1
DEPARTMENT OF HEALTH AND HUMAN
SERVICES
FOOD AND DRUG
ADMINISTRATION
CENTER FOR DRUG EVALUATION AND
RESEARCH
ADVISORY COMMITTEE FOR PHARMACEUTICAL
SCIENCE
CDER Advisory Committee
Conference Room
2
PARTICIPANTS
Arthur H. Kibbe, Ph.D., Chair
Hilda F. Scharen, M.S., Executive Secretary
MEMBERS
Patrick P. DeLuca, Ph.D.
Paul H. Fackler, Ph.D.
Meryl H. Karol, Ph.D.
Melvin V. Koch, Ph.D.
Michael S. Korczynski, Ph.D.
Marvin C. Meyer, Ph.D.
Gerald P. Migliaccio,
Ph.D. (Industry
Representative)
Kenneth
R. Morris, Ph.D.
Cynthia
R.D. Selassie, Ph.D.
Nozer
Singpurwalla, Ph.D.
Marc Swadener, Ed.D.
(Consumer Representative)
Jurgen Venitz, M.D., Ph.D.
SPECIAL GOVERNMENT EMPLOYEES SPEAKERS
Judy Boehlert, Ph.D.
Gordon Amidon, Ph.D., M.A.
FDA Staff
Gary Buehler, R.Ph.
Lucinda Buhse, Ph.D.
Jon Clark, M.S.
Jerry Collins, Ph.D.
Joseph Contrera, Ph.D.
Ajaz Hussain, Ph.D.
Monsoor Khan, R.Ph., Ph.D.
Steven Kozlowski, M.D.
Vincent Lee, Ph.D.
Qian Li, Ph.D.
Robert Lionberger, Ph.D.
Robert O'Neill, Ph.D.
Amy Rosenberg, M.D.
John Simmons, Ph.D.
Keith Webber, Ph.D.
Helen Winkle
3
C O N T E N T S
PAGE
Call
to Order
Arthur Kibbe, Ph.D. 5
Conflict of Interest Statement:
Hilda Scharen, M.S. 5
Committee Discussion (Continued) 8
Science in Regulation - Visionary
Overview
Arthur Kibbe, Ph.D. 43
The "
Regulatory Policies
Ajaz Hussain, Ph.D. 74
Organization Gap Analysis
Helen Winkle 75
Scientific Gap Analysis
Ajaz Hussain, Ph.D. 103
Policy Gap Analysis
Jon Clark, M.S. 135
Generic Pharmaceutical Association
Perspective
Shahid Ahmed, M.S. 152
Pharmaceutical Research and Manufacturers
of
Gerry Migliaccio, Ph.D. 164
Committee Discussion and
Recommendations 182
Pharmaceutical Equivalence and
Bioequivalence
of Generic Drugs
The Concept and Criteria of
BioINequivalence
Concept of BioINequivalence
Criteria of BioINequivalence
Qian Li, Sc.D. 222
4
C O N T E N T S
(Continued)
PAGE
Committee Discussion and
Recommendations 234
Bioequivalence Testing for Locally Acting
Gastrointestinal Drugs
Topic Introduction
Scientific Principles
Gordon Amidon, Ph.D. 275
Regulatory Implications and Case
Studies
Robert Lionberger, Ph.D. 308
Committee Discussion and
Recommendations 324
Conclusion and Summary Remarks
Ajaz Hussain, Ph.D. 341
Helen Winkle 349
5
P R O C E E D I N G S
Call to Order
DR. KIBBE: Ladies and gentlemen, I would
like to call the meeting to order. The first item
of business is the reading of the
Conflict of
Interest Statement.
Conflict of Interest
Statement
MS. SCHAREN: Good morning.
The following
announcement addresses the issue of conflict
of
interest with respect to this meeting and
is made a
part of the record to preclude even the
appearance
of such.
Based on the agenda, it has
been
determined that the topics of today's
meeting are
issues of broad applicability and there
are no
products being approved. Unlike issues before a
committee in which a particular product
is
discussed, issues of broader
applicability involve
many industrial sponsors and academic institutions.
All Special Government
Employees have been
screened for their financial interests as
they may
apply to the general topics at hand. To determine
6
if any conflict of interest existed, the
Agency has
reviewed the agenda and all relevant
financial
interests reported by the meeting
participants.
The Food and Drug
Administration has
granted general matters waivers to the
Special
Government Employees participating in
this meeting
who require a waiver under Title 18,
Code, Section 208.
A copy of the waiver statements
may be
obtained by submitting a written request
to the
Agency's Freedom of Information Office,
Room 12A-30
of the
Because general topics impact
so many
entities, it is not practical to recite
all
potential conflicts of interest as they
apply to
each member, consultant, and guest
speaker.
FDA acknowledges that there may
be
potential conflicts of interest, but
because of the
general nature of the discussions before
the
committee, these potential conflicts are
mitigated.
With respect to FDA's invited
industry
representative, we would like to disclose
that Dr.
7
Paul Fackler and Mr. Gerald Migliaccio
are
participating in this meeting as
non-voting
industry representatives acting on behalf
of
regulated industry. Dr. Fackler's and Mr.
Migliaccio's role on this committee is to
represent
industry interests in general, and not
any other
particular company.
Dr. Fackler is employed by Teva
Pharmaceuticals
employed by Pfizer, Inc.
In the event that the
discussions involve
any other products or firms not already
on the
agenda for which FDA participants have a
financial
interest, the participants' involvement
and their
exclusion will be noted for the record.
With respect to all other
participants, we
ask in the interest of fairness that they
address
any current or previous financial
involvement with
any firm whose product they may wish to
comment
upon.
Thank you.
DR. KIBBE: Thank you.
8
Yesterday, we concluded with a
suggestion
that we might want to continue our
discussion about
the questions that the Agency has raised,
and I
think Dr. Meyer has asked Dr. Hussain to
come up
with a straw man, and we have it ready,
so I think
we should go there first and then go back
to the
scheduled agenda.
Ajaz.
Committee Discussion (Continued)
DR. HUSSAIN: Good morning.
I think the
discussions towards the end of yesterday
started
honing down on some of the key challenges
we face
in the designing of a Critical Path
Initiative in
OPS.
I think, reflecting back on the
discussion
yesterday, clearly, I think we have a
wide range of
research capabilities and programs
already in
place, and the challenge would be to sort
of direct
these in a very focused way to help the
Critical
Path Initiative, keeping in mind that all
of our
research will not be focused on critical
path,
there are other aspects that we have to
focus on.
9
I will sort of reflect back on
the PAT
Initiative and how that sort of
evolved. Clearly,
if you recall, the PAT Initiative led to
the GMP,
and there is a whole sequence of
initiatives that
have occurred. The PAT Initiative was a model and
we can learn some things from that as a
model also.
I will sort of summarize my
thoughts here
with a hypothesis statement that Jerry
proposed
yesterday, that Critical Path Initiative
will
improve the efficiency and effectiveness
of drug
development process. That is the hypothesis that
sort of really we are engaged in trying
to fulfill
or trying to confirm.
The challenge would be then how
do we
measure efficiency and effectiveness of
drug
development. That is one of the keys, how do you
measure drug development in terms of the
failure
rate, or the time it takes, or the cost
of drug
development.
All of these are relevant metrics, but for
the purposes of a hypothesis, what and
how should
we approach and define that, because
unless you can
10
measure something, you cannot improve
it. So,
measurement and metrics would be a key
factor of
that.
The second aspect then would be
what are
the root causes of low efficiency and
effectiveness
in all the three dimensions. A number of factors
were put up looking at 1999 or 1991 or
2000, and so
forth.
They were indicators of what
may be
happening, but you have to keep in mind
that is
partial information, information
available to FDA
and available in public is just limited
because the
companies have far more information about
the root
causes, and so forth. So, you have to sort of
factor that into our decisionmaking.
The next question is who is in
the best
position to address these root cause
factors that
we identify, what is the role of FDA,
what should
FDA do and what should some industry,
academia, and
other agencies should be doing is the key
there.
I think based on information and based on
experience at FDA, clearly, we are in a
good
11
position to identify many of the
problems, not all
of the problems, but many of the
problems, and FDA
has the responsibility to communicate
these
findings in some way or form.
If you look at John Simmons'
presentation,
he laid out, as a part of the critical
path,
strategic meeting points during the drug
development process. That is one aspect,
communication between sponsors of
applications and
our review scientists, if that is timely
and in a
coordinated manner, that is one effective
means of
that.
So, communication through
meetings for
specific drug applications, broader
communications
with workshops, and then eventually
guidance
documents outlining FDA's current
thinking on a
given topic are the communication
mechanisms that
we have.
For that, clearly, I think you
have to
think about resources and how do you
facilitate
that process. If you want to sort of move towards
more meetings and more interactions between
12
reviewers and sponsors, then, you have to
build the
time for that, and so forth. That has to be
considered, too.
Workshops and guidances also take
a
significant amount of effort, and so
forth, so as
we improve our communication channels,
where will
we find the resources and time to do that, I think
that is also a management aspect that has
to be
discussed.
FDA's knowledge base, I think was
clearly
an asset in the sense we have a lot of
information.
If we are able to create a knowledge base
that can
be useful, not only for identifying
problems that
we see, but also for improving of a
predictive
ability in all three dimensions, safety,
efficacy,
and industrializations, what are the
practices that
lead to success, what are the practices
that may
not be as efficient, and so forth.
So, this knowledge base would be useful
for that purpose, but again I will remind
in the
sense we have to be cautious, there are
limitations
of that knowledge, because we don't
always have all
13
the information, so you have to factor
that in.
But based on our knowledge base
and based
on communication, and so forth, I think
the
laboratory and the research functions
clearly have
to
focus on improving methodologies. There
are
many aspects of laboratory work that only
FDA is in
a good position to do, and others either
don't have
the interest or don't have the focus to
sort of
address some of the challenges.
For example, in the case of
regulatory
decisionmaking, risk-based
decisionmaking, the
decision process itself often needs
support of
science, and so forth, so that is where
the
research really could focus on.
Also for development and
validation of new
methodologies, standards development,
methodology,
validation, say, from biomarker to any
new
technology, unless FDA has a role in
achieving
that, it may not be fully appreciated
within the
Agency, and some of the Agency concerns
would not
be addressed if it is done totally
outside, so
there has to be some means of linking our
14
laboratory work to standards development,
validation of new methods, and so forth.
Our postmarketing experience
again is
unique because that is where I think we
have a lot
of information, how do we capture that as
lessons
learned and how do we use that. You saw some
examples of how we were learning from
that and
going retrospectively and said how could
we have
improved the process. Those experiments would be
very valuable.
One aspect is in terms of
innovation, in
terms of new technologies, what is an
important
aspect of standard setting? Standard setting and
guidances are slightly different in my
opinion.
For example, in the PAT Initiative, we
opted to
move towards ASTM International as the
body for
standard setting.
What that does is allows
industry,
academia, every stakeholder to be part of
that, and
actually identify what standards are
needed, and
actually develop those as quickly as
possible.
That relieves the burden on
FDA, and FDA
15
simply adapt or adopt those standards
after
evaluation, so that might be an option
that seems
to be moving forward in the PAT
Initiative for new
technologies, new methods, and so
forth. So, that
could be considered at the same time.
But at the same time, I think as
we look
at FDA's role, what is the role of
industry and
what is the role of other agencies and
academia
really have to sort of come together.
The role of industry I think is
knowledge
sharing. Clearly, it has far more
information, and
reluctance to share knowledge will
inhibit the
progress, and how do you do that is a key
challenge.
At the same time, I think in
order to
bring all of us together, focused on a
given goal,
we really I think have to define clearly
the
metrics, the desired state, and so forth,
and come
on the same page, so that we can
coordinate all of
these activities.
In some ways, FDA could play
that role of
coordination, as Viad declared yesterday,
not only
16
from the prospective of we are not
competing in
this arena, eventually, we have to be
involved, so
coordination function for FDA would be an
important
function for all these activities.
If we have a clear
understanding of what
are the issues and what are we trying to
achieve,
then, the coordination and synergy would
sort of
evolve naturally.
So, those are the sort of
thought process
that I could capture.
DR. KIBBE: Anybody?
Marvin? You asked
for the straw, you got the straw man.
DR. MEYER: The virus, are you talking
about that later?
DR. HUSSAIN: No, that is a very specific
example.
DR. MEYER: I thought I had more time to
think then, since I was waiting for the
virus.
DR. KIBBE: Does somebody else want to--
DR. MEYER: No, no, I have something to
say.
DR. KOCH: Marvin, just before you--I
17
think I need a point of clarification
because some
of what came out yesterday was the desire
to have
either shorter development time, more
compounds
coming out that could be effective, new
pharmaceuticals, and it seems to be a
push towards
industry to try to become more effective,
et
cetera, but I saw a couple times
yesterday where
things developed within the Agency to
improve the
ability to go after materials through
some of the
databases and things were certainly ways
to help
that process.
The other thing, though, is
that when you
looked at that chart that showed an
increase in
cost of materials and a few other things,
toxicity,
you know, is something that shows up
there, and I
think something a little bit insidious
over time
has been with improved technologies and
increased
concerns over pharmaceuticals, there are
new tests
that come in that prolong the evaluation,
that
anything the Agency can do to pull things
together
to make those things, immunogenicity or
other
things that have, you know, you go back
two
18
generations ago and if you come up with a
new
material, would you put it through all of
the same
tests, so anything that can be done to
simplify and
do more predictive studies in that
regard, I think
would help.
DR. HUSSAIN: Definitely, that is an
important point. For example, I think as we move
towards more complex materials, the
material cost
is, as you saw, is already showing up,
and so
forth.
Introduction of new excipients
or new
adjuvants, and so forth, is a significant
challenge, and as we go towards
nanomaterials,
nanodevices, and so forth, if we still
have to rely
on the traditional pharmaceutical
excipients, it
would be a very limiting aspect, so I
think that
that is clearly on our agenda.
One aspect that I do want to
mention, as
we think about this, the patient has to
be foremost
in our minds, what are the unmet needs,
and as we
sort of develop this, I think clearly,
the patient
needs have to be kept in mind as we move
forward,
19
because there are many diseases, many
aspects where
we don't have effective drugs, and so
forth, so we
shouldn't forget that aspect.
DR. KIBBE: Marvin, are you ready now?
DR. MEYER: Yes. I
was just thinking of a
simple example where the Agency I think
played a
major role and really expedited drug
approval, and
that was back when we were battling over
assay
method validation.
The hypothesis was if we had a
better way
of validating assays or a uniform way of
validating
assays, things would get approved without
recycling
and redoing, and, in fact, FDA then, and
APS and
others, convened several workshops, had
white
papers, ultimately put out a guidance,
and I
suspect that hypothesis has been tested,
that there
are much fewer problems in the local
methodology,
so I think that is a good model, and you
alluded to
that.
DR. KIBBE: Anybody else would like to
make a comment?
DR. SINGPURWALLA: Yes. I
repeatedly hear
20
from individuals like yourself asking
industry to
share more information with you. What is the
incentive to the industry to do so,
because there
is a penalty to do so? Recently, those of us who
read the Washington Post can see the
number of
pages devoted to the Merck and also to
Pfizer
having a similar drug going to be tested,
and
things like that.
So, unless the legal pressures
that are on
industry are defused or removed, industry
is going
to be foolish to share all the
information with
you.
I wouldn't. It's like me going to the IRS and
saying look, this is how I have cheated,
catch me.
It doesn't make sense, does it?
DR. HUSSAIN: No, I think it is not in the
context of sort of cheating, and so
forth. This is
in the context of how much we know and
how much we
don't know, to start filling the gaps
where the
knowledge exists. Clearly, that is the aspect, and
it is complicated by the fact that the
way it gets
entrenched into the legal and political
scenarios,
those are significance challenges, no
doubt about
21
that.
DR. SINGPURWALLA: What is the industry's
response to this?
DR. MIGLIACCIO: Well, it is complicated
by obviously, intellectual property
rights, which
is the life blood of a commercial
business, but it
is also complicated by--you just used the
word
"trust."
I will give an anecdote
here. I went to
an internal FDA meeting to provide
training, and
during that training, provided knowledge
about
products, which normally, would not have
been made
available.
The reaction was the
traditional
predictable reaction, not the forward
thinking
reaction, by certain elements of the
audience. So,
that was a risk, that was a poorly
thought-through
risk on my part.
We have to reduce the risk
associated with
sharing knowledge. That is the fundamental issue
is if we share knowledge and expose
ourselves to
compliance action where that knowledge is
22
essentially reflecting what is the
scientific
truth, and we can now measure that, we
can now see
that where we couldn't before, and if you divulge
that knowledge and risk compliance action
versus
scientific discussion, then, the
knowledge will not
be transferred.
DR. KIBBE: Ajaz, anything?
DR. HUSSAIN: No.
DR. KIBBE: Anybody else?
Let me just put
three things on the table and perhaps you
can think
about them as how they respond to the
questions we
were left with yesterday.
One is that I think I heard
from around
the room that the Agency has a limited
resource
base, and it truly should focus on those
aspects of
the critical path that only the Agency
can do, that
no one else has the wherewithal or the
capability
or the information to do that.
Secondly, that we try to get
others who
are even more capable of responding to
certain
aspects of the critical path to take on
that
burden.
I am thinking primarily of industry and
23
perhaps industry/academia together
looking at those
aspects of it.
But the third thing that I
think that the
Agency and both the industry will have to
look
forward to is that the rate of
technological
advance is such that 10 years from now,
the
questions that you are trying to answer
now will be
ancient history, and the questions that
you are
running into are going to be dramatic and
clearly
different, and I really look forward to a
paradigm
shift in the way we approach therapy, and
I would
recommend to the industry that they
change their
name from drug companies to companies
that provide
therapeutic agents and processes, because
they
could be caught up in the same system
that the
railroads did. They were railroad companies, and
not transportation companies.
I don't know how the Agency can
respond
effectively without having some type of
internal
committee that is constantly looking at
four or
five years out and the technology that
they are
going to have to deal with then.
24
So, that is where I think the
critical
path kind of initiative ought to be
looking.
DR. DeLUCA: Let me just comment. I made
some notes here from yesterday, and I
think that
this is based on collaboration, I think I
am going
to really focus on that, and as Jerry
mentioned, I
think trust. Certainly, trust is essential in
collaboration, and as we get into talking
about the
science-based approach here and research,
research
is a search for the truth.
I would like also to commend
the Agency in
their research efforts. I mean yesterday was, I
think, I would say overwhelming to learn
the type
of research that is going on and the
collaboration
with NIH, so I really have to commend the
Agency
for this.
I would like to also talk about
the
presentation by Monsoor Khan where he
talked about
critical path research and some of their
efforts,
and I think Gerry Migliaccio had
responded that
industry takes this approach.
I have to say that that is true
from my
25
experience to an extent. I know, I am involved in
the novel drug delivery area and the
research in
that area, and working with a company
that was
scaling up or transferring some
technology, that
that approach was taken, the critical
path approach
was taken, and worked with them, and they
did solve
the problem at hand, but there were other
things
that still needed to be done to define
some of the
process variables, that once the problem
was
solved, they went on, they didn't want to
go any
further with that.
So, I think there is a limit to
where they
go and I think this is where
collaboration is
important, and I think there is a need to
continue
on and to search those things out, and
probably the
place for that is in academe.
I don't know if it was Jerry
Collins, when
he talked about the science-based
approach to
critical path issues and the research,
that it is
probably essential that the research that
is going
on, that it is going to be hypothesis
driven. I
think this is something that many times
the
26
research that takes place, and if it is
in an
industrial setting, may lack the
hypothesis driven
type of research, and that probably I
think is
important.
I guess my feeling is in
hearing all the
things, and I think what Art had said,
FDA can't
really overreach, I mean there is a limit
resourcewise, but I think more
importantly is the
idea that they can't do it alone, so I
think that
collaboration is important.
The FDA has been collaborating
more with
NIH, and I think translational research
issues,
taking the drug product development, that
portion
of it along, but I think there is a gap
there with
the critical path, in the formulation and
taking it
and the manufacturing science, and I
guess I think
that the collaboration has to be there
between
academe, industry, and FDA, and FDA could
really
set the stage for this. I think there is a very
important role.
Just to bring out my experience
with the
journal, the APS on-line Pharmaceutical
Science
27
Technology Journal, that the submissions,
we get
about 50-50 from abroad and the United
States, but
about 90 percent of the submissions--now,
this is
in the Pharmaceutical Technology Journal,
and you
would really think that you would get
more from
industry--but about 90 percent of the
submissions
come from academe.
I have to acknowledge that
probably in
about 40 percent of those, there is a
collaboration
between academe and industry, so that
there is a
tie-in and that it is all those being
submitted
from the academic institution, the
industry is
involved in it.
But I think that this kind of
sends a
message, and I have tried to encourage
more, more
submissions from industry, and there is
intellectual property situations involved
in that,
but I think there is a need.
I know, being in academe and
graduating
Ph.D.'s, the majority of them will go
into the
industry, few of them publish after they
are in
industry.
Before they left, they had eight or nine
28
publications, and then they went in and
stopped
publishing, so, I am not sure that is a
good thing.
What I wanted to emphasize here
is that
there is an essential need for
collaboration. I
think the whole science-based approach to
the
regulatory arena is great, and I have to
commend
the Agency in this, but they just can't
do it
alone, and I think there is an essential
need that
this collaboration occur.
It may be that with that type
of
collaboration, you know, for a long time
in
academe, we have talked about the NIH and
trying to
get them involved with supporting drug
product
research and development to little avail,
so maybe
now is the time.
Certainly, I think it is
essential that
this type of approach be taken in the
manufacturing
sciences, because it certainly will benefit,
I
think, our society, so I think it is in
the
interests of the country, so that
hopefully, the
NIH will look a little bit more favorably
on
supporting this type of research. I think the
29
collaboration between FDA and NIH may
help in doing
just this.
DR. KIBBE: Thank you.
Ken, go ahead.
DR. MORRIS: Thanks, Art.
Welcome to your
last day.
A couple of things I wanted to
say first
to Nozer's point. In the face of the data that I
think Merck generated or was generated
and then
shown to Merck, I don't think they would
need the
Agency to tell them that they needed to
pull the
drug or to modify it. They are really very
responsible about that sort of
thing. I understand
your point.
DR. SINGPURWALLA: The lawyers made them
do it.
DR. MORRIS: Well, the lawyers made them
do it, but the drug companies in general,
the
innovators of generics, when they see
problems like
that, are still I think honor bound and
have
historically done a good job of
monitoring
themselves with respect to public health when
there
30
is a clear and present public health
injury issue.
But beyond that, let me just
comment, if I
can, I have just five points here,
relatively
short.
The first is, is that the
unique
opportunity afforded by the FDA massive
database, I
think is absolutely invaluable and needs
to be
exploited to the maximum. I mean that, in my mind,
is perhaps the number one initiative in
terms of
getting down the critical pathway with
all due
deference to proprietary data, of course,
as we saw
yesterday.
There are some issues I think,
for
instance, tox, where the database would
probably
not be nearly as good as even the
sampling of the
Big Pharma companies' databases on tox,
because you
don't have the tox information on
compounds that
never made it to filing, so they may
actually have
a bigger database there, which would
really help in
the really interesting work we saw
yesterday on
modeling.
The second point is it look
from a
31
nonbiological and obviously blue collar
tablet
smasher, that there is a fairly large
disparity
between the amount of internal biological
research
versus product development research. I am not
really in a position to judge what the
priorities
in the biologicals should be. It is all obviously,
very high-caliber research.
I am not in any way commenting
on that,
nor am I capable of it, but I think it
does point
out the fact that there are developmental
research
agendas that probably would be better
handled in
part at least, or at least administered
through the
Agency, that aren't being, and we can
talk about
specifics, and have with John and you and
others,
of course.
But I think that points out an
opportunity
if you were on the critical path, and
that is the
prioritization that I was asking about
yesterday,
is that given the breadth of projects and
the
dearth of resources, I think the
prioritization,
particularly the internal research
projects,
becomes your biggest challenge and one
that I think
32
could be helped by the committee, and I
think has
been in part, hopefully, in these days.
It also points out, to
reinforce Pat's
point, and this is a little bit
self-serving, but
the amount of research that doesn't or is
not as
logically done within the Agency, needs
to find
federal support in terms of public health
initiatives, as well as the obvious
advance of just
basic science.
To address a question that
Gerry had
raised with respect to the possible
putative
consequences of sharing information,
there is a
mechanism that we have been sort of
developing,
which is to, through blinded
intermediates, to be
able to discuss general topics without
filtering.
I am not talking about
filtering data or
hiding data, but to bring data to light
to the
Agency in a blinded manner to say, you
know, is
this the sort of data that would be
useful, or is
this the sort of data that would give you
cause to
think that there was no particular reason
to review
it, and it would just be a waste of the
Agency's
33
time.
So, I think there are mechanisms to do
that.
They are not formal mechanisms, but through
consultants and whatnot, I think you have
already
got that as an opportunity, so you can't
be too
specific, of course, because once you are
too
specific, then, you have already revealed
what it
is you are asking about.
Finally, with a question of
metrics, I
think the suggestions you made yesterday,
the
multiple review cycles I think is a great
metric.
That was what the Manufacturing
Subcommittee, at
Judy's last meeting, we talked about the
idea that
in the new or the desired state, instead
of having
minimal data that the reviewers have to
try to
piece together into some sort of
Frankenstein
rationale, if you get the rationale in a
piece from
the companies with summarized supporting
data to
make it a compelling argument, then, the
reviewers
just have to assess the sufficiency of the
rationale as opposed to trying to piece
together
one on their own.
34
So, I think the review cycles
are an
excellent metric. The time to approval, of course,
is a low hanging fruit there in terms of
a metric
although it is not independent, and for
generics,
of course, that is compounded by the
workload
itself.
Maybe you could normalize it by
normalizing the time to approval to the
number of
pre-filing and pre-approval meetings, the
off-line
meetings that John talked about
yesterday, John
Simmons talked about yesterday.
The other one--and I don't know
if we have
talked about this before--is to track FDA personnel
turnover.
I think it is not a bad metric to look
at retention of the FDA reviewers
themselves. I
mean it is a very high-pressure job, it
is not all
that celebrated a position, but obviously
of key
importance.
I think that does two
things. One is it
gives us a metric of how effectively the
program
works, and the other is that it gives an
internal
metric for the personnel management, so
you don't
35
burn out your best and brightest.
DR. HUSSAIN: Ken, the whole aspect of the
critical path was in a sense that review
cycles
have really come down, so that review
cycle is not
the rate-limiting step in the critical
path. So,
there are different metrics for that
purpose.
DR. KIBBE: Marvin, do you have something
else?
DR. MEYER: A quick comment. Helen was
saying last night that on the ANDA side,
that the
generics are now, or shortly going to be,
required
to submit all studies they did, not just
the 1 out
of 12, the test.
Maybe, and this is terribly
naive because
I don't know all of the complications,
but maybe
there is some way down the line of having
the NDAs
be accompanied by a synopsis, at least,
of what
they tried and what failed, a one-pager
perhaps.
We tried doing virus filtration
this way,
and it failed because, we think it
because, and
this might be attached with the NDA, but
reviewed
independent of the NDA. There might be a
group at
36
FDA that evaluates failures, if you
will. So, some
way of getting the data to the Agency
that wouldn't
impact on the NDA and yet would provide
the Agency
with I think some valuable information.
DR. KIBBE: I am concerned that as much as
we in academia value getting all the
information,
industry values having information that
their
competitors don't have, and if they have
a lot of
failures they corrected, and they know
what
mistakes not to make, they generally
think they
have an edge on doing it right, and they
are not
really excited about turning that over to
someone
else, let them make their own mistakes
and figure
it out.
I think the time that we will
actually be
able to share all the information about
all the
drugs that have ever been approved is
when Glaxo
finishes buying everybody or Pfizer has
merged with
whoever is left, and there now is
International
Therapy Development Company.
DR. KIBBE: Ajaz, anything to wrap up
with?
Okay.
37
DR. SINGPURWALLA: Mr. Chairman, I have a
few thoughts. Ajaz, back.
[Laughter.]
DR. KIBBE: It's okay, Ajaz, you can
escape if you would like.
DR. SINGPURWALLA: Ajaz, you asked three
questions here. To be quite honest with you,
yesterday, I couldn't focus on these
because I
couldn't get my mind straight as to what
we are up
to and what is happening.
But subsequently, I think I can
answer
some of your questions very directly.
I looked at TR Critical Path
Initiative
Challenges document, and to be quite
honest with
you, I think you are on the right track,
and I
think you are thinking along the proper
lines.
Two, three things come to my mind. Your
mention or at least the mention of design
of
experiments that was discussed is one of
the right
ways to go about things.
You also mentioned the use of
bayesian
ideas. That is the best way to reduce time cycles
38
because you are taking advantage of all
other
sources of information, but you don't
want to use
that only for clinical trials, but you
want to use
it throughout the entire process. Again, you have
highlighted it, so I think again you are
on the
right track.
The one thing is you cited
examples from
manufacturing. That is fine, but I seriously
consider you also look at the area of
weapons
development. They face problems very similar to
yours and you may want to see what they
are doing
and how they are developing their
particular
processes, and the weapon development
process has
much in parallel. The two communities are very
alien to each other, but I urge you to
look into
what they are doing, and I think I can
say that you
are on the right track. You are focusing on the
issues that I would focus about, that is
all. I
wanted to reaffirm it.
DR. HUSSAIN: Thank you.
DR. KIBBE: Paul.
DR. FACKLER: Let me just offer a couple
39
of thoughts to those questions, you know,
are you
on the right track. Of course, I am speaking for
the generic industry, but it is difficult
to give
people help when they haven't asked for any,
and I
can't speak for PhRMA, and I don't know
if PhRMA
has come to the Agency and said, help, we
can't
develop new drugs.
So, I think you face a very
difficult
challenge trying to assist a process that
maybe the
people actually doing it don't feel is
broken. The
economics of drug development in 2004 is
significantly different than it was in
1994.
You know, if you have a company
selling
$50 billion in drugs a year, and they
want to grow
by, say, 5 percent, which isn't
acceptable by any
means, they need to get an additional $2
1/2
billion in revenue out of the new drugs
that they
are developing.
So, you know, a product that
has some
marginal value, say, 50- or $100 million
that would
benefit society probably just gets put in
an
envelope somewhere, and not brought
out. It is a
40
problem with the situation in industry, but I
am
not sure FDA is going to be able to do
anything to
assist that.
Let me speak to the generics
because there
was a presentation yesterday, and
speaking for the
generic industry, we have communicated
with FDA
where we think we need help. We have asked about
topical products, we have asked about
inhaled
products, biologics, of course, are an
issue, and
time to approval is a real issue for us.
So, the question was are you on
the right
track, and at least from the generic
perspective,
the answer is yes. I think you are trying to
overcome the hurdles that we face, that
would
assist us in bringing products to the
market
earlier.
I know it is not really the
main thrust of
the Critical Path Initiative, but for our
portion
of it, the answer is yes.
DR. KIBBE: Thank you, Paul.
DR. HUSSAIN: I think just the point
generics are equally important for us, so
they are
41
part of the critical path from an OPS
perspective.
DR. KIBBE: Gary.
MR. BUEHLER: Well, we have had a number
of mentions of our workload. It is significant.
We did receive 563 applications I believe
the last
fiscal year, 449 the year before, and 361
the year
before, so we are increasing by about 100
a year.
It is a bit scary, but we are
dealing with
it, and we are communicating with the
industry
significantly on what we can do to make
their
applications better and to make our
responses to
them more predictable, so that they know what
we
want.
As part of the critical path,
and we are
trying to work in providing the
information that
the industry needs to develop their
products, and
this is through the dissolution methods
and the
bioequivalence methods that we get tons
of letters.
We got over 1,000
correspondence last
year, over half of them requesting what
is the
bioequivalence method for a particular
drug, what
is the dissolution method for a
particular drug, so
42
we can begin to develop our products.
We are trying to get that up as
a
web-based program, so that they can
actually access
these methods. We have people working full time in
our office to research this information,
so we can
make it available to the firm.
Now, this isn't revolutionary
stuff. This
is stuff that we have always provided
them. We
just want to provide it to them
faster. We want to
make it easier for them to access this
information,
and we don't want to get as many
letters. The
letters that we get obviously take up our
resource
time, and we want those resources to be
put toward
application review.
We hope to be able to get these
up soon.
I know I promised them I think six months
ago to
the industry. Things are never as easy as you
would like them to be in the Agency. A lot of
people have to sign off and make sure
that we are
not giving away the farm, and we don't
want to give
away the farm, but we do want to give
away
information that is needed by the generic
industry.
43
The generic industry is a very
viable,
very robust industry right now. A lot of new
players are getting into it, a lot of
people want
to put applications in as evidenced by
our
workload.
We welcome that workload, we are glad.
This country needs generic products. A lot of
people out there can't afford
prescription drugs
out there.
So, we welcome the work and we welcome the
challenge.
DR. KIBBE: Anybody else?
Okay.
We have an opportunity now to
hear from an
absolute genius. They asked me to give a talk on
visionary overview, and I will get up
there, then,
I will pontificate for half an hour, and
I hope you
all enjoy it.
Science in Regulation -
Visionary Overview
DR. KIBBE: I need a soapbox. I have six
slides.
This is to reduce some of the slide
overload that we are suffering from. You all have
copies of these slides. You can tell that the
slides are really informative because
they are
44
filled with words. I look at slides and I say,
hey, there are 22 slides and each one has
180
words, how am I going to get through it,
so I put
up a couple of simple slides.
First, the title was given to
me by the
Agency.
I looked at it and I said Visionary
Overview, I guess they think I am a
visionary, why
would they think that. So, I thought long and hard
about why they think I am a visionary,
and I
realized it was because I live in
Pennsylvania,
which is the home of the world's most
well known
and renown visionary, the seer or all
seers, the
procrastinator for all good things,
Punxsutawney
Phil, who comes out and tells you whether
you are
going to have winter for another six
weeks or not.
I also would like to make a
disclaimer, we
do lots of disclaimers. All the ideas that I
express today are strictly my ideas, and
I would
not saddle anyone in the scientific
community and
industry over the Agency with any of
these
cockamamie ideas. So, they are all mine and
hopefully, they will stimulate your
thinking
45
without putting you completely to sleep.
So, what has the FDA and we
been doing for
the last few years? I actually went out and got a
copy of the agenda for the first meeting
I was at,
and there wasn't PAT mentioned in the
agenda, but
when you looked through the agenda, you
saw the
beginnings of what was I think a
wonderful three-
or four-year push in an area that can
significantly
impact industry's bottom line, and
hopefully, the
industry will be in a mood of generosity
and have
that bottom line, some of those savings
reflected
in the cost of goods produced.
The effort I think was an
opportunity for
me to view the way that scientists from
industry,
both the generic and innovator companies,
scientists within the FDA, and scientists
from
academia, and those consultants who serve
all of
us, could get together, look at a
problem, develop
a reasonable approach to it, something
that would
work in the community that we work in,
and really
come up with something worthwhile.
I could go on about the
successes we have
46
had, but they don't make great news, and
the news
media always wants failures and disasters
to report
on, and so I will move directly into
those.
First, is it the Agency's role
to apply
science to regulation? Of course, we all agree it
is.
The application of the scientific method to
goalpost generation for the industry is
extremely
important, and I am going to try to look
at what we
have done and where we are going, and
perhaps make
some projections out.
If we are going to regulate a
science-based industry with science,
then, we need
to use a scientific approach to where we
are going.
We all are familiar with linear
regression, and we know that there is a
certain
amount of error associated with it, but
in order to
project beyond the data that we already
have, we
have to have a significant amount of data
going
backwards to draw a line through, so that
as we go
out in the future, we get closer to the
truth.
We know that the further out in
the future
we can project, the less reliable the
answer is,
47
but we do it anyhow, and I am going to do
that.
So, where have we been in terms
of
regulating the quality of drug products
and
therapies in the United States? Of
course, we start
in 1817 with Dr. Spalding, and he decided
that we
ought to get the physicians together and
say why
can't we have quality products to give to
our
patients, let's set up some standards,
and the USP
was formed.
So, we started the regulation
of the
quality of how we treat our patients by
getting the
health care providers who treated
patients together
to decide what quality was and how to
arrive at it.
After the Civil War, the
pharmacists got
together and decided that while the USP
had
standards for individual ingredients, it
really
didn't have standards for how to mix them
together
and make them useful, so they decided to
publish
the National Formulary, and I was
instrumental in
the first edition, and I brought my copy
with me.
This is the sum total of how to
make
pharmaceuticals in 1888, and compare it
with what
48
we know today and how many shelves it
takes up, and
how controversial each little, tiny issue
is. Of
course, we also know that you have to
learn Latin
to use this, so it's dead along with the
dead
language that it is written in.
At the same time, the industry
actually
regulated itself. There was a comment made here a
little while ago, which said that lawyers
make them
do things. I would argue that in the current
litigious society, companies act slower
to remove
drugs from the market when they have
worrisome data
than they would if there wasn't a
litigious
society.
I think they worry more about
what it
means to their future class action suits
to
actually admit that there is a problem
until they
have all their lawyers lined up, so they
know how
to defend themselves, and if they weren't
worried
about the fact that the American public
has an
exaggerated misconception of what drugs
do and
work, they would act quicker.
I think the American public in
general
49
expects drugs to be safe and effective,
and they
don't recognize that drugs can be safe
and
effective if used correctly, but in the
wrong way,
are dangerous and shouldn't be used, and
they don't
get that.
They just don't get it.
I put E.R. Squibb down because
I know a
little bit about E.R. Squibb as an example
of the
leadership that the industry had back in
the 19th
century.
Dr. Squibb, a physician, wanted a higher
quality ether for anesthesia. This was an
extremely important drug in those days,
and so he
founded a company for the express
purposes of
making sure he had high quality ether.
He built it in Brooklyn, and
then his
company started making other things and
then he
noticed that there were other companies
that were
copying his products, calling them the
same thing
and putting them out there less
expensively, and he
said the public might be at risk if they
aren't
made correctly.
So, he did something unique
which I don't
think any of the companies would do
today. He got
50
all his formulas together, how he made
everything,
and he published them in the Journal of
the
American Pharmaceutical Association with
the
proviso that if anybody wanted to make a
product
that E.R. Squibb sold, they should make
it the way
we make it, so it would be of the same
quality, so
at least the public would have a good
quality
product, and if they could make it less
expensively
than we could, good luck to them.
Well, I wonder how many
companies are
ready to jump into that game. At that same time,
of course, Eli Lilly was producing well
over 100
generic products. It was the largest generic
manufacturer in the United States. It produced
everything that could be made that was
listed in
the USP or NF, extracts, and what have
you. It was
an interesting time.
Now we get into government
regulation.
Now, why did the government get into
regulation?
Well, it bought quinine that wasn't
quinine and it
got upset. So, in 1848, with the troops attacking
Mexico City, their quinine didn't work
like it was
51
supposed to, they said what's in here, it
wasn't
quinine, it was something else, I don't
know what
it was.
They said that's terrible,
terrible,
terrible, and so we needed to find a way
to make
sure that when something was labeled
quinine, it
really was indeed quinine. That was the first shot
out of the cannon.
We finally had the Food and
Drug Act of
1906, which really just said that if you
are going
to sell something and call it a drug, and
name it,
it ought to be what you call it. Right about that
time we got into the concept of
misbranding, which
was putting something in something and
calling it
something that it wasn't, and that is
basically
what misbranding is.
We have a lot of meetings for
misbranding
now, but the bottom line is that it is
not what it
is supposed to have been.
The Agency wasn't really
founded then, but
the government said that if we wanted to
take
action against the company that
misbranded a drug,
52
that it was incumbent upon the government
to prove
that the drug was indeed not what it said
it was,
and that there was intent to
defraud. If you do
that to the government, we can't enforce
any
quality on anybody, because we don't have
any
information to use for it.
But I want you to remember that
concept
that came about in the early 1900s,
because when we
get to the end of the 1900s, we have
another law
that brought us right back to that place.
So, 1938, we killed a bunch of
kids in the
New York City area with antifreeze as a
sweetening
agent in a sulfa drug preparation. That was the
end of a company's reputation, and well
it should
have been, and everybody was in an
uproar, so we
now have a new regulation. You will notice the
trend here - disaster, new regulation,
disaster,
new regulation. It's kind of a recurring theme.
So, we know said, okay, it has
to be what
it says it is, it has to contain what it
says it
contains, and it has to be safe, but it
doesn't
have to work.
53
Homeopathic remedies are
exactly that.
They are 1 to 100 dilutions of something
done 1,000
times.
You end up with a bottle of water, which
they claim contains the essence of the
power of
whatever drug was in the first bottle of
1,000
dilutions before. All right.
So, we can claim it
works, and it contains a diluted,
diluted, diluted,
fine, that is what it really
contains. You can't
find a molecule because you have diluted
more than
Avargordo's number, so we have products
on the
market.
By the way, the Food, Drug, and
Cosmetic
Act says specifically that drugs are
things that
are contained in the homeopathic
pharmacopeia,
which means that they are precluded from
acting
against products that are in the
homeopathic
pharmacopeia even though we know they
don't work.
We are still working with
things that are
just safe, but at least they are branded
right, you
know.
Nowadays we have people who claim that water
solves medical conditions. What the heck, you
know, 1938, that would have worked, put a
label on
54
water, say, if you will bottle water and
pay for it
at a rate higher than you pay for
gasoline, then,
it is better for you than the water you
get for
free out of the tap, and you will do
better.
Well, I don't know, I wonder
about things
like that. I have a problem my students always
complain about. I have one of those minds that
kind of wanders, and so I do that.
Let's get back to misspelled
words and
regulation. So, in 1951, two pharmacists
got
together, a guy named Carl Durham and a
guy named
Hubert Horatio Humphrey--I love his
name. They
were pharmacists. One was in Congress and one was
in the Senate. Hubert came from Minnesota. He
ultimately became vice president, ran for
president, didn't make it.
I often wonder what would
happen if the
president of the United States was really
a
physician or a pharmacist, a health care
worker,
what difference that would make in their
approach
to the health care problems.
So, they got together and they
said, you
55
know, there is a lot of drugs out there
that are
pretty dangerous, that the average person
really
can't understand, and maybe we ought to
have
somebody help them figure out what to
take, so they
established two criteria, prescription
drugs and
over-the-counter drugs. We still, by the way,
don't have to have them work. You know, God
forbid, they actually should work.
We are a unique country among
the
developed nations of the world. We only have two
categories of drugs. Most of them have
many more
categories of different levels, and, in
fact, I
like the Australian system. They are listed in the
group of things they call poisons, so we
clearly
know where they belong, right? They are the poison
list.
In 1962, we finally got around
to hoping
that we could figure out that the drugs
were both
safe and effective, so in 1962, we said,
okay, new
drugs have to be safe and effective. The Agency
was kind of curious. It said, but you can't tell
people that this is an approved drug,
because that
56
gives you a marketing advantage over the
drugs that
haven't been approved by us, and then we
don't know
are effective. Hmm, that's interesting.
Then, Congress, in its infinite
wisdom,
jumped right in there with DSHEA, and
DSHEA says
that if you aren't really a drug, but
kind of imply
that you are a drug, then, you can go
back to the
1906 regulation which says that it only
has to be
what it says it is, and it doesn't have
to be
proven to be safe or effective, and if
there is any
problems with it, the Agency has to
compile the
data before they can make you take it off
the
market.
I just love that, you know, retrograde
regulation, I just wonder about the
wisdom of that.
I am sure it has to do with the need that
the
public has for unsubstantiated claimed
herbal
remedies.
All right. Here is where we really get to
where the rubber meets the road, and that
is the
cost of drugs. I grew up in a pharmacy family. My
father was a pharmacist, my uncle was a
pharmacist,
I became a pharmacist because I didn't
know
57
anything else.
I grew up in a drugstore, and
when I was
at
a young age, I worked in my father's drugstore
as a soda jerk. Some people think that I
have never
gotten over the second half of that.
But in those days, the average
cost of
drugs that my father filled--he has a
wonderful
ledger, handwritten in ink pen where he
wrote down
the name of the patient, the
prescription, the
physician, and then the cost--and if you
look at
it, you will find that the average charge
to his
patient was $1.75.
I asked him one day, being a
nosy
teenager, how do we make money to live on
at the
store here, and he says, "Well, I
charge $1.75, but
it costs me about 25 cents of
goods." So, I said I
thought that was pretty good.
Of course, nowadays, the
average charge of
a prescription can be in the $50 or $60
range, and
the pharmacy gets $3.50. There has been a shift
here somewhere.
At that time, Tetracycline came
out. It
58
was about 50 cents a capsule. The price of
Tetracycline has gone down dramatically,
but we
keep bringing out new drugs, and I think
each time
we bring out a new drug, we say what was
the price
that we charged for the last new drug,
and we
multiply by 1.5.
You also understand that there
is
absolutely no relationship between the
charge for
the drug and the cost of actually
manufacturing it,
and that they factor in all of the other
costs to
maintain the corporate entity that
creates new
drugs.
So, they need to have this huge inflow of
money in order to float all of the
research and the
marketing, and all the other efforts that
go on,
and so that there is some disconnect.
Waxman and Hatch got
together. We
recognized back in the 1980s that the
cost of
health care was going up quickly. Uwe Reinhardt
has a wonderful graph that he puts up, a
Princeton
economist, that shows the gross national
product
and its rate of increase and the cost of
health
care and its rate of increase, and then
he predicts
59
some date in the future where the two
will meet.
Then, he has a cartoon where he
has two
physicians lying in beds in the hospital
together,
prescribing for each other, and he said
that is
going to be the entire productivity of
the United
States is going to be this.
So, we know that there is a
disaster in
the future and what are we going to do
about it,
and we have a culture in the United
States where we
don't regulate the price of drugs. We are again
unique.
Very few developed countries have that
compunction. So, we try to regulate it through
competition.
So, the Waxman-Hatch Act or the
Hatch-Waxman Act, depending on whether
you are a
Republican or a Democrat, came into
being, and it
was a compromise that was supposed to
benefit the
innovator companies by ensuring them a
reasonable
patent extension or exclusivity time
frame in order
to recoup the investment to bring the new
drug to
the market, and established rules and
regulations
for the development of generic drugs.
60
It seems to work in some areas
and we hope
for the best. However, it is not going to be the
end of the issue, and if we start to make
projections out in the future, we are
going to have
to do more than that in terms of cost,
but it was
the first time that the FDA was an active
participant in controlling costs.
I think I see that as something
going
forward.
We have a problem, of course, with other
issues associated with cost, and, of
course, here
comes re-importation, and we are going to
get into
that in a little bit, but I don't want to
beat a
dead horse.
My wife is Canadian and my
inlaws are in
Canada, and they see the U.S. news come
across that
says that Canadian drugs are bad for
American
citizens, and they say, oh, and they call
their
son-in-law, the expert, and they say,
"What's wrong
with Canadian drugs?" Of course, i am hard pressed
to say anything about it, because there
is nothing
wrong with Canadian drugs. So, that makes an
interesting argument. I think we can go down that
61
road as long as we want.
I think the next level of
regulation is
going to be the line on top. People are going to
want information that shows that the next
new drug
is not only safe and effective, but
better. I
don't know how long it is going to take for
Congress to do that, but that is what is
coming.
We have a history of producing
lots of
drugs that might be different, but not
necessarily
an improvement, where are we going to go,
and I
think both the industry and the Agency
should be
prepared to think about how they would
handle that
situation.
Remember that we are trying to
regulate
according to best science, and sometimes
we lose
track of best science. There are some classic
equations that we use that we depend upon
to help
us decide what is good science. One is the
Noyes-Whitney expression. The Noyes-Whitney
expression describes dissolution profile,
and it
was developed by these gentlemen using a
very
interesting standard material. It was a fused
62
cylinder of material.
So, their apparatus and how
they did it
were standardized based on one solid hunk
of an
individual chemical in the cylindrical
form, so
they could accurately determine the area
exposed to
the fluid and therefore, from it,
determine all of
the
equations. Nowadays we use a
standardized
compressed tablet. I would argue that the
dissolution apparatus is probably less
variable
than an individual tablet coming off a
tablet run.
If you wanted to standardize an
apparatus,
you ought to standardize it with
something which is
less variable than the apparatus you are
standardizing. I wonder about that. I guess we
could ask our colleagues down the street
what they
think about that, but let's go back to
the basic
science and figure out what is going on.
The other one I like to talk
about
occasionally is Arrhenius. Arrhenius developed a
relationship between temperature and rate
of
reaction that was developed for reactions
that
happened in dilute solutions.
63
We apply them, same rules, too,
tablets,
ointments, creams, and lotions. We put things
aside for three months at elevated
temperature, and
we say this is going to predict what is
going on in
two years. We will give you two years, just send
us the real data later. I would argue that if we
went through the data that the Agency
has, that we
would be hard pressed to get a
correlation
coefficient much over 0.3 for that data.
The other thing is what is the
rule and
regulation. When a rule or regulation gets out
there and purports to be doing something, and
it
doesn't, it makes you wonder. We have a regulation
that says you have to do accelerated
stability at
40 degrees and 75 percent relative
humidity, but
you can take the humidity and temperature
chamber
and you can put in it a tablet container
that is
sealed with a descant in it and do the
study.
That is kind of like saying
let's see how
fast ice cream can melt in the kitchen,
but you are
allowed to put it in the freezer. I wonder, you
know, I just wonder. I am just kind of curious
64
about those kinds of things. You sit around in an
academic office, you are a tenured full
professor,
what are they going to do. You wonder about those
things.
I think that there is going to
be a lot of
international regulation. I think that we are at
the stage where the companies are truly
international. The largest provider of generic
drugs in the United States is in Tel
Aviv. Most of
the big developmental innovator companies
are
really housed everywhere.
In fact, the numbers of workers
at
pharmaceutical plants in the world has
shifted from
the United States out. If that is true,
then, we
really have to have cooperative control
on quality.
I am sure that England and Germany and
France want
the same high quality of drugs as we do,
as the
Canadian Health Protection Branch insists
that they
do.
So, we need to go in the
direction of what
is truly a harmonized or
internationalized
regulation of quality. We need to somehow control
65
the cost to the consumer, and if we don't
find a
way to do that, it will be imposed on us.
One of the problems I have with
all of
this is that drug costs to consumer seems
to make
the news way more than the cost of a bed
in the
hospital.
Now, I will just ask you, how much does
it cost to be in a hospital bed. Does anyone know?
No, but you sure know how much it costs
for a
bottle of Viagra--oh, excuse me.
DR. SINGPURWALLA: I don't.
DR. KIBBE: Oh, there is a man with
confidence.
[Laughter.]
DR. KIBBE: The reason is that most of us
in the public are covered by some
insurance plan
that covers the cost of the hospital bed,
but we
aren't covered by drugs, and drugs
represent 8 to
10 percent of the total cost of health
care in the
United States, and if you look at it, it
is much
cheaper to give reasonably expensive
drugs to
patients than to put them in a
hospital. But the
patients don't pay for it out of pocket.
66
I wonder why the huge lobbying
efforts of
the pharmaceutical company isn't applied
to getting
drugs covered by Medicare and Medicaid
instead of
anything else. If they could ever do
that, they
could forget about the arguments in the
newspaper
about the cost of drugs.
I am sure there is lots of
economic issues
associated with that.
The last three things I have
are
continuous quality improvement, PAT, and
federally-funded efficacy testing. I don't know
whether we are going to get the right to
demand
that you do an efficacy test against
seven or eight
of your competitors in order to get
approval, but I
think that the world deserves a chance to
look at
what is those relative efficacies in an
abstract or
at least impartial way.
PAT has been fun for me. I think it's a
wonderful initiative, it has its own
journal now,
those of you who are interested in
it. It has got
a forward written by--oh, my
heavens--Ajaz. It has
some beautiful pictures in here.
67
I went through it immediately
and wanted
to see if I knew anybody that actually
was involved
in PAT, and there is a whole bunch of
really pretty
pictures of all sorts of people that were
actually
on the committee with it, if anybody is
interested
in
it. I thought that was pretty neat.
I think that there are things
in the
horizon that really threaten the way we
do business
both at the industrial level and at the
regulatory
level.
One of them is the development of
nanotechnology and computational power.
We are looking forward to a
singularity in
computational power, a point beyond which
we cannot
predict or even understand the
future. In
approximately 2014 or 15, the computer on
your desk
will have not only digital computational
power, but
parallel processing, and will be able to
think
better than you can. We will be able to process
data, come up with new ideas, and, in
fact, at some
point in time, it will be the most
intelligent
being on the planet, and we humans will
relegate
ourselves to second place.
68
When that happens, what do we
do about
health care? And let's look at nanobots and what
they can do. If aging is truly a degradation of
the DNA strand within people, if we can
inject
nanobots who know how to count DNA
strands and
repair them, how are we going to age?
If we have the capacity to scan
individual
molecules and relationship in the neural
net, can
we then scan down a person's entire
knowledge base
and personality, and shift it from a
carbon-based,
short term, to a silicon-based, long term holding
facility?
How many of us would be willing
at the
ends of our days to become virtual us in
a virtual
environment?
Where are we going? Challenges to the
FDA.
In our experience over the last several
millennium, an ever- increasing rate of
new
technological development. It was 20,000 years
from the time we developed hand-held rock
until we
actually made a bow and arrow with a
processed
rock, and the rate at which we develop
things now
69
is astronomical.
We need to have improved
productivity in
the industry, but that needs to be
related to an
improvement in the cost of goods sold to
the
American public, and the Agency needs to
maintain
public confidence. It needs to not say things that
are clearly difficult to defend in the
public
environment.
It needs to be responsive to the
public needs and realistic, so that the
public
understands the expectations of drugs.
If there was any advertising
that the
Agency could do, that I think would help
in the
long run, it is to get the American
public to
understand that drugs are not safe, that
they can
be used safely.
The American public has an
unrealistic
expectation for their medications and an
unrealistic expectation of how they
should feel as
they go through life, and they expect
that these
little pills will do it for them, and it
won't, and
we need to get them to stop thinking that
way.
We need to maintain and improve
70
international cooperation in both
regulation and
harmonization, and we need to, in the
final
analysis, decriminalize Grandma. When she crosses
the border to pick up drugs, she needs to
understand that we don't think that she
is
committing a heinous crime against
society, that we
understand that the economics are driving
her to
it, and we need to find a way of making
it happen
for
her, so that she can get the drugs she needs at
the price she can afford.
Does anybody have any
questions?
[Applause.]
DR. KIBBE: Thank you, Dr. Kibbe, for that
exhilarating presentation. I am sorry, I just love
those kinds of things.
Now, we are going to get into
some serious
stuff here, because Ajaz is going to get
up to the
podium.
DR. HUSSAIN: Could we just take a break
now
and then start after the break?
DR. KIBBE: I am still fired up, you know,
whatever you want to do. You know the energy level
71
after making a presentation. I really want to
complain about the lack of a
soapbox. I asked for
a soapbox up there because I knew I was
going to
get on my soapbox.
Ajaz wants to take a
break. Let's try to
get back and get back to work at two
minutes to
10:00.
[Recess.]
DR. KIBBE: We have comments on my talk
that some members would like to make, and
then I am
going to be more than happy to add to my
talk a few
other issues, so we might have a lot of
fun today.
As is the tradition with this
year's
committee, Nozer has a comment.
DR. SINGPURWALLA: I don't have a comment,
I have a question for you. The question is what
would your reaction be to the idea of
nationalizing
the drug industry?
DR. KIBBE: That is a wonderful question
and I think the answer to it resides with
our
colleagues over there. I know that if they ever
did that, I would volunteer to be drug
czar. There
72
are a couple of issues that I didn't hit
on in my
thing.
One of them is direct-to-consumer
advertising. I think the issue of why the public
has the misconception that drugs are not
safe can
be tied directly to direct-to-consumer
advertising.
Many years ago, in my one
opportunity to
appear on the Today Show, I was
interviewed by
Debra Norville on the topic, and I was
debating an
industry representative, and I said that
it would
completely change the dynamics of
prescribing and
using of drugs in the United States, and
I think it
has.
Two days ago, I was sitting at home
watching TV, and for an hour and a half,
every
single ad on TV, every single ad was for
a
prescription drug, and it just has to
have a
dramatic effect on the way patients
interact with
their physician and how they get health
care. I
think it was a mistake, but we can
comment on that,
too.
Does anybody want to throw a
few cents'
worth in while we are prognosticating?
73
MR. CLARK: You mentioned something about
E.R. Squibb challenging the world to meet
his
efficiency in his products that he
manufactured. I
was just trying to point out that while
he
challenged the world, that challenge
could prove
fatal today, because today, Mr. Squibb or
Dr.
Squibb would be required to freeze his
manufacturing technique, whatever it may
have been,
and that while his challengers came in
with new
techniques, he would be burdened with an
approval
process that would slow down his ability
to
compete, and we should be able to create
a
regulatory environment that protects the
public as
it still encourages innovation, and not
just
encourages the innovation for
innovation's sake,
but encourages applying it to the
products and to
improve the entire environment.
DR. KIBBE: Clearly, he couldn't do what
he did then now, because the Federal
Government is
in his business now.
MR. CLARK: Exactly.
DR. KIBBE: And that has happened after
74
World War II. Before World War II, the Federal
Government stayed out of everybody's
business, and
that is a dramatic change in the way we
do business
in the United States.
We need to get to the desired
state--I
recommend Pennsylvania, far less
hurricanes--the
desired state, however, is going to be
defined by
Ajaz.
The "Desired
State" of Science and
Risk-based Regulatory
Policies
DR. HUSSAIN: I will do it from here. In
a sense, what we wanted to do, sort of
build on the
Manufacturing Committee discussion that
was
reported quite elaborately by Judy
Boehlert, the
chair of that committee, but to sort of
now do a
gap analysis, what we see as gaps between
the
current state and the desired state from
an
internal FDA perspective, what are the
challenges
we face internally, and get your feedback
on that.
So, what we have are three
presentations,
one, Helen will sort of look at the
organizational
issues, I will try to identify some of
the
75
scientific gaps, and Jon will identify
some of the
policy gaps and how we intend to sort of
fill those
gaps.
If you could sort of give us
feedback on
are we missing in our gap analysis, it is
a
preliminary gap analysis right now, and
then how we
proceed, and then this will be followed
by
discussions and presentations by PhRMA
and GPhA
perspective on how they see the progress
we have
made and some of the challenges that
remain.
So, that is the discussion for
this
morning.
DR. KIBBE: That means you are passing the
ball to Helen.
DR. HUSSAIN: Yes.
DR. KIBBE: Let's see how you follow my
act.
Organizational Gap
Analysis
MS. WINKLE: Believe me, in 100 years, not
only could I not only follow your act, I
wouldn't
know where to begin, and my topic is so
boring
anyway.
76
I am going to talk about
organizational
gap in the Agency right now as far as the
desired
state is concerned, and as Ajaz said,
this is sort
of a follow-up to some of the things we
talked
about at the Manufacturing Subcommittee,
and I
think it is really important that we look
at these
gaps and talk more about them, and sort
of discuss
how we can possibly fill some of them.
I have some ideas on filling on
some of
them, but I think there is a lot more
that we will
need.
DR. KIBBE: There appears to be a gap in
the computational problem, too.
[Pause.]
DR. KIBBE: We are passing around a
transportation note for people who need
help to get
to the airport.
MS. WINKLE: Is everybody leaving now?
Gosh, you could at least have given me a
chance.
[Laughter.]
MS. WINKLE: As I said, I am going to talk
about the organizational gaps and
reaching the
77
desired state, and I wanted to start off
with, just
a second, showing you the organizational
chart of
OPS, because I think as I talk about
organizational
gaps, you need to know a little bit about
what the
organization looks like, and I think you
have a
good idea, but I just wanted to point out
we do
have four offices.
You actually heard from all
four of those
offices yesterday, but you say, in
yellow, where
the CMC is done in all four offices, so
almost
every part of the organization in some
way is
affected by the changes that are being
made by the
new paradigm and what we are trying to
accomplish
with the desired gap, which complicates
the issues
somewhat.
It is very important as we look
at the
organizational gaps that it is
multi-dimensional,
it goes across all of the
organization. It is
between organizations and it is within
organizations. There is lots of gaps here and we
need to look at all of these gaps and
figure out
how we are going to handle them.
78
It is outside of OPS and other
parts of
CDER.
We really do a lot of work with products
with devices with CBER, so we need to be
sure that
those gaps are closed as we move forward
in trying
to accomplish the desired state.
So, basically, what I am going
to be
talking about here is what we need to
consider and
resolve in our process or processes
before we can
adequately implement regulatory direction
and
support through applications process and
review of
what we are calling the desired state.
I also want to point out as I
talk, and
you will see this a lot, that the
organizational
gaps that I am going to point out really
intersect
with the science gaps that Ajaz is going
to talk
about and the policy gaps that Jon is
going to talk
about, and you probably can't really
address any of
these separately although that is what we
are
making an attempt to do here. But again as I go
through the organizational gaps, you will
see a lot
of the intersections.
What constitutes the gap in OPS
and what
79
are actually the process issues for
implementing
the desired state, and how we will review
at
different levels? This is really some of what we
need to talk about.
One of the big problems here is
the
appropriate utilization and focus of
available
resources. I am reading it wrong. This is why I
am having problems. It is the resources. We have
a lot of human resources. You have already heard
some of the issues that we have had with
how to use
our best resources and how to focus those
resources
on those issues that are most
important. So, that
is really one of the things that we have
as part of
the gap.
We are not always focused on
those issues
which are the most important, and we
don't always
have the science expertise available to
focus on
the gaps or focus on the issues
correctly. So,
this is a big gap that we have across the
entire
organization.
There is a difference in
products and
regulatory requirements and review
processes. We
80
are regulating ANDAs, NDAs, and BLAs, and
BLAs even
fall under a different act than the ANDAs
and the
NDAs.
So, there are some complications and some
gaps there that we are going to have to
look at and
determine how best to handle.
The organizational structure,
the way it
is set up really creates a really large
gap in how
we are going to move forward.
I think we have made it clear
in the past
that in ONDC, I know Moheb has talked
about this at
different times, Dr. Nasr has talked
about this at
different times, that we have chemists
from the
Office of New Drug Chemistry that are
located in
the different clinical divisions, so that
we lack
consistency in how they make decisions often,
because it is done outside of the whole
chemistry
structure, so to speak. It is done within the
Clinical Division, and we also lack the
flexibility
of being able to use our staff and to
utilize the
science and the staff because of these
collocations.
Actually, we have chemists in
18 different
81
teams across the Clinical Divisions, and
they very
rarely interact with each other, so it
really
causes a lot of complications in how we
do our
work, and it will cause even more
complications
when we get into the new paradigm.
I think one of the main gaps is
that we
are very process driven, not science
driven. This
goes back to the earlier comment by Dr.
Kibbe. We
are regulating a science industry. It is a science
industry that we are regulating through
process.
Some of the things that contribute
to this
is PDUFA in generic drugs, first in,
first
reviewed.
We have a tiered approach to our
reviews.
We have heavy backlogs. I think
that
Gary has made that point several times to
this
committee. The workloads are big, the backlogs are
big.
So, that is really driving us, too, to focus
more on process than science. So, this is causing
us to really have to rethink how we want
to do
things.
Part of what is adding to that
workload
and to the backlog is that we get too
many
82
supplements. We require supplements on little
changes that really have no significance
in the
manufacturing process.
Also, part of the gap is the
interaction
with inspection. We have a lack of appropriate
reviewer involvement, and we get no
feedback. We
do not get copies of 483s. Once they have been in
to the industry with the observations, we
don't
even have any correspondence in most
cases with the
inspection people on things that they
find when
they go out on inspections.
So, how you are supposed to
really have
knowledge about the products that you are
reviewing
in the future or where you can use that
knowledge
that has been gathered and incorporate
that into
your thinking about reviews and products,
you can't
do, so that really creates a lot of gaps.
One of the things that is going
to create
a gap in the future is the possibility of
having a
two-tiered system. As we talk about the
desired
state, as we talk about the things that
are
required under the desired state, we
don't have
83
regulations that are going to require
manufacturers
to submit pharmaceutical development
information.
We don't have regulations that are going
to require
them to do this thing or that thing, and
in some
places, I am not even sure we have the
carrot to
encourage them to do that.
So, you are going to have some
companies
that are naturally going to submit this
stuff, or
naturally going to move toward PAT and
toward other
aspects of improving on their
manufacture, but you
are going to have companies that don't,
so what we
are looking at is the possibility of having
a
two-tiered system which is going to
create a gap
even within one reviewer.
He is going to have to be able
to look at
both tiers and make decisions, and this
is going to
complicate issues a lot when we move
ahead.
We use guidances to accomplish
consistency, and those guidances are
sometimes very
prescriptive, and this adds to the whole
gap, and
also not only are we using guidances for
consistency, they are also up for
interpretation.
84
Unless they are prescriptive, they are
interpreted
differently by different people, so
obviously, we
have some concerns about this.
Organizational components are too
reactive, and not proactive. Now, this is caused
by workload, and the workload continues
to
perpetuate the problem.
You have to be reactive because
you have
so much work piled up in your In box that
that is
what you have to focus on, and it is very
hard to
be creative and innovative and think
about those
issues and problems that you are going to
have down
the road, think about, as Dr. Kibbe was
talking,
new therapeutics that are coming along or
new novel
delivery systems or different things like
that, too
busy moving the freight from day to day.
Use of available scientific
expertise and
scientific collaboration. Often within especially
in ONDC, because they are broken up
according to
the clinical divisions, you may not have
the
necessary scientific expertise to look at
an issue,
to look at a problem, to know really what
the right
85
direction is for making a decision on a
product.
Also, we do not go out and use
a lot of
scientific collaboration. I mean we have a lot of
SGEs, we have several in this room that
are helping
us on different scientific issues of a
broader
nature, but we could be taking advantage
of some of
those and calling and getting more
information in
the future.
There is a challenge in focusing on the
appropriate questions or what are the
right
questions. Reviewers have a tendency--and this is
not any kind of negative against
reviewers--but
they do have a tendency to look at all
the data
that is provided, and we have not focused
down on
what the appropriate data is, and,
therefore, the
appropriate questions that we need to
have
answered.
We have a lack of utilization
of
appropriate tools. We could be using statistics
more, all of us, to get better answers to
some of
the questions that we have around
review. There
are other tools, as well, that we could
be using
86
that we are not. One of the big areas I think that
causes a gap is the lack of communication
between
disciplines, but I do want to add to
this, there is
also a lack of communication between
organizations
or
components of the organization, and this is one
of the things that we need to focus on to
help
close the gap.
So, I did take a look at what
we had done
so far for closing the gap, because I
think it is
important to emphasize some of the stuff,
because
we do realize that we have some big gaps
here.
I do want to upfront say,
though, that
these are not all of things that we need
to do. I
know that there is a lot more down the
road that is
going to come along, and I am really
looking for
advice from the Advisory Committee as to
some of
the things that we need to be thinking
more
carefully about, or make suggestions for
some of
the things that we could be doing to help
close
this organizational gap.
One of the things we have been
doing is
making some structural changes in the
organization
87
In
the Office of New Drug Chemistry, which Judy
talked about yesterday for the
Manufacturing
Subcommittee, we are reorganizing the
Office of New
Drug Chemistry. We are actually doing away with
the collocation and making one Office of
Chemistry
when we move to White Oak.
We feel that this is going to
give us a
lot of consistency or at least more
consistency,
and give us the opportunity to have more
flexibility as to how we look at the
review
process.
We feel that this is going to have some
real advantages to us.
We are also, in our Office of
Generic
Drugs, we have set up a third division
for doing
chemistry. The workload is so heavy in the office
that we felt like if we had more
divisions where we
could spend more time and focus more on
some of the
issues, that we could help in some of the
gap
problems, having reviewers on inspections
or as
consultants to inspection, so have
complete
knowledge on products and the results of
inspections.
88
This is something that we have
been
working on. We have been working with our Office
of Compliance and with our field
component. We
feel like we would like to have reviewers
on
inspection. We think it is very important for them
to go out and provide some of the
scientific
knowledge to the inspectors as the
inspections are
being done, but I don't think that part
of the
question even came up yesterday on
resources to do
this.
This is a resource issue. You are taking
people away from their desk to go and--we
have
already talked about the workload being
high--taking people away from the desk to
go out on
inspections and spend time away from
their desks,
but also this is costly, and like it or
not, we are
not flush with money, so we won't be able
to do
this in every inspection.
But I do think it is important
that before
the inspections are done, even though the
reviewers
can't go out, that they provide
consultation to the
inspectors and talk about some of the
issues that
89
they have seen in the reviews of these
particular
companies and give them some advice on
what they
may want to focus on more in the
inspections.
One of the big things that is
really
necessary, in my mind, to closing the
gaps, and
again this sort of goes across the whole
concept of
the science gap and the guidance gap, and
our
policy gap, as well, is that we have a
lot of
questions we still need to answer and
address.
This is only a few of them, but
I think
there is a lot of things that we have not
come to
grips with on manufacturing science and
how that is
going to affect our review and what our
review
process is going to be to handle these
things in
the future - things like quality overall
summaries.
Dr. Nasr talks about
incorporating this
into the process of ONDC. We have not come to any
conclusions on this. We are still in the proposal
stage.
We need to decide, if this is the direction
we want to go, what is the benefit of it
to us, to
the industry, and what we really need to
see in
that QOS.
So, that is a question that we need to
90
look at.
What we need in the way of
pharmaceutical
development information. There is a lot of
information that these companies have in
the
pharmaceutical development arena, and do
we want
all of that information. If we get that
information, what are we supposed to look
at, what
would we focus on. We need to answer those
questions, it is very important, before
we start
asking for this information.
We have to have addressed that
before, I
think, companies are going to feel
comfortable in
providing it. I think many companies see this as
just more information they are sending us
to look
at and more questions they are going to
get from
us, so we really have to develop our
processes.
We also need to look at things
like how
industry will determine critical
attributes, so as
we look toward the desired state, that we
are able
to regulate that and include those in our
process,
and we need to know in all these cases
what we will
do as far as reviewing these.
91
This is just a small part of
the
questions, I think, that we need to be
addressing.
Also, as far as closing the
gap, we need
to have a better understanding of what
constitutes
the design space across all products, and
once we
have a good feel about that, or
understand that, we
need to know when notification to FDA is
necessary
for change in manufacturing.
We have not reached these
conclusions yet,
and we need to have working groups,
whatever it
takes, to really develop our own internal
thinking
on this and work with industry to make
sure that
the direction we are going is going to be
useful to
them.
We need to have a better
understanding of
what risk for a product is and develop a
systematic
risk approach to review processes. I keep seeing
time and time again, people talk about
risk
management or risk processes, and stuff
like that.
I think when you talk about risk
management, you
are talking about something different for
every
person.
92
I mean what is in my mind, what
is in
everybody else's mind in this room could
be
entirely different, and we at the Agency
need to
narrow down as far as review is
concerned, decide
what that means, how we are going to use it,
and
have a very, very concise program for
ensuring that
we do look at this in a fair and
equitable way.
Guidances. Again, Jon is going to talk
about guidances, but it is really
necessary for us
to go back. This will help us close the
gap, but we
need to go back and look at our
guidance. We have
many guidances out there that are
outdated, many
guidances that are overly prescriptive,
many
guidances that don't fit into the new
paradigm at
all.
Jon is in the process, he and
his staff,
of going through the guidances, removing
some,
redoing some, and looking at what
guidances we will
need for the future in order to
incorporate some of
the changes.
Training. This goes to the question Mel
asked, training, training, training is
necessary
93
here.
We have so much to train. We are
doing some
training, and I will talk about some of
that, but I
think it is really important and part of
what we
need to do to close the gap, is determine
what
really training our reviewers need and
what
training is necessary for the industry,
and I don't
think we have come to those conclusions
yet.
We need to determine how we are
going to
work under a two-tiered system if, in
fact, that is
the direction that we go, and we need to
have
developed the processes for doing
that. We need to
develop an internal system for handling
differences
in Review Divisions.
I met a couple of months ago
with PhRMA on
a RAC Committee on a dialogue session,
and this was
one of the things that came up was the
need for a
dispute resolution process, some kind of
mechanism
where, when there are differences in
review, that
we are able to handle those decisions and
get
information out as to how these issues
are
resolved.
The last thing I have on
here--and
94
actually, I wrote this slide before we
even had
some of the conversations yesterday--was
what is
really important is we need to have
appropriate
metrics for measuring things.
Today, in the review, we
measure by what
we accomplish, how many supplements we
get. Well,
let me tell you when you are measuring
supplements,
and that is an indication of how good you
are doing
your job, you are going to want more
supplements.
We have got to really back off
of the
metrics that we currently have and look
for those
appropriate metrics to help close these
gaps.
So, some of the current steps
we have
across OPS, we mentioned before that we
are setting
up a working group under the
Manufacturing
Subcommittee of the Advisory Committee to
begin to
address many of these questions that we
have. We
are also setting up some CRADAs to get
some case
studies to help us, too, in getting a
better
understanding of these issues and how we
are going
to handle them in our processes.
ICH, too, is going to be an
important part
95
of helping us handle some of our
organizational
decisions especially Q8 as it looks at
the desired
state and implement some guidelines on
it.
We are doing some
workshops. We have a
Workshop of Specification Setting and
looking at
how we need to have a mechanistic understanding
of
setting specifications in line with the
direction
that we are going.
That particular workshop is set
up in
March, but I will be upfront with the
fact that I
think there is still going to be issues that
come
out of that workshop where we are going
to have to
look more specifically at some of the
specification
areas and really dig deeper into them, so
I really
anticipate more workshops than this just
in the
area of specifications, but I think a
number of
workshops are on the horizon in order for
us to be
able to address many of these questions.
I actually think, too, one of
the things
that we are probably looking at having a
workshop
on is quality overall summaries. If that is the
direction we want to go, I think we need
input from
96
the industry and others, so that we have
a better
understanding of what we need for using
that type
of process and what it means to us in the
industry
when we do.
We already have some
collaboration with
academics. As I said, we are involved in
several
CRADAs or in the development of several
CRADAs. I
think these are going to be very helpful
for us in
getting a better understanding of some of
the areas
that we need to, or some of the questions
we need
to, answer, so that we can fill some of
those gaps,
and we have been doing some work with the
Product
Quality Research Institute.
As I already said, we have an
internal
review of our current guidances, which is
very
important to helping us have the
appropriate
guidances out there. We are developing a program
for team interactions for inspections.
We are sort of basing this on
how we have
handled PAT and the team approach, and we
are
working with Compliance in the field to
develop a
better way of handling inspections and
including
97
the Review folks in those inspections,
and also a
better way of getting the findings from
those
inspections.
Training for reviewers. We have already
had a number of scientific seminars. We have
started that, especially, OGD has one
every six
months or so, and those seminars have
been very
beneficial to our staff in helping us have a
better
understanding of where we need to go, but
we need
more seminars and we need, again, really
a set
training program for all of our
reviewers.
We did form an OPS Coordinating
Committee
within the immediate office, and
actually, Keith
Webber and Gary Buehler are the chairs of
that
committee, and we will be looking at all
the issues
that come into OPS to try to ensure that
we have
consistency throughout all of OPS on
handling
these.
One of the other things that we
are in the
process, I think, that will help with the
gap is
finalizing the guidance on comparability
protocol.
Also, in ONDC, as I said, we
are really
98
changing the organizational structure,
but much
more than changing the organizational
structure, we
are changing from a review program to an
assessment
program, and that assessment program will
focus on
quality attributes of the manufacturing
including
chemistry, pharmaceutical formulation,
and the
manufacturing process.
So, the significant thing here
is that we
will be looking at much more the
chemistry, the
CMC, and that is where the assessment
program is
focused.
We have the proposed QOS. We have also
implemented a team approach. We are establishing a
peer review process. We have already done this on
a limited basis to provide more
scientific input to
our scientists in their review processes,
and
helping everyone have a better
understanding and
sharing the knowledge that they learn
from the
reviews.
We are implementing a Quality
Systems
approach. One of the things, too, that
ONDC is
doing, is they are developing a mock NDA
under the
99
new paradigm, under the desired state
paradigm, so
that they will have a better feel for
some of the
questions and issues that can come up,
and they are
looking at reducing supplement
requirements.
OGD has reorganized, as well. As I
mentioned, they have an additional
Chemistry
Division.
They are looking at changes in the
supplement review and evaluation to
determine if
some of the supplements can be eliminated.
They have also taken on the
team approach
on some applications, so that they have
better
utilization of scientific expertise and
ensure
consistency across similar product areas,
and they
are also looking at efficiencies in
review to
eliminate redundant or non-essential
review
activities. So, they are very much involved, too,
in some of the things that we need to be
focused on
in order to eliminate the gap.
OBP, which is, of course, our
newest
review organization, is looking at
supplement
requirements to determine where we can
eliminate or
reduce supplements.
100
Some additional steps. I think we need to
involve stakeholders in the review of
guidances.
Maybe under the Manufacturing
Subcommittee we need
a group that looks at some of the
guidances. I
don't know how we need to do this, but I
think it
is a step we need to do.
We need to determine how to
handle the
two-tiered approach, if we do it at
all. I have
mentioned this before, and I think it is
going to
be important to involve industry and
others in
doing that.
We need to have external
workshops,
develop a dispute resolution
process. One of the
things, too, besides looking at regular
GMP
inspections, we really need to look at
better ways
to
handle pre-approval inspection process.
I would really like to see
reviewers more
involved in making some of the decisions
on whether
to do pre-approval inspections and to set
better
criteria for getting those done plus
participate on
the inspections if we feel they are
necessary, and
develop appropriate metrics. These are things we
101
haven't started on, but are obviously
necessary,
and I am sure there is others.
Just to finish up, observations
and
conclusions. I think we need to continue to
address and analyze the organizational
gap issues.
I think they are going to be really
important to
us, to have determined what they are and
to resolve
how we are going to handle them in the
future in
order to move in the direction that we
need to do
to be able to regulate under the desired
state.
One of the things I think that
is very
important that we need to think about is
culture.
When I talk about culture, I am talking
about the
culture within FDA, and I am talking
about the
culture outside of FDA.
There are a lot of changes that
need to be
here.
I realize that the culture is always a
different area of thing to handle. I thought
Jerry's example was an excellent example
of how the
culture is a problem in some of the
things that we
are trying to do, and this is one of the
things
that we have got to manage and figure out
how best
102
to handle.
All of this, all of this, the
changes in
the organizational gap will take time,
and we need
to be dedicating the time to make it
happen.
Also, as I said, a lot of this
depends on
resolving some of the science gaps that
we have.
We need to include stakeholders in making
some of
the decisions and developing some of the
procedures. We need to work closely, I think, with
the stakeholders or we are really not
going to have
the answers that we need.
The training of reviewers is
important,
and the thing I think that is going to be
something
that we have got to be open to is that as
we move
forward, we are going to see other gaps
that we
haven't anticipated, and we are going to
have to be
ready to fill those.
That is all I have. Thank you.
DR. KIBBE: Thank you, Helen.
Should we take questions or you
want to
move to the--
MS. WINKLE: Let's go through all of it,
103
yes.
DR. KIBBE: I will hold my really tough
and incisive questions until later.
Scientific Gap Analysis
DR. HUSSAIN:
Last night I had a phone
call and I couldn't answer that, but this
morning,
at 7:00, I had another phone call from a
company
which recently got an approval for an
inhalation
product, and they were ecstatic. They had
submitted
a complete development pharmaceutics
report, and
that process went extremely well. This was a
one-cycle review approval for an
important product
including all the development report.
So, I think that is a wonderful
example
that shows ONDC has already moved and
things are
moving in this direction already. We probably will
make a case study out of it, and so
forth.
Anyway, I would like to sort of
focus on
the
scientific gaps. I will use some
background
information. I know a number of members on this
committee who are new, so I thought I
will spend
some starting with the basics.
104
I think, as Dr. Woodcock had
come to the
Manufacturing Committee, her articulation
of what
is quality has really come to almost
fruition, and
she is publishing this article soon. The
definition of quality is fundamental as
we move
forward, and there are some challenges as
we move
forward.
Good pharmaceutical quality
essentially
means an acceptably low risk of failing
to achieve
the desired clinical attributes. That is the goal
of achieving quality.
The challenge has always been,
and you
heard many of the discussions yesterday,
saying the
weakest link--and the weakest link is
what we have
to strengthen and address--how do we link
measurements and risk?
I think what we believe quality
by design
approach that we are developing under ICH
Q8 is a
way to help that. It is a multivariate
model. It
is characterized during development. You have to
think about, when you think about quality
by
design, the clinical is a confirmation of
that.
105
That is the fundamental aspect that I
think is
going to be a significant challenge in
how we
achieve that.
At the same time, you have to
remember
that development is only one part of
it. You
essentially have to make sure you have a
quality
system.
The final link between product and
customer-driven quality attributes really
means you
have a good quality system for
manufacture that
brought us consistently also, that
requires
integrated product and process knowledge
on an
ongoing basis, because you don't stop
learning at
the time of approval. In fact, you learn quite
significantly after manufacturing status.
You have to assure ongoing
control, and
you have to enable continuous
improvement.
In summary, I think Dr.
Woodcock
articulated this at our ONDC scientific
rounds on
October 6th. The future definition of quality
should be probabilistic in nature. That
is the
fundamental aspect, and we are not there
yet.
106
Science management, risk
management, and
quality management are critical aspects,
and I
think we really would like to be leaders
in this,
and I personally believe that we are.
But let me take a look
backwards from
beginning with the end in mind. Since we started
the PAT Initiative, the cGMP Initiative,
our focus
has been on looking at the entire system,
and we
have been looking backwards from a
manufacturing
end to the entire discovery development
product,
and it is what do we learn from that.
But before you look, before you
measure,
it
is always good to make sure your measurement
system is good, so you get your eyes
checked. That
is the symbol there.
The reason I was so sensitive
to that is I
think the dissolution variability from a
manufacturing end, we really fully
appreciate it
when we are putting that white paper
together, that
today, companies don't have the ability
to document
lower variability than that of the
calibrated
tablet, and which is made with the
conventional
107
method, and so forth. So, that was I think a stark
reality that a lot of us fully understood
during
this process. Art mentioned that, and so forth.
So, what are the challenges
here in the
sense the challenge is organizational
communication, and knowledge sharing and
information sharing. If you work in silos, the
boundaries between organization, which I
call
interface, the quality of the interface
between
functional unit, that means the
effectiveness and
efficiency of the process, the interface
can be
handoffs between functions, and often is
in need
for better coordination, and that is what
we
learned through our GMP Initiative.
The rapid and broad movement of
information and knowledge sharing is
necessary for
process optimization between
organizations, within
any organization itself, so we have to
move from
technology transfer to knowledge
transfer.
But just toward the stage,
reliability is
a phrase that we often don't use in
pharmaceuticals, but reliability has a
very
108
distinct body of information, body of
knowledge
associated with that.
So, if you look at this figure,
you have
performance, you have life, shelf life,
and you
have a desired specification on both
sides, on the
performance.
The first, Figure A is good
performance,
but poor reliability because the
performance
changes significantly over time, and the
variability of the spec changes, too.
The second one is good
performance and
good reliability over the life.
The third is poor performance
below spec.
So, I think the key is when we
think about
performance, we are thinking about
performance of
the shelf life, the bioavailability, and
so forth,
remaining unchanged throughout the shelf
life.
Just to keep that in mind.
In the current state, today,
chemistry,
manufacturing, controls, design
information
available in applications is limited and
varied.
Our reviewers have a high degree of
uncertainty
109
with respect to what is critical and what
sort of
process controls are necessary.
Our reviewers have significant
doubt on
the level of process validation and
process
understanding. So, they have no option but to
focus on in-process and product
testing. So, in
the pharmaceutical manufacturing from an
engineering sense, testing is control,
but in an
engineering sense, control is very
different. It
is a dynamic method. We don't do that.
Risk coverage post-approval is
a
challenge, and supplements are a means
for risk
mitigation. That is the way we have approached it.
Traditional use of market
standards--these
are the pharmacopeial standards--as
release tests
are not effective for process
understanding and
continuous improvement. In fact, by definition, if
you have attribute data or so-called zero
tolerance, continuous improvement is
impossible by
definition. That is the definition in QS 9000,
because we can only have continuous
improvement
when the product is already in spec.
110
If you have zero tolerance
criteria, by
definition, the product is not in
spec. So, that
is the fundamental thing. Also, we understood and
wrote about that in our Manufacturing
Science white
paper.
We have variable test methods
for physical
characteristics, less than optimal
systems
perspective and approach, low efficiency
and high
cost of drug development and
manufacturing, and
continuous improvement is difficult, I
would dare
to say not possible.
So, the success of the cGMP
Initiative was
to get a consensus desired state
statement, so I am
not using the exact words that we
developed. They
are modified and the desired state
statements
adopted by ICH, these are the ones.
Product quality
and performance are achieved and assured
by design
of effective and efficient manufacturing
processes.
Since we are looking back from
the
manufacturing side, manufacturing goals
are kept
first.
Product specifications based on
111
mechanistic understanding of how
formulation and
process factors impact product
performance, and an
ability to effect continuous improvement
and
continuous "real time"
assurance of quality.
Now, let's start looking at
this in the
sense what are the gaps and how do you
fill those
gaps.
Information and knowledge for
regulatory
assessment and decision process based on
the
desired state is information related to
quality and
performance and how the design impacts
that. So,
we need to know impact of formulation and
process
factors on performance.
We need information to judge
and develop
specifications based on mechanistic
understanding.
We need information to evaluate and
facilitate
continuous improvement, and continuous
"real time"
assurance of quality.
The focus is on design, and if
you are a
formulator, especially one trained a few
years ago,
or if you have been in the design
business, this is
simply logical extension, so this is
nothing new
112
about this, but if you are not, then, you
have to
think differently the design process.
Design is about doing things
consciously,
and not because they have always been
done in a
certain way. It is about comparing alternatives to
select the best possible solution. It is about
exploring and experimenting in a
structured way.
So, that is what design vocabulary brings
to us.
So, in the context of drug
product
development, design is about doing things
consciously, so you start with the
intended use.
That is the fundamental issue. You cannot forget
the clinical use of the product that you
are
designing. That includes route of administration,
patient population, and all other things
that
impact on the intended use.
That intended use defines for
you what the
product design should be. You have options to
select, and you select a product
design. That
design leads you to design
specifications, and
those design specifications define the
manufacturing process and its control
necessary to
113
develop those design specifications back
to deliver
the intended use.
So, you have product
performance, design
specification that reliably and
consistently
deliver the therapeutic objective, and
you have
manufacturing capability, ability to
reliably and
consistently deliver the target for a
design. This
is straightforward, logical, no rocket
science, and
we have been making and doing this for
100 years.
So, that was the basis that we
said we
will develop the ICH Q8, and ICH Q8
document, which
will go to Step 2 in Yokohama, I am
confident about
that, will essentially bring this type of
information.
It will not deal with the drug
substance
manufacturing part of it, but it will
start with
drug substance characterization.
So, it will bring in
characterization of
components of drug product. It will bring in
aspects of manufacturing compatibility,
and so
forth.
Much of this information is sort of missing
or varied in the current submission, and
we are
114
hoping, although the sections in CTD-Q
(P2) are not
ideal, we have to live with that because
that is
how everything goes in green, we felt
that the
sections provide enough room for bringing
all the
information to bear on that.
So, we have made significant progress,
and
I think the draft 4 we are working on has
captured
most of this.
What is the importance of
design thinking?
Design thinking makes the user paramount,
ensuring
that services we end up will do the job they
are
supposed to, as well as delighting the
customer.
Design thinking and methods
provide new
routes to better public services that
meet people's
needs and deliver value for money. That is the
key.
We have been making tablets for 100
years
or more.
It is a design problem. We
essentially
have not used the vocabulary, we haven't
brought
that in.
Tools, such as pre-formulation
characterization, and so forth, literally
have
become [inaudible], but that information
often is
115
missing in our assessment.
So, if I distinguish between
conventional
and novel design for the sake of
distinction in
terms of how we use prior knowledge, the
key aspect
of this design and quality relationship
is utility
of prior knowledge. For similar drug products, you
have probably more prior knowledge, and
for novel
designs, you have to rely more on the
experiments
you generate, but prior knowledge is the
key.
If you are going to come with a
new tablet
formulation, and you have 300 similar
tablet
formulations on the market, how much more
information do we need? If you leverage the prior
knowledge correctly and characterize your
drug
substance in a way to say all right, this
is the
way it is, you can leverage that
knowledge.
Level of mechanistic understanding will
depend, will vary. Pre-formulation programs, many
good pre-formulation programs get to the
mechanism
of degradation, get to the mechanism of
absorption
including Bioform Classification System,
characterization, that is the
fundamental. That
116
defines literally every aspect of the
manufacturing
process and other things.
So, if you have that
information, if you
will not be mechanistic completely, but
you have
valuable information, that moves forward.
The challenge I think is to
think about
design during drug development. As you develop
your characterization and your
development program,
you have to keep in mind the ability to
reliably
predict performance, confirm as you
progress.
Every experiment you do next, say, a
scale-up, is
adding to your knowledge base, is a means
to
evaluate the predictability of your prior
knowledge, and so forth.
So, if you think about
designing the
entire development project from a design
prospective, and capturing your
predictability, you
actually have an opportunity to move forward
very
quickly in terms of regulatory aspects,
as well.
So, level of understanding
increases over
time, and I think we have to recognize
that.
Structured empirical approach is often
necessary
117
because you often are not mechanistic.
Use of prior knowledge to
identify and
select a design space for
characterization is
fundamental, and I think Ken Morris
mentioned this
yesterday. People often jump into design of
experiments without knowing what design
space they
want to explore.
If you miss the prior
knowledge, you
actually increase your workload, you
increase your
cost by not being intelligent enough to
say what
are the critical variables upfront, and
sort of
exploring the design space. You cannot approach
this in a blinded fashion.
For example, now, if you have
multiple
number of variables that you have to
study,
obviously, you cannot study all of
them. That is
where risk comes in. Prior knowledge and risk
assessment is the way to address that,
for example,
failure mode effect analysis would be a
means to
say all right, these are the critical
variables, at
least these are potential critical
variables, these
are the ones we will select and move
forward.
118
So, initial conditions for
screening
experiments and then experimental
conditions are
then dependent on this prior knowledge
and risk
assessment.
Impact of formulation and
process factors
on performance, why can't we leverage and
be more
intelligent about our clinical trial
material
itself, and how do we design clinical
trials,
because that is where the connection
between
quality and clinical comes together, and
I will
show you an example as we go.
Similarly, with shelf
life. If you are
getting mechanistic understanding, and so
forth,
prior knowledge and shelf life, I think
is a
wonderful opportunity which we don't
utilize today.
Just to give you an example,
these are
standard procedures in industry. Here, is a work
from Amgen in a sense how do they address
the large
number of variables as they go through
process
characterization, pre-characterization
experiments,
is to bring the prior knowledge to bear
on this.
So, process characterization
studies start
119
with pre-characterization work, screening
experiments, interactions, and
combinations of key
parameters leading to process redundancy.
They sort of covered that with
a formal
risk analysis. So, these are standard procedures,
and in many, many aspects, the
formulation
development process has built in robust
approaches,
but it is not formalized, it is not well
understood.
What is a robust design? A robust design
is not removing the source of
variability, but
designing a process or product to reduce
the
variability.
A very simple example that in
pharmaceuticals we have, is we know
magnesium
stearate is a wonderful lubricant, but it
has a
drawback of affecting dissolution, we
know that.
Half of the formulations that
we have in
our submissions actually have a robust
design built
in.
They will use a smaller model on sodium lauryl
sulfate.
That negates the negative effect of
magnesium stearate. We have known that as
120
pharmacists, formulators, and so forth, a
long
period of time, but we never captured
that as a
knowledge base.
If you are making a tablet, you
are
compacting. Compaction has an effect on
dissolution. If you have right amount of
disintegrating agent, you remove the
effect of
compaction force. It's as simple as that. That is
what a robust design principle brings to
bear on
that.
What is troubling often is, if
you look at
the SUPAC guidance, and if you look at
the way we
have regulated, the way we have done
experiments
often is to define our input or
independent
variables in terms of equipment. Say, for example,
if you look at the SUPAC guidance, we say
equipment
of same design and operating principles,
you can do
this, and so forth.
That is not a
quantifiable. It is an
identifier. So, we know that performance
of a unit
operation depends on material
characteristics,
particle attributes, equipment design, and
121
operating conditions.
Instead of sort of defining of
input as
equipment A, equipment B, and equipment
C, and then
doing a design experiment, if you are
smarter, you
will say all right, what are the forces
acting on
the particle irrespective of the
equipment design.
That removes that and improves your
ability to
generalize. So these principles have been there
for 60 years.
Let me explain, in a sense, I
think the
key aspect here is risk-based
specifications.
Here, is an ICH Q6A decision tree. Let me walk you
through this.
What specific test conditions
and
acceptance criteria are appropriate for a
conventional or immediate release dosage
form?
Now, Professor Nozer corrected
me before,
so I will correct myself again. He said this is
not a decision tree, this is an event
tree.
Question 1 is: Dissolution significantly
affects bioavailability? That is the Question 1.
If the answer is yes, develop
test
122
conditions and acceptance criteria to
distinguish
between unacceptable bioavailability.
But if the answer is no, you go
down. Do
changes in formulation or manufacturing
variables
affect dissolution?
If the answer is no, go
down. Adopt
appropriate test conditions and
acceptance criteria
without regard to discriminating power,
to pass
clinically acceptable batches.
The first question is how do
you know
dissolution significantly affects
viability. Most
NDAs, not all, have a simple test that
they do. It
is called a "Related bioavailability
study." They
will compare a solution with a solid
dosage form,
and often you see they are superimposable. That
means dissolution is not rate
limiting. So,
dissolution is not likely to affect that.
Do changes in formulation
variables affect
dissolution? Yes, all of them do, most of them do.
If the answer is no, for
heaven's sake, if
you can find a formulation that doesn't
have that,
but you still put a dissolution
test. The question
123
should be why, why do you need that
dissolution
test?
So, some of the questions, how,
why, and
what really have not been addressed
adequately, and
our dissolution specification setting is
one, two,
three, these were your three batches,
this is your
specification. Often, it is limited to that.
I want to remind you that
variability is
inherent, and I did include a paper in
your packet.
This was published recently from
Cambridge
University and Pfizer.
It said if you don't account
for
variability and you assume that meeting
specification means you are
bioequivalent, that may
not always be true. In fact, if you do this
analysis, if your specifications are not
set right,
you have a 50-50 chance whether you are
bioequivalent or not, or whether you meet
specification or not.
So specifications for
dissolution are not
likely to be the ones ensuring
bioequivalence. It
is the entire control process that does
that, but
124
we focus so much attention on just one
test, we
miss the whole point.
Let me come back to this
decision tree. I
think in a quality by design thinking,
this is what
are my questions. So, dissolution significantly
affect bioavailability is a product
design issue.
You start with your pre-formulation, your
biopharm
classification, the solubility,
permeability, and
all that aspect, and you have an
anticipation
whether it will be affected or not, so
when you do
your related bioavailability study, you
are
conforming your past prior knowledge.
So, postulate-confirmed based
on mechanism
or empirically, and that can apply to the
question
dissolution significantly affect
bioavailability or
do changes in formulation affect
dissolution or
not.
But Jurgen points out, the key
question is
we have a mind-set 80 to 125, and that is
the magic
number.
Where did that magic number come from?
I
think this is where the clinical
relevance comes
in, what is a relevant acceptance
criteria to judge
125
whether an acceptable bioavailability is
there or
not, and that is a clinical pharmacology
question.
That is where we link to the clinic.
If you have a good PK/PD
assessment, and
so forth, you have far more information
available
to make a more rational judgment, and
that is the
question there.
So, design of manufacturing and
controls,
and the question is how reliable those
are, do
changes in formulation and manufacturing
variables
affect dissolution? If the answer is yes,
Are these
changes controlled by another procedure
and
acceptance criteria?
If the answer is yes, we come back
and put
a dissolution test. My question is why? If the
dissolution test itself is variable, and
so forth,
why would you want to put another
test? You have a
series of tests, and so forth, your
chances of
failure keeps increasing.
So, the questions that we need
to ask are:
How good or how reliable are your design
and
controls that you have put in place for
particle
126
size, morphic form, and so forth, to
address these
conditions?
So, overall risk-based CMC
would ask why
for these questions, but also, so
what? If
dissolution is not rate-limiting, the
question
should be so what, why do we need a
dissolution
test, and so forth.
So, this is how it all sort of
comes out.
So, quality by design thinking brings an
overall
CMC systems approach, for example, link
to morphic
form, particle size, stability failure
mechanisms,
and so forth, to address this in a
systematic way.
Continuous improvement is not
possible
today, because any movement is a
change. This is a
direct cut-and-paste from our SUPAC guidance.
Level 1 change, definition of change is
this
category includes process changes
including changes
such as mixing times and operating speeds
within
application/validation ranges.
If you need to have validated
those
ranges, any movement within that is a
change today.
So, it requires to be reported. If you change
127
outside those ranges, you not only have
to report
it, but then you have to do a Case B
dissolution,
which is a profile comparison, and the
supplement,
and the stability, and so forth, so
today, it is
not possible.
Our law and our regulations
provide
provisions for those approaches, and this
is a
Section 506A of the Act and 314.70 that
we issued.
We are required to make decisions based
on
potential to have an adverse effect on
identity,
strength, quality, purity, or potency of
the drug
product.
We have used the phrases
"substantial,"
"moderate," and
"minimal." They are not very
useful, they are not probabilistic, and I
think
that is where we have to work at.
But also, if you look at CFR
314.70, there
is a provision no change means no
reporting beyond
the variations already provided in the
application,
and that is where the design space comes
in.
So, what is this design space? The design
space is simply a space of knowledge or
information
128
where you know you will not affect your
bioavailability, you will not affect your
stability, and you will be in
specification, but
you are improving the manufacturing
efficiency, you
are improving the manufacturing process
through new
equipment, better controls through
process
adjustment in response to incoming input
variables,
and so forth.
So, that is what continuous
improvement
is, and Box defined this years ago as
evolutionary
operations.
So, that is how ICH Q8
information that
brings reliability to your deliver design
information, ICH Q9, which will develop
the failure
mode effect analysis and risk
communication, too,
all of them come together to define a
design space
for continuous improvement, and that
design space
will depend on the company's information
that sort
of comes about.
You will know which area is the
change,
which area is not a change, and that is
the map of
Maryland, a weather map, so you shouldn't
be in the
129
red area.
That's about it.
Yesterday, Steve showed this
slide that I
had developed for thinking about the
entire system,
how do you connect the dots. I am not going to get
into that, but I think the key aspect
there is the
knowledge space, and the knowledge space
in
relation to the clinical knowledge space
and in
relation to the manufacturing knowledge space
all
have to come together to sort of address
this.
A personal learning that I had
going
through the GMP process is a better
appreciation
for quality system. I am still an academic at
heart, and when you put me into a
documentation
mode, I get nervous, and great mounds of
paper is
something I want to avoid.
The quality system that we have
worked out
in the GMP Initiative is actually quite
nice and
simple.
It says say what you do, do what you say,
prove it, and improve it. Those are the
fundamental
principles.
So, say what you do to FDA, is
your
pharmaceutical development information
that you
130
share with us? If you say this is all I know, so
that is what you are going to get. If you say this
is how much I know, and so forth, you get
benefits
from that, but then you have to do what
you say
consistently. You have to prove it, and if you are
unable to prove it, you have to ask why,
and you
have corrective actions.
If you are unable to ask why,
unable to
answer why, then, there is a risk profile
that
increases. And prove it is more optional, there is
continuous improvement in innovation sort
of comes
in there.
The challenges, I think in
pharmacy, in
pharmaceutical education, we have been
doing this
all along. What has been missing is a
formal
structure and communication tool.
I draw some similarity
here. If I look at
what has happened in chemical
engineering, and now
I think chemical engineering is going
through
soul-searching activities to redefine
themselves,
but this is how chemical engineering
evolved.
It started with industrial
chemistry, unit
131
operations, material and energy balance,
chemical
engineering thermodynamics and control,
applied
kinetics, process design, transport
phenomena,
process dynamics, process
engineering. Now, they
are in molecular transformation,
multi-scale
analysis, and systems view.
So, they went from industrial
chemistry to
unit operations, to chemical engineering
science,
to system engineering.
Industrial pharmacy is still
industrial
pharmacy in the U.S., not as well in Japan,
China,
Europe, it's a pharmaceutical engineering
degree
literally. So, we are still in that, and I think
we can catch up on that, going to bring
some of
those principles.
It is important to do that, it is
important to bring a systems engineering
perspective because not only we have to
deal with
the traditional goals of quality, the GMP
Initiative offered new, non-traditional
goals, that
is, risk-based, flexibility, robustness,
scalability, continuous improvement,
innovation,
132
and efficiency. These are typically
non-traditional goals.
The characteristics of these goals are
complexity and uncertainty associated
with that. The
relationship between goals and
characteristics
that we are seeking is knowledge and
information
centric relationships.
There are fundamental issues there,
because if you don't get to this, our
quality
system will continue to be a paper chase
exercise,
and not really get to the heart of it,
because we
don't want to be lurching from fact to
fact, from
one quality system to another. Unless this process
is in the same sciences there, this will
not
happen.
I will skip that and focus on
where we
are.
My assessment is this. This is
not rocket
science, this is straightforward and simple
for
those who have been in this area for
quite some
time.
For those who are not, there is a need for
education training.
There are signs that I
see. The phone
133
call this morning from a major company,
and, in
fact, I should have asked that I can
share the name
or not.
Their positive experience with the
development report already in a
four-cycle review
is a good example that our folks can
manage this
process well, but consistency and making
sure it
happens consistently is a challenge.
So, the immediate education
need that I
see going through the PAT training, and
so forth.
Now, for a broader training is
introduction to
statistical quality control. That is fundamental.
We are missing that, and I emphasize it
is not
biostatistics. There is a distinction between
statistical quality control and
biostatistics,
hypothesis testing, and where we keep
missing that.
I meet with the PhRMA
Statistics Group,
and so forth. It comes back we are missing the
quality dimension here. We have to understand
variability. We have to focus and put training
programs on molecular pharmaceutics and
biopharmaceutics.
We have gone to the molecular
level in
134
most of those areas. Engineering principles is a
key aspect. Risk assessment and communication
would be a program. All of this will come together
quite nicely with the ICH Q8/Q9 training
program
itself, but I think we would like to add some
additional training.
I know Ken has been working
with us quite
constantly on focusing on what the right
questions
are for the review process, but I think
we can put
a
more formal training program on all of these
aspects.
I would like to say systems
approach and
thinking is important. Unfortunately, most of the
training programs that we go through, our
BS/MS/BA
programs actually takes us away from
systems
thinking to focus as narrowly as
possible, and so
forth, but in an applied area in the
regulatory
setting like this, systems thinking is
important.
Unfortunately, I go and talk
about Deming,
many people in the industry have never
heard of the
name Deming. I think we need to introduce
people to
Deming and others.
135
Team building and communication
will be
the key.
I will end my talk saying that
coming
together is a beginning, keeping together
is
progress, working together is success.
The GMP Initiative brought us
together. I
think the PAT Initiative took us further,
and a
smaller group is actually making
progress.
I am fairly positive. I went through a
quite depressive cycle in some of the
challenges,
and so forth, but I am fairly positive
that I think
we are on the right track, and we will
achieve this
rather quickly.
Policy Gap Analysis
MR. CLARK: I am going to deliver a talk
about something of the policy gap, but
what I will
really be talking about is a guidance
development
process and some changes that we have
done there.
My talk will be quite pedestrian and
quite short,
which I hope to be some relief.
I noticed in the agenda, at
10:30 we were
supposed to break, but now it's after
11:00, and
136
were this agenda an application, we would
be found
in violation of an agreement, and if we
showed a
pattern of being late, well, we might
just be under
a consent decree. So, I think we are at some risk,
so I will move us along and try to get us
back on
the path of righteousness, and such.
I want to point out that
somebody
mentioned earlier today about failure,
about
failure data, I am sorry I didn't quite
catalog who
it was that brought it up, but the
failure of data
to point stayed in my mind.
One of the things that we will
be talking
about in Yokohama in Q8 is what role that
plays in
an application, and to help define a
design space,
is there a place for us to use that in an
evaluation of an application, and if you
have
determined where your system fails, can
that offer
you some relief as to where you
operate. I wanted
to just bring that point up as I start in
this
speech.
In the GMPs for the 21st
Century, some CMC
guidance documents are out of
synchronization with
137
that rollout that you all have seen by
now, but the
guidance process that developed these
documents has
strong and weak points, and one of the
main
strengths of this is the technical input
from our
staff, from our review staff.
One of our weaknesses is in the
decisionmaking process for actually
moving the
documents from step to step and getting
them out,
which causes it to be very slow.
I would like to really dwell on
the
strength.
One of the things that we need to take
away from our previous guidance
development process
is that these deliberative processes are
well
meaning, and these people are highly
trained, and
they are experts at what they do. At every step,
they are trying to articulate the things
that are
on their minds and how to get
applications approved
in
the best way.
They may have become
proscriptive and
prescriptive, but that is not a failure
in their
attempts to articulate the best way to
get an
application approved.
138
We believe that there may be a
better way
to articulate that point, and obviously,
with the
rollout, and you compare the rollout to
the
documents that we have on our guidance
page, there
is
the gap, and most people know that, that are
familiar with the two sets of documents,
so I will
move on from there.
The draft cycling that was the
weak point,
I will point out this is the old draft
cycling, and
you
will see that there was a CMCCC working group
assigned from a CMCCC committee
body. That CMCCC
would define a group to work, and we will
call that
the body for now, to develop a document.
They would go ahead and develop
a
document, and this might take six months,
and it
might take two years, and it might take
five years,
but they would develop an articulation of
the areas
of interest for that document.
They would then proceed to take
that and
go through each review team, through some
kind of a
hierarchical structure in the
organization. Those
review teams would then have comments,
not unlike
139
the public comment system, and those
comments would
go back to that review group, and they
would
redraft the document, which might take
another year
or two.
Because these people are not
dedicated to
that task, they are also reviewing drugs,
they are
also involved in a lot of other efforts
like ACPS,
and they are also involved in guidance
development.
So, they go back to the review
teams, goes
back to that CMCCC body, and then it goes
back up
to the working group or back up to the
committee
for review, and then from the committee, it goes
to an OPS editor.
Now, that process, those steps
might take
as much as six months, it might be a
year. The OPS
editors then have a go at making sure
that the
legal language is current with the
desires of our
legal staff, and they might pass it on to
the legal
edit if their suggestions are minimal, and
so on.
If not, if the suggestions are
strong and
get into the body of the document to a
large
degree, it might actually go right back
up to the
140
body and have to go all through all that
stuff I
just mentioned all over again, and were I
mean-spirited, I could go through it a
second time,
but I won't.
Well, you go to the legal edit,
then, it
goes out to public comment. Then, you have public
comments dockets come back. You might have 1,000
comments if you are lucky. It might be a
couple
inches thick if you are lucky, and it
might be a
foot high if you are not so lucky.
If that happens, well, it will happen,
then, you have to catalog all those
comments,
address each of them somehow, address
them by
groups or individually, and then if you
have to
make substantial changes to the document in
order
to address those comments, go back up to
the top,
and if I were extra mean-spirited, we
could go
through this whole thing again.
I think you can understand that
that could
be a laborious process, and that is my
excuse for
why guidances take so long to get out of
the
Agency.
141
When somebody says that a
guidance will be
available soon, they may think it is
going to be
available soon, because they think they
are near
the end of the process, or what they
don't maybe
not understand is that there is an
iteration that
they hadn't predicted. So, that is something we
have been dealing with over the years.
This is a slide that puts into
words some
of what we just discussed, but it also
points out
that there is a rapid change in FDA
thinking over
the last couple of years. It leaves
slower efforts
to catch up. The guidance development is, in fact,
a slower process.
There is an investment of time,
and the
documents may be slow, but they have a
lot of
momentum.
You have that many people working that
hard, that many smart people working that
hard over
that many years, and the document gets
momentum.
The guidance content and the
new direction
are managed actually by two different
groups. We
have this OPS policy direction setting
group, which
had some involvement from the reviewers
and staff,
142
but the guidance content was being
managed by the
review staff directly with reciprocal
input from
the higher end, the OPS level.
We have taken some actions to
try to
remedy the situation. We have moved coordinating
and decisionmaking from the CMCCC working
group to
an OPS office level group, and have
created the OPS
CC.
Helen mentioned that a little earlier.
They
have a lot of tasks, one of which will be
to help
us coordinate these guidance documents.
It will move guidance content
management
up
to OPS with some lookback down to the review
level when we need that expertise that is
at the
review level.
We disbanded the CMCCC as such,
as words.
Of course, we have the same people
wearing hats
with new letters on them, OPS CC, but it
has a
different function, and I will show you
that in the
next slide.
The mandate through OPS CC is
to recruit
technical input from scientists when we
need it.
The new process looks something
like what
143
I have on the board now, where you have
OPS CC
actually managing the creation of the
documents,
getting science input from selected
teams, getting
input back to OPS CC. Then, you have a smoother
process on the high end of that document
formation
process.
So, we hope that that will shorten things
up a bit up there. It might take 10 or so years
off the process, which should improve
things.
The OPS edit and the legal edit
have never
really been hugely time-consuming except
that they
can cause reiterations, but it is that
iteration
that we have tried to smooth out. Of course, the
public comment isn't changing. We still
have the
dockets and dockets management.
We are trying to synchronize
the effort,
but there are a number of techniques we
intend to
use to synchronize the effort of
guidances with the
rollout of the two-year effort in the
GMPs for the
21st Century.
One is, of course, obviously, a
revision
of drafts. We have drafts out there. We have
public comments that are available for
many of
144
those drafts, bring those in, and we
might actually
put out for a second draft some of these
documents,
because we may incorporate public
comment, we may
incorporate some of the new thinking, and
if the
document is substantially changed, we may
not go
from draft to final. We may go from draft to
public draft to get more comment.
Another option is the withdrawal
of
documents. I would see that option mainly for
finalized guidance documents that just
have become
obsolete, and perhaps no longer do any
good for us,
so we may actually consider withdrawing
some of
them.
We have done this in the past, and we may
seek to do it in the future.
Another effort is enforcement
discretion.
We may actually use guidance to bring
into line
more of our practice where sometimes a
practice has
found its way into regulation, where the
regulation
will require you to do something that we
really
don't necessarily believe is constructive
at this
point anymore, and we may use guidance to
iterate
an enforcement discretion over some terms
to
145
address some of the regulation
requirements. That
would pretty much be viewed as a
precursor to
actually revising the regulation itself.
There is also a consideration of options
in OTG other than guidance. The question and
answer format where we have a simple
question and
we want to answer it with a paragraph and
hopefully, it doesn't turn into a
chapter, we post
it on public Internet, try to keep it
simple.
It would go through our
guidance
procedure, but hopefully, be much
quicker, and not
try to address a large range of things,
but just to
keep it to one topic, keep it simple.
We also want to look into a
Manual of
Policy Procedures, to expand that. I don't believe
that we have actually exploited our
Manuals of
Policy and Procedures enough to
articulate what we
would like our internal staff to be
doing, have it
in an if/then format, directed toward OPS
staff.
What is nice about the Manuals
of Policies
and Procedures is that they are publicly
available
once we are done, and they are rapid. They can be
146
rapid to get out, and we have more
flexibility in
what we say in them.
Back to guidances in
particular. We
really would like to have a more
risk-based
approach expressed in the guidance,
reverse recent
tendency toward proscription and
prescription,
which I think that if anyone has read
some of the
recent drafts, you can see there is a
difference
between what was published in 1987 and
what we have
been publishing in recent times, and that
they tend
to be much more detailed and much more
instructive
as to exactly what you should be doing.
We try to get up to a higher
level and try
to rely more on the science for the
specific issues
to drive the decisions that are made.
The manufacturer should choose
the
technology and the approach to problem
solving
where we should be evaluating more of the
science
behind whether or not the quality, the
ongoing
quality is short, and as Ajaz put up
there earlier,
what is the acceptably low probability of
not
meeting the clinical intent.
147
Focus efforts on assurance of
reliable
product quality, and not on technology.
Input into our processes is one
of the
things I would like to propose here, is
for this
body to seriously consider, creating a
fact-finding
group to help us with guidance processes.
Is there a potential gain from
formation
of this fact-finding group to help us
determine how
we move forward to the desired state?
Some of the questions that
fact-finding
group would be asking or asking and
answering would
be:
Do we need all the guidances that we have now?
Where are they incongruent? Provide advice as to
what guidance industry needs as we work
toward a
new regulatory paradigm, and what should
be the
prioritization of those documents if we
have any of
them at all.
So, I would like to propose
that this body
be engaged in that effort at some level
to be
deciding who would be on that body and,
of course,
it would be within the rules of the
advisory
committee system.
148
That is the end of my talk. Since I am
the last of the three, I guess I will
stay at the
podium and be the target for the
questions.
DR. KIBBE: It depends on how much time
the other two speakers will need and how
much time
we allow for questions. We can always come back to
the topic after the lunch break, but if
there are
some quick questions, we can handle,
Nozer?
DR. SINGPURWALLA: Not quick.
I have lots
of questions for Ajaz, and I am sure he
expects
those.
So, perhaps we should postpone them until
another avenue, or I can privately
communicate.
DR. KIBBE: No, we will get to them. I
want them on the record.
Go ahead.
DR. MEYER: Just one quick comment. I
like the idea of a question and answer
format in a
public Internet, frequently asked
question kind of
thing.
I think that would save the Agency some
time.
It will be easy for the sponsors.
You won't
have to deal with repeat questions, so I
think that
would be a step forward to have that.
149
MR. CLARK: Thank you, Committee, thank
you, Dr. Kibbe.
DR. KIBBE: Meryl, go ahead.
DR. KAROL: I have heard throughout this
morning about the risk-based approach
towards your
changes, and I wondered what do you have
in mind
for training and education of the
personnel in
order to make them knowledgeable about
risk-based
approaches?
MR. CLARK: The training of the personnel
is really an ongoing thing. It isn't really a here
it
starts and here it ends. We have the
people
involved.
We have a peer review system that Moheb
has set up where they come together after
they have
finished a review, and they present it to
their
colleagues.
We, from the Policy Groups, we go in and
we listen. We have comments. We steer them, we
try to steer them toward the things that
have been
done well, and to bring that out. A lot of the
practices that are involved in a review
of an
application are, in fact, risk-based
practices, and
150
sometimes a reviewer just doesn't bring
that up to
the surface and make that well known.
Part of our role in that particular venue
is to point that out, to say, well, you
took a
risk-based decision here, we would just
like you to
bring that up to the surface and make it
the theme
of the document as opposed to going down
into the
technology and making that the backbone.
So, it isn't really a big
stretch for
these people to grasp onto it. They are all very
highly trained.
We do have training programs in
mind. We
haven't really implemented a lot of them
upfront
yet, but the goal is to have a top-down
approach.
Dr. Morris has been involved at some
level at
helping us train managers into the
thinking and to
work that down through the management
levels.
We have gotten down to the team
leader
level at their last presentation, but the
rollout
did take up all our resources up until
just now.
So, we will be re-initiating that effort.
DR. KIBBE:
Jurgen.
151
DR. VENITZ: A few comments, one more
fundamental comment to Jon's presentation
that has
to do with my favorite word, and that is
risk. You
talk about how risk is being assessed,
and you have
emphasized primarily the probabilistic
component,
that you have tried to predict the
likelihood of
something happens.
Well, risk has a second
component, and
that is a judgment, how bad is what you
measure, or
how potentially bad can it be. I think one of the
things that you will face is within the
Agency, is
this culture of always assuming the
worst, anything
that happens we can assess whether that
happens
often or not is bad.
Well, risk really requires you
to make a
judgment as to how bad or how not as bad
it might
be.
So, I am just pointing that out, risk is not
just measuring probabilities, but it is
also
assigning utility values to those
probabilities.
Two small comments to Helen's
presentation. I would encourage, especially with
all the changes that are occurring, that
your staff
152
participate in all those industry
meetings,
particularly the pre-IND and the end of
Phase II
meetings, because I am pretty sure that
is where a
lot of things are being discussed that
are relevant
to the change in GMP and PAT, and what
have you.
As far as the briefings are
concerned, you
might consider something like a
question-based
review where not the entire material gets
reviewed,
but you are actually focusing in on
things like
source variability on things that are
potentially
at risk.
DR. KIBBE: Thank you, Jurgen. What we
really were hoping for is if you had a
quick
question, because I want to get the next
two things
out.
If you have got a long discussion, we will do
it right after lunch.
Let's go ahead and get Mr.
Ahmed to talk
about the FDA Critical Path Initiative, a
generic
industry perspective.
Generic Pharmaceutical Association
Perspective
MR. AHMED: Good morning, everybody.
Sorry for this delayed
presentation. I
153
know a lot of you must be looking for a
break, but
just bear with me. It's not that bad. It will be
fairly short because I am presenting from
a generic
perspective, and clearly, the Critical
Path
Initiative does not really look into the
generic
industry the way other presentations or
the
Internet on the FDA web site.
First of all, I would like to
thank Helen
and everybody at the Agency for providing
GPhA the
opportunity to present their view
regarding the
Critical Path Initiative.
Gordon actually asked me to do
this
presentation on behalf of GPhA. I am from American
Pharmaceutical Partners, which is a
direct
injectable manufacturer. We have a proprietary
drug under review right now at the FDA.
What I am going to talk about,
first, is
what is the overview of generic industry,
what part
the generic industry really plays in the
health
care system in the United States, the Critical Path
Initiative concept, the benefits of the
Critical
Path Initiative to both the generic
industry, as
154
well as the innovator companies.
I think one of the most
important
component of the U.S. health care system
is generic
industry, as you can see. Fifty-one
percent of the
prescriptions dispensed in the United
States were
generic drugs. This data is based on the 2003 IMS
data.
I am sure the utilization of drugs will
increase as time goes by.
About 8 percent of overall cost
of
prescription drugs is contributed by the
generics.
There are several off-patent products
which still
don't have any generic competition, and
that is
really hurting the consumer in a lot of
respects,
because these products have been off
patent for a
while, and I will elaborate on those as I
go on.
When we reviewed the Critical
Path
Initiative, we found that it really
provides
significantly improved tools for
evaluating
basically safety, efficacy, and
characterization of
the drug substance like the
product/manufacturing,
because I think this is a very bold
concept
proposed by the Agency, because over the
last
155
15-plus years I have been dealing with
the Agency,
this is really a breath of fresh air,
because I
still hear that titration is a better
method than
is HPLC, because it's in the USP, so this
is really
very good thinking on the part of the
Agency.
We have got to get out of that
mode,
because, on the one hand, we hear from
FDA good
science, good science, good science. We file an
application and provide a better method
to the
Agency.
There is now we like potential metric
titration because in the USP, your HPLC method
or
LCMS method is not good.
So, I think this is a good
shift. At
least from our standpoint, we think that
science
really holds a lot more promise and can
hoe a lot
more ground than regulations.
We like FDA's approach. We like the
collaborative approach. I think it is important
that we involve the academics, the
patient groups,
the industry, as well as the regulatory
bodies,
because really at the end of the day, all
of us
have the same function, which is to
protect the
156
American public and to provide a service,
because
the drugs used by folks is really for
people who
are unfortunate, because they have to use
those
drugs, so we have to make sure that we
provide the
highest quality product, which is the
most
efficacious and has the lowest side
effect.
The desired outcome meaning yes,
faster
approvals. Again, a little bit of a
misnomer
because just to give an example, we filed
for a
CB30 to extend expiration dating for a
very
critical product.
It took FDA 60 days to first
acknowledge
that we filed the CB30, and then it will
be another
six months before we make a decision, so
what is
the point in filing a CB30? That is what I really
wanted to bring to Helen. If we want the desired
outcome to be faster, let's please be
more
pragmatic, because if it is built around
the system
the way it is right now, the desired
objective is
really very far away.
Again, the primary emphasis is
on
discovery, and it really helps pharma,
and clearly,
157
we are a big proponent of that because
what pharma
produces eventually has the opportunity
to become a
generic, so we are imploring the Agency
to work
with pharma to enhance drug development
process and
give the pharma folks faster approvals.
From a generic industry
perspective, I
think the Critical Path Initiative is
also very
helpful, and I request Helen and Ajaz and
everybody
else to really think it hard because the
generic
industry represents a big opportunity for
the
American health care system, as well as
for the
Agency itself.
We would like to be part of this
initiative because there are lot of
opportunities
where we can minimize the pains we go
through on a
daily basis, as well as you folks.
Again, the goal here is to have
timely
submissions as well as timely approvals,
so the
consumer can benefit. As you know, the prices,
when the product becomes generic, almost
within the
three months, the price of the generic
drug is
almost 90 percent less than the innovator
product,
158
so very clearly it helps the American
public
significantly from a financial
standpoint.
Again, we very earnestly
request the
Agency to really look into providing
guidance to
MDIs.
As all of you know, it took seven years for
Albuterol to become generic. We don't
want to go
through that scenario again. There are still a lot
of products which there is no guidance
for.
The products, especially the
transdermal
products, which have no guidance, the
inhalation
products, there is still no guidance, and
also,
again, I don't want to take anything away
from
Ajaz's PAT Initiative, but in the entire
PAT
Initiative, there is no mention of
sterile
products, and currently, in the United
States,
about 20 percent of prescription drugs
are sterile
products, which includes parenterals,
which
includes inhalation products and topical
products.
So, I would really again urge
FDA to look
into how they can help the sterile
product
manufacturers in terms of providing
guidance and
reducing regulatory burden. Clearly, we are
159
looking for better collaboration with
FDA. We
would like to be partners with FDA in
every respect
we can, but again it's a two-way street.
Sometimes we get a lot of resistance
from
OGD in terms of pestering them too much
or in terms
of arguing on scientific issues, so we
really would
like to have a more collaborative effort.
As part of GPhA, I would really
urge FDA
to look into it in terms of what our
objective is,
to make the safest product, but
apparently there
are some scientific disagreements at
times, and we
want to minimize the pains we go through
in terms
of the review process.
Lastly, I would like to talk
about the
21st Century Pharmaceutical GMP
Initiative. I
think it's a step in the right direction,
but there
are a lot of issues we really need to
look very
hard at.
One of the issues is inspection.
There is so much inconsistency
in the
inspection process, and to give you an
example, an
inspector going to a sterile
manufacturing facility
in New Mexico really does not even cover
one-tenth
160
of what is covered in the Chicago
District.
So, like we are from the
Chicago District
and we get the highest scrutiny, I can
tell you
that for a fact. Susan Bradley basically really
keeps us on our toes, but there is so
much
inconsistency as I go from district to
district.
Please look into this because this really
hurts the
generic manufacturers especially because
of the
patent issue, and things like that.
The other think is I keep on
hearing the
Science-based Initiative, Science-based
Initiative,
but we are still getting questions from
the
reviewing chemists, which is really a
total waste
of time for us and you both, the reason
being that
we cannot change basic science, and
nobody can
change basic science.
But to give you an example, not
too long
ago I got a request from a reviewing
chemist asking
me to provide a chromatogram for mannitol
where the
quantitation was 229 nanometers, and I
picked up
the phone and called the reviewing
chemist and I
said, "Please, mannitol has no pi
bonds, they are
161
all sigma bonds," and I [inaudible]
off any sigma
bond which the electrons could be
transferred to
sigma star at 220 nanometers."
He said, "I still want to
look at the
chromatogram." So, please, please be open-minded,
please listen to the folks who really do
this for a
living.
They know the science. They would
not
present data which does not make sense.
So, that is part of the 21st
Century
Pharmaceutical GMP Initiative. Let's be
open-minded, let's discuss with the
industry,
because we folks, we put a lot of time
and effort
in terms of manufacturing batches,
testing batches,
and
we want to produce the highest quality product
and to have the minimum risk.
However, we need to be very
open-minded.
We need to depend on science rather than
regulation.
Thank you.
DR. KIBBE: Judy has a question.
DR. BOEHLERT: I have a comment. I am
going to help you out. You said you had a problem
162
with USP methods, titrations. Well, I have several
solutions for you.
First of all, USP would love to
get your
HPLC method. Second of all, USP allows the use of
alternate methods, so as long as your
HPLC method
is equivalent to a better titration, you
may use
it.
Now you have got the FDA to
deal with.
They are talking about risk, and they are
talking
about science. I would hope that the FDA would
look at a submission favorably if you
have
demonstrated that your HPLC method is equivalent to
the titration method and the compendium,
and you
can go forward.
I know you probably do get
questions along
those lines, but I think we need to move
past that
and
ask the right questions.
MR. AHMED: Judy, what happens is that
they say yes, HPLC method is better
assurance than
equivalence data, so what is the
point? If you are
showing equivalence data, then, what is
the point
of using HPLC method?
163
DR. BOEHLERT: Well, because you don't
have to do the titration, and you really
do get
better data with the HPLC method.
MR. AHMED: That is what my point is.
DR. BOEHLERT: Titration might be a great
method for releasing product, that you
might want
to use HPLC particularly during
development to
learn more about your product.
MR. AHMED: And like we are providing a
stability indicating method, and we are
telling the
reviewing chemist, please, titration
cannot predict
the stability indicating nature of the
product, and
still we are asked to provide comparative
data.
So, that is what my point is.
DR. BOEHLERT: I know that that happens,
and it happens to the pioneer industry,
as well as
the generic industry. We need to get past those
questions that really aren't meaningful and
deal
with really the scientific issues and
look at what
the risk is of some of these decisions
that are
made.
MR. AHMED: Exactly.
164
DR. BOEHLERT: So, no, I support what you
are saying.
MR. AHMED: Thank you, because we know FDA
has very limited resources. We don't want to waste
your time in terms of providing
meaningless data.
DR. KIBBE: Thank you.
MR. AHMED: Any other questions? Thank
you.
DR. KIBBE: Gerry, you have the last word.
Pharmaceutical Research and
Manufacturers
of America (PhRMA)
Perspective
DR. MIGLIACCIO: Three weeks ago, the FDA
released their final report on the 21st
Century
Initiative. This morning, Helen, Jon, and Ajaz
talked about some gaps to achieving what
was
defined in that final report.
That final report did a very
good job of
defining the desired state, and I am not
sure how
many of the committee members have gone
through the
entire report. I had 12 hours on a flight to
Tokyo, and I couldn't get through it all,
but it
does a very comprehensive job of painting
the
165
desired state. Yes, there are gaps in the FDA and
yes, there are gaps in industry.
In the time that I have, I am
not going to
give PhRMA perspectives on every element
of the
final report. What I will do is touch on
certain
elements of it.
I want to reinforce what Helen said
earlier about culture change, make a
couple of
general comments on the report, and then
talk about
three of the documents in the final
report, the
Quality Systems guidance, the risk-based
inspection
model, and the ONDC risk-based quality
assessment
system, and then finally, summarize
before lunch.
Three years of culture
change. It's a
two-year process, but we have been
working on this
for a lot longer than two years. In fact, I
contend that we have been working on this
for three
to four years, and the culture is
changing.
It is changing, it's gradual,
in some
cases it is dramatic and others, but
there has been
in this unprecedented period of
communication and
learning for both the industry and for
FDA, and
166
that has really been driven by the, first
of all,
having a shared vision of what the
desired state
is, maybe not always agreeing on how to
get to the
desired state, but at least having the
shared
vision of where we wanted to get to, and
having
open communications about that.
PAT was the model, and much has followed
based on the model.
We, in industry, are moving
from a fear of
data to a passion for process
understanding. Now,
let me describe that fear of data. I grew up in
this industry, I have been in it for 25
years, and
I remember the days in the laboratory
where I did
some testing that was not required by the
documents
in my NDA, and getting reprimanded for
doing any
testing that was not required and where I
did not
have a clear understanding of how I was
going to
deal with the data.
Well, we are getting much
better now, we
are using much more sophisticated tools
at
converting data into knowledge, and therefore,
the
fear of just data generation is going
away, and we
167
are also learning very rapidly that that
data, if
converted properly into process
understanding,
process knowledge, can tell us a lot about our
processes, can tell us where the
variability is,
and can help us remove that variability
where it is
excessive.
The next point, science, not
blind
compliance, is winning the day more
often. I am
seeing this both internally and I am
seeing it
during inspections by FDA, that there is
much more
discussion. There is much more discussion about
the science, and I certainly, from my
perspective
in the industry, and I see my colleagues,
as well,
in other companies, they are using the
science to
make decisions much more frequently than
simply
sitting back and saying I am not going to
take a
risk, blind compliance, this is what the
Agency
expects, and just ignore the
science. So, that is
going away.
I think you are finding in many
companies,
innovation is accelerating, and a lot of
it has to
do with the open dialogue and the
encouragement of
168
the Agency obviously in this area.
Interactions during inspections
are
changing, and I think some companies have
seen
dramatic changes, others subtle changes,
but Helen
talked I guess yesterday about the
dispute
resolution process, and I think that the
discussions leading up to the guidance,
the draft
guidance, have had a significant impact
on the way
both our firms and the FDA investigators
approach
an investigation or an inspection.
We are far more willing to put
much more
scientific knowledge on the table during
an
inspection. Why?
Because basically, the dispute
resolution process says that if you are
going to
dispute a scientific issue, you have to
use what
you have presented during the inspection,
so we
want to make sure we get that information
out.
Most of used to sit there and
say, all
right, we will wait to see if the
investigator puts
it on the 483, and if they put it on the
483, then,
we will provide a written response to
dispute it.
Well, now, we are dealing with
it during
169
the inspection, and now the investigators
are
taking the time to look at the science,
and they
are calling in outside help when they
need it to
review the science, and we are seeing
more and more
issues resolved during the inspection, or
shortly
after the inspection with the district.
So, I think the dialogue about
dispute
resolution has taken us 90 percent of the
way to
dealing with scientific disputes.
So, some general comments on
the final
report.
I think this is a Churchill quote.
"This
is not the end, it is the
beginning." The
infrastructure is in place, but we have a
lot to
do.
We applaud the magnitude of
what has been
accomplished, and not so much the volume
of work
that was accomplished by FDA with
assistance from
industry and academia, but the rigor.
If you look at these documents carefully,
it is not individuals sitting around and
putting
their own thoughts on paper, but bringing
in
experts from various fields, including
risk
170
management, and ensuring that the
documents had the
technical content that was required to
achieve the
objective.
We obviously, in both industry
and FDA,
need to continue to work together to make
sure that
what is in those documents is fully
understood
throughout the Agency and throughout
industry, and
modified as appropriate, and then
implemented. So,
we have a lot more to accomplish.
Let me now turn to three of the
documents
that were in there.
First of all, the Quality
Systems
guidance, and this is the Quality Systems
guidance
for industry. We were very pleased to see the
Quality Systems document for use within
FDA. We
think that is going to contribute to
their
operation significantly, but the guidance
for
industry is very comprehensive, and we
think it is
of great utility.
I know many companies are
already using it
to benchmark themselves, to look at our
own Quality
Systems and say how do we match up
against this
171
document.
Now, there are some key issues
in the
document, and PhRMA will be commenting on
this
document and many others. I am just trying to
point out some high-level issues, but
certainly
process understanding, what level of
process
understanding leads to flexible
continuous
improvement in the regulatory
environment? It's a
key issue and I think both Helen and Ajaz
pointed
that out this morning.
We believe that it is
imperative that
these concepts be addressed on the global
level.
Somebody pointed out earlier today, most
of us are
in global businesses, the FDA guidance on
Quality
Systems, we believe really lays that
starting point
for ICH discussions on the proposed
Q10. So, we
really believe that this needs to go on a
global
basis and can contribute globally.
There are some parts of the
guidance that
we need to ensure we look at, potentially
revise.
We need to move away from some of the
current
compliance systems, and get us into a
Quality
172
Systems mode. So, I will just give you one
example.
The guidance calls for the
trending of
data, encourages the trending of data as being
part
of a good Quality System structure. The problem is
some of the data we generate for
compliance
purposes provides no value on a trending
exercise.
How do you trend "none
detected" or below
level of quantitation? How do you trend data that
is rounded down to a point where it no
longer has
any meaningful benefit to gaining process
understanding? And why trend data that is
generated that has no relationship to
critical
quality attributes or critical process
parameters?
Our current specification
system drives
the collection of that data. So, we talked about
specifications earlier. I think this workshop in
March is going to help us move these
discussions
forward, but our current specification
system has
to evolve. It has to evolve so that we are
collecting meaningful data on things that
are
critical to quality.
173
Now, let me move to the
risk-based
inspection model, certainly a solid step
forward to
allow FDA to focus on higher risk
sites. We do
question some of the elements of the
algorithm, and
we think they require further discussion.
Again, as I raised at the
Manufacturing
Subcommittee meeting, and Judy talked
about
yesterday, the concept of volume is
problematic to
us,
because if you have a manufacturing facility
that makes two or three products at very
high
volume, actually, they understand those
processes
much more than a facility that may make
20 products
at lower volumes, and they are probably
much better
controlled. Therefore, the risk
associated with
high volume is probably, in many cases,
much lower.
So, the use of a strict volume
metric in
the algorithm is problematic, and we
would like to
discuss that.
But our most serious concern is
about the
transparency of the system. These are words from
the risk-based inspection model document,
and what
they say is that the FDA brought in
experts, and
174
the experts have established risk
weightings for
various processes, products, unit
operations I
assume, we don't have that information,
and that
has all been used to determine the risk
factors or
the risk level of various sites, but they
don't
intend to publish or disclose that
information.
Now, when we started this three
years ago,
we said one of the key objectives was
that we
understood where the risk was in this business, and
that both industry and FDA focused on the
high
risk.
Another document that was
issued in the
final report is this defining the
customer in a
regulatory agency, and we are, the
industry is, a
customer in certain case of the Agency,
especially
when it comes to their guidance documents
and
position papers.
If you look in the middle, when
FDA
develops guidance documents representing
the
Agency's current thinking on a particular
subject,
we provide clarity and understanding to
firms.
Well, there is no clarity and
175
understanding on what the FDA considers high
risk
in manufacturing coming out of this
document.
Also, in the final report, the FDA will
now be
using a Quality Systems approach to
improve the
predictability and consistency. Again, there is no
predictability here.
We believe that industry and
FDA have to
have a mutual understanding of what
processes, what
unit operations are considered higher
risk. We
believe experts in industry should have
the
opportunity to discuss and potentially
debate what
is considered higher risk by FDA, so that
both
industry and FDA can focus their
resources on what
we mutually agree are higher risk.
We think transparency is
essential here,
and I truly believe this is a
win-win-win, a win
for FDA, a win for industry, and a win
for the
patients if we all agree and focus on
what is high
risk, but if it's not going to be
published, if we
are
not going to understand whether the FDA
perceives certain facilities as high
risk, we are
going to do our own risk assessment, and
it may be
176
different. It may be different, and I don't think
that is going to be an effective use of
anyone's
resources.
So, we feel very strongly that
transparency here is essential, and we
hope FDA
will continue the dialogue with us on this
specific
model.
Finally, the ONDC risk-based
assessment,
quality assessment system, it represents
a very
profound change in the organization and
review
process.
You heard a lot about that both yesterday
and
today. We support the objective
strongly, but
a lot of work is required to achieve the
end state.
Now, on the PAT Initiative and
then on the
GMP or the Drug Quality Initiative, the
FDA
demonstrated a willingness to put the
leadership in
place and the resources in place to make
it happen,
and I think this two-year report is an
evidence of
that.
The question we have, a
question that will
we have the same commitment of leadership
and of,
in this case, review availability, to
deal with
177
this initiative, and as importantly, will
industry
have the same input into this process as
we had
into the PAT and the general Drug Quality
Initiative over the past two to three
years. So,
it is a key issue for us.
There is this potential that
this proposal
creates this expectation. I think, Judy, you said
something about this yesterday in your
summary of
the Subcommittee meeting, that there is
going to be
a significant increase in the knowledge
provided
upfront.
We will be providing different
information
and
more knowledge, yes, not necessarily more
information, it is going to be more
knowledge.
But again, there will be a
greater degree
of process understanding, but it still
will be
limited with the original NDA submission. It will
be sufficient to establish that we have a
robust
process, but process understanding is a
continuum
and will accelerate post-approval.
So, you know, you are going to
get the
design space in the NDA, but that design
space may
178
be limited. It will expand tremendously in the
first one to two years of commercial
manufacturing.
We believe FDA should work
through ICH to
establish a global partnership for the
risk-based
quality assessment system. Right now the
assessment systems in Japan, the EU, and
the U.S.
are very different, and we certainly
should not
exacerbate the differences by moving
forward with
this initiative without including our
global
partners.
One area which we understand
the
motivation to go to the pre- and
post-marketing
approach in the quality assessment system,
but it
does require some further evaluation.
As I said just a few minutes
ago, process
understanding significantly increases
during the
first one to two years of marketing, and
potential
for continuous improvement is really the
highest
during that period of time, and there is
certainly
a value in having the reviewer who
reviewed the
original NDA involved in those initial
continuous
improvement changes.
179
As I said, you have got a
limited design
space in the NDA, and that design space
is going to
grow exponentially in the first couple of
years.
Having to shift gears from one
reviewer to
another may be somewhat
counterproductive. So, we
think this requires further discussion as
to is
there an appropriate time frame where the
handoff
goes to postmarketing.
Helen talked this morning about
two
systems, potentially two review systems,
one where
you have the science process
understanding in the
application, one without. The problem is that
process understanding is not a yes or no
answer,
it's a continuum.
So, every application, Company
A may
believe in this concept of manufacturing
process
and process understanding, but Company
A's NDA for
product X and for product Y will have
different
levels of process understanding.
How will a reviewer determine
where the
application is in that continuum and how
it should
be addressed? Key questions we think need to be
180
addressed between FDA and industry as we
move this
quality assessment system forward.
So, to summarize at two minutes
before
noon, Mr. Chairman, the infrastructure is
in place,
but some modification is required. We are very
pleased with the two-year report. We think it is
extremely comprehensive and really helps
us focus
on that desired state.
The culture is changing, it
needs more
changing, but I think the steamroller is
moving and
I don't think it is going to stop either
in
industry or in FDA.
The desired state is on the
horizon, we
are starting to see it in little bits at
manufacturing sites and in research and
development
sites, you are starting to see the
desired state
coming to fruition, and I think it is not
as far
away as some might think.
Thank you.
DR. KIBBE: Thank you.
I see by the clock
on the wall that we are in plenty of time
to
actually get up and go to lunch. We have no
181
individuals who seek time during the open
forum, so
that we will return and at 1 o'clock we
can
entertain questions and comments about
all of the
speakers that we have heard prior to
lunch break,
and then we can jump right into the
information and
presentations in the second half of the
day.
I know that there are some
members of the
committee who need to get out of here by
3:30 or
4:00 and we are going to try out best to
make sure
that the bulk of the work is done before
they have
to leave, and I appreciate your attendance.
[Whereupon, at 12:00 noon, the
proceedings
were recessed, to be resumed at 1:00
p.m.]
182
A F T E R N O O N P R O C E E D I N G S
[1:00
p.m.]
DR. KIBBE: I see by the clock on the wall
that we should be turning on our
equipment, so we
can start recording the wonderful
brilliance that
is coming forth from these meetings.
We agreed before we went to
lunch that we
would spend some time responding to this
morning's
activities and give an opportunity for
those with
extensive comments and questions to
discuss the
issues, and I know, Nozer, you are ready.
Committee Discussion and
Recommendations
DR. SINGPURWALLA: This is to Ajaz. Would
you like to sit down while I ask the
questions, or
would you like to stand? You will stay here.
DR. KIBBE: It's safer, you are further
away over there.
DR. SINGPURWALLA: Anyway, I enjoyed your
presentation. The ideas again are visionary and
stimulating, and the comments I want to
make are
purely to improve upon what I think is
something
positive.
183
The first question I want to
ask you is a
particular viewgraph that you put up in which
you
said the future definitions of quality
should be
probabilistic in nature. This is a comment
attributed to Janet Woodcock, although I
am pretty
sure you may have had the thought.
What do you mean by that? I subscribe to
the view, but I would like to see what
you want to
say to it.
DR. HUSSAIN: I think, in part, that sort
of follows a continuum of our discussions
at FDA
Science Board, and that is the IPAP RS debate
also,
is we often establish acceptance
criteria, and so
forth, in a more deterministic type of
thinking in
the sense of we don't utilize variability
as a
means of decisionmaking, and our criteria
are 75 to
125, that's it, and no unit outside that
for a
given sample, so it's an outgrowth of
that thinking
to saying that at one level, to say let's
utilize
the statistical distribution variability
in making
decisions.
That is one level of that, but
184
probabilistic also is linked to the
risk-based
aspect, and risk is probability for harm
and
consequences of that, and I think to
bring that
level of thought into how we approach quality.
DR. SINGPURWALLA: I think this requires
much more thought and discussion, but I
was curious
as to what you had in mind.
On your viewgraph 7, you have
those
3-dimensional illustrations of critical
parameters,
reliability, and life, shelf life is what
you had
in mind.
I could not get the essence of what it is
that you were saying there, and the main
reason I
am raising this particular issue is next
time you
present it, you may want to rethink as to
what it
is that you want to communicate, and I
didn't get
the message.
DR. HUSSAIN: The message was in the sense
I think when we often think about shelf
life as a
point estimate or as a endpoint
determination
perspective, and so forth, and what I
liked about
that viewgraph was it brings in the
variability, it
brings in an understanding of variability
from that
185
context and how variability changes.
So, that was an illustration,
and so it
provides a means to think about
reliability and
shelf life together. So, that was the
attractiveness of that viewgraph.
DR. SINGPURWALLA: You mean the
reliability of a drug?
DR. HUSSAIN: Product.
DR. SINGPURWALLA: Of a product.
DR. HUSSAIN: Yes.
DR. SINGPURWALLA: Well, when we normally
talk about reliability of a product, it
is the
probability of it either living to some
life length
or doing it or performing satisfactorily
or not, so
you had a similar notion of reliability
in mind?
DR. HUSSAIN: Yes.
DR. SINGPURWALLA: Let me go to your
viewgraph No. 24. I want to introduce a new
concept based on that viewgraph. This is on page
12.
Perhaps I am going to introduce a new buzz
word, which you may find attractive in
the future.
Are you ready?
186
I think that buzz word may also
apply to
some of Gerry's presentation. You are talking
about substantial, moderate, and minimum
as not
being probabilistic. You said something to that
extent.
DR. HUSSAIN: We haven't defined the
probability aspects to that.
DR. SINGPURWALLA: I want to introduce you
to the following. You perhaps know that in our
presidential debates, President Bush used
the word
"fuzzy math," and I think now
Kerry is talking
about "fuzzy calculations." But there is a body of
knowledge called "fuzzy sets,"
and fuzzy sets are
those whose boundaries are not sharply
defined.
So, the idea of substantial,
moderate, and
minimum would contribute to what is
called a fuzzy
set, whose boundaries are not sharply
defined, and
I think Gerry said something, I forget
exactly
where, but he was also alluding to a
fuzzy set.
Now, what I am recommending to
the FDA and
to the drug industry and whoever else is
listening,
that the notion of fuzzy sets may be very
useful
187
here, particularly with your nasal sprays
thresholds, with the idea of a threshold.
You
cannot have a precise threshold. You cannot say
that if something is more than 4 inches,
it is
good, and if it is less than 4 inches, it
is bad.
The boundary is fuzzy.
There are ways by which you can
endow
probabilities on fuzzy sets, and I would
like to
alert you to that particular body of
knowledge, and
I would like to alert you to looking into
it as a
possible venue for the kind of problems
that you
are working on.
This is very recent, very
new. The idea
of fuzzy sets has been around. Control engineers
use it, but to make it precise, and to
endow
probabilities on it is something very,
very new,
and the last issue of the Journal of the
American
Statistical Association has a paper with
discussion
on that particular topic. I urge you to look at
it.
The last comment, and this is a
comment,
on your last slide, on immediate
education needs.
188
First, I want to say what else you need
into this
package.
You really don't need introduction to
statistical quality control. What you need is
introduction to statistical process
control,
because when you take quality control, it
goes back
to acceptance sampling, and you are
really
interested in a process.
You say you should understand
variability.
I think the point is if you understand
variability,
variability goes away. Once you understand it, you
know what to do with it, and therefore it
vanishes.
What you really need as far as
an
educational component is some training
and
appreciation of what do we mean by
uncertainty, and
how do you work with uncertainty. I think that is
an important component. Once that is understood,
all other issues, process control,
biostatistics,
non-biostatistics, all those become very
clear.
The last question is how do you
plan to
implement this immediate education needs,
what are
you going to do, hold workshops?
DR. HUSSAIN: No, I think we have
189
different mechanisms to sort of
training. We
actually have formal training programs
that
actually go on, on a regular basis at a
training--we used to have what I think we
called
staff college. There are aspects of that. There
is an Office of Training and
Communication, which
is thinking of a course on design of
experiments,
and a whole series of courses which are
sort of
built in.
People can take them whenever they have
time, so these are ongoing.
But immediately what we have
been focused
on is workshops, internal seminar series,
and so
forth.
We are starting in a step-by-step fashion.
As we move towards the final guidances,
for
example, we generally have a training
around the
guidance.
Much of that we also get captured there.
But my thoughts are I think we
need a
portfolio of courses that are available
every
quarter or every semester, or whatever,
that people
can take on a regular basis, so we need a
curriculum that is available for our
staff whenever
they choose to take it, and so forth.
190
DR. SINGPURWALLA: My reaction to these
things, available on a
semester-by-semester or
quarter-by-quarter is that they
fail. What you may
want to do is have special courses like
this
throughout the FDA, not just your
organization,
provided they agree that this is what is
needed,
and have them, you know, once, and maybe
twice, and
then stop, but then require that the
people who
should need that exposure go.
If you keep it optional,
somebody says,
well, I know statistics, I know what
r-square is, I
don't need to take a course. That is where the
tragedy is, that everybody thinks they
know it.
So, I propose that you design something
specific
and go through that.
Again, I am not volunteering
since I made
the suggestion.
DR. KIBBE: Do you have a suggestion for a
college that they could take all these
courses at?
DR. MORRIS: Someplace in Indiana maybe,
Notre Dame.
DR. KIBBE: I was hoping for Pennsylvania
191
myself.
DR. MORRIS: Pennsylvania, okay.
Actually, the University of Hawaii I
think would be
good.
I actually had a couple of
comments I
think that go across Helen's, Ajaz's, and
Jon's
talk, but I want to preface it by saying
that in
the time I have been spending with you
all, it
occurs to me that there is a lot that is
right,
right now, with the reviewers and the
reviewing
process that we don't want to lose. I don't get
paid any extra for that, by the way.
In that sense, the idea of a
two-tiered
approach doesn't really seem to have legs
to me.
In a sense what we are saying is, is that
the
reviewers would have to be somehow two
tiered, and
that is really not the way they review.
What we talked about at the
last
Manufacturing Subcommittee was the idea
that when
companies can provide a rationale with
supporting
data, that that automatically registers
as a level
of understanding as opposed to when the
reviewers
192
have to develop a rationale based on the
data that
they have available, but at the end of
the day,
they are still looking for rationales,
still
looking for flags in terms of safety and
efficacy
and rationales.
So, I don't really see that
there is a
real need for that. I mean we can certainly
comment on it, and probably should hear
from Paul
and Gerry on it.
I have a couple of other
comments, but go
ahead.
DR. HUSSAIN: No, I think that has been a
concern, and this was an extensive topic
that we
discussed at the London meeting of ICH,
especially
working group meeting on that. The language we
have crafted, I mean that took a long
time to
craft, to say there won't be any two
systems or not
tiers at all, in the sense it is a smooth
transition.
But the reality is you will
have much more
varied types of submissions.
DR. MORRIS: Absolutely.
193
DR. HUSSAIN: But that is managing that,
and so forth.
DR. MORRIS: And I agree, but I think in
that sense, you know, there were a couple
of
things.
You know, the evaluation of the
Pharmaceutical Development Report that
you had
mentioned, Helen, and really it should create
less
questions if you assume that most
questions come
from the lack of a presentation of a
rationale, so
hopefully, we will realize some
efficiencies from
that.
Similarly, Ajaz, I think the
review of the
critical variables and attributes should
be
improved by the Development Report, the
P2 and the
CTD, as well.
I guess this point you had made
in your
presentation, as well, I can't remember
if it was
Jon or you, Ajaz, but the idea that there
are
already in industry, the scientific
expertise. It
is a question of how they are organized
and linked
in terms of communication.
I think the same is true in the
Agency.
194
There is a good bit that is already in
place, that
just needs to be wired, and that is
really all I
had to say.
DR. KIBBE: Melvin, did you have
something?
DR. KOCH: I had a comment or question for
Jon.
It had to do with the new process or actually
even the old process in terms of
creating, say, a
draft or a guidance, and after you go
through the
legal audit, the public comment, and
eventually
have a final draft, I am just wondering
if there
isn't--maybe I should ask the
question--is there
anything that happens before the final
draft in
terms of explaining how the comments were
used?
I know they all show up on a
docket, but I
know with the most recent PAT final
guidance, there
were questions that those who are hoping
to use it
have, mainly because there is some part
of it which
seems to refer back to doing things the
way they
used to be, and there is a little bit of
inhibition
initially in how to use it.
I am just wondering, in the
process, you
195
mentioned something called a public
draft, and if,
say, after the public comments are taken,
you know,
the almost final draft shows up for some
quick
response.
MR. CLARK: The comment that we receive
when--this is true of the old process and
true in
new, as well--public comments received
are treated
as if they are FOIable whether they are
or not.
So, there is a documentation of comments,
an
indication of how they are collated, and
the
response to them if they are grouped
together as a
group or individually depending, that is
up to the
group whether they group them together or
not.
The FOI rules, they shift a
little on me,
I am not sure what level they are
available, but we
treat them as though they are. It's a public
draft, it was not meant to be any
different than
the current draft that we now publish. We
provide a
Notice of Availability in the Federal
Register and
then it is provided on the web. That draft is what
receives the public comments.
What I was referring to, the
main crux of
196
what I wanted to have accomplished here
was to
involve this body in evaluation of
prioritization
of guidances in general and where we
would need
them.
In other areas where we talked
about this,
it had come up that we needed to have a
better way
of looking at whether or not we needed a
guidance
in certain instances, because our methods
for doing
that now usually involve public
workshops, very
driven by the FDA, and perhaps we should
include
some more outside opinion as to where we
would need
guidance.
DR. KOCH: I guess I was wondering is
there a process for revision and/or
interpretation
of a final guidance?
DR. HUSSAIN: No, I think the policy in a
sense, we do not plan to sort of outline
how we
address the comments, only for regulation
changes
that is a requirement. For guidances, comments,
essentially, we don't share that
information.
MR. CLARK: If we receive some indication
that there are interpretations of the
guidance that
197
aren't going the way we anticipated, this
is a
point of having a question and answer
associated
with that document or with that topic.
DR. KOCH: Right, and it is in
that
context.
Anyway, we maybe could talk at some point
in terms of some specifics.
DR. KIBBE: Pat.
DR. DeLUCA: I just want to comment, and I
guess I best comment using Ajaz's slide,
the cGMP
Initiative. It is on page 14. It has been my
experience that most times in the
pre-IND, pre-AND
product development process development,
that these
is a hesitance at actually getting too
innovative
or pursuing new things and trying to
improve a
product with the idea that, oh, well,
after it is
marketed, then, we will do this, which I
don't
think in my experience that it happens
very often.
So, I like this. It is because you have
in the circle here that after you get to
improve
it, and there is a slot there for
continuous
improvement and innovation, and I think I
may have
said this before at previous meetings
that I think
198
even after postmarketing, that there
should be
vigilance in trying to improve the
process, and
sometimes you have to try to change both
in order
to do one or the other.
I am certain, seeing here that
the
expertise certainly exists in the
industry the
expertise exists here, and kind of
encourage that
it looks like the mentality may even be
moving in
that direction, too.
So, that is something that I
would
encourage as the desired state, that
postmarketing,
that there is this effort, continued
effort to
improve the product and the process.
DR. KIBBE: Gordon, you have a comment.
DR. AMIDON: Yes, I have a comment on two
of the slides that Ajaz presented. I mentioned one
to you at lunch, Ajaz. The other was--
DR. KIBBE: Just give a page number.
DR. AMIDON: On page 11 in your
presentation, the top slide. Interpretation and
Optimization of the Dissolution
Specifications for
a Modified Release Product with an In
Vivo-In Vitro
199
Correlation.
I am not sure what implication
the authors
were making with the dissolution passed,
the
bioequivalence passed and the dissolution
passed,
bioequivalence failed, but my question
would be
what dissolution did they do.
The impression I had from your
presentation was it was suggesting that
dissolution
was inadequate in predicting
bioequivalence, and it
obviously wasn't, but that is because
they did the
wrong test, the dissolution test.
So, this implies to me that we
need to
really be more specific when we are
talking about
dissolution, what is the test, did the
test reflect
what is going on in vivo, because if your
in vitro
test reflects your in vivo process, that
has to
predict what is going on. Otherwise, there is
magic in between.
DR. HUSSAIN: I think the theme of that
paper was twofold in the sense to
evaluate--if you
have the handout, it is part of that--the
team was
to evaluate the linkage between in vitro
200
dissolution and in vivo bioequivalence
and how
variability or random variability sort of
plays
into that.
Authors looked at what they
call a
non-optimal specification and an optimal
specification, optimized specification
based on in
vitro correlation, then essentially
simulated with
random variability built in, in vitro, as
well as
in vivo, what is the likelihood of
finding failed
dissolution, so that has been that analysis.
DR. AMIDON: I didn't mean for you to have
to defend the paper, and I will have to
take a look
at it, but there is something wrong here
in my mind
at least based on this one slide, because
if your
in vitro dissolution reflects the in vivo
dissolution process, it will predict
bioequivalence.
DR. HUSSAIN: No, that is the whole point
here in a sense. The point here, all we can do
with the current dissolution
methodologies are the
mean values. We have no idea on the variability.
So, once you factor in the variability,
then, you
201
see these aspects. That is the crux of that.
DR. AMIDON: The second comment would be
on just the previous page, page 10, at
the bottom
of the slide, which I did mention to you
at lunch.
I am confused whether this is a flow chart
or an
event chart of something.
The very first step, does
dissolution
significantly affect bioavailability, I
mean the
answer, I have problem with the answer
"No,"
because I can make a tablet no matter what
the drug
properties, it will not dissolve and
disintegrate,
so dissolution always affect
bioavailability, it is
a matter of how much.
I guess my broad comment, and I
will make
this again later in my presentation, is
that we
need more research on dissolution
particularly in
vitro to in vivo. That is where we have
the big gap
in our scientific understanding today.
DR. HUSSAIN: I think this is an ICH Q6A
decision tree, and now looking back after
years of
this, I do want to give credit to
Professor Nozer
Singpurwalla that he really pointed out
this is an
202
event tree, not a decision tree, and I
actually
before that had not paid attention to
that fact.
Now, when we look at it after
years, some
of the questions really don't make sense.
DR. KIBBE: Michael.
DR. KORCZNSKI: This is just a comment
related to the past two days of
discussion. I
think we really heard some excellent data
being
generated from the FDA labs, and I might
add, after
spending a number of years in the
industry, I wish
some of my retired colleagues could have
heard the
discussion. I think they would have marveled at
the progress being made scientifically.
A thought comes to mind. I think this is
a very significant question. I have tried to
distill a significant thought here. Can innovative
drugs, as a result of where we are going
and the
data we are seeing, eventually be reduced
in scope
relative to clinical trials and without a
loss of
patient safety or efficacy?
Just to kind of encapsulate
that, if,
indeed, you do have some very excellent
innovative
203
types of assays at the preclinical level,
such that
they perhaps might even mitigate to some
degree or
reduce Phase I studies, because they are
so
thorough and better predictors, and then
if you got
to that stage, could there be a
consideration of
collapsing Phase I and Phase II clinicals
into one
clinical trial basically, then, with all
the data
that we saw from the ICSAS staff, the
computerized
data, is it possible in some way to
network with
sponsor companies, such that more drugs
could
probably be placed in an orphan drug
clinical
pathway as opposed to the current
conventional
Phase III.
So, long term, there are some
real
opportunities, I think, in condensing and
collapsing that costly and lengthy
clinical trial
procedure, and the data to a large extent
that is
being collected.
Just another two items. One, we talked
about innovative drugs. I think what is going to
be important, this is kind of futuristic,
would be
the identification of target sites for
disease
204
states, and very important with the
development of
these nanomolecules and small peptides
will be the
delivery systems. What will be the micro delivery
system to the host site especially as we
talk about
gene therapy?
So, the development of that
technology has
to occur in concert with the development
of the
molecules, so we are going to have the
molecules
and not the means to deliver to disease
state host
sites.
So, that is just an
encapsulation. Thank
you.
DR. HUSSAIN: On the last point, I think
there has been an amazing growth in that
particular
area, vehicles, nanovehicles, nanodrug
delivery
systems, and so forth, and we hope to
publish a
paper soon on the topic of dendrimers and
then how
dendrimers can sort of shrink promers
[ph], and so
forth, so there has been wonderful
progress even on
that side, on the delivery part of it.
DR. KIBBE: Anybody else?
Okay. So, we
kind of beat that into the ground, and we
got that
205
under control.
We need to move forward. I see that
Lawrence is getting ready to go. He has two sets
of presentations, and he agreed that if
we would
just do exactly what he says in the first
set of
presentations, he would be really fast on
the
second set.
Pharmaceutical Equivalence and
Bioequivalence
of Generic Drugs
The Concept and Criteria of
BioINequivalence
Concept of
BioINequivalence
DR. YU: Good afternoon. This afternoon
we have to deal with two topics. One is follow-up
topics which we have presented this
committee six
months ago, and a second topic,
introduction topic,
is called Bioequivalence of Locally
Acting GI
Drugs.
Let me go through the first
topics of
bioINequivalence, concept and definition,
which
this topic, as I said, was presented to
you six
months ago on exactly April 14, 2004.
The bioavailability is defined
as the rate
206
and extent of drug absorption. Regularly, we use
the Cmax as a surrogate for the rate
absorption and
the AUC, all to the exact area under the
curve used
for the extent of the drug absorption.
So, the bioequivalence is
defined as
absence of a significant difference in
the rate and
extent absorption, and many other things,
for
example, become available at the site of
drug
action when administered at the same
molar dose
under similar conditions in an
appropriately
designed study. So, this is basically the
CFR
definition.
With bioequivalence, we usually
use 90
percent confidence interval for AUC or
the extent
of absorption, or Cmax rate of absorption
between
80 and 125 percent.
The passing bioequivalence
criteria or
confidence interval allows market access.
Certainly, we have to, for example,
generic firm
have to meet other quality
standards. If you pass
the confidence interval, or we call the
bioequivalence for generics, this means
the generic
207
approvals, for innovators, this means
demonstrated
to be marketed formulation to be
equivalent to
clinical formulation.
Those concepts are well
understood, well
developed, widely used.
The question remains why do we
need to
define bioINequivalence? It is because FDA
receives studies, as we discussed six
months ago,
receives studies that attempt to reverse
a previous
finding of bioequivalence, in other
words,
companies, whether innovator companies or
generic
companies out there, to conduct a study,
say, the
generic products or products on the market, in
fact, is a bioINequivalent.
The bioINequivalent definition
is not very
well defined. In many cases, to be scientific
term, those studies actually should be
defined as
failed to demonstrate bioequivalence to
be exact.
That is part of reason when you define
the criteria
of a bioINequivalence, when you define
the concept
and criteria of bioINequivalence.
The question certainly, what
should
208
bioINequivalence mean? Bioequivalence leads to
market access, bioINequivalence leads to
market
exclusion. That is what
bioequivalence/bioINequivalence basically
means.
Now, come back to say look at
the results,
what does it specifically mean when I
said failed
to demonstrate bioequivalence, failed to
demonstrate bioINequivalence,
bioequivalence or
bioINequivalence, those concepts.
In the center, the top one is
demonstrate
bioequivalence just because the
confidence interval
for this study is within the
bioequivalence
interval of 80 to 125 percent.
The bottom one is demonstrated
bioINequivalence, I use BIE here, it
stands for
bioINequivalence. Now, the middle one, basically,
the left side or right side, we have fail
to
demonstrate bioINequivalence and fail to
demonstrate bioequivalence. This basically is
because neither, whether either outside
of the
confidence interval, outside of
bioequivalence
interval, 80 to 125 percent or not
completely
209
inside.
Now, since the April
discussion, I
received several phone calls and also we
had a
discussion within FDA with my colleagues,
some
question came back that by definition of
bioINequivalence may be too far. Some people
suggest maybe use mean of the point
estimate,
confidence interval, while the suggestion
was that
if you use the point estimate or mean
ratio outside
or above 125 percent, it should be
defined as
bioINequivalence.
Now, certainly this makes a lot
of sense.
However, if we look at it deeper,
statistically, it
is not.
Let's look at an example here.
The top
one, obviously, it failed to demonstrate
bioINequivalence. Now, failed to demonstrate
bioINequivalence could have been for many
reasons.
One of the reasons you would think is not
enough
subjects, for example theoretically, they
should
have used 100 subjects or 50 subjects,
and this
study only used 10 subjects.
Because of the small number of
subjects,
210
it makes the confidence interval of the
final
results much wider, therefore, end up you
have a
failed to demonstrate
bioINequivalence. We see we
cannot use this result, failed to
demonstrate
bioINequivalence as definitive answer, is
because
if you use large number of subjects,
there is two
possibilities as you can see here.
Could be bioINequivalence
completely below
the 80, outside of the bioequivalence
confidence,
either below 80 or above 125 percent.
There is another possibility
that even
though the study, this is a small number
of
subjects study shows bioINequivalence, in
reality,
they could be bioequivalent, as you show
here on
the left side, right side or left side.
So, the
right side one is because once you use
large
numbers of subjects, could give you a
definitive
answer, you end up even though the study
itself
fails to show bioINequivalence, but at
the end you
have two outcomes, and that is the power
that could
be
demonstrated bioINequivalence or demonstrated
bioequivalence.
211
That is part of the reason we
say we may
not be able to use these results to have
a
definitive answer to make regulatory
decisions.
The objective is the same which
we were
discussing back to April 14th, to develop
a
bioINequivalence criteria that are
scientifically
sound, statistically valid, fair to all
parties,
and easy to use.
I want to remind you that the
comments or
conclusion draws back to the last
discussion. The
first question, back to April 14th: Does the ACPS
agree with the distinction between demonstrating
bioINequivalence and failure to
demonstrate
bioequivalence?
The answer was yes. That was the
conclusion reached back to April 14,
2004.
The second question: Does the ACPS
recommend a preferred method for
evaluating the
three pharmacokinetic endpoints for
bioINequivalence?
There are many
sub-questions. Here are
the comments which were discussed.
212
The committee agreed on a
general
understanding of bioINequivalence to move
forward
by recognizing this is not a simple
matter. In
addition, the members felt that there is
an
important concept, especially now it
applies to
entire regulatory scenario. There was no consensus
at this point as to a final criteria
pertaining to
the three pharmacokinetic endpoints.
We will present you today our
recommendations based on those
discussions and hope
we can follow the comments or discussion.
DR. MEYER: Could we ask a question before
we get confused with statistics?
DR. KIBBE: Okay, great.
Why don't we ask
the question before we get statistical.
DR. MEYER: I like to approach things in a
very simple manner. If you could get slide 7 back,
Lawrence.
It strikes me that you are
trying to use
phraseology to fit a subsequent
statistical
analysis, and I am wondering if there
isn't a
simpler way to go about it. I am kind of, of the
213
school that you are either pregnant or
you are not,
and if you haven't taken the test yet,
you can't
say.
So, that is basically where I am going to go.
I am going to number these 1,
2, 3, 4, 5.
I think we can all agree that No. 1 is
bioequivalent, and No. 4 and 5 are
bioINequivalent,
just as you have shown.
DR. YU: Yes.
DR. MEYER: I say No. 2 is
bioINequivalent, not failed to
demonstrate
bioequivalence, the
pregnancy/non-pregnancy thing.
I say that because the mean is well
outside 80
percent, and increasing the numbers may
shift the
mean one way or the other. You shifted it, of
course, to the left to make it worse.
I would say maybe it will stay
the same.
All I know is that the means are
terrible, and the
confidence limit kind of extends over to
acceptable
range, but in my view, if somebody came
to me with
that data and said I ran the study on 40
subjects,
should I do 80, I would say no,
reformulate.
If they came to me with a
confidence level
214
that lopped that over, I would say, yeah,
do an
added number.
So, I think the N problem, which could
expand confidence limits, can be solved
by
requiring anyone that dose a
bioINequivalence study
to fix their N at the same as the person
that got
the approved ANDA. So, if it took Teva 485
subjects to do their study, then, I think
Pfizer
ought to do 485 subjects if they are
going to try
to prove that Teva is no longer
bioequivalent for
some reason.
No. 3, I would say that fails
to
demonstrate bioequivalence by the current
standards, but it also fails to
demonstrate
bioINequivalence, because the means are
well below
125.
That is a wash.
You can't tell one way or the
other,
therefore, whoever is doing the study
can't make a
claim of bioINequivalence, because there
is
ambiguity in the data. That is a much simpler
approach than trying to come up with a
new metric
of 3 parameters or 1 parameter or what
have you.
215
DR YU: Indeed, Marvin, you have excellent
questions. That is true, I guess, we have too many
discussions on this topic.
Well, come back to your
question, is that
does fail to demonstrate
bioINequivalence, top, No.
2, is actually demonstrated
bioINequivalence here,
that is the question.
Indeed, you point out it
probably could be
bioINequivalence if the sponsors, whoever
conduct
the study, you have sufficient power, was
sufficient in number of subjects.
I guess the question comes back
under this
scenario, that for top one, which is
clear,
demonstrate bioequivalence, and the
bottom one is
clear, demonstrate bioINequivalence,
however, for
No. 2 or No. 3, this does not
necessarily--especially for No. 2--if we
receive
the data, this does not necessarily
suggest, we are
not going to take any action whether or
not you
look at it, simply because you fail to
show
bioINequivalence, therefore, we are not
going to
take a look at it, we are not going to
take any
216
action, that is not the case.
Certainly, this case, if you
submit it to
the Agency, and the study is well
conducted, well
powered, we have to look at all of the
scenario and
then to draw a scientific decision, or I
guess what
I am saying is top 1, and top 4 and top 5
give us
definitive answer, top 2 and top 3, we
have to look
at it case by case. We cannot draw very decisive
conclusive decisions is part of the
reason you have
assumption here, you have a sufficient
number of
subjects.
If you don't use a sufficient
number of
subjects, say, you only use 1 or 2
subjects,
certainly, we were not able to say you
have
demonstrated.
I hope I answered your
question. I guess
we answer your "if" questions.
DR. MEYER: How about fixing the N at the
same as the ANDA submission used?
DR. YU: That is one of the options, yes.
DR. KIBBE: You are responding to a
question that I don't think the Agency is
asking.
217
You are responding to a question that a
sponsor of
one of those studies would ask in order
to get the
study to do what it wants to do.
Then, you are saying, well, the
Agency
should say if they did this, then, they
probably
would show this and what should we do
about that.
The
difficulty for me in this whole scenario is
that the failure to demonstrate
bioequivalency
doesn't necessarily prove
bioINequivalency, and if
a product is already on the market
because it has a
bioequivalency approval, what level of
information
do we need to reverse that decision.
I agree with Lawrence, those
two in the
middle wouldn't justify in my mind as a
regulator a
change in the previous decision, whatever
it was.
Okay?
DR. MEYER: If the orange one, No. 3, was
done under the same conditions as the
ANDA holder.
DR. KIBBE: If it was done under the same
conditions as the ANDA holder, I wouldn't
reverse
my decision on anything.
DR. MEYER: That is an exaggeration
218
perhaps because let's say it's 127, you
can't get
real excited about that. No, it would have to be
125, but let's say it was 145.
DR. KIBBE: What I am saying is we already
have a product that passed once.
DR. MEYER: Right.
DR. KIBBE: That study doesn't help me
decide to reverse that decision.
DR. MEYER: The product has changed.
Remember, Gary Levy published The Clay
Feet of
Bioavailability or Bioequivalence
Testing. You do
it once on a hand-picked lot against one
lot of the
innovator product, a fresh lot of yours,
and one
you have selected maybe out of 12 of
theirs, and
then somebody else comes along with an
older lot of
theirs or vice versa, and no longer are
you
bioequivalent.
DR. KIBBE: I understand the argument. I
am just saying that, as a regulator, I
wouldn't
change anything I have got on the books
based on 1,
2, 3.
Okay?
DR. YU: I guess I will make one more
219
comment. When you conduct bioequivalence
studies,
you look at availability, you look at the
power,
you look at the subject. You are basically always
saying here that the bioequivalence
criteria is 80
to
1 to 25 percent. Agency never defines
how many
subjects you use. You could have used 24, 48, 96,
500, for example, for clinical endpoint
studies.
So, when you define the number
of
subjects, then the confidence interval
could be
variable, one way or another. You define one, and
you have another criteria.
I think that for the
bioequivalence
criteria, we define the confidence
interval instead
of define the number of subjects.
DR. KIBBE: Go ahead, Nozer. We are
having a lot of fun here.
DR. SINGPURWALLA: I think it is always
fun when committee members disagree, but
something
bothers me about this whole concept.
DR. YU: That was not the--
DR. SINGPURWALLA: That was not the
question, I understand.
220
DR. YU: That was not the question.
DR. SINGPURWALLA: I understand, that was
not the question.
DR. YU: If we don't want to live with it,
I guess the decision will have to be
made.
DR. SINGPURWALLA: I am sorry, my comment
here is you are building a castle on
sand, I think
this whole idea--
DR. KIBBE: Or clay, right?
DR. SINGPURWALLA: Whatever you want to
use, it's a castle that cannot hold
up. What you
have done is you looked at Cmax and you
looked at
AUC, and if the Cmax and AUC are not
significantly
different, you say there is
bioequivalence. Not
true?
What is it then?
DR. KIBBE: If you are going to give an
answer, Don, you might as well get on a
microphone.
MR. SHERMAN: Lawrence showed a quote from
the regulations on the definition of
bioequivalence, and it said rate do not
show a
significant difference. The word "significant" in
that sentence does not mean statistically
221
significant, it means significant the way
the word
is used in the English language,
substantial,
important.
DR. SINGPURWALLA: Then, my comment
becomes even more acute. The whole thing should be
relooked, revisited, because I kind of
agree with
Lawrence about those two, and I agree
with our
chairman about the two middle ones as
demonstrating
failure to demonstrate, then
demonstrating. I mean
I wouldn't change anything, but I think
the whole
concept of doing all this through this
particular
vehicle of setting confidence limits and
looking at
the little tail falls here or there seems
completely capricious to me, and you may
want to
revisit this whole topic.
DR. KIBBE: We have been revisiting this
since 1970 at least.
DR. SINGPURWALLA: I was not there.
Change the paradigm.
DR. KIBBE: We are not revisiting the
paradigm.
MR. SHERMAN: I just want to correct the
222
notion that you look at Cmax and AUC and
you
approve products as inequivalent if there
is not
statistically significantly
difference. That
hasn't been true for decades.
DR. SINGPURWALLA: So, it has been
nonsense for decades.
MR. SHERMAN: You are entitled to your
opinion, sir.
DR. YU: We will move on to the next
topic.
Thank you.
Criteria of
BioINequivalence
DR. LI: Good afternoon. My name is Qian
Li.
I am from Office of Biostatistics in CDER,
FDA.
I am going to present a statistical criteria
for evaluating bioINequivalence using
multiple
endpoints. I know there have been citing
for using
one endpoint, but I decided to move on to
talk
about multiple endpoints.
Before I start to talk about
the criteria,
I would like first to discuss the
question why we
need to use multiple PK endpoints to
assess
bioINequivalence.
223
To answer this question, we
need to
understand that for systematically
delivered drug
product, bioequivalence established by
comparing
the rate and extent of drug
absorptions. The rate
and extent are usually represented by
Cmax, AUCt,
and AUCinfinity.
In statistical terms, Cmax,
AUCt and
AUCinfinity, I refer to as PK
endpoints. For
bioequivalence assessment in generic drug
approval,
it has evolved to use AUCt, AUCinfinity,
and Cmax.
As Lawrence has mentioned before, that
you have to
prove that 0 to 3 PK endpoints to be
equivalent in
order for the generic drug product to
have market
access.
Now, for bioINequivalence, it
can be
established if one of the PK endpoints
are
inequivalent in truth.
Now, in reality, we do not know
the truth,
so we have to perform statistical
analysis to test
all the PK endpoints. That is why we need to
assess multiple PK endpoints. I hope this is clear
to everybody now.
224
Now, this is the outline of my
presentation. I will give a brief review on the
criteria for bioINequivalence using one
PK endpoint
and then present strategies for assessing
bioINequivalence using three PK
endpoints, and I
will discuss available approaches and
then present
power comparisons of those approaches.
Then, I will discuss FDA's
recommendation
of using the three PK endpoints in
assessing
bioINequivalence.
Now, let's first look at
definition of
bioequivalence and the inequivalence
using one PK
endpoint.
We use the ratio of geometric
means Mu
T/MuR
to define bioequivalence and
INequivalence. Mu
T
represent the geometric mean of the test
product,
and the Mu
R is the
geometric
mean of the reference
product.
The bioequivalence is true when
the ratio
is between 80 percent and 125
percent. We call
this bioequivalence interval. Outside this
bioequivalence interval is defined as
225
bioINequivalence region.
Now, to test the
bioINequivalence, the
null hypothesis is the bioequivalence
interval.
The alternative is the bioINequivalence
regions.
Similar to bioequivalence test, we will
perform
two, 1-sided test, as well for
bioINequivalence
assessment.
The null hypothesis for 1-sided
test is to
have the ratio reached and equal to 80
percent, and
the alternative is less than 80
percent. There is
another test that now is less than or equal to 125
percent, and the alternative is driven
125 percent.
We can claim bioINequivalence if one of
the two nulls is rejected. We perform the test on
the significance level of 0.05. Now, this is the
equivalent to form 2-sided 90 percent
confidence
intervals.
The criteria to claim
bioINequivalence is
when the 2-sided 95 percent confidence
interval
lies completely outside the
bioequivalence
interval.
I would like to remind
everybody here that
226
using 2-sided 90 percent confidence
interval for
bioINequivalence test, the error rate is
not always
protected at 5 percent level. This is different
from bioequivalence test.
If we have a reasonable
conducted study,
say, it's a 2-sequence and a 2-way
crossover
design, and the subject is more than 20,
and the
within-subject standard deviation is less
than 0.7,
then, we can safely control the error rate
to 5
percent, however, if the subject sample
size is
less than 20, we have large variance for
the test
statistics, then, the error rate may not
always be
controlled at 5 percent. In this case, we might
have to consider to use 2-sided 95
percent
interval.
Let's move on to the definition
of the
bioequivalence and the inequivalence
using three PK
endpoints. As mentioned before, the three PK
endpoints is Cmax, AUCt, and AUCinfinity.
The definition of
bioequivalence is the
cubic region in this 3-dimensional
diagram.
Outside this cubic region will be the
227
bioINequivalence region.
For the criteria for
bioequivalence
assessment using three PK endpoints is to
require
all 3, 2-sided 90 percent confidence
interval for
the ratios of our geometric means has to
be
reaching the bioequivalence limit.
Now, the error rate of
wrongfully
rejecting bioequivalence using this
criteria is
protected at 5 percent level.
Now, for bioINequivalence
assessment using
three endpoints, we are looking for a strategy
that
can control the error rate of wrongfully
rejecting
bioequivalence at a rate of 5 percent.
Also, we want to control the
error rate
under all correlation structures because
we do not
know the correlation structure of the
three PK
endpoints.
Now, to develop those
strategies, we
assume the variances of test statistics
are not
large.
Now, there is a common
misconception when
assessing bioINequivalence. The common
228
misconception is that to claim
bioINequivalence,
when one of the three PK endpoints
satisfies the
bioequivalence criteria, which is the
2-sided 90
percent confidence interval is outside of
the
bioequivalence interval.
Now, we will not accept this
kind of
approach to assess bioINequivalence
because it will
inflate error rate of wrongfully
rejecting
bioequivalence. The error rate can be up to 15
percent if three endpoints are
independent, can be
about 8 percent if the three endpoints
are highly
correlated. We consider that approach is quite
liberal.
Now, people may want to think
about we can
use quite tough criteria, which is to
claim
bioINequivalence if all the three PK
endpoints
satisfy the bioINequivalence criterion,
which is
2-sided 90 percent confidence interval
outside of
the bioequivalence interval.
These tough criteria will
tightly control
the error rate under all correlation
structures,
however, it won't provide good power to
demonstrate
229
bioINequivalence, therefore, we are not
recommending to use this criteria either.
What we would like to recommend
are the
following strategies. The first strategy we
present here is to pre-specify one of the
three PK
endpoints for bioINequivalence test. For example,
if you decided to use AUCt to test the
bioINequivalence, then, you can perform
analysis on
this endpoint only, ignore the other two.
The requirement for this strategy is
that
you have to pre-specify this endpoint in
your study
protocol.
Otherwise, you could end up switching
endpoints, which may inflate error rate,
which we
don't like to see that.
Now, this strategy is ideal for
situations
when one knows that one specific PK
endpoint is
more likely to show bioINequivalence than
others,
but it may have poor power if you
misspecify the
endpoint.
Another strategy we would like to
recommend is called Bonferroni
corrections. There
are many versions of Bonferroni
corrections. One
230
example of using Bonferroni correction is
to use a
2-sided 96.7 percent confidence intervals
for three
endpoints instead of 90 percent
confidence
interval.
If one of the three 96.7
percent
confidence interval fall in the
bioINequivalence
regions, we will say the bioINequivalence
is
demonstrated.
This strategy is ideal for
scenarios when
one knows that one PK endpoint is more
likely to
demonstrate bioINequivalence than others,
but do
not believe that all the endpoints have
good power
to demonstrate bioINequivalence.
Another strategy we would like
to discuss
here is to use three confidence intervals
with
different lengths. This can be considered a
variation to the approach that requires all
the
three endpoints to satisfy the
bioINequivalence,
which is the very tough criteria. This criteria
will control the error rate no more than
5 percent.
One example of this criteria is
to use 94
percent confidence interval, 98 and 96
percent
231
confidence interval for the three
endpoints instead
of all three, 90 percent confidence
intervals.
This approach is ideal for
situations when
one has no idea which PK endpoint has the
best
power, but you know that all the three
endpoints
could show bioINequivalence.
To support what we have discussed
for the
three strategies, I would like to show
you some
power examples for several scenarios for
the three
strategies.
We calculated power under two
correlation
structures. The first scenario is that only one
endpoint has good power to demonstrate
bioINequivalence. One example is that AUCt has
only 5 percent power, AUCinfinity has 20
percent
power, and Cmax has the best power, which
is 90
percent.
In this case, if we choose
pre-specified
strategy, we could have a power between 5
percent
and 90 percent, so this example clearly
shows that
if you know Cmax is the endpoint that
could give
you the best power to demonstrate
bioINequivalence,
232
then, you should choose the strategy to
pre-specify
Cmax in your protocol.
In this example, Bonferroni
correction
also can give you quite a robust power,
but the
varying confidence interval approach is
not as good
as the other two approach.
Now, for second scenario, which
is all
three endpoints have reasonable power to
show
bioINequivalence. Now, here, one example
is AUCt
has 60 percent power, AUCinfinity has 70
percent
power, and Cmax has 80 percent power.
Now, the per-specified approach
give you
60 to 80 percent power. The Bonferroni correction
will give you about 64 to 72 percent
power, and
under the varying confidence interval
approach, we
will give you about 70 percent power.
So, in this case, if you feel
that all
three endpoints could demonstrate
bioINequivalence,
then, varying confidence interval
approach might be
a good choice.
This scenario is that all three
endpoints
have equal power. All has 80 percent power. In
233
this case, if you know exactly that all
has 80
percent power, then, you can choose
pre-specified
approach, but this is probably unlikely
known to us
before we do the experiment.
Now, in this approach,
Bonferroni
correction will give you decent power,
but varying
confidence interval will give you
probably the best
power if you don't know that all the
endpoints has
equal power.
This leads to the summary of our
recommendations on using three PK
endpoints for
assessing bioINequivalence. When one knows which
endpoint is more likely to show
bioINequivalence,
Strategy I should be used, which is to
pre-specify
the endpoint in study protocols and use
the
endpoint to test bioINequivalence. For this
approach, you should use a two-sided 90
percent
confidence interval.
When one knows that one
endpoint may have
good power, but do not believe that all
of them,
all of the endpoints have good power,
then, we
suggest to use Strategy II, which is the
Bonferroni
234
correction.
One example of Bonferroni
correction is to
use a two-sided, 96.7 percent confidence
interval.
If you believe that all three endpoints
could have
reasonable power to show
bioINequivalence, then,
Strategy III should be recommended.
The example of Strategy III is
to use 94
percent and 98 and 65 percent confidence
intervals.
Thank you.
DR. KIBBE: Questions?
Go ahead.
Committee Discussion and
Recommendations
DR. GLOFF: Art suggested over lunch that
I say something this afternoon to earn my
keep
here.
I have a question, it is probably an
uninformed question, so I apologize for
that.
Is there any provision in all
this to look
at the results for the reference product
relative
to the results for the reference product
that were
obtained either by the company that
submitted the
ANDA in the first place or prior results
submitted
by the holder of the NDA?
DR. LI: To my knowledge, I don't think
235
so.
I don't know.
DR. YU: I guess the bioequivalence
confidence is defined 88 to 125 percent,
so that is
the criteria we use. Under certain
circumstances,
when, for example, the variability is
significantly
high, we will get clinical studies. We will look
at the availability, how much they impact
the
confidence interval, but the criteria
still remains
80 to 125 percent.
DR. GLOFF: Well, I am sure you probably
have thought of why I am asking that
question. A
prior speaker made the comment that when
a generic
is being developed and submitted, that
they could
hand-pick the lot that the generic
company uses to
compare to and hand-pick the lot of the
innovator
product.
I am sitting here thinking,
well, why
couldn't the innovator company do the
exact same
thing to try to demonstrate that the two
were
bioINequivalent or were not
bioequivalent, and I
don't know if the Agency came to that
conclusion,
what they would then do for sure, but I
am
236
concerned about that, that you do that,
and then
where are you.
DR. YU: I guess the information for
bioavailability, bioequivalence, as
clinical
pharmacology sections, it is available in
the
public domain, so any sponsors, any
companies out
there, you can request, go through
Freedom of
Information and get those information
from FDA
before you conduct any studies.
DR. KIBBE: Go ahead, Ajaz.
DR. HUSSAIN: I think the question that is
being asked is how do we relate one study
finding
to what happened in the previous
one. I mean that
is the fundamental question.
We actually don't do that. We often don't
do that.
But I think that is an important point,
and I have tried to look across different
ANDA
submissions, especially when Gordon
Amidon was at
FDA and we did a lot of the data mining
for our BCS
classification, we looked at all that.
I think there is value to that,
but often
we find that absolute numbers that you
see in terms
237
of percentages is fine, but the absolute
values
that you see depends on the assay
variability, and
so forth, differences, and so forth.
But I think no matter what you
see, one
study, the second study being done to
show
bioINequivalence in the first study,
there is an
aspect of selecting the thing, and that
has been
discussed as Marvin said, profoundly by
the father
of biopharmaceutics, Garrett Levy, so
that is part
of the systems.
DR. KIBBE: Jurgen.
DR. VENITZ: I have a question on Slide
11.
This is where you were discussing using the
three PK parameters and you are
commenting that you
don't recommend that because there is not
adequate
power.
I would like for you to explain that
statement to me.
DR. LI: This criterion to require all
three endpoints to show bioINequivalence,
the
bioINequivalence criteria for one single
endpoint
is 2-sided 90 percent confidence outside
of the
bioequivalence region.
238
So, if you remember our
previous talk, we
showed a power of showing bioINequivalence. It is
pretty hard to show bioINequivalence for
one
endpoint.
DR. VENITZ: How do you define power?
DR. LI: Power is the probability to show
bioINequivalence if bioINequivalence is
true.
DR. VENITZ: So, you are ignoring the fact
that you have a previous study that says
the two
products are bioequivalent, in other
words, your
power is only defined post hoc after this
single
experiment that you are trying to
address?
DR. LI: No, the power is not defined by
the experiment. It is a probability which you
don't know actually, you do not know.
DR. VENITZ: But you do have prior
information that two products are
bioequivalent,
right?
DR. LI: Right, you could have. What has
driven power is how far the
bioINequivalence away
from the bioequivalence interval. The further away
from bioequivalence interval, you have
back-up
239
power, and also it depends on how many
sample size
you use, which is sample size basically
reduce the
variability.
DR. VENITZ: But don't you then ignore, as
I said before, in your power, the way you
define
power, the fact that you have prior
information?
You are just basing it on a single
experiment. You
already have an accepted study that says
those two
products are bioequivalent, and now you are
saying,
well, I need more power to show that they
are
bioINequivalent. Isn't that kind of a
contradiction?
DR. LI: I didn't see the contradiction.
What I am trying to explain to you, that
to use
three endpoint to show bioINequivalence--
DR. VENITZ: Is hard.
DR. LI: --is harder.
DR. VENITZ: Right.
DR. LI: Because the power is lower.
DR. VENITZ: And I am saying you already
have evidence to suggest that they are
bioequivalent, shouldn't it be harder.
240
DR. LI: Well, no, if the drug is truly
bioINequivalence, if you design your
study
properly, you should have good power to
show that,
but if you choose this criteria, you
probably won't
have good power. You could choose the better
criteria that give you better power.
DR. KIBBE: You are beaten.
DR. VENITZ: Okay.
DR. KIBBE: She is talking about
statistical power of the individual study
presented
to her.
DR. VENITZ: I am talking about the
overall power to rule whether something
is
bioequivalent or not in the totality of
the
information that you have, not just the
specific
study, which is what you are talking
about. You
are talking about a specific study where
you look
at the three parameters individually, you
correct
it or all three of them.
DR. YU: I guess, Jurgen, you are
absolutely correct. Actually, we have many, many
debates and discussions, Qian knows that,
talk
241
about when the statistic versus you have
a prior
knowledge about bioequivalence or
bioINequivalence
or quality.
Certainly, what we are trying
to address
here is actually, you have five potential
options.
One of the options is you have no prior
knowledge
whatsoever. That is one of them we have to present
as a complete picture, we are not recommending
this.
One of the options, we say--all
the option
scenario out there, I guess, one of the
scenarios,
in your mind, you have a prior knowledge
that is
impossible, but for the completion of the
picture,
that was presented.
DR. KIBBE: We have got Ken and then Paul,
Nozer, and me, and Marvin.
DR. MORRIS: A quick comment.
Irrespective of whether this is innovator
or
generic, I think the fact that you can
hand-pick
lots that are this different says more
about the
process that is being used to make our
products
than it does our testing for
bioINequivalence. I
242
think this was Levy's point either
directly or
indirectly is that it is probably more to
the point
that we need to control our processes to
the point
where Jurgen's observation becomes the
rule in a
sense.
DR. KIBBE: Paul.
DR. FACKLER: Part of the problem, I think
is we are trying to do an exact science
here where
the whole issue is so variable, it is out
of our
control.
If we run the same study in the same set
of subjects twice, we will get two
different
results, the same drug product, the same
people,
and it's different.
The variability is just
unmanageable, so a
generic company will take a lot of
innovator
product and a lot of generic product and
run it in
a certain number of subjects, and that
number is
calculated to give us 80 percent power.
Four out of
five times, those products will
statistically
appear to be bioequivalent, and one out
of five
times they won't, they are not
bioequivalent.
It's statistics, and we should
not spend
243
too much time trying to get an exact
measurement
here of what bioequivalence is, nor what
bioINequivalence is. I think the question is
really when is a product not going to
perform for
the patient who is taking it, and we have
arbitrarily said 80 to 125 works, and
there is some
anecdotal evidence over the last 25 years
that the
generic products on the market work.
So, I would just caution the
committee to
be careful defining when you would want
to pull one
of those products off the market.
DR. KIBBE: Nozer.
DR. SINGPURWALLA: Well, first is I think
you have done a very thorough, detailed
analysis
given a badly defined problem. You have done very
well.
Thank you.
Now, I am going to comment. I am going to
ask you two questions. You have this AUCt. How do
you pick the t?
DR. LI: The t is the time point that you
can still identify the drug concentration
in your
blood.
244
DR. SINGPURWALLA: It's the last.
Is it possible that for one t,
you will
arrive at one decision, and for another
t, you will
arrive at another decision, which goes
back to what
Paul has been cautioning us about?
DR. LI: Uh-huh.
DR. SINGPURWALLA: That is one comment.
The last comment is your last
viewgraph. When you
say "recommendations," you have
three bullets.
When one knows this, you do this. When one know
this, you do this. When one knows this, you do
this.
Well, what do you do when one
knows
nothing?
DR. LI: Actually, this is a very good
point.
If you see my example, the 3 example, if
you know nothing, actually, Bonferroni
correction
give you quite reliable robust power even
it is not
the best power for that situation, but it
give you
pretty good power, we might recommend
this, and
actually, this probably will be the
default
approach for us to review if the sponsor
didn't
245
specify any approach.
DR. SINGPURWALLA: So, maybe you should
put a fourth bullet, if you know nothing,
do this.
But now let me go back to the
main
discussion that has been spawned by
Carol's
question.
I think both Jurgen and Ajaz have been
dancing around the issue, and not coming
out right
and saying what is on their mind.
Basically, what is on their
mind is if you
have prior information, which you do
have, what do
you do, and really what we should do is
not address
the problem in the manner in which this
is
addressed. No criticism intended to you of the way
you have done it. You have done it very well.
I think you should formulate this
as a
problem in making decisions. You either declare
bioequivalence or you declare
non-bioequivalence,
and the declaration of one or the other
is a
function of what risks it may entail if
you make
the wrong decision, so that takes care of
Paul's
argument that there is so much
variability.
You make decisions in the face
of
246
variability, so that would recast this
whole
problem, reformulate it, and readdress
it. It's a
serious issue, because really, what you
are all
doing is building a superstructure on
something
that is not carefully defined.
Thank you.
DR. KIBBE: Marvin, do you want to jump in
or do you want me to jump in?
DR. MEYER: Well, let me just comment. In
my experience--and there is probably
exceptions
certainly--if I had to pick a parameter
to show
bioINequivalence, I would go with
Cmax. That tends
to be a lot more variable, wider
confidence limits,
so I think you could probably go a priori
with Cmax
and use Strategy I if you wanted to do
that.
You are saying if I do that,
then, the
confidence limits has to be totally
outside of 80
or 125.
DR. LI: Right, yes.
DR. MEYER: Otherwise, you can't tell
perhaps.
DR. LI: Yes, it is exactly the picture
247
Lawrence showed you before, and if you
feel
uncomfortable, I think there is a second
picture
that has failed to show bioINequivalence.
Actually, I would like to come
to answer
that question from a statistical point of
view.
You like to see, you know, the picture,
if the
confidence interval overlap to about
bioequivalence
interval, which is the second case.
DR. MEYER: I am really more worried about
means.
DR. LI: The mean is outside the
bioequivalence. Let me tell you about the
statistical concerns. If we claim, be clear, this
is bioINequivalence, then, you end up to
make the
error that is a modern FAQ [ph] event,
which for
statisticians, we do not like to see this
happen,
and if you have the confidence interval
completely
outside of the bioequivalence interval,
then, we
are comfortable that if you would claim
this drop
is bioINequivalence, we won't make error
more than
5 percent.
I know for pregnancy test, you
could claim
248
that lady not pregnant, even 51 percent
sure, and
you could make a 49 percent error, but
for
statistical decision, for regulatory
decision, we
cannot make more than 5 percent
error. That is why
we define it has to be outside the region.
DR. MEYER: I guess I worry about too much
rigor in that. Let's say the point estimate was 60
percent, pretty bad, and the confidence
limit,
because of high variability, went over
onto 80.2
percent.
You have a bioINequivalent
product, there
is no question about it. You are not going to fix
that by anything other than
reformulation, but
because of variability, you managed to
slop that
righthand tail over above 80.
DR. LI: What if people tell you the study
is conducted using only five subjects,
and you see
the point estimate is 60 percent, and you
have a
confidence limit, you know, almost
everywhere, can
you claim this is bioINequivalence?
DR. MEYER: No, that is why I think you
ought to fix N, too, to avoid that kind
of an
249
issue, and maybe even fix the mean must
be less
than the mean ratio in the ANDA.
DR. LI: If we fix, that will lead to
stagnation [?] of bioequivalence and
bioINequivalence. Maybe we will fix our approach
after the problem is redefined.
DR. YU: I do not see actually any
difference. I personally perfectly understand your
concern.
For example, there are two scenarios we
can talk about to this figure, which
figure No. 2.
One of the scenarios is point estimate is
79, the
confidence interval is 78 versus 81. Another
scenario is the point estimate is 60,
confidence
interval is 40 to 80, for example.
Under these two scenarios, from
statistical perspective, we cannot give you
a
definitive answer, however, the first
scenario, do
you know the drug. Certainly, we can definitely
use prior knowledge with respect to
safety and
efficacy of this drug, and we will make a
scientific decision.
The first case, you might have to
think
250
about it, because this drug is still on
the market,
you are perfectly okay, and you will make
a
scientific decision on the second
case. Obviously,
we will not close eyes and say let it go,
definitely not, and chance to be pulled
to the top
of the company is high.
Even the first case, we will
inform the
company, we will discuss with the company
what to
do with that case. Certain case, the probability
to be pulled is higher. I would not say 100
percent definitely as the first case,
that
bioINequivalence case, this case,
certainly we will
take a look at it and discuss it with our
clinicians within FDA, discuss it with
our sponsors
outside of FDA to take a proper action as
issues
occur.
I hope this answers your
question. Thank
you.
DR. KIBBE:
Let me just throw a few random
thoughts in on the table with the
intention of
keeping us all past our flights, so
everybody
misses their flight.
251
First, if a product has already
been
established as bioequivalent, it has been
on the
market for a while, and we have a lot of
confidence
in the product, then, I think to pull the
product
off the market, we have to make a clear
and
distinct argument that the product is
indeed
failing to live up to the criteria that
was
established for it.
It is hard for me to imagine a
product
that got on the market with a
bioequivalency study
where the mean values were, say, 97
percent or 103
percent of the mean, the innovator, and
well within
the confidence interval, and all of a
sudden you
are going to find a lot that is going to
be a
disaster.
However, it is possible. If that is the
case, then, you have before you two
experiments
with opposite results, and in most
laboratories
that I have been involved with, when you
have two
experiments with opposite results,
everybody looks
at each other and says we have got to do
it again,
we can't just leave it like this.
252
So, no matter what you do as an
agency
setting up guideposts for the innovator
to come
forward with a bioINequivalency study, I
think then
the Agency says thank you very much for a
second
piece of information, and we now must
resolve the
discrepancy, not by trumping their study
with your
study, or trumping your study with their
study, but
doing the critical study, which is now
the Agency
should go out in the marketplace and buy
100 of
each of the products off the shelf
somewhere, maybe
St. Louis, maybe Kansas City, somewhere.
I would be afraid that if we
went to
Canada, we would get much higher quality
products,
and we want to stay with the quality
level here,
and do the third study, and then say,
okay, your
company did it with whatever biases that
might have
been involved in the selection to get on
the
market, and your company did it with
whatever
biases or not that you had and to show
that it was
off market, and we did it, and now we
have the
definitive result, and we have to limit
the number
of times you can come forward and do this
with us.
253
So, we have done the third study.
Otherwise, I think it is really one of
those he
trumps you, and you trump him, and if
they come
forward with a bioINequivalency study
that seems to
pass the criteria, whatever you pick, and
I come
and give you a second bio study with the
product
and say, look at that, it really is good,
what do
you do?
Let them trump and trump and
trump, and I
know the CROs are all saying oh, shut up,
let them
do it, because we can do these studies,
you know,
once a month, it would be okay, but it is
not going
to get you the final answer.
There is a couple of other
things that you
might want to keep in the back of your
mind. Drugs
which are non-linear, are easy to
manipulate.
If you do a study with dilantin
at 50
milligrams per patient, and you get a
bioequivalent
result where one is just slightly higher
than the
other, then, get a group of patients and
give them
400 milligrams, and you will run that
mean right
off the table, and what would be an
insignificant
254
difference at a reasonable therapeutic
level would
not be a insignificant difference at an
elevated or
above normal therapeutic level.
There is lots of things we can
do, so that
if we are going to get into this, I would
go with a
nice tight bioINequivalent study, and we can
argue
the value of the statistics, but that
can't be the
end.
That absolutely cannot be the stopping point.
That is just one more piece of
data, as
Jurgen correctly points out. We already have data,
now we have new data, and to resolve it,
we have to
have an impartial arbiter, and the Agency
has to do
its own biostuff.
Gordon wants to disagree with
me. Go
ahead.
DR. MEYER: Actually, I think that is a
good idea, but I think that given the
resources,
that there is no reason that an innovator
can't be
expected to do a study properly. They will get
inspected on the first one anyway, the first
bioINequivalence study, they can be
inspected on
the second one. They are not going to risk their
255
reputation by messing around with the
data, so I
don't think the Agency, I mean that would
be
impractical for the Agency to have to run
out and
do a confirmatory study.
DR. YU: We come back the April 14th
discussion, and I think all of us heard
Gary made
the presentation back in July that our
submissions
this year increased 25 percent, and then
we talk
about risk management, we talk about
where we put
our resource in, and all we are doing for
this
bioINequivalence type is Agency have
defined the
criteria for bioequivalence, we want to
define the
criteria for bioINequivalence to make a
clarification out there.
That's all come back because in
the
literature, scientific literature, people
tend to
conduct a bioequivalence study, now the
confidence
interval is 79 or 126, claim as
bioINequivalence.
We say this is not
scientifically valid.
So that is the whole purpose is wanting
to give a
clear definition with respect to
bioequivalence
versus the bioINequivalence, as well as a
fail to
256
demonstrate bioequivalence and a fail to
demonstrate bioINequivalence.
Then, from here, you are
absolutely
correct, when we see a study like that,
if we
cannot--we are trying to put our
resources in the
NDA reviews. Just in case this happened, cannot be
very clear, there is an ambiguity in the
gray area,
certainly, Agency will have to put the
resource
whether we like it or not. I think we agree.
Thank you.
DR. KIBBE: Go ahead, Gordon.
DR. AMIDON: I am not going to disagree
with you, Art, but the question I have
regarding
the scenario where a product is
bioequivalent and
on the market, and another company comes
in showing
potential bioINequivalencies, has the
product
changed.
If we had a good dissolution
criteria in
evaluation, we would have some underlying
possible
more scientific hypothesis to make rather
than run
out and test another set of products.
So, I think it comes down to
dissolution.
257
DR. KIBBE: Just so you know, bayesian
dissolution.
DR. GLOFF: One quick comment on what
Gordon just said. He said has the product changed,
and my question would be, and if so, which
of the
two products has changed. It is not necessarily
just the generic.
MR. BUEHLER: Well, Lawrence made a good
point, and this is a resource issue for
us. We
haven't gotten that many challenge
studies
recently.
When we do get them, they are usually
out by, like Lawrence said, a little,
127, 79,
something like that, but the letters that
accompany
them are very profound.
They are big public health
issues, they
are always presented as huge issues, and,
of
course, we have to look at these, we have
to
address the issue and resolve the issue
because I
agree with Gordon, we don't want generic
products
out there that are bioINequivalent. That is a
problem for that particular product, it's
a problem
for the entire industry to have products
where the
258
American public can't have confidence in
those
particular products.
We want to know about those
products, and
we want to know when products are truly
bioINequivalent, but we don't want to
have to deal
with all of these studies that come that
are just a
hair out one way or a hair out the other.
We like rules in the Office of
Generic
Drugs, and we are sort of bound by our
rules, and
however, you know, they are criticized by
some
statisticians, we do have our 80 to 125
rule, and
we stick by that very rigidly, and to not
do that
would mean a tremendous creep, you know,
a
tremendous I think lack of confidence in
the
generic process.
Is 79 okay? Well, you know, sure, okay,
79 is okay. But what about 78, what about 77? You
can to down, you can go up, and the next
thing you
know, you have a confidence interval you
can drive
trucks through.
So, that is why we are very
particular
about rules, and in this particular case, you know,
259
this is what we are trying to get for
bioINequivalence, is some kind of a rule
that
companies won't send these in if they
know that we
are not going to deal with them.
If they know that our rule is
it has got
to be this far out or this far over--
DR. KIBBE: Let me ask you a question.
The ones that you have gotten, would they
have
passed this rule that you are putting--
DR. YU: The one we have right now--how
many addition we have? Probably 41 additions
already back and forth, four or five
people
involved with the lawyer and the
scientist
involved.
The case we have, we hope we resolve
very soon, but this case submitted to us
back to
'99, I think, submitted again in 2002,
and you can
see how many resources we are putting in,
more than
two
years already passed.
The issue is this case we are
here, the
confidence interval is--I have got a
lower one--the
top one is 126.
DR. KIBBE: And it's Cmax.
260
DR. YU: It's Cmax.
DR. KIBBE: So Marvin is right.
DR. YU: Oh, it's always Cmax.
DR. KIBBE: Of course, it is. Would this
proposed rule have said upfront that you
haven't
established your case?
DR. YU: Absolutely, yes. You can see
that, two people for two years.
DR. VENITZ: Can I make an observation and
then give my recommendation? First, I agree with
Nozer, we are trying to squeeze a
bayesian problem
into a frequentist scenario, but given
the fact
that we have been hearing this time and
last time
that those are the rules that have been
in effect
for 20-plus years, that we don't find
people dying
on the streets, or they might be working
actually
in terms of providing safe and effective
generic
products, you are stuck with the system
the way it
is right now.
So, I way I tried to approach
it, not
being a statistician, we have a body of
evidence to
suggest that the generic and the
reference product
261
are bioequivalent. That is the reason why it got
approved in the first place.
I assume as part of your
review, you are
going to look for things that might have
changed,
creep in either the product or the
reference
product.
So, now you have a claim being
made that
seems to contradict that, and then, in my
mind, the
burden of proof is with the person or the
organization that files that claim. So, the burden
of proof to me means that it has to be
difficult
for them to overcome what you already
know, so I
personally would go with the toughest off
your
recommendation, and you have excluded my
favorite
one, which is all three of them have to
pass:
Cmax, AUCt, and AUCinfinity have to pass,
because
that is toughest route.
If you can overcome that, then,
I think
you can argue, well, that is about as
much evidence
as you need given the fact that you
already had
pre-existing bioequivalence. Then, you have enough
to overrule.
262
So, I would recommend what you
didn't
recommend, that all three parameters, all
three
metrics have to pass in order to conclude
bioINequivalence.
DR. KIBBE: What do you recommend, Marvin?
DR. MEYER: I might say I am going to
apologize for not being a statistician--I
once said
pharmaceutical scientists all apologize
for not
being statisticians, but statisticians
don't
apologize for anything. I think that probably
applies here.
DR. KIBBE: I will let you comment on that
later, Nozer.
DR. MEYER: Under Strategy I, Lawrence,
prespecify one of the three PK endpoints,
and then
analyze that. Now, if you do a PK study, you are
going to have all three at hand. Why don't you
just do Cmax and then do AUC, and then do
AUCinfinity, and look at the data? What is the
problem with doing that?
DR. YU: Let me explain that first, that
the criteria, when we define,
statistically
263
significant or not, is 5 percent of
criteria. If
below 5 percent, statistically
significant; above,
it is difficult to say.
For you to not prespecify
anything, you
conduct a study, the chance to be wrong
is higher
than 5 percent. In fact, change on one slides, I
believe it could be high, like 14
percent.
DR. MEYER: Isn't that if you use all
three?
DR. YU: If you use any of three. You
don't not prespecify any of them, they
are just
looking for one. For examples, these are the
slides, the error rate could be 14.7
percent. If
that is the case, scientific speaking, is
too high.
Certainly, scientific speaking will look
at a case,
you just present it and make a scientific
decision.
DR. MEYER: But if you pick one and pick
the wrong one, you also have a chance of
being in
error, which isn't in there somewhere.
DR. YU: That is correct. You are
absolutely correct, Marvin. Actually, you
understand very well in my judgment. If you
264
prespecify and if you use AUC, the wrong
one, you
could have a probability power.
The power could be
5 percent to 90 percent, however, you
just said you
have a prior knowledge, so most likely
you have
probably picked the correct one.
Now, let's put the stack back. Indeed,
there is a company out there. Pick the wrong one,
but even the wrong one you pick, for
example, you
pick the Cmax. The case I will talk about is very
theoretical.
You pick up a Cmax, but it end
up an AUC,
you show the confidence interval, for
example, 60
to 79.
This is certainly the case is,
statistically speaking, you do not see,
the error
could be a lot higher, but this does not
mean we
are going to close eyes, the Agency will
not take
an action, probably not, absolutely not.
We certainly will
investigate. As we
said, we look at a formulation change for
both
innovator and the generic side. We look at all the
scenarios. We will have a bunch of people sitting
in the conference for many hours,
probably many
265
meetings, and discuss with many parties
and trying
to make the best decision for the public.
DR. MEYER: Let me just say that if you
pick an area under the curve, you are not
getting
any information about rate, because area
under the
curve can be quite independent of rate.
If you pick Cmax, you are
getting
information about rate and amount, and
also
information about how different your
population is,
and lots of that kind of information,
because Cmax
has bounced all over the place especially
when Cmax
has also moved around Tmax, and the
Tmaxes aren't
constant, so your Cmax from one patient
is going to
be happening at half an hour, and then
the next
patient's Cmax is going to be happening
at one
hour, and you are going to have lots of
fun.
I really think Jurgen is right,
that you
need to establish a criteria that looks
at all of
the three parameters for
bioINequivalency, because
we look at all three parameters for
bioequivalency,
and pre-existing information has to be
trumped
effectively. I still like the idea of doing a
266
third study.
DR. YU: Thank you.
DR. KIBBE: Nozer, do you want to comment
on his statistics?
DR. SINGPURWALLA: Oh, he was absolutely
brilliant. I am disappointed he went into
pharmacy.
DR. KIBBE: Pat, go ahead.
DR. DeLUCA: In your diagram, Lawrence,
you know, to me, there is only one here
that
demonstrates bioequivalence, the others
are not
bioequivalent, so however you turned
them, the
other four are not bioequivalent.
But the question I would ask,
in
determining that the product was
bioequivalent, you
need Cmax and area under the curve.
DR. YU: Correct.
DR. DeLUCA: If they came in with just
Cmax or just area under the curve, you
wouldn't
have approved that as being
bioequivalent, is that
right?
DR. YU: Absolutely.
267
DR. DeLUCA: So, if they just had one that
wouldn't be, you wouldn't approve it if
they just
had one. So, I can't see why, then, if
you are
looking at a product that is
bioINequivalent, why
you can't just use Cmax.
You can just use one of them to
me,
because they had a pass, both of them, at
the
start, so if they didn't have both of
them, they
wouldn't have passed bioequivalence, so
why isn't
one enough? Why isn't just Cmax enough to show
bioINequivalence?
DR. KIBBE: Paul.
DR. FACKLER: Just one quick point. The
generic has to pass Cmax, AUC zero to t,
and AUC
zero to infinity under fasting
conditions, under
fed conditions, and for capsule beaded
products,
under sprinkle conditions.
So, for some products, it is
nine
parameters that need to pass, for others,
it is
six, and for a relatively small group of
products,
it is three. I just put that out there for what it
is worth.
268
DR. KIBBE: Marvin.
DR. MEYER: You asked for a
recommendation. In the spirit of harmonization, I
would suggest that Lawrence's figure
there is
perfect, that under standard conditions,
no
monkeying with the confidence limits, not
86, not
three different ones, keep our 80 to 125,
90
percent, and declare the two bottom ones
bioINequivalent, and therefore bad, and
therefore
need investigation, and the rest of them
are all
either unknown or bioequivalent.
DR. YU: Thank you.
That is actually what
we are recommending.
DR. KIBBE: Anybody else want to jump in
on the consensus recommendation wagon?
What do you think, Carol?
DR. GLOFF: I have a question for Marvin.
Do you mean that all three parameters
need to fall
outside?
DR. MEYER:
No, this could be any of the
three parameters, any one of the three
exhibiting
either of the two bottom ones,
bioINequivalence.
269
DR. SINGPURWALLA: I would like to make a
comment from mathematics. To disprove a theorem,
all you need is one counter example, so
if you want
to show bioINequivalence, all you need is
one
violation. If you want to show bioequivalence,
then, you may have to go and do everything
else.
Does that rhyme well with your
view,
Jurgen?
DR. VENITZ: I am not sure because I still
think that the hurdles that you have to
overcome to
get an approved generic on the market,
not just
looking at Cmax, I mean the other things
as you
have heard, and it may just not be a
single Cmax,
it may be other things.
Given the fact that, as you
have heard me
talk about earlier this year, the 80 to
125 percent
is really an arbitrary goalpost. I do believe that
the burden of proof should be high for
somebody to
get this reversed, to get an approval
reversed.
DR. SINGPURWALLA: But the burden of this
proof need not be so high.
DR. VENITZ: Think about what
270
bioequivalence means. It basically means you don't
have enough evidence to reject a null
hypothesis,
to use statistical lingo. You are basically trying
to prove the impossible. You can never prove that
something is equal. So, you are just bounding.
You are saying, in my mind, I put
arbitrary bounds
on, and say, well, as long as it fits
those bounds,
we consider it to be bioequivalent.
So, to disprove that, I think
you have to
disprove it on all the three metrics, the
metrics
that you used in the first place, to get
approval.
DR. KIBBE: The argument that you are
making is that because the criteria says
that all
three of these parameters have to meet
the criteria
to get approval, doesn't necessarily mean
that we
shouldn't require all three to meet the
criteria to
get unapproval.
By saying okay, in the original
submission, all they had to do is fail
one to not
get approval, that's fine, but now we
have an
already approved product, and we are
doing a test
to show that it is not equivalent, so I
would like
271
to see it demonstrated that it is not
equivalent on
the same three parameters, and if it
can't do it on
all three, then, it has failed that test,
just like
it would have failed originally to get
approval by
failing one of the three, and that is my
argument.
DR. SINGPURWALLA: Off the record. To
really look into this issue, it is a much
more
serious issue that what meets the
eye. There is a
rule that has been set up, and you have
to live
with that rule, I agree with you, but
what is to
stop the FDA from looking to the future
and
changing the rules?
DR. HUSSAIN: I think that point is well
taken, and, Helen, actually that is
exactly what my
new instructions were from her, so we
will take
this further into discussion, and so
forth.
I think this was very valuable,
and I am
not fully sure exactly that we have come
to a
conclusion on this yet.
DR. KIBBE: Jurgen and I have come to a
conclusion. Marvin's conclusion is slightly
variant, but not too much.
272
DR. MEYER: But I am retiring.
DR. YU: I guess that we are back to the
April 14th situation where there is 1
versus 3.
DR. HUSSAIN: Lawrence, I think it is time
to stop the discussion.
DR. KIBBE: I think the Agency has to step
up to the plate. We have given you the best advice
we can.
DR. YU:
Okay. Thank you.
DR. KIBBE: I think it is appropriate at
this stage, since my schedule says we are
taking a
break, to take a break.
We are breaking 15 minutes
early. We will
give you 10 minutes. We expect you back in your
seat at three minutes to 3:00, and at 3
o'clock, we
will have our discussion about the
locally acting
gastrointestinal materials, and we will
wrap that
up in short order because Gordon has the
exact
answer we need right here.
[Recess.]
DR. KIBBE: Ladies and gentlemen, the
clock on the wall says it is three
minutes to 3:00,
273
and as it is my tradition, I will remind
you to
gather and begin.
We have one more topic area.
Lawrence, are you going to set
the topic
up?
You have got three minutes.
DR. YU: Yes, I will.
Actually, I can
finish within two minutes.
Bioequivalence Testing for
Locally Acting
Gastrointestinal Drugs
Topic Introduction
DR. YU: The bioequivalence testing for
locally acting drugs was I think
presented to you,
and I saw the comments. Well, today, we are going
to discuss the real issue of
bioequivalence testing
for locally acting drugs. I will introduce this
topic, and Gordon from the University of
Michigan
will give the talk on Scientific
Principles, and
Robert Lionberger from the Office of
Generic Drugs
will give you specific examples.
Again, bioequivalence is
defined as the
absence in the rate and the extent of
drug
absorption.
274
As I said yesterday, the
pharmacokinetic
measure for bioequivalence method for
systematic
drugs is well understood, well used. We have
pulled many, many products.
The issue remains for locally
acting
gastro and GI drugs. The part of reason for that,
because the plasma concentrations may not
be
relevant to locally delivery
bioequivalence, for
example, a topical, nasal, inhalation,
which I
presented to you yesterday.
The point we want to make sure
that the
dissolution controls the delivery to the
site of
action, whether it's the jejunum, jejunal
ileum, or
colon.
The drug concentrations in plasma are
downstream from the site of action,
unlike for
systematic drugs, the drug ending
systematic first,
then, get the side effects action, for
example,
heart and liver, and so on, the heart and
the
brain.
For GI acting drugs, there is
no alternate
absorption path because already they have
to be
absorbed from the gastrointestinal
tract. So,
275
bioequivalence approach the Agency has
used, for
example, the clinical study of
vancomycin,
pharmacokinetics study is sulfasalazine,
the in
vitro study is cholestrylamine.
What we want is want to develop
a
scientific basis for the choice of BE
method,
bioequivalence method, which we need your
input on
role of pharmacokinetic studies, role for
in vitro
dissolution studies, role of the clinical
studies.
With that short introduction,
Mr.
Chairman, I finished it within two
minutes, I turn
the podium to Gordie.
DR. KIBBE: Thank you, Lawrence.
Dr. Amidon.
Scientific Principles
DR. AMIDON: Thank you.
I am glad we are
recovered from that discussion of
statistics. I
know I get glassy-eyed. I think Qian Li did a
great job, and then we turned it into
chaos, but
that is our job, I guess.
MR. CLARK: Chaos theory is our goal.
DR. AMIDON: What you received in the
276
handout was the unedited version of my
presentation, because it was done before
I knew
what Lawrence and what Rob were doing, so
I am
going to skip a lot of the slides that I
have
because points are already being made,
and talk
about the highlights, the essentials of
my point,
and I will give you the executive summary
right
now.
First, bioequivalence is the
question of
dissolution. What else is it? The same drug in
different products. Once the drug is absorbed, it
is the same except in the unlikely
scenario, there
is maybe a competitive metabolism
inhibitor or an
excipient that might alter
permeability. Evidence
for that is limited in vivo in humans.
So, bioequivalence is a
question of
dissolution. That is where the science needs to be
done.
So, the bottom line for GI drugs, for all
drugs, is that we should put more
emphasis on the
science of dissolution and what I think
of as a
bioequivalence dissolution test. So, that is going
to be my bottom line, big picture
conclusion.
277
The subconclusion for GI drugs
is that I
think what we need is a bioequivalence
dissolution
test with some type of in vivo test,
perhaps not a
confidence interval test, maybe a point
estimate
and an interval requirement, say, between
90 and
110, so that we don't have this confidence
interval
issue.
You could argue do we need the
in vivo
test.
I think in some cases, we do not, and
probably we need to go drug by drug for
locally
acting drugs, and Rob will talk about
specific drug
examples.
One of the issues in setting up
a policy
issue is try to be very general, and you
get into
trouble because some things aren't
generalizable
very easily. So, I think we will have to regulate
GI drugs, of which there may be a half a
dozen that
are very important, more on a
drug-by-drug basis,
or maybe classify them, I don't know.
So, that is the bottom
line. Dissolution
is what we should be looking at, and a dissolution
test plus an in vivo test, perhaps a
point estimate
278
would be enough.
I am going to skip over most of
these
slides, because the points are already
being made.
The one point that I will make,
that I
make all the time, and I think is
generally
accepted, at least no one has argued, is
that I
think bioequivalence is maybe the single
most
important regulatory standard for
virtually all
products on the market today.
That is, products on the market
today, are
on the market because of either proven or
assumed
bioequivalence. If not, what could we say about
the clinical? We have to make that connection, the
connection between the product in the
bottle and
the label is bioequivalence.
Yes, of course, we have to have
the
potency, the impurities, and we have to
have the
standards, but bioequivalence, so this is
I think
one of the most important issues in drug
regulatory
standards in the world today, because it
pertains
to all products. My interest, of course, is in
oral products.
279
There is some caveat in the
Orange Book,
just to point out kind of the legal
basis. If you
look at where I think the most up-to-date
definition of bioavailability and
bioequivalence,
it is in the preface to the Orange Book.
It has been revised
periodically over the
years, I think no one has noticed it,
because they
slip kind of changes into the Orange
Book, it just
comes out, and life goes on. Right?
Is that what
you do?
At any rate, for locally acting
drugs, it
says, "Where the above methods are
not applicable,
e.g., for drug products that are not
intended to be
absorbed into the bloodstream, other in
vivo or in
vitro test methods to demonstrate
bioequivalence
may be appropriate."
That is where we are at here
with GI
locally acting drugs.
Again, I am going to skip through
these.
You have seen this. Rob is going to use this, so I
am going to skip it, the disconnect for
locally
acting drugs.
280
Now, classically, we do Cmax,
AUC, of
course, and AUCt, and we have confidence
interval
test, but if the levels in blood are very
low, we
have a problem, so we have a practical
problem.
So, I am going to come back to
the
paradigm for bioequivalence. I think of the
paradigm of bioequivalence today as being
the
following, starting at the top. Similar plasma
levels, similar pharmacodynamics, similar
efficacy
to the label. I mean that is the implication,
similar pharmacodynamics.
Then, similar in vivo
dissolution, similar
plasma levels. For oral products, I think maybe
for all products, but certainly for oral
products,
similar in vivo dissolution. When we think about
the physiology of oral absorption, the
drug
dissolves and is spread along the
gastrointestinal
tract and absorbed. We think of absorption as into
the intestinal mucosal cell.
Subsequently, systemic
availability is
later, and there is more stuff, drug
stuff, you
know, liver in between. That is part of
the
281
complication, we are doing bioequivalence
based on
plasma levels, which is systemic
availability. We
are doing a systemic availability test
which is
distant for GI drugs from the site of
action.
So, similar in vivo
dissolution, similar
plasma levels. So, then, where the science is
today is in vitro dissolution. We need to be more
rigorous in how we do dissolution when we
use it
for bioequivalence materials.
We think that there is no
reason why we
cannot establish better dissolution
methodologies
that reflect the in vivo dissolution
process. I
think that would have a number of
implications
including accelerating the drug
development
process, because when you are making a
product and
then testing it in humans, you would like
to have a
good idea you are going to succeed.
In order to do that, you have
to have a
dissolution test that reflects what is
going on in
vivo.
So, I think better dissolution can be a big
step in advancing and accelerating drug
development.
282
So, that means doing
something. Often now
today, we have what we call biorelevant
dissolution
media or biorelevant dissolution.
I think we need
to use that term carefully because, you
know, to
take some natural surfactants and a
little bit of
phospholipid and put it in water and
shake, you
either have a drug delivery company or
you call it
biorelevant dissolution media, but what
is it?
There is no evidence that it is relevant
to the in
vivo dissolution process.
So, we need to establish that
connection
between the in vitro dissolution
methodology and
the in vivo dissolution process, and I
think that
is where there is a big gap in our
knowledge today,
not just for GI drugs, for all drugs.
So, my point here is broader
than just GI
drugs, but similarly, if we had
confidence in an in
vitro dissolution test, that is all we
need to do.
That is all we need to do. So, we should be
focusing the science on that dissolution
test.
I am preaching too much here,
so I am
going to skip most of my slides, but I
have to show
283
at least an equation. I noticed that on my badge,
they had MA. It took me a while to remember that I
had a Masters in mathematics, and I was
impressed.
The FDA is so thorough in their
investigation of my background, you know,
every
year I have to fill out all these
conflict of
interest things. They actually put MA. That is
the first time that has ever happened, so
I have to
compliment the FDA and their thoroughness
in
investigating my background. But I did pass, and I
am here.
Anyway, this is the equation of
bioequivalence, but I am not going to
talk about
it.
We talk more about the
physiology of
gastrointestinal tract and product
disintegration,
dissolution, and spreading along the
gastrointestinal tract. That is where the
investigation is. That is where we need to do more
investigation.
Again, I am skipping most of
the slides.
So, to kind of come to the
conclusion on
284
the bioequivalence for locally acting
drugs. I
mean obviously, plasma level is
downstream from the
site of clinical effect, which is
local. The local
site of action is in the GI tract.
So, dissolution and transit in vivo
controls the presentation of drug to the
site of
action.
So, this is where plasma levels are
probably less good than a good
dissolution test.
Now, we could, with intubation,
measure
concentrations along at least part of the
gastrointestinal tract, not easy to do,
and, yes,
you have got tubes in, so it is not
normal. You
could argue is that feasible.
Now, I think most locally GI
drugs are low
permeability, but I now would want to
caveat that,
I am not sure that is the case. There is certainly
low systemic availability in general, and
that is
probably more the issue, because that
makes the in
vivo test plasma levels more difficult to
measure.
So, for locally acting drugs,
in vivo
dissolution is the key determinant. So, for the in
vitro dissolution test, we should cover
the range
285
of in vivo variables.
So, here is the
hypothesis. If a product
dissolves, two products could be the same
manufacturer, but they just did some
reformulation.
If those two products dissolve at the
same rate
under all in vivo conditions, such as pH,
6.5, 7,
7.5, maybe a couple more if you want to
be really
rigorous.
That means that the two
products will
perform the same regardless of what the
pH profile
is in an individual subject. That is what we have
to ensure, and I think we can do that
better with
an in vitro test than an in vivo test.
Now, might want to debate do we
want to do
6.5, 6.75, you know, but we need to first
accept in
principle that a dissolution test is a
key crucial,
I would say an essential component of
setting a
bioequivalence criteria for a GI drug,
because I
think that is the case.
Now, one place we could start
is that
biowaivers for Class I drug, if a GI drug
is a
Class I drug, high solubility, high
permeability,
286
and rapidly dissolving, it is all over,
it doesn't
matter.
In this case, the GI drug would
be low
permeability or certainly low systemic
availability, it may or may not be low
permeability. So, I think that is the equivalent
to extending biowaivers to Class III
drugs--I am
sorry--that is Class I drugs.
What we are talking about for
low
solubility drugs or particularly low
permeability
drugs would be extending biowaivers to
Class III
drugs, which has been proposed. I don't know if
there is any examples, and maybe Rob will
talk more
about that.
I think for pH, we want to look
at
dissolution as a function of pH. The one product
that I am going to show just some data
on, I think
Rob is going to show the same data, so I
will be
quick, is mesalamine. It is in enteric coated
dosage form.
The question would be do we
need
surfactants or not, and that would depend
on the
287
drug, because if the drug is poorly
soluble, then,
we get into the spiro-relevant [ph]
dissolution
media, and that is a bigger question.
So, I think we should require a
dissolution test for bioequivalence in
the
bioequivalence criteria for acceptance
criteria for
GI drugs, that we need to consider the pH
and time
that the drug will spend in the stomach
and in the
gastrointestinal tract. I can propose those if you
want to discuss them.
I think what is more important
is
accepting or at least advising and
recommending to
the FDA that in vitro dissolution testing
should be
part of considering the bioequivalence
requirement
or testing for a local GI drug.
You would have to use the
similarity, you
know, the 10 percent difference or F2
comparison
for dissolution profiles.
The dissolution test actually is a
difficult criteria, I think. Mesalamine, I will
show just a few slides on that, but
mesalamine is
an enteric-coated, local acting drug, and
are some
288
dissolution profiles done by Jennifer
Dressman, now
at Frankfort, published in European
Journal of
Pharmaceutics and Biopharmaceutics, but
here are
different products and simulated gastric
fluid.
Here is a pH 6.8. You see, none of the
products would be similar. They all dissolve at a
different rate, and that is a surprise to
development scientists, and these are
different
products.
I don't know if they were approved as
bioequivalent or not actually, but I do
not think
they would be bioequivalent in the
gastrointestinal
tract even if they were bioequivalent in
vivo.
There are some other profiles
again
showing that they are quite
different. If you
increase the pH to 7.8, more of them
become similar
because they all dissolve rapidly above
the pH of
the enteric coating. So, if you do a dissolution
at a high enough pH, you can make things
look
mostly similar, but at the critical pH
where
dissolution is occurring, they would be
different.
So, I think that the pH
dissolution
profile requiring similar dissolution at,
let's
289
say, in this case we are looking at I
think like a
pH of 6.57, 7.0, 7.5, that pH range. We can debate
whether you should do pH 6.0 would be
critical.
I am going to skip these
because these
questions are already up.
So, what I want to propose that
the
committee consider and perhaps recommend
to the
FDA, I am not sure, I guess we are going
to go
through a list of questions that Rob is
going to
discuss, but that we do require in vitro
dissolution as part of our bioequivalence
testing
for drugs that are locally acting, and
that then
the in vivo test, do we need an in vivo
test for
safety purposes, for safety assurance
purposes, and
do we need a confidence interval test for
in vivo.
I think that may be on a
drug-by-drug
basis.
You may not want to try and make a decision
for all locally GI drugs. I think depends on the
pharmacology and metabolism of the drug.
But I can say that if had a
rigorous
enough in vitro dissolution test, in vivo
testing
would not be required.
290
That's it. Thank you.
Do you want to have questions
now?
DR. KIBBE: Shall we do that, because I
think Marvin has a question and so do
I. Go ahead.
DR. MEYER: Real quick, Gordie.
Do you think that F2 is an adequate
parameter to use in making a
bioequivalence
decision?
DR. AMIDON: The answer is I don't know,
Marvin.
I have suggested this to a couple of
people, that we need to evaluate that,
and whether
F2 or 50, where did that come from, and
is it
enough, and the answer is I don't know.
I think that the statistics of
dissolution
and the dissolution variability that you
could
allow, that would keep you within the
bioequivalence 80 to 125. Now, that is not so easy
to answer, and it is going to depend on
drug
properties, but I agree, that the F2 or
50 needs
more investigation.
DR. KIBBE: Anybody else?
Go ahead, Judy.
DR. BOEHLERT: I finally have a question.
291
By making this suggestion that
the in
vitro dissolution is a factor, does that
presuppose
that the clinical efficacy of this drug
only occurs
in a very narrow pH range, so that this
pH
difference you see on dissolution is
meaningful,
because they could be clinically
equivalent or have
the same action, and have different
profiles at
different ph's because where the drug
acts is
across the GI tract, and not just one
location.
DR. AMIDON: They could be clinically
equivalent with a different dissolution
profile,
but you would have to prove it to
me. No one is
going to do that.
DR. BOEHLERT: And how would you do that?
DR. AMIDON: But my answer, I think more
to the point, I think, Judy, is the pH
profile
changes through the GI tract, stomach,
duodenum,
jejunum, goes 5.5, 6.0, 6.5, and it
varies from
subject to subject, and in a fasted/fed
state,
during the different Phase I, II, III, of
the
fasted/ state.
So, what you want to do is
ensure the two
292
products will dissolve under any of the
pH
conditions that we would see. So, if they dissolve
the same, let's say at pH 6.5, 7.0, 7.5,
you could
say, well, maybe we should 6.0, maybe
5.5, maybe we
should pretreat for 15 minutes in 10th
normal HCL
for a while, pretreat for 2 hours, maybe
we should
pretreat at pH 4.0, because that is more
like the
average pH in the stomach with food.
So, I think that is more of an
issue of
the specific test, and that is going to
be a more
elaborate test than our typical so-called
dissolution test.
DR. BOEHLERT: Actually, I guess my only
concern was if, indeed, the drugs were
equivalent
at pH 7.8 where it all dissolved at the
same rate,
is it really meaningful that it was
different at
lower pH's.
DR. AMIDON: I would argue that it is,
because the pH in the intestine in humans
is more
like around 6.5 to 7.0, and it goes down
to 5.0 or
5.5 in the duodenum where you have got
the mixing
of gastric acid and pancreatic and bile,
so that I
293
would say 7.8 is actually not relevant.
I wouldn't do that. The highest I would
go would be about 7.5.
DR. KIBBE: Paul.
DR. FACKLER: If I can just make a couple
of points. First, I agree about the dissolution
and how inappropriate the dissolution we
do today
is to certainly the way orally absorbed
drugs are
taken.
I can't remember the last time I saw
somebody drink 900 milliliters of water
with their
tablets.
DR. AMIDON: Or even 250 ml.
DR. FACKLER: Or even the 240 or 250 that
we use in the clinics. But let me ask a question.
For systemic drugs, we look at the plasma
concentrations of the compartment just
prior to the
site of effect, and for these drugs, we
might look
at the plasma compartment as adjacent to
the site
of effect just after it, and so in either
case, it
might be a good surrogate measurement for
the
amount of drug at the site of action in
our
traditional case where we have been doing
it for 20
294
years, it is just prior to, but for these
drugs, we
might consider it just post, and maybe as
a
surrogate marker for the amount of drug
that was at
the site of action.
DR. AMIDON: I would say the following. I
think that is a good question, Paul, and
I thought
some about that. Two things I would say. One is
the drug as it spread along the
gastrointestinal
tract, depending on how it is released
may be
absorbed in different segments, so the
drug might
actually be presented to different sites
from
different formulations, and still meet a
Cmax and
AUC criteria.
Now, if you were to add in Tmax
and/or
some absorption rate measurement in
comparison,
then, I would agree with you, it would be
equivalent, but that is complicated. I mean the
FDA has looked into measures of
absorption rate
other than Cmax and concluded that there
isn't
really any good measurement.
So, the answer is if you measured
rate and
Tmax and had criteria on that, I would
say then
295
plasma would be just the same basically,
but if the
drug is very highly metabolized, so the
systemic
availability is low, and you are
measuring parent,
and assuming there is no problems with
metabolite,
which you can't assume, then, you have
got the
variability issues on the plasma site, so
the
plasma site could be a much harder test.
DR. KIBBE: Jurgen.
DR. VENITZ: I think in general I buy the
idea that in vitro dissolution could
predict in
vivo dissolution of those products, but
you
mentioned kind of in passing excipient effects on
permeability or metabolism. What about food
effects?
In other words, food constituents would
considerably at least have the same
effects on
either permeability and/or metabolism,
which may
affect locally what is going on.
DR. AMIDON: For bioavailability, I would
agree, Jurgen, it would have a big effect
obviously, but for bioequivalence, I
would say the
effects would tend to be washed out by
the dilution
of the food of any excipient effect.
296
So, I think when we are
comparing two
products, the same drug, I would say it
is a less
important issue.
DR. KIBBE: Anybody have something because
I have got a whole bunch?
DR. MORRIS: I have just one quick one.
Gordon, were you assuming that this would
be done
in the normal dissolution apparatus or is
that an
open question?
DR. AMIDON: That is an open question in
my mind, yes.
DR. MORRIS: Because I think there is a
fair amount of concern, hydrodynamic at
the very
least, with the current apparati. I just wondered
if you are not restricting it to that.
DR. AMIDON: I guess my position is
changing dissolution apparatus should be
done with
great care and with good justification.
DR. MORRIS: Oh, absolutely.
DR. AMIDON: But I believe that a
bioequivalence methodology needs to look
at
reflecting in vivo processes, and we
should start
297
with that.
DR. MORRIS: I agree with your premise.
DR. KIBBE: I follow up on Ken and go down
a road with a lot of variables in
it. I love in
vitro tests if I have control of all the
variables,
and when I start losing control of the
variables,
then, I start to get worried about the
test.
I can imagine two products that
have
slightly different excipient compositions
who
appear, in a dissolution apparatus, to
dissolve
equivalently, but that they aren't
presenting the
same amount of drug to the surface of the
membrane
for one reason or another.
There might be an excipient in
there that
forms cells that trap some of the drug,
there might
be an excipient that binds the drug, but
in a
dissolution apparatus, the excipient is
small
enough so that it goes into solution, but
then it
doesn't allow it to release.
I can imagine interactions
between food
substances that are different for this
dosage form
versus that dosage form. I mean I really would
298
like to go back to the if you could show
me that
both formulations have the same inert or
inactive
or excipient ingredients and in the same
basic
dosage form, then, I am really happy with
dissolution studies as a mimicker because
we are
really looking at the levels of drug in
that lumen,
that then gets presented to the surface
where it is
supposed to have its effect.
The other thing that I wonder
about is if
some of these excipients are permeation
enhancers,
than one drug, they both would dissolve
the same in
the in vitro dissolution apparatus, but
the one
with the permeation enhancer would start
to leave
the site where you want it and end up in
the blood
supply where you don't want it, and you
really have
to link that to some marker at the back
end to see
how fast it is leaving where I want it to
be.
If one gets into the body in
this case
better, that's worse.
DR. AMIDON: It could be, at least in the
systemic circulation.
DR. KIBBE: So, the questions that I come
299
back to is what is the dissolution
apparatus, what
are the criteria that we have to put on
the product
to narrow it down, so that we know that
the
dissolution apparatus can tell us if the
two are
behaving the same before we put it in a
person, and
then when we do that, aren't we better
off still
getting minimalist levels in the plasma
just to
look for the odd chance that one of them
is
permeating better than the other, so that
we can
either prevent levels going up, so that
we might
have toxicity in one or the other, or
prevent
levels being too low, so that we have
some
secondary measure.
Now, if we are going to do
that, which one
does a better job of actually measuring
the drug at
what I would call the biophase where the
drug is
having its real action, and I really
think they
both measure it incompletely.
DR. AMIDON: One side and the other. We
have got it sandwiched in between, right?
DR. KIBBE: Right.
So, I would argue that
you have to do both somehow.
300
DR. AMIDON: Maybe we need an intermediate
step here to get more experience. So, I could see
where we might require some type of
pharmacokinetic
measure, plasma levels, as well as
dissolution,
both.
That is, I am recommending both.
Now, I believe that when we
know enough,
and maybe for some, maybe most, but
probably not
all drug products, dissolution would be
enough, but
we are not there yet.
DR. KIBBE: I could support that once I
start narrowing down my variables.
DR. AMIDON: So, we get into the
discussions of dissolution, but I mean
and I could
throw things out, but it is going to take
more than
me to kind of evaluate what might be a
good
criteria here and dissolution
methodology. I think
what is important at this stage for the
committee
is to say, yes, we think this is the
right path to
go down and tell the FDA to figure it
out, like
Congress does, you know, you go figure
out what to
do with bioavailability.
DR. KIBBE: Ken wants to say something
301
else.
DR. MORRIS: I have two caveats I guess,
the first of which is that in that
dissolution is a
manifestation of the product itself, as
we have
discussed, I like that because it is
looking at
changes in the product.
But aside from the apparatus
issues and
the tactical issues, there is the
statistical
sampling issue that we still face with
our normal
dissolution testing that I think is
probably only
addressed by the consistency that we had
hope to
achieve.
DR. AMIDON: Statistics will be here, too.
DR. MORRIS: Yes, sir.
I was going to
insult Nozer, but he has left already.
DR. KIBBE: Marvin.
DR. MEYER: Gordie, what do you with the
situation where your in vitro is too
discriminating? You have three different products
and clinically they all work, maybe to
different
degrees, but close enough, and yet your
dissolution
says there is a 20, 30 percent
difference.
302
DR. AMIDON: Good question, Marvin. I
have tried when I talk about dissolution
and talk
to the dissolution people at the FDA, say
do not
use the word discriminating, because that
is not
what your job is. Your job is to ensure in vivo
bioavailability.
If the manufacturer wants to be
discriminating because of his quality
standards,
that is fine, the manufacturer can do
what he wants
as long as he meets the standards that
ensure
bioequivalence.
So, I don't think the FDA
should be
regulating on the basis of discriminating
dissolution tests. I think they should be
regulating on the basis of a test that
will ensure
in vivo bioequivalence. I don't know if anyone
agrees with me, maybe you do, but I don't
like the
term discriminating because you can
always get
discrimination.
DR. KIBBE: Ajaz wants to say something.
DR. HUSSAIN: Gordon, I think you put
something right on, and that is a
challenge. The
303
last several months we have been
struggling with
this in a sense. I did share with you my
presentation to the USP meeting, and so
forth.
Since we do not have a good handle on
variability
because of the calibration issues, and so
forth, I
think the tendency has been at the Agency
to say,
all right, how do you minimize that.
I want a big difference, so I feel
confidence, and I think the time has
changed to say
what is the intended use, and so forth,
so I think
the point you made, I don't think we have
consensus
throughout right now, but it is a very
important
point, and I think we have to move in
that
direction and sort of discuss and debate
that.
DR. KOCH: I just had a quick question
that goes back to some involvement I had
maybe over
20 years ago where there was a dog model
set up,
and I imagine it could be most animals
where they
actually had a trapdoor on his stomach,
where you
were able to, in this case, primarily
watch
disintegration and subsequent effects by
monitoring
plasma levels on dissolution.
304
But is there any development
that could be
called pseudo in vivo to check for
absorption, not
to do what we are talking about here, but
sometimes
there is a question somewhere between the
in vitro
and the in vivo?
DR. AMIDON: I think my answer to that
question is no, there is no way to make a
step
forward dosage forms. I would say we regularly
test in dogs, recognizing, though, that
the average
pH in the upper small intestine of dogs
is about 1
pH unit higher than humans, so therefore,
enteric-coated products are not going to
be
reflected by the dog.
We do controlled release in the
dog, but
we only look at the first 6, 8 hours at
most,
because of the shorter transit time, and
you have
got to just discard any data after that.
The dog, as I understand, is
not
basogastric acid secreter, so the pH in
the stomach
is much more variable. So, the answer is I don't
think it will do the job for this, and
any smaller
animal you can't do because you can't
scale the
305
dosage form down very well, or can't at
all maybe,
I guess.
I mean make an enteric-coated tablet for
the rat and scale it up to humans, you
have got
bioequivalence issues again.
DR. KIBBE: The pig is really good.
Meryl?
DR. KAROL: Am I assuming correctly that
the question really is are we going to
test for
bioequivalence locally, that includes the
pulmonary
tract, as well as the GI tract?
DR. AMIDON: Me, no, I am not ready to go
that step in terms of topical or
inhalation
although I could make some case that the
principles
apply.
The difficulty with topical, for example,
is that you are putting your formulation
in direct
contact with the absorbing surface, so
the
formulation will affect, let's say, the
permeability of the skin.
The big saving grace about the
gastrointestinal tract is there is this
big
dilution in the stomach, and that is why
I think
the excipient effects are small. I mean we have
306
got this big dilution in the stomach, so
all the
excipient effects are diluted out.
DR. KIBBE: We can talk about that.
DR. AMIDON: Inhalation, also, there is
the physics of the dosage form particle
size and
deposition along the GI tract, so I am
not an
expert at that, so I am not very
knowledgeable
about that. This is focused on GI.
DR. KIBBE: Ajaz.
DR. HUSSAIN: I think the other aspect is
if you recall the discussion Rob
Lionberger had
presented on topical skin products, there
is a
fundamental principle that is evolving,
it is
quality by design.
What that means is in a sense,
comparing
formulations at anything worth critical
and then
trying to relate that to that, and I
think the
aspects of excipient similarity, and so
forth, a
lot of this formulation information can
be
supportive information that can give you
all this,
and so forth.
So, I don't want to discount that
in the
307
sense I think Rob Lionberger's
presentation on the
topical decision tree had those elements,
so in
many ways, I think the proposal for
looking at an
in vivo relevant bioequivalence test
using in vitro
method is a confirmation of all that
similarity or
all the design that sort of comes
through.
So, keep that in mind as you
think about
it.
DR. KIBBE: Ajaz, that was clearly the
point we get to. Once we get an understanding of
the formulation itself, and can look
critically at
what is in there besides the active,
then, we are
much more confident in what Gordon is
proposing as
a way of measuring what is happening
because if we
know there is nothing in there that has a
habit of
doing the things that might disrupt it,
then, we
are done.
I still think that we are going
to find
early on it is going to be very
comforting, if you
will, to get some blood levels just to
make sure
that there is something that we haven't
expected
that is happening, but I think we need to
go in the
308
direction of dissolution testing. I think that is
a wonderful way to go in a situation.
Anybody else?
We have another presentation,
right?
DR. AMIDON: Right.
Thank you.
DR. KIBBE: Thank you.
Dr. Lionberger.
Regulatory Implications and
Case Studies
DR. LIONBERGER: What I am going to talk
about is just give some examples that are
illustrates of how we apply some of these
scientific principles to several
different products
with the intention of sort of spurring
discussion
although we have already had some very good
discussion.
The first scientific issue is
dissolution,
we are not really going to talk about
because you
have had a very good discussion about
dissolution.
The second scientific issue
that I want
you
to keep in mind through my presentation is the
issue of sort of how we should interpret
pharmacokinetic measurements that we make
on the GI
309
acting drugs.
Certainly, they are related to safety, but
I also want to just indicate that, you
know, you
often hear that the peak pharmacokinetics
of
locally acting drugs aren't correlated
with
therapeutic effect, so I want to sort of
focus your
attention on the sort of last point here,
on how
the pharmacokinetics of the GI acting
drug is
related to formulation performance, and
that will
sort of lead into some of the discussion
that we
would like to have on how we should use
pharmacokinetic data in evaluating
bioequivalence.
So, the first example that I
want to talk
about today is for the drug
mesalamine. This is an
anti-inflammatory drug mainly targeting the
colon.
It turns out that it is actually pretty
rapidly
absorbed from the intestine and this drug
can be
measured in the plasma. There is also some
extensive metabolism pre-systemic
circulation, as
well.
The one thing that is sort of interesting
about this drug is sort of a case study,
is that
310
there is a wide variety of formulations
of this
drug that are currently on the market, so
you can
do a lot of sort of comparisons to see
which
different types of tests can actually
distinguish
between these different formulations.
The sort of key scientific
issue that is
driving these formulations is you want
the drug to
target basically the colon, but it is
rapidly
absorbed, so the formulation technology
is either
pro-drugs or some sort of delayed release
enteric
coating are designed to sort of keep the
drug from
being absorbed until it reaches the
target.
So, this sort of raises the
issue of
targeting different areas of the
gastrointestinal
tract and some of the issues that that
might raise.
The first product is
sulfasalazine. This
is the oldest mesalamine product. It is the third
molecule down in the chemical structures,
and it is
a pro-drug that consists of mesalamine
moiety and
then the other moiety, sulfapyridine.
For this case, the mesalamine acts
locally.
The other moiety of the product is
311
actually rapidly and quickly
absorbed. So, this
drug is old enough that OGD has actually
approved
ANDAs for this product, and the basis for
the
bioequivalence determination in this case
was
pharmacokinetics, but it was the
pharmacokinetics
of the inactive part of the pro-drug,
primarily
because it's rapidly absorbed and it has
much lower
variability than the active moiety
itself.
Also, for this drug, the
sulfapyridine
itself has pharmacological activity, and
there are
also its systemic exposures is highly
related to
some of the safety issues with this
product.
So, for this product, this
moiety was used
primarily because, as we will see later,
there is a
high variability associated with the
pharmacokinetics of the active ingredient
itself.
So, that is sort of just one
example, and
Lawrence sort of pointed these out, that
sort of in
the past, FDA, for these GI acting drugs,
has used
sort of a wide variety of different ways
to
evaluate bioequivalence, and we would
like to sort
of put together a sort of more
fundamental
312
scientific framework on sort of when we
should use
which aspect.
So, a second mesalamine is the
Pentasa
product, and this is a slow release,
microgranular
formulation, sort of releases
continuously through
the intestine. It is not really pH dependent on
how it releases.
During some of the development of the
evaluation of the NDA for this product,
there are
PK studies attempted for bioequivalence
between
pilot and production scale products. Again, here
the issue was we weren't really able to
conclude
bioequivalence because of the high
variability of
the active ingredient, but they were able
to
establish in vitro/in vivo correlation
between
dissolution and the pharmacokinetic
studies, and
that was used to demonstrate equivalence
between
different pilot and production scale
formulations
for that particular product.
So, a third mesalamine
formulation is the
Asacol product. This is a delayed release, coated
formulation, and here, there is pH sensitive
313
dissolution that allows it to target the
colon.
So, when you have all of these
different
products, we can look at sort of the
different
possible ways of testing these
products. The
discussion points that were presented to
the
committee were what is the role of
dissolution,
pharmacokinetics, clinical studies, so
that we can
look at these types of comparative
studies between
these different formulations to sort of
get at
least some sort of solid basis for
discussion.
Gordie in detail showed this
data on the
dissolution studies of mesalamine products,
basically different a pH-independent
product.
These are mainly European formulations,
they are
not the formulations that are marketed in
the U.S.
You can definitely see that at
the low pH,
you see only the sort of slow release
product
dissolving. As you raise the pH, the different
enterically-coated products start to
dissolve
depending on what pH they are
particularly
targeting.
So, by choosing appropriate dissolution
314
conditions that are relevant to the in
vivo
conditions where the product started, and
you can
distinguish between the different
products and what
region of the intestine they may be
targeting.
There is also some studies that
have done
comparative pharmacokinetic studies. Again, here
is sort of a pellet and tablet type
formulation,
not the sort of currently marketed
formulations,
but sort of similar ones.
You see in this case, you look
at the Cmax
and AUC.
Definitely, in this case, the
pharmacokinetic studies actually show a
large
difference between the products, but again,
here,
this is just scientific publications for
these. It
is small sample sizes, so they didn't
evaluate
confidence intervals, but the variability
of the
measures for these drugs are very high.
So, that sort of complicates
the
determination of bioequivalence using
pharmacokinetic measures.
So, with all these different
products on
the market, there has been some interest
in trying
315
to say which one clinically is more
effective. So,
there have been, not complete
head-to-head trials
between all products, but there have been
a large
number of clinical studies, and sort of a
recent
review, came to the conclusion that
clinical
studies haven't been able to demonstrate
significant differences in the efficacy
between
existing mesalamine formulations and
stuff, two
examples here, but there are sort of many
studies
available in the literature.
So, if you were trying to sort
of
determine equivalence between two
formulations, you
might have a very hard time using the
clinical
study to have a sensitive discrimination
of
formulation performance, because these
existing
formulations, which use very different
technologies, aren't very well
distinguished by
sort of clinical studies using the usual
efficacy
endpoints.
So, if we just summarize the
mesalamine
example, if we want to sort of
distinguish the
current products, we would probably yes,
if we
316
chose the right dissolution criteria, we
could
clearly see the difference between the
products.
Pharmacokinetics, it looks like
again we
could see the difference especially
because if the
products release early, they are rapidly
absorbed.
If they are delayed into the colon, then,
the
absorption is much slower, so differences
in local
release actually do show up in the
pharmacokinetics
through especially Cmax for this case.
But again, with this particular
product,
there is the issue of pharmacokinetics of
highly
variables. If you want to do clinical comparisons
in terms of the bioequivalence study,
again, that
would be very challenging in terms of
getting a
sensitive test of the formulation
differences.
So, just to bring in a slightly
different
example, another example of a locally
acting
product is Acarbose. This is an
intestinal enzyme
inhibitor that acts to reduce glucose
absorption.
For this product, there is no measurable
absorption, so you can't use
pharmacokinetic
studies, however, there are
pharmacodynamic
317
endpoints available.
Again, you can look at changes in
glucose
or insulin level in response to a meal
with the use
of this drug or a comparator product.
Here, one sort of interesting
thing to
think about is when we think about the
pharmacokinetic studies, like the
pharmacodynamic
endpoints, are usually downstream
measures of the
formulation performance, in the same way
that the
pharmacokinetic measurements are
here. So, there
is some sort of mathematical similarity
between how
we might interpret pharmacodynamic
endpoints and
pharmacokinetic measurements for GI
acting
products.
Another example of a product
where there
is not much detectable absorption is
cholestyramine. This is a bile acid sequestrant,
essentially binds to cholesterol in the
intestine.
This is sort of an older
product. In
1993, FDA guidance recommended using an
in vitro
binding assay to demonstrate equivalence
of these
products, so there is no dissolution of
318
pharmacokinetics. These assays measured affinity
and capacity.
One of the issues, here, we
talked a
little about the role of excipients. For these
types of products that are involved in
binding
there, then, you sort of worry that if
there is
differences in the excipients in the
formulation,
that that might make a difference in how
they bind
to other products, and these types of in
vitro
binding assays can be valuable and
interesting,
those types of concerns if they are
relevant to a
particular product.
So, before I lead into our
discussion, I
just want to sort of return again to some
of the
scientific issues that were raised by the
GI acting
drugs.
The first is the BCS
classifications and
biowaivers again for systemic drugs. If we have
high permeability and high solubility
drugs in
rapidly dissolving dosage forms, we often
consider
waiving in vivo bioequivalence studies.
So, the question is how should
we extend
319
this to GI acting drugs. I think one more question
is what should the permeability
play. If
permeability doesn't play any role in the
absorption process, should we extend the
biowaivers
to sort of all highly soluble drugs in
rapidly
dissolving forms irrespective of what
their
permeability is, or is there something
about the
classification of drugs in terms of high
permeability, low permeability that may
make that
more risky, perhaps interactions with
excipients or
the role of absorption in the intestinal
tract.
So, that is one of the issues
that we
might like to have some discussion on.
Again, I just want to come back to the
issue of the role of pharmacokinetic
studies in the
absorption from the GI tract. Again, for a
systemic drug, what you see is that the
formulation
performance is what we are really trying
to make a
determination about. Again, this essentially is
controlled by the dissolution.
The drug goes through the
absorption
process, reaches the plasma
concentration, and that
320
is the place where you take a
pharmacokinetic
sample, and that is also the place where
the effect
of the drug takes place.
So, the difference for a GI
acting drug is
essentially the relationship between sort
of the
formulation performance, pharmacokinetic
sample,
that is still the same, but the only
thing that
sort of moved is where the effect is
taking place.
If you think about how we conduct
bioequivalence studies, we usually
conduct them in
healthy people. We don't really concern ourselves
with the effect, so that whatever
connection we are
making for systemic drugs between PK
sampling and
formulation performance, the connection
is still
there for the GI acting drugs.
So, I think the big concern
with looking
at the pharmacokinetic studies in the GI
acting
drugs is essentially when, for the
systemic drugs,
since we know the effect is taking place
where we
are taking the sampling, we have some
sort of
intrinsic way to know what a significant
difference
is.
321
We sort of say 80 to 125
percent
difference in the plasma concentration is
sort of a
general definition of clinical
significance, so we
know that if we meet that, that sort of
gives us an
idea of what equivalent formulation performance
is.
When we go to the GI acting
case, where
the effect is now separate, we don't have
that
connection as to what the calibration is
between
difference in pharmacokinetic sampling
and a
significance clinical effect in the same
way that
we do for the systemic drugs, so that I
think is
the issue for interpreting the
pharmacokinetic
studies is we don't have the sort of
intrinsic
calibration of how significant the effect
on
formulation performance is.
But still the relationship
between
formulation performance and
pharmacokinetics is
still there in both cases in a way that
maybe it is
not for other locally acting drugs, say,
inhalation
products where you have the case where if the
product reaches the lungs, it also can be
absorbed
by different pathways, and you are not
sure that it
322
is passing through the exact same
pathway.
For the GI acting drugs, you
know that in
both cases, it is being absorbed through
the same
site.
Just to lead into sort of the
specific
questions that we wanted you to discuss,
just sort
of outline, a sort of potential framework
for
thinking about bioequivalence of locally
acting
drugs, sort of the first point would be
again the
sort of critical importance of
dissolution testing
in conditions based on understanding of
the
formulation and how it interacts with the
in vivo
environment, so choosing the right
conditions for
in vivo and in vitro dissolution testing.
The second part of that is
pharmacokinetic
and pharmacodynamic studies. Again, the sort of
potential we always see for those is
really, they
sort of confirm the dissolution
testing. They
confirm the relationship between the in
vitro
dissolution testing and the in vivo
dissolution,
which determines product behavior.
They are also important for
assessing
323
systemic exposure in terms of any type of
safety
concern.
The third element of the
framework is
concern about excipient interactions, and
if, say,
the product's mechanism of action is
binding to
like the cholestyramine mean example,
binding to
something in the GI tract where an
excipient could
be competitive or inhibitory for that
binding
process, it might be useful to require
some sort of
in vitro assay for that type of process
for
particular products if there is a
mechanistic
reason why the excipient interaction may
be
important.
So, here, I just want to remind
you of the
discussion questions that we suggested to
you.
We wanted your input on the
role of
pharmacokinetic studies, the role of the
in vitro
dissolution tests, and the role of
clinical studies
in these particular products.
I have sort of listed out sort
of slightly
more detailed versions of these
questions, you
know, how should we use the
pharmacokinetics data,
324
if it's measurable, to evaluate
formulation
performance? What drug specific information would
be valuable in sort of calibrating our
interpretation of pharmacokinetic
studies, when it
would be valuable, when it would not be
valuable?
When is it possible to use
dissolution
testing alone to demonstrate
bioequivalence, and
when do we actually need the confirmatory
data from
pharmacokinetic or pharmacodynamic
studies, as
well?
When should comparative
clinical trials be
conducted, what types of issues there?
The final question on who we
should look
at extending the BSC-based biowaivers for
GI acting
drugs.
With that, hopefully, we will
be able to
have some more discussion on some of
these issues.
Committee Discussion and
Recommendations
DR. KIBBE: Marvin is ready.
DR. MEYER: Kind of relating to your in
vitro question, are there any drug
products that
are known to act locally in an
undissolved state,
325
particles, fine particles?
DR. KIBBE: What about Sucralfate?
DR. MEYER: Yes, Sucralfate. Dissolution,
PK, would that work?
DR. LIONBERGER: I think you have to
consider the mechanism of action. I am thinking
primarily here of drugs, you know, where the sort
of mechanism of action is distinct from
the
formulation. I think when you have products where
the sort of mechanism of action is very
connected
to how the drug is formulated, then, you
have to
have some measure of the formulation
performance in
vivo.
So, if you were thinking mainly
of drugs
where the drug is released from a
formulation
before it reaches the site of action, and
I think
there are sort of other issues where the
formulation acts, just the formulation
or, you
know, manufacturing acts directly.
I think that issue probably
will come up a
lot when we look at nanotechnology.
DR. MEYER: I sort of subscribe to
326
Gordie's point of view. There is probably a
dissolution test that will work, and I
think we
know enough about dissolution testing if
we want to
use paddle and basket and three different
RPMs and
five different pH's, and surfactants, and
everything that anyone has ever done, and
two
products are equivalent under a myriad of
conditions, they are probably going to be
bioequivalent. "Probably" is not a good regulatory
word.
DR. AMIDON: It would be very low risk of
bioINequivalence.
DR. MEYER: Somebody would argue, well,
that's overkill, you shouldn't have to do
89.
Well, that is a lot cheaper than doing a
clinical
trial with these products, so that might
be one
approach.
DR. KIBBE: I think Paul would agree that
it is cheaper than doing a clinical
trial.
DR. FACKLER: Could I make a couple
comments?
DR. KIBBE: Oh, please.
327
DR. FACKLER: Just to answer your
question, there are some rectal
suspensions that
might fall into the category of drugs
that are
effective without dissolving.
My understanding of the lower
part of the
colon is that it is relatively
water-free, and
these undissolved-- mesalamine being one
of them,
by the way--products, hydrocortisone is
another
one, both of which, by the way, OGD has
approved on
the basis of pharmacokinetic
comparability.
The other way that mesalamine,
for what it
is worth, is delivered to the colon is by
suppository in the United States, I
think. So,
there is an oral tablet, there is an oral
capsule,
there is a rectal suspension, and a
suppository,
all of which are equally efficacious and
making you
wonder about the utility of clinical
studies. I
will just leave it at that.
DR. KIBBE: That is a valid point. Once
you have lots of different routes of
administration, and for a local effect,
and the
formulation effects clearly go away when
you look
328
at the clinical impact, but remember that
clinical
endpoints are very wide goalposts, and we
tend to,
I guess being slightly anal-retentive,
want
narrower ones for regulatory purposes.
Just a small point. Most drugs don't act
in the central blood supply, they act
someplace
else, and even though we measure, we
measure
central because we assume, having never
actually
verified this, assume that they are in
relative
strict proportionality to the amount of
molecules
of drug at the actual biophase, and that
is the
whole basis for kinetics and what Marvin
and I have
done for all our lives, so we are
reticent to give
that up, but you can't say that that is
where the
drug works, because that is not where it
is working
either.
DR. KIBBE: Ken.
DR. MORRIS: I just had a question because
I think, Art, actually, you mentioned
Sucralfate.
What is the criteria for bioequivalence? That is
not absorbed at all, right?
DR. KIBBE: Right.
329
DR. MORRIS: So, what is the criteria for
bioequivalence for that?
DR. AMIDON:
I know originally, they were
doing clinical studies. There were clinical
comparison studies.
DR. MORRIS: It just seemed to me I mean
it would make no sense to do
pharmacokinetic
studies on cholestyramine, which never
dissolves
even when it is active.
DR. KOCH: Just to add to that, the
cholestyramine is an interesting one,
because from
an in vitro type, you can't really
duplicate the
sequestering. I mean bile acid is just one of the
things that it sequesters it. Basically, it's a
handful of ion exchange resin, and ion
exchange
resins are trained to go after a lot of
things
where it can pick up that ion.
So, that would be a difficult
one I think
to just run one simple in vitro test.
Another point that I thought of
when we
talked about the suppository, Ajaz, when
you go to
Europe, do you do a check in terms of
dosage forms,
330
as well, because it was very interesting
on a
European assignment, it turns out that
more of our
drug deliveries were suppositories than
they were
tablets, and someday we will start to see
that as
another way of administration,
particularly as
dosages get smaller and smaller.
DR. KIBBE: I think suppositories are much
better accepted among the populace in
France and
Germany than they are here.
You had a comment.
DR. AMIDON: I worked with the FDA
extensively and did a lot of in vitro
testing on
the bile acid resins, with different bile
acids
under differing conditions, and so it is
a fairly
rigorous test in terms of the capacity of
the ion
exchange resins to bind relevant bile
salts.
I think if you look at the
guidance, you
would look at it and say if two resins
appear the
same under all of these conditions, they
are
likely, likely, the risk of
bioINequivalence is
low.
Of course, the questions in
plasma, what
331
are you going to measure? You wanted to do
cholesterol lowering. That is a long, extensive
study, so clinical studies, at least I
interpret
here, clinical studies meaning efficacy
studies are
much more complicated, much more variable,
and I
think quite insensitive to formulation
differences.
So, I think we can make an
adequate case,
and if you come up with something that
you think
might affect the in vivo performance, we
can
enumerate what happens in all of the
components in
the GI tract, and they can be tested, so
we can do
an in vitro test to see if it has an
effect and
decide whether it is relevant or not.
DR. YU: I just wanted to comment on
Paul's comments for rectal suspensions,
especially
for the mesalamines, when we look at the
pharmacokinetics, we also look at a
dissolution
very closely. Thank you.
DR. KIBBE: So, you have a whole body of
data then on pre-existing products of
varying
formulation, so the Agency actually has a
real good
handle on whether or not there are a
diversity of
332
excipients and could actually do
dissolution
testing on samples of all the products
already on
the market in various environments to
come up with
a criteria.
DR. YU: That's correct.
DR. KIBBE: Anybody else got anything?
Ajaz has a comment. Good.
DR. HUSSAIN: As I think about these
questions, I think Dr. Amidon and Ken
Morris both
have pointed out in a sense I think what
is in
vitro test conditions and how appropriate
they are.
That is a significant challenge.
I don't want to sort of jump in
and say
all right, BCS Class I was highly
soluble, highly
permeable, 900 ml, and so forth, because
the volume
and the hydrodynamics, and so forth, I
think we
have to give some thought to how we would
approach
that, so it is not a trivial matter.
DR. KIBBE: Go ahead, Ken.
DR. MORRIS: I guess I would just echo
that.
The principle I think is sound, you know,
which is no surprise coming from Gordon,
but the
333
tactical aspects of that really are quite
a
challenge. There is still a lot of work to do in
terms of dissolution testing. As Gordon said,
redesigning the dissolution test is no
seed for the
faint hearted, I mean that is something
that is
going to really take some serious
scientific and
engineering work.
DR. MEYER: Is it correct, Gordon, that a
Class I BCS is likely to appear, at least
to some
extent, systemically, and a Class III
similarly,
maybe not so much so, but still, because
of high
solubility, you are likely to have
something you
can measure, and therefore, you could do
a PK
study?
DR. VENITZ: If it has high first pass
effect, it might not be systemic.
DR. MEYER: That's true.
DR. KIBBE: Detectable levels are going to
be a problem.
DR. AMIDON: Class III drugs tend to be
not very highly metabolized, so it would
probably
work there, and that is where I think it
is the
334
most important, because for a low
permeability
drug, there is obviously permeability
dependence
along the GI tract, because there is some
absorption, and then it stops. It has to because
it is not fully absorbed. So, I think Class III
drugs is where it is more critical.
DR. MEYER: Therefore, you wouldn't need
to give a waiver, you could do PK.
DR. AMIDON: You could.
I am not saying
you can't do PK. In fact, PK plasma levels in
general, even for GI drugs, I mean if you
can
measure something, you know, it would
give you the
highest assurance. The question is what is the
best test, and broadly, for
bioequivalence, Cmax,
AUC is our gold standard. I think our focus on
Cmax, AUC has kind of preempted us from
thinking
about what is the real issues here, which
are for
oral absorption and/or, for GI, the GI
locally
acting drugs, the dissolution process is
where the
action is at, and when we want to set
standards,
some drugs are going to be simple and
some drugs
are going to be complicated, so let's try
and
335
decide where we can simplify the standard
and make
it maker, and then where it is
complicated, well,
that is where science is today.
DR. KIBBE: I agree with Gordon. I think
if you have a lot of background data, we
can safely
go to a dissolution test with something
like this.
In the absence of it, it is always nice
to have a
little bit of PK data, blood level data,
maybe a
simplified study just to get a sense for
the
levels, because we want to be careful of
toxicity
and equivalence, and it is going to be
case by
case.
Anybody else? Jurgen.
DR. VENITZ: I was just going to speak and
for being a former clinician, in favor of
clinical
studies.
I mean everybody here is mentioning a
true statement. They are not very sensitive to
formulation effects, but on the other
hand, they
are the ultimate relevant test. I mean they make
what we are doing clinically relevant.
So, as much as I am personally
in favor
and moving along with looking at
dissolution
336
testing as the base of your surrogate of
in vivo
bioequivalence, there is a price to be
paid, and
that is, we are going to find differences
in those
dissolution tests between formulations
that
clinically are irrelevant.
So, we are looking for
discriminating
tests--excuse the term--that
discriminates between
formulation differences that are
clinically
probably meaningless.
DR. KIBBE: And the question I guess boils
down to an economic one, do I want to
spend the
money to do a clinical test to show
obviously that
the differences are meaningless, or can I
do a
fairly well designed dissolution test
which doesn't
cost me much and is very discriminating,
and if I
pass that, I am guaranteed that I will be
okay on
the clinic, and that is really what these
tests
are.
These are surrogates for the ultimate use of
the drug in 400,000 people.
DR. VENITZ: We had a similar discussion a
couple of years ago when we talked about
intranasal
products, and I guess this committee
voted in
337
favor, and the FDA ultimately accepted
the fact
that the only way to assess
bioequivalence of
intranasal products is the walk in the
park, in
other words, a clinical test.
Given the fact from my
perspective that I
think we understand much more about GI
dissolution,
GI
absorption, all of which you presented, Gordon,
I am personally comfortable in moving
along with
that, and not making the clinical gold
standard a
requirement, but I am just cautioning
that in the
process, you are going to throw out
formulations
that are clinically probably equivalent.
DR. KIBBE: To ahead, Ajaz.
DR. HUSSAIN: That is one of the reasons,
I think, why we pushed the concept of
quality by
design, and so forth, because all the
relevant
formulation information and all that has
never been
brought into that discussion.
It was simply a test to test
comparison
discussion, so over the last several years, we
have brought that discussion up, and then
as you go
towards understanding your formulation,
338
understanding what pharmaceutical
equivalence could
mean from that perspective, we actually
can open
that debate again because of that.
DR. KIBBE: Ken.
DR. MORRIS: Just a quick question,
Jurgen.
You are still doing dose-ranging studies
when you are doing the initial
development, I
guess, so if you are going to use the prior
knowledge, use the dose-ranging studies,
doesn't
that help you when it comes time to
determine
whether or not the tests are
over-discriminating or
not?
DR. VENITZ: But I mean most of the time,
even a two-fold dose range, you may not
be able to
distinguish clinically, so you are
talking about
100 percent difference in formulation
performance,
and clinically, you may not be able to
tell the
difference.
DR. KIBBE: Paul has another one.
DR. FACKLER: I just want to correct
something. On the nasal products, the clinical
study is required, but in addition to
that, the
339
only way to get a generic product
approved is to
also pass a PK study and also pass in
vitro plume
geometry and spray pattern.
So, one of those three isn't
good enough,
two of those three aren't good enough,
all three
need to pass in order for the nasal
products,
which, in my personal opinion, is
overkill for
demonstrating that the two products are
equivalent.
The clinical study alone should have been
enough.
If the patients are being benefited
equally, the
products to some extent are
bioequivalent.
The other thing I wanted to
correct was
just that the variability of mesalamine
is
admittedly high, but not so high that it
can't be
dealt with in a pharmacokinetic sense.
DR. YU: At the last Advisory Committee
meeting, we had a topic on how to deal
with highly
variable.
DR. FACKLER: Only that I know there are
some relatively new data on some
mesalamine
products available to the Agency, so that
the
variability, at least from rectal
suspensions, is
340
defined and was manageable in an BE sense.
DR. VENITZ: Just to follow up on that,
you are correct. I mean for the intranasally
administered drugs, they have to pass all
three.
What I would like to see for this in terms
of the
future progress, would be not to get to
that level.
If we accept in vitro dissolution as a
surrogate of
in vivo dissolution, as a surrogate of in
vivo
bioequivalence, let's stick with it.
If you decide that we don't
mechanistically understand enough what is
going on
and we require clinical study, let's
stick with the
clinical study.
DR. KIBBE: Thank you, Jurgen.
Anybody else? I see by the clock on the
wall that we are running out of
time. Have we
given you enough guidance on this one to
move
forward without taking any formal
votes? Lawrence
wants some more information.
DR. YU: That's correct. Thank you.
DR. KIBBE: He wants to thank me. That's
good.
341
I have on my calendar of events
that there
is a summary and conclusion, summary
remarks, and I
have two names, and they are looking at
each other
like which one of you is going to say
anything. I
would be happy to just rule you of order
and close,
if you don't have anything to say.
Go ahead.
Conclusion and Summary
Remarks
DR. HUSSAIN: I think again as the
previous meeting, I think the discussions
were very
valuable and I think help us think
more. The one I
think you probably for the first time got
an
opportunity to see the range of
laboratory and
other such activities that we have
ongoing, and I
think the Critical Path Initiative
clearly is not
just lab based, it is much broader than
that, but
that discussion allowed us to think more
carefully
about how to approach the Critical Path
Initiative.
It also helped us to start
thinking about
how do you align such programs,
especially the
laboratory programs, to be targeted and
the key
questions. The challenge is great and I think we
342
want to maintain as many best practices
as we have,
and you did see a number of best
practices sort of
come out in the discussion, and maintain
that, and
bring all the offices in OPS to be
aligned
together.
The immediate office project
that we
articulated, the three projects, all
interrelated,
I think will be a means to not only
identify the
best practices, but also to bring a
system approach
to address uncertainty and complexity,
and I think
that would be the key aspect and in many
ways, that
allows us to approach bioequivalence,
follow-on
proteins, generic drugs, all of those
challenges
next year in a systematic manner.
So, I think in a number of
cases, I think
irrespective of what that pathway for
these
products might be, the follow-on
proteins, the
scientific framework for the
decisionmaking process
should be common irrespective of that,
and it
should be related to the uncertainty and
complexity
of the dosage form set of products that
we have.
At the same time, I think you saw an
343
impressive array of laboratory research
from
biology to quality, and how do we align
that, I
think is a significant challenge. We have sort of
summarized that discussion early this
morning, and
the key questions, the metrics, I think
will be the
key part for sort of making sure whatever
approach
we use is measurable and then
quantifiable in terms
of its benefit to the critical path, and
so forth.
But I do want to emphasize in
the sense
that FDA is only one part of that
critical path.
Industry, academia, and other agencies
play an
equally important role. Our role will be more also
of coordination, but the need for
research,
especially fundamental research in this
area and
need for public funding is acute.
I do want to go back and say
the
formulation development, manufacturing
has been a
neglected area, especially in the U.S.,
and if you
don't bring the focus on that, I think we
are
already 10 years behind Europe and Japan
in many of
these areas, so we will lost that part of
the
industry, so that is important, so seek your
help
344
to make that case also.
I think in terms of the gaps
going to the
desired state, Helen outlined some of the
key
fundamental organizational gaps. Some
reorganization is already occurring. White Oak
provides an opportunity to really bring a
team
approach and peer review process to the
CMC
function in Office of New Drug Chemistry,
but at
the same time, I think one key aspect is
a
question-based CMC review process which
focuses on
risk.
That is already in the works,
but also
support that with tools that we have not
often
utilized in this arena, and that is
chemometrics
modeling and other aspects of that.
I have a virtual team for
chemometrics
right now, but I think we are adding
people with
computational fluid dynamics, and others,
to really
make a core team that will support the
review
function in many aspects.
For example, computational
fluid dynamics
is the issue of hydrodynamics, issue of
inhalation
345
products, and so forth. So, hopefully, we will
have that team up and running soon.
I think the science gaps are
significant
from a training perspective, but those
are not, in
my opinion, unsurmountable. I think we have seen,
we have the expertise within the Office
of New Drug
Chemistry, Generic Drugs, and so
forth. It is
simply identifying and aligning that
expertise bear
on some of those challenges, but also provide
a
training program for all of our
reviewers.
In fact, I really think the PAT
training
program opened up a lot of opportunities
for our
staff to excel in areas and become
leaders
worldwide, and although we cannot do the
entire
period, training for all of our staff
immediately,
especially the practicum part of it, but
as we go
to the next PAT training program, we
intend to open
the didactic sessions that we have
locally to all
of
CMC reviewers to be part of it. So, we
will
bring that onboard.
I think Jon outlined for you
some of the
directions we will make more in
policy. Jon is
346
aggressively moving in that direction and
I think
his leadership will help us align. A number of
people have joined his group. He has a group that
is focused on that now.
So, policy alignment and the
OPS
Coordinating Committee with Gary Buehler
and Keith
Webber co-chairing that, sort of brings
all in one
place now to sort of make sure that the
policies
that evolve are aligned with where we
want to go.
The issue I think I do want to
mention,
the issue of two-tiered approach, I think
there is
a risk of that, clearly, there is a risk
of that,
but I think we will try at least at the
draft 3.1
for Q8, at least I felt that the language
was
written not to invoke a two-tiered
approach. It's a
continuum, it will be a challenge to
manage, but I
think we will get there.
I am hoping in Yokohama, Japan,
starting
November 12th, we will bring Q8 to Step
2. I am
keeping my fingers crossed. As soon as that
happens, I think things will start moving
rather
quickly.
347
Pharmaceutical development
information is
already coming in. Many companies were willing to
take the first step, and have done so,
and the
initial experience is positive, but we
will have a
quality system, along with peer review
process, to
make sure consistency and proper
utilization of
that comes in.
I thin, the other two topics
are probably
fresh in our minds. I think bioINequivalence is a
significant challenge, but it is a
challenge right
now we are facing to minimize our
resources being
spent in things which we think are not
value added,
and I think as we move forward, the
discussion here
will be helpful.
We will probably not bring that
topic back
and we will probably come with an
approach, and
then solve that in a way which is
consistent with
the way we do it, so I think the
discussion will be
helpful, but I still feel, I think Jurgan
and
Professor Nozer Singpurwalla, I think we
have to
start using prior information, prior
knowledge more
effectively, and especially with
biostudies that we
348
will have access to, I think it allows us
to be
more proactive and make decisions more
quickly, so
that Gary and his staff really don't have
to spend
so much time in answering these questions
in a
legal perspective, and so forth.
Locally acting products clearly
are part
of
the critical path for the generic drugs.
It is
not only GI, inhalation, topical, it is
an entire
area of research that Lawrence will have
to sort of
spearhead and move forward, and that is a
critical
path research for generic drugs.
Approval of generic drugs in a
timely
manner hinges upon that. I think that PAT concept,
the cGMP, the Quality by Design are all
positioned
right to help generics and help
innovators all
together, and so you will see that
happen.
With dissolution testing, I do
want to
sort of say that I think dissolution
testing, we
have to think carefully about the
variability
aspects of that and how we calibrate. In
many ways,
I think our labs have started putting a
document
together.
They feel that mechanical calibrators
349
and others are sufficient and relying on
an
external calibrator of poor quality
actually is
diverting attention away and actually
creating
problems, unnecessary problems, so we
will issue
something on that soon hopefully.
With that, I will hand it over
to Helen.
MS. WINKLE:
Ajaz did a wonderful job of
recognizing, I think, the contributions
that the
committee made. I think there were some excellent
discussion over the last two days and
some
excellent recommendations that have been
made to us
on things that we need to focus on more
and areas
that we need to do more planning and even
more
research.
I did also want to mention I
thought that
the presentations made by Dr. Boehlert
and Dr.
O'Neill on the two workings groups, the
subcommittee for Dr. Boehlert and the
working group
for Dr. O'Neill, were very beneficial to
those
discussions. I think both groups are working hard
to accomplish a lot and to get answers
back to us
that are really necessary.
350
The Manufacturing Subcommittee
at the last
meeting I thought did an excellent job,
and think
their report back to you yesterday was
indicative
of how much effort they are putting in to
helping
make some of the recommendations that we
need to
move forward.
Also, with the Dose Uniformity
Working
Group, they, too, have worked very hard
during the
year, and I think that the report Bob
made was well
accepted by the committee and is a good
indication
of how these working groups, too, can be
beneficial
to the committee in making
recommendations to us in
the future.
I do have some other little
things,
though, I want to talk about, and that is
the two
people that are leaving the
committee. It is a sad
time for us, I think here at FDA, because
we have
appreciated both Marv and Art's contributions.
They have been very, very significant in
helping
direct us at the Agency in the directions
that we
really need to go, and they have also
provided a
great deal of scientific expertise and
knowledge,
351
not only on the committee, but in other
aspects,
too, and they have been very valuable to
us.
The one thing, too, I would
like to add is
they have also added a great deal of
humor to this
committee, which I think many of us are
going to
miss.
Now, how long we will miss it is
questionable, because they are both SGEs
and can be
called back at anytime--just like Gordon
is over
there shaking his head--they can be
called back at
anytime to participate in different
discussions,
and I think that we are probably going to
continue
to take advantage of them.
But today, being their last day
on the
committee, I do want to present them with
these
plaques in recognition of their service
to the
Advisory Committee.
The first one is to Dr.
Kibbe. Not only
has he been an excellent, excellent
member of the
committee, he has also been a very, very
thought-provoking chair even though a
little
schizophrenic, I worried about today,
when he
thanked himself. But we certainly appreciate all
352
your service and everything.
Thank you.
[Applause.]
MS. WINKLE: And the other plaque goes to
Marv, and the one thing I want to say
about Marv, I
have enjoyed having dinner with Marv in
the
evenings.
Last night, for you who weren't at
dinner, he had this huge, huge plate of
food, and
he said, "It is actually bigger than
it looks."
He has contributed a lot at
this meeting,
and we are going to miss him.
[Applause.]
DR. KIBBE: Marvin, would you like to make
a comment, a last shot across the barrel?
DR. MEYER: It took I guess 30 years to
get on this committee, but I have
thoroughly
enjoyed it. Everyone around the table brings a
different perspective, and that is what
makes it
good for the FDA and fun to be part of.
We have good chairs, Vince Lee,
and then
Art Kibbe, and so I would highly
recommend this
position for three years to anyone who
wishes or
353
gets invited to participate. Beyond three years
may be questionable.
So, thank you.
DR. KIBBE: Thank you, Marvin.
I have a whole series of points
to make.
First, is that this has been a real joy
and an
opportunity to serve and do what I think
are useful
things, and to work with people who are
dedicated
to having positive outcomes for the
American
public.
Most of that, of course, goes,
the blame
for how well it turned out goes to Ajaz
and Helen,
who lead a great ship and have become
close
friends, as well as good working
colleagues.
I think we did quite a bit over
the years
and I think there is quite a bit more to
do, and
the committee needs to move forward, and
I would be
happy to help in whatever manner I can.
One of the things that I think
you need to
be careful about is that the speed of
change is
ever increasing, and like Alice, you are
running as
hard as you can just to stay where you
are, and to
354
get ahead, you have to jump off of that
treadmill
and get on a different path.
One thing that I didn't mention
this
morning that you need to keep in the back
of your
mind is that according to the U.S. law,
treaty
trumps law. If the Senate and the President want
to sign a treaty with some country that
allows for
something to happen, the Food, Drug, and
Cosmetic
Act is trumped by the treaty, and
whatever rules
and regulations you have, the treaty
wins.
Yes, it is absolutely
true. Treaty trumps
law, and the President signs it, and the
Senate
agrees to it, and the House of
Representatives can
complain all they want, and the
regulatory agent
have to readjust.
I would love to see the
industry take some
of its money that it spends on
direct-to-consumer
advertising and put it into, first, getting
the
American public to understand how cost
effective
drugs are relative to other therapeutic
moieties,
because they don't understand that,
because they
see the bill in front of them and they
don't see
355
the other bills.
The other thing is to get them
to
understand that drugs aren't safe, and
they
shouldn't just use them because somebody
says it
might be a good deal. I don't know whether we can
get the industry to do that, because I
know that
the ads are meant to sell things rather
than not
sell things, and taking out ads to tell
people not
to do things is hard to get them to do,
but I would
love to do that.
We need to change the criteria
for how we
evaluate who well the FDA is doing. I think there
is way too much pressure on them to
produce new
drugs, to produce new reviews, to produce
new
things, and I don't know what the right
productivity criteria is.
I know we need to change the
productivity
criteria for the U.S. Patent Office, that
we have
got to stop them from just issuing
patents to make
sure they have issued three patents, and
give them
credit for not issuing a patent that
shouldn't be
issued.
356
Then, lastly, the
goldpost. When you use
science to establish the goldpost for
regulatory
approval, you have moving goalposts
because science
moves, science progresses, current best
thinking is
always better than it was 10 or 15 years
ago, and
both the Agency and the industry have to
understand
that that is not a threat, that is an
opportunity.
I truly have enjoyed myself and
I hope
that what little I have done has
contributed to
everybody else having a good time.
I think that it is 4:30 and it
is an
appropriate time for us to adjourn. If there are
not other dramatic statements that need
to be made
by anyone else, I will see you next time
maybe.
[Whereupon, at 4:30 p.m., the meeting was
adjourned.]
- - -