1
DEPARTMENT OF HEALTH AND HUMAN
SERVICES
UNITED STATES FOOD AND DRUG
ADMINISTRATION
CENTER FOR DRUG EVALUATION AND
RESEARCH
PULMONARY-ALLERGY DRUGS
ADVISORY COMMITTEE
Monday, June 6, 2005
8:00 a.m.
5600 Fishers Lane
Room 1066
Rockville, Maryland
2
P A R T I C I P A N T
S
Erik R. Swenson, M.D., Chairman
Teresa Watkins, R.Ph., Executive
Secretary
MEMBERS:
Mark L. Brantly, M.D.
Steven E. Gay, M.D., M.S.
I. Marc Moss, M.D.
Calman P. Prussin, M.D.
Theodore F. Reiss, M.D., Industry
Representative
Karen Schell, RRT, Consumer Representative
David A. Schoenfeld, Ph.D.
SGE CONSULTANTS AND GUESTS (VOTING):
Jeffrey S. Barrett, Ph.D.
Lawrence Hunsicker, M.D.
Allan R. Sampson, Ph.D.
Jurgen Venitz, M.D., Ph.D.
Mary Lou Drittler, SGE Patient
Representative
GOVERNMENT EMPLOYEES (VOTING):
James Burdick, M.D.
Roslyn B. Mannon, M.D.
Michael A. Proschan, Ph.D.
John Tisdale, M.D.
FDA STAFF:
Mark J. Goldberger, M.D., M.P.H.
Renata Albrecht, M.D.
Marc Cavaille-Coll, Ph.D.
Arturo Hernandez, M.D.
Jyoti Zalkikar, Ph.D.
3
C O N T E N T S
Call to Order and Opening Remarks,
Erik R. Swenson, M.D. 4
Conflict of Interest Statement,
Teresa A. Watkins, R.Ph. 7
FDA Introductory Remarks,
Renata Albrecht, M.D. 8
Sponsor Presentation:
Introduction, Michael Scaife,
Ph.D. 14
Current State of Lung
Transplantation,
Jeffrey Golden, M.D., University of
California, San Francisco 20
Clinical Evidence of Efficacy and
Safety,
Sarah Noonberg, M.D., Ph.D. 28
Statistical Considerations,
Ronald W. Helms, Ph.D.,
Rho, Inc.; University of North
Carolina 59
Safety and Benefit-Risk,
Stephen Dilly, M.D., Ph.D. 65
Questions from the Panel 70
FDA Presentation:
Overview of Clinical Trial Efficacy
and
Safety Evaluation Discussion of
Analysis,
Arturo Hernandez, M.D. 88
Safety Considerations and
Conclusions,
Marc Cavaille-Coll, M.D., Ph.D. 105
Statistical Evaluation,
Jyoti Zalkikar, Ph.D. 111
Questions from the Panel
(Continued) 121
4
C O N T E N T S
(Continued)
Open Public Hearing:
Esther Suss, Ph.D. 151
John C. Sullivan 156
Bill Stein 158
Renee Moeller 162
Charge to the Committee, Renata
Albrecht, M.D. 163
Committee Discussion and Vote 168
5
P R O C E E D I N G S
Call to Order
DR. SWENSON: Good morning, everyone. I
am Dr. Erik Swenson, from the University
of
Washington, and I will be chairing this
session.
This is the meeting of the Pulmonary and
Allergy
Drugs Advisory Committee and today we
are going to
be discussing inhaled cyclosporine, a
product to be
presented by Chiron.
Let me begin with just a few
items to keep
us on schedule and for organizational
purposes.
One, I would request that everyone with
cell
phones, please turn them off or at least
down to
some vibrating or some innocuous
mode. Then, we
will go around and introduce everyone
here at the
table.
I would ask that when you are questioning
anything during this meeting to please
identify
yourself first. The transcriber will need to know
who is speaking. We have microphones here. All
you need to do is simply push down
"talk" to go
ahead and be heard but, please, turn it
off when
you have finished. If we get more than three
6
microphones on at one time things get
confusing.
Without any further ado, I am going
to
turn the meeting over to Dr. Teresa
Watkins for
some introductory comments.
Introduction of the
Committee
DR. WATKINS: Let's first go around the
table, starting with Dr. Reiss, if you
will
introduce yourself and your
affiliations, please?
DR. REISS: My name is Ted Reiss. I am
vice president of clinical research at
Merck
Research Labs. I am the non-voting industry
representative.
DR. BRANTLY: My name is Mark Brantly. I
am from the University of Florida. I am a
professor of medicine.
DR. TISDALE: My name is John Tisdale and
I am in the intramural program of NIDDK.
DR. PRUSSIN: My name is Calman Prussin.
I am a clinical investigator with
National
Institute of Allergy and Infectious
Diseases.
DR. MANNON: I am Roslyn Mannon and I am a
transplant nephrologist and medical
director of the
7
intramural solid organ transplant
program at NIDDK.
DR. GAY: I am Steven Gay, assistant
professor at the University of Michigan,
associate
director of the lung transplant program
and
director of clinical support services.
DR. HUNSICKER: I am Larry Hunsicker, from
the University of Iowa. I am a transplant
nephrologist and professor of medicine,
and I am a
member of the Chemical Immunosuppression
Advisory
Committee but guesting on this one.
DR. VENITZ: I am Jurgen Venitz. I am a
clinical pharmacologist and associate
professor at
Virginia Commonwealth University.
MS. DRITTLER: I am Mary Lou Drittler. I
am a
lung transplant recipient and I am a patient
representative from here, in Silver Spring.
DR. BURDICK: I am Jim Burdick. I am
director of the Division of
Transplantation and
Healthcare System, HRSA and a transplant
surgeon.
DR. MOSS: I am Mark Moss. I am an
associate professor of medicine at Emory
University
and section chief at Grady Memorial
Hospital.
8
DR. BARRETT: I am Jeff Barrett. I am a
clinical pharmacologist from the
University of
Pennsylvania and Children's Hospital of
Philadelphia.
DR. PROSCHAN: I am Mike Proschan and I am
a statistician from the National Heart,
Lung and
Blood Institute.
DR. SCHOENFELD: I am David Schoenfeld. I
am a biostatistician and professor of
medicine at
Harvard Medical School and Massachusetts
General
Hospital.
DR. SAMPSON: I am Allan Sampson,
professor of statistics, Department of
Statistics
at the University of Pittsburgh.
MS. SCHELL: I am Karen Schell. I am a
respiratory therapist from Emporia
Kansas, and I am
the consumer representative.
DR. CAVAILLE-COLL: I am Marc
Cavaille-Coll, medical team leader,
Division of
Special Pathogen and Immunologic Drug
Products.
DR. ALBRECHT: I am Renata Albrecht,
director, Division of Special Pathogen
and
9
Immunologic Drug Products.
DR. HERNANDEZ: I am Arturo Hernandez, a
medical reviewer for FDA, Division of
Special
Pathogens and Immunologic Drug Products,
and I am a
transplant surgeon.
Conflict of Interest
Statement
DR. WATKINS: With that, thank you.
Welcome everyone. I am now going to now read the
conflict of interest statement.
The following announcement
addresses the
issue of conflict of interest with
regard to this
meeting and is made a part of the record
to
preclude even the appearance of such at
this
meeting.
Based on the submitted agenda
for the
meeting 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 respect to FDA's invited
industry
10
representative, we would like to
disclose that Dr.
Theodore Reiss is participating in this
meeting as
a non-voting industry representative
acting on
behalf of regulated industry. Dr. Reiss' role on
this committee is to represent industry
interests
in general and not any one particular
company. Dr.
Reiss is employed by Merck.
In the event that the
discussions involve
any other products or firms not already
on the
agenda for which an FDA participant has
a financial
interest, the participants are aware of
the need to
exclude themselves from such involvement
and their
exclusion will be noted for the record.
With respect to all other
participants, we
ask in the interest of fairness that
they address
any current or previous financial
involvement with
any firms whose products they may wish
to comment
upon.
Thank you. With that, we will
have opening
remarks from Dr. Albrecht.
FDA Introductory
Remarks
DR. ALBRECHT: Thank you, Dr. Watkins.
Good morning, everybody. On behalf of the Division
11
of Special Pathogen and Immunologic Drug
Products
and the Office of Drug Evaluation IV, I
would like
to welcome everyone to today's meeting.
We wish to thank the members
of the
Pulmonary Advisory Committee, the Chair,
Dr.
Swenson, and our consultants for taking
the time
out of their schedules to come to
Rockville and
join us here to discuss this
application. I also
wish to express our appreciation to
Chiron and the
investigators for the time and effort
that they
have put into developing this drug
product and to
the Chiron staff for their willingness
and
preparation for this advisory committee
meeting. I
would also like to recognize the
dedication of the
Division staff and the long hours they
have put in
for reviewing this application.
Let me speak briefly about
this new drug
application for cyclosporine
inhalational solution
and why we are bringing this application
to the
advisory committee. Could I have someone run the
slides?
I apologize, there are some slides that go
with this presentation so that you may
follow
12
along.
Let me continue. There are currently no
FDA-approved products for the prevention
of chronic
rejection in patients with lung
allografts. There
are approximately 1100 transplants done
in the U.S.
annually and the survival at five years
is lower
than survival in other organ transplants
such as
heart, kidney or liver transplants. Prevention of
rejection and increase in survival are
critical
and, therefore, there is a clear need
for safe and
effective therapy.
Next slide. Chiron has submitted the NDA
for Pulminiq and requested that the
cyclosporine
inhalational solution be approved for
the increase
in survival and prevention of chronic
rejection in
lung transplant patients. The drug
development
program and the NDA for this product are
not
conventional. Unlike applications for
immunosuppressants in kidney, heart of
liver
transplants for example, this NDA
contained results
from one single Phase II study conducted
at one
center.
This trial enrolled 66 patients out of a
13
planned 136 patients. However, we learned that
there was a survival advantage, 88
percent survival
in patients who received aerosolized
cyclosporine
plus a tacrolimus-based systemic
immunosuppressive
regimen compared to 53 percent survival
in patients
receiving aerosolized propylene glycol
vehicle in
addition to a tacrolimus-based systemic
immunosuppressive regimen.
Therefore, the agency agreed
to file and
review this NDA application. Based on the NDA
review of the information in the
application, we
were unable to conclude that the
observed
difference in survival and chronic
rejection was
due to study drug. Therefore, we determined it was
important to bring this application to
the advisory
committee for the following reasons:
This would represent the first
drug for
immunosuppression in patients with lung
transplantation to garner FDA approval. This is a
new drug application. Although oral and systemic
cyclosporine are well characterized,
cyclosporine
inhalational solution is a new
formulation of
14
cyclosporine. It is seeking a new indication. It
is administered by a new route and it
requests a
new dosage regimen. As I mentioned, we weren't
able to conclude that the differences in
chronic
rejection and survival were due to the
study drug.
For these reasons, we determined it was
important
to have this application discussed in an
open
public forum.
We have asked the help of the
pulmonary
product advisory committee because it is
a standing
committee with expertise in pulmonary
disease. We
have invited experts in statistics and
transplantation to help with the
deliberation, and
we are very much interested in the
committee's
input regarding the adequacy of the
clinical and
statistical evidence whether aerosolized
cyclosporine is safe and effective for
the proposed
indication.
This morning Chiron will
present most of
the background information, starting
with Dr.
Michael Scaife's presentation on the
drug
development program. Then Dr. Jeff Golden will
15
provide an overview of lung
transplantation. Dr.
Sarah Noonberg will discuss the results
of the
efficacy study, followed by Dr. Steven
Dilly's
presentation and Dr. Ron Helms' views.
The FDA presentation will
follow the
Chiron presentations and we will focus
on those
areas that proved challenging during the
course of
the review. Dr. Arturo Hernandez will discuss the
study design, various clinical issues and
outcome
demographic characteristics and
dosing. Dr. Marc
Cavaille-Coll will provide a summary of
the safety
issues and Dr. Jyoti Zalkikar will give
the
statistical presentation.
Then, in the afternoon, we
would like you
to discuss, give advice and vote on a
few
questions. So, as you listen to the presentations
this morning, please keep these
questions in mind
for later discussion. The first question: Is
there sufficient information to make the
determination whether the observed
survival
difference in study ACS001 is due to
study
treatment or some other factors?
16
In your deliberations, we will
ask you to
recall the statistical issues that were
raised by
the application; differences in baseline
donor and
recipient characteristics; whether the
product
demonstrated an effect on various
clinical outcomes
or things such as acute rejection,
bronchiolitis
obliterans syndrome, obliterative
bronchiolitis.
Depending on whether you
conclude that the
answer is yes or no, we have a few
additional
questions, namely, if the answer is yes
we would
like you to talk about the
generalizability or,
more specifically, the labeling issues
that you
would recommend be put into a product
label. If
the answer is no we would like you to
consider what
additional studies you would recommend
be
conducted. In these discussions we would also like
you to give us some suggestions
regarding patient
population, drug dosing regimen, as well as
efficacy and endpoints that could be
included in
such studies.
The next question would be
whether the
safety of the product has been
adequately
17
characterized for its intended use. Again, in this
particular question we would like you to
also
consider the amount of preclinical and
clinical
information that is available in this
application;
infection about the cyclosporine and the
vehicle,
as well as the number of patients who
have been
exposed to the proposed dosage regimen.
If the answer to this question
as well as
the preceding one is yes, then we would
like you to
give us suggestions about what
population the
product should be labeled for; what
information we
should include in labeling on dosing
regimen, dose
preparation and administration, dosing
intervals
and duration of treatment. In addition, if you
could give us guidance on what should be
included
in the labeling regarding the expected
benefit on
acute rejection, BOS, OB and so
forth. If your
answer to the latter question is no,
then we would
like you to give us some advice about
what
preclinical and clinical information
would be
needed.
With that, thank you and I
will turn it
18
back to Dr. Swenson.
DR. SWENSON: Thank you, Dr. Albrecht. We
will proceed now with the sponsor
presentation and
I would like Dr. Michael Scaife to go
ahead and
begin this, and I will let him introduce
his
colleagues and their different
presentations.
Sponsor Presentation
Introduction
DR. SCAIFE: First of all, good morning,
ladies and gentlemen. My name is Michael Scaife.
On behalf of Chiron, I would like to
thank the FDA
as well as members of the advisory panel
for this
opportunity today to present to you on the
safety
and efficacy of an inhalable form of
cyclosporine
that will be referred to throughout the
talk as
either CyIS or the product's trade name,
Pulminiq.
The first point I would like
to make is
that currently in the United States
there are no
drugs or combination of drug therapies
approved for
the treatment of chronic rejection
following lung
transplantation. The prognosis for these patients
is really poor. Despite aggressive care, only 45
19
percent of lung transplant recipients
will be alive
five years following transplantation. This is much
worse than for other solid organ
transplant
recipients. This is an orphan population in the
U.S.
On average less than 1100 lung transplants
are performed each year.
We are here today to talk
about certain
aspects of Pulminiq, a medication that
is an
aerosolized form of cyclosporine
dissolved in an
inert vehicle, propylene glycol. As you all know,
cyclosporine is not a new chemical
entity.
Cyclosporine was approved by the FDA in
1983 and
currently has been approved in most
countries of
the world. It is available in oral, IV and ocular
forms.
In the U.S. it has been approved for the
prophylaxis of allogeneic heart, liver
and kidney
graft rejection, and for the treatment
of
refractory rheumatoid arthritis and
plaque
psoriasis. In Europe cyclosporine is also approved
for use following bone marrow and
pancreatic
transplantation, as well as for a
variety of
immune-modulated pathologies such as
nephrotic
20
syndrome, atopic dermatitis and Bessay
syndrome.
Pulminiq is simply an inhalable
form of
cyclosporine so, in essence, we are here
today to
talk about a well-known drug given by a
new route
of administration to enable delivery to
the
required site of action. As I mentioned, Pulminiq
is a simple formulation consisting of
cyclosporine
dissolved in propylene glycol, with no
other
ingredients. Propylene glycol is also not new to
pharmaceutics. Since the initial inhalation tox
studies of propylene glycol in the '40s
it has been
widely used as a compounding agent for
intravenous
and oral pharmaceuticals, as well as
foods. In
fact, it is currently listed by the FDA
as an
approved inactive ingredient for use in
inhalation
products.
Several preclinical inhalation
studies
have been performed both with Pulminiq
as well as
with the vehicle alone. Specifically, you will see
in
the briefing book that we make mention of two
one-month studies in the rat and the dog
and a
three-month study in the rat. I won't go into the
21
specific details but, as you will see,
doses given
in those animals were in multiples 15,
17 times the
dose that we expected in man. The
histopathological findings again are
detailed in
the book. You will find that aside from some small
punctate findings in the larynx in a few
of the
animals, there were no long-lasting
changes and, in
our view, the results are not
significant.
How did Chiron first become
aware of the
work on inhalable cyclosporine at the University of
Pittsburgh Medical Center, which we will
refer to
from now on as UPMC? Well, in fact, from a sales
rep who was detailing our inhalable
topromycine
product, TOBI, which is used for the
treatment of
pseudomonas infections in cystic
fibrosis patients.
The slide here details the
development
activities at UPMC. The preclinical study started
in '88, followed in '91 by human studies
in lung
transplant patients with chronic
rejection. In '97
UPMC started a randomized, double-blind,
placebo-controlled study of cyclosporine
that ended
in August, 2003. The results of this and other
22
studies will form the discussion of
today's
meeting.
You may ask why did Chiron
want to acquire
the rights to develop this product. Well, we
looked at the results of 15 years of
work at one of
the largest lung transplant centers in
the U.S. We
asked ourselves the same questions,
frankly, and
had the same concerns as anyone would
have had. It
is a single-center study. It was being conducted
by a single lead investigator. Has the study been
conducted appropriately? Are the data robust? Are
the striking effects seen on survival
benefit real?
And, if so, are they due to cyclosporine
or some
other factor or factors?
We did our initial due
diligence of the
data and how it had been collected and
we concluded
that the effect is real. Based upon our
conviction, we acquired the right to file an
NDA
for the product. As you know, the FDA encourages
the filing of applications for products
that
address a clear unmet medical need with
a
demonstrated significant clinical
benefit and an
23
acceptable safety profile. We went to UPMC and we
extensively audited the hospital
records. We went
in and we collected all of these data on
standardized forms, and we analyzed the
data in
every possible manner, as you will hear
later.
In may, 2004 we met with the
FDA. We
posed a very simple question, would the
agency
consider the positive findings from one
clinical
study, conducted by one principal
investigator to
be registerable? The FDA response, and I think Dr.
Albrecht referred to it so she will
forgive me for
paraphrasing I hope, was assuming that
the data are
robust--and I happily stress the word
"robust"--we
encourage you to file. It is rare for us at the
FDA to be provided with significant
survival data
for such a product. Based upon this positive
meeting, Chiron filed an NDA for
Pulminiq in
October, 2004.
I would like to acknowledge
the
collaborative position taken by the FDA
throughout
the NDA process. We have been encouraged to
maintain a dialogue with the reviewers
and it is in
24
this spirit that we are here today.
The finding was accepted by
the FDA and
priority review status was granted in
December,
2004.
Ladies and gentlemen, Chiron is here today
because we believe the data on survival
benefit are
real and clinically relevant, as well as
statistically significant. We will present data
that confirm that CyIS is safe and
efficacious for
the requested indication, which is to
increase the
survival and prevent chronic rejection
in patients
receiving allogeneic lung transplants in
combination with standard chronic
immunosuppressive
therapy.
With that, I would like to
introduce to
the panel and the audience the agenda
for the
Chiron presentation as well as the
speakers, their
background and affiliation. The first speaker is
Dr. Jeff Golden who is professor of
clinical
medicine and surgery at the University
of
California in San Francisco. Dr. Golden is also
the medical director of the lung
transplant program
at UCSF.
25
We have asked Dr. Golden to
speak to you
today for two main reasons, firstly,
because he is
an eminent practicing physician and
scientist who
actually treats and cares for lung
transplant
patients, as well as being an active
researcher
into the mechanisms of acute and chronic
lung
rejection phenomena. Secondly, because he was not
involved in the study and we wanted his
independent
views on the clinical findings. Dr. Golden will
address the current status of lung
transplantation.
He will be followed by Dr.
Sarah Noonberg
who is the clinical leader at Chiron for
this
project.
Dr. Noonberg will present to you the
clinical evidence for the efficacy and
safety of
CyIS.
Dr. Noonberg will be followed
by Dr. Ron
Helms, an emeritus professor of
statistics at the
University of North Carolina. Why did we ask him
to be here today? As statisticians and physicians
have analyzed the data from every
possible angle
and found the positive effect of
Pulminiq on
survival to be clinically as well as
statistically
26
robust, the FDA statisticians expressed some
concerns about our analyses and so we
asked Prof.
Helms to look at our approaches,
assumptions and
methodologies, as well as those of the
FDA
reviewers, and to let us have his candid
opinion.
He will share those views with you
today.
The final presentation by
Chiron will be
given by Dr. Stephen Dilly. He is the chief
medical officer for Chiron
BioPharmaceuticals. He
will review the case for approval of
Pulminiq
including a discussion of our proposed
postapproval
study.
We will then hand over the meeting to the
Q&A session that will be moderated
by myself.
Finally, we have a list of
additional
experts, both internal and
external. I would like
to make the special point that we have
the pleasure
of having Dr. Trulock here who is a
world renowned
expert on lung transplantation and,
again, as you
know, you are free to ask any of our
experts for
additional information. With that, I would like to
hand over to Dr. Golden. Thank you very much.
Current State of Lung
Transplantation
27
DR. GOLDEN: Thanks.
I am extremely
delighted to be here as somebody who
takes care of
patients after transplantation. I am here really
to give an overview of the current state
of lung
transplant.
Just a brief statement about
myself, about
15 years ago I helped start a lung
transplant
program at the University of California
in San
Francisco. In the past few years we have been
doing about 30 transplants a year, and
this year we
are on a pace for 40 transplants. Just to give you
a perspective, this puts us in about the
top 10
percent in terms of volume of annual
transplants in
the world.
About two years ago I was
asked to visit
Chiron and give a review of lung
transplant. At
that time I was first shown some data
from the
University of Pittsburgh on aerosolized
cyclosporine. Subsequently, as some of you may
know, I did attend the first FDA meeting
in 2004
where I similarly presented an overview
of lung
transplant. Well, I am back and actually nothing
28
has changed.
I would like to summarize on
the next
slide the main points in terms of where
we are in
lung transplant. First, the long-term survival of
lung transplant is 50 percent by five
years. This
is a poor survival. Second, bronchiolitis
obliterans, or chronic rejection, is the
primary
cause of this poor survival. Third, the future of
lung transplant really demands that we
learn how to
prevent bronchiolitis obliterans.
By way of history, before
cyclosporine
there had been approximately 40 lung
transplants in
the world. Looking at their survival, the median
survival was somewhere around 10 days. One patient
lived 10 months. After the introduction of
cyclosporine there were one-year
survivals, such
that eventually there was 75 percent
one-year
survival in lung transplant. With this large
improvement compared to the
pre-cyclosporine era,
the interest in lung transplant really
took off.
As you can see from this
slide, early on
in 1985 there were about a dozen
transplants and as
29
of 2003 there are somewhere around 1700
transplants
in the world, about 1100 in the United
States.
These are done for various recipient
categories you
see listed here. Approximately half are for
emphysema or alpha-1 antitrypsin
deficiency, cystic
fibrosis, and another large area is
idiopathic
pulmonary fibrosis. Although in this registry
analysis it is 17 percent, at UCSF 60
percent of
our lung transplant patients have
idiopathic
pulmonary fibrosis, a disease for which
there is no
therapy and a disease that has a
five-year
survival, somewhat similar to lung
cancer.
However, despite this
increased one-year
survival and this tremendous increase in
the number
of transplants done around the world, we
are,
unfortunately, still stuck at a low 50
percent
survival of around 4.5 to 5 years. Although one
might say emphysema has a slightly
better outlook
at 4 and 5 years than idiopathic
pulmonary
fibrosis, in general lung transplant
survival is
about 50 percent at 4.5 to 5 years.
To give you some perspective,
if you look
30
at kidney transplant at that interval of
4.5 years
after transplant there is 90 percent
survival. And
if you look at heart and liver
transplant it is
about 75 percent survival. Not only do we have
this 50 percent survival at 4 to 5
years, but this
has not changed in almost 20 years. We have
plateau'd in terms of poor survival in
that period
of time.
As I say, the problem
responsible for this
poor mortality clearly is bronchiolitis
obliterans.
In this histology section, with an
artery here, the
obliterative lesion that is established
as a
fibroplastic plug diminishes the airway
diameter
such that, instead of being this size,
it is
reduced and constricted down to this
tiny lumen
here secondary to this
fibroproliferative process
of the lesion of bronchiolitis
obliterans.
Bronchiolitis obliterans or
chronic
rejection is diagnosed in two ways,
histologically
through a transbronchial biopsy or
clinically. The
problem with the histologic diagnosis of
a
transbronchial biopsy is that it is a
specific
31
finding but it is not very sensitive.
Transbronchial biopsy is simply not
sensitive
sufficiently to diagnose this chronic
airway
process.
Therefore, over the years we have
developed a clinical diagnosis in the
absence of a
histologic finding on the transbronchial
biopsy
such that we look at specific decrease
in air flow
when there is no alternative cause, and
we label
this bronchiolitis obliterans syndrome.
It is important to stress that
obliterative bronchiolitis and
bronchiolitis
obliterans syndrome, or BOS, are really
histologic
and clinical manifestations of the same
airway
process.
Patients develop progressive shortness of
breath with this graft failure,
progressive airflow
obstruction and recurrent pulmonary
infections.
Regrettably, once this chronic rejection
develops
the airway damage is progressive and
irreversible
and patients die of graft failure and
related
infections.
The registry for transplant
would say that
somewhere around 5 years the percent of
patients
32
dying from different etiologies would be
bronchiolitis obliterans about 30
percent, but
actually you cannot separate this from
infections
which are always present in the setting
of this
airway damage. Furthermore, in this registry
setting where they describe organ
failure, that is
obviously bronchiolitis obliterans. So, when you
add up these categories of bronchiolitis
obliterans
organ failure and related airway
infections,
including pseudomonas, aspergillus,
etc., let me
simply state that bronchiolitis
obliterans
complications relate to the vast
majority of deaths
at 4.5 to 5 years after lung
transplantation.
No matter what we have done in
the last 18
years, we have not prevented this
development of
chronic rejection, this airway process,
whether we
give tacrolimus, different combinations
of
cyclosporine, micofenolate,
azathioprine,
prednisone and, in fact, I could put
rapomyacin up
there and various other lytic therapies
and various
approaches to prednisone pulses for
acute
rejection, etc. Despite all this systemic
33
immunosuppression, we really have not changed
the
incidence of chronic airway rejection
closely
related, unfortunately, to the poor
survival at 4.5
to 5 years.
We now appreciate that there
are
non-immune factors that relate to airway
damage, be
these infection or reflux disease. These
non-alloimmune factors clearly relate to
immune
activation. In fact, I believe we are now
understanding that when we see chronic
airway
rejection and we increase systemic
immunosuppression we actually are
helping to
promote such non-alloimmune factors,
especially
infections which cause further airway
immune
activation and actually make the process
worse.
We have always known that there are
alloimmune factors such as acute
rejection that
relate to damage of the organ. We are now
appreciating these non-alloimmune
factors, again,
be it early airway damage with transplant,
various
infections, reflux disease which is a
very new
concept in terms of what injures the
airway--that
34
these non-alloimmune stimuli, in consort
with
alloimmune rejection, together damage
the graft
leading to progressive, additive
epithelial injury,
inflammation and fibroblastic repair
culminating in
the picture I showed you of
bronchiolitis
obliterans.
One newer concept in terms of
immune
factors is called lymphocytic
bronchitis/bronchiolitis. One might call it airway
rejection. This histology reveals lymphocytic
bronchitis/bronchiolitis and airway
disease wherein
you have submucosal lymphocytes working
their way
into the mucosa. Let me point out that lymphocytic
bronchitis/bronchiolitis has been highly
related to
the subsequent development of the more
fibrotic
bronchiolitis obliterans. This concept of an
airway inflammation based on immune
reaction in the
airway, lymphocytic bronchitis, was not
on the
radar screen 15 years ago when the
concept of
inhaled cyclosporine was conceived.
We have always known that
acute rejection
is one of the factors that relates to
the
35
subsequent development of bronchiolitis
obliterans.
However, I want to separate out the
airway process
from acute rejection which is a
perivascular
process diagnosed by transbronchial
biopsy. I
should emphasize that a transbronchial
biopsy has
variable adequacy for obtaining small airway
samples to diagnose whether it is
bronchiolitis
obliterans or the early airway
inflammation of
lymphocytic bronchitis.
If I was designing a study
today of any
inhaled immunosuppressant therapy I
would try to
learn more about the biology of the
airways. We,
at UCSF, and some other institutions
have been
doing endobronchial biopsy. This is not standard
but we are learning a lot more about the
airway
biology in terms of lung
transplantation.
On my last slide I want to
just emphasize
that although I might expect systemic
immunosuppression to clear up a
perivascular
process, I am suggesting that
bronchiolitis
obliterans, chronic rejection, is an
airway process
and it makes eminent sense to employ
inhaled
36
cyclosporine to treat the
epithelium. It is clear
now that the epithelium is key to the
development
of bronchiolitis obliterans. Bronchiolitis
obliterans is an airway disease.
Just to finish, my colleagues
in the lung
transplant world are very excited about
the
potential benefit of inhaled
cyclosporine. As I
say, the epithelium is key and it makes
eminent
sense to develop a system of local
immune
suppression to the airway and the
mucosa. Frankly,
given the poor survival of our
transplant
recipients which, as I already
mentioned, has not
changed in almost 20 years, I personally
feel that
inhaled cyclosporine fulfills an unmet
need.
I questioned whether I was
going to say
the following but I think I will. On a personal
note, for people like myself who take
care of these
patients, who see them terribly short of
breath in
various diagnostic categories who go on to
have a
lung transplant and then regain a normal
life,
including family life, going back to
work--to all
of a sudden see these patients once
again slowly
37
develop progressive airway rejection,
chronic
rejection and shortness of breath is
extremely
disheartening to the patients, to say
the least,
their family and, frankly, for their
physicians.
Thank you for your attention.
Clinical Evidence of Efficacy
and Safety
DR. NOONBERG: Good morning.
My name is
Sarah Noonberg and I am the clinical
leader for the
inhaled cyclosporine project. Over the next 45
minutes I will be reviewing the clinical
data
supporting the use of inhaled
cyclosporine in lung
transplant recipients.
I will begin with a brief
discussion of
early preclinical and open-label
clinical trials of
inhaled cyclosporine at UPMC. These trials
generated a lot of interest in inhaled
cyclosporine
and really set the stage for the pivotal
randomized, double-blind,
placebo-controlled trial
which we, at Chiron, refer to as ACS001.
I will then describe the study
design and
baseline characteristics of patients in
ACS001
before moving into a discussion of
efficacy,
38
focusing primarily on the endpoints of
survival and
chronic rejection. I will then switch gears and
summarize the safety data that has been
generated
for inhaled cyclosporine from a safety
database of
102 patients. Although the favorable safety
profile is clearly an important aspect
of the drug,
I am going to be spending much less time
reviewing
safety listings as this is an area of
general
agreement with the FDA.
Finally, as with all studies,
there are
limitations to ACS001 both with respect
to study
design, as well as choice of the primary
endpoint.
I
am going to end this presentation with a
discussion of some of those limitations
and how we
view them in light of the clear
strengths of the
study.
As Dr. Golden has described,
the
introduction of cyclosporine as an
immunosuppressant truly revolutionized
lung
transplantation and allowed for the
possibility of
long-term survival. Within a few years of FDA
approval investigators at UPMC began to
develop an
39
aerosolized formulation, and within five
years they
initiated preclinical trials.
In the first set of
experiments
non-transplanted dogs were given a
single dose of
inhaled cyclosporine. The dose was well tolerated
and revealed that pulmonary
concentrations were
10- to 100-fold higher than
concentrations in other
tissues.
In addition, there was no change in lung
function and no histologic
abnormalities.
In a canine lung transplant
model dogs
were given single agent
immunosuppression with
inhaled cyclosporine and investigators
reported a
dose-dependent decrease in the frequency
and
severity of allograft rejection.
In a rat transplant model rats
were given
an identical dose of either inhaled
cyclosporine or
intramuscular cyclosporine. Inhaled cyclosporine
was found to be at least as effective as
intramuscular cyclosporine in causing a
dose-dependent decrease in
proinflammatory cytokine
production, as well as a decrease in
allograft
rejection but with far lower systemic
exposure to
40
cyclosporine.
These encouraging preclinical
results led
to the development of a series of
open-label
non-comparative trials with inhaled
cyclosporine at
UPMC.
These trials enrolled two different groups
of patients, both with established
complications of
lung transplantation. In the first set of
protocols lung transplant recipients
with
documented chronic rejection were given
inhaled
cyclosporine in addition to their
standard
immunosuppressive regimen. Investigators reported
improvement in rejection histology and
stabilization of pulmonary function
relative to
pre-enrollment data. But, more importantly, these
patients had improved survival both
compared to
contemporary UPMC unenrolled controls as
well as
controls from a historical lung
transplant
registry.
In the next set of protocols
patients with
refractory acute rejection, defined as
acute
rejection that failed to respond to
immunosuppressive intensification--this
represents
41
a step earlier in the disease process as
acute
rejection--as a risk factor for the
subsequent
development of chronic rejection and was
the
logical next population to study. When these
patients were given inhaled
cyclosporine, again, in
addition to their standard
immunosuppressive
regimen, investigators reported an
improvement in
rejection histology, a reduction in
proinflammatory
cytokine production, and a
dose-dependent increase
in pulmonary function, all relative to
re-enrollment data. Once again, these patients had
improved survival compared to
contemporary UPMC
unenrolled controls.
Despite the non-comparative
nature of
these trials and their inherent
limitations, they
made quite an impact in the transplant
community,
and have led to unregulated compounding
of inhaled
cyclosporine by a number of U.S.
transplant
centers.
In a survey of 2002, published in Chest,
of transplant practices 10 percent of
U.S.
transplant centers already used inhaled
cyclosporine. They compound it in their pharmacies
42
and they give it to patients with
progressive
chronic rejection.
These open-label trials were
clearly
provocative but their interpretation is
limited by
the lack of an adequate control
group. However,
they laid the framework for the very
first and one
of the only randomized, double-blind,
placebo-controlled trials in the lung
transplant
population. Unlike the previous protocols that
enrolled patients with established
complications of
lung transplantation, this trial was
designed to
test the efficacy of inhaled
cyclosporine in
preventing rejection and improving
outcomes when
given prophylactically to patients
shortly after
their single or double lung transplant
procedure.
The trial had two phases. In a pilot
phase, the first phase, 10 patients were
given
open-label inhaled cyclosporine and were
followed
prospectively. They formed a cohort designed to
test the safety and tolerability of the
drug in
this patient population. In the second phase, the
randomized phase, 58 patients were
randomized and
43
56 were randomized and treated with
either inhaled
cyclosporine or placebo, which in this
case was
inhaled propylene glycol, the vehicle
used to
create the inhalation solution. The primary
endpoint of the study was rate of acute
rejection,
and secondary, prospectively defined
endpoints of
survival, rate of chronic rejection and
pulmonary
function.
The criteria for enrollment
into ACS001
were fairly straightforward. To be included, you
had to be a recipient of a single or
double lung
transplant and be 18 years of age or older.
Exclusion criteria included the presence
of active
fungal or bacterial pneumonia or
anastomotic
infections prior to the initiation of
appropriate
antimicrobial therapy. Patients with bronchial
stenosis greater than 80 percent had to
be treated
with standard techniques prior to
enrollment.
Patients who failed to wean from
mechanical
ventilation and women of childbearing
potential
unwilling to use birth control were also
excluded.
It is important to note that all
patients met study
44
inclusion and exclusion criteria.
All patients in ACS001 were
treated with
standard-of-care immunosuppressive
therapy
following transplantation, and all were
randomized
and enrolled within the first 7-42 days
following
their transplant surgery. A total of 26 patients
were treated with inhaled cyclosporine
and 30 were
treated with placebo. All patients underwent an
initial 10-day dose escalation period
where they
were initiated on low dose inhaled
cyclosporine at
100 mg, and that dose or equivalent
volume of
placebo was gradually increased to a
maximally
tolerated dose up to a
protocol-specified maximum
of 300 mg. The dose or equivalent volume that they
reached on day 10 was the dose that they
continued
3 times a week for a period of 2 years.
After completion of dosing
patients
continued to be followed for study
endpoints up to
the study end date of August 21,
2003. This
corresponded to 2 years after the last
patient was
enrolled and, therefore, could complete
their
2-year period of dosing. Therefore, the total
45
length of follow-up per patient depended
on the
timing of enrollment and ranged from 24
months for
the last patient enrolled up through 56
months for
the first patient enrolled.
ACS001 was a randomized trial,
and the
randomization scheme was developed by
the
Department of Statistics at the
University of
Pittsburgh. The randomization was stratified by
CMV mismatch, defined as donor
positive/recipient
negative, versus all other
combinations. This was
chosen because international registry
data has
demonstrated that patients with CMV
mismatch have a
32 percent increased relative risk of
death in the
first year compared with other
combinations, with a
p value of less than 0.0001. Therefore, the
assertion that the randomization was not
stratified
by any variables known to affect outcome
is
incorrect. The randomization was also stratified
by enrollment period and distinguishes
patients who
generally had a less complicated
postoperative
course, were stable and met exclusion
criteria by
7-21 days versus those that had a
relatively more
46
complicated postoperative course and met
exclusion
criteria and were stable between 22 and
42 days
after the surgery. In line with ICH guidelines, it
is impractical and often
counterproductive to
stratify by more than 2 factors in a
study of this
size.
This slide illustrates the
baseline
characteristics of patients enrolled in
ACS001.
Overall, the two groups were well
matched with
respect to the majority of relevant
baseline
demographic characteristics. Donors were similarly
well matched for clinically relevant
variables.
However, as can be expected from any
randomized
study, there were a few important
imbalances. The
two variables where clinically relevant
imbalances
existed were with respect to primary
diagnosis and
transplant type.
As Dr. Golden has demonstrated,
the
primary diagnosis leading to
transplantation can
have an important impact on
survival. Patients
with COPD have traditionally been
associated with
better outcomes, especially within the
first year,
47
and this is statistically
significant. Nearly
twice as many placebo patients had this
more
favorable diagnosis. In addition, patients with
idiopathic pulmonary fibrosis or IPF
have
historically had among the worst
survival, both
short-term and long-term, and this is
statistically
significant at one year and at five
years, and
there were far more patients with IPF in
the
inhaled cyclosporine group compared to
placebo.
Both of these factors together could
potentially
bias results for better outcomes in the
placebo
group.
By contrast, double lung
transplant
recipients have historically had
marginally
improved survival compared to single
lung
transplant recipients in the first
several years,
and this difference becomes increasingly
pronounced
with time but is not statistically
significant at
one year or at five years, the time
period of
interest for ACS001. However, there were more
double lung transplant recipients in the
inhaled
cyclosporine group and this could
potentially bias
48
results towards better outcomes in the
inhaled
cyclosporine group. Therefore, although imbalances
exist, they are split between groups and
would not
be expected to strongly influence
results in one
direction or the other.
The protocol specified that
patients were
to continue study drug for a period of
two years.
However, due to the nature of the
patient
population with its high mortality rate,
frequent
complications and frequent
hospitalizations, not
all patients could complete the two-year
period of
dosing and this is not surprising. Roughly
two-thirds completed at least one year of
therapy
and roughly half completed the full two
years of
therapy.
As the protocol specified that dosing
should be held temporarily in the
presence of an
infection not responding to treatment,
not all
patients had each and every one of their
scheduled
doses.
However, this just reflects the protocol
rather than any lack of compliance.
The median duration of dosing
was
comparable among the two groups. Of the patients
49
that did prematurely discontinue dosing,
the
primary reasons were adverse events in
the placebo
group and withdrawal of consent in the
inhaled
cyclosporine group. Of the six who withdrew
consent, two were due to early
tolerability
problems; two were primarily due to
unrelated
medical problems; and one was due
primarily to an
unrelated social problem and for one the
reason was
unknown.
Although no patients were lost
to
follow-up, five patients, three in the
inhaled
cyclosporine group and one in the
placebo group,
were taken off the study, the randomized
trial, and
crossed over into an open-label rescue
protocol of
inhaled cyclosporine. Their data was censored at
the time of crossover and the treatment
groups
remained blinded. In both groups there were
patients that were withdrawn due to
protocol
deviations and violations that largely
included
medical non-compliance and smoking.
This slide summarizes the
important
efficacy and safety results from study
ACS001.
50
Treatment with inhaled cyclosporine led
to
significantly improved survival and
chronic
rejection-free survival compared to
placebo but did
not affect the rate of acute
rejection. Treatment
with inhaled cyclosporine was not
associated with
increased risk of nephrotoxicity,
infections,
malignancies or any systemic toxicities
known to
occur when cyclosporine is given orally
or
intravenously. However, similar to other inhaled
drugs, inhaled cyclosporine was
associated with
mild to moderate respiratory tract
irritation and
bronchospasm.
I will first discuss the
effect of inhaled
cyclosporine on survival. Using an unadjusted
analysis, inhaled cyclosporine was
associated with
a significant survival advantage
compared to
placebo, with a relative risk of death of
0.213 and
a p value of 0.007. This corresponds to a 79
percent decreased risk of death in
patients treated
with inhaled cyclosporine compared to
placebo.
This slide is the Kaplan-Meier plot of
survival
duration from the time of
transplantation to the
51
study end date, and is the primary
reason that we
are all here today.
During the period of the study
there were
3 deaths in the inhaled cyclosporine
group compared
to 14 deaths in the placebo group. The results are
not only highly statistically
significant but also
clinically very important. This is the first time
a cohort of lung transplant recipients
has had
survival comparable to recipients of
other solid
organ transplants and marks a major
advance in
outcomes for this patient population.
The importance of an
unadjusted analysis
rests on its robustness and how well it
compares to
analyses that control for other baseline
characteristics that might affect
outcome.
Therefore, we performed univariate
analyses
adjusting for potential risk factors
that might
affect survival, and found that the
relative risk
of death and the p values were
remarkably
consistent.
This graph illustrates the
relative risk
of death and 95 confidence intervals
when the
52
survival data is adjusted by a number of
different
factors that have been documented in the
literature
to potentially affect outcome. We also include two
factors suggested by the FDA, ICU time
after
transplantation and the use of donors
who at some
point during their hospitalization prior
to
harvesting were treated with an
inotrope. Neither
of these two factors is supported by the
literature
or registry data as having an impact on
survival.
For the case of donor inotrope use, it
is not
considered in guidelines for optimal
donors or
marginal donors. However, the key message is that
regardless of the baseline
characteristic none of
these factors appreciably impacts the
relative risk
of death and lends strong support to the
validity
of the unadjusted analysis, and this is
what is
meant by a robust endpoint.
In order to further test the
robustness of
the survival endpoint, we performed
multivariate
analyses which adjust for clinically
relevant
baseline characteristics
simultaneously. As not
all characteristics can ever be
simultaneously
53
input into a single statistical model,
the job of
the clinician is to decide which of
these are the
most clinically relevant.
In order to determine the most
clinically
relevant factors we searched through the
literature
to determine those that had been
documented to be
short-term or long-term prognostic
factors. We
then reviewed registry data to determine
the level
of significance and, finally, we
discussed these
factors with transplant physicians who
care for
these patients. The general agreement was that the
most clinically relevant factors were
transplant
type, CMV mismatch, primary diagnosis,
early acute
rejection--all shown in green. We also include in
our model the variable of enrollment
period as this
was a randomization stratification
variable and it
is in accordance with ICH guidelines.
This slide also illustrates
the relative
distribution of 16 different baseline
characteristics that have been
documented in the
literature to potentially affect
short-term or
long-term outcome. As is evident, the majority are
54
balanced or, if anything, would favor
better
outcomes in the placebo group.
This slide illustrates the
results of the
multivariate analyses when these factors
are
successively added into a Cox
proportional hazards
model.
The key point is the consistency of the
treatment effect. The addition of the five most
clinically relevant factors into this
study does
not have any appreciable impact on the
relative
risk of death or the p values, and
provides even
further support for the robustness of
the survival
endpoint.
Robustness was further
evaluated by
performing a number of sensitivity
analyses around
the survival endpoint. When we did so, we found
that the relative risk of death remained
consistent. The top row illustrates the unadjusted
analysis on the full data set. When we include
patients who were randomized and treated
the
results are essentially unchanged. When we look at
survival relative to first dose of study
drug
rather than time of transplantation,
again the
55
results are essentially unchanged. When we exclude
three placebo patients who had early
mortality and
died within the first three months--when
we just
take them out of the analysis and we
only analyze
the remaining 27, it remains
statistically
significant. When we take out 14 patients who did
not receive at least 80 percent of the
protocol
maximum dosing adjusted for death, we
lose 25
percent of the sample size but still
maintain
statistical significance and the
relative risk of
death is barely altered.
The FDA has raised concern
about the
effects of early pneumonia. So, if we remove from
analysis 15 patients who had an episode
of
pneumonia within one month of initiation
of study
drug we have lost greater than 25
percent of the
patient population and, therefore,
expect that the
p value is going to increase but the key
point is
that the relative risk of death, the
treatment
effect, is barely changed.
Questions have also been
raised about the
effects of ICU time after transplantation. If five
56
patients who were in the ICU greater
than 14 days
were removed from analysis, the results
are
statistically significant and in favor
of the
inhaled cyclosporine group. Therefore, we have
looked at the survival data from a
number of
different angles and found the survival
data to be
robust.
To assess the duration of the
survival
benefit we collected additional survival
data 10
months after the study ended, and we
found that the
survival benefit persisted. At that point there
were 5 deaths in the inhaled
cyclosporine group
compared to 15 deaths in the placebo
group, with a
p value of 0.017.
This post-study follow-up is
important and
it is useful and supportive data. However, it has
its limitations. The first is that the study had
ended and it ended almost a year
earlier. The data
was analyzed and patients were
unblinded; treatment
groups were known. In addition, except for those
patients who had crossed over into an
open-label
protocol, all patients were off study
drug for a
57
substantial period of time, ranging
anywhere from
10 months to a maximum of 3.5
years. When you
consider that the median time to
diagnosis of
chronic rejection is 16-20 months, it is
going to
confound the results. Also, there were placebo
patients that had crossed over and were
now
receiving inhaled cyclosporine so the
net effect,
as expected, is that it is going to
trend toward
the null.
This is what the FDA refers to
as the
five-year data and believes that it is
the most
appropriate time point to analyze the
survival
data, but for the reasons that I have
just
described we disagree and we believe
that the data
is best analyzed at the prospectively
defined study
end date.
In order to verify that the
placebo
population was representative of what would be
expected in a larger U.S. transplant
population,
the placebo survival curve was compared
with data
from the United Network for Organ
Sharing, or UNOS,
that maintains a large transplant
registry.
58
Placebo patients were matched with UNOS
controls
who were transplanted during the same
period of
enrollment as ACS001, and they were
matched by the
variables on the slide. Matching also excluded
patients who died before they could have
possibly
enrolled into ACS001.
This slide illustrates the
results and
shows that both early mortality and late
mortality
in the placebo group are extremely
consistent with
what is expected in a larger multicenter
patient
population. Roughly 50 percent survival at 4.5
years is exactly what has been
documented in the
literature for years. Therefore, any analyses that
exclude early deaths or late deaths or
deaths due
to particular causes have to be viewed
with caution
as they would no longer lead to a
placebo group
whose survival is representative. By comparison,
when the ACS001 inhaled cyclosporine
group is
compared to the UNOS controls the
relative risk of
death of 0.252 is very comparable to
what was seen
in ACS001 where the relative risk of
death was
0.213.
59
This is a busy slide but it
makes a very
important point and brings us into our
next topic,
namely, the primary reason for the
improved
survival in patients treated with
inhaled
cyclosporine is that inhaled
cyclosporine prevented
chronic rejection. This slide illustrates the
timing and cause of death for both
groups. As
expected, early deaths were
predominantly due to
infectious causes. However, subsequently nearly
all deaths are associated with chronic
rejection.
Of the five deaths that the agency calls
attention
to in the mid portion of the graph as
driving the
statistical significance, four out of
the five had
chronic rejection. By contrast, in the inhaled
cyclosporine group the curve becomes
flat and late
mortality is not occurring.
One question that has been
raised is why
is the survival difference statistically
different
at two years when all patients would
have completed
their study drug. The reason, as evident from this
graph, is that chronic rejection is the
predominant
cause of death in the first year so you
wouldn't
60
expect to see early large separation of
the two
curves.
However, after a year it is the major
contributor, as Dr. Golden has
demonstrated, to
mortality.
To review, chronic rejection
is an
umbrella term for patients with
histologic evidence
of bronchiolitis obliterans, or OB,
documented by
transbronchial biopsy. It is also representative
of patients with clinical evidence of
bronchiolitis
obliterans syndrome, or BOS, using a
sustained and
unexplained decline in FEV1 as a surrogate
marker.
It is not uncommon for patients to have
bronchiolitis obliterans but, due to the
progressive nature, they haven't met
clinical
criteria for BOS. It is also not uncommon for
patients to have BOS but, due to the
insensitive
nature of transbronchial biopsy in
making the
diagnosis they don't have OB. So, these two
groups, patients with OB and patients
with BOS, are
overlapping but they all represent
patients with
chronic rejection. So, looking at each group
individually may be informative but it
has to be
61
viewed as a subset analysis. Consistent with
direct delivery to the airway
epithelium, the site
of chronic rejection, treatment with
inhaled
cyclosporine led to a 72 percent
decrease in the
risk of chronic rejection or death. As you will
see, when we performed the same
univariate and
multivariate analyses, the results are
even more
robust.
This slide illustrates the
Kaplan-Meier
estimate of chronic rejection-free
survival and
uses a composite endpoint of first
diagnosis of OB,
first diagnosis of BOS or death. There are two
important points here. One is that there is
general agreement with the FDA that the
rate of
biopsy and the rate of pulmonary
function testing
is comparable between the two groups so
that the
difference isn't driven by increased
testing in one
group or the other.
The second is that the use of
a composite
endpoint of chronic rejection and death
implies
that patients who die and, therefore,
can't go on
to be diagnosed with chronic rejection
are counted
62
as
events rather than censored in the statistical
analysis. To censor deaths in a statistical
analysis of chronic rejection would
require the
assumption that there is no relationship
between
chronic rejection and death, an
assumption that we
know to be invalid.
The agency issued guidelines
in April of
2005 endorsing a progression-free
survival analysis
for similar oncology endpoints to avoid
a type of
bias known as informative
censoring. As with the
survival endpoint, we found a remarkable
consistency of the chronic
rejection-free survival
endpoint when we performed a series of
univariate
analyses. None of these baseline characteristics
had any appreciable impact on the
treatment effect
or its significance, which speaks to the
robustness
of this endpoint as well.
This slide illustrates the
result of
multivariate analyses on the chronic
rejection-free
survival endpoint. Once again, the addition of the
5 most clinically relevant factors in
this
study--adding them into a Cox
proportional hazards
63
model has essentially no real impact on
the
treatment effect of the confidence
intervals and
the p values remain highly statistically
significant.
Valid questions have been
raised about
whether the survival benefit is so
strong that any
composite endpoint that includes
survival would be
statistically significant. Therefore, for
exploratory reasons we performed an
analysis of
chronic rejection with death
censored. This
clearly biases results against the
inhaled
cyclosporine group due to the larger
number of
deaths in the placebo group. As mentioned, this is
referred to as informative
censoring. However that
said, when we performed the analysis the
results
were still statistically significant and
in favor
of the inhaled cyclosporine group. Chronic
rejection occurred in 50 percent of
placebo
patients and 27 percent of inhaled
cyclosporine
patients.
This slide illustrates the
Kaplan-Meier
estimate of time to chronic rejection
with deaths
64
censored and clearly illustrates a
statistically
significant effect on chronic rejection
independent
of death despite the large bias inherent
in the
analysis. This analysis is important because it
leads to the conclusion that treatment with
inhaled
cyclosporine prevents chronic rejection,
the
leading cause of late mortality in lung
transplant
patients.
However, the primary endpoint
of the study
was not survival or chronic rejection but rate of
acute rejection and this endpoint was
not met.
Approximately 70 percent of patients in
both groups
had at least 1 episode of documented
grade 2 or
higher acute rejection prior to study
termination.
After the start of dosing rates were
comparable
between the 2 groups, with a p value of
0.73.
Dr. Golden has explained the
paradigm
shift that has occurred in the transplant
community
in terms of how acute and chronic
rejection are now
understood. Acute rejection is primarily a
vascular process so an immunosuppressant
with low
vascular exposure would not be expected
to have a
65
significant effect, and that is what we
are seeing
in ACS001. By contrast, chronic rejection is an
airway process. It is mediated in the airway
epithelium so an immunosuppressant
delivered
directly to the airway epithelium would
be expected
to have an effect, and that too is what
we are
seeing in ACS001.
Now I am going to switch gears
and briefly
discuss safety. This slide illustrates the
relative systemic exposure to
cyclosporine when
given by an inhalation route compared to
an oral
route.
A 300 mg dose of inhaled cyclosporine has
been demonstrated to lead to a mean peak
blood
concentration of 206 ng/mL, roughly
11-14 percent
of what you would expect in an oral
dose. the
levels at 24 hours are barely detectable
by
standard assays, and these numbers are
reflected in
the mean AUC, or area under the curve,
which
suggests a roughly 8-fold lower systemic
exposure
to cyclosporine when it is given by an
inhaled
route compared to an oral route. This low systemic
exposure explains why no additional
systemic
66
toxicities were seen in the inhaled
cyclosporine
group compared to placebo.
Data to support the safety of
inhaled
cyclosporine and propylene glycol come
from
multiple sources, and this is outlined
in much
further detail in the briefing
book. The first are
preclinical toxicology studies in dogs
and rats,
performed both by Chiron as well as
referenced in
the literature. These studies show that no
unexpected toxicities were seen when
animals were
treated at many-fold higher doses than
what would
be used clinically.
The next source is the randomized,
placebo-controlled ACS001 trial where
safety data
from 30 placebo patients were compared
with safety
data from 36 inhaled cyclosporine
patients, the 26
randomized and the 10 placebo.
The next source is ACS002,
which was a
retrospective safety analysis of 70
patients
enrolled in 1/7 different open-label
protocols of
inhaled cyclosporine in patients with
refractory
acute and chronic rejection. The ISS, or
67
integrated safety summary, is a
combination of all
patients treated with inhaled
cyclosporine in
either ACS001 or ACS002 and represents
102 unique
patients in our safety database.
To summarize our clinical
safety data,
review of the adverse event listings in
ACS001
revealed that inhaled cyclosporine was
safe. There
was no increased risk of nephrotoxicity,
neurotoxicity, infections, malignancies
or any
other toxicities that occur with oral or
intravenous cyclosporine. In addition, there were
no new or unexpected systemic toxicities.
So, the key point is that
treatment with
inhaled cyclosporine led to a 79 percent
decreased
risk of death compared to placebo, with
no systemic
toxicity. However, inhaled cyclosporine was
associated with respiratory tract
irritation and
bronchospasm and this will be discussed
in the next
slide.
Review of adverse event data in ACS002 and
the ISS confirmed the safety findings of
ACS001,
and no new safety signals were seen
after review of
the serious adverse event data.
68
After review of the ACS001
adverse event
listings and case report forms, it became
clear
that there were two distinct but
interrelated
safety signals that appeared to be a
direct result
of inhaled cyclosporine. The first was
bronchospasm manifested primarily by
cough,
exacerbated dyspnea and wheezing. The second was
respiratory tract irritation manifested
primarily
by pharyngitis but also laryngitis and
non-cardiac
chest pain. In general, these events were mild to
moderate. They occurred early in the patient's
treatment course and diminished with
time, and once
they resolved it was rare for them to
recur. But,
most importantly, there was no
progression to more
serious respiratory complications such
as acute
respiratory failure or ARDS. The adverse event of
lung consolidation was noted in higher
frequency in
the inhaled cyclosporine group but the
clinical
relevance of this finding is unclear as
underlying
causes such as pneumonia, lung mass,
atelectases or
other underlying causes were comparable
between the
2 groups.
69
Having reviewed the most important
clinical results for inhaled
cyclosporine, it is
appropriate to take a step back and take
a look at
some of the outstanding issues
surrounding the
data.
Study ACS001 was conducted at a single
center, and this was discussed with the
FDA well
before Chiron decided to move ahead and
file the
NDA.
However, it is important to note that no
other transplant studies or registry
analyses have
ever shown a survival benefit comparable
to what
was seen in the inhaled cyclosporine
group of
ACS001.
We also looked at the placebo
group and
found that survival was comparable to a
multicenter
matched database. Single-center trials are not
ideal.
However, they do have one important
advantage. Because confounding due to differences
in patient care is minimized,
single-center trials
are actually better at determining a
treatment
effect than multicenter trials of the
same size.
Finally, Chiron has committed to a
multicenter
postapproval trial to further study the
efficacy
70
and safety of inhaled cyclosporine.
The sample size of N equals 56
for
efficacy and N equals 102 for safety is
small.
However, the lung transplant population
is
exceedingly small, with 1100 lung
transplants
performed in the United States each
year. Despite
the small sample size, the survival and
chronic
rejection data are highly statistically
significant
so the sample size was sufficient to
test the
hypothesis that inhaled cyclosporine
improves
survival and chronic rejection-free
survival.
Cyclosporine and propylene
glycol are
well-known and well-characterized, and
the safety
profile of inhales cyclosporine is
extremely
favorable, especially in light of the
survival
benefit.
Again, Chiron has committed to creating a
larger efficacy and safety database
through a
postapproval trial.
The randomization code was
susceptible to
unblinding and CRF assembly was
retrospective. The
randomization code used a patient
subject number
followed by an A, B, C or D designation,
with A and
71
D referring to placebo patients, B and C
referring
to inhaled cyclosporine patients, and it
is
possible that the study could have
become unblinded
due to the simple nature of this
designation.
However, there are several factors that
make this
very unlikely. First is that the principal
investigator was never exposed to the
subject
numbers.
Second, the investigator removed 3
inhaled cyclosporine patients from the
inhaled
cyclosporine arm only to cross over into
an inhaled
cyclosporine open-label rescue
protocol. In
addition, the pathologist reading the
transbronchial biopsies and making the
determination of bronchiolitis
obliterans was never
exposed to study numbers.
The issue with retrospective
CRF assembly
is whether somehow in the retrospective
nature of
filling out these forms an assessment of
an outcome
is
altered. However, when the outcome is
death, or
the presence or absence of bronchiolitis
obliterans
on an original histopathology report, or
whether
FEV1 has declined by 20 percent or more
from a
72
post-transplant maximum, these are hard
endpoints
and would not be expected to be altered
by
retrospective CRF assembly.
Treatment groups were not
balanced on each
and every baseline characteristic. The purpose of
randomization is not to eliminate all
imbalances
but, rather, to randomly distribute them
between
groups.
The two treatment groups are comparable,
and of the clinically relevant baseline
characteristics we examined the majority
are
balanced or, if anything, would favor
better
outcomes in the placebo group.
Finally, when imbalances do
occur in
clinically relevant variables
statistical models
can be used to adjust for these both in
univariate
or multivariate analyses, and we have
presented
such analyses that show that the data is
robust.
So, we feel extremely confident in
saying that
baseline imbalances did not explain the
efficacy of
inhaled cyclosporine.
The study did not meet its
primary
endpoint of decreased rate of acute
rejection.
73
However, scientific understanding has
evolved since
the design of ACS001 and the lack of an
effect on
acute rejection is consistent with low
systemic
exposure. The design of the study doesn't impact
the assessment of survival or chronic
rejection or
alter how the data is obtained. It is also
important to note that survival and
chronic
rejection were prospectively defined
secondary
endpoints. These analyses are not post hoc nor do
they constitute data mining.
Finally, the survival and
chronic
rejection data are clinically important,
statistically significant and
scientifically sound.
Inhaled cyclosporine is delivered
directly to the
airways, the site of chronic
rejection. Inhaled
cyclosporine prevented chronic rejection
and, in
doing so, markedly improved
survival. The
importance of this data is illustrated
by the fact
that physicians from 30 different
transplant
centers in the United States, which
represents
almost half of all active lung
transplant centers,
have requested early access to inhaled
cyclosporine
74
as part of our early access program.
We have been advised to make
it clear to
the advisory committee where there are
differences
of opinion between Chiron and the FDA,
and that is
really why we are here today. So, this slide
illustrates five of the most important
areas where
we disagree.
First, we believe that covariates
in a
statistical model should be chosen based
on an
association with the clinical outcome
rather than
because of an imbalance. In the case of ICU time,
the use of ICU time greater than ten
days, there is
an
imbalance toward the placebo group.
However,
this is not documented to be associated
with
survival. If an ICU time greater than seven days
is chosen that imbalance is minimized,
and if an
ICU time greater than four days is
chosen the
imbalance is reversed. We believe that it is
important to differentiate patients who
had an
earlier, easier postoperative course
from those who
had a harder postoperative course, but
believe that
this is best accomplished by the
randomization
75
stratification variable enrollment
period, early
versus late.
In the case of donor inotropic
support, we
have yet to find a single reference that
even
considers this variable, much less finds
it
clinically relevant and the FDA has
called this one
of the most clinically relevant factors
in the
study.
We do have variables and we do
have data
on donor quality through other variables
that have
been documented in the literature to be
clinically
important, such as donor age, donor
bacterial
colonization, donor graft, ischemic time,
and these
are balanced between the two
groups. The important
point is that the use of a covariate
that is
imbalanced but not clinically relevant
will always
cause results to trend toward the null
and that is
what we have seen with the FDA analyses.
Second, in analyses of
survival we
disagree that patients whose use of
donor inotrope
or the donor inotrope data is
missing--we disagree
that these patients should be excluded
from
76
analyses. In the FDA analysis, by excluding
long-term survivors in the inhaled
cyclosporine
group, the treatment effect and p values
are going
to be altered inappropriately.
Three, we believe that
survival is best
analyzed at the prospectively defined
study end
date rather than one year after--or
nearly a year
after the study was over. I have already discussed
our reasons for this.
Four, we believe that patients
with
bronchiolitis obliterans, or OB, should
be included
in an analysis of chronic
rejection. The diagnosis
of OB has a specificity of over 95
percent.
Patients with BOS and OB represent
overlapping
subsets and, therefore, to look at
either one
alone, we believe, is a subset analysis.
Finally, five, analyses of BOS
should not
censor deaths. This is clearly informative
censoring, and when deaths are not
censored and
BOS-free survival is analyzed the
results are
statistically significant and remain so
when
controlled for by CMV mismatch, primary
diagnosis
77
and early acute rejection. Analyses that censor
death can be informative but we have
shown in our
chronic rejection that although they can
be
informative they shouldn't be used as
the primary
analysis.
So, I would like to end with a
summary of
the clinical data that I presented. In the lung
transplant population with no
appropriate approved
drugs, very few randomized clinical
trials and a
dismal prognosis that hasn't changed in
almost 20
years, treatment with inhaled
cyclosporine was
associated with a 79 percent decrease in the risk
of death. Treatment with inhaled cyclosporine was
associated with a 72 percent decrease in
the risk
of chronic rejection or death. We have
demonstrated that our efficacy results
are robust
through a number of different
analyses. We have
also demonstrated that the ACS001
placebo
population is representative of a larger
U.S.
transplant population. We have demonstrated that
treatment with inhaled cyclosporine was
not
associated with any systemic
toxicities. Finally,
78
inhaled cyclosporine was associated with
local
respiratory tract irritation and
bronchospasm, a
relatively small price to pay in light
of the
profound survival benefit.
Thank you. I would like to end and turn
this presentation over to Dr. Ronald
Helms,
Professor Emeritus of Biostatistics of
the
University of North Carolina, who is
going to spend
a few minutes discussing the statistical
considerations of the study.
Statistical
Considerations
DR. HELMS: Thank you, and thank you for
the opportunity to come and address this
group here
this morning. My time is short so I am going to
dive right in, if I may.
Why are we here? Well, this survival
curve tells why we are here, the
profound
difference in survival in these two
treatment arms,
as has been discussed at length already.
A second reason I am here is
that this is
a very interesting project, a very
interesting
project.
Let me first establish a disclaimer and
79
my conflict of interest issue. The views that are
expressed in this presentation are mine
alone and
do not represent either the FDA or
Chiron or Rho,
my current employer, or the University
of North
Carolina, my former employer. It is possible that
these views may represent the best
interests of
future lung transplant patients. In terms of
financial conflict of interest, neither
Rho nor I
have any financial stake in the outcome
of this
submission. Less than half a percent of Rho's
total income this year will come from Chiron.
Chiron pays Rho an hourly consulting fee
for my
time plus travel expenses and, in fact,
my board of
directors told me they would prefer that
I work on
other projects that are more financially
rewarding
to the company.
[Laughter]
So, I am here despite
that. Also, neither
Rho nor Chiron has edited my
presentation and I
have reviewed the briefing documents
that you have
seen from both the FDA and Chiron, plus
some other
more comprehensive documentation. So, I feel
80
unconflicted here.
So, why am I here? Well, coming back to
the results of this study and the fact
that it is a
very interesting project--it is a very
interesting
project and we have a problem. By "we" I mean the
professionals sitting here around the
table, the
FDA professionals, the Chiron staff--we
have a
problem.
This Kaplan-Meier graph tells
that this
product has the potential to save the
lives of a
statistical number of lung transplant
patients.
The NDA does not meet the usual
regulatory
requirements for approval. Should it be approved?
Well, there are advantages and
disadvantages to approval in this
case. The
results indicate that if approved,
widespread use
of this product would probably save the
lives of
around 300 to 350 lung transplant
patients a year.
Now, I should just comment that my
comments here
are really aimed at the
non-statisticians on this
panel.
The statisticians know how to interpret
relative risk and those kinds of
things. I thought
81
it would be useful to translate this
into lives
saved after a period of time when the
product was
in widespread use. It appears to improve the
survival probability by about 30 or 35
percentage
points.
You see the numbers there, somewhere
around 50-90 percent, and there are
about 1000 or
1100 transplant patients so if you do
the
arithmetic it comes out to around 300 to
350 lung
transplant lives saved a year.
Another advantage is--and this
is a
practical advantage--if this product
were approved
FDA could require Chiron to conduct the
sufficiently large follow-up study that
Chiron has
proposed. If the study were negative the approval
could be withdrawn and, as a practical
matter,
without approval the follow-up study
will never be
done.
Off-label use of the product would
ultimately become a standard of care and
failure to
use it would be considered unethical and
subject to
lawsuits and those sorts of things. And it is an
interesting aside that we have a very
closely
related case. Cyclosporine, which is used
82
universally in the treatment of lung
transplants,
is not approved for that indication; it
is all
off-label use. The studies have never been done.
There are some obvious
obstacles to
approval. We have the results of only one small
unconfirmed study. This is a serious problem. It
is a serious problem. This one study has a number
of flaws that have been noted by both
Chiron and
FDA.
Here are some opinions, one of these is very
important; some are potentially
important; and some
really are inconsequential in my
opinion.
The very important flaw in
this clinical
trial from a statistical perspective is
that the
stated primary outcome was acute
rejection, not
mortality or survival. The statistical methods
that we routinely use for Phase III
confirmatory
studies aren't very helpful with this
problem, the
problem of switching the primary
endpoint from what
was stated in the protocol to a
secondary endpoint.
But good, old-fashioned common sense can
be
helpful.
When you see that big an effect on
survival you very likely made an
important
83
discovery.
Now, we could use, as
statisticians, a
branch of statistics called decision
theory for
formal risk-benefit analyses here but
the fact is
that if we did that the analyses would
be based on
a number of assumptions and if you are
strongly
opposed to approval here you challenge
the
assumptions, and rightly so. The result is so big,
the difference in survival is so big
here that we
can tell what the outcome would be
anyway, that it
would lead to a decision in favor of the
product.
Some potentially important
flaws--let me
address those. My time is brief and I won't go
into statistical details but there is an
important
side note here. At least as of a few weeks ago,
the FDA and Chiron biostatisticians had confirmed
each other's statistical
calculations. The point
is that there is no issue about
correctness of
populations. Now, you are going to hear different
perspectives obviously from Chiron and
FDA. In my
opinion, the issues here are about how
to use and
interpret the statistics, not the actual
results,
84
and I think that is good to know.
There are some potentially
important flaws
that have already been mentioned and you
will hear
some more about that in the FDA
presentation. The
randomization, if done improperly, could
be an
important flaw; the lack of balance with
respect to
important baseline characteristics;
unmasking or
unblinding--we used to call it
unblinding but then
I worked with some ophthalmologists and
they taught
me to use the word
"unmasking." The study was
conducted in such a manner that the
investigators
could have been unmasked essentially,
and the study
was conducted at a single clinical
center, not
multiple centers.
I want to cut to the chase
because my time
is limited. The bottom line is I reviewed each of
the potentially important flaws and my
conclusions
for each one were that each was either
not a flaw
at all or was relatively
unimportant. For example,
the randomization failed to balance with
respect to
all the baseline factors. It rarely does in
clinical trials, even large clinical
trials. It
85
has been my experience over the last 15
years since
I began looking at this that only one
out of
hundreds of clinical trials was balanced
with
respect to all important baseline
factors. So, it
is not a case of failure. On request, on somebody
else's time, I will be happy to talk about
some of
these issues.
There are some unimportant
flaws in the
clinical trial, and they are listed
there. We
don't have to spend time on that.
Let me raise an important ethical point
for the members of the panel. Suppose the data
from this study were the results of an
interim
analysis half way through the study, and
suppose
the members of this advisory panel were
instead
sitting as the study's data and safety
monitoring
board, would we be ethically bound to
terminate the
study to protect future patients who
might be
assigned to placebo? I suspect that many of you
have sat as members of data and safety
monitoring
boards and faced precisely this question
in the
middle of a study. I have.
And I believe that
86
everyone on the DSMBs in which I participated
would
have stopped this study to protect
placebo
patients, the results of the study are
that
compelling.
I think there is another
important ethical
point.
I think the people in this room--again, the
FDA staff, the advisory panel, the
Chiron
staff--are ethically bound to find a way
to make
this product available on-label to U.S.
lung
transplant patients. It will be used off-label.
It already is being used off-label but
without
approval for some years this product
will only be
available to people who can afford to
pay for it
from their own funds because it won't be
covered by
insurance. So, we have a product
that would be
made available to wealthy people and not
others.
We also, I think, are
ethically bound to
find a way to make it necessary for
Chiron to
conduct the proposed postapproval
follow-up study.
If we don't, it won't be done. Realistically, it
can only be done as a postapproval study
for
financial reasons that Chiron can talk
to you
87
about.
What an interesting
project! Thank you
for the opportunity to talk to you.
Safety and
Benefit-Risk
DR. DILLY: Thank you, Prof. Helms and
thank you, everyone, for your patience
in following
through our presentation. I am going to conclude
with about five minutes of remarks to
end the
Chiron presentation.
What I would like to do is consider
some
of the issues relevant to the potential
approval of
Pulminiq. Clearly, we believe the best way to help
lung transplant patients now is to make
CyIS
available. Lung transplant, as you heard, is in
many ways the poster child of the orphan drug
indication. Despite the incentive provided by the
orphan drug designation, no drugs have
been
developed for lung transplantation,
probably
because the economics simply don't work
for a
conventional development program. So, if we are
looking for new drugs, it is going to
come from
sources like this.
88
Now, we are not suggesting for a moment
that the burden of evidence is any
different for an
orphan indication. Rather, what we are suggesting
is that we must consider the evidence
that exists
on its merit and, in fact, the case for
approval of
this drug is very strong.
The scientific premise for
inhaled
cyclosporine is extremely
straightforward. We are
giving an effective drug, with systemic
toxicity,
by inhalation to achieve higher lung
levels. This
has been done, of course, successfully
in asthma,
in COPD, in cystic fibrosis. It is a well
precedented approach. In fact, as you heard, the
idea is so straightforward that many
lung
transplant centers were already using
inhaled
cyclosporine empirically before the
clinical data
of ACS001 were known.
Of course, these are the
essential
clinical data. Patients who received inhaled
cyclosporine in the pivotal trial lived
significantly longer than those who did
not. You
have heard compelling arguments that the
difference
89
in survival was due to inhaled
cyclosporine and
that the benefit is highly likely to be
generalizable to other patients in other
treatment
centers.
Publication of these data will rightly
have a major impact on the treatment of
lung
transplantation with or without approval
of
Pulminiq. The case for benefit is very strong.
Also as you have heard, there is very
little risk
of harm.
This is a known drug. Local
toxicity in
the lung is minor and systemic exposure
is not
clinically important. Finally, this is a very
small population with an entirely
clear-cut
diagnosis, lung transplantation. So, the chances
of a major public health problem from
broad usage
is very, very small. In other words, the
demonstrated benefit far outweighs the
potential
for harm. The bottom line is patients will live
longer if inhaled cyclosporine is made
available to
them.
Of course, some questions
remain open
because of the nature of the clinical
program
conducted to date. So, the right thing to do for
90
patients is to approve inhaled
cyclosporine now and
conduct the appropriate postapproval
study to
address those outstanding
questions. So, I would
like to finish the Chiron comments by
considering
what that postapproval study should look
like.
The central question really is
how to give
inhaled cyclosporine. We have seen benefits from
therapy lasting for up to two
years. All logic
dictates that for a chronic rejection
endpoint
chronic therapy should be better. We need to study
that.
We need to study dosing beyond two years.
We need to work on making the first few
doses as
tolerable as possible so we can get as
many
patients as possible onto an effective
dosing
regimen.
We would also love to know
more about the
interplay of the key clinical endpoints,
survival,
rejection, lung function. You can only interpret
so far based on a single, relatively
small study
with such a bright line survival
effect. We
believe that 300 mg of inhaled
cyclosporine by
nebulizer three times a week is a
perfectly
91
appropriate inhaled regimen and the
right thing to
put in the label, but there are some
questions we
need a bigger study to answer.
How do patients do if they
actually
tolerate a dose below 300 mg--100 mg or
200 mg? Is
the need for systemic dose
intensification reduced
with effective long-term inhaled
therapy? What is
the best way to deal with treatment
interruptions,
for instance during concomitant
illnesses? Of
course, it will be informative to have a
much
bigger safety experience.
So, here is our proposal,
essentially this
is a very large single-arm study with
external
controls. We believe that we could draw the
control arm now from the UNOS
database. From the
comments you heard from Dr. Golden and
others, we
know what happens to lung transplant
patients
treated with current standard of
care. So, 250
patients will be treated with a labeled
regimen of
inhaled cyclosporine for 5 years. A placebo group
is not appropriate and not necessary
given the
robust survival advantage already
demonstrated with
92
inhaled cyclosporine. There will be 2 external
controls, firstly, about a thousand
matched
patients with long-term follow-on data
drawn from
the UNOS database. Secondly, a group of
contemporaneous controls who will not
receive
inhaled cyclosporine. The exact size of this
group, of course, will be somewhat
dependent on the
rapidity of uptake of inhaled
cyclosporine therapy.
So, we would expect that the
availability of those
patients would go down over time.
What I am attempting to
describe to you
here is a study that is entirely doable
in the
postapproval context. The primary endpoint will be
chronic rejection-free survival, with
all-cause
mortality and lung function as secondary
endpoints.
We see three safety endpoints as
particularly
interesting: Firstly, infections requiring
hospitalization because we believe that
that signal
in favor of the lower incidence of
pneumonia on the
inhaled cyclosporine group in ACS001 is
probably
real and due to decreased lung damage
from chronic
rejection, making the lungs less
susceptible to
93
infection. Secondly, we want to look at renal
dysfunction and malignancy as readouts
of systemic
immunosuppressive status, as well as
diligent
follow-up for the other safety
events. In fact,
this will be the largest study ever done
and the
longest study ever done in the lung
transplant
setting.
In conclusion, based on what
we know now
lung transplant patients will clearly
live longer
with inhaled cyclosporine. The outstanding
questions can be addressed in a
postapproval study
and so we believe that inhaled
cyclosporine should
be approved now.
Now I would like to invite Dr.
Scaife to
the podium as well and we can take your
questions.
DR. SCAIFE: Thank you very much, Dr.
Dilly.
We can open to the FDA and the panel for
questions.
Questions from the
Panel
DR. SWENSON: Go ahead.
DR. SCHOENFELD: I just had a few
questions on acute rejection since that
endpoint
94
wasn't exactly described. How is that diagnosed?
DR. SCAIFE: Dr. Sarah Noonberg?
DR. NOONBERG: It is diagnosed by
transbronchial biopsy and it is graded
0-4. So, it
is the same transbronchial biopsy that
can be used
to make the diagnosis of bronchiolitis
obliterans.
DR. SCHOENFELD: So, was there sort of a
program of periodic transbronchial
biopsies in
these patients during the study?
DR. NOONBERG: Yes, approximately the
first month and then three to four
months afterward
for a period of two years and then as
clinically
relevant. It should be noted that the mean greatly
exceeded that. All patients had the minimum and
the mean was far higher.
DR. SCHOENFELD: Another question about
acute rejection, once a patient has bronchiolitis
obliterans can they have acute rejection
also?
DR. NOONBERG: Yes.
DR. SCHOENFELD: I see.
So, it can happen
after the chronic rejection has begun.
DR. SWENSON: Dr. Hunsicker?
95
DR. HUNSICKER: I would like to ask Dr.
Golden if he would be willing to comment
on this.
Let me give perhaps a little bit of a setting for
my concerns here. We have a study in which the
primary outcome was not met and the
secondary
outcome is met that at the time the
study was
conceived didn't correspond to biology
that was
understood. The understanding of biology has
changed but--I would like to say I am
not a
pulmonary person but I am a
transplanter--is still
not very well understood. So, I think I need to
have somebody who really understands the
pulmonary
rejection business to tell me a little
bit about
the preclinical information on the
impact of local
immunosuppression for chronic rejection
in the
lungs.
Right now the general assumption is that
most of the effects of immunosuppression
are
central.
I grant you that there is some very real
interest in the possibility of local
immunocytes
being locally immunosuppressed but this
is not what
I would call a robustly well understood
part of
science.
So, since we can't look at this really in
96
most of the forms of transplantation, it
may be
that we have some better understanding
of this from
the pulmonary point of view and I would
like to get
the best understanding I can have of
what is
currently understood about the impact of
local
immunosuppression for pulmonary rejection.
DR. GOLDEN: First of all, nobody knows
with precision exactly where you are
treating
locally along the airway. I would infer, given
that there is a difference in chronic
rejection,
that that is generally a more peripheral
airway
portion.
DR. HUNSICKER: Let me clarify that. I
wasn't talking where along the airway, I
was
talking about central immunological events
as
opposed to peripheral immunological
events. Most
of us have assumed that the primary
effects of
immunosuppression are central rather
than in the
peripheral organs, particularly of the
calcineurin
inhibitors. So, what I want to know is, is it
known what the effects of local
immunosuppression
in lung rejection are in experimental
models for
97
instance?
DR. GOLDEN: Let me make sure I understand
the question. You want to know when you give
systemic immunosuppression centrally how
that might
affect the airway.
DR. HUNSICKER: Actually, it is the other
way around. Let's assume that when cyclosporine
gets into the body where it really is
doing its
thing is in the lymph nodes and the
spleen, and
stuff like that where the cells are
being
developed. Then it doesn't make a whole lot of
sense that local application should be
effective.
If, in fact, there is local effect on
the lymphatic
cells that are in the bronchi, then it
might make
sense.
Right now this is something that is not
understood in other forms of rejection
because we
can't get at the local tissues quite so
well. What
is known about this?
DR. GOLDEN: I think this is a new area.
To answer it the best I can, one would
have to
infer that systemic therapy does not
reach a level
of mucosal benefit, that applying the
medicine
98
locally, as you say, must have some
local immune
benefit.
The slide I showed of the mucosa with
lymphocytes moving into the submucosa--I
can only
infer that systemic therapy or having a
central
effect on lymph nodes, etc., as you say,
is not
reaching a level of immunosuppression
along the
airway that is benefitted by a direct
local
application to the epithelium of an
immunosuppressant.
I must say that there are
ongoing studies
now with other agents, like inhaled
rapomyacin, to
also try and treat this. That is an animal study,
very preliminary. So, the best answer is I really
don't know. I infer that there is a benefit
locally to applying, as you can uniquely
do in the
lung as you said, to a mucosal process.
DR. PRUSSIN: Calman Prussin, NIAID. Just
to follow-up, in all immunologic and
allergic lung
diseases I know T-cells are being
activated in the
lung locally and expressing cytokines
locally. So,
if you are applying that drug locally
you would
expect that it would have an effect
there as
99
opposed to cells that are in the spleen
which are
mostly resting and not producing
cytokines. So, it
does make sense immunologically.
DR. SWENSON: Dr. Gay?
DR. GAY: Steve Gay, University of
Michigan. I had a question concerning the early
stoppage of the trial. Pittsburgh is a fairly
aggressive transplant institution and it
seems as
if the study was initially powered for
120
patients.
The study was stopped at 56 patients.
I
was wondering what factors led to the
early
stoppage with the fact that the primary
endpoint
was clearly not achieved at that point.
DR. DILLY: The original sample size
estimate was based on the availability
of patients
during the predefined study duration,
and the study
ended on the day that the study was
intended to
end.
That was not influenced by the primary
endpoint. It was simply that there were
approximately 120-odd patients during
that period
who were transplanted at Pittsburgh and
around half
of those patients went on to the
study. So, in
100
fact, this was a pretty good enrollment
of eligible
patients at the site.
DR. SWENSON: Dr. Proschan?
DR. PROSCHAN: I also have a question
about that because you say it was not
influenced by
the results. Does that mean the results were not
known at that time?
DR. DILLY: The study was done blinded at
that time so the results were not known
and the
blind was well preserved. We really became aware
of those results after the unblinding.
Another thing that we have
looked at in
some detail--and perhaps Dr. Noonberg or
Dr. Capra
would like to talk about this--is
whether there was
something special about the patients
that went into
the study. Was there something about the placebo
group and whether these were a selected
group of
patients? All the evidence says is that these were
the same kind of patients as were not
enrolled in
the study.
DR. NOONBERG: When we compared the
placebo and ACS001 to UPMC unenrolled
controls we
101
found that the survival curves were
comparable,
with a p value of 0.99, so these didn't
represent a
select group of patients. One of the reasons for
the poor enrollment is that there just
simply
weren't enough transplants performed
during that
period of time. During those three years there was
a far lower time for--I am just going to
stop and
show this slide quickly that demonstrate
the
survival of screen failures, so patients
who were
not enrolled in ACS001 and those that
were enrolled
into the placebo group.
But to go back to my previous
thought, I
mean, they couldn't have enrolled 136
patients.
There were 105 transplants performed
during the
enrollment period. The enrollment period didn't
stop early; the enrollment period had a
three-year
duration and it stopped at that
three-year
duration. It just didn't enroll the requisite
number of patients that it anticipated.
DR. SWENSON: I believe Dr. Proschan has
another question, but for the members of
the panel
here, if you will just simply hit your
"talk"
102
button we will be able to see the light
on and you
needn't raise your hand. That will probably be
easier for us. Dr. Proschan?
DR. PROSCHAN: I guess I was just
following up on that because, you know,
usually
even if it is the primary endpoint to
stop early
there are boundaries that you use and,
you know,
the commonly used boundary is called the
O'Brien-Flemming type of boundary, and
this trial
would not have met that level of
evidence. But
that is a concern, mainly motivated by
my thinking
that the results were known at the time
you stopped
and, therefore, the possibility on a
random high.
DR. SWENSON: Dr. Moss?
DR. MOSS: I have a question I guess for
Dr. Noonberg but you, guys, might answer
it too.
It has to do with the generalizability
of your
results and I think you showed it on
that slide.
Normally when you have figures on a
study you say
we screened this many people; these many
were
excluded and we were left with 10
percent of the
population. That wasn't included in any documents
103
but I think you brought it out a little
bit there
so could you just go over that and say,
you know,
these many people were screened and
these many were
excluded and you were left with what
percentage of
the patients that were actually enrolled
in the
study, so we can get an idea about the
generalizability of your data?
DR. SCAIFE: Dr. Noonberg?
DR. NOONBERG: You want to go back to that
last slide?
DR. MOSS: I think the data was there but
you never mentioned it before. You don't need the
slide, just how many people were
screened and how
many were excluded and you were left
with this many
people so we can see how generalizable
your data
are.
DR. NOONBERG: Right.
There were 105
transplants performed during the roughly
3-year
enrollment period and there were 68
patients--actually, 58 patients enrolled
during
that 3-year period; 10 were enrolled the
year
previous. So, approximately half and, as I say,
104
the survival in the enrolled and the
placebo
survival in the unenrolled group is
comparable,
with a p value of 0.99.
DR. SWENSON: Dr. Venitz?
DR. VENITZ: I want to follow-up on Dr.
Hunsicker's question in a different
way. He was
questioning the biology supporting
localized
administration versus systemic administration. You
obviously looked at exposure to
cyclosporine after
inhalation relative to oral or systemic
administration. Did you look at exposure to the
lung in either clinical or preclinical
models and
compare systemic administration to
inhalation?
DR. DILLY: We actually have access to
data on a scintigraphy study looking at
labeled
inhaled cyclosporine, conducted by Dr.
Corcoran at
the University of Pittsburgh, and I
think it would
be extremely relevant to show you those
data. I
will give you the editorial comment
while Sarah
retrieves the slide.
But with the 300 mg dose put
into a
nebulizer, what we have seen is that
about 25 mg is
105
the applied dose to the lung. That is achieving
dose levels in the lung that would
require
approximately doubling of the systemic
immunosuppressive dose, and that is our
central
premise, which is that that is not
something that
you could routinely do in clinical
practice because
of the toxicities.
DR. NOONBERG: Again, just going back to
the first animal experiments in 1988
where they
just gave single doses of inhaled
cyclosporine,
they found that pulmonary concentrations
were
10- to 100-fold higher than
concentrations in other
tissues.
In the rat model that I described
pulmonary concentrations were at least
3-fold
higher than systemic
concentrations. So, that is
the data that we have for preclinical.
DR. VENITZ: Again just to follow-up, how
does that compare if you give
cyclosporine
systemically? You are talking about what happens
after inhalation. Right?
The levels in the lung
are higher than in other tissues, higher
than in
plasma?
106
DR. NOONBERG: Right.
DR. VENITZ: And I am wondering how would
that compare if a dose of cyclosporine
was given
intravenously to those animals. What lung
concentrations would you be able to
achieve?
DR. DILLY: What we showed was a 25 mg
dose applied to the lung through
inhalation. You
have to remember that when you put 300
mg into a
nebulizer an awful lot goes into the
atmosphere and
an awful lot doesn't get into the
lung. That 25 mg
applied dose, in terms of mg/g lung
weight, equates
to approximately an 8-fold higher systemic
dose.
If you assume 100 percent
bioavailability of the
systemic dose you have given
parenterally, that
would mean that you are looking at
something like a
200 mg dose given orally to get to the
same lung
levels.
That is based on AUC calculations.
If you
are thinking about peak levels, then the
difference
is far greater because, of course, you
get the
early distribution phenomenon into the
lung.
DR. VENITZ: And that is in humans? Any
preclinical data to back that up?
107
DR. DILLY: Actually, that is in the
briefing book. The best data we got is in humans.
It is actually in the briefing book.
DR. SWENSON: Dr. Burdick?
DR. BARRETT: In Dr. Golden's presentation
he showed some data looking at BOS as a
disease
progression marker. However, in the documentation
provided both BOS and FEV1 were not
determined to
be significantly different between the
two groups.
So, assuming chronic rejection as the
indication
here for this product, can you give some
reasons
why you think that occurred?
DR. SCAIFE: Dr. Bill Capra is the lead
statistician for Chiron.
DR. CAPRA: Actually, CyIS did show an
effect on BOS, specifically BOS-free survival. The
reason why our results are different
than the FDA's
is that the FDA censors BOS in their
analysis and
this is informative censoring. Because the reasons
for death are disease related, it is
invalid to
censor deaths in a disease progression endpoint.
The FDA has recently issued a
guidance on
108
this type of endpoint for oncology
studies where
they recommend using a progression-free survival
endpoint in such an analysis rather than
time to
progression analysis. If you do such an analysis
with this BOS what you see is an effect
of
cyclosporine on improving BOS-free
survival with a
p value of 0.99.
DR. BARRETT: Could you comment on the
FEV1 though?
DR. CAPRA: Sure.
We looked at FEV1 in a
number of ways. We looked at change from baseline
to the final value; change from
post-transplant to
the final value. We looked at time adjusted area
under the curves and we looked at
slopes. In none
of these analyses did we see a
statistical
significance. However, in each and every analysis
the point estimate favored the active
group. As an
example, up here I have the results of
the change
from baseline to the final value and we
see that
the placebo group increased by 0.15 L
and the
active group increased by 0.40 L. So, there seemed
to be a trend, however it was not
statistically
109
significant.
We think there are some limitations to the
FEV1 analysis and we think one of the
major
limitations is the informed
censoring. Because
there is such a large number of deaths
and because
the FEV1 values cannot be obtained from
subjects
after they die it goes against
censoring. Also,
FEV1 itself is highly variable. Any single subject
might have short-term fluctuations and
what BOS
does is it basically ignores those
short-term
fluctuations and looks for a sustained
20 percent
decrease. So, when you look at BOS, removing some
of that variability, and when you
address the
informed censoring by use of
progression-free
survival endpoint rather than time to
progression
endpoint, we see an effect of
cyclosporine on lung
function, namely, BOS-free survival with
a p value
of 0.019.
DR. DILLY: Can I just add one
supplementary comment? This is exactly the kind of
question that we need to nail down in
the next
study because what we want to do is take
a large
110
group of patients, enroll them, nail
down what
their lung function is and follow them
over time
because, remember, the objective of this
treatment
is to preserve the lungs in a good
condition. So,
actually a no-effect on FEV1 in that
context in a
large group of patients would be a great
outcome,
and that is what we want to show next.
DR. SWENSON: Dr. Gay?
DR. GAY: My question is to follow Dr.
Moss' question from a while ago. I am still not
clear on the number of patients, why the
number is
so small, the number of patients that
were included
in the study. It is essentially a single-site
study in which every therapy is an
off-label one
for the treatment of rejection in
transplantation.
I am trying to get a grasp of why there
were so
many screening failures, essentially 50
percent
screening failures in the study over the
course of
the three years. Why weren't more patients
included or made available to be
included in the
study, and what were the reasons for
that?
DR. DILLY: In fact, what we would
111
consider the 50 percent enrollment of
eligible
patients as quite good in a clinical
study. Our
experience has been typically when we
are trying to
enroll clinical trials, which is what we
do for a
living, that we see something like 25-40
percent
enrollment into the study. So, when we went into
Pittsburgh and we looked at this whole
body of data
we were quite reassured that the
patients had gone
to the study in an elegant way; that
about half of
them got into the study; and there was
nothing
particularly strange about the patients
that did
and the patients that didn't. So, we did not see
that as an issue and we came back to the
fact that
we saw the data as robust.
DR. SWENSON: Dr. Prussin?
DR. PRUSSIN: I was impressed by the
heterogeneity in terms of the
cyclosporine group in
terms of the dose that they
received. You know,
some of the subjects received all the
doses for the
full length of the study, and various
documents
suggest that something like 9/36
received 1 month
or less.
So, my question is did you ever stratify
112
the analysis for survival based on how
much drug
they received? It is pretty impressive that 9 of
these patients received only a month of
drug and
yet presumably had a fairly good
survival.
DR. SCAIFE: Dr. Noonberg?
DR. NOONBERG: There are several responses
to that question. The first is that ACS001 wasn't
a dose-response study and we don't have
formal
dose-response data. However, in one of our
sensitivity analyses we did exclude
patients, 14,
who didn't receive at least 80 percent
of the
protocol maximum dosing and they are
excluded from
analysis. As would be expected, the p value is
going to go up due to loss of power,
however, the
treatment effect is essentially
unchanged.
DR. PRUSSIN: But on the flip side, why
did the patients who essentially didn't
receive
drug respond to a drug they didn't
get? That is
what I am more concerned about, not the
ones that
did receive the drug. Yes, they responded even if
the p value is going to be higher, but
the ones
that essentially were on the active side
of the
113
protocol but who effectively did not
receive drug
still had an effect in their survival. Correct?
DR. NOONBERG: I mean, we used an
intent-to-treat analysis so we include
those
patients, but there are placebo patients
that have
long-term survival too. This isn't a uniformly
fatal diagnosis so you would expect to
see
variability in survival. But we include the
intent-to-treat analysis in accordance
with
guidelines.
DR. SWENSON: Dr. Noonberg, I have a
question that somewhat follows up on
that very same
one but is occasioned by one of your
graphs here,
and that is you continue to see and, in
fact, you
even highlighted that more patients
seemed to be
prevented in their chronic rejection appearance
following the cessation of their
two-year therapy,
if I read this graph correctly. Can you explain
why this drug may, in fact, have
benefits beyond
its cessation?
DR. NOONBERG: The CR that is in green on
this graph doesn't represent new
diagnoses of
114
chronic rejection but, rather, deaths
associated
with chronic rejection. So, they are not
necessarily new rejection episodes. So, this just
highlights the strong association of
chronic
rejection with death and the fact that
you don't
see that in the inhaled cyclosporine
group. But
the chronic rejection episodes are
actually
occurring throughout the process.
DR. SWENSON: Okay.
Dr. Hunsicker?
DR. HUNSICKER: On that same graph, it was
not clear to me when you put that
up--you don't
have to put it back up again, I think we
have all
seen it--how you made the diagnosis of
chronic
rejection in those cases. Was that either well
defined BOS or a biopsy, or was that a
clinical
definition of chronic rejection based on
the fact
the patient had died with lung disease?
DR. NOONBERG: It is either by
transbronchial biopsy with histologic
proof of the
lesion of bronchiolitis obliterans or
clinical
BOS--
DR. HUNSICKER: So, all of those patients
115
that had the CR in green there either
had one or
the other?
DR. NOONBERG: Correct.
DR. HUNSICKER: I have a couple of other
questions just to be sure I am correct
on this, you
referred to the analysis plan. First of all, there
was a prospective analysis plan that
specified that
the total survival at the end of the
study was to
be used as the primary outcome rather
than the data
at the end of two years of
treatment? I wasn't
quite sure. There were three or more different
types of analysis that were discussed in
the
briefing books. What did the original prospective
analysis plan say was to be used as the
primary
evaluation? Was it total survival at March 31, or
whatever it was, or was it supposed to
be at the
end of the two years of treatment?
DR. NOONBERG: It should have been at the
end of the study. Dr. Aldo Iacona, the principal
investigator is nodding his head so,
yes.
DR. PROSCHAN:
But it was not survival; it
was acute rejection.
116
DR. HUNSICKER: Well, I understand--
DR. NOONBERG: Right, but the survival is
the secondary endpoint--
DR. PROSCHAN: We have so many secondary
endpoints to look at, we have to figure
out which
endpoint we are looking at.
DR. HUNSICKER: And the second question I
have is that I thought I found in the
briefing book
that of the ten patients who were put
into the
so-called pilot thing, five of them had
eventually
died.
Is this correct?
DR. NOONBERG: That is correct.
DR. HUNSICKER: So, five out or ten
patients, and they received treatment
for the full
two years or at least as much of the two
full years
as one would have expected them to get?
DR. NOONBERG: Correct.
When those
patients are included in the statistical
analysis,
and that was one of the sensitivity
analyses that
we performed, the results were still
statistically
significant. They died but the timing of death is
very important, as well as the fact that
they
117
died--
DR. HUNSICKER: Sure.
DR. NOONBERG: Here is a Kaplan-Meier of
survival from time of transplantation to
study end
date including the randomized and the
pilot, with a
p value of 0.018.
DR. SWENSON: At this time we should
break.
I know there are more questions and they
can be taken up in our other discussion
sessions
later today. We will reconvene at 10:15.
[Brief recess.]
DR. SWENSON: We should make a start on
the next session, and Dr. Hernandez, of
the FDA,
will lead the discussion.
FDA Presentation
Overview of Clinical Trial
Efficacy
and Safety Evaluation
Discussion of Analysis
DR. HERNANDEZ: Thank you.
Good morning.
During this presentation I will describe
the
Division's perspective on the
application for
cyclosporine inhalation solution. I will start by
118
saying that this is not a regular NDA
application.
The study, submitted to support the
proposed
indication, is a small Phase II study
that failed
to meet its primary endpoint for the
prevention of
acute rejection. However, the potential for the
prevention of chronic rejection and
improved
survival are very important aspects for
the lung
transplant population for which
long-term survival
is mostly limited by chronic rejection.
The agency considered that the
potential
survival benefit in this specific
transplant
population was reason enough to accept this
new
drug application for review. The proposed
indication for cyclosporine inhalation
solution
requested by Chiron is for increase in
survival and
prevention of chronic rejection in
patients who
receive allogeneic lung transplantation,
in
combination with standard
immunosuppression.
In my presentation I will give
an overview
of the data submitted in this NDA. Then I will
summarize study ACS001 objectives, outcomes
and
limitations. I will describe the FDA review which
119
will address the following
subjects: Acute
rejection, obliterative bronchiolitis,
bronchiolitis obliterans syndrome and FEV1
data,
and survival. Then I will discuss the recipient
and baseline characteristics, donor
baseline
characteristics, the primary causes of
death,
available autopsy results, dosing
information and
related outcomes and, finally, Dr.
Cavaille-Coll
will give you a summary of the safety
considerations and our summary
conclusions.
The data submitted to support
this
application was derived from two reports
generated
by Chiron Corp. That report was referred to as
ACS001 and ACS002. The study ACS001 is actually
the name given by Chiron to the study
report that
summarizes the findings from the
University of
Pittsburgh Medical Center, protocol
003. In this
protocol a total of 68 patients were
studied in two
phases.
First, 10 patients were enrolled in an
open phase and treated with cyclosporine
inhalation
solution. Then the total of 58 patients were
randomized to cyclosporine inhalation
solution
120
which contains propylene glycol as a
vehicle or
propylene glycol vehicle alone.
From here I will refer to
these groups as
cyclosporine inhalation solution as CyIS
or
propylene glycol group as PG. Twenty-six patients
received CyIS and 30 patients received
propylene
glycol vehicle. This was administered by
inhalation with a nebulizer. It should be noted
that all patients received concurrent
tacrolimus-based systemic
immunosuppressive
therapy.
Study ACS002 was the name that Chiron
Corp. gave to the study report that
summarizes the
findings on adverse events in 70
patients selected
from seven open-label studies conducted
at UPMC. I
will refer to these study reports
later. Also, I
will refer to the ACS001 study and study
ACS002 to
avoid confusion.
The rest of my discussion will
focus on
study ACS001, and the primary objective
of this
study was to determine if cyclosporine
delivered to
the lung allograft by inhalation
prevents the
121
development of acute cellular rejection.
As you can see from this
slide, the study
failed to show superiority of
cyclosporine
inhalation solution over PG
vehicle. The mean
number of acute rejections of grade 2 or
higher per
patient was 1.3 in the cyclosporine arm
and 1.2 in
the PG arm. The median number of acute rejections
grade 2 or higher was 1 in both
arms. Therefore,
the study failed the primary endpoint.
However, we noted that the
sponsor
reported a difference in mortality and
obliterative
bronchiolitis between the two arms. In the study
report and database OB was reported as 1
for its
presence or 0 for its absence. No additional
histopathology information was provided. The
specimens for diagnosis of OB were
obtained by
transbronchial biopsies.
The reporting mortality was 12
percent in
the CyIS arm and 40 percent in the PG
arm. The
applicant noted that this represents a
79 percent
decrease in risk for mortality in this
specific
population. The reported rate of bronchiolitis
122
obliterans or death was 19 percent in
the CyIS arm
and 60 percent in the PG arm, with a
reported p
value of 0.003. It should be noted that this
difference is mostly driven by the
difference in
mortality.
In study ACS001 all patients
were followed
up for three years after enrollment, and
thereafter
were followed up to document
mortality. At the
time of the study end when the last
patient
completed two years of aerosolized
treatment in
August, 2003, the mortality was 12 percent
in the
cyclosporine arm and 40 percent in the
PG arm.
Follow-up data obtained through July,
2004 was
submitted in the NDA and it showed
mortality of 19
percent in the cyclosporine arm and 50
percent in
the PG arm. Additional information submitted in
the safety update in May, 2005 showed a
mortality
rate of 31 percent in the CyIS arm and
50 percent
in the PG arm.
At the time the NDA was
submitted to the
agency, the limitations of the study
were known to
us.
These included the following:
This was a
123
single-center Phase II study. There was a small
sample size. The study intended to enroll 136
patients. The case report forms were created
retrospectively. Therefore, some important
recipient and donor implementation was
not
captured. Some data were not systematically
collected, for example, prospective
routine
transbronchial biopsies. Some data were not
available, for example, some donor
characteristics
or information on management on acute
rejection
episodes grade 2 or higher that appeared
prior to
enrollment.
FDA concerns included the lack
of effect
on the primary endpoint. We also shared the
sponsor's concerns that the study may
have become
unblinded. For example, patients at UPMC with
identification numbers ending in letters
B or C
received cyclosporine inhalation
solution, while
those patients with numbers ending in A
or D
received PG.
This may have allowed the
investigators to identify if a given
patient was
receiving propylene glycol or
cyclosporine
124
inhalation solution.
Protocol documentation was
limited.
Chronic rejection or survival were not
designed as
the primary endpoints. Furthermore, the protocol
for this study did not specify how
secondary
endpoints would be analyzed, and there
was no
pre-specified statistical analysis plan.
There were nine protocol
amendments. The
study was stopped before completing
enrollment.
There were various protocol violations
and there
was no stratification by risk factors
important for
chronic rejection or mortality. We can give an
example such as double lung versus
single lung.
Despite randomization, there were
imbalances in
baseline characteristics.
Now I would like to describe
our approach
to the analysis of chronic rejection and
mortality
in study ACS001. Acute rejection is considered a
major risk factor for the development of
chronic
rejection or obliterative bronchiolitis,
and a
number of acute rejection episodes
experienced
early after transplantation are
considered to have
125
a significant impact on the subsequent
development
of OB.
Even though acute and chronic
rejection
represent different histopathology and
pathophysiology, there is general
consensus that
the frequency, intensity and duration of
acute
rejection episodes are correlated with
subsequent
development of obliterative
bronchiolitis.
Strong evidence suggests that
acute
rejection is the principal cause of
chronic
allograft dysfunction. However, the role of other
immunologic and non-immunological
factors have to
be considered. Therefore, we examined the
following data on acute rejection,
obliterative
bronchiolitis histological findings,
FEV1 and BOS
clinical manifestations of the disease,
and
mortality as a clinical outcome.
Obliterative bronchiolitis is
an important
cause of mortality after the first year
from
transplantation, accounting
approximately for 30
percent of deaths. FEV1 is the best surrogate
marker available for OB, and has been
proven
126
successful in describing--very
important--the
pattern of lung function decline,
described as
acute or chronic BOS onset; the
identification of
the main risk factors for BOS; and the
extent and
the rate of progression of OB.
The International Society of
Heart and
Lung Transplantation subcommittee has
recommended
that the slope of serial FEV1
measurements over
time, before and after a therapeutic
intervention,
should be used to compare treatment
responses.
Therefore, if chronic
rejection is
effectively prevented, we should expect
to observe
an evident therapeutic effect on FEV1
and BOS.
Obliterative bronchiolitis, as defined
in the study
report, was documented by transbronchial
biopsies
and was found in 12 percent of the CyIS
patients
and 30 percent of the propylene glycol
patients.
Now there are three points
that I would
like to make regarding FEV1. First, as you can
see, FEV1 values pre-enrollment, that
is, after the
transplantation but before randomization
to the
cyclosporine or PG arms, were not
available in 40
127
percent if the patients. This data is shown in the
first row. Second, by 3 months there is FEV1 data
on essentially all patients, all 26
patients in the
CyIS arm and 26/30 in the PG
patients. Third, you
will notice that there is a difference
in mean FEV1
values between the 2 groups. At all point times
the mean FEV1 values are higher for the
CyIS group
as compared to the PG group. Even before treatment
assignment higher mean FEV1 values were
observed in
the cyclosporine inhalation group. This difference
may be attributable to the greater
number of double
lung transplants that were performed in
this group,
which we will discuss later in greater
detail.
Here is a graphical
presentation of the
data shown in the previous slide. You can see that
even though the FEV1 values in the
cyclosporine
inhalation group are higher than the PG
group, the
yellow line below, the two curves are
essentially
parallel. Therefore, it does not appear that
cyclosporine inhalation solution has an
effect on
FEV1.
Complete FEV1 values were not
available so
128
BOS, bronchiolitis obliterans syndrome,
as defined
by the International Society of Heart
and Lung
Transplantation could not be calculated
using these
criteria. Therefore, an alternative definition of
BOS, defined by the sponsor and
qualified by an
independent investigator was used.
As seen in this graph, the
time to BOS
between the 2 arms is similar, and the
log-rank b
value is 0.214. This also indicates that the
cyclosporine inhalation solution has no
effect on
BOS.
Patients who died without double-blind of
BOS, as defined by the applicant, were
censored at
the time of the last follow-up for BOS.
We observed a difference in OB
and
mortality at the end of the study in
August, 2003.
OB was present in 12 percent in the
cyclosporine
inhalation solution versus 30 percent in
the PG
group.
Mortality was 12 percent in the CyIS arm
versus 47 percent in the PG group. No difference
was observed in acute rejection, FEV1 or
BOS. As a
clinician, FEV1 values are really,
really
important. Questions like "how are you
breathing"
129
are really important questions.
The association between acute
rejection
and chronic rejection and the effect on
patients
and graft survival is well documented in
registry
and published literature. Acute rejection is a
major risk factor for the development of
chronic
rejection or obliterative
bronchiolitis. In light
of the strong association between acute
rejection
and chronic rejection, the difference
observed in
OB was not expected in the absence of differences
in acute rejection, FEV1 or BOS, and
this warrants
further exploration.
Therefore, we asked the
question is the
mortality difference between
cyclosporine
inhalation solution and PG in the absence
of
differences in acute rejection, FEV1 or
BOS due to
treatment effect or could other factors
account for
this difference? For example, difference in
baseline characteristics of donors and
recipients
between the study arms, or other factors such
as
study conduct.
I want to remind you that
there was no
130
difference in acute rejection grade 2 or
higher at
randomization to the drug or to the
placebo arm.
In contrast, there is a clinical and
meaningful
difference in acute rejection grade 2 or
higher
before treatment assignment. Thirty-one percent in
the CyIS arm and 42 percent in the PG
patients had
grade 2 or higher acute rejection prior
to
enrollment. Although data were incomplete,
approximately 40 percent of the CyIS
allografts and
50 percent of the PG allografts were
colonized with
bacteria or fungi. So, this data is incomplete but
I still think it is worth mentioning
it. So, if we
assume that patients who had acute
rejection grade
2 or higher prior to enrollment received
some type
of steroid treatment or any other
treatment
augmentation, they could be predisposed
to
infectious complications such as
pneumonia or
sepsis.
Now I will discuss other
imbalances in
patient characteristics. There is well documented
association between the type of lung
transplant and
survival. In this study there is an imbalance in
131
the number of single lung and double
lung
transplants between the two arms. Single lung
transplants were done in 58 percent in
the CyIS arm
and 80 percent of the PG patients. Conversely,
double lung transplants were done in 42
percent of
the CyIS patients and 20 percent of the
PG
patients. This difference is statistically
significant at a level of 10
percent. FEV1
pre-enrollment was lower in the PG arm
and may be a
reflection of more single lung
transplants in this
group.
The imbalance between single
and double
lung transplant is important. The literature and
registry data show an advantage for
long-term
survival and freedom from BOS in double
versus
single lung transplants. Single lung
transplantation is associated with lower
exercise
tolerance, poorer pulmonary mechanics,
and higher
infectious complications such as
pneumonia.
The International Society of
Heart and
Lung Transplant registry data show that
the there
is a difference in survival between
single and
132
double lung transplant patients. The half-life of
double lung transplant patients is 5.3
years, as
shown in the top line, while the
half-life for
single lung transplants is 3.9
years. The average
survival is shown in green in this graph.
As noted before, the
information on donor
characteristics was incomplete. Therefore, we
examined the data available that was
informative
about the state of the donor lung, and
we noted a
difference in donor inotropic
support. Fifty
percent of the donor lung
transplantations to the
CyIS patients and 83 percent of the
donor lung
transplantations to the PG arm came from
donors
that received inotropic support.
PaO2/FiO2 ratio is an
indicator of the
severity of acute lung injury and it is
useful to
indirectly assess the degree of ischemic
re-perfusion injury sustained by an allograft.
PaO2/FiO2 ratio of greater than 200
percent
indicates limited alveolar damage and
gas exchange.
Another difference between the
two arms
was the time in the ICU. While most of the
133
patients stayed in the ICU for less than
10 days, 4
percent in the cyclosporine arm patients
and 20
percent in the PG patients were in the
ICU for more
than 10 days, and this is kind of
important in a
single center where the criteria for
keeping the
patients in the ICU pretty much remained
the same
The other important thing is that it
will reflect
how the patients are in terms of degree
of severity
of the disease. Patients are not allowed to go out
of the ICU if there is something that
still needs
to be taken care of. So, it is a good reflection
of the degree of sickness that these patients
have.
PaO1/FiO2 ratio is an
indicator of the
ability of the lung to perform adequate
gas
exchange, and it is useful to indirectly
assess the
severity of acute allograft injury. The baseline
PaO2/FiO2 ratio on ICU admission was
worse in the
PG group, suggesting a major degree of
ischemic
re-perfusion injury in these
allografts. Also,
perioperative renal dysfunction was in 4
percent in
the cyclosporine inhalation solution and
13 percent
in the PG patients. Prolonged ICU stay, inadequate
134
gas exchange and perioperative renal
dysfunction
are factors that reflect a more severe
condition
after surgery.
We also looked at the time to
the first
pneumonia. As noted, there were more cases of
pneumonia in the PG arm and this was
within the
first one to two months of the
study. The outcome
in patients with these pneumonias is
summarized in
the next slide.
A large number of patients in
the PG arm
had early pneumonias and there was a
strong
relationship between pneumonia and
death. The
relationship is not surprising given
what we know
about the causes of death after lung
transplantation. The occurrence of these early
pneumonias is not likely to be related
to any
treatment effect but may be related to
baseline
donor and recipient characteristics or
other events
which occurred prior to enrollment. These events
include but are not limited to episodes
of acute
rejection requiring additional
immunosuppressive
therapy or microbial colonization of the
graft.
135
I would like to underline that
early
pneumonia may lead to histopathological
findings
compatible with obliterative
bronchiolitis. This
has been documented to be a risk factor
for the
development of obliterative
bronchiolitis. There
were five patients in cyclosporine inhalation
solution arm and two patients in the PG
arm who
developed pneumonia in the first
month. By two
months there were an additional three PG
patients
with pneumonia. Of these patients that developed
pneumonia, 2/5 died in the cyclosporine
arm and
7/13 in the PG arm; 1/5 developed OB in
the
cyclosporine inhalation solution and
7/13 in the PG
group; and BOS was observed in 3/5 in
the CyIS arm
and 3/13 in the PG arm.
I want to make two
observations. There is
a strong association between early
pneumonia and
risk of death. Second, early pulmonary infections
and early acute rejection episodes are
well
recognized risk factors for the
subsequent
development of chronic rejection.
This table show the primary
causes of
136
death by July, 2004. Three patients in the
cyclosporine inhalation solution arm and
seven
patients in the PG group died of
infections,
pneumonia or sepsis. In the CyIS arm one patient
died of graft failure and in one patient
the cause
was unknown. In the PG group two patients died of
OB; one patient died of pulmonary
embolism and
another from congestive heart failure,
and one from
lung cancer. There were three patients in which
the cause of death was unknown. The distribution
of causes of death is consistent with
registry data
where infections remain the major cause
of death
during the first year after
transplantation while
chronic rejection begins to become an
important
cause of death after one year, as seen
in table 3,
reference 1 in your background package.
Autopsy results--from the
available data
in the application CRFs, narratives and
data sets
we learned that some patients who died
had autopsy
performed. In the cyclosporine inhalation solution
arm one patient had autopsy and OB was
not
reported. In the propylene glycol arm 15 patients
137
died and there were six autopsies. In two of these
OB was reported and four of them died of
infection,
and there was no OB reported out of the
six
reports.
The protocol specified that
patients
should receive treatment for two
years. The dose
should be titrated from 100 mg to 300 mg
for the
first three days of treatment, then
daily dosing up
to three consecutive days with the maximum
tolerated dose, and thereafter three
times weekly
dosing for two years. There was a lot of
variability in individual patient dosing
in this
trial.
This table shows the number of
doses
received by patients. The protocol dosing schedule
was not followed in many patients. In fact, six
CyIS and five PG patients received less
than 25
doses, as you can see circled in this
slide. The
large variation in the number of doses
received
makes it difficult to establish a
relationship
between the specific treatment regimen
and the
improvement in survival.
138
Six cyclosporine inhalation
solution
patients who received less than 25 doses
are shown
on this table. Two patients received a single
dose; others received 3, 12, 13 and 24
doses
respectively. The doses show that not all patients
succeeded in titrating up to 300
mg. Five out of
these six patients experienced adverse
events
directly related to the administration
of the
cyclosporine inhalation solution, and three
patients discontinued due to adverse
events, and
three additional patients withdrew
consent. We
noted, however, that all six patients
survived and
all are included in the mortality
calculations as
cyclosporine inhalation solution
successes.
There were five patients in
the PG arm who
received less than 25 doses and, as can
be seen,
four/five died. Could these be attributable to the
lack of cyclosporine inhalation
solution? All
these deaths are included in the
mortality
calculation as PG failures.
In addition to the 3
cyclosporine
inhalation solution who withdrew consent
after
139
receiving 1, 3 and 13 doses, 3
additional patients
withdrew consent--these are the last 3
rows in this
slide--1 at 4 months and 2 others at 20
months.
The right-hand column shows that 2 of
these 3
additional patients survived.
At this point I would like to
turn the
podium over to Dr. Cavaille-Coll to
discuss our
safety considerations and give our
conclusions.
Safety Considerations and
Conclusions
DR. CAVAILLE-COLL: Good morning.
We are
in general agreement with the applicant
that the
systemic safety profile of cyclosporine
after oral
or intravenous administration is well
characterized
and that the amount of systemic exposure
to
cyclosporine, meaning what was deposited
in the
lung and entered in the bloodstream
before being
eliminated, was not associated with
detectable
increases in systemic toxicity. There is more
limited information on the safety of
cyclosporine
when administered by inhalation in a
propylene
glycol solution.
As you have heard, propylene
glycol is
140
classified as an additive that is
generally
recognized as safe for use in food,
mainly through
studies using oral and dermal
exposure. It is used
to
absorb extra water and maintain moisture in
certain medicines, cosmetics or food
products. It
is a solvent for food colors and
flavors. However,
information on the inhalation toxicity
of propylene
glycol is more limited. There is no approved
product for inhalation containing nearly
100
percent propylene glycol such as this
product.
The applicant has submitted
some
preclinical safety data, including a
28-day study
in dogs and a 28-day inhalation study in
rats. The
28-day inhalation study in dogs
demonstrated lung
irritation, alveolar and interstitial
inflammation
in all cyclosporine dose groups and the
vehicle
control.
Laryngeal inflammation with ulceration
was seen in the mid-dose group
males. Inflammatory
cell infiltrates, lymphocytes, plasma
cells,
monocytes were seen in the control and
treated
group as well. The dog studies did not contain a
sham control. Thus, this confounded the separation
141
of the extent of pulmonary toxicity due
to
cyclosporine versus that of the propylene
glycol
vehicle.
No additional cyclosporine inhalation
toxicity was observed in the
animals. Dose levels
in the dogs were limited, however, by
the maximum
feasible dose. However, serum cyclosporine levels
in
the high dose group exceeded the human exposure
by 2.5-fold.
Again, there were also studies
that were
done in rats which showed similar
findings, except
that the doses in rats did exceed about
80-fold the
human exposure and there was evidence of
increasing
toxicity with increasing doses of
cyclosporine.
The rat studies did include an air
control and did
show that even in the propylene glycol
group there
were findings that were not present in
the sham
control animals.
I would like to address now
the clinical
safety.
In the usual safety review we would look
at the rates of adverse events, the
grade of
severity, the duration of the events and their
reversibility, as well as the temporal
relationship
142
to dosing with the study drug. Collection of such
information is facilitated by the use of
prospectively designed case report
forms. The
latter often provide another very useful
source of
safety information in the form of
handwritten
comments by the investigators on the
margins of the
pages of the case report forms. Such forms and
comments were not available and it is in
the
context of these limitations that we
must evaluate
the safety of this product. Evaluation of safety
in
this fragile population receiving systemic
immunosuppression and numerous
medications is
admittedly complicated.
There are no prospectively
designed case
report forms to guide the systematic
collection of
safety data throughout the conduct of
the study
including but not limited to the use of
concomitant
medications used to prevent or treat the
complications associated with the
administration of
study drug. Clinical safety data was collected
retrospectively from source materials
from one
double-blind, controlled study and a
number of
143
small open-label, uncontrolled studies
at the
University of Pittsburgh Medical Center.
Comparative safety data is available on
only 26
randomized subjects in study ACS001, or
36 subjects
that include the first 10 non-randomized
subjects
from the study. Additional non-comparative safety
data was obtained in report ACS002 by
pooling data
from seven open-label, uncontrolled
studies that
enrolled 70 lung transplant recipients
who were
receiving similar tacrolimus-based
systemic
immunosuppression.
Subjects in study ACS001 were
titrated in
a double-blind fashion to a maximum
tolerated dose
not to exceed 300 mg or the propylene
glycol
control equivalent. That dose was then to be
administered three times a week for up
to two
years.
As mentioned earlier, there was a great
variation in dose, 100 mg to 300 mg per
day, the
number of doses administered and,
consequently,
duration of exposure. I think we have seen those
slides before. Subjects also received per protocol
premedication with aerosolized lidocaine
and
144
bronchodilators to improve tolerance.
This slide comes from the
integrated
summary of safety and lists basically
the adverse
events that occurred with a statistical
significance of greater than 10
percent. I think
we are in general agreement with the
applicant's
description of the safety data they were
able to
collect.
We do note that there seemed to have been
more respiratory, and thoracic adverse
events in
the cyclosporine group compared to the
propylene
glycol group. In all these categories, of course,
as I mentioned before, the significance
was greater
than 10 percent. As in the 28-day preclinical
animal studies, there was a sham treatment
group to
help discern the potential contribution
of inhaled
propylene glycol to the respiratory
tolerability in
both treatment groups. Here we do see that more
events occurred in the cyclosporine
group. These
findings in general are consistent with
the
respiratory safety findings that were
found in the
28-day preclinical animal studies.
Another thing we look at when
we are
145
evaluating safety is the
discontinuations and
withdrawal of consent. Although a greater
proportion of subjects in the propylene
glycol
group, 33 percent, were reported to
discontinue
study drug due to an adverse event,
other than
death, than in the cyclosporine group,
15 percent,
this comparison must be interpreted with
caution.
Six patients in the
cyclosporine group, or
23 percent, were reported to have
discontinued due
to withdrawal consent compared to none
in the
propylene glycol group. Further examination of the
individual case report forms revealed a
number of
respiratory adverse events associated
with the
study drug administration which could
have
influenced their continued willingness
to
participate in the study. Taken together, a
similar proportion of subjects
discontinued study
drug due to adverse events or
tolerability in the
propylene glycol group and the
cyclosporine group.
We also have some
non-comparative data
that was presented in report ACS002 from
a pool of
70 lung transplant recipients. Again, these
146
represent a variety of lung transplant
types,
mostly patients with refractory acute
rejection
and/or OB who were treated with
cyclosporine
inhalation solution in seven open-label,
uncontrolled studies at UPMC. They were also
receiving systemic tacrolimus-based
immunosuppression. These, again, represent an
experience of a wide range of dosing and
duration
of treatment, which is really very
difficult to
interpret. Patients were generally administered
the maximum tolerated dose which was
individualized
and depended on the characteristics of
the patients
and their response to medication.
In summary, the overall safety
database is
smaller than usually expected in a
commercial
application. Respiratory adverse events were
common despite premedication and limited
the
maximum doses used and the durations of
the
treatment. Data available in the study report and
case report forms did not allow us to
fully
evaluate a temporal relationship between
study drug
administration and particular adverse
events.
147
Ultimately the acceptability of the
safety
information in this NDA must be weighed
against the
degree of certainty of the potential
clinical
benefit.
These are briefly our
conclusions: We
have a single small study. There is no effect on
the primary endpoint of acute rejection
or on
measurements of pulmonary function of
FEV1 or
bronchiolitis obliterans syndrome. There are
differences observed in mortality and
bronchiolitis
obliterans. There also important imbalances in the
donor/recipient baseline
characteristics. There
are also variable causes of death, most
with no
evidence of bronchiolitis obliterans
syndrome.
There is variable dosing in both groups,
and we
have limited safety information.
At this point, I would like to
turn the
podium over to Dr. Jyoti Zalkikar who
will present
the statistical perspective of this
review. Thank
you.
Statistical Evaluation
DR. ZALKIKAR: Hello.
My name is Jyoti
148
Zalkikar. I am the statistical reviewer for the
application for Pulminiq, which is the
trade name
for cyclosporine inhalation solution.
I will be focusing on the
efficacy of the
product during this presentation. As you know, the
efficacy of this product is based on
just one small
Phase II, single-center study conducted
at the
University of Pittsburgh Medical
Center. This was
not designed as a confirmatory
study. The planned
sample size was 136 patients as per the
applicant's
study report. The investigators did not use case
report forms during the conduct of the
study.
These were generated retrospectively by
the
applicant. Also, there was no prospective
statistical data analysis plan and no
formal
stopping rules.
The study began in 1997. The first 10
patients were who were enrolled received
cyclosporine as part of the open-label
pilot phase.
The next 58 patients were randomized to
either
cyclosporine with propylene glycol as
vehicle or
propylene glycol alone. Two of these patients did
149
not receive any dose of the study
medication and
were excluded from all analyses. The applicant
says that the enrollment was stopped in
August,
2001 at 68 patients. The study was vehicle
controlled. As you know, all subjects received
systemic immunosuppression.
Now I will briefly go over some aspects
of
the study design. As per the initial design, the
patients were to be enrolled from 7-21
days after
the transplant. But later a 22-42 days window was
added to speed up the enrollment. Three patients
out of the total of 56 were enrolled
past 42 days
after their transplant. Randomization was
stratified by the enrollment window and
CMV
mismatch. Donor positive/recipient negative was
defined as a mismatch and all other
combinations of
donor and recipient CMV status were
called a match.
Two out of 56 patients were incorrectly
stratified.
Patients were to be on treatment daily
for the
first 10 days and 3 times a week
thereafter for a
total of 2 years. Thirty out of the 56 patients
discontinued treatment early. All 56 patients were
150
followed for at least 33 months in the
data
submitted with the NDA. That is the database I
will be using in this presentation.
Evaluation of survival and
chronic
rejection were not the primary
objectives in the
study.
The prespecified primary objective was
superiority over propylene glycol in
terms of the
rate of acute rejection. The study failed to
achieve that objective. In fact, the mean number
of acute rejections was slightly higher
in the
cyclosporine group.
Here is the graph for survival
distributions in the two arms. This graph is based
on all the data submitted with the NDA
and,
therefore, is slightly different from
the one based
on the study end date that was
previously
presented. The 24-month line indicates the end of
protocol specified treatment
period. When we saw
this dramatic survival difference at the
pre-NDA
meeting we were excited and encouraged
the
applicant to submit the application.
When the NDA was submitted,
given that the
151
trial had failed on the primary endpoint, the
challenge for the review team was to
determine if
the observed survival difference was due
to
cyclosporine inhalation solution.
One of the first things we
found was the
baseline imbalance between the two
treatment
groups.
Although randomized, due to a small sample
size, the study failed to benefit from
randomization. Several baseline factors that
clinicians consider to influence patient
survival
are not balanced between the two
groups. This
means that the two groups are not
comparable for
evaluating survival or chronic
rejection.
Here are the factors that show
imbalance.
All of these are statistically
significant at the
10 percent level, with the exception of
grade
2-plus acute rejection prior to
dosing. Dr.
Hernandez has discussed the clinical
importance of
the influence of these factors on
patient survival.
All patients in the study had at least
one of these
risk factors.
Here is the nature of the
imbalance. The
152
yellow bars represent patients in the
propylene
glycol group and the red bars represent
patients in
the cyclosporine group. As you can see, the
majority of the cyclosporine patients
are on the
left, with two or less risk factors,
whereas the
majority of the propylene glycol
patients are on
the right side, with three or more risk
factors.
This can occur in randomized trials, and
more so in
the trials with small sample sizes such
as this
one.
That is why it is important to prespecify the
primary endpoint so that appropriate
stratification
may variables can be used at
randomization to
control for at least some factors known
to
influence that primary endpoint. The study could
not accomplish that since the primary
endpoint was
not survival or chronic rejection.
In this situation statistical
methods can
be used to adjust for these imbalances,
but using
these methods to adjust for factors
individually
one at a time is not appropriate as the
groups are
still not comparable due to imbalances
with respect
to the other factors. For simultaneous adjustment
153
for all the factors the sample size is
too small in
the study.
We also considered methods
that in some
situations allow us to handle a large
number of
factors such as propensity scores, but
these have
limitations and there are underlying
assumptions
that need to be validated.
So, given that the trial
failed to show a
difference in terms of the primary
endpoint of
acute rejection, given the absence of
prospectively
defined analysis plan including
statistical details
such as multiplicity adjustments and
stopping
rules, and given the baseline imbalances
with
respect to many factors, the validity of
any
further inferential statistical analyses
on the
data from this trial is questionable and
lends to
caution.
Now I would like to draw your
attention to
another problem the review team
faced. The study
was designed as a double-blind study
but, as you
have heard, a code A, B, C, D was added
to the
patient number to aid pharmacy in
preparing study
154
medication. Patients with A and D in their patient
number received propylene glycol and
patients with
B and C received cyclosporine. The applicant
stated in their study report that this
may have
revealed a treatment assignment to the
investigators. This fact, together with the
retrospective nature of the study, makes
it
vulnerable to the introduction of bias,
as
inadvertent as it may be. For example, if the
investigators knew that patients were
getting
propylene glycol they may have adjusted
systemic
immunosuppression to compensate for the
lack of
cyclosporine, inadvertently predisposing
the
patients to infections like pneumonia
and sepsis
which were among the leading causes of
death in the
study, as discussed by Dr.
Hernandez. Detection of
the presence and magnitude of bias is
difficult,
but its possibility certainly lends to
caution when
interpreting the study results.
Now I would like to illustrate
how small
perturbations in the assignment of the
subjects in
a
small study such as this one can change the
155
picture dramatically. Here is a graph of survival
distributions again. Please notice that there were
three early deaths in the control arm due
to
pneumonia and sepsis. These patients received very
few doses of the study medication. The review team
felt that these deaths were not expected
to be
influenced by the treatment
assignment. There are
eight different possible ways to assign
these three
patients to the two groups. The current assignment
is the only assignment that results in
statistical
significance in terms of the p
value. For
assignments in which one of these
patients gets
assigned to the cyclosporine arm the p
value is
over five percent. For assignments in which two of
these patients get assigned to the
cyclosporine
arm, the p value is over 12 percent. And one
possible assignment in which all 3
patients get
assigned to the cyclosporine arm uses
this picture.
This is completely different and no
longer shows a
survival difference. This illustration shows that
due to small sample size the results of
the study
are extremely sensitive to very small
perturbations
156
in the assignment of the patients.
Now let's look at the
discontinuations.
More than 50 percent of the patients in
the study
discontinued for various reasons. Please note that
there were six patients who withdrew
consent. They
all were in the cyclosporine arm. One of them was
crossed over to the open-label
cyclosporine study.
All of these patients were followed for
survival
even after the withdrawal of consent.
This table shows the number of
doses
received by the patients in the
study. As you can
see, there is large variation in the
amount of
treatment received. Please note that there are 11
subjects who received less than 25
doses, which is
less than 10 percent of their protocol
specified
treatment. Four of these 11 subjects received only
one or two doses. As discussed by Dr. Hernandez,
the review team felt that this short
duration of
treatment would not be expected to
result in
significant benefit to the patients.
Therefore, we conducted
sensitivity
analyses to assess the impact of the
data from
157
these patients. In the first analysis we would
exclude all 11 patients. Here are the results.
The survival difference is no longer
statistically
significant.
In the second analysis we will
treat the
six cyclosporine patients as a control
patients.
So, we will have 20 patients in the
cyclosporine
arm and 36 patients in the control
arm. And here
are the results. Once again the survival
difference is no longer statistically
significant.
In the third sensitivity
analysis we
define treatment failure as death or
discontinuation within 25 doses due to
either
adverse events or withdrawal of consent,
and we
analyzed time to treatment failure. Here are the
results.
These results do not show a difference
between the two arms in terms of time to
treatment
failure.
I recognize that all these
sensitivity
analyses are subject to criticism, but
the point is
that the data from subjects who
discontinued very
early and received very little treatment
has a big
158
impact on the study results, as shown by
these
sensitivity analyses. This certainly lends to
caution when interpreting the study
results.
Now moving away from survival,
I will
briefly go over some results related to
lung
function. Here is the descriptive data on FEV1
previously discussed by Dr.
Hernandez. Please
recall that data on pre-enrollment is
missing for
43 percent of the patients in the
study. But the
available data shows lower FEV1 values
on average
for the propylene glycol arm at
pre-enrollment
compared to the cyclosporine arm. This may very
well be the effect of observed baseline
imbalance
with respect to the type of lung
transplant among
other factors.
Here is the graph of these
numbers. The
lines for the cyclosporine and the
control arm are
parallel over the length of the study,
showing that
treatment with cyclosporine had no
effect on lung
function in terms of FEV1. This, together with the
observed survival difference, raises an
interesting
question. Is it plausible that the patients in the
159
PG arm were predisposed at
pre-enrollment due to
compromised lung function to a higher
probability
of getting serious infections such as pneumonia
and
sepsis and then dying from them? If so, how can
one adjust for this phenomenon to get an
unbiased
estimate of the treatment effect in
terms of
survival in this small data set, where
nearly 43
percent of the pre-enrollment FEV1
values were
missing?
Here is a graph of time to BOS
as defined
by the applicant when patients who died
without
diagnosis of BOS were censored at the
time of last
follow-up for BOS. This analysis is subject to
criticism of informative censoring but
it helps to
assess the effect of cyclosporine on BOS
in the
absence of a survival difference. Again,
consistent with the analysis of FEV1,
the
difference between the two arms is not
significant.
Please recall that there were
10 patients
who received cyclosporine as part of the
open-label
phase of the study. There was very limited
information on these patients but here is some
that
160
was available. Please note that these 10 patients
showed remarkable similarity with the
propylene
glycol group in terms of 2 important factors,
the
type of lung transplant and emphysema as
the
underlying condition requiring lung
transplant.
So, we overlaid the survival
curves from
these 10 patients. Please keep in mind that the
patients were not randomized to this
group so
direct statistical comparison is not
possible. But
given that, this group does not seem to
support the
survival difference between the
randomized
cyclosporine group and the propylene
glycol group.
In summary, the study failed
to
demonstrate effect in terms of the
prespecified
primary endpoint of acute
rejection. Several
baseline factors considered to affect
patient
survival were not balanced between the
two arms.
There were concerns that blinding may
not have been
adequately preserved in the conduct of
the study.
The study is also highly sensitive to small
perturbations in the assignment of
patients.
Patients who discontinued early and
received very
161
little treatment had a big impact on the
study
results. The study failed to
demonstrate a
treatment effect in terms of FEV1 and
BOS, and
additional data were not supportive.
So, the question still remains
is the
observed survival difference between the
two arms
in the randomized portion of the trial
due to
cyclosporine inhalation solution or due
to other
factors?
This concludes the FDA presentations.
Questions from the Panel
(Continued)
DR. SWENSON: We are ahead of time here
and there were a number of questions
that were left
from the previous session. I believe we could
probably open it up to both the
applicant and the
FDA.
So, let me begin with Dr. Sampson.
DR. SAMPSON: I had a number of questions,
this one to both the applicant and the
FDA. It is
still not clear to me about the actual
data that
was collected in the original UPMC study
and the
data that was collected
retrospectively. I
understand the safety data was primarily
retrospective but, for example, was the
designation
162
of
BOS at a specific time point designated in the
charts as occurring at that time point,
or was that
made by reviewing the charts and saying
yes/no
there was BOS at that time point? I think I have
asked that question appropriately.
The other thing is when the
charts were
reviewed, were these reviewed blinded to
treatment
or did the reviewer know the treatment
the patient
was on?
I have several other questions in addition
but just give me some information on
that, please.
DR. CAPRA: You are correct that the
safety data was collected
retrospectively. The
efficacy data consisted of survival, the
results of
the histology, the FEV1 data which was
used to
calculate BOS. The survival data was obtained on
CRFs but retrospectively but it was
confirmed
through the source data with the autopsy
reports.
The results of the transbronchial
biopsies were
sent electronically from the University
of
Pittsburgh to Chiron. So, we didn't collect that
information on case report forms. They basically
maintained the database throughout the
study and
163
they sent that database electronically
to Chiron.
Similarly, the FEV1 data and their lab
data as well
were collected on a central database at
the
University of Pittsburgh Medical
Center. Those
data were sent electronically to Chiron
as well.
So, none of the efficacy data, with the
exception
of the actual survival days, were
collected on
retrospective case report forms.
DR. SAMPSON: There is something there
that says 20 percent reduction in FEV1
in the
presence of no other clinical
symptoms. How would
that be determined just purely from
FEV1? Is it
just such an automatic algorithm that
you don't
look at other clinical symptoms, or was
there a
judgment made on that?
DR. NOONBERG: Well, it is a definition
that was determined by a consensus group
from the
International Society of Heart and Lung
Transplantation and the definition of
BOS grade 1
or higher--and there is successive
grading--is 20
percent decline in FEV1 from a
post-transplant
maximum.
Now, post-transplant maximum generally
164
doesn't occur for three to six months
due to
postoperative factors. So, the missing early
FEV1's are not expected to influence at
all the
designation of BOS because all patients,
unless
they die early, increase their FEV1 up
to a period
of time, generally an average of about
three to
four or five months. Then that post-transplant
maximum, a rolling average, is used as
their
baseline from which you determine a 20
percent
decline.
So, that is how the definition is and
that is not set by us; that was a
definition
proposed by the International Society
for Heart,
Lung Transplantation. The 20 percent was just
programmatically defined and then we
reviewed the
cases to determine that there were no
other
clinical causes which were defined as
acute
rejection, concurrent pneumonias or
other clinical
causes that would impair lung function.
DR. SAMPSON: Again, were all the
retrospective chart reviews done by
people--was it
possible to blind them to the treatment assignment,
or did they know the treatment
assignment when they
165
did the chart reviews?
DR. NOONBERG: Treatment assignment was
known at the time of chart review.
DR. SAMPSON: I would like to switch gears
and ask a different question that I have
been
puzzling a little bit over, and that is
the study
design stratified by two variables, the
CMV
match/mismatch and the start date. I don't recall
seeing survival data--there is possibly
something
for the CMV mismatch, but does the
sponsor have
survival data done by strata? In particular, one
of the standard questions when you have
a
stratification factor in any clinical
trial is the
question of whether or not there is
strata by
treatment interaction; whether or not
the survival
is comparable across strata; is it
something that
is poolable? And I have not seen any demonstration
of that, I don't think, for this data
either for
CMV match/mismatch or the start post day
21. I was
hoping you might have slides on that
some place as
backup.
DR. CAPRA: Can we see slide CE-18,
166
please?
Unfortunately, I don't have the analysis
here just limited to two stratification
factors but
the results are consistent. But here, in the last
row of this slide, we show the survival
analysis
where the 2 stratification factors, as
well as the
other known risk factors that were
imbalanced and
were agreed to by both Chiron and the
FDA, were
included, namely, single versus double
lung
transplant, prior acute rejection, and
primary
diagnosis. The results show a significant value of
0.032 with a consistent p value. The
results aren't
shown here but the results are very
similar when
you just limit it to CMV mismatch and
the primary
diagnosis.
DR. SAMPSON: I did some sketching of my
own.
I don't have it with me for the CMV
match/mismatch. I think you have some of that
data, don't you, in the document? It is page 17 in
the document.
DR. CAPRA: Can we put slide BD-5 up,
please?
You are correct, in the briefing document
there is a table showing the number of
deaths by
167
the CMV match and mismatch. For those who had a
CMV match, 51 percent of those--I am
sorry, this is
on chronic rejection-free survival. Can we have
BD-3?
For survival data, 39 percent of the placebo
subjects who had a CMV match died versus
14 percent
of the cyclosporine. Among the mismatches, 71
percent of placebo, 0 percent for
cyclosporine.
The numbers are small but what we are
not seeing is
we are not seeing the survival effect
limited to
just one of the two subgroups.
DR. SAMPSON:
Again, I agree that the
sample sizes are small, but it looks
like there is
quite a bit of difference in the effect
of CyIS
depending on whether you are a match or
mismatch.
I realize that is the log-rank but I am
trying to
figure out if it is comparable across
both arms.
Would you have just a log-rank statistic
on the
matched group alone for the folks that
had the CMV
match?
Is there really a significant difference in
survival based on 9 and 23 and 3 and 21?
DR. CAPRA: I don't have that log-rank
limited to that subgroup.
168
DR. SAMPSON: And then you don't have
anything on the other stratification
variable?
DR. CAPRA: We looked at that similarly
where you see that in the case of the
breakdown is
limited to 1 into 2 subgroups. But, you know, the
numbers are small and what we did was we
did a
stratified log-rank rather than do a
log-rank by
subgroups because the numbers are
limited.
DR. SAMPSON: There wouldn't be an
interaction here though. I mean, that is clear at
least for the CMV mismatch, but you
don't have that
data for the other--
DR. CAPRA: We looked at the interactions
between the major risk factors and we
didn't see
any significant interactions, nor
interactions with
those major risk factors with the
treatment effect.
DR. SCHOENFELD: I am curious about one
other thing. You know, I looked at the AR data.
There is a real difficulty even
analyzing that
data.
For instance, the three patients who died
early, they may have been the bad actors
and they
may have been the ones that would have
had a lot of
169
acute rejections. So, I don't know if either you
or the FDA made any attempt at trying to
analyze--you know, it is a strange
situations when
the primary endpoint is essentially an endpoint
which is un-analyzable. It comes from the fact
that I assume you assumed there would be
no
survival difference so you chose an
endpoint that
is very hard to analyze in the face of a
survival
difference. So, was any attempt made to model
that?
I don't see the study as negative in terms
of the AR difference.
DR. CAPRA: We looked at AR-free survival
but what happens is the ARs are
occurring very
quickly and I didn't feel that that was
necessarily
as meaningful on average because the
effect seems
to be limited quite early in the first
couple of
months.
DR. SWENSON: Dr. Mannon?
DR. MANNON: I had a couple of questions
and a continuation so let me just pose
one to both
the applicant and FDA. There was a comment made
about drug levels between the two
groups. That
170
information isn't always clinically
obtained. Do
you have a sense of mean levels over
time between
the two groups both for tacrolimus
and/or
cyclosporine between the two groups?
DR. NOONBERG: We looked at mean blood
levels of tacrolimus. It is really dosed by levels
rather than by actual doses, and at the
three-month
time points the results were always
comparable
between the two groups. We also looked at
prednisone doses between the two groups
and they
were comparable as well. We looked at
immunosuppressive intensifications
between the two
groups and that was also comparable.
DR. MANNON: But these are three-month
levels that you were measuring. So, then looking
at the comparability later of
intensification, that
is based on dose again or based on
level?
DR. NOONBERG: I am sorry, can you repeat
that?
DR. MANNON: So, the levels that you had
at three months where you saw no
difference between
the two groups--
171
DR. NOONBERG: Three months, six months,
nine months, all the way up through
final.
DR. MANNON: My second question deals with
nephrology insofar as there appeared to
be a
significant increase in perioperative
renal failure
between the propylene glycol versus the
CyIS. What
was the sense of serum creatinine or
calculated
creatinine clearances between the two
groups, say,
at six months and 12 months? Also as a second part
of that question, was there a difference
in the
possibility of dialysis or intervention
of dialysis
between the two groups?
DR. NOONBERG: We looked at creatinine
levels because we had all the laboratory
values
transferred electronically to Chiron,
and we didn't
see any difference in creatinine levels
at the same
specified time periods. moxifloxacin
DR. SWENSON: Dr. Hunsicker?
DR. HUNSICKER: Yes, I would like to
address a question both to Drs. Zalkikar
and also,
if he is still willing to talk about
things, Dr.
Helms.
I am interested in what I might call study
172
ascertainment bias. This is a small study. Let me
again start out by saying that the
reason that we
are here is that there was a striking p
value for
the difference in survival. So, that p value is
reasonable if it is a true p value that
is not a
selected p value. The company has described how
they became involved in this as having
been told by
a rep. that there was a study that was
remarkably
positive. So, we have here now a selected study.
This is a classic issue in selection
bias of
studies, or what is called publication
bias, and I
would like to have both Dr. Zalkikar and
also Dr.
Helms discuss how seriously we should
take this
primary p value given that it was a
study that was
selected on the basis that the p value
was highly
significant, without knowing how many
studies it
might have been selected from. What I am basically
doing is challenging whether the primary
p value is
a real p value. Did you understand my question,
Dr. Zalkikar?
DR. ZALKIKAR: The primary p value was
actually the p value associated with
acute
173
rejection because that was the primary
endpoint.
DR. HUNSICKER: No, I am sorry, I am not
talking about the designed study's primary
p value.
I am saying that we are looking at this
study
because it has a p value of 0.007, as I
recall. Is
that something that we can rely on, that
this is
truly an unexpected result based on
chance given
that the study was selected based on the
fact that
there was a very highly significant
primary p
value?
DR. ZALKIKAR: From the only analysis that
I have seen in the data, I would be very
concerned
that the study was selected in terms of
the p
value.
DR. HELMS: You remind me of a couple of
students in my linear models class who
always ask
the tough questions. I think the short answer is
that the p value that is reported with
the
Kaplan-Meier curve is not a real p
value. I have
warned you in one of my slides that the
opinions I
express are not those of Chiron or FDA,
or anybody
else perhaps. There is some obvious selection
174
going on here.
But if you will let me be
informal for a
moment, the difference is so big it
passes the
intraocular trauma test--it hits you
between the
eyes.
So, if there were a good statistical
procedure--I also made the point in my
slides that
this is the primary problem here. This is the real
problem that we are facing because we
don't have a
real p value. As I said, we could do the decision
theory analysis. But, again, if you didn't want to
approve it you could question the
assumptions, and
so
on. But the difference is so huge that
we
cannot ignore it. So, the short answer to your
question is no, the p value--I mean, you
could put
in as many decimal places as you wanted;
it is
computed, but it is not a real p value
in that
sense.
Let me make a point that might
help, and
my statistical colleagues can correct
me.
Basically, this is high stakes gambling,
and the
best people at gambling in the world are
the
casinos and there are certain bets in a
casino
175
where you get partial data and then you
can place
an additional bet. Black Jack is one of those. I
am not an expert at that. We are in that kind of a
situation. We now have the results. How striking
are these values? And it is very difficult for
statisticians to come up with real
answers to that
question.
We have to bounce back to common sense.
DR. HUNSICKER: If I might, I do have one
other question. There seems to be a difference
between the survival curves that we have
been shown
by the company and by the FDA based on
time to
either BOS or death, depending upon
exactly how the
BOS has been defined. I am not sure I understand
whose graph is the one that I should be
looking at.
As I understand it--and I am putting
this question
to both groups--as I am understanding
it, the
company has shown us that if you look at
time to
BOS, when you include death when it is
presumed to
be due to BOS there is a significant
difference. I
think you showed 0.01, or something like
that.
When the FDA showed this based on
censoring all of
the patients whose death was not proved
to be due
176
to BOS, when that BOS was defined
according to a
specific thing, it was no longer
significant. I do
not understand what the discrepancy is
here. Can
this be clarified for me by the company
and by the
FDA?
DR. CAVAILLE-COLL: Well, I think that all
the BOS-free survival is largely driven
by the
difference that is observed in
mortality, and we
tried to look at it differently. Is there a
possible way of seeing an effect on
BOS? Looking
at the FEV1 plots, we don't believe that
there
really is a treatment effect on
BOS. Certainly, we
were not able to use the definition of
the
International Society of Heart, Lung
Transplantation because the baseline would
require
the average of two maximum values more
than three
weeks apart. We didn't have that type of database.
So, we were able to look at it basically
based on
the applicant's definition, but we are
really very
concerned about the completeness of the
collection
of that data.
I also want to go back to
another issue
177
having to do with the background
immunosuppression.
That information was not available in
the original
application. We had to request some of that
information from the company, and they
even said
there was a lot of difficulty in
collecting that.
Therefore, we asked them to provide us
at least
with some summary statistics of what
would be the
mean exposure at certain time points to
at least
give ourselves a qualitative impression
of whether
there was similar treatment between the
two groups.
But this is certainly a very
unconventional
application and we did not have the
level of detail
that we normally get in an application.
DR. HUNSICKER: Could I get the company to
clarify for me then? If you use your definition of
BOS, which we understand cannot be
ISHLT's
definition because you don't have the
baselines--if
you use yours and do time to BOS,
whether it was
ascertained at death or otherwise, do
you or do you
not find a significant difference?
DR. CAPRA: We find a significant
difference and there are two differences
between
178
our analysis and FDA's. Number one reason is the
informative censoring because we are
including
deaths as an endpoint, as occurs by
FDA's own
guidelines for oncology studies. I also want to
make the point that we did get all the
FEV1 values
electronically from the Pittsburgh
database. There
were no values that were missing from
our data.
The second difference is that
we are
including in a definition of chronic
rejection both
BOS, as determined by a sustained
decrease in FEV1,
and OB, defined by presence of OB or
change on
bronchial biopsy.
If we could show slide CE-30, please.
Combined in that chronic rejection
analysis we are
including as chronic rejection either OB
or BOS.
When we censor those patients and,
granted,
informative censoring is going to bias
against the
treatment groups, we still get a
significant value
of 0.015 in favor of the treatment
group.
This is an analysis of chronic
rejection-free survival with deaths
censored, and
it demonstrates that inhaled
cyclosporine prevented
179
chronic rejection. So, the effect of cyclosporine
on chronic rejection was not determined
solely by
mortality.
DR. HUNSICKER: If I could be clear on
this because I think it is a fairly
important
thing, Dr. Cavaille-Coll, do you agree
that this is
an accurate survival curve, assuming
that you are
now looking at time to chronic rejection
as
evidenced either by BOS in accordance
with the
applicant's definition or biopsy
documented OB? Is
that fair?
DR, CAVAILLE-COLL: This is one of the
retrospective analyses. This was not from
prospectively designed data analysis
plan, and it
is combining two things, OB as diagnosed
by
transbronchial biopsy, for which we know
the
sensitivity is fairly poor--the
specificity is
great but the sensitivity is very poor,
and also
bronchiolitis obliterans syndrome, which
is the
clinical description of that. As we have seen
before in the FEV1 plots, there really
is no
treatment effect. If you want to look at it this
180
way I am not sure how we would interpret
that p
value when you are combining two
different types of
things.
DR. HUNSICKER: But I am taking a
clinician's judgment here that if there
is anything
that you can say, it is that when you
have a biopsy
documented OB, that is OB, for God's
sake. So, if
you add biopsy documented OB into your
analysis, do
you then get this graph?
DR. CAVAILLE-COLL: We didn't do this
particular analysis and we didn't mix OB
with BOS
because BOS tells you something about
the rate and
the extent of progression. The transbronchial
biopsy doesn't tell you anything about
the extent
or the rate of progression so we didn't
mix the
two.
DR. SWENSON: Dr. Hernandez, do you have a
quick question here relevant to this?
DR. HERNANDEZ: Yes. I
just want to make
a comment from a clinical
perspective. When you
are analyzing chronic rejection and you
combine OB
and BOS there are several things that
you have to
181
take into consideration. The first is
the natural
history of the disease of chronic
rejection. It is
something that can be rapidly progressing;
it could
be slowly progressing. So, the diagnosis of OB by
transbronchial biopsy in a patient that
has a very
slow progression is not of the same
significance as
the patient that has, you know, OB
biopsy diagnosed
that is rapidly progressive. That is why the best
surrogate marker is FEV1.
So, when we are trying to
analyze by
putting together two of these
definitions for
chronic rejection we come to that
problem. So, the
problem of this natural history of the
disease that
we still do not understand is not only
due to
immunological causes. There are other non-immunological
causes that contribute to how this
disease behaves. So, basing our
diagnosis and
taking that as an endpoint as OB by
biopsy becomes
clinically not really meaningful because
the bottom
line is how well the patient is
breathing, and FEV1
as a diagnosis of BOS is really
relevant.
DR. NOONBERG: I just want to make the
182
point that Dr. Golden made initially,
which is that
OB and BOS are not two different
entities. They
are clinical and histologic
manifestations of the
same process. The problem with BOS is that you
have to use an unexplained decline in
FEV1 and we
already know that BOS is intricately related
with
pneumonias so patients will often have
pneumonia
and, by definition, you have to say,
"okay, not yet
BOS," wait for the pneumonia to be
treated and see
whether the decline in FEV1 is still
there and the
patient gets another episode of
pneumonia.
So, the diagnosis of BOS,
exactly when you
get BOS, is very hard to make, unlike
transbronchial biopsy where you have
that
diagnosis. Really it is a spectrum of disease and,
therefore, we feel that both have their
strengths
and limitations in the diagnosis and the
specificity and sensitivity, but they
are the same
clinical process.
DR. SWENSON: Dr. Proschan?
DR. PROSCHAN: Yes, I think that was a
really good question Dr. Hunsicker asked
and I just
183
wanted to clarify. In your analysis--you may want
to pull up a chair--
[Laughter]
--in the analysis that you
did, are you
saying you effectively used the
composite endpoint,
death or BOS, or did you only count
deaths for
which you could confirm there was BOS?
DR. CAPRA: It was death or BOS. In the
analysis of chronic rejection-free
survival we
include all-cause mortality. So, it is death or
chronic rejection or both.
DR. HUNSICKER: But is it not the case
that the non-prospective--all of that
stuff--analysis that showed time to
either death
with documented BOS or OB with death is
not due to
those documented, censored?
DR. CAPRA: Yes, 0.015.
So, we looked at
a sensitivity analysis where we censored
deaths.
So, we included just time to first case
of chronic
rejection, either OB or BOS, and we
basically
ignored death and that was significant.
DR. BRANTLY: I still have one concern.
184
The definition that the sponsor is using
for BOS is
20 percent drop. Is that correct?
DR. CAPRA: It is the 20 percent drop from
the previous peak FEV1 that is
unexplained by other
clinical manifestations.
DR. BRANTLY: My concern in using that is
basically the difference between a
double lung
transplant and a single lung transplant
because,
obviously, if you use that 20 percent it
is going
to take a lot longer to drop to 20
percent on
double lung than it is on single lung.
DR. CAPRA: Not necessarily. Dr.
Zalkikar, in fact, showed that the
double lungs had
higher FEV1s and we agree with
that. But we are
looking at a change, a decrease in 20
percent. So,
a double lung transplant subject who is
starting
with an FEV1 higher has more room to
drop.
Basically, what happens is you are
controlling for
that in the analysis because you are
looking for a
change from the previous values.
DR. SAMPSON: Excuse me, we have seen OB
or BOS censored by death. We have seen BOS
185
censored by death. Does somebody have OB censored
by death?
DR. CAPRA: We have that.
DR. SAMPSON: Did you show it to us? Did
I miss that?
DR. CAPRA: No, we haven't presented any
of that.
It is one of our backups. Can we
look at
SA-35?
Again, this is censoring deaths.
It
includes informative censoring so it is
going to be
biased against the treatment group. But when we
look at this analysis we get a p value
of 0.06 on
time to first case of OB.
DR. SAMPSON: And this is the deaths that
resulted from OB that are included as an
OB event?
Correct?
DR. CAPRA: No, but they would have had OB
before.
It is only cases of OB documented through
the histology. We are ignoring deaths in the
analysis. We are censoring those subjects. So,
those subjects who died for other causes
are
basically censored in the analysis. Subjects who
had an OB and later died are included on
the
186
Kaplan-Meier curves at the time of first
case of
OB, first diagnosis of OB.
DR. SAMPSON: I am not asking the right
question but I am going to try, were
there subjects
that did not show OB before death but on
autopsy
were diagnosed as dead primarily from
OB?
DR. NOONBERG: In our autopsy reports all
patients that were said to have had OB
were called
OB, and none of the patients that we
didn't say had
OB have OB on autopsy.
DR. SWENSON: Dr. Burdick?
DR. BURDICK: I would like to think about
the biology of what happens in the lung
a little
bit.
We have this huge difference between the two
groups and you would have thought that
if the huge
difference is due to the only thing we think
cyclosporine can be doing you would have
a huge
effect on what you think cyclosporine is
supposed
to be doing. We have just had a long argument
about OB and, leaving that aside, I am
also
concerned about the absence of effect on
acute
rejection. If the tests we are looking at for both
187
OB or chronic rejection and acute
rejection are
meaningful, it is of some concern.
In particular, we have talked
about
systemic versus local effects and the
biology of
that, and I think that is a bit
speculative at this
point, as people have said, but one
thing that is
very clear is that cyclosporine is
chemically very
lipophilic and it ought to move rather
quickly
through the small distances between the
bronchioles
and the small vasculature. I am concerned about
the interpretation that somehow this is
only
affecting what is going on in the
respiratory space
and not in the vasculature nearby where
it ought to
have a big effect on acute rejection if
it is an
immunological effect that the cyclosporine
is
having.
So, I just wonder if there is anything
more to be said about that.
DR. DILLY: Because of the very phenomenon
you describe, the lipophilicity, we know
that when
cyclosporine is given directly into the
lung the
plasma half-life is actually 40 hours as
opposed to
when it is given systemically where the
plasma
188
half-life is six hours. So, it is clearly sticking
around in that compartment for a long
time.
Now, one admittedly simplistic
way of
looking at this whole picture is that
everyone was
getting full dose of systemic immunosuppression.
If you accept that acute rejection is
largely
driven from the vascular compartment you
would
expect it to be treated similarly in
both groups
because we believe, for the reasons you
say about
the short diffusion distances, there is
really
little difference between getting
immunosuppression
into the vascular compartment and
diffusing out on
both sides in terms of the vascular
compartment.
In fact, we think the very
lack of an
effect on acute rejection makes the
chronic
rejection difference more meaningful
because we
don't have the confounding
variable. Imagine if
were trying to interpret this study and
there was a
huge signal that cyclosporine had
suppressed acute
rejection, then we would all be standing
here,
saying, well, you can't then surmise it
did
anything to chronic rejection. So, our hypothesis,
189
put really simply, is acute rejection
was treated
systemically; we are treating chronic
rejection.
The fact that there was no difference in
acute
rejection actually makes the effect on
chronic
rejection and survival more
interpretable rather
than less interpretable. We are deliberately
keeping it simple right now because that
is the
extent of our biological knowledge if we
are really
truthful about it.
DR. SWENSON: Dr. Tisdale?
DR. TISDALE: I am having trouble here.
There are two pieces of data--I didn't
mention when
I introduced myself that I am a
hematologist so I
don't know anything about the way that you grade
BOS in this clinical context, but there
are two
pieces of objective data that I find
very striking.
One is the survival curves that you
presented and
the other is the analysis of FEV1, which
are
parallel. So, to me these seem almost
irreconcilable. On the one hand, we have mortality
decrease presumably due to decrease in
chronic
rejection, which should be shown by change
in the
190
slope of the FEV1, much like you would
see for
creatinine in chronic allograft
nephropathy.
So, I am wondering, number
one, did you do
the analysis for the FEV1 with all
patients, or was
that censored in some way? If so, maybe somebody
could comment to me on what is the
proposed
mechanism of action of inhaled
cyclosporine if it
doesn't prevent a decline in FEV1?
DR. CAPRA: We did look at FEV1 in a
number of ways. We looked at changes from
baseline. Could we have slide SA-27 up, please?
We saw a trend in favor of the active
group. This
is a representative example, a 0.15
increase in the
placebo group versus a 0.40 increase in
the
cyclosporine group. It was not significant, as
were all the analyses, but all the
analyses did
trend towards a favor of the active
group.
Could we have the next slide,
please? We
think the reason why is because of two
reasons.
Number one is informative
censoring. We are not
able to get FEV1s on subjects after they
are
deceased. Secondly, FEV1 is highly variable and it
191
is subject to short-term
variations. The BOS
analysis basically ignores short-term
variations
and looks at a sustained 20 percent
decrease in
FEV1.
By sustained, it has to be for a period of
at least 3 weeks.
So, ignoring the short-term
variations and
addressing the informative censoring
what you have
is an analysis of BOS-free survival and
that was
significant, with a p value of
0.019. We are not
claiming a direct effect on FEV1 because
we did not
hit that statistically in the analysis,
but we do
think that these data support that there
seems to
be some improvement in lung function,
namely,
through BOS-free survival.
DR. HUNSICKER: Could I address that
specific issue very quickly? You said that you
can't get the data after the people have
died, and
that is true, but you can do a mixed
model and
still get an accurate and relatively
unbiased
assessment of whether there is a
difference in
slope and I am surprised that that
wasn't presented
by either of you.
192
DR. CAPRA: We did look at slopes through
a mixed model in the NDA and, again,
that was not
significant but the point estimate
favored the
active group.
DR. HUNSICKER: That should eliminate some
of the problems of missing data.
DR. NOONBERG: One of the pieces that we
looked at, since it is the heart of your
question,
is biological plausibility and, to
simplify it,
chronic rejection is mediated in the
airway
epithelium. Dr. Golden has described that the
epithelium is key to this process and we
are
delivering an immunosuppressant directly
to the
airway epithelium. Chronic rejection is a
progressive problem and so it is decline
in FEV1
but it is also recurrent pneumonias and
all sorts
of other complications. Graft failure as you see
in the chronic rejection process is
really a setup
for all sorts of mortality and,
therefore, the key
point that we want to make, and what is
so
important about the analysis where we
censored the
deaths with our chronic rejection, is
that
193
treatment with inhaled cyclosporine
prevented
chronic rejection, and by preventing
chronic
rejection we improve mortality. I mean, it is
clearly more complicated than that but
that is the
simplified story and to me, as a
clinician, it
makes good sense.
DR. SWENSON: Dr. Barrett?
DR. BARRETT: I was really struck by Dr.
Zalkikar's simulations or the analyses
looking at
the assignment, in particular one
looking at
partitioning out subjects who got little
drug. I
know there was no a priori statistical
analysis
planned, but I guess I was curious from
the
standpoint of the sponsor, would you
declare
evaluable subjects based on the limited
number of
doses received with inhaled cyclosporine
based on
this data? And what is your take on that?
DR. DILLY: Our take on that is that when
we go from 26 patients treated with
active and
long-term follow-on to 10 times that
number we will
be able to draw a much more accurate
confidence
interval around the trajectory of the
patient. We
194
think that is really the bottom line of
what needs
to be done now because I think Prof.
Helms'
statement about is the p value real--who
can say?
We want to put a real p value on what
happens to
respiratory function; what happens to
survival;
what happens to chronic rejection. At the moment,
we need to include in the prospective
study that we
want to do an intent-to-treat analysis
because we
want to describe what happens when
patients are
prescribed 300 mg three times a week of
inhaled
cyclosporine. So, yes, we would evaluate those
patients. Does that answer your question?
DR. BARRETT: No, I guess I was looking at
if you were going to define criteria on
evaluable
subjects based--I mean, in this
situation you had
difficulties. Again, it was a study that was done
already at the University of Pittsburgh,
but very
heterogeneous dosing exposures.
DR. DILLY: So, one of the things that we
have clearly not done, we have clearly
not got a
formal dose-response study sitting in
front of us.
Right?
One of the analyses that we want to do is
195
an achieved dose duration analysis
versus benefit.
Absolutely, that is one of the critical
parameters
in the study.
DR. ZALKIKAR: Can I just put in a couple
of words regarding FEV1? We looked at the FEV1
data as much as we could. The pre-enrollment data
was missing on a large number of
patients, as I
mentioned. We tried to impute the pre-enrollment
data by the type of lung transplant that
the
patients received and see the difference
between
the final FEV1 and the pre-enrollment
FEV1, and
none of those analyses has shown any
difference
between the two arms.
We also looked at the three-month
data and
treated that as some sort of baseline
because it
was available, and saw a difference from
that to
the final FEV1 that was indicated in the
database
and, again, there was no difference
between the two
arms in terms of that.
DR. PROSCHAN: Yes, I thought it was very
interesting to see the sensitivity
analyses, and it
seems like both parties did something
196
reasonable--you know, not that you
haven't been
doing other reasonable things. You basically threw
out those early deaths and did a
sensitivity
analysis and showed that the results
were still
significant. On the other hand, you switched the
treatment labels and showed that things
become not
significant once you do that.
You know, I want to go back to
their
analysis. You would think that if you threw out
those three placebo patients with the
early deaths,
you are throwing out sick patients and,
therefore,
the remaining placebo patients should be
healthier
than the patients left in the CyIS
arm. So, it
seems like that is somewhat reassuring,
that even
when you throw those out it retains the
significance. I guess it is not really a question
but a comment.
DR. SWENSON: At this stage we are close
to the end of the morning session--oh,
Dr. Reiss?
DR. REISS: I just have one question. It
might have been discussed and I might
have missed
it, but the bronchial biopsies for the
diagnosis,
197
were they done in a routine, standard
manner or
were they done for cause?
DR. NOONBERG: The short answer is both.
There was a protocol specified minimum
and then
also for clinical cause.
DR. WATKINS: At this point, I just have a
brief statement before we adjourn for
lunch. I
would like to remind the committee that,
in the
spirit of the Federal Advisory Committee
Act and
the Sunshine Amendment, discussion about
today's
topic should take place in the form of
this meeting
only and not occur during lunch or in
private
discussions. We ask that the press honor the
obligations of the committee members as
well.
Additionally, any open public
hearing
speakers who have not yet checked in,
please do so
at the registration desk. And, if everyone could
return just prior to one o'clock we will
reconvene.
[Whereupon, at 12:00 p.m., the
proceedings
adjourned for lunch, to reconvene at
1:00 p.m. this
same day.]
198
A F T E R N O O N P R O C E E D I N G S
DR. SWENSON: Welcome back to the
afternoon session. This session will have
initially an open public hearing to hear
from
interested parties outside the company
and the FDA.
Then we will move back to general
discussions of
concerns and problems and issues, and
then
ultimately we will vote on the
particular questions
that are posed to us by the FDA.
Before starting the open public hearing, I
would like to read this particular
message: Both
the Food and Drug Administration and the
public
believe in a transparent process for
information
gathering and decision-making. To ensure such
transparency at the open public hearing
session of
the advisory committee meeting, FDA
believes that
it is important to understand the
context of an
individual's presentation.
For this reason, the FDA
encourages you,
the open public hearing speaker, at the
beginning
of your written or oral statement to
advise the
committee of any financial relationship
that you
199
may have with the sponsor, its products
and, if
known, its direct competitors. For example, this
financial information may include a
sponsor's
payment for your travel, lodging or
other expenses
in connection with your attendance at
this meeting.
Likewise, FDA encourages you, at the
beginning of
your statement, to advise the committee
if you do
not have any such financial relationships. If you
choose not to address this issue of
financial
relationships at the beginning of your
statement,
it will not preclude you from speaking.
So, with that said, I would
like to
introduce Miss Esther Suss. She will be our first
presenter. Miss Suss, are you here? You can use
that microphone over there.
Open Public Hearing
MS. SUSS: Before I start my statement I
would like to point out to the board members
that I
was one of five persons on Team
Pittsburgh who
agreed to be filmed and interviewed by
Chiron
during the 2004 Transplant Games as part
of an
internal campaign by Chiron
to...[microphone
200
off]...
As a follow-up to that
interview that I
gave at the games, Chiron asked me to
demonstrate
the use of inhaled cyclosporine as part
of a
demonstration DVD to be given to new
users of this
therapy.
Thus, in March of 2005 Chiron paid my
travel and other expenses to Orange
County,
California for the taping of my portion
of the DVD.
Now, ladies and gentlemen, I
appreciate
the opportunity to tell my story in this
public
forum because I feel that the FDA
approval of
inhaled cyclosporine is extremely
important to
persons like me. For over 30 years I worked for
the International Monetary Fund and
traveled
extensively to Latin America, Europe,
the Soviet
Union and Africa. In fact, in 1992 I opened the
IMF's rep. office in Latvia and was
essentially the
ambassador there and I lived there for
two years.
After working in Latvia, I then was
based in
headquarters but traveled to Uzbekistan
where I
worked for two and a half years, and
then to
Armenia where I worked for two years,
and then I
201
returned to work in Latin America.
During this time I often had
trouble with
some bronchitis, some asthma but I was
away from
Washington for over half of the
year. In January,
2000, after a vacation to Florida and a
visit to
Pittsburgh to see my family over the
holidays I
came down with what I thought was the
flu and had
trouble breathing and my inhalers
weren't working.
At that point, I finally went to the
doctor. I
went to the hospital and I was admitted
to the
hospital where my doctor told me we
think it is
just an exacerbation of your asthma by
the flu;
your x-rays look fine. But the next day I had a
classic asthma attack in the hospital
and went into
total respiratory failure. After not improving for
the next two weeks, they brought in a
pulmanologist. The pulmanologist told me "you're
in end-stage emphysema and you will
never go off
oxygen again." And I was devastated. I was 52
years old. I couldn't believe that this could
happen so quickly and I asked him,
"how could this
happen?" And he said, "first of all, your x-rays
202
were terrible. Secondly, some people push
themselves so that they use up every margin
that
they have and when they fall, they fall
completely
off the cliff." And, that was the situation that I
was in.
At 52 I faced the possibility
of never
being able to travel again, quite
honestly. I
spent about a month in the hospital and
was able to
improve significantly more than they had
anticipated. In the end, by the time I was out in
a month, I was on oxygen only when I was
doing any
real physical activity. But I returned to work
full-time and during the day I didn't
really need
to use my oxygen. But if I needed to walk--today,
for example, to walk from this building
across the
street to the cafeteria would have just
about
killed me even using my oxygen.
In June I asked my doctor to
check, to do
PFTs again to see if anything had really
improved
and the answer was no, it hadn't. That was as good
as I was going to be. I was told then the only
other thing to do was to have a
transplant. So, I
203
asked my doctor to write the necessary
letters of
introduction and I was evaluated both at
the
University of Pittsburgh and at NOVA
Fairfax. I am
originally from Pittsburgh. I have my Ph.D. from
the University of Pittsburgh. My younger brother
is a physician from the University of
Pittsburgh
Medical School. So, I was pretty familiar with the
programs in the hospital there and I
felt quite
comfortable with the people there.
Well, 22 months and five calls
later I got
my transplant in August, 2002. It was just
incredible. On August 13, 2002 I had a new
beginning. I had a new lung. I no longer had blue
lips, and for one week my progress was
impressive.
Then I had a major acute rejection, was
in the
intensive care unit on a ventilator for
several
weeks, and I just kept taking one
rejection after
another.
I had an agreement with the doctors and
Dr. Iacona was wonderful--"if you don't
give up on
me I'll keep fighting" and that
model has worked
very well. They haven't given up on me. I kept
fighting and here I am today.
204
I returned to Washington in
November of
2002 to have Thanksgiving with friends
but then had
to be rushed back to Pittsburgh again at
the
beginning of December with a major
infection which
resulted in another rejection episode,
and came
back to work full-time in March, 2003
and I started
traveling again. I went to South Africa to visit a
very dear friend in October but,
unfortunately, in
December, 2003 I came down with a viral
infection
which got my immune system roughed up
again and
gave me another acute rejection. At that point the
doctors decided, once they got the acute
rejection
under control, to put me on the inhaled
cyclosporine therapy.
Since then I have spent the
whole year of
2004 out of the hospital and, with the
exception of
an intestinal flu on New Year's Day of
2005, I have
been out of the hospital and I have not
had a
rejection episode for over a year. I have been
able to resume my life that I had before
I got
sick.
In October, 2003, as I said, I went to South
Africa to visit a dear friend and in
June, 2004 I
205
took my nephew to Paris and Madrid for
his 13th
birthday. I participated in the 2004 Transplant
Games where I ran an 800 meter, women's
800 meter
and got a silver medal, and I did a 5K
in about one
hour.
Pre-transplant I would never have been able
to do that even with oxygen in two
weeks, the 5K.
The inhaled cyclosporine is
the first, and
at this time the only drug which is
aimed at
preventing or at least postponing the
onset of
chronic rejection. I am a classic profile for
someone who will get chronic rejection
because of
my multiple acute rejections. For those of us who
have been given a second chance for
having a good
life it is so important to have this
treatment
available. I am eternally grateful to my donor for
the gift that I received and I feel that
wherever I
go I am taking my donor with me, and I
hope that he
enjoys traveling because he is doing a
lot of it.
I am aiming at setting the
record of being
the longest survivor as a lung transplant
recipient, not withstanding the survival
statistics
that you all know quite well. As we know,
206
statistics are for large numbers of
people and for
any individual they don't necessarily
apply, and I
intend to be on the plus side of those
statistics.
And, I believe that by using the inhaled
cyclosporine I am doing everything
possible that I
can to preserve and protect this
wonderful gift
that has been given to me and that it
will help me
accomplish my goals.
I have had no adverse
reactions or
problems using the inhaled cyclosporine,
and my
hope is that your approval of this drug
for the
general population will allow many other
people to
have the same long-term goals for
themselves.
Thank you.
DR. SWENSON: Thank you Miss Suss. Our
next speaker is Mr. John Sullivan.
MR. SULLIVAN: My name is John Sullivan.
I appreciate the opportunity to address
you today.
I have not had any financial rewards
from UPMC or
Chiron to come here and speak.
I would like to tell you a little bit
about myself. I had alpha-1 anti-trypsin
207
deficiency when I was 35 years old. That was in
1981.
In 1991 I had a single right lung
transplant. I did fairly well for about six months
when I came down with pancreatitis which
they for
some reason thought immune suppression
caused. So,
they had to reduce my immune suppression
and I am
still alive but after three weeks my
lung capacity
had greatly been reduced and I was
rejecting. I
had a couple of bouts of acute rejection
then I had
chronic rejection.
In 1994 I came down with
aspergillus and I
ended up there at the UPMC for three
months on a
ventilator. Then I went home for three years on a
ventilator and I was put back on the
list again. I
was lucky enough in February, 1997 to
receive a
double lung transplant. I had to have a double
because of the aspergillus. I was in a coma for
two weeks after that and I also lost my
kidney
during that operation. As soon as I came out of
the coma, Dr. Iacona and also Dr.
Griffith put me
on inhaled cyclosporine because I
already had shown
that I was a chronic rejector. I have been on
208
inhaled cyclosporine to the current
time. I have
been reduced to once a week.
The only time I have had any
chronic
rejection or any type of rejection was
in June of
1997.
I had fallen and broke my left hip.
So
then, when they put the pins in I came
home and I
was doing fine for about 10 days. All of a sudden
I had a grand mal seizure and I drove my
femur
straight through my pelvis on my right
side but and
on the other side bent the ball right
over. But
because of the grand mal seizure they
put me on
dilantin which reduces or dilutes your
immune
suppression. So I went into acute rejection. I
was actually in Michigan and then
Pittsburgh had me
switched over to Norantin. That is the only
rejection I have had since 1997.
In January of this year I had
the flu and
my immune system kicked in. My temperature went up
to 102, which isn't very good. And, I was on it
once a week. They jumped it up to twice a week,
hoping to stop any kind of
rejection. I have had
my PFTs done as well as biopsies and I
show no
209
signs of rejection at all.
I feel that the main reason I
have done so
well is because of the inhaled
cyclosporine. I
think that has helped me. I don't show any
rejection and there hasn't been any adverse
effect
on me at all. Thank you very much.
DR. SWENSON: Thank you, Mr. Sullivan.
Our next presenter is Mr. Bill Stein.
MR. STEIN: I want to address the
committee today and tell them that I
appreciate you
allowing me to take the time to tell a
little bit
about my past disease history and what
the
transplant has done for me to date.
I am a 32 years old and I had
a double
lung transplant in August of 2002, after
suffering
devastating effects of cystic
fibrosis. I was
diagnosed at the age of eight and I
lived a fairly
healthy childhood in adolescent years
until I was
approximately 20 years old. That was my first
hospitalization with the complications
arising from
cystic fibrosis. I thought after the initial
hospitalization that things would get
better and I
210
would just go on my merry way with my
life, but it
didn't turn out that way. That first year I was
hospitalized over five times and
progressing over
the next eight years my lung function
went from
approximately 95 percent to 18 percent
at the time
of the transplant.
In January of 1999 I was
listed for
transplant at the University of
Pittsburgh Medical
Center and in the years leading up to my
surgery I
faced the reality that my own lungs may
not sustain
my body until new lungs became
available. In 2002
my health declined to a point so bad
that I was on
oxygen 24 hours a day until August of
2002 when I
received my double lung transplant. I was called
four times for transplant. I had three false
alarms until I got the actual call in
August of
2002.
It was a very emotional ride, as well as a
very physical one, you know, dealing
with the trips
to the hospital and all the prep
procedures where
you would have your surgeon come in and
tell you
that, you know, the operation wasn't a
go.
It has been almost three years now
since
211
my transplant and in that time I have
managed to
reclaim my life and venture forward with
my goals.
Less than three months after my transplant I
returned to the ice hockey rink. I was an avid ice
hockey player and I also played golf
and, sort of
against a lot of people's wishes, I
wanted to go
back and enjoy the sport I loved. I took the
necessary precautions; I wanted to know
that I
could do it again.
Less than six months after my
transplant I
decided, well, it is time for a career
change so I
decided to go back to school and take
preparatory
classes to pursue my ultimate goal of
being a
physician assistant. The long hours have paid off
and after all the long and hard work
that I have
spent for the past two and a half years,
I will be
accepted to the class of 2007 physician
assistant
studies program this August.
Words cannot express the
gratitude and
appreciation I have for the support that
was given
throughout my journey by my friends, my
family and
my medical team and, most importantly,
the gift of
212
life given to me by my donor and the
decision by
her family.
I realize that my disease did
not allow me
the convenience of making an easy
decision
regarding my transplant but in the years
that
followed my transplant I have been able
to overcome
the hurdles associated with the
complexities of the
surgery, and even the care that
followed. I knew
going into transplant that infection and
rejection
would be issues that I would have to
address in
future care. The statistic that was provided to me
prior to transplant was not very
impressive, given
that the five-year survival rate was
approximately
50 percent, but what was the
alternative? I chose
the transplant and took my chances at
that point.
I have been very fortunate and
have only
had to deal with a few cases of acute
rejection
that were resolved with conventional
steroids and
adjustments to my immune suppression
medication. I
was able to battle through these minor
hurdles; I
always worry about what happens in the
future, what
happens the next time I get
rejection. Is it going
213
to
be treatable? Or, what happens if it
just turns
into chronic rejection? Do I have something to
look forward to, to help me treat
it? Just given
what I have heard with the inhaled
cyclosporine, it
just seems like it is a tremendous asset
to the
transplant community to have it
available for
patients like myself that might run into
problems
and conventional therapy might not carry
you. I
have seen a lot of my friends be in a
situation
where they are in intensive care for
months on end.
A few of them have passed away. It is very
heart-breaking knowing that I am well
and that they
are gone because there wasn't a drug
available to
combat the effects of rejection.
A new drug such as inhaled
cyclosporine
will help benefit transplant patients
such as
myself and prevent or reduce the
probability of
rejection occurring in transplanted lungs. The
drug can also help eliminate some of the
traditional side effects that patients
like myself
deal with. My creatinine level is currently 1.5,
1.6 so I am teetering on the sea-saw as
to which
214
way I am going to go. So, will inhaled
cyclosporine give me that opportunity to
lower my
current immune suppression in the future
and maybe
have dual medications to help alleviate
some of the
toxicities on the other organs? Those are
questions on our minds and, you know, I
am sure
there are a lot of transplant patients
who feel the
same way, and I believe the long-term benefits
of
the inhaled cyclosporine will provide
new hope and
opportunities for many patients allowing
them to
live long and prosperous lives without
worry of
additional complications. Thank you very much.
DR. SWENSON: Thank you, Mr. Stein. Our
last speaker is Miss Renee Moeller.
MS. MOELLER: I am a double lung
transplant survivor of three years. Although I am
doing very well, I am concerned about the
long-term
risks associated with my current
rejection drugs.
I feel that taking cyclosporine as a
preventative
medicine will benefit my long-term
survival rate
and I will tell you why that is
important to me.
I am a 29 year-old young woman who
for all
215
of my life coped with cystic fibrosis
and took care
of myself for years. I lived a pretty normal life
except for the daily routine of taking
50 pills a
day.
After I finished college my health declined
dramatically. I thought my life was over. I was
completely devastated knowing that I was
so young
and I had to deal with death. A transplant was my
only hope. My whole life I dealt with an incurable
illness.
Finally I could breathe like a normal
human being after this wonderful gift of
a
transplant. I don't want to risk losing this new
life that I absolutely love and have
longed for.
I am afraid of getting
rejection. One of
my best friends died because she
rejected to the
point of no return. It affected me a great deal
and I don't want this to happen to
myself. This is
a huge concern of mine and I would love
to have the
opportunity to take cyclosporine. By taking this
drug my other organs can be saved and,
hopefully, I
can live a longer and prosperous
life. Thank you.
DR. SWENSON: Thank you.
At this point I
will turn the meeting over to Dr.
Albrecht to give
216
us a charge for the rest of the
afternoon's
considerations.
Charge to the
Committee
DR. ALBRECHT: Thank you.
Before I read
the questions let me just spend a couple
of minutes
summarizing some of the salient points
and some of
the questions that we still have
remaining that we
would like the committee to talk about.
Let me just reiterate, as we
have heard,
that a single Phase II study was
performed and
Chiron submitted this, and we accepted
it, knowing
the size of the study, given the
dramatic
difference in mortality that was
seen. As you have
heard before, the reason for this is
that the
incidence of lung transplants is fairly
low. The
current mortality is higher than with
other
transplants. There is no approved FDA therapy and,
clearly, there is a need for safe and
effective
therapy in this population, as we have
heard so
eloquently stated.
In the Chiron presentations
you heard
quite a bit of information presented
about this
217
study and you heard Chiron's assessment that
they
felt that these results were really
quite robust
and supported the survival difference
and warranted
a positive regulatory action.
During the FDA presentation
you heard a
somewhat different perspective or
interpretation of
the very same data in the following
way: We
acknowledge that there was a difference
seen in OB
and mortality. But OB is a histologic finding and
a
histologic finding without some clinical signs
and symptoms causing mortality doesn't
quite
follow.
So, when we looked at some clinical signs
and symptoms, for example characterized
by FEV1 or
BOS, and did not see that difference we
were
puzzled and continued to do additional
analyses.
Again, just to remind everyone, the
primary
endpoint of acute rejection also was not
demonstrated to be significantly
affected by the
aerosolized cyclosporine.
So, we looked for other
explanations that
could possibly say why was there a
difference in
mortality without seeing differences in
FEV1 or BOS
218
and found a number of characteristics
that were
imbalanced in the two populations which,
as we
heard, of course, had been randomized
but sometimes
the risk of a small study is that randomization
may
still not lead to an adequate balance in
characteristics.
One of the main ones was
double lung
versus single lung imbalance. Then you heard a
difference in the incidence of early pneumonias
that happened in the single lung in
patients to a
greater proportion than the double lung
patients.
Dr. Hernandez then admitted that we do
not have a
lot of information on the donors but
used the
available information that we had to try
to get a
sense of the donors, as well as the
early courses
in the recipients, and pointed out that
the
inotropic to support to donors was
longer; the
PaO2/FiO2 was less than 300 in the PG
arm whereas
it was greater than 300 in the double
lung arm.
Then the recipients also had a longer
stay in the
ICU which correlates with their clinical
course
post-transplant.
219
So, seeing these imbalances,
we thought
could this be part of the explanation
because the
mechanism of action of cyclosporine and
the
incidence of pneumonia did not seem to
be directly
correlated. Then we see from the literature that
there is a strong association between
acute
rejection and chronic rejection. The assumption is
that acute rejection episodes through
various
cycles do lead to BOS and chronic rejection
and
mortality.
We also heard that non-immune
causes may
be responsible for this and again, as I
mentioned,
we did see that there were differences
in some of
the donor and recipient characteristics, as well as
differences in pneumonia, so perhaps
this was in
part accounting for the differences we
were seeing.
In my reading I actually came
across a
statement by the pathology group from
Pittsburgh
that occasionally organizing
pneumonia-like
reactions in airways and air spaces may
induce
bronchiolitis obliterans. So, perhaps these are
some of the known immunologic features that
may
220
predispose to bronchiolitis obliterans.
Then you heard about the risks
of a small
study, such as the fact that when you
have a small
study the small sample size results are
highly
sensitive to small perturbations in
assignment of
patients, and you heard some of the
results of the
sensitivity analyses and their impact on
changing
the significance of the detected
mortality and,
finally, you also saw that if we looked
at patients
who received less than 25 doses, which
was less
than 10 percent of the intended two-year
dosing
regimen, the difference in survival was
also no
longer significant.
So, having said all that, let
me now go
ahead and turn to the questions that we
would like
the committee to deliberate. Having read them
earlier, I will not read all of
them. I will start
with question one. I just did want to mention that
this is also attached to the agenda for
ease of
reading and it is now being displayed on
the
screens.
So, the first question, is
there
221
sufficient information to make the
determination
whether the observed survival difference
in study
ACS001 is due to study drug or to some
other
factors?
In your deliberation we would
like you to
keep in mind the information that you
heard this
morning, including the statistical
issues that were
raised; the differences in baseline
donor and
recipient characteristics; the effect
that was
demonstrated or the differences or the
lack of
differences that were demonstrated on
various
endpoints, including the survival
endpoint, acute
rejection, BOS, FEV1 and OB; and whether
you think
some other endpoint showed a
demonstrated benefit
or difference.
Let me just mention parts (a)
and (b) of
question 1 which is that if after you
deliberate
you believe that the answer to the first
question
is yes, what we would be interested in
is hearing
about the generalizability or, more
specifically
the labeling recommendations or the
labeling
summaries that would then evolve from
the results
222
of this study.
If, on the other hand, you
believe that
the answer to the first question is no,
we would be
very interested in hearing a discussion of what
additional clinical studies you would
recommend be
conducted. Here we would be interested in specific
recommendations regarding the patient
population,
drug dosing and regimens, drug
administration and
efficacy endpoints. I will stop there and then we
can return to the second question.
Committee Discussion and
Vote
DR. SWENSON: For ease in going through
these discussions, I would ask that you
just tap
your "talk" button and we will
see the red light go
on and then we will call people in turn.
I would like to first step
back and allow
for some potential questions that were
possibly not
brought up in the morning session and
maybe to
spend some 15 minutes or 20 minutes on
that because
I think that will be part and parcel of
your
answers to these questions. I know that Dr. Mannon
had some unanswered questions and we
will start
223
with her. But anybody else that has new ideas or
something unanswered from this morning,
please let
us know.
DR. MANNON: This is a question sort of
involved in Dr. Proschan's question
earlier this
morning on page 84 of the Chiron
binder. I was
struck by the number of patients that
completed
therapy.
Maybe I am misinterpreting it but only
about half of the patients, maybe less
than half in
the placebo group but about half of the
patients in
the CyIS group were able to complete
therapy. So,
I question both the sponsor and the FDA
about what
were the reasons for the lack of
completion of
therapy.
Also, from a clinical perspective, how do
you interpret the outcome when half the
patients in
your treatment group did not complete
full therapy?
DR. NOONBERG: To address the first part
of your question, the reasons for early
drug
discontinuation are shown in the table
below. The
primary reasons for the placebo group
are due to a
variety of adverse events. We have already
discussed the withdrawal of
consent. Again, there
224
were some problems with early tolerability
that
could be fed into the adverse event
group of the
inhaled cyclosporine group. The unsatisfactory
therapeutic response, those are patients
that were
taken off study drug in order to cross
over into
the open-label protocol. Protocol deviations and
violations were largely due to medical
non-compliance and smoking. So, those are the
range of reasons. There were no patients that
discontinued because they died. They discontinued
for other reasons prior to death.
DR. MANNON: I guess my second question is
when only half of your patients complete
the
proposed therapy, what is your
interpretation of
the outcome of those patients when only
half
completed really the full therapeutic
part of the
trial?
DR. NOONBERG: Well, my sense is that
inhaled cyclosporine helps patients
early on and
continuously because chronic
rejection--it is not
like acute rejection where one day you
have it and
the day before you didn't and two weeks
later you
225
don't.
It is a gradual process that probably
starts very early on and only after a
year it is
able to be detected by transbronchial
biopsy and
certainly by a decline in 20 percent of
FEV1.
Again, there is nothing magical about the
20
percent decline and, therefore, it is
not an
episodic process and, therefore,
patients who
appear to get more drug did better. How we justify
this really is using an intent-to-treat
analysis.
DR. ZALKIKAR: Can I respond to that
question? We agree with the reasons for
discontinuation that were displayed here
a while
ago.
As for the sensitivity, I have shown you some
sensitivity analyses where we treated
the patients
who received very little therapy in
three different
manners, one, excluding all 11 patients
who
received less than 25 doses; secondly,
treating the
cyclosporine patients who received very
little
treatment as the control patients; and,
thirdly,
defining treatment failure as
discontinuation or
very early discontinuation due to
adverse event or
withdrawal of consent. All of those sensitivity
226
analyses show that the survival benefit
is no
longer significant. So, the interpretation is
really hard.
DR. NOONBERG: Well, I would comment that
one of the reasons that patients didn't
get 25
doses is that a couple of them died and,
therefore,
couldn't have gotten 25 doses. When we did a
sensitivity analysis and only looked at
patients
who had 80 percent of protocol maximum
dosing
adjusted for death we found the results
were
statistically significant. I don't know if you
want to bring that slide up again just
to show the
results of our sensitivity analysis.
DR. SWENSON: I think you have made the
point well and I think we remember it.
DR. ZALKIKAR: Can I counter that? Again,
the few early deaths--is it really
possible to
attribute those deaths to lack of cyclosporine?
That is also a question we had a hard
time dealing
with.
DR. SWENSON: Dr. Burdick?
DR. BURDICK: Just for the record, I would
227
like to know just a little bit more
about how the
randomization actually occurred to the
extent that
that is reconstructible. About half of the
patients done during that interval
actually were in
the study. Were all approached? If not, what were
the criteria? Who made the decisions, and when?
And was it blocked by two, by four or
six? So,
what was that process like?
DR. IACONA: I am the principal
investigator of the study. About 120 transplants
were done at the University of
Pittsburgh at the
time of the three-year limited
enrollment period.
Out of those patients, about half
enrolled.
Unfortunately, we had about 20 percent
death rate
during that interval which excluded a
large number
of patients, which we could do nothing
about. We
did screen through other patients and,
to be quite
frank, there were a lot of things going
on. We
didn't want to push the drug because we
had it for
other studies and it was an open-label
protocol.
That was one issue. The other issue was that the
patients didn't like the idea of
receiving a
228
placebo.
So, when you pushed them, their gut
reaction was why should I go into the
study when I
can get the drug if I should
reject? So, that is
how it was done. That is how I approached patients
and my colleagues did.
As far as randomization is
concerned, I
had no part in randomization. It was done by the
University of Pittsburgh statisticians
and I had no
idea as to how blocks were
assigned. I never
received the coding for randomization
and there was
no way any investigator in the program
had any idea
as to who was receiving drug or placebo.
DR. SWENSON: Dr. Hunsicker? I am still a
little bit foggy about exactly how the
analysis
plan was constructed. We have a study for which
the primary outcome was something that
is not now
really the issue of evaluation. So, the
presumption is that you noticed a very
substantial
difference in the death rate and that is
what now
powered the rest of the analysis that
was going on.
So, my question is, and I don't know how
I can put
this quite precisely, but to what extent
was the
229
analysis plan developed before the data
were looked
at so that you knew what you were looking
to test?
Because, clearly, if the analysis plan
knew the
outcomes it is no longer really an
unbiased
analysis. So, I have to ask at what stage in the
development of this thing was the
analysis plan
finally formalized by which the analysis
specified
that we were going to look at these
things? Was it
after you knew the frequency of BOS and
so forth,
or was it simply when you recognized
that there was
a difference in the rate of death?
DR. CAPRA: As we have heard earlier in
discussion, the enrollment was for three
years in
the randomization portion of the trial,
up to
August of 2001. The last subject was followed for
an additional two years for
treatment. At that
point the study was unblinded and at
that point the
analysis took place, and there was no
prospective
analysis plan because it was an
investigator-sponsored trial done at
Pittsburgh.
What they had in their
protocol was they
had a primary endpoint of acute
rejection. The
230
prospective endpoints were chronic
rejection and
overall survival duration. There was no
prospective analysis as to what kind of
covariates
to include in the survival duration
endpoint. So,
the trial was over; the results were
this;
treatment assignments were unblinded so
we looked
at survival duration from the time of
transplant to
when the study was closed in August of
2003 and the
results were unblinded. And, that is why we looked
at a number of analyses. The first thing is
straightforward log-rank test. It was significant.
Then we went from there and started
looking at
other sensitivity analyses which
basically
confirmed the overall result.
DR. HUNSICKER: The specific issue that I
am getting at is what you chose to use
as your
endpoint beyond simply survival. I am going to
assume that that is well defined, and
everybody
knows whether you are alive or
dead. But when you
create an analysis like this you can
have any
number of things that you put into a
composite.
You know, you are looking at a whole
series of
231
composites and if you know what the
outcomes are
before you build your composites, then
you are not
blinded as you create your test and it
means that
you can't really test it that way. So, I am trying
to
get a sense of how you developed what you were
going to do--you understand what I am
talking
about.
DR. CAPRA: Sure.
The secondary endpoint
was chronic rejection. So, in discussion with the
doctors both at Chiron and then with Dr.
Iacona at
Pittsburgh I understood that chronic
rejection is
measured both histologically through OB
that is on
the biopsies, as well as the BOS which
is the
sustained 20 percent decrease. From there, we
looked at time to first occurrence of
chronic
rejection, again, first doing it in an
unadjusted
analysis, looking by log-rank test, and
then
forward from there.
DR. HUNSICKER: But were there analyses
done so that the people at
Pittsburgh--and if I had
been there I am sure that I would have
done these
things so that as you did this you would
have known
232
what the rate was of OB and BOS before
you began
putting together your composites. Is that the
case?
DR. CAPRA: I guess I will have to yield
to people from Pittsburgh. I haven't seen that so
I will let them discuss it.
DR. IACONA: So, the question, as I
understand it, is what was prespecified
in our NIH
grant.
We had an NIH grant that supported the
study at the time. The primary outcome, as written
by Howard Rockette [?] at the University
of
Pittsburgh who is the head of our
biostatistics
section, states that we would use acute
rejection
as our primary endpoint evaluating acute
rejection
by Poisson analysis.
We also evaluated survival and
freedom
from rejection as secondary endpoints,
as stated in
that grant. The freedom from rejection was
specified for both acute and chronic
rejection as
determined by Kaplan-Meier and
evaluating by
log-rank analysis.
DR. HUNSICKER: So, it was freedom from
233
both acute and chronic?
DR. IACONA: Freedom from acute rejection
was primary--
DR. HUNSICKER: No, no, no, I understand,
but your secondary endpoint was--
DR. IACONA: Freedom from chronic
rejection. It is stated in the grant, yes.
DR. HUNSICKER: And was the definition of
chronic rejection in there? The major issue here
is that you are depending a fair amount
on BOS.
DR. IACONA: Yes.
DR. HUNSICKER: You know that because of
the way you did things--you can't
use--what do you
call it?--the heart, lung transplant
definition
because you didn't have the baselines. Did you
have your definitions in place before
you started
building an analysis?
DR. IACONA: Yes, the definitions were in
place as specified in that NIH grant,
and we can
certainly provide you the statistical
section of
that grant. We evaluated chronic rejection by two
means, by BOS and by histology.
234
DR. HUNSICKER: But those endpoints were
defined before you did the analysis?
DR. IACONA: Yes, in the NIH grant.
DR. CAPRA: I think there is also a
misunderstanding about the baseline
BOS. Baseline
is a running average as a function of
time so
because subjects were enrolled as early
as seven
days after transplant, sometimes a
subject didn't
have an FEV1 value available in that
first week;
they had one maybe on day eight or day
nine. We
were able to calculate a baseline BOS,
which is a
running average, and then look at change
from that
value for calculation of BOS.
DR. IACONA: Could I make a statement
about the initial PFT? Before these patients go on
aerosolized cyclosporine or placebo they
are
postoperative patients, some of whom
have
tracheostomies; they have chest pain;
they are not
candidates to actually journey to a
bronchoscopy
laboratory to get a PFT and if they did
get a PFT
they would be highly inaccurate. So, that initial
PFT, in most cases where it was not
present, would
235
be impossible to get under the clinical
circumstances of a postoperative
transplant
patient.
DR. SWENSON: Dr. Moss?
DR. MOSS: I think this is all really
interesting. I think Chiron has done a great job
in presenting the data and actually the
question is
not for you but you can stay up there if
you want.
The data that Pittsburgh and Chiron have
presented
is extremely interesting and very thought
provoking. There was a comment made that if the
decision was made by this committee/FDA
that
further studies needed to be performed,
maybe a
larger multicenter trial to really
determine the
potential answers, that Chiron would not
be able to
do that or was not committed to do
that. I just
wanted to understand your reasoning
behind that
because if you are this committed to
this and you
really, really believe that this really
works, then
why would you throw in the towel
now? I hope the
answer is more than, you know, it costs
a lot of
money.
Let me prop this up by saying that there
236
are other ways of funding studies or
co-sponsoring
studies like that with the NIH, etc.
potentially.
DR. DILLY: The answer is not it costs a
lot of money. The answer is we don't think it is
an appropriate thing to do. Just think through the
practicalities of running another
placebo-controlled trial even in a
multicenter
site.
That involves IRB approval. That
involves
informed consent. Frankly, it would be
inconceivable--you have heard from the
patients
already--that patients would be willing
to go on
placebo in a randomized, controlled
study with this
kind of clinical signal. We believe that that
question is behind us.
What is important is to really
address
some of the questions about the
relationship of the
endpoint. Now, there is no such thing as absolute
certainty in science. Therefore, we think that the
right thing to do is to run a large
group of
patients, prospectively defined, follow
them for a
long period of time so we can really
characterize
the natural history of patients treated
with
237
inhaled cyclosporine so that we can
address those
questions once and for all. So, it is just not the
right thing to do, to do another
controlled trial
before approval.
DR. SWENSON: Dr. Sampson?
DR. SAMPSON: It is actually a small
point.
I just want to follow-up on Larry's
question about endpoints because in the
Chiron
report it says chronic rejection as
manifested by
OB and BOS, and the variable that you
used was OB
or BOS in your primary analysis. Is that right, or
am I just getting caught up in
"and/or's" in this?
Again, it is the same issue, you know,
about
looking for the right variable to
produce the
answer.
DR. CAPRA: The answer is it is and/or.
So, it is OB or BOS or both. I don't know if one
of the doctors wants to comment on the
clinical
relevance of the use of the composite.
DR. NOONBERG: Again, I just want to
reiterate that chronic rejection can
only be
diagnosed in two ways at present,
histologically
238
and clinically. There will be patients that have
histologic OB, and they will have it on
autopsy,
that don't subsequently go on to BOS in
the time
period of the study. However, like I said, there
is nothing magical about 20
percent. They may have
18 percent decline, 19 percent, or 25
percent
decline but they have an episode of
pneumonia. So,
there are also patients that have BOS
but, due to
the insensitive nature, it is not picked
up. So,
really chronic rejection encompasses
both groups of
patients but they are not really
distinct groups.
They have the same clinical
process. It is just a
matter of how it is detected. And, the whole BOS
criteria became necessary because of the
insensitivity of transbronchial biopsy,
not because
of the poor specificity or the thought
that these
patients who had it on biopsy didn't
really
actually have bronchiolitis
obliterans. It is just
that there was a sense that there were
other
patients who also had chronic rejection
that
weren't coming up as OB.
Again, I want to clarify this
question
239
about Chiron's definition of BOS because
it is
really not Chiron's. We used the programmatic
definition developed in 2001 by a
consensus
committee, and it uses a post-transplant
maximum.
As you may remember from the FDA's
slides, the
initial FEV1 is rarely, if ever, your
post-transplant maximum. It usually really doesn't
come for several months and, therefore, it
doesn't
affect the diagnosis of BOS. So, we didn't come up
with any special criteria for BOS. We used the
same programmatic definition that is
used in the
transplant community.
DR. SWENSON: Dr. Proschan?
DR. PROSCHAN: I want to get back to the
monitoring issue because that is still
bothering me
a little bit. It is hard for me to imagine that in
this study that was done at the
University of
Pittsburgh there was no group looking at
the
results on an ongoing basis. I mean, is that what
you are saying, that there was no data
and safety
monitoring board or no group or individual
that was
looking to see whether, for example, the
mortality
240
was in the active treatment arm?
DR. DILLY: There are two points, one
which I would like Dr. Iacona to comment
on, the
one about monitoring of endpoints. One of the
questions that we have been discussing
is just how
much did the people that entered the
endpoints in
the CRs know about the treatment
assignment. I
think there was some confusion in the
way this was
communicated this morning. The endpoints that
appeared in the case record forms, that
appeared in
the data that constituted the analyses
you have
seen, were taken from source patient
records,
source documents. The people managing the patients
were blind to treatment assignment, as
you have
heard from Dr. Iacona. The people transcribing the
data from the source document into the
database
could theoretically have known what the
treatment
assignment was. There is no evidence that is the
case.
However, both we and the FDA went in an
audited the data in the database against
the
original source documentation and it is
a matter of
common understanding that the data in
the database
241
reflect the source documentation. So, we believe
that the data that you see have very,
very little
liability to be biased.
DR. PROSCHAN: No, I am not worried about
that.
What I am worried about is that the study
was stopped on a possibly random
high. That is,
someone is looking at the data, seeing
the results
and then saying, "gee, if we stop
now, look, it's
going to be a p value of 0.007."
DR. DILLY: Remember, the study was
stopped on the date that was
prospectively defined
for the study stop. There was a completion
enrollment date in August, 2001 and
completion of
the study in August, 2003.
DR. IACONA: Correct.
So, Dr. Proschan,
the enrollment period went on for a
duration of
three years. There was a safety and monitoring
committee during those three-year
intervals that
met at yearly intervals, to the best of
my
recollection, and they looked at safety
and the
primary endpoint, and they reported that
safety
events were no different between the
groups looking
242
at
renal function and infectious processes between
the groups, which is of concern, and
they looked at
acute rejection. At the end of the three-year
block the safety and monitoring
committee ceased to
meet.
The patients who were enrolled at that
three-year interval went on to receive
their
two-year course of therapy, and all the
patients in
the study were continued to be
monitored. At some
point, right before the end of the study
in June,
Dr. Corcoran had done lung function
analysis and he
was unblinded. At that point Dr. Corcoran also
looked at survival and it became
apparent that
there was a difference in survival. Up to that
point we had no idea that survival was
different.
I knew that survival was significantly
better in
the aerosol group and I wasn't unblinded
yet. I
was told of this information. I contacted the
members of the FDA, Dr. Albrecht and Dr.
Cavaille-Coll, and we had a
conversation. I also
contacted our IRB. And, the decision was made to
proceed to finalization of the trial in
August of
2003 until the last person received two
years of
243
therapy as prespecified in our NIH
grant.
I would like to make one more
comment in
reference to the differences in AA
gradient between
the donors, between the immediate post-transplant
patients. Since more of the cyclosporine patients
had double lungs you would expect that
their
PaO2/FiO2 ratios to be higher. It may not reflect
anything of the donor per se, just the
fact that
they had double the lung surface volume.
DR. SWENSON: Dr. Hernandez?
DR. HERNANDEZ: I am going to make several
comments. The first is regarding the study
termination.
I have the impression that the study
was terminated early and what I see is
that an
amendment was implemented to terminate
the study
earlier, before reaching the 126
patients that were
initially planned. Because of reasons that were
not in the hands of the investigators,
the study
needed to be terminated earlier, before
126
patients.
DR. IACONA: Yes, Dr. Hernandez, I
appreciate the point. Logistically, we were not
244
able to proceed with the
enrollment. We had
requested additional funds from the NIH
and the
funding supply was exhausted at that
point. We
recognize that we were not able to
enroll the
number of patients as prespecified and
extend the
study.
I made a specific request to the NIH to
extend the funds and they were not
allocated. So,
we were forced to make do with what we
had.
DR. HERNANDEZ: This is a very difficult
thing to me, but I think because I am a
clinician I
would like to insist on this, when I
changed to the
area of transplantation I became a
transplant
surgeon and very committed to my
patients, and I
believe that most transplant patients
are very
educated about what the transplant
is. I think
they are getting explained perfectly
what is going
on and they understand clearly. The thing is that
I don't want to raise false expectations
in my
transplant patients. First, I don't want them to
get the impression that this treatment
is a
calcineurin-sparing drug because there
is no data
to imply that. This drug has not demonstrated an
245
effect on recurring acute
rejection. So, I think
that we need to clarify that with
patients where
they can understand exactly what we are
looking at.
What matters to me is the
function of
these lungs. Histology has several limitations.
Transbronchial biopsy doesn't tell you
anything
about how extensive the disease is; FEV1
does.
Transbronchial biopsy doesn't tell you
anything
about the clinical course of the disease
and, as we
know, chronic rejection can go very,
very slowly
over time and just maintain like that and
not give
too much disturbances. Other clinical courses are
really, really dramatic that go from a
very fast
onset.
What is indicated in all this is that there
are several factors that are not only
immunological
but are also non-immunological factors,
namely,
infections that can, you know, go
through this
vicious cycle where the patient has
rejection, gets
immunosuppression, gets infection, more
rejection
and finally all these stimuli and
calcineurin
damage may lead to chronic
rejection. But what is
really, really important is the effect
on the
246
function of the graft--FEV1 and BOS, and
this has
been addressed by the International
Society of
Heart, Lung Transplantation that has,
you know,
clearly specified. And, when we are looking at the
treatment we have to look at the effect over
time
of serial measurements of FEV1 before
and after the
treatment in order to assess an effect
of this
drug.
DR. DILLY: If I may comment, our
perspective is we endorse--we want to
know the
effect of inhaled cyclosporine on lung
function,
but we would suggest that the cardinal
endpoint is
whether the patient is alive or dead
and, frankly,
you know, whether they are breathing
easily is
secondary compared to whether they are
alive or
dead.
And, we have a very strong signal in favor
of mortality here which makes
interpretation of
some of the other endpoints rather
difficult. That
is why we are suggesting that the right
thing to do
is to benefit from that finding, take
this drug
forward and study it in a much larger
group of
patients where we can really nail what
the
247
trajectory of FEV1 is in patients who
have their
rejection controlled by inhaled
cyclosporine.
DR. ZALKIKAR: I want to make just one
point regarding monitoring the
study. The study
report that was submitted to us
mentioned an
interim analysis that was conducted by
UPMC during
the conduct of the study. However, we did not
receive any report. There was no adjustment.
There was no information about the
boundary used or
anything like that. So, we requested that
information. In response, what we received was
that the interim analysis that was
mentioned in the
study report was really the so-called
final
analysis from UPMC. But prior to that during the
course of the study there was another
interim
analysis that was also conducted. The data was
blinded at that point. University investigators
were blinded at that point and that data
did
suggest a difference in survival,
although not
significant at that point. Again, there were no
formal statistics done on these interim
analyses.
DR. CAPRA: Let me comment on that.
248
Again, Chiron came into the study at the
end of the
trial basically when the results were
already over.
When we had the discussion with Dr.
Iacona he
talked about this analysis, and what he
was
referring to late and which we learned
later was
that this was the analysis looked at by
their data
monitoring committee. There was a misunderstanding
between Chiron and Dr. Iacona over
terminology
between basically the data monitoring
committee
versus formal interim analyses. You know, in the
analysis that was done by the data
monitoring
committee no adjustments were made as
far as we
could tell or Dr. Iacona communicated to
us, and
the study continued as planned.
DR. ZALKIKAR: Also, I want to address an
issue that came up about the definition
of the
endpoints. Although the endpoints may have been
defined, on reading of the document, it
suggested
that the definition was simply survival,
chronic
rejection and lung function in terms of
FEV1.
Also, the data analysis plan was created
retrospectively after unblinding of the
data as to
249
how to analyze those endpoints.
DR. CAVAILLE-COLL: I want to address some
of the comments or some of the questions
by Dr.
Hunsicker. I think that we share with you concerns
about when this data was really looked
at, and we
do know that the study was presented at
the
International Society of Heart, Lung
Transplantation at an open session. So, there were
some analyses done at least before that
presentation. I don't recall the exact date of
that and I don't recall when that took
place with
respect to the involvement of Chiron and
this
application.
We do not have a prospectively
defined
data analysis plan and, certainly, we
were willing
to look at survival because that is certainly
an
unequivocal endpoint. But when it came to the
secondary endpoints, such as those used
to define
chronic rejection, there really wasn't
anything
that was prospectively discussed with us
and that
was defined before the information on
survival had
become public.
250
DR. PRUSSIN: I have a question as far as
the charge to the committee. Do we go over that
now?
Is this a reasonable time to do that?
DR. SWENSON: Let me hold your question.
I have several and we may have one other
but we
will get to that. My question is to both sides
here, and that is that in all of this we
haven't
discussed any issues around the
potential toxicity
over a longer period of time of these
very high
concentrations of drug on the
airway. The animal
data rest largely on one-, two- and
three-month
data.
As I look at those data, I see that
transiently at peak there are 100-fold
greater
concentrations of drug applied to the
airway
epithelium. It does tail off but it still remains
high. And, the risk, as I see it
in this field, is
the issue of malignancy developing and
the problem
of post-transplant lymphoprolific
disorder. So, I
wonder if both sides might touch on this
issue
about this dosing and down the road
possible
adverse consequences.
DR. JOHNSON: Dale Johnson, from Chiron.
251
One of the things that we did, of
course, was to do
toxicology studies in normal animals at
high doses
and look at those particular
findings. What you
actually saw on the listing, for
instance, doesn't
actually describe the severity effects
which are
barely detectable. So, from an acute standpoint it
is fairly clear it is not an issue. Chronically,
there have been studies and they have
been done
with monkeys, and what I would like to
do is to
refer you to Dr. Allan Singer, who is
from Battelle
Institute, who actually has knowledge of
those
studies.
DR. SINGER: I am Al Singer and I am from
Battelle, which is a not-for-profit
organization.
We actually do very little work for
Chiron so I
don't really have a whole lot to gain
one way or
the other here. They are not paying my salary,
Battelle is. They are reimbursing Battelle for my
time for today.
The monkey studies were
actually run with
propylene glycol. Those were run over a period of
13 months at a saturated solution. They caused no
252
effect. But your question really relates to
cyclosporine. The main contribution I can make
here--I was not the pathologist,
veterinary
pathologist, toxicologic pathologist; I
was not the
pathologist in either of those studies
but I did
peer review both of those studies, the
rat and the
dog study. The lesions which were diagnosed--and
there were lesions--there were minimal
inflammations, sometimes mild
inflammations in the
alveoli and in the interstitial areas of
the lungs
and, frankly, it is within the realm of
normal. We
did not run an air control on the dog
study, and
that is simply because we have run so
darned many
air control dogs through the years that
it is just
a waste of dogs. At some point you have to decide
why you are just putting more animals on
a study;
it is just a waste of animals.
The lesions that were seen in
the
propylene glycol controls are typical of
what you
would see in an air control in that
study. The
pathologist reviewed those dog slides
and we cut
35, 40 slides. He spent most of his time on 40X
253
looking for a few cells.
The only thing that we
actually had any
trouble with was the pinpoint ulcers in
the larynx
of some of the dogs, and we typically
find those in
controls as well. It is my belief that it is the
function of the way the mouthpiece works
in the
dog.
The rats are put into a tube and they have to
breathe through the nose. But in the case of the
dog it is actually a tube that goes in
the mouth
and they have to breathe through the
mouth, and I
think that is somewhat drying. But they are all
within normal limits. There are no acute findings
within 28 days to speak of.
DR. SWENSON: Well, I can imagine that
they are typical. What we are talking about is a
concern raised in the literature that
cancer and
malignancies will be a problem with
long-term
chronic immunosuppression, and
cyclosporine has
been pin-pointed as a possible risk
factor for
that.
So, I don't think the animal studies are
going to answer that. Cancer just takes too much
longer to evolve. So, it is going to be an issue
254
only in the humans because what you are
proposing
is to take this out for more than two
years, maybe
five years or the life of the patients.
DR. DILLY: Exactly right, and it may be
worth showing slide PK-26. First of all, there are
few safety endpoints that trump survival
and
patients have to survive long enough to
see the
toxicity. Second of all, what you are talking
about are going to be low incidence
signals. If it
was a high incidence signal it was
likely to have
been picked up by now. Those will only be found
with a significant sized study for a
long time, and
that is why we say that actually to nail
questions
exactly like the one that you are
posing, the right
way to do that is a five-year study
following a
large cohort of patients through treatment
so we
can look at the incidence.
We also have the very nice
reference point
of, for instance, the UNOS database,
where we can
look at the incidence in people not
receiving
inhaled cyclosporine. So, that is why we see in
this context that a postapproval study,
running for
255
a long time to gather this kind of data,
is
entirely the right way to go.
DR. SWENSON: Before you step down, let me
raise just one issue. If, in fact, this committee
votes positively to proceed here, what
assurance do
we have that this study will be
done? The track
record, as was highlighted recently in an
article
in the LA Times last week pertaining to
fast track
approval, and in essence this is
something similar
to fast track approval, and many of
these
promissory notes have not been delivered
and,
therefore, if this does go to approval
it is really
a contingent approval.
DR. DILLY: I will give you three answers
to that.
The first one is that it is entirely
within the FDA's regulations, FDA's own
regulations
to hold us to a postapproval commitment,
not
withstanding whether it is accelerated
approval or
not.
Second of all, this is part of
a much
broader commitment of Chiron to lung
disease, and
we have referred to our treatment of
cystic
256
fibrosis with TOBI. We followed that population.
One of the reasons we got here in the
first place
was patients with cystic fibrosis
receiving TOBI
for pseudomonal infections. We have been very
diligent in furthering that treatment,
for instance
with our drug powder inhaler program
where we
followed through with a program to
optimize that
therapy.
Third of all, and probably the
most
important, is think about how you
commercialize a
drug like this. Trying to get broad adoption of a
treatment based on a single-site study
in 26 active
patients versus 30 control patients is
an uphill
proposition. This is exactly the study that we
want to do to generate publications, to
generate
more data to talk about, and to
underwrite the
widespread adoption because if we can do
the study,
then we can optimize the dosing
regimen. We have
the option to optimize the
formulation. We have
the option to just get it right in the
long-term
for patients with lung transplantation.
In the absence of this kind of
study this
257
could be the sum total of the clinical experience
with inhaled cyclosporine that is
interpretable,
for some of the reasons that we talked
about
before.
So, it is in our mutual interest to do
this study.
DR. HUNSICKER: Just an extension of that,
you argued before that one of the major
reasons for
requesting approval now, rather than
doing a
definitive study that you just talked
about, would
be that it would be both unethical and
impossible
to recruit patients to a study if there
was this
publication, which is undoubtedly going
to be out.
But you just said now that you don't
think you can
commercialize this unless, in fact,
there is better
data.
That would imply that there would be
patients who might be available and that
it
wouldn't be unethical. So, is there a
contradiction here?
DR. DILLY: Not really.
We are talking
about 250 patients over five years
compared to
about approximately 2000 patients
treated worldwide
every year. So, during the conduct of this study,
258
say, we take a year to enroll 250
patients and then
five years to follow them up, that is
six years
worth of treatment and that is 12,000
people
worldwide getting transplanted. It is a relatively
small subset of the overall population
that we
would be putting in this study. We do see that
having a robust, ongoing program of
research to
optimize the drug is the right thing to
do because,
as we said from the get-go, not all the
questions
have been nailed down by the single
study right
now.
DR. ZALKIKAR: If I may just say something
that came to my mind as I looked at this
slide,
PK-26 that the applicant put up, we saw
this slide
and an effort was made to match the
placebo
patients to the UNOS data, but we didn't
see any
effort to match the cyclosporine
patients to the
UNOS data in terms of baseline factors.
DR. DILLY: In fact, you did see that
slide during the presentation this
morning, and the
p value and the point estimate were
almost entirely
the same as in the original analysis.
259
DR. ZALKIKAR: That was a comparison of
the cyclosporine arm in the study to the
UNOS
matched controls.
DR. CAPRA: As Dr. Zalkikar was
mentioning, we did compare the
cyclosporine
subjects to the UNOS matched controls
and we used
the same matching criteria. We matched by
transplant type, CMV match or mismatch,
early acute
rejection, age where we took the ages in
brackets,
and sex and, not surprising--I mean,
this result is
positive given that the comparison from
the placebo
to the UNOS was nearly identical. So, here the
p value is 0.19 and the relative risk
estimate of
0.252 is nearly identical to what we saw
in the
primary analysis of 0.213.
DR. SWENSON: Dr. Tisdale?
DR. TISDALE: I had a question that was
percolating some time ago. We seem to be inferring
from the analysis of this study that it
is all post
hoc and the primary endpoint was not
met. But I
imagine the primary endpoint was decided
upon
because it was felt that it was more
likely to be
260
impacted by this.
But I want to be sure that I
understand
that in the original study that was
written for
three years--it wasn't re-funded,
probably because
the primary endpoint wasn't met and
there was no
difference--survival was in fact one of
the
endpoints, one of the secondary
endpoints, and in
the analysis of one of the secondary
endpoints that
was a planned part of this study there
was a highly
statistically significant difference, and
that is
the reason why we are all here.
DR. IACONA: Unfortunately, I don't have
the statistical section of the NIH grant
with me to
show the audience. It is written as an endpoint of
survival as a secondary endpoint, as is
chronic
rejection and bronchiolitis obliterans
syndrome.
One of the analyses that I did
independently was to
actually look at change in FEV1 from
peak, as is
conventionally done, and calculate
individual
slopes.
I calculated individual slopes between the
cyclosporine and the placebo patients
and, at least
by my independent analyses, I am showing
a
261
difference between the rate of decline
in the
cyclosporine versus the placebo. Now, I don't have
the background that Chiron does, but
that was also
mentioned as part of the analysis. I am not sure
why that is not being presented or
accepted, and I
can show you how I did it
independently. But the
chronic rejection is a secondary
endpoint and we
looked at chronic rejection in our study
by
bronchiolitis obliterans syndrome and by
histological rejection. If one should read through
the transplant literature, our pathology
department
is probably the best in the world in
diagnosing
chronic rejection by histology and we really
emphasize that in our grant, that we are
going to
look at histological OB.
The second point that you made
was what we
should pick as a primary endpoint. Well, when you
look at transplant drugs it is
believable and
acceptable by the NIH, if one should
mention acute
rejection as a primary endpoint and that
was our
thinking, Dr. Griffith's and my own
thinking, that
acute rejection is a conventional
endpoint for a
262
transplant drug. It is taken as an endpoint
because it is a surrogate marker of
chronic
rejection. If I went and I suggested that
aerosolized cyclosporine would be the
first drug
ever to prevent chronic rejection my
grant would
not get funded.
DR. SWENSON: Dr. Burdick?
DR. BURDICK: I would like to just explore
the alternative to the proposed study,
which would
be perhaps a smaller number of patients,
a
randomized trial. I think a lot is being hung on
the perception that around the country
it would not
be possible to do that. I am not totally
convinced. I think there is obviously a great deal
of interest around the country in this
potentially
radical advantage. But I think that a lot of
patients don't complete the course;
don't get very
far in the course. It is a huge expense--there are
a lot of things about it I think. I would like to
hear perhaps more of the argument that a
randomized
trial, much smaller in scale but powered
to provide
capacity to see the difference that we
are hearing
263
is believed this trial gave us from the
data we
have now, why that wouldn't be a viable
alternative.
DR. DILLY: There are several points.
First of all, our analysis of the data,
as we have
said from the very start of our session
today, is
that there is a robust survival advantage
for
inhaled cyclosporine. Therefore, we think the
right thing to do is to make this
treatment
available to patients with lung
transplantation.
Therefore, we actually do not see
withholding
therapy, even in the context of the
clinical trial,
as a desirable way to go if there is an
alternative.
We believe the very strength
of the
available data on the natural history of
survival
after lung transplantation means that we
don't have
to randomize patients to placebo. So, we took what
we saw as a much more sound approach of
saying
let's do a big study rather than a small
study
which will allow us to also look at some
of the
secondary phenomena like the effect on
pulmonary
264
function, like the effect on chronic
rejection, as
well as survival. So, we felt it was a more
elegant way forward than another
placebo-controlled
study and a more appropriate way
forward.
We also had advice from our
advisers that
it would, in fact, be rather
problematic, and we
did our own internal soul searching
around how
would you write an informed consent form
with that
Kaplan-Meier shape. And, drawing on my own
personal experience in oncology where,
in the face
of this kind of survival endpoint, you
do
everything you can to get patients onto
active
therapy.
DR. BURDICK: So, in essence, what the
proposal is--the point that it is
effective is
proven and we are going on from there to
look at
the details. I think that that happens a lot,
especially in transplantation where the
numbers are
small and there are a variety of sort of
opinions
around the country about exactly how to
best treat
patients and you end up with a series of
papers
written showing exactly how this or that
effect is
265
optimized by this or that approach, and
a lot of
good things, and ultimately truth comes
out of
that.
But that would happen anyway if the FDA just
said, okay, approve it.
The trial you propose with an
OPTN
control, which is going to have some
problems--I am
not sure exactly what is in the OPTN database that
would provide you with what you need for
your
control group. Maybe it is good. Maybe you could
address that. But I am concerned that we will have
a series of opinions about details
around the
treatment but it is not going to extend
how
convincing the big picture is, which is
does it
make a difference or not since you are
hearing some
concerns about whether that really has been
proven.
DR. DILLY: Our take on that is that
currently the five-year survival for
lung
transplantation is quite well nailed
down at around
50 percent, and that has not been a
point of major
contention. The control group in the ACS001 study
behaved very like a control group in
external
controls. It has all been about the remarkable
266
trajectory of the patients who received
inhaled
cyclosporine, and what we want to do in
the
confirmatory study is confirm that
remarkable
trajectory. Therefore, we can draw statistical
plans around the point estimate and the
confidence
intervals of one year, two years, five,
and so
forth, and we see that as the most
powerful way of
generating the data that are necessary
to address
this question.
DR. BURDICK: I don't mean to argue this
extensively, but you are caught in the
same
situation as we are in transplantation
otherwise.
By the time you are calculating your
endpoints you
are going to have patients on MMF being
moved back
and forth on rapomyacin and you are going
to end up
with a series of observations that
characterize
certain subsets or certain areas and
seem
convincing, and it is not again going to
provide
you further strength against the critics
who might
say this is very interesting but we
haven't seen
that it is really making a difference
that is
proven.
267
DR. DILLY:
If in seven years time we are
looking at data that says that the point
estimate
on five-year survival with inhaled
cyclosporine
really is 85 percent, it will be
absolutely clear
that that is different from the current
practice.
And, we are confident enough in these
data having,
as you have seen stress tested them
every way, that
we could think of that as appropriate,
that we are
going to see an effect that is clearly different
from past practice and that is what we
are
interested in pursuing.
DR. GOLDBERGER: Dr. Swenson, I was
wondering whether it would be
helpful--the company
has Dr. Golden here and another expert,
if maybe it
would be helpful to the committee to
hear their
views on the types of trials that could
be done
with and without prior approval because
they are
the kind of people who actually have to
do those
trials practically.
DR. SWENSON: Fair enough.
Dr. Golden, do
you want to take that?
DR. GOLDEN: I got most of your question.
268
Do you mind just summarizing it?
DR. GOLDBERGER: Yes, there has been a lot
of talk about what kind of trial could
be done. In
particular, whether a placebo-controlled
trial
could be done at this point in time with
this
information already public and and
probably will be
more public when an article comes out
about it, and
what your take is about that and about
the
situation for further studying this
therapy.
DR. GOLDEN: I have to believe that given
the change in survival in this study
that that is,
(a), going to be sustained in another
study and
(b), it will be very hard to look at my
patients
and ask them to be in a placebo
arm. This is
always the problem with any study. But I happen to
think that in my institution my IRB will
likely
say, especially after a publication,
that this
information is already out there in
abstract form
and otherwise. I have to explain to a patient that
this inhaled therapy, a novel approach,
has been
the first absolute study to show a
change in our
dismal outcome at five years.
269
I must say to the point of
rapomyacin
trials, now at 24 months, it hasn't
shown any
difference in obliterative
bronchiolitis. I think,
as a physician, that I am not going to
be able to
entice patients to go in a placebo group
in the
context of this inhaled therapy.
I would very much actually
like to ask
Bert Trulock his opinion on that
question as well.
DR. TRULOCK: Thanks, Jeff.
Bert Trulock,
from Washington University. I don't have very much
to add to what Jeff has already
said. But
presented with this evidence, I am not
sure that
there is an IRB in the country that
would approve a
randomized, placebo-controlled
trial. Certainly,
at our university when presented with
this kind of
information these days, it is very
difficult for
the IRB to approve a placebo-controlled
trial.
Aside from that, there are the
emotions
that Jeff described in convincing
patients to go
into a placebo-controlled trial when
this
information is already in the public
domain. So, I
don't believe that it is practical to do a
270
placebo-controlled study.
DR. SWENSON: Instead of a placebo arm,
what about the possibility of some lower
dose that
might achieve at the airway level
essentially the
same levels that are obtained with oral
dosing?
This would then offer patients at least
something
active, and it is still a question that
is relevant
because we don't know what the
dose-response
relationship is.
DR. TRULOCK: I can't speak for Chiron,
but I think those kind of dose-ranging
studies are
obviously important, and there are many things
like
that that can and need to be done in a
postapproval
study.
DR. SWENSON: And it doesn't seem as if
you would need 250. If this is really as robust as
it is, simply another trial of 50-75
patients, if
it showed exactly the same--I think the
question
would be moot about survival
advantage. Dr.
Proschan?
DR. PROSCHAN: If I had anything to say, I
have forgotten it now.
271
[Laughter]
DR. SWENSON: Dr. Barrett?
DR. BARRETT: As I was listening,
particularly as the sponsor pointed out
what their
plans were next, I am not so sure that
an
additional confirmatory trial, ethical
reasons
aside, is going to be necessarily
informative here.
There are so many other questions with
regard to
how often this agent should be given;
what dose;
and the disconnect between acute
response and the
chronic response data that you see
here. There is
some disconnect as far as disease
progression. The
FDA pointed out that these two maybe
should be
closer linked and the overwhelming
evidence that
you have a signal here, although
unanticipated, is
very large in magnitude. Those seem to be more
relevant questions that would be
addressed not from
repeating this with a bigger N but with a
study
design that is perhaps more exploratory
in nature.
Maybe just to ask the FDA, you
know, you
put in your slides the expectation of
the
correlation between acute rejection and
chronic
272
rejection but I didn't see any
references there.
Could you comment a little bit more
about what you
would expect to see there?
DR. HERNANDEZ: Yes, this is well
documented about the relationship
between acute and
chronic rejection. This has been documented in
literature and registry data. As a matter of fact,
the sponsor agreed that one of the
factors that is
important to take into consideration is
to take
into consideration acute rejection. Basically, I
would say the natural history of these
patients
will be developing acute rejection;
getting
treatment for this acute rejection and
infections.
So, this meets the category of the
multifactorial
nature of chronic rejection.
If we are looking at a drug
that is
supposed to prevent or decrease the
degree of
chronic rejection, which is one of the
leading
causes after the second year of
transplantation, it
is completely reasonable to look at
long-term
outcomes and look at correlation with
the degree of
disease progression which would be
reflected in the
273
histology of the lung. As I pointed out before,
the limitations of defining chronic
rejection only
by biopsy and giving it weight similar
to BOS, from
my clinical perspective, is not adequate
because,
as I said before, it doesn't have the
ability to
predict the extent of the disease, the
rate of this
disease progression, and those are things
that are
very relevant questions. When we don't see this
correlation between acute and chronic
rejection,
and if chronic rejection is really
prevented, we
would expect to see at least some
evidence on the
preservation of lung function and we
don't see
that.
And that is one of the disturbing things
that put a question mark in this
difference in
mortality.
It is not that we are
questioning that
these patients are not dying. There is no
difference in the graft. The problem is are the
patients in the placebo group dying
because of
other causes, because of the baseline
characteristics of these patients, or
are these
patients surviving because they have
better
274
starting points and not because of the
cyclosporine? I mean these kind of questions when
you don't see these correlations, when
you don't
see that everything is going in the
right
direction, then make you think about
what is going
on over here.
DR. TRULOCK: There has been an
association between acute rejection and
chronic
rejection in most of the multivariate
analyses that
have been done. However, many patients develop
chronic rejection without ever having a
single
episode of acute rejection. Furthermore, just
because there is a statistical
association in
multivariate studies, that doesn't mean
there is a
cause and effect relationship between
acute
rejection and chronic rejection. That is, it is
not a requirement that acute rejection
occurs
before chronic rejection can
develop. All of these
things are probably in the continuum of
allo-reactivity that is not well
controlled by the
current immunosuppressive medications.
DR. HERNANDEZ: Yes, I agree with that
275
fact.
Acute rejection could be not clinically
evident and could be subclinical, and I
agree that
chronic rejection could be developed
without acute
rejection but that is not the rule. I mean, I
understand that, for example, there are
other
factors that could be taken into account
for the
development of chronic rejection and
acute
rejection is only one but the rule, what
you would
like to see, is the patient who had
acute rejection
correlating with chronic rejection, not
the other
way. It is not the rule that the
patient doesn't
have acute rejection and develops
chronic
rejection. That is my point.
DR. SWENSON: At this juncture I think we
probably need to get to the questions at
hand. Dr.
Prussin, you had a question about the
charge and
maybe we could have you ask your
question and then
Dr. Albrecht will proceed with the
discussion to
those questions.
DR. PRUSSIN: Dr. Albrecht, the charge as
written has a fairly binary yes or no
answer to the
question and I think what we have seen
here is that
276
this is a fairly unusual
application. There is a
lot of grey area. We were talking about p values
and the fact that while these p values
give you an
indication they really shouldn't be
thought of in a
quantitative sense. So, I guess my question for
you is if you could elaborate a bit and
clarify--and I know from statute that
approval of a
drug such as this has the same
requirements as
drugs for larger patient populations
that are not
orphan drugs, but how should we manage
that?
Clearly, this is not the typical binary
situation
where we have Ns of thousands, and this
is a
population of patients that clearly has
a huge
mortality. Can you give us a little more
instruction on how to factor that into
our
thinking? Thanks.
DR. ALBRECHT: You raise good points and
you have very well summarized the
challenge before
us.
The truth is our decision does have to be
binary.
As I mentioned in the introduction, we did
have difficulty reaching the conclusion
that would
allow us to make that decision. So we did, in
277
fact, want to bring this to the
committee to hear
your perspectives on this with all the
challenges
that you have identified.
In the event that it is
difficult to come
to sort of a yes or no recommendation,
it would be
very important for us to understand both
where you
believe the data are convincing and
where you
still, in your own assessment of these
data, have
questions and where you believe the
unanswered
questions are.
I don't know if that helps but
I think we
all acknowledge that this is a very
difficult
scenario. I think we heard earlier during the day
about there being promising potential here,
something to that effect, and I think we
certainly
concur with that. I think we heard some analogies
to gambling and I think, being
regulators, we are
not willing to gamble with patients'
health and,
therefore, we take these questions very
seriously
and are very seriously looking into your
perspectives and how much has been
demonstrated;
how confident are we that the questions
have been
278
answered about the efficacy of the
product, mainly
the safety, or how confident are we that
there is
still further information that needs to
be
gathered.
Parenthetically, I know we
talked about
the pharmacology/toxicology
information. I didn't
know if we needed more information on
that, but our
pharm/tox review is available. Just to repeat, the
FDA only received those one-month
toxicology
studies.
I could parenthetically say that for
products that are administered
chronically the FDA
actually does request carcinogenicity
studies.
Those can be done preapproval or those
can also be
done as postmarketing commitments after
approval.
But we do not at this point have an
aerosolized
cyclosporine carcinogenicity study. There were
carcinogenicity studies done with the
systemic
formulation.
DR. SWENSON: Dr. Venitz?
DR. VENITZ: I have a follow-up question,
not regarding the process as much as the
consequences. Let's assume that the committee
279
votes in favor of approving the product
and let's
assume you follow our recommendation,
what are your
tools to enforce a follow-up study as
suggested by
Chiron?
DR. ALBRECHT: As you have heard allusion
to accelerated approval, that is a
regulation that
allows us to approve a product on a
surrogate
endpoint and then request confirmatory
studies. At
this moment, I do not believe that is
the question
on the table. So, in a setting of approving a
product in the conventional fashion what
we can do
is request that a company commit to
conducting a
postmarketing study and we put that in
the action
letter.
We put that in the action letter after
receiving a letter of commitment from
the company.
I don't have the data as far as
postmarketing
studies for what the record of
completion of those
is but, again, our expectation would be
that the
postmarketing study would be done as
committed to
by the company.
DR. VENITZ: The reason why I bring that
up is that somebody mentioned the survey
that was
280
published, I think it was in "The
Los Angeles
Times," and only 50 percent of
those promises are
kept.
I think that is obviously something we have
to consider as we discuss the responses
to the
question. You are basically saying a letter is
going to be written that is going to be
agreed upon
by both parties, but from that point on
there is
nothing that you can do.
DR. ALBRECHT: I am not aware that, other
than having that commitment, there are
other sort
of regulatory steps that the agency has
available.
DR. SCAIFE: May I just comment on that?
I obviously defer to the FDA regarding
legal
requirements but I think it is true,
what you say,
that the commitment that the FDA issues
to a
sponsor is a legal commitment. This is not "would
you like to do this?" This is not an option.
Whether the Office of the Inspector
General, in
other words the chief counsel of the FDA
decides to
enforce, that is really a decision of
the general
counsel.
If you actually look at the trends over
the last few years and if you actually
look at some
281
of the congressional activity, there are
very
strong signals that they are going to
tighten this
up.
If you look, for example, at warning letters,
a warning letter is issued by the FDA
that is
sanctioned by the OIG which is basically
telling
the company you have got to do
something; if you
don't there will be a legal
consequence. If you
look at the trending over the last few
years, those
letters have changed in tenor. So, those letters
now almost invariably are actually
issued and go
through the OIG. That is basically saying an
attorney is underwriting this. So, I think if you
actually want to look at the statistics,
when those
letters are issued those postapproval
commitments
would be enforced. It has a legal meaning. It is
legally binding. Up to now the question has been
whether the FDA has been prepared to
enforce it or
not.
But if you look at the statistics, OIG is
getting a lot more blunt about it.
DR. SWENSON: Dr. Schoenfeld?
DR. SCHOENFELD: Are we at a point now
where people can begin making their
comments on the
282
data, or are we not yet at that point?
DR. SWENSON: We are fast approaching it.
Do you wish to start that? We could go ahead and
focus on question one. You are first.
DR. SCHOENFELD: First I want to talk only
about the mortality endpoint and the
discussion
that we have had about various covariate
adjustments of that endpoint and various
subgroup
analyses.
First, people should realize
that if you
do a randomized study a p value is valid
if you
close your eyes entirely to the patient
characteristics. You don't need to look at patient
characteristics to justify a p
value. What ends up
happening is that if there is no
difference you can
be wrong five percent of the time, and
rolled up
into that five percent are all the times
that you
get a bad break or a good break with
covariates.
So, that is the first thing to
realize. Covariates
don't count. You don't have to look at them.
The second thing is if you do
look at
them, if you decide to split your data
by one
283
covariate and to analyze within each
covariate
group and pool the results, in a
randomized study
the randomization carries over to each
subgroup.
That is very nice. That means that when you split
there is no presumption--let's say you do
a study
and you do have imbalance in one group
and you
decide, well, I am going to split, you
are now
saying I am not going to use the
original p value.
I am going to do another p value, and
that p value
is also valid which, of course, is a big
contradiction in people's minds. How can two
different p values be valid? But they are.
The
fact is that there should be no
presumption that
because they are unbalanced in one
covariate they
will be unbalanced in other covariates
because they
were randomized within whatever subgroup
you view,
and this is something else.
Now, people often expect that
imbalances
will propagate down subgroups because
they are used
to observational studies, and in
observational
studies this isn't true. In observational studies
you don't have that factor that
randomization
284
protects balance through subgroups, and
when you
see imbalance in one subgroup you
presume it is
going to be imbalanced in other
subgroups and you
presume it is going to be imbalanced in
things that
you didn't measure. But a randomized study is not
that kind of study. A randomized study is a study
that if you didn't measure something you
should
assume it is balanced in it and not that
it is
unbalanced in it.
These are just sort of factors
that people
should try to understand. So, I thought that the
survival analysis was robust and the
reason is
that, first, all of these subgroup
analyses or
these stratified analyses are somewhat
post hoc and
you can search for good p values or bad
p values.
You can make just as much of a mistake
trying to
kill a p value as you can to sort of
create a p
value.
So, when you have a lot of them they are
all suspect. So, you end up having to use your
reason, watching them and watching them,
and it is
clear that the FDA did admit that they
were driven
to look for imbalances by another
problem, which is
285
the problem that the rationale didn't
fit. Okay?
Then they were looking for something
that would
throw away the p value, and I think that
that is
kind of not a good idea. So, I think that survival
benefit can't be thrown away because of
all these
various analyses. I think that you have to see it
as standing. So, I wanted to say that first.
Now let's deal with the other
analyses
because I think the thing that the
agency was sort
of concerned about was sort of mechanism
of action
issues.
That is, they said, well, so there is a
survival difference and the mechanism of
action
doesn't seem to work because there
wasn't an acute
rejection difference. This is probably a harder
problem.
First, one of the things is
that I am used
to survival being the primary endpoint
and one of
the reasons that survival is often the
primary
endpoint in studies is that when there
is a
survival difference you never can
analyze for
anything else because what happens is
that the
deaths cause biases. So, it is important to
286
understand this phenomenon. For instance, if there
were three people who were quite sick
and died
right away even if it had nothing to do
with the
treatment, those are three people who
are bad
actors who have been pulled out of the
placebo
group.
Okay? That automatically means
that if you
just remove those people and look
subsequent to
those three deaths every analysis is
biased by the
fact that the patients aren't balanced
in the two
groups.
In fact, this is a problem in
general,
that any analysis that is based on
subsets that are
defined after treatment began is
biased. Those are
problems that you have in really even
analyzing
FEV1, acute rejection, chronic
rejection, all of
those things where you try to remove
survival. For
instance, I was sort of scratching my
head about
the FEV1 because I couldn't really know
whether it
was due to the different amounts of
follow-up time
in the two treatments. This is also acute
rejection. I mean, the FEV1 is particularly
strange because actually the treated
group had much
287
better FEV1 during the whole course of
treatment,
but this could actually be a benefit of
treatment
but we don't know because we don't have
a baseline.
Actually, it is not clear that baseline
is even
that meaningful. So, it is difficult to analyze.
So, when you look at these secondary
endpoints, if the deaths that occurred
without
these secondary endpoints had nothing to
do with
the secondary endpoints, then those
analyses would
become unbiased but nobody really
believes that is
true.
So, we have a bunch of biased analyses, and
most of them actually would be biased
towards
placebo basically because I think it is
reasonable
to assume that the people who died would
tend to be
the bad actors and that would bias
FEV. It would
bias acute rejection, and so on.
Anyway, when I sort of count
those up,
FEV1 was negative. Acute rejection was negative.
Chronic rejection was positive including
both
endpoints. So, I think the situation really is
that we have a fairly robust survival
benefit and
our problem really simply is that we
don't really
288
understand this acute rejection well
enough to
really use it.
DR. HERNANDEZ: Well, my main concern is
acute rejection, of course, but my main
concern is
this, if I am following a patient with
FEV1's and I
have a drug that is going to graft
rejection I
would like to see that this FEV1, which
is the
physiology of the lung, to be preserved
regardless
of at what level they start, whatever
the baseline.
And, what I would like to see is that if
the group
that has not preserved that function, I
would like
to see the natural history of decline of
these FEV1
declines over time. But if I see these two graphs
to be parallel and the slope analysis
doesn't show
any significance and the slope is
similar, then I
don't see any preservation of the
function.
DR. SCHOENFELD: Those graphs are biased
anyway because in one group people are
being culled
out by death and in the other group
people aren't
being culled out so fast. So, a longitudinal graph
over time where you are just looking at
the
marginal comparison isn't actually valid. It isn't
289
a valid statistic. That is, in other words, if you
look at the FEV for all patients at,
let's say, two
and a half years you have six or seven
people who
are not in one group and who are in the
other
group.
That is, the only valid analysis would be
somehow to subset the analysis by the
people who
would have been alive on both treatments
if they
had had both treatments. You can create models to
estimate that but they are all models
and nothing
is perfect. But just looking at the graph isn't
really a fair comparison.
DR. HERNANDEZ: The only thing is not only
looking at the graph but also to look at
the
incidence of BOS as defined by the
sponsor. Really
this is something that is bothering
me. If you
look at a drug that is preserving the
lung
function, that is preventing chronic
rejection, you
need to see a benefit in the function of
the lung.
That is something that you would like to
see. I
mean, you can't say to the patient I am
going to
give you this drug that is going to
prevent chronic
rejection but I am not sure the lung
function is
290
going to be improved or is going to be
maintained.
That is my main concern.
DR. ZALKIKAR: I want to address a couple
of things. There was pre-enrollment FEV1 available
in the patients and the final FEV1
available in the
patients and we took the difference in
each patient
between the final and the pre-enrollment
FEV1, and
even those differences didn't show any
significance.
DR. SCHOENFELD: But that was already
subsetted. That was only about half the patients
who had pre-enrollment FEVs.
DR. ZALKIKAR: That is right. Of course,
we had to impute the missing data.
DR. SCHOENFELD: I mean, I think it might
be reasonable to say--I mean, I really
am
supporting this application now but I would
think
it is reasonable to say that there is
something
that we don't understand here. Okay?
And, that is
true but I think that then the question
is do we
have to understand everything in order
to approve
drugs?
If things were completely the other way, if
291
there was no difference in survival but
you did see
a difference in FEV we would be in the
same
situation in a sense. We would still have the
question of, well, how come it produced
a
difference in FEV and how come it didn't
produce a
difference in survival? Especially with a small
study, we can't understand
everything. Even with a
big study we tend not to be able to
understand
everything.
DR. HERNANDEZ: My problem is this, if you
are going to give a drug it does need to
have a
clinical effect. If you are giving this drug to
prevent chronic rejection, this drug
should have a
clinical effect.
DR. SCHOENFELD: You mean a physiological
effect.
DR. HERNANDEZ: Exactly.
DR. SWENSON: We should move on to other
members.
Dr. Hunsicker?
DR. HUNSICKER: First, I would like to
suggest, Dr. Swenson--I may be totally
out of
order, but if we are going to go back
and forth
292
between either the FDA or the other side
arguing
each point, we are never going to get
anywhere, and
I would really suggest that we limit the
discussion
for the time being to the committee
members unless
we have specific questions of the other
people.
DR. SWENSON: I agree.
We need to move on
to a vote and I would like to at least
open this up
to the rest of the panel members here
for any other
questions or observations before we make
that vote.
DR. HUNSICKER: I have a technical
response to the question from--I don't
know what
your name is--
DR. SCHOENFELD: David.
DR. HUNSICKER: David.
There is, in fact,
a way to look at the FEV1 data that
hasn't been
done so we can't use it today. You can do a mixed
model analysis so long as you include the
last
measurement that was associated with
what caused
the failure. And it has been shown that that
unrelieves the bias of informative
censoring.
DR. PROSCHAN: That is not true. I mean,
it depends on whether it is missing at
random
293
versus not missing at random.
DR. HUNSICKER: We can discuss this later
on but there are additional analyses
that could be
more informative than what we have.
DR. PROSCHAN: That is true.
DR. HUNSICKER: I would like to give my
approach to this which I offer up
because it is
really opposite to yours, Dr.
Schoenfeld. Where we
start out in this application is that
there was a
very well formulated hypothesis that use
of this
agent would affect acute rejection. We didn't find
that.
We find now no effect on that.
So, we start
with the usual presumption that if you
have
something that turns up in a secondary
analysis
this is considered hypothesis-generating
and needs
to be then retested.
I can practically quote
Flemming on this
who says that if you have a secondary
outcome that
is significant or if you have a subset
analysis
that is significant, almost irrespective
of the
p
value for it, if you repeat it only about a third
of those are confirmed when they are
then repeated
294
because there is an effect of regression
to the
mean.
You have chosen them because they were
further, more deviant from the mean.
So, we start here with why we
should not
or should look at this as being
different from the
usual circumstance where the primary
outcome was
not recognized, was not achieved, but a
very
striking secondary outcome has been
found either in
a subset or with a secondary thing.
This really is related to Dr.
Helms'
comment where he says that this is
sufficiently
strong that it strikes you between the
eyes. That
is true.
There is a strong thing here. So,
how do
you evaluate this kind of a thing when
it was not
the primary question?
There are two ways that you
look at this.
First of all, does this have a credible
underlying
hypothesis? Now, I would say that with respect to
the hypothesis that local administration
of an
immunosuppressant agent might be
effective in
preventing chronic rejection is credible
but it is
certainly not something for which there
is any
295
experimental data. This is something that is
really de novo. It is not impossible. It is a
rather attractive hypothesis but it is
not based on
the same kind of data that we have for
the
assumption that immunosuppression would
affect
rejection in a broader sense.
So, that is a weakness in the
hypothesis.
Then the second question, and this is
why I do
believe that looking at the relationship
between
the deaths and all the other stuff is
relevant--we
have a circumstance where there is no
question; it
is absolutely clear that there were
fewer deaths in
the group that was treated. That is not a matter
of statistics; it is a matter of fact. So, then we
can ask ourselves is this difference in
fact
something that was the result of the
application of
the treatment or is this a fluke, if you
will.
This is a small study so flukes are more
possible.
We have the problem that I referred to
before of
having sort of chosen this to bring
forward because
it was an attractive, unexpected
finding. That is
why we are here. That is not true for Dr. Iacona.
296
I don't mean to imply that he is in
quite the same
situation. He put out the hypothesis that there
would be a good outcome. But we are looking at it
because this is one of the unusual
circumstances
where you have a really striking p value
in an
unexpected circumstance. So, we have to deflate
the p value for that.
Then we look at the
circumstances that
deal with the question of whether, in
fact, it is
credible that this thing is affecting
chronic
rejection and, thereby, is preserving
people's
lives.
That the lives were preserved is
unequivocal. Then one question that could come up
is, well, there was no difference in
acute
rejection and here I have to say that
what I
consider to be an appropriate analysis
actually
favors the company rather than the
FDA. There is
no reason why acute rejection should be
the
prerequisite for chronic rejection. As the FDA
probably knows from other circumstances,
I have
argued in the past that what we call
chronic
rejection is generally a mixture of
immune and
297
non-immune events. Everybody has agreed upon this,
and you don't ever quite know how much
of this is
immune and how much is non-immune. In fact, even
earlier failure, whether it is graft
failure in the
case of the kidney or death in the case
of lung
transplantation, could be the result of
a process
added to something, what I have called
in my area
which is nephrology interceptant [?]
slope. If you
start lower you are going to get to your
endpoint
sooner than if you started higher and it
has
nothing to do with the rate at which
chronic damage
is being done. So, there may be a difference in
chronic rejection or fibrosis that has
nothing to
do with acute rejection.
On the other hand, what we
have seen is
parallelism with all of the limitations
of the
methods that we are using--parallelism
in the
outcomes in terms of the FEV1 and other
things like
that.
So, at the bottom what I wind up with is
that this is a very promising therapy
but it is not
a therapy that is now established. This is in a
way very strong hypothesis
generating. So, has
298
this reached the status of substantial
evidence
from well designed and executed clinical
trials, or
something like that? But I don't think we have
gotten there.
I do want to say one last
thing, which is
the Helm's comment. I find myself worried about
the idea that we would be chased into
approval
because if we didn't approve nothing
would ever
happen again. I think that would be a very bad
precedent because then all that a
company would
have to do in coming to this group is to
put you in
the position where, if you don't approve
the drug,
nothing is ever going to happen. There are lots of
ways in which this drug can be studied
further. It
is not this group's business to figure
out what
happens if we don't approve it. The question we
have to deal with is whether it has
reached an
approvable status or not.
DR. SWENSON: Dr. Gay?
DR. GAY: We have been given a challenge
that our cardinal task is to consider
alive or
dead.
There are significant things, in my opinion,
299
that compromise our ability to decide
this. We
cannot underestimate the potential bias
that occurs
with the fact that there are more
patients who
underwent double lung transplantation in
the
treatment group as opposed to the
placebo group.
We have different criteria for listing
for double
lung transplantation than we do for
single lung
transplantation. The patient population is
younger.
There can be potential selection bias
because the wait list time tends to be
longer with
the double lung transplant
population. Thus, if
they can get to transplant they tend to
be
potentially healthier at the time that
they are
transplanted with a lack of other
potential
problems and medical issues that could
compromise
their survivability.
I am concerned that this puts
into this
process a profound amount of bias, and
my concern
is, as optimistic as I would like to be
for this to
be a therapy that we should consider,
that the
explanation of the change in
survivability can't be
easily explained by that and that alone
as opposed
300
to factors that we cannot determine with
the lack
of change in FEV1, the lack of change in
the
presence of acute rejection, and with
this I do
feel that a significant amount of
increased
investigation is going to be required to
make
absolutely sure that the difference we are
seeing
between alive and dead are secondary to
the
therapy.
Because once we move to approve our first
therapy for lung transplantation I think
we can
guarantee that it will be utilized in
full force
and we must be cautious in doing so.
DR. SWENSON: Dr. Proschan?
DR. PROSCHAN: You know, this is a very
difficult decision for me. One of the things I was
struck by is that really small trials go
both ways.
That is, in this particular trial you
would think
that because it is a small trial and you
are seeing
more placebo deaths, then those people
who are
dying are the sicker patients. Therefore, the
remaining patients in the placebo arm
should be
healthier than the ones in the
cyclosporine arm.
So, I would expect in a small trial,
even if you
301
see a benefit at first, that the curves
would tend
to come back together because of that
survival of
the fittest phenomenon--you know, you
have only the
fittest surviving in the placebo arm and
everyone
surviving in the other arm. So, I would expect
them to come back together and the fact
that they
didn't come back together is something
that I think
influences me when I try to think about
is it real
or
not.
The statement has been made
that no IRB
would allow another randomized
trial. Earlier than
that a statement was made that if you
were on the
DSMB you would definitely vote to
declare treatment
benefit.
I think that is not true at all.
I think
that a DSMB would be struggling just as
we are and
that it would not be a slam-dunk. In the cardiac
arrhythmia suppression trial there were
19 events
in one arm and 3 in the other arm and
they didn't
even want to be unblinded yet. They said whichever
way it went, it is too early to
tell. So, that was
19-3 and they didn't even want to become
unblinded.
So, I think the idea that this
is a
302
slam-dunk and the question has already
been
answered is not true. I struggle with this and I
do agree that the survival result seems
to be
robust to various sensitivity analyses
and, you
know, that makes me feel a little
better. I agree
with Dr. Schoenfeld. That does make me feel a
little better.
Then the question is what
degree of
evidence is required. If it is like a civil trial
where you just need a preponderance of
evidence
there is no question in my mind that
that has been
shown.
If it is like a criminal trial where you
need proof beyond reasonable doubt, I am not sure
that that has been shown. So, to me, it is
somewhere in between those two and, you
know, the
question becomes how much evidence do
you need.
DR. SWENSON: Miss Drittler?
MS. DRITTLER: I am a lung recipient and
this is sort of getting away from the
questions
presented to the committee but what I
wanted to
know--the first year and a half after my
transplant
I went through approximately three
rejections, one
303
very serious one. Of course, nothing was available
at that time. After that year and a half I have
not had a rejection but, as we all know,
rejections
can occur at any time. If I were, say, in my
fourth year to encounter this problem,
would the
inhaled cyclosporine be beneficial to me
at that
time?
Because I am assuming that these people who
were in the study were put on the
inhaled
cyclosporine immediately after
transplant and
followed for the two years. So, I want to know is
it beneficial to somebody--Dr. Suss stated
that she
was put on it a year and a half or two
years after
transplant. Is this going to be the case, that it
will benefit those of us? If I hit my five-year
survival rate, which I don't believe in,
will I be
able to take it to prevent or stop
rejection?
DR. SWENSON: Well, I think that is a
pretty hard question because it wasn't
studied in
that fashion. Maybe someone from the company might
just spend a minute or two, not any
longer, just
answering her question if you think you
have
something to either agree or to put a
different
304
light on it.
DR. DILLY: Our belief is that we are
preventing or slowing the trajectory of
chronic
rejection so our hypothesis, which is
worth testing
and we have already seen the RAR data
being very
provocative, is that the drug would
benefit
patients already some way down the road
from their
transplant. But that is not the basis of ACS001.
MS. DRITTLER: Thank you.
DR. SWENSON: Dr. Barrett?
DR. BARRETT: I just wanted to again
reiterate that this was really not
designed as a
registration trial. It is a Phase II trial, not
even with the benefit of a formalized
statistical
analysis plan prior to proceeding. So, applying
the conventional statistical criteria,
as has been
done by both the agency and the sponsor,
makes it
very difficult and I think is the cause
of all our
unhappiness as we sit here and see the
data
summarized.
Having said that, I do believe
Dr.
Schoenfeld's assessment of the
forgiveness of the p
305
value given--maybe short-sighted--but
the attempt
to randomize patients as was done in
this trial.
However, I also feel the way the FDA has
done their
subset analysis was absolutely relevant
in order to
define which criteria are in fact
sensitive to
these results. Because of that, it makes it very
difficult. I believe that the signal is, in fact,
valid and robust but I don't know if I
would feel
comfortable generalizing the magnitude
of this
effect from this one trial in a
subsequent study.
DR. SWENSON: At this juncture I think we
should go ahead and take the vote,
unless there are
any committee members that feel that
they really
need to say one more thing.
DR. VENITZ: Yes, one more.
DR. SWENSON: All right.
DR. VENITZ: I am not a statistician but I
am trying to use informative decision
analysis to
approach the problem that we are facing
because I
think everybody would agree it is a
close call.
The way I approach things like that that
are a
close call is that I am going to look at
what is
306
the likelihood that we are right or
wrong and what
are the stakes. In other words, I do what Helms
suggested. I am a rational gambler. And, I think
to some extent, whether we admit it or
not and that
includes the FDA, we are rational gamblers
when we
review evidence to decide what the
significance is
both in terms of the statistics as well
as the
consequences.
So, I think, like most of you,
I am
obviously not convinced that this is beyond any
reasonable doubt for proving that the
mortality
benefit that was demonstrated was due to
the
treatment as opposed to something else
that we
don't know about, no matter how much we
try.
However, I do think that it is
plausible, and I do
think that it meets the civil trial
standard. It
is the preponderance of the evidence;
not beyond
reasonable doubt but preponderance of
the evidence.
So, there is some 30 percent likelihood
of coming
up with the wrong answer, that we find a
difference
that really doesn't exist.
Counteracting that is the fact
that it is
307
a large difference in terms of the
mortality. That
is all I am talking about. And, counteracting that
is the fact is what is the competition
that we are
trying to beat right now? There are no drugs
approved. It is a dismal disease. Right now we
have a 50 percent mortality rate with
the current
treatments. Maybe there are other ones being
investigated. So, I think the stakes are high and
I am willing to take a chance on the
likelihood
that this might be a study that doesn't
prove
beyond any reasonable doubt that the
drug actually
works.
So, in my final summary then,
I would
conclude, even though I have this doubt
about the
likelihood that this is real, the
stakes, to me,
favor saying that this is sufficient
evidence.
That is a word that is in the question
here, I
think there is sufficient evidence to
say that the
survival benefit is due to the treatment
that
patients might benefit from.
DR. SWENSON: We will go ahead and begin
the voting. I will ask Miss Schell to give us your
308
vote.
MS. SCHELL: I am Karen Schell and I am
the consumer representative. I would just like to
add one comment before I vote. Looking at the
question on sufficient evidence, I am a
practicing
respiratory therapist and I also do
perform FEV1
pulmonary functions on a daily basis,
and I think
those objective measurements are very
important in
a
study. And, because of the limitations
of this
study I think the study needs to be
bigger. I
think that whenever there are unanswered
questions
more research needs to be done. So, I am voting as
a no.
DR. SWENSON: Dr. Sampson?
DR. SAMPSON: It is obviously a very, very
difficult decision. I have listened to my
statistical colleagues and my clinical
colleagues
and have wrestled with this more than I
care to
admit, and I think all I can conclude
from this
single trial is that the observed
mortality
difference may be due to treatment. There is the
other uncontrolled data, the ACS002,
that is
309
supportive but it is certainly
uncontrolled.
My basic problem, as has been
stated by so
many other people, is if it is due to
treatment the
question is why, and what has been
demonstrated to
cause the survival difference. I don't see strong
enough evidence, certainly not in acute
rejection.
In fact, at least by the sponsor's own
analysis,
acute rejection rates are higher in
the CyIS
group.
And, I think there are a lot of other
statistical difficulties in the analysis
of the
chronic rejection data, both in the
definition of
the endpoints and the analysis.
DR. SWENSON:
Your vote then is no. Dr.
Schoenfeld?
DR. SCHOENFELD: I guess I will be very
brief.
My vote is yes.
DR. SWENSON: That was brief. Dr.
Proschan?
DR. PROSCHAN: To me, what you want is
compelling evidence to offset any
potential harm
that there might be that is hard to see
based only
on 100 patients on the safety data. So, my short
310
answer would be no, that it is not
sufficient.
DR. SWENSON: Dr. Barrett?
DR. BARRETT: I am going to key in on two
words, "sufficient" and
"survival" and because of
that I am going to say yes.
DR. SWENSON: And I will reserve mine till
the end and move on to Dr. Moss.
DR. MOSS: The thing in making the
decision for me is that based on the
small study
the outcome of five or so patients, if
they had
gone the other way is going to change
the standard
of care for a thousand patients each
year based on
that outcome of just a few. I think there is some
precedent with the gamma interferon
trial, with the
steroid trial for ARDS of two small
studies that
when the companies were able to go ahead
and do
larger trials, it showed that those
things were not
true.
On the flip side, with the ventilator
strategy for ARDS the paper in The New
England
Journal of Medicine it showed that the
ventilator
strategy worked and work now shows that,
in fact,
311
it was true. So, I think it was possibly due to
the studies. But I don't think we should change
the practice based on a small study
where a few
patients can weight the whole outcome so
I would
say no.
DR. SWENSON: Dr. Burdick?
DR. BURDICK: I sort of approach this
question in the setting of the thousand
shades of
grey that were dealt with by something
that others
have said, and that is where is the
greatest harm?
I think the tradeoff we have is
believing the
survival or believing just about
everything else
that would make sense about it. I think, in my
view, the greatest harm would be to not
proceed
with approval. So, it is a yes.
DR. SWENSON: Miss Drittler?
MS. DRITTLER: Well, of course, as a
patient I have to look at the survival
rate and the
application that it has on chronic
rejection, and
my vote will be yes.
DR. VENITZ: I have said my piece. My
vote is yes.
312
DR. SWENSON:
Dr. Hunsicker?
DR. HUNSICKER: I have said my piece and
my vote is no.
DR. SWENSON: Dr. Gay?
DR. GAY: Considering all the factors
involved, understanding that this is a
therapy that
will be adopted across the board,
understanding the
limitations of the sample size and the
potential
for a skew on this, I must vote no.
DR. SWENSON: Dr. Mannon?
DR. MANNON: With due respect to the
thoughtful presentations by both Chiron
and the
FDA, and I also empathize with the
patients for
coming here, as a solid organ
transplanter, you
know, facing life and death is a
difficult issue
but, be that as it may, my vote is
against
approval. Again, it is because, to me, though the
survival effect may seem to be a
statistically and
a reality-based issue, it is not clear
to me what
caused that and it is not clear that the
drug
itself was responsible, particularly
when only half
of the individuals on that treatment
completed the
313
treatment.
DR. SWENSON: Dr. Proschan?
DR. PROSCHAN: I usually like to think of
things in mechanistic terms, but in this
disease we
don't really understand the mechanisms
enough and I
think ultimately the uncertainties are
much greater
but ultimately it appears relatively
safe. We
haven't seen evidence of toxicity. There is a
potentially huge upside so I will vote
yes.
DR. SWENSON: Dr. Tisdale?
DR. TISDALE: Well, I am basing my
decision on the fact that this was a
secondary
endpoint in the study, that it did come
out in the
original planned analysis of the
study. There was
a clear survival advantage. I am not saying that I
believe that there is a connection
between the drug
definitively and the survival advantage
but, when
considering everything on balance, I am
going to
have to vote yes.
DR. SWENSON: Dr. Brantly?
DR. BRANTLY: Having been involved in
treating individuals with rare diseases
for the
314
last 20 years, I crossed this bridge
about 20 years
ago actually when I first started. My vote is yes
for the reason that I believe this
offers probably
the best promise at the present time and
has
sufficient data to support that.
DR. SWENSON: My vote is no, not
overwhelmingly no but for reasons that
have been
stated here, I don't think it is proven
and I think
that the issue could possibly be
resolved faster
than what has been stated in a smaller
trial. If
this is really as strong an effect as it
is, it
should be borne out very quickly. So, I don't
think the evidence is yet there.
What this means is that we
have a tie
vote.
Dr. Albrecht, how do we proceed here with
the second portion to this question?
DR. ALBRECHT: I think next time we will
invite an odd number of consultants--
[Laughter]
If I may ask you to please
move to the
second part of the question, I think it
would be
very helpful for us for those who
believe the
315
evidence is sufficient to please address
the yes
part of question 1(a). For those who feel that the
data are not in, to please give us
suggestions
about additional studies. So, please discuss
either option yes or option no.
DR. SWENSON: What I would ask then--and
we will go through the order just as we
did--is
that those voting yes please answer (a)
and those
voting no to offer their
suggestions. So, Miss
Schell?
MS. SCHELL: I voted no and I would like
to have more data concerning the donor
and also
post-operation, some of the other
factors that
would be sufficient like the FEV1 and
less decline.
Basically, I would like to know more
about the
donors' lungs before they were donated.
DR. SWENSON: Dr. Sampson?
DR. SAMPSON: I voted no and, as I started
to say before, I think there is a
struggle and I
think there are certainly very strong
arguments as
to why CyIS would work. But to do another two-year
treatment study with a survival endpoint
and
316
reasonable follow-up would be a lengthy
undertaking. I would encourage the agency and the
sponsor to work together to design a
prospective,
randomized, blinded trial but with a
more
innovative design. I don't know if that means
doing something like low dose versus
high dose to
increase enrollment, as has been
suggested, or
introducing a reasonable surrogate
endpoint for
survival that could be measured earlier
than
waiting for it two or three years post
treatment.
I have heard arguments pro and
con why
FEV1 might be such a surrogate. I think it is
really critical to identify and agree
upon
important baseline variables, and
perhaps even
employ some sort of stochastic, dynamic
balanced
randomization scheme that would try to
keep all
these variables between treatment groups
reasonably
comparable and maintain the blinding so
that we
would not have a discussion like this
again.
As is usual in regulatory
studies, the
study protocol and the SCP should spell
out the
objectives very clearly; the primary
variables for
317
analyzing these; and the primary
analytical methods
for the primary variables. I also think that this
is the kind of study, given the kind of
difference
that we have seen up to now, that would
benefit
from some sort of sequential design
which would
allow for an early stopping if, in fact,
the
differences persist of the magnitude
that we are
seeing so that a quick decision could be
reached.
DR. SCHOENFELD: I won't be as
brief this
time.
I am going to look--
DR. SWENSON: I should say that is Dr.
Schoenfeld. We need to keep that for the record.
DR. SCHOENFELD: I am sorry.
I think that
in regards to 1(a) it is sort of a
difficult
situation because this study would
indicate that
the biggest difference created by this
treatment
was survival. I mean, this would be the thing you
would want to look at as the thing that would be
the most likely endpoint to show a
difference in a
subsequent trial. As I have said before, even
analyzing any of these other endpoints
that would
give you a good idea of the mechanism is
very
318
difficult in the face of a big survival
difference.
This was the big difference, by the way,
in the
ARDS.
We still don't know why 6 mg/kg works.
Because, in fact, all the oxygenation
parameters
and all the various parameters don't
really show a
difference between 6 mg/kg and 12
mg/kg--I am
referring to the study of low versus
high pressure
ventilation for ARDS. None of those mechanistic
things work because you can't really
compare them
because of the difference in mortality.
So, I think that this leaves a
big problem
for someone coming to a future trial
because their
best shot is mortality and the other
things are
going to be probably fuzzy, like they
are now. I
think that is something to realize. So, the big
advantage they will get in doing another
trial is
that mortality won't be a secondary
endpoint and
there will be now two trials, and maybe
a
multicenter trial so they will get some
of these
other things dealt with.
So, I think that in terms of
1(b) you end
up having to repeat the trial and it
would sound
319
very funny if you then argued--you know,
you
wouldn't have any of those secondary
endpoint
arguments at that point.
I don't know how you could
balance it
because I am sure that, no matter how
carefully you
stratified the trial, if people were
really trying
to knock it down by finding different
covariates
you could find some. So, I don't think there is
any way of perfectly balancing a
trial. So, it
would be a confirmatory trial and that
is probably
what should be done if you want to go
forward.
Now, in terms of 1(a), this is
always a
problem, how to generalize clinical
trials either
from one institution or from--so I just
answered
1(b) and I guess the answer to that is
you are
going to have to do another survival trial,
bigger
and multicenter.
In terms of generalizability,
that is
always a problem in clinical trials,
generalizing
from the few patients you treated to the
patients
you didn't treat. This trial randomized about five
percent of the patients in the United
States that
320
have this disease. That is, you randomized about
five percent of the people which is much
better
than we did in ARDS network. So, I don't know how
you can answer these questions. Nobody knows
whether you can generalize. Usually what you end
up doing with clinical trials is you
close your
eyes and you generalize.
DR. SWENSON: Dr. Proschan?
DR. PROSCHAN: For what additional
information would be needed, to me, I
think you do
have to do another randomized
trial. I don't think
you can get away with a single-arm trial
where
mortality is the primary outcome. And, I don't
agree with a comment made earlier that
it could be
much smaller. I think that is an invitation for
disaster again because then you could
get a p value
that is 0.10 or something and you might
also get
the baseline imbalances. So, I do think it has to
be a larger trial, perhaps multicenter.
I am not as concerned about the issue
of
generalizability. To me, the issue of
generalizability is, you know,
University of
321
Pittsburgh does a lot of these and maybe
they have
learned something about how to deliver
this
aerosolized cyclosporine and maybe other
centers
that are just starting might not know
that. So, to
me, that is the issue of generalizability. Is
there anything special that Pittsburgh
learned and
other centers that try to do this might
learn the
hard way? So, to me, you have to do another trial
where survival is the primary outcome, a
randomized
trial.
DR. SWENSON: Dr. Barrett?
DR. BARRETT: I voted yes.
However, as I
mentioned before, I think there is very
little that
you can generalize from the previous
study, and for
all of the analyses that the FDA showed
there are
several factors which make the existing
data,
primarily due to the small sample size,
very
suspect.
One of the most important
things I thought
was the idea that the acute rejection
doesn't
necessarily correlate with the chronic
rejection.
And, one of the things that I am more
suspicious of
322
is the maintenance of the survival
benefit in the
face of limited dosing and/or
exposure. It simply
may be that we don't understand how to
dose that
particular facet of rejection
progression. It is
an interesting finding. I don't know if I would
walk back into the quicksand and do
another
survival study though. I think there are better
studies to do now with the idea of
really figuring
out how to dose this agent.
I think there was a rationale
for inhaled
cyclosporine at the very beginning that
prompted
this trial, which I think is probably
still there.
There is, for sure, a regional delivery
advantage,
however, it is probably not maintained
by
conventional pharmacokinetic behavior so
we simply
don't understand what is the local
exposure
requirement in order to maximize the
benefit of
this route of administration.
So, as far as studies go, I
think a
combination of preclinical models as
well as
different dosing scenarios where you
consider low
and prophylactic regimens to be more
beneficial.
323
DR. SWENSON: I voted no, and I have
mentioned my thoughts here. I think another trial,
maybe two to three times larger than
this one that
might incorporate several centers and,
of course, I
think with the problems that develop in
the design
and then ultimately coming here could be
corrected
from the start. That is, we would have all the
preliminary data. There would be no question about
loss of data in a prospective
study. I think
another study of that magnitude would be
what would
be necessary, and possibly a stopping
point could
be agreed upon early. If, in fact, this major
survival advantage persists the study
could be
concluded earlier on the basis of that,
given what
we have seen with this study. Dr. Moss?
DR. MOSS: I really don't have a lot to
add to what Erik just said. I think that the
results need to be confirmed in a larger
multicenter study where the study is
performed more
rigorously and, of course, add some
longer-term
outcomes that won't preclude people from
voting yes
to approve the drug, but that that would
be kept in
324
mind in terms of carcinogenic
possibilities.
DR. SWENSON: Dr. Burdick?
DR. BURDICK: The big issue in
generalizability is what was observed
before, that
about two-thirds of the patients in this
trial
weren't on up-to-date
immunosuppression. So, for
all programs, including Pittsburgh, I
think that is
a question that remains in terms of
using this
regimen today.
I think the other major issue
is that if
this is effective, if the results of
this trial
haven't misled us due to happenstance,
or whatever,
then there needs to be some inventive
thought about
how to look at it with things other than
biopsy,
FEV1 and other things which have not
correlated
with the outcome.
DR. SWENSON: Miss Drittler?
MS. DRITTLER: Again from a patient
viewpoint, I feel that what I am reading
in the
data with the survival rate, the safety
problems
don't seem to exist. I just think that having been
very sick for ten years, hoping
something would
325
come along as many patients are doing,
grasping at
straws actually, this seems to be
something that
could be very profitable to the lung
transplant
population. And, I just feel that it is an
important thing at this particular
moment not to
preclude necessarily further
studies. I think that
could be necessary but in the meantime
there will
be patients dying who might profit from
this.
Thank you.
DR. SWENSON: Dr. Venitz?
DR. VENITZ: Obviously, I voted yes so I
am down to "yes-(a)" I
guess. Confirm efficacy;
establish mechanism of action; expand
safety; and
optimize the dose. Those would be the outstanding
issues in spite of the fact that I did
vote yes.
Confirm efficacy in a larger, more
heterogeneous
population. I think mortality is definitely going
to continue to be an outcome that is of
interest
because you have to come up with
labeling language
that reflects whatever the committee
recommendation
is and the FDA final judgment on this is
going to
be, but I think also an assessment of
mechanism of
326
action would look at FEV1's. It would look at the
relationship between some of the markers
of disease
progression relative to survival to
understand what
actually happened in this study. Expanding safety,
I think that is question number two, is
to look at
long-term safety five-year safety
data. It was
already mentioned that there is a
concern about the
incidence of cancer as a result of
chronic
immunosuppression even though you would
think, due
to the low systemic exposures, that is
minimized
but, nevertheless, you would like to
have some
empiric data.
Lastly, I don't think the
dose--and I
agree with Dr. Barrett on that, I don't
think this
is the optimal dose, as it usually is
not. I would
like to believe that it is a safe and
effective
dose but I don't think it is an optimal
dose. So,
I think in addition to efficacy, safety
and
mechanism of action, additional dose
optimization
studies are appropriate. In particular, what I
would be interested in would be the
dosing
interval. Right now I don't know whether the drug
327
has to be given three times a week. It could well
be that you could give it once a
month. It could
well be that you don't have to give 300
mg and
expose patients to the hassle and the
inconvenience
of being bronchodilated, getting
lidocaine and then
sitting in front of a nebulizer for ten
minutes.
So, I think there are some dosing
questions that
should be addressed in clinical as well
as possibly
some of the preclinical studies as well.
DR. SWENSON: Dr. Hunsicker?
DR. HUNSICKER: I was a no voter. One of
the committee members down the line
there, whose
name I can't see, suggested that we
should not
exclude the possibility of preclinical
studies. I
am, for one, not absolutely sure that we have
an
established mechanism that we are
working towards.
It may be that it is obvious to some
that locally
administered immunosuppression should
work. It is
not so obvious to me. So, I would not forget the
possibility of doing preclinical studies
that are
really properly designed to see whether
topical
administration of this in the lung
transplant
328
setting--which you can do and it is
really unique
in the lung transplantation that you can
do
this--is effective in preventing chronic
rejection.
I think this is a very interesting
question and
that shouldn't be forgotten.
With respect to the clinical
trial, it
seems to me very straightforward, as has
already
been said, that the next trial should
have
mortality as its major endpoint. That is where we
are, and if the next trial shows an
advantage in
mortality it doesn't make any difference
what else
is shown. That confirms it. At this point, this
is now starting from "we think that
this might
work" and that would just nail it
down. But I
would hope that that trial would have
defined in
advance very well how to at least look
at the
issues of fibrosis and function, these
being the
two major things that are involved in
chronic
rejection of lung, as well as the other
organs.
So, I think of things like the
endobronchial biopsy
that was being discussed--some knowledge
of what is
actually happening to the lining of the
bronchi
329
which is where the fibrosis is going on
would be
very helpful in eventually being able to
analyze
this.
Finally, with respect to the
study size, I
would just caution you that virtually
always when
you have a striking effect like this the
next time
you look at it the effect is much
smaller. So, do
not assume that you are going to get this
size of
effect the next go around even if this
is a real
thing.
You have to power the next trial
sufficiently to be able to detect a
difference that
would be convincing to you and important
to you
irrespective of it is the size that we
see in this
trial.
So, I think it is going to wind up being a
fairly large trial.
Then you get into the issues
of can you
ameliorate the ethical quandaries by
having a
different randomization scheme, 2:1 rather
than 1:1
and things like that. These are well within
Chiron's and Pittsburgh's ability to
figure out and
I don't think it is going to help us any
for me to
go on about it. So, basically, it needs a new
330
clinical trial and I would love to see
some basic
studies.
DR. SWENSON: Dr. Gay?
DR. GAY: I voted no because of the
problematic nature of attempting to
define
mortality in the transplant population
and the
causes for mortality. Unlike the oncology studies
where mortality usually results from
progression of
the primary malignancy, with factors
related to
both rejection and infection and side
effects from
the immunosuppression, I believe
mortality becomes
very hard to power for. Where a significant impact
may be present is the fact that we have
no therapy
whatsoever that intervenes on chronic
rejection,
and if we can have a proven theory that
affects
BOS-free interval, if we can show that
you can go
longer without developing BOS with a
therapy as
opposed to without it, I think you will
find a
therapy that will have significant
positive benefit
and a reasonably easy course for
possible approval.
So, I would attempt to put together a randomized,
controlled trial that would deal as the
primary
331
endpoint with BOS-free interval in
enrolling a
patient population post-transplant.
DR. SWENSON: Dr. Mannon?
DR. MANNON: Again, I was a no vote. To
reiterate, I think a proper prospective
collection
of data, so a randomized trial, and also
with
balance of the two critical components,
that is
double versus single and acute rejection
rates. I
think that is a big sticking point for
me in the
analysis of the initial data.
I also think it would be
important to have
a spelled out standardization of
immunosuppressive
strategies whether it is one center or
multicenter,
as well as standardized strategies for
patients
with acute rejection so that all
patients are
receiving similar therapies. Then, if they fail
those therapies there are appropriate
outcomes.
Obviously, a formal safety
data monitoring
board, with an appropriate stop point,
the endpoint
being survival. But I also think, as Dr. Hunsicker
pointed out, an opportunity to collect
mechanistic
data as additional information,
including
332
standardized time points for FEV1 and
bronchoscopy,
but also maybe localized as well as
systemic
measurements of allo-immunity because
you have a
small group on specific therapy and that
may
actually shed some light even if the
study is not
huge.
Finally, I think we need some long-term
safety data. There was a discussion of potential
malignancy from this drug, but
cyclosporine also
stimulates TGF-beta and whether this
local effect
may in the long term actually promote
fibrosis
indirectly, which we don't know, is
something else
that I think needs to be answered.
DR. SWENSON: Dr. Prussin?
DR. PRUSSIN: Most of my comments have
already been addressed. In terms of
generalizability of the study, I think
the
population that is being looked at could
be more
ethnically diverse than the small study
in
Pittsburgh, and that probably would be
addressed if
it
is a multicenter study.
Also, the whole issue of an
inhaled
333
formulation in terms of SOPs for
nebulization. I
gather that inhaled lidocaine is not
approved so
that whole usage and how that is
delivered should
certainly be standardized.
DR. SWENSON: Dr. Tisdale?
DR. TISDALE: I was a yes but, as I said
the first time around, my decision was
based mostly
on a risk-benefit assessment. Here the risk seems
to be extraordinarily low and the
benefit is
possibly extraordinarily high. So, that weighed
me.
I sat right on the fence the whole way around
the table until it came to me, and the
whole
morning as well.
But when it comes to the
question of
generalizability, that is the yes-man's
question.
Certainly the control group correlated
well with
the experience that is out there, the
broad
experience for transplant. This represented a very
sizeable fraction of the transplant
patients over
that period of time. So, any new therapy that is
tried, for example in bone marrow transplant,
is
always a very small fraction. A hundred things are
334
different about each protocol so in this
I was
actually encouraged by the fact that
there was a
big percentage of patients that were
studied and
that the control group looked just like
the broad
experience of transplant patients. So, I felt like
there was at least some generalizability
of this
information.
We have had a lot of people
bring up the
issue of the collection of the data
being
problematic in this study and,
certainly, the study
wasn't designed as a pivotal study for
FDA approval
of this drug so there are a lot of
problems with
the collection of the data but the
endpoint that we
are looking at, which is survival, was a
secondary
endpoint and this is not a subtle
endpoint. I
mean, you are alive or you are dead so
that is
simple to look at. And, whether you look at it a
week after the study closed, a year
after the study
closed, there is no arguing about
whether or not
there is a survival advantage. So, to design
another randomized study with a survival
endpoint
is the study that we already have. It was a
335
randomized study with a survival
endpoint, and the
survival endpoint was statistically
significant.
So, you know, what is
necessary for me to
move it on--you know, not everyone is
going to
adopt this. People will look at the study when it
comes out and realize that there are
problems with
the study and not everyone is going to
do it.
Certainly, it requires further
validation in
another study I think which doesn't
necessarily
need to be, in my mind, a randomized,
controlled
trial.
There are many different ways to get at the
questions that still remain beyond the
survival,
and the survival can be compared, I
think, to the
existing survival data because we have
heard over
and over again that this survival curve
has not
changed in more than a decade.
So, I certainly feel
comfortable with
comparing whatever is seen in a
subsequent trial to
the existing 50 percent survival
rate. I don't
think that is likely to change in a big
way, and if
the next study is a negative study
everybody is not
going to be giving inhaled
cyclosporine. It is a
336
pain for the patients; it is a pain for
the
placebos. They have all demonstrated that it is
not something that people are going to
wildly run
to without further information. So, I think a
subsequent trial that is not randomized
is
certainly sufficient to get at the
questions that
weren't addressed in the original
randomized,
controlled trial.
So I agree that the things
that need to be
addressed next is that we need a study
that
confirms the efficacy on survival. That is easy
enough to do in a study that is not
randomized. We
need to better establish the safety. That is easy
to do in a study that is not
randomized. And then
optimize dosing, which I think would be
very
difficult to do in another randomized,
controlled
trial.
So, those are what I would see as the ways
to move forward on whether or not this
is actually
generalizable.
DR. SWENSON: Dr. Brantly?
DR. BRANTLY: I was a yes voter, and I
think that, number one, my greatest
concern with
337
the study was the fact that there was an
imbalance
in the double lung/single lung
transplants and that
that plays strongly into survival
benefit.
As far as generalizability, I have
questions about how generalizable these
studies
are, and I believe that that should be
expanded in
a good controlled study in the
future. I would
recommend basically having four arms,
which would
be double lung, single lung, a low dose
versus the
standard dose, which would answer some
of the
questions regarding both the ethics
since we really
don't know what the dose is in this
particular
study group. I think it would be acceptable to the
patients and probably get to the answer
about
survival.
DR. SCHOENFELD: I just want to say
something because this keeps coming up
as double
lung and single lung. I think that the
stratification by double lung and single
lung as
was done in the analysis pretty much
shows that the
benefit occurred in both groups, and
when you
control for single and double lung the
relative
338
risk stays exactly the same and the
estimates stay
the same. So, to consider that as a major problem
with the study I think is not borne out
by the
data.
I think that the FDA would even agree with
that, that that isn't the primary
problem with this
data.
Their stratified analysis on that simple
stratification was still highly
significant. If
that was the problem, it would just go
away, the
significance.
DR. ZALKIKAR: The model that included
just the single versus double lung, yes,
the data
was significant, just that one factor at
a time.
DR. SWENSON: Dr. Albrecht, hopefully, we
have satisfied your charge here on this
question.
It looks like it is a split and I don't
think the
yes/no are really that far apart. We all have
concerns and we all have, obviously,
great hopes
for something like this to work and hope
that a
speedy answer could be obtained. Shall we move on
to our second question?
DR. ALBRECHT: Yes, please, and I think I
will elaborate a little bit on it to see
if I can
339
probe for some more comments and
advice. The
second question is fundamentally focused
on safety,
specifically, has the safety of the
product been
adequately characterized for its
intended use?
I know we have heard some of
these
comments already during the past
discussions, but
in these deliberations please consider
both the
amount of preclinical information and
the clinical
information that is available on the
administration
of cyclosporine, as well as the vehicle,
through
this route, and the number of patients
that have
been exposed to the inhaled cyclosporine
in this
application at the proposed recommended
dose.
Here is where I would like to
just
elaborate a little bit more, if I can,
which is if
you answer yes to this, and especially
this in
addition to question one, here is where
I would
really like a lot of comments from you,
if
possible, about what you would envision
in the
labeling, I think both the positive as
well as what
kind of cautionary or negative language
you might
include.
So, specifically for what population
340
would you suggest, if you are advising that
the
product be approved, that it be labeled
for what
patients perhaps should receive the
product and
where the data may in fact be limited or
absent as
far as patients that perhaps shouldn't
at this
point be advised or recommended to
receive the
product.
Also, it would be very
beneficial if you
could comment on the dosing regimen, the
preparation, the administration route,
the dosing
intervals and the duration, again basing
this on
the information that we have in the
application.
How comfortable would you be summarizing
the
information in labeling and, again, what
kind of
cautionary information do you believe or
would you
recommend that we consider putting in
the labeling
regarding what isn't known about the
product.
Also, if you could, because I
think as we
have heard and has been stated, the
difference in
mortality that has been presented is fairly
dramatic and, at the same time, there
are not
differences in some of the other
endpoints like
341
acute rejection, BOS, FEV1--how much of
this
information would be valuable in
labeling; how much
of the information might not be ready
for labeling
based on the comments you made about
need for
corroborative data, and so forth.
So, I would like to hear as
much
discussion as you can give to the
"yes" part of the
question. I notice that we were getting
recommendations both on "yes"
and "no" so on the
"no" side it would be really
useful--and I know we
heard a lot of comments about
traditional studies,
both efficacy and safety and dosing and
preclinical, but if you could just
elaborate in any
way that you can specifically on study
designs. We
heard some suggestions for randomized,
some
non-randomized, but if you could perhaps
give some
further details about study designs that
may be
coming to mind as you go around the
table.
DR. SWENSON: Well, we could do it in the
same fashion as before, a vote, but
maybe as much
as we have talked already, if the
committee would
be willing just to go ahead and take
both in turn
342
individually. Do I have any strong dissent for
that?
Just for fairness, we will start on the
other direction and I will ask Dr.
Brantly to start
off.
DR. BRANTLY: Well, if yes, I would argue
based on the study I read--I would say
the
population is lung transplant. I would make it no
more specific than that.
As far as the dosing regime
and dosing
intervals, I think they have to stick
closely to
what was done in the original
trial. At the
present time we have no other
information.
Regarding 2(c), I believe that
this can
only be labeled as having a survival benefit.
There is no evidence to suggest that the
other
things were changed by this drug.
DR. SWENSON: Dr. Tisdale?
DR. TISDALE: I think I would have to echo
almost identical statements. If yes, only for
prolongation in survival in recipients
of single
and double lung transplants.
The same dosing regimen should
be listed
343
in
the labeling even though it wasn't completely
followed by all the patients.
There should be I think no
expected
benefit with respect to the other
endpoints since
they weren't shown definitively to be altered
by
the cyclosporine inhalation.
I think there also should be
caution that
there is no safety data regarding
carcinogenesis
effect on lung scarring or that sort of
thing. So,
I think there should be caution that
there is
incomplete data on carcinogenesis in
laboratory
animals and in humans, and that it is
too early to
know in the recipients that have been
treated with
this regimen whether that is a concern.
DR. SWENSON: Dr. Prussin?
DR. PRUSSIN: Along the same lines I
guess, you could put for the dosing 300
mg three
times a week as tolerated. In terms of duration,
obviously the study was for two years
and so you
would put in some caveat that use of
this has not
been shown to be of value beyond the
two-year
point.
344
There is the whole quandary of
the
lidocaine use and how do you put that
into the
labeling since that is an integral part
of the drug
but is not an approved drug itself. I think, other
than that, that is the only unique
thing.
DR. SWENSON: I should step in here and
say that another party has entered this
position
about lidocaine and I am not sure that
100 percent
of patients had to have lidocaine as
pretreatment.
It
was a large fraction. Could the company
just
comment on that? Did everyone get lidocaine as
part of this or was it only as needed?
DR. NOONBERG: It is my understanding that
most, if not all, patients were
initiated on
inhaled lidocaine without butyryl. However, how
long they continued with it was
variable. We don't
have exact information on duration but
certainly it
wasn't uniformly continued.
DR. SWENSON: So, this probably should
just be a point for further discussion
and
decisions about how lidocaine might fit
in this and
whether it needs its own labeling. Dr. Prussin?
345
DR. PRUSSIN: Obviously, these people have
been followed closely but, because this
has been
used in such small numbers of patients,
something
about follow-up and close follow-up and
long-term
safety has not been established.
DR. SWENSON: Dr. Mannon?
DR. MANNON: From the perspective of
long-term safety, we do now have a
patient
population that is available that has
been on the
therapy for a number of years, and
perhaps we will
be able to incorporate them insofar as
giving us
some of the more long-term data rather
than relying
on a dog model or a rodent model where
the dosing
is different. You know, the opportunity to keep
dogs around for five years may not be
feasible.
I am not really sure how to
address the
tolerability of dosing. It wasn't clear to me what
was the cut-off. Were patients having either
sufficient bronchoconstriction that they
had to
stop, or coughing or wheezing,
whatever? But I
think that we need to have a better
understanding
of the tolerability of dosing and what
is
346
appropriate pre-medication. As we alluded to, it
is not clear to me what the dose
escalation
guidelines are. It sounds as if you got 100 mg and
you tolerated it, then you went to 200
and by 10
days you got 300. But I think that those
guidelines need to be defined and it is
not clear
to me whether there was additional
intolerability
or bronchospasm or cough or wheezing due
to a
higher dose, or whether it was the
vehicle itself.
It would be helpful I think to
know what
the tissue levels are. We have DTPA and
scintigraphy that suggest that there are
very good
levels but it would be nice to know what
the actual
levels are because I am not really sure
how you
dose the drug. We are talking about can we get
away with one time a month. I mean, I think we
need to know what the level of the drug is
and get
a sense of the outcome of the drug. Since
obviously systemic levels appear to be
lower by
inhalation, I am not exactly sure how to
do that.
DR. SWENSON: Dr. Mannon, could you
clarify which patient population?
347
DR. MANNON: It would be lung transplant
recipients. You know, how would you label it?
Preventing death? I mean, I think that would be
really the only thing that you could
honestly put
in the label. Clearly there was no prevention of
rejection. There clearly was no prevention, based
on the data presented, of BOS, and there
may have
been an effect on OB but it is not clear
to me that
it was that, or whether you can say
these are the
limitations of this on the label.
DR. SWENSON: Dr. Gay?
DR. GAY: Most of my comments actually
echo Dr. Mannon's. I think short-term safety seems
to be reasonable. I have concerns about long-term
safety issues, and with the fact that
this will be
a long-term drug I think that needs to
be studied
more fully and evaluated more fully.
It would have to be labeled
for lung
transplant patients and at this point,
yes, clearly
for improving survival. But with future studies it
will probably be more for prevention of
chronic
rejection.
348
Dosing regimens have to be
standardized at
some point. With the concern that half the
patients did not receive really
long-term therapy,
greater than 24 doses, that has to be
looked at
more closely with potential long-term
risk of
therapy associated with the disorder.
DR. SWENSON: Dr. Hunsicker?
DR. HUNSICKER: First, the question put to
us is, is the agent safe? I just have to comment
that if the survival benefit is real the
drug is
safe with reference to its benefit. But if the
survival benefit is not real, then we
have
substantial issues, or if the benefit is
much
smaller than it seems we have
substantial issues
with local tolerability because, as has
already
been commented, it is striking that half the
patients couldn't tolerate it for the
long term.
With respect to what should be done
further--
DR. SWENSON: Let's get your vote. Yes or
no on that?
DR. HUNSICKER: I would vote, since I
voted no on the first, that I don't
think that it
349
is shown to be safe relative to its
known efficacy.
If the efficacy is repeated and is
there, then
obviously it is safe because being alive
is better
than being dead.
DR. SWENSON: And this is a no vote?
DR. HUNSICKER: It is a no vote for the
question or whether that issue has been
answered.
With respect to what more should be
done, first of
all, I think that there does need to be
some more
animal data specifically with respect to
the lung
toxicity of cyclosporine, and it is
striking that
there is no really long-term animal data
on the
tolerability of cyclosporine. That should be easy
to be done. It isn't a definitive thing but it
certainly is something that should be
done along
the way.
I am one who believes that
things like
cancer can really only be answered in
postmarketing
surveillance. I think it has to be done as part of
postmarketing surveillance, along with
all sorts of
other bad things but I don't think there
is any way
that kind of question can be answered
prior to its
350
approval.
With respect to question one, for which
population, I agree with everybody that
it should
be lung transplants.
What information should be
included on
dosing, we have only had one dosing
regimen and,
therefore, I think we can only talk about
one
dosing regimen.
I do think that for duration
we should say
that this has only been shown to be
useful up to
two years. I think we can add on that there is no
reason why the sponsor could not give us
evidence
about tolerability after two years. But right now
what we have is data on tolerability,
which seems
to be actually sort of marginal, up to
two years
with half the patients going off of it
at that
time.
What information should be
included on the
labeling, I have sympathy with those who
say the
only thing you should say is that it
prolongs life,
however, it seems to me that to say that
it
prolongs life but we don't have any
information
351
about what it does to chronic rejection
or acute
rejection is a little naive. I don't see how we
can say that and not sound silly. So, if the
agency is going to approve it, we should
say that
the outcome is that it prolongs life
presumably due
to suppression of chronic
rejection. I think that
would be the way I would phrase it. And, I think
that you need to have an explicit
statement in
there that there is no evidence that it
affects
acute rejection because I am afraid that
some
people will think that this is something
you should
use for acute rejection, and if there is
anything
we do know is that it does not seem to
affect acute
rejection.
DR. SWENSON: Dr. Venitz?
DR. VENITZ: Again, I am a yes-man so I
would vote in favor of safety having
been
demonstrated. I agree that the population would be
the population of lung transplant
patients at
large.
I don't think there is any evidence for
subpopulations at a particular risk but,
hopefully,
that is a question that can be addressed
in the
352
follow-up study that we alluded to.
As far as the dosing regimen is
concerned,
I think the dosing regimen that should
be labeled
should be the dosing regimen that was
used, and
that does include the initial escalating
dose study
to get patients up to 300 mg before they
then start
their three times weekly maintenance
dose.
Something that I found
interesting is that
one of the patients during the open
session
mentioned that, because of seizures, he
was put on
phenytoin and he developed a
rejection. Obviously,
phenytoin is a known enzyme inducer
which increases
the clearance of drugs such as
cyclosporine and
lowers their systemic levels. So, as far as drug
interactions are concerned, I do think
that you
have to stick with the current
cyclosporine label
for oral use even though we think the
systemic
experiences are low and unlikely to
contribute to
this effect.
I also have some concerns
about the fact
that at least the studies that we looked
at were
done with the same nebulizer. So, I think you are
353
looking at a fixed drug-device
combination here. I
don't think we can extrapolate that
beyond any
other devices since I have no clue what
the aerosol
dynamics--the mean diameter for
example--might be
in a different device. So, I think the device is
pretty much locked in, and you have
already heard
some of the comments about the lidocaine
briefly.
As far as 2(c) is concerned, I
agree with
my neighbor to the left. I believe they have
demonstrated survival benefit. They have
demonstrated no benefit on acute
rejection and,
therefore, maybe a benefit on OB.
A couple of additional
comments, being a
kineticist by training, I guess I miss
some
exposure measures in the preclinical
studies that
allow me to compare--and I am talking
about
pulmonary exposures, not drug levels or
PG
levels--some of those studies in the
lung so I can
make some extrapolations as to what that
might be
in humans. Is there any way that we can use some
of those biopsies from lung to actually
measure
drug levels and to ascertain in humans
what the
354
relative ratios of systemic exposure
over topical
exposure are?
On the last question that Dr.
Albrecht
wanted us to address about the study
design, again
given the fact that I am a yes-man on
both
questions, I don't see any reason why
the follow-up
clinical trial could not be an
open-label trial. I
happen to be a member on an IRB and I
would think
that most of my committee members,
probably as a
consensus, would not approve a
randomized,
placebo-controlled trial with this
product. So, I
do think an open-label trial ethically
is more than
justifiable, and I think most of the
information,
at least that I am looking at,
especially as far as
dose optimization concerns, could be
gotten out of
a non-randomized, open, prospectively
designed
trial.
DR. SWENSON: Miss Drittler?
MS. DRITTLER: On 2(a), I don't know what
else to say except transplant
patients. What
information should be included on dosing
regimen--I
question whether each patient might have
a
355
different dosing regimen as opposed to
what has
been shown in the study with the 300 mg
for all
patients. I just wonder if each patient could be
different, and in that case it would
have to be as
directed.
Included in the labeling
regarding
expected benefit--so, that has not been
proven. I
agree that it should refer to
longevity. As far as
the carcinogenic impact, I pay little
attention to
that because everything I take has
carcinogenic
impact so we know that we are at greater
risk for
cancer when we are taking these drugs
but that is
one of the benefits of still breathing
and living.
That is it.
DR. SWENSON: And your answer to the
original question is yes?
MS. DRITTLER: Yes.
DR. SWENSON: Dr. Burdick?
DR. BURDICK: Well, presuming
approvability, the answer to the first
question is
yes.
It would be for long allo-transplant
recipients.
356
For 2(b) I think it would be
important to
note that it would be used in the
setting of
standard systemic immunosuppression,
perhaps with
an insert noting that the data were
essentially in
patients on azathioprine, which is not
going to be
the standard practice now.
For 2(c) it should say that
this may
improve survival. It is not expected to be
beneficial for acute rejection, and
leave out
anything else.
Can I just mention that
postmarketing--you
know, the FDA has one advantage here
that you have
built in 100 percent postmarketing
surveillance
through the OPTN. So, that is different from
looking towards approval of other drugs
where you
sort of lose control of things a little
more, and
that is an important dimension to keep
in mind.
DR. SWENSON: Dr. Moss?
DR. MOSS: I kind of agree with Dr.
Hunsicker. I am not as concerned about the safety
as whether there was really efficacy
demonstrated
in this trial. So, I would say no but I thought
357
the safety data was reasonable.
I really don't have a lot to
add to what
the other people said, except I think it
is a good
idea that there is already a cohort of
patients you
can follow-up for long-term follow-up.
There was some data about
irritability and
irritation of the airways and that could
just be
followed up better in a better
controlled,
randomized, multicenter trial.
DR. SWENSON: I will vote yes on this
question and all the points that I would
raise have
already been raised by preceding
members. Dr.
Barrett?
DR. BARRETT: I will vote yes to the main
body and for the population of lung
transplant,
like has been mentioned earlier.
With respect to parts (b) and
(c), I
think, again if we are going to
recommend approval
based on a single Phase II study, the
description
of the experimental findings unique to
this study
have to be clearly stated. However, in addition to
that, I think there has to be
information that says
358
explicitly that dose response has not
been shown
with this product, and I think it is
going to be
very important to decide how much of the
oral or
other route of administration of
cyclosporine gets
added to this label because I think we
are going to
have to make some choices. Dr. Venitz pointed out
that in some cases the drug interaction
piece will
be very pertinent, but with others I
don't think
you can confer some of the information
that is
contained with the other routes of
administration
to this product. We are specifically fighting the
issue of very limited patient exposures
with this
product and I think that is where the
primary
safety liability comes from.
Having said that, on (d), even
though I
answered yes, I think we still need to
do more
follow-up studies with respect to the
propylene
glycol as well as the product itself in
longer term
tox studies.
DR. SWENSON: Dr. Proschan?
DR. PROSCHAN: On the safety, I don't
really feel like I have a lot to say on
the safety.
359
I would feel a lot more comfortable with
more data
than is available so far. I think that if there is
an additional randomized trial that is
larger, that
would go a long way toward establishing
safety.
So, to me, I am not convinced that the
safety has
been adequately characterized.
I guess I would label it for
survival and
chronic rejection-free survival for lung
transplant
patients.
In terms of additional information, I
sort
of changed my mind on what kind of study
would be a
good follow-up study if, in fact, this
is not
approved. So, the question is what would you do
next if it is not approved. I think that the
essential uncertainty here, in this
group, is due
to the fact that is one study; that it
is a small
study; it is a single institution study;
and that
the endpoint was this unexpected endpoint. It
seems to me that, given the actual data,
you would
choose a follow-up trial using the most
powerful
possible endpoint and that is basically
chronic
rejection-free survival. That had the lowest
360
p value.
If you use that as an endpoint you would
have the additional advantage that if
patients then
rejected, had a chronic rejection, you
could give
them drug so you wouldn't actually
necessarily have
to treat people until death. Treating people until
death is a very unpopular way of running
a study.
DR. SWENSON: Dr. Sampson?
DR. SAMPSON: I actually don't have much
to add to what has already been
said. I voted no
on number one and I am ambivalent about
the amount
of safety data. I would certainly like to see more
information on dosing to understand the
relationship of the severity and
occurrence of the
adverse experiences that the patients
experienced
relative to dose.
DR. SWENSON: Miss Schell?
MS. SCHELL: My vote is no also, mostly
because of the limitations of the study,
including
the variable dosing when the patient
fell out and
didn't take medication anymore, and also
the length
of the study, the long-term
effects. So, it is no.
DR. SWENSON: Well, that concludes the
361
official questions and I will turn it
back to you,
Dr. Albrecht, for any further thoughts
that you
might want to pose to us. We are a little ahead of
schedule--no one is going to complain,
but we do
have a bit of time for other pressing
issues to be
raised by members of the committee.
DR. ALBRECHT: If I could just ask a
follow-up question, and I know I have
heard a lot
of discussion on this but I wanted to
actually just
specifically ask this one more time in
case there
were further comments. As you heard, Dr. Dilly
proposed that the company, in the event
of
approval, would commit to conducting an
open-label,
multinational study of 250
patients. I know going
around the table I heard different
opinions about
open-label studies, not comparative
studies. I
wonder if perhaps we could go around the
table to
see if the committee members would
believe this was
enough, or this in addition to
comparative studies,
randomized studies, dose-ranging
studies. Could I
ask the committee to sort of think about
the
proposal versus other studies that I
sort of think
362
I heard mentioned, either dose-ranging
studies or
some other randomized studies?
DR. SWENSON: Well, if I could at least
try to paraphrase the company's plan, it
was 250
patients. Certainly it would be a broad number of
patients that would address issues of
confounders
and all of that, but there really wasn't
any
explicit discussion of dosing either as
to
frequency per week or absolute amount
per dosing.
So, those questions sort of remain. I am sure the
company probably doesn't know exactly
what the
answers are. Please?
DR. DILLY: We explicitly suggested 300 mg
inhaled cyclosporine three times a week
or the
patient maximum tolerated dose. So, our explicit
suggestion was--
DR. SWENSON: Just the way the first study
was performed?
DR. DILLY: Yes.
DR. SWENSON: Okay.
DR. HUNSICKER: But it says for five years
and, at this point, we have no
information on five
363
years and at least a presumption that
you don't
need five years since the people who got
short
courses seemed to have benefited
similarly. So, I
would not agree with five years at this
point.
That is something that you might look to
change as
you accumulate more information.
DR. DILLY: One consideration here is we
put in a single dosing element, if you
like, 300
mg, three times a week for five
years. It is
perfectly within the realms of
appropriate design
to take a 100 mg arm into this
design. It is
perfectly appropriate to have a two-year
cohort and
a five-year cohort so we can actually
test.
Because, you know, we do agree that
providing
patients with access to potentially
efficacious
therapy is entirely appropriate at this
stage so
comparison of 100 versus 300 would be a
very
appropriate way to go; a comparison of
two years
versus five years we would see as both
addressing
our concerns about the ethics of
withholding
therapy, but also answering some of the
key
questions around safety of long-term
administration
364
and what is the right dose for these
patients. So,
we are willing to refine this design.
DR. SWENSON: Rather than go around the
table, I will just look to anybody with
comments.
Dr. Hunsicker?
DR. HUNSICKER: There are two issues and,
therefore, I am not quite sure how to
answer your
question. One is what, in my mind, would be needed
for approval given that I voted no. Then, the
second is what is needed by the
community, assuming
that this does eventually become
approvable for
intelligent use. We have to keep these things
separate.
With respect to approval, what
I want is
more evidence that, in fact, there is a
real
benefit here. I would actually agree rather
strongly, surprisingly perhaps to some,
that it
need not be necessarily a
placebo-designed study.
I would be willing to accept, because
there has
been no change in chronic rejection if
there is no
change in chronic rejection over the
next little
while, an active treatment study against
a
365
registry, contemporaneous registry
control. I
don't think that it really is needed to
set it up
as a placebo-controlled trial. So, that might ease
the issue of doing another study if, in
fact, the
drug does not get to be approved.
In terms of its intelligent
use, I think
the outstanding issue is the amount of
the drug and
the duration of the drug. Actually, what is to me
a rather attractive design is what has
just been
discussed which would be 100 mg, or some
other dose
that was designed to be somewhat similar
to what
you would have given with oral dosing as
opposed to
respiratory dosing, versus the higher
dose. If you
get a dose response--you know, if 300 is
better
than 100 then you have a clear answer to the other
question as well.
The second is duration. I personally
would like to see a trial that has two
years--how
shall I say?--if you are going to
approve, I would
approve for two years and then let them
design a
trial to take it to five years and see
if there is
an additional benefit.
366
DR. TISDALE: I just wanted clarification
on your question. I think your question was,
assuming approval, what study design
would you
envision as being appropriate as the
next step, or
did I understand you wrong?
DR. ALBRECHT: I apologize.
I said that.
Actually, I was thinking that the
company's
proposal was for a postapproval study
but I was
more interested, because the company had
proposed a
study that was an open-label study
comparing to the
registry, whether that type of study
would be
convincing to the committee or whether,
when I
heard discussion around the table about
randomized
studies, in fact, the committee felt that
while
that study might be informative,
randomized
studies, in fact, were what I was really
hearing as
the preferred approach to gaining
further
information, regardless of whether it be
preapproval or postapproval.
DR. TISDALE: Well, I happen to agree that
it would likely be problematic--you
know, putting
on my other hat as an IRB member looking
at a study
367
that is a randomized study design, to
repeat a
randomized study that has already shown
a striking
survival benefit. I think most IRBs are not going
to be capable of doing the post hoc
analysis of the
statistics and understand the
limitations as they
have been put forward today, and the
majority of
them would say, "look, you've
already got a
striking survival benefit, how are you
going to go
back?" So, I would look at that investigator and
say, "hm, I wonder what it is
they're trying to do?
They want to look at mechanism. They want to do
something else. And, they're going to take these
poor patients and randomize them to
placebo so that
they can look at lung biopsies and try
and figure
out how cyclosporine is working locally
when the
question for survival has already been
answered."
I would really be wondering at that
point whether
they were putting their science ahead of
the
patient's benefit. So, putting on that other hat,
I think it would be very difficult for
me to look
at a randomized, controlled trial in
this setting
with the survival being the primary
outcome, with
368
my IRB hat on.
So, I think it is perfectly
appropriate to
go forward, either preapproval or
postapproval,
with a study that compares to
contemporary
patients. There will be plenty of patients who
don't go on trial. I mean, I think it is striking
that you got 50 percent. I think that is a really
high percentage of patients to go on a
clinical
trial so I don't think you will have any
trouble
having controls. They will be the majority.
DR. SWENSON: Dr. Burdick?
DR. BURDICK: I mentioned this before, I
am concerned that the sort of study that
was
proposed by the company, using cohort
controls
through OPTN data and so forth, is going
to leave
one, no matter how carefully it is put
together,
with a series of confounding variables
to be
interpreted and a more or less random
conclusion
that certain clinical settings work and
certain
others don't unless it is an
extraordinarily
positive benefit which is being
predicted. And, as
we have said, probably the next time
around there
369
won't be quite such a big difference,
then you are
not going to be able to have the
fundamental
information about basic utility of the
drug.
So, I think another randomized
trial is
the way to find that out. The argument that there
is a huge difference in the trial we
have been
considering is true but, remember, we
have seen
that huge difference based on just a few
patients
because the numbers were so small. And, it is not
quite as convincing as everybody is
making it out
to be.
It is just too convincing to ignore but it
is not that convincing. The way you really find
out for sure is to do another randomized
trial,
somewhat larger numbers, pick your
power, and then
the rest of what is proposed in sort of
cohort
study, no control group study or dose escalation
will go on in the transplant community
anyway and
we will get that information. It will evolve as
immunosuppression evolves. It will have to be
redone every two or three years, just as
everything
else works.
DR. SWENSON: Dr. Proschan?
370
DR. PROSCHAN: I think Chiron has done as
good a job as you could possibly do with
this data
that they have, and I am still not
convinced. So,
for me, I would need to see another
randomized
trial.
I think the analyses that they have done,
the sensitivity analyses, are very good
and that is
what made this so difficult for me. You pushed me
right to the edge but you didn't quite
push me
over, although I am sure some of you
probably want
to push me over a cliff. But, to me, it is not
strong enough and, to me, I would need
to see
another randomized, controlled
trial. It would not
be sufficient to see a one-arm trial for
example.
DR. SWENSON: Dr. Moss?
DR. MOSS: Actually, I like the suggestion
that Dr. Schoenfeld recommended. You know, I think
the people that voted no, they would
need to say
that a randomized clinical trial needs
to be done.
I think having the endpoint be chronic
rejection-free survival makes a lot of
sense to me
in the sense that then if people did
develop the
endpoint they could be done with the
study, and if
371
they wanted to then do something
off-label that
would be a reasonable thing to do. So, I thought
that was an excellent suggestion for a
randomized,
clinical trial.
DR. SWENSON: Dr. Hunsicker?
DR. HUNSICKER: It is always preferable to
do a randomized clinical trial and my
endorsement
of the possibility of a one-arm study
does not take
away from the fact that is always
preferable to do
a
randomized clinical trial. Sometimes it
is not
possible or feasible, or whatever you
want to call
it.
A way to get around the
problem that Dr.
Burdick raised, which is that if you use
a registry
control you are going to have some
explanation to
do, and if you have a close call it is
going to be
very unconvincing. One way to get around this is
to stipulate a minimum difference. You know, what
we are trying to do now is to see
whether this idea
that there is 70, or whatever it was, 75
percent
reduction in mortality is credible. If you say
that the mortality absolute
difference--let's say
372
between 20 percent and 50 percent, I
don't remember
exactly what the numbers were--that
there has to be
at least a 15 percent difference in
mortality that
is observed, that will take care of an
awful lot of
messes in non-equality and will also get
at the
issue of whether this is really a robust
finding.
So, I think that there is, in fact, a
rigorous way
that one can do a single-arm, registry
controlled
trial by defining exactly what your
outcome is.
DR. SWENSON: Dr. Proschan?
DR. PROSCHAN: In terms of the single arm
or the non-placebo study, if you did
dose ranging I
would suggest trying to actually
decrease that
volume of propylene glycol. It seemed that even in
the placebo patients it was poorly
tolerated. So,
rather than decreasing--you know, it is
a small
point--the concentration of cyclosporine,
actually
try to decrease the volume--it is
something like 5
mL, which is a lot of stuff to inhale
even though
most of it is not being inhaled.
A second point is that if you
are going to
compare to the registry, is there a way to
control
373
or at least partially control for
systemic
immunosuppressive drugs so that at the
end of this
clinical trial you don't have radically
different
systemic immunosuppression?
DR. HUNSICKER: The registry data on doses
of immunosuppression is non-existent and
is very
low quality. I think all you can get is sort of
what they were on.
DR. SWENSON: Dr. Schoenfeld?
DR. SCHOENFELD: I think, first, if you
think that it is unethical to do a
randomized trial
we should approve this. I think if you can't do a
randomized trial, then the drug should
be approved
now.
As a postmarketing study, I
think the
single-arm study is a great idea because
basically,
to use a baseball metaphor, the
postmarketing study
would tell us whether we hit a home run
or just got
to first base and that is an important
thing to
know because you want to know where to
go from
here, and a long postmarketing study
would show us
that.
If survival was really, really good compared
374
to historical controls we would know
that we had
gotten a home run.
On the other hand, the fastest
way to get
approval has got to be a randomized,
controlled
study organized to be as quick as
possible, with a
sequential stopping rule, with an
endpoint that
occurs as early as possible. In this case it would
be chronic rejection with lots of
biopsies done so
you can see it. If they do a single-arm study it
will be a long, long, long time and then
I think
the results will be somewhat confusing.
DR. SWENSON: Dr. Sampson?
DR. SAMPSON: I just wanted to echo again
the controlled, randomized, double-blind
trial.
DR. MANNON: I just wanted to make a
comment regarding the number of comments
made
around the table and also by the
investigators
regarding the ethics of this and what an
IRB might
potentially go along with. But I have to tell you
that if I were sitting on an IRB and an
investigator wanted to present a
double-blind study
and said, "I have this initial
study and look at
375
the striking event," I don't think
that I would
find it inappropriate if I understood
that the data
presented represented some concerns and
flaws and
how the data was interpreted. And, look at us.
There are 18 of us sitting here and we
can't even
come to an agreement one way or the other. We are
all sort of split. So, that clearly means a number
of intelligent people are interpreting
this data in
a variety of ways. So, I think that a standard
IRB, presented with that kind of
information, would
understand the necessity for going on.
Then, as far as the patients,
I think if I
presented my patient with what they
should take--we
use a lot of things off-label--I
shouldn't even say
this but we use a lot of drugs in
immunosuppression
for transplant that are off-label based
on a couple
of case series, and you if present it to
a patient
and say this is the option and we are
going to try
this.
Most will do what you recommend and it is
not clear to many of the patients that
this is the
beneficial therapy. Therefore, I don't feel that
everybody is going to run and say I
can't be on
376
placebo.
DR. SWENSON: Dr. Moss?
DR. MOSS: I just want to echo what Dr.
Mannon said. I think there is a lot in the
literature about smaller Phase II trials
which show
that there appears to be efficacy and in
a lot of
those cases, or most of those cases,
there were
then larger multicenter studies done and
they all
got through IRB committees. So, I think there is
very good precedent in the literature
for orphan
diseases, such as idiopathic pulmonary
fibrosis,
where this same scenario has happened
where IRBs
approved larger multicenter, randomized
clinical
trials.
So, I agree with you, Dr. Mannon.
To say
that no IRB is ever going to approve a
multicenter
trail based on the data that we were
presented
today I think is incorrect.
DR. SWENSON: Well, that being the extent
of comments here, Dr. Albrecht, anything
more?
DR. ALBRECHT: No, just to thank everybody
very much for all your thoughtful
comments. You
have given us a lot of food for thought
and we will
377
take them back and discuss them
internally.
DR. SWENSON: I would like to reiterate my
thanks to everyone, the company, the
FDA, the panel
members, the audience and the patients
who came
here.
[Whereupon, at 4:30 p.m., the
proceedings
adjourned.]