trying to look at this to

see if I could understand it. Obviously, it could

just be a statistical fluke, which my own bias says

is the most likely. But one issue that has come up

before is the possibility of drug interactions.

So, I am assuming that you looked and found that

U.S. and non-U.S. subjects were treated similarly.

A second one has to do with whether the statistical

power was sufficient within the U.S. population,

and I think we addressed that by the question

117

earlier. The third question has to do with what is

unlikely but possible, that perhaps the people in

American who have systolic dysfunction are already

treated with an ACE inhibitor. There is a potency

issue about the way candesartan works compared to

the way it works in similarly described patients

outside of the U.S.

One of the ways I would like to get a

handle on that is by seeing what the AE effects

were. If you were to tell me that by region the

North Americans had a very low rate of renal

insufficiency, a very low rate of hypotension, then

I would have the bias that perhaps we are looking

at a differential potency in a population. I

wouldn't understand why. Perhaps I am putting

myself at risk of attack from pharmacologists but

that is the way I have thought about this and I am

wondering if you looked at that data.

DR. PFEFFER: Let me first start by what

it is. I reject that there is anything here

personally. So, if you are asking me to defend

what it is, I can't do that because I think there

118

was nothing there.

But if you want to explore something where

I don't believe it, I can tell you there are a lot

of differences between U.S. patients and non-U.S.

patients at baseline. You heard that their

outcomes are pretty much the same, and you heard

that in the trial of 7599 in the program the effect

of candesartan was pretty much the same.

But just to explore, in CHARM-Added, yes,

there were some differences, fairly minor, in the

medication use but here is medication use. Now, I

mentioned that being at the recommended dose, and I

think Ralph pointed out that the arrow went in a

good way so you can see the inconsistency, there

are more people at the recommended dose. So, there

are a lot of inconsistencies here.

Let me show some more differences between

U.S. and non-U.S. We do more procedures. That is

no revelation. Coronary procedures, we are very

good at that. That did not influence anyone's

outcome. We do more angioplasty. We have ICDs and

pacemakers. So, procedures we do more of. I am

119

off the cuff going to ask Dr. McMurray, did we do

any quality of life? Did U.S. patients feel better

with all this hardware?

DR. MCMURRAY: Only in the U.S.

DR. PFEFFER: Quality of life was only

done in the U.S. Now, for AEs we can give you this

by North America. Is that okay? So, we can go

across the border and I think it is important to

look at placebo. Placebo in the U.S. were more

likely to tick a box, and we really asked these

questions--renal function, 6.3 versus 2.9. I am

not going to make anything of it but numerically

more. Obviously, the agent increased that in both

North America and the rest of the world. Here is

the hyperkalemia, increased in North America;

increased by the same factorial in the rest of the

world.

So, Jonathan, I don't see that there is a

clue here that they are under-treated,

over-treated; that the SAEs are helping us with

this. I go back to your first statement of fluke

but I don't even say fluke because I don't make the

120

observation.

DR. NISSEN: It is intriguing though,

Mark. I mean, some of the therapies like

defibrillators do have an impact and, you know, the

fact that there were more defibrillators used in

the U.S.. You know, one of the mechanisms of

death--you know probably better than I--in these

patients is sudden death. So, it is possible that

there is a competition for benefit between

defibrillators and more effective heart failure

treatment. If, in fact, there is more

defibrillator use in the United States there may be

less opportunity for benefit from candesartan. So,

some of these hypotheses, and they are just

hypotheses--I basically agree with you but when you

see an observation, it is our responsibility

obviously to explore that and make sure we

understand it, that there is some strong signal

here and I think there is not a signal; I think

there is an observation. I think you can see how

the defibrillator use could certainly drive some of

this.

121

DR. PFEFFER: And defibrillator use is

something that in the year 2005 we are much smarter

than in 1999. I don't know what the balance would

be around the world now but these are heart failure

patients and I have some of my heart failure

colleagues telling me to turn these things off

sometimes too. So, I don't have the answer for

that.

DR. TEMPLE: Of the U.S. differences you

showed, one of them is sort of tempting. More U.S.

patients were on the full dose so maybe that would

explain why the addition didn't work as well, but

your overall data shows that people who were on a

full dose on the whole did better.

DR. PFEFFER: I mentioned that as an

example of a confounder. The point estimate for

being on full dose moved in the right direction.

More U.S. were on the full dose so it is a perfect

confounder--

DR. TEMPLE: Right.

DR. PFEFFER: A fluke, and I just think it

is a great example. Dr. Granger has something to

122

add.

DR. GRANGER: We did look at this. One of

the obvious things is procedure use and prior

re-vascularization. When we looked at prior ICD or

prior re-vascularization the point estimates were

almost identical for the treatment effect of

candesartan. So, it doesn't appear to be that.

DR. NISSEN: I would have guessed that

having angioplasty would increase the event rate

because we all know that angioplasty is bad for

you--

[Laughter.]

--but I guess you didn't see that.

DR. PFEFFER: We don't know how long ago

the angioplasty was or where it was done.

DR. NISSEN: Did you enroll any patients

at the Cleveland Clinic?

DR. PFEFFER: Cleveland Clinic was a

vigorous proponent of conducting the CHARM trial

and Jim was the U.S. lead investigator. He

probably asked you about some of your patients.

DR. PICKERING: Could I raise the issue of

123

African Americans? I think you had 2.8 percent and

the issue is if this gets approved what are we

going to say about its use in black patients?

Because there is evidence that blocking the

renin-angiotensin system may not be so effective in

African Americans. At two meetings ago we reviewed

a drug which was basically killed because of

adverse effects, angioedema, which is commoner in

African Americans, and there seems to be a total

void here. Should clinicians be using it in

African Americans or not? Or, what are we going to

say?

DR. PFEFFER: I think we have as much

confidence in our data as any that have been

presented here, and let me walk through that.

This is self-designated as

black--self-designated. In CHARM-Added, of the

70-something patients there is the point estimate.

I am not saying that it is this way or that way.

That was Alternative. That was not on an ACE

inhibitor. In CHARM-Added, in the few patients we

had you can see the point estimate here. But if

124

you go through our whole program I think there is a

consistent message here that designating yourself

as black and then being enrolled in our study there

is no loss of efficacy, and my interpretation would

be we are offering an opportunity to reduce

someone's risk regardless of this designation, and

that is the best estimate even the point goes this

way. When we do the total, we are talking about

over 300 patients.

DR. NISSEN: While other people are

thinking, Mark, let me tell you what triggered my

request for the time to all-cause hospitalization.

I did some sort of simple numerics and I see there

were 56 fewer deaths or CHF hospitalizations in the

primary endpoint. So, you avoid 56 deaths in the

primary analysis. Then I looked at the hypotension

and there are 33 more people hospitalized for

hypotension. So, at least in your mind, until you

see an analysis you have to say, well, you kept 56

out of the hospital and from dying but you had 33

that had excess hospitalizations and you have 20

excess hospitalizations for renal dysfunction and

125

then you have another 10 in the hyperkalemia.

So, when you add all the numbers up, you

know, you sort of see an analysis that says, well,

you are keeping people out of the hospital for

heart failure but you are admitting a lot more to

the hospital for AEs, so isn't the hospitalization

data kind of a wash? I know it is not the correct

analysis because once you have that first heart

failure or hospitalization you may have more. That

is why I am so keen on seeing that.

DR. PFEFFER: I think it is a key number

to get you but we do have it without Kaplan-Meiers

and Dr. McMurray would like to tell you about total

hospitalizations.

DR. MCMURRAY: I am afraid I don't have a

slide of this but I do have the numbers so you

might want to write them down. I was intrigued for

my own interest to figure out how it balances up.

On the benefit column what we actually have, and I

will give it to you per 1000 patients treated over

the duration of the study--on the benefit column

there were 46 fewer patients hospitalized for heart

126

failure. There were 100 on ACE fewer heart failure

hospitalizations and 35 fewer cardiovascular

deaths.

On the risk side there were 26 more

patients hospitalized with hypotension, but when I

say with hypotension that means hypotension was

just on the list of possible causes for that

hospitalization. For example, amongst those there

were people with septicemia, people with GI

bleeding, and this is true for all the AEs. There

were 16 extra hospital admissions for renal

dysfunction and there were 8 extra hospital

admissions with hyperkalemia. Again, some of those

groups overlap but we weren't able to quite tease

that out.

In summary, the balance was substantially

in favor of candesartan and, in fact, I can give

you sort of a handle on that because we have done

an economic analysis in Europe and a resource

utilization economic analysis, and over the course

of the study for every 1000 patients treated with

candesartan there were 1900 fewer days in hospital

127

with worsening heart failure. There were

significantly fewer days in hospital for any reason

whatsoever in the candesartan group. So, yes, of

course, there is a trade-off but it is

substantially less on the benefit side in terms of

morbidity and resource utilization.

DR. TEMPLE: I just added up your numbers

leaving deaths out of it for the moment, not that

you necessarily want to. There was a 46-patient

benefit for heart failure hospitalizations.

DR. MCMURRAY: Forty-six patients, yes.

DR. TEMPLE: And 49 extra hospitalizations

for hypotension, renal dysfunction and

hyperkalemia.

DR. MCMURRAY: Okay, the difference there

is--well, there were several differences. First of

all, you have picked patients as opposed to

admissions and, secondly, on the risk side when I

said hypotension, when I said renal dysfunction,

when I said hyperkalemia I really do mean that if

those terms appeared anywhere on the long list of

reasons for admission we counted that just in case

128

it could be a risk. Also, there was overlap. The

best estimate I can give you of overlap, and I

really don't know the proper numbers but the best

estimate of overlap is two-thirds of those patients

were counted more than once.

DR. TEMPLE: Okay, but those were extra

hospitalizations in the treated group.

DR. MCMURRAY: Extra hospitalizations,

yes. So, the contrabalancing number for that is

188.

DR. NISSEN: Bob, I understand what he is

saying and I want to just see if I can rephrase it.

You know, if you take the number of patients that

had a hospitalization for either heart failure or a

drug AE, it is fairly balanced. But once you got

admitted once for heart failure you are very much

likely to be admitted again and again. So, what

they showed us was the Kaplan-Meier for cumulative

incidence of all-cause hospitalization. And I

understand that. And it is very important and I am

not minimizing it at all. But, you know, it did

strike me that there was a cost for that, and the

129

cost is that a fair number more patients--when you

talk about AEs I look at hospitalized AEs rather

than I do incidental AEs that are sort of

discovered on a laboratory test. If you have

hyperkalemia sufficient to land yourself in the

hospital, that is a pretty serious AE, and if you

have hypotension that gets you in the hospital,

that is a pretty serious AE. So, that is why I am

so keen on seeing that time to first event because

that is an important objective. Now, I know that

over time the hospitalizations are clearly less in

the candesartan arm. But I am going to guess

that--

DR. TEMPLE: Yes, and the implications are

different. One is transient, you fix it and it is

over--

DR. NISSEN: Yes.

DR. TEMPLE: But being hospitalized for

heart failure means you are on the way to troubles.

DR. NISSEN: You are on a downward spiral.

Don't misunderstand me, I am not placing equal

weight on them.

130

DR. MCMURRAY: I was trying to give you

actual numbers.

DR. NISSEN: Yes. Obviously, FDA is going

to have to write a label and we have to understand

this as well as we can in order to help them

understand it.

DR. PFEFFER: Dr. McMurray was looking at

pharmacoeconomics and multiple admissions. I think

what he was explaining is that for hyperkalemia and

hypotension, we can count both of those for the

same admission just to be on the safe side. But I

have also learned something--when I go over there

and sit down I become a little smarter, and people

have now fed me the numbers for the total

hospitalizations as a function of time with your

question about the early hazard. I didn't know

this answer so it is new for me too, and it was a

very appropriate question, what happens in the

first month. May I share that slide or do I read

numbers--I don't have a slide; I read numbers.

So, at the first month, which is that

up-titration phase, for hospitalization for any

131

reason, 69 of the candesartan patients and 80 of

the placebo. At 6 months it is 297 and 304. Then,

as a function of time we get better, as you see.

That doesn't mean we didn't hurt somebody early but

in the overall, all-cause hospitalization for any

reason numerically people were on the candesartan.

Then you did see the curve of the cumulative

hospitalizations.

DR. NISSEN: It actually does sort of

support the hypothesis that you are really picking

up the benefits once you get outside of that early

sort of titration. Once you have proven you can

tolerate the agent, then you are starting to

accumulate lots and lots of benefit.

DR. PFEFFER: Well, I really have trouble

with when was the benefit. I know you spent some

time on that last week--when is the benefit. I

don't know what statistical tool one uses to do

that besides your eyeball. So, why don't we look

at our two low EFs combined? You know, we did a

lot of statin work, as you have, and for the most

part, except for a few studies, you need a little

132

time to see the benefit unless you are very, very

aggressive with your statin use. Treating heart

failure, symptomatic heart failure, you tend to

start to see things early.

So, Dr. Nissen, I don't know what to make

of this, of when, but I do think we are starting to

see the benefits that you would ask for in a

medication for the treatment of people with

symptomatic heart failure and, yes, there are other

things that we must be vigilant to look for. It

happens in placebo too so I think we need to raise

our standards of how to monitor patients whether

they are on the triple therapy or not.

DR. SACKNER-BERNSTEIN: In terms of the

endpoint of heart failure hospitalization that was

part of the primary endpoint, I am wondering if you

might comment on how you can be confident that you

captured all the heart failure hospitalization that

occurred appropriately. Literature, including the

RESOLVe trial has shown that about as many as 11

percent of heart failure hospitalizations are

associated with pulmonary processes. So, I am

133

curious about how you made sure that the endpoint

committee saw all the hospitalizations that an

investigator may have thought were just bronchitis

or pneumonia and may have actually been given IV

diuretics. Another issue is that of worsening

renal function which certainly can be a sign of