DEPARTMENT OF HEALTH AND HUMAN SERVICES
FOOD AND DRUG ADMINISTRATION
CENTER FOR DRUG EVALUATION AND RESEARCH
ADVISORY COMMITTEE FOR PHARMACEUTICAL SCIENCE
Wednesday, May 4, 2005
CDER Advisory Committee Conference Room
5630 Fishers Lane
Charles Cooney, Ph.D., Chair
Hilda F. Scharen, M.S., Executive Secretary
Patrick P. DeLuca, Ph.D.
Paul H. Fackler, Ph.D., Industry Representative
Michael S. Korczynski, Ph.D.
Gerald P. Migliaccio (Industry Representative)
Kenneth R. Morris, Ph.D.
Marc Swadener, Ed.D. (Consumer Representative)
Cynthia R.D. Selassie, Ph.D.
Nozer Singpurwalla, Ph.D.
SPECIAL GOVERNMENT EMPLOYEES
Carol Gloff, Ph.D.
Arthur Kibbe, Ph.D.
Thomas P. Layloff, Jr., Ph.D.
Marvin C. Meyer, Ph.D.
Gary Buehler, R.Ph.
Kathleen A. Clouse, Ph.D.
Jerry Collins, Ph.D.
Ajaz Hussain, Ph.D.
Robert Lionberger, Ph.D.
Robert O'Neill, Ph.D.
Keith O. Webber, Ph.D.
Lawrence Yu, Ph.D.
C O N T E N T S
Call to Order
Charles Cooney, Ph.D. 5
Conflict of Interest Statement
Hilda Scharen, M.S. 5
Parametric Tolerance Interval Test for Dose
Current Update on the Working Group
Robert O'Neill, Ph.D. 8
Quality-by-Design and Pharmaceutical Equivalence
Ajaz Hussain, Ph.D. 10
Using Product Development Information to
Extend Biopharmaceutics Classification
Ajaz Hussain, Ph.D. 30
Using Product Development Information to Address
the Challenge of Highly Variable Drugs
Lawrence Yu, Ph.D. 79
Using Product Development Information to Support
Establishing Therapeutic Equivalence of
Robert Lionberger, Ph.D. 140
Summary of Plan
Ajaz Hussain, Ph.D. 169
Committee Discussion and Recommendations 192
Criteria for Establishing a Working Group for
Review and Assessment of OPS Research Programs
CBER Peer Review Process for
Kathleen A. Clouse, Ph.D. 221
CDER Peer Review Research
Jerry Collins, Ph.D. 243
C O N T E N T S (Continued)
Committee Discussion and Recommendations 258
Conclusion and Summary Remarks
Ajaz Hussain, Ph.D. 273
Helen Winkle 280
P R O C E E D I N G S
Call to Order
DR. COONEY: I would like to call the
meeting to order this morning.
Conflict of Interest Statement
MS. SCHAREN: Good morning. I am going to
be going through the Conflict of Interest
The Food and Drug Administration has
prepared general matters waivers for the following
Special Government Employees: Drs. Charles Cooney,
Patrick DeLuca, Carol Gloff, Arthur Kibbe, Michael
Korczynski, Thomas Layloff, Marvin Meyer, Kenneth
Morris, Nozer Singpurwalla, who are attending
today's meeting of the Pharmaceutical Science
Advisory Committee, to:
1. Receive an update on current
activities of the Parametric Tolerance Interval
Level PTIT Work Group;
2. Discuss and provide comments on the
general topic of considerations for assessment of
pharmaceutical equivalence and product design;
3. Discuss criteria for establishing a
working group for review and assessment of Office
of Pharmaceutical Science Research
The meeting is being held by the Center
for Drug Evaluation and Research. Unlike issues
before a committee in which a particular product is
discussed, issues of broader applicability, such as
the topic of today's meeting, involve many
industrial sponsors and academic institutions.
The committee members have been screened
for their financial interests as they may apply to
the general topic at hand. Because general topics
impact so many institutions, it is not practical to
recite all potential conflicts of interest as they
apply to each member.
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.
Paul Fackler and Dr. Gerald Migliaccio
participating in this meeting as non-voting
industry representatives acting on behalf of
regulated industry. Dr. Fackler's and Dr.
Migliaccio's role on this committee is to represent
industry interests in general, and not any other
one particular company. Dr. Fackler is employed by
Teva Pharmaceuticals, Dr. Migliaccio is employed by
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
DR. COONEY: Thank you, Hilda.
I am Charles Cooney, the chairman of the
committee, and will preside over the
today. We have several topics on the schedule.
The first of these is a current update of the
working group Parametric Tolerance Interval Test
for Dose Content Uniformity by Robert O'Neill, who
I believe has just come in.
Parametric Tolerance Interval Test for
Dose Content Uniformity
Current Update on the Working Group
DR. O'NEILL: I am here to just update you
on what we promised you from the last meeting on
October 19th. As you know, there is a technical
working group that has been put together with folks
from FDA and folks from the IPAC group.
We have been working diligently since
then. We thought we would have some
recommendations for you today. We do not, but we
have had approximately eight get-togethers, five
telecons and three face-to-face meetings, the last
of which was about a week, week and a half ago.
What those discussions have been about is
the agreement of the statistical formulations of
the problem. There were a number of
regarding the coverage probability and symmetry of
coverage, off-target operating characteristic curve
agreements, and things of that nature.
There has been some computer programs that
have been shared back and forth, validation of each
other's methods, and I believe there is now
agreement on that aspect of it and that the working
group is turning towards the agreement on where the
operating characteristic curve ought to be placed
and how it might handle certain particular
situations, particularly off-target means.
That is essentially where we are. I think
everyone is optimistic that probably the next time
we report to you, that there will be actual
recommendations for you to respond to, but that is
essentially where we are right now.
I would be willing to take any questions
if you have any.
DR. COONEY: Any comments or questions
from the Committee? If not, thank you very much
and we look forward to the next step.
DR. O'NEILL: Thank you.
DR. COONEY: We will immediately begin
with the second topic this morning, which is
Quality-by-Design and Pharmaceutical
to be introduced by Ajaz Hussain.
Quality-by-Design and Pharmaceutical Equivalence
DR. HUSSAIN: Good morning. I would like
to introduce to you the Topic No. 2,
Quality-by-Design and Pharmaceutical Equivalence.
We believe we have an opportunity here to explore
and what we plan to do with you is to share some of
our initial thoughts and hopefully, engage the
Advisory Committee in discussion to help us make
sure we are on the right track.
In many ways, the discussions continue
from yesterday where, in essence, we are looking at
opportunities that have been created and
re-examining some of our current policies and to
see how we can realize opportunities to move
towards a desired state.
I would like to put this in the context of
moving from a reactive to a proactive
system for pharmaceutical quality, and recognizing
that this is only a journey, not a destination. I
think in the world of continuous learning and
continuous improvement, really, continuous learning
is your destination in some ways.
I think over the last four years, we have
focused on discussing some of the opportunities in
general, but if I look at some of the reactive
examples on this chart, yesterday, in some way, we
talked about testing to document quality, repeating
deviations in our specification investigations, and
in some ways we will start focusing on the other
aspects of prior approval supplement for process
optimization and continuous improvement, multiple
CMC review cycles, but more importantly, I think,
how can we leverage the opportunity of quality by
design for demonstrating therapeutic equivalence of
I think we struggle often in this arena,
and we have struggled and unable to sort of find
ways of approving generic products and the degree
of complexity has increased such as
products and inhalation products, and so forth, so
how can we leverage the pharmaceutical development
information to seek out ways to find mechanisms for
approving complex generic products, I think is a
I think the proactive examples I think is
quality by design and real time release, right
first time, process optimization and continuous
improvement within the facilities quality system.
But I think in some ways, today, we will
focus on single CMC review cycle, less so, but I
think quality-by- design approach for demonstrating
therapeutic equivalence of generic products would
be a focus.
Therapeutic equivalence. The definition
from the Orange Book is as follows. Drug products
are considered to be therapeutic equivalents only
if they are pharmaceutical equivalents and if they
can be expected to have the same clinical effect
and safety profile when administered to patients
under conditions specified in the labeling.
drug products are considered to be pharmaceutical
equivalents if they contain the same active
ingredients, are of the same dosage form, route of
administration, and are identical in strength or
Pharmaceutically equivalent drug products
are formulated to contain the same amount of active
ingredient in the same dosage form and to meet the
same or compendial or other applicable standards of
strength, quality, purity, and identity, but they
may differ in characteristics such as shape,
scoring configuration, release mechanisms,
packaging, excipients, expiration time, and within
certain limits, labeling.
So, I think we have certain flexibility
built into the issue of pharmaceutical equivalence,
and one of the desired states is to leverage that
and to say that we want to set specification based
on mechanistic understanding, so if you have a
different mechanism, that is perfectly fine, but
then you set your specification on that.
The discussion yesterday was
not to force
the generic to a particular specification, but
recognize that as part of that.
Today, I think I would like to put this in
the context of risk, uncertainty, and variability,
and I think that framework will help us think more
clearly about the issues.
FDA classifies as therapeutically
equivalent those products: That are approved safe
and effective, so you have to have a reference
product which is safe and effective, and has an
approved pharmaceutical equivalents against the
repetition of that, and are bioequivalent, that
they do not present a known or potential
bioequivalence problem, and they meet an acceptable
in vitro standard, or if they do present such a
known or potential problem, they are shown to meet
an appropriate bioequivalence standard.
In absence of pharmaceutical development
information and quality-by-design aspect, we have
to assume that they present a bioproblem, so we
often go to the second bullet in most cases; are
adequately labeled, and are manufactured
compliance to Current Good Manufacturing Practice
In a sense, the last four years, our
initiative goals have been to focus on
science-based, risk-based approaches, and there are
certain challenges I think in our articulation of
the problem today.
I want to sort of share with you what the
challenges might be.
Risk-based scientific decisions on
pharmaceutical quality is clearly our goal. Risk
is a combination of the probability of occurrence
of harm and the severity of that harm. The reason
for focusing on that is an aspect that I feel that
we often get entangled in and unable to really
articulate the problem carefully.
Uncertainty with respect to severity of
harm and/or probability of its occurrence and their
modulating factors is that challenge that we face,
what are the critical quality attributes, and what
is an acceptable variability.
I have argued, and I think I
will ask the
Committee to think about this, is my argument sort
of correct, in some ways, in the current decision
system which tends to be reactive, one contributing
factor for the reactive decision system is we
confound uncertainty, variability, and risk.
This is, by nature, how we develop our
products to a large degree. The current paradigm
for product and process development tends to do
this, because our entire effort in new drug
development, for example, is focus on safety and
efficacy of a molecule, but we use a product to
The connection between product quality and
the clinical is generally focused on the molecules
rather than the product. So, that is a part of
Often intrinsic safety and efficacy of a
new molecular entity is confounded with its product
and manufacturing process just by the nature of our
product development and process itself.
We have multifactorial aspects of
pharmaceutical products and manufacturing
processes, and there is increasing complexity, and
if we can find a way to articulate our problem more
carefully, this may help moving forward more
Establishing constraints based on prior
knowledge and limited development experiments that
are done in the development cycle.
What I have argued is there is a need to
entangle or, as I call it, de-convolute
uncertainty, variability, and risk, and then to
achieve truly a scientific integration of these for
quality decisions, how we set specification.
Yesterday, for example, what we proposed
was an assessment of variability, example, begins R
and R. That is an attempt to start characterizing
the variability and attempt to start teasing out
what comes from what.
This may appear to be paradoxical, and it
probably is without the concept of quality by
design, and that sort of links to Dr. Woodcock's
paper that we talked about.
Let me illustrate that. When we approve a
new drug application, we bring assessment of all
the disciplines to bear on a decision which says
the risk to benefit ratio of this proposed drug
product is acceptable when used in accordance to
the label. That is the pivotal decision criteria.
Now, that is based on the pivotal clinical
trials along with the toxicology, and along with
all the studies that go along with that, but the
pivotal clinical trials play the major role there.
What do we use for the pivotal clinical
trials? We use a product, and we often do not have
the opportunity or do not have the intent to gauge
whether the product is modulating the safety and
efficacy of the molecule beyond that of exposure
less the bioavailability.
So, from that aspect, the quality of the
product has to be built in before you get into the
clinical trials, otherwise you confound the
clinical trials with quality problems, and
actually, I will illustrate one example in my
second presentation of that.
As that happens, then, the
approved and then manufacturing is transferred to
production and you have seven production lots, but
that goes to the patient population.
So, you see the disconnect between
pharmaceutical quality or product quality. I am
not talking about the safety and efficacy of a
molecule, I am talking about how a product
So, in many ways, if I look at drug
development program, and here is an illustration of
that, this is actually a real case study. The only
difference I have here is I have a line that made
it more linear. That is the only change I did, but
here is a development program.
The initial formulation was a capsule.
Then, they went to a table, wet granulation tablet,
and each star is a bioequivalent study. So, they
qualified that change from capsule to tablet using
a bioequivalent study. Then, they add a film coat,
then, change the site of manufacture of the drug
substance, and so forth.
You see the changes that were
this drug development program, some before clinical
trials were initiated, some during the clinical
trials, site changes, and so forth.
But this company actually qualified, what
they call bridging studies using a bioequivalence
trial. Towards the end of the review time, to
qualify and bridge the clinical trial, pivotal
clinical trial material with the to-be-marketed
product, they opted to use multiple-dose studies of
the traditional thing, and that failed at the last
minute. Actually, it didn't fail, it just failed
to establish bioequivalence, so they went back and
actually repeated the study with a larger number of
subjects, but it delayed them.
The point here is all these changes are
being qualified on the basis of their traditional
bioequivalence trial. In new NDAs, we generally
see from 3 to 6 clinical bioequivalence studies,
and that was our number.
Here is a number that was shared with us
by Jack Cook sometime, this is the year 2000, about
7 per approved compound. In a sense, if I look
at--this is Gary Buehler's slide--I think in many
ways, the difference between the generic drug
approval and the new drug approval is minimal when
it comes to bioequivalence trial.
In many ways, we use bioequivalence to
gauge information which we don't have in the NDA,
but in reality, in some cases, it is simply the two
are comparable. So, from that aspect, I think when
I look at oral products, immediate release
products, and here is my demarcation in an attempt
to sort of categorize what is uncertainty, what is
variability, and what is risk.
The goal of our generic drug approval
process is to approve a generic product. An
approved product is expected to have the same
clinical effect and safety profile when
administered to patients under conditions specified
in the labeling.
Based on the previous slide, I have shown
the characteristics of this decision system is,
one, the first one, the product must be
pharmaceutically equivalent, and here the
are has the applicant demonstrated that it's the
same active, identical amount, same dosage form,
route of administration, and so forth, identity,
strength, quality, purity.
So, that is an assessment process of how
good the identical methods are and how good they
have qualified, so it's a knowledge-based decision,
there is an uncertainty aspect associated with
Then, we have to define acceptable
variability for that product, and we rely on the
compendial or other standards to do that.
The risk in this case, I am talking about
risk from a clinical perspective, is the prior
knowledge that come from NDA.
Need for bioequivalence assessment for
oral products is that next question, and again the
same words from the Orange Book is if you do not
present a known or potential bioequivalence
problem, acceptable in vitro standard is fine. But
you saw the debate with dissolution, we often don't
go there, we often go to a bioequivalence study.
If you go to in vitro standard compendial
dissolution test method, if you go to a
bioequivalence standard, then, the
variability is our bioequivalence standard, 90
percent confidence interval for test or reference
ratio for rate and extent of absorption is within
80 to 125 percent.
It has to be adequately labeled and
manufactured in conformance to cGMPs, and in that
case, acceptable variability is what we tolerate in
terms of deviation or specifications, and so forth.
So, that is one way of looking at trying
to partition uncertainty, variability, and risk, so
that we can formulate the right questions.
Clearly, I think the quality-by-design
thinking is intended to focus on the intended use,
and design is about doing things consciously. I
showed this slide to you before.
I think what we would like to consider is
in the context of pre- and post-approval changes,
generic drugs, and even extending that to the
concept of follow-on protein
primary goal of a scientific decision framework--I
am not talking with a regulatory process--the
decision criteria that we bring, need to understand
the complexity and uncertainty, but the decision
process should be consistent.
I think that is the fundamental basis that
we work under. Furthermore, I think our goal is
also to identify and eliminate unnecessary human
and animal testing for this decision framework,
keeping in mind that most bioequivalent studies are
done in normal healthy subject volunteers, new
drugs and so forth.
If we can avoid exposing normal healthy
subject human volunteers, it is desirable, and that
is part of the regulation. I will share that
regulation with you.
So, Topic 2 today, the premise that we had
in mind was that a quality-by-design approach via
pharmaceutical development information can
potentially provide an excellent means to address a
number of challenges previously discussed at ACPS
meetings without complete or satisfactory
resolution, for example, bioequivalence of highly
variable drugs, bio-in-equivalence criteria,
pharmaceutical and therapeutic equipment of locally
acting drug products, such as topical products.
Today, Lawrence will bring his thoughts to
you, and these are our preliminary thoughts, and I
think we just wanted to put our preliminary
thoughts on the table to engage you and engage the
community to help us think about this, so that as
we spend our time thinking about this, we already
have some feedback and we are also on the right
Rob Lionberger will come back. He
presented a decision tree to you before, but he
will recast that decision tree for topical products
in the context of quality-by-design.
Yesterday, we focused on dissolution
testing, and as the past chair of the BCS Working
Group, I took it upon myself to sort of go back and
re-examine what were my personal thoughts and what
held me back to make the recommendations that you
see in the guidance and see how we can
discussion into the quality-by-design thinking, so
that I can take this discussion and give that as a
recommendation to the current Technical Committee
on Biopharmaceutical Classification System.
Again, as I said, these are initial
thoughts, and our goal is to engage you in a debate
and discussion to hopefully give us some
perspective are we on the right track. The three
topics today are:
How can pharmaceutical development
information help to extend the applications of
BCS-based waivers of in-vivo studies for immediate
How can pharmaceutical development
information be utilized to address the challenges
of highly variable drugs?
How do we use this to establish
therapeutic equivalence of topical products?
Those are the three topics that we would
like to present, and the general question is are we
on the right track, but then more detailed
recommendations on how we should proceed
three topics or other topics that we should have
considered instead of this.
So, that is the Topic 2.
DR. COONEY: Thank you, Ajaz.
We might pause for a moment for any
questions particularly around clarification of
these opening comments from the committee. Art?
DR. KIBBE: The question I always struggle
with is how do we define highly variable drugs.
Are we defining them in clinical outcomes, because
then the dosage form might not be involved in it at
DR. HUSSAIN: The definition hinges on the
bioequivalence drug, the variability that we
estimate from the bioequivalence drug.
DR. KIBBE: But that could be a function
of intersubject variability, subject population
selection, and have nothing at all to do or
minimally to do with the actual product that you
DR. HUSSAIN: That is the discussion
Lawrence will bring to you, so if you
question for Lawrence.
DR. KIBBE: Thank you.
DR. COONEY: Ken?
DR. MORRIS: The two comments is that the
Topic 2 premise, the other part of that premise is
that the proper development information is being
generated at the companies which, of course, you
have limited control over, and that is being
shared, just as a caveat.
DR. COONEY: Nozer?
DR. SINGPURWALLA: Two comments. Slide
DR. HUSSAIN: I am sorry, I don't have
DR. SINGPURWALLA: I know. Components of
the Challenge. The second bullet. That second
bullet is wrongly worded, it has to be changed, and
I will tell you why.
The more important reason is you, on your
eight bullet, are talking about confounding of
uncertainty, variability, and risk. They should
not be confounded. Who is confounding them and
DR. HUSSAIN: The current system has a
tendency to do that. That is what I mean.
DR. SINGPURWALLA: But that simply means
that the system is not educated enough, because
variability is the cause of uncertainty, is one of
the causes of uncertainty, and risk is a measure,
is a way to measure uncertainty and its
So, why is there so much confusion about
these very elementary ideas in the industry and
perhaps in the pharmaceutical community?
DR. HUSSAIN: I don't know how to answer
DR. SINGPURWALLA: Well, they need to be
DR. HUSSAIN: But let me propose this in
the sense, uncertainty is not risk.
DR. SINGPURWALLA: I agree with you. I
agree with what you are saying completely. What I
am asking is, what is the cause of this confusion,
and it is so easy to remove this confusion?
DR. HUSSAIN: I understand the concern you
are expressing, and my premise is for years we have
not utilized the pharmaceutical
development, and we
have treated that as an art rather than a science,
and that is the way to get away from that
confusion. So, that is the premise.
DR. COONEY: Any other comments at this
point? We will have ample opportunity for further
Using Product Development Information to
Extend Biopharmaceutics Classification
DR. HUSSAIN: Let me go on to the
Biopharmaceutics Classification System. In
preparing for this presentation, I actually went
back and reviewed all the publications that have
occurred in this arena in this discipline, in this
topic area for the last five years, and there has
been a tremendous number of progress in this area.
For example, more recently, Professor Les
Benet's article was published on how you can
actually start predicting metabolism, and
and how you can sort of add that additional
There has been a paper published in Pharm
Research on quantitative instead of, you know,
rigid boundaries, and so forth.
But instead of sort of trying to summarize
the progression signs, what I wanted to do was to
go back and re-examine my own thoughts that were
expressed to the Advisory Committee in the year
2000, so I am actually going to repeat an old
presentation, but in light of what we have learned
in the last four years.
My goal here is to share with you some of
the concerns we had when we proposed the BCS-based
waiver guidance in the year 2000, and to what
extent those concerns remain, and to what extent
quality-by-design may be able to alleviate some of
this concern, and the discussion with you, I intend
to use that as recommendation to the current
Technical Committee on BCS. So, that is the game
This is an old presentation
minor modification. When I had made this
presentation, I was completely focused on risk, and
the title was "Biopharmaceutics Classification
System: A Risk Management Tool."
In light of the learning that I at least
personally had, I want to sort of bring in the
uncertainty and variability components to this.
Since this is a presentation, probably my
last presentation on the BCS topic before I handed
over the reins of responsibility to Lawrence and
Mehul, the new BCS Technical Committee was formed,
when Helen asked me to move to OPS and the PAT
process got started, so my focus went to PAT for a
reason which connects back to this one.
So, this BCS Technical Committee has been
in place under the leadership of Lawrence and
Mehul, and they have been diligently addressing a
number of implementation issues trying to
coordinate all the submissions, and so forth, and
there has been significant activity on this
guidance on the new drug side, very little, if any,
on the generic side.
You also heard from Mehul the database is
now almost ready, is being audited, database and
prospective research for extensions,
links to PQRI
and FIP, but the PQRI program really didn't take
off, and our thoughts were we wanted to explore
extension of BCS-based biowaivers to Class III and
Class II drugs.
Further research at the FDA, which we
completed, and we did extend the BCS-based
biowaivers to "fed" bioequivalence studies, and
that was part of the thing, and that work was done
with collaboration with Tennessee.
Continuing education initiatives and
practitioners and public, and the group has been
busy. International harmonization was an aspect,
but to the extent the definition of high solubility
and rapid dissolution, we got into ICH Q6A, the
European Guideline also adopted much of the BCS
recommendation to some extent. There are certain
differences, though, it is not fully harmonized.
With that as a background, let me trace
for you the evolution of the
recommendations in the
BCS guidance that we released in the year 2000.
Regulatory Bioequivalence: An Overview,
from my perspective, this is a graphical
representation of our regulation.
If you look at the dosage form that we
deal with, solutions, suspensions, chewable
tablets, conventional tablets, and modified release
products, for solutions, we consider bioequivalence
essentially is self-evident, bioavailability is
self-evident, and biowaivers are granted, condition
being excipients do not alter absorption, and that
is an assessment based on historical data.
For any product that contains drug in a
solid form, we have a concern, and for pre-1962
drugs, we call DESI drugs, in vivo evaluation for
bio-problem, that was the original biopharm
classification system, if you really look at it,
that had many of the elements of therapeutic index,
PK, the solubility, and so forth.
For post-1962 drugs, generally, in vivo,
some exceptions with IVIVC. Then, we introduced a
SUPAC-IR guideline in '95, and we
elements of BCS guidance, BCS system in that to
give a waiver for minor changes and moderate
changes that Mehul talked to you about yesterday.
For modified release, we don't have a
classification system, bioequivalence has to be
demonstrated in vivo with the exception of SUPAC
modified release for within a product. If you have
in vitro bioequivalence, you can make changes.
Again, Mehul summarized that.
I want to trace back the discussion to a
bioequivalence hearing, which I did not attend. I
was just graduating in '86, but I was connected to
this because I made the slides of a number of
people who presented here, so I knew what was
This was a pivotal discussion and I think
set the stage for what evolved as bioequivalence
standard. There were two comments that I want to
share with you.
One was Dr. Bob Temple. He said, after
the end of this discussion, it seems sensible to
think that swallowing something that
turns into a
solution rapidly would be difficult to lead to
differences from one product to another.
So, the clinicians were arguing you don't
need biostudies for everything. Arnold Beckett had
made that argument years at that time, so he said
you shouldn't go with in vivo for everything, but
Milo Gibaldi, an eminent pharmacokinetic
professor, "I have learned that there is no support
here for attempting to provide such assurance
solely with in vitro data." So, that was a pivotal
aspect, I think, and I went back and sort of tried
to examine the thoughts and the concerns that were
expressed at that session.
The other aspect that I do want to put on
the table is need to reduce our reliance on in vivo
bioequivalence studies. Why? Ethical reasons. 21
CFR 320.25 says, "No unnecessary human research
should be done." Science continues to provide new
methods to identify and eliminate unnecessary in
vivo bioequivalence studies.
Focus on prevention,
into products - right first time." So you see the
PAT initiative and how this will connect to that
was in the thought process and why we aggressively
moved in that direction, "right first time," I used
Time and cost of drug development and
review is a key, because if you see that we have
three to six bioequivalent studies in our NDAs. We
actually at some point said we don't even review
some of those because they are redundant, so why
expose normal healthy subjects to a new drug which
is under development with all the risks associated
So, prior to SUPAC-IR/BCS, in vivo
bioequivalence assessments to justify a majority of
manufacturing changes. So, this was a significant
hurdle, and that changed. In the SUPAC-IR guidance
in '95, we brought in the classification system and
provided a tiered approach for changes based on in
For example, highly soluble, highly
permeable drug, the critical processes
emptying, dissolution is not likely, and
dissolution is not likely to be rate limiting, but
we said 0.1 single point, 85 percent, and so forth.
So, you can see that for each class, we
recommended a tiered approach for waiver of
biostudies for minor changes, and so forth. We
excluded the Narrow Therapeutic Index drugs from
waiver consideration, but we never defined what
narrow therapeutic index was, and we still haven't.
The guidance in 2000 really extended that
and put that as a waiver for first time approval,
and also provided the methods to classify your
drug, and so forth. The pivotal recommendation in
that was waiver for in vivo bioequivalence studies.
I do want to stop here and say the title
of this guidance was debated to the nth degree
before we agreed on this internally. The word
"waiver" was to signify that we want an in vivo
study for everything that is in solid, so the title
was very carefully chosen to reflect that.
Anyway, it's waiver for in vivo
bioequivalence studies is recommended for
oral test product that exhibit rapid and similar in
vitro dissolution under specified conditions to an
approved reference product when the following
conditions are satisfied: products are
pharmaceutically equivalent, drug substance is
highly soluble and highly permeable and is not
considered to have a narrow therapeutic range and
excipients used are not likely to affect drug
The class membership, the boundaries that
you see, which are rigid, high solubility, the
highest dose strength is soluble in less than or
equal to 250 ml of aqueous buffers over the pH
range that we had 1 to 6.5 or so, whatever that
thing is I forgot.
The reason for 250 ml is the glass of
water that we take when we do a bioequivalence
study. High permeability, the extent of absorption
in humans is determined to be greater than 90
Rapid dissolution is 85 percent dissolves
within 30 minutes in three different
HCL, pH 4.5 and 6.8 buffers using traditional
settings of dissolution apparatus.
Now, clearly, I had approached this as a
risk to bio-in-equivalence because since we started
with the premise that you needed bioequivalence
trials for approval of changes, and so forth, so
the risk factor for me was the proposal the
recommendations should not result in
The risk factors that we had in mind were
clearly manufacturing changes, poor process
capability, high between and within batch
variability, but the thing we focused on, reliance
on in vitro dissolution tests and how reliable that
is, single point specification, sampling,
predictability were the issues.
Clearly, the other factors were there,
deficiencies in BE study design, Type II errors,
and so forth.
Now, assessment of risk, what is the risk
of bio-in-equivalence between two pharmaceutically
equivalent products when in vitro
comparisons are used for regulatory decisions?
That was the question we asked.
What is the likelihood of occurrence and
the severity of the consequences?
Severity was not meeting the
bioequivalence criteria was unacceptable, but what
was the likelihood, so we needed some information
Regulatory decision, whether or not the
risks are such that the project can be pursued with
or without additional arrangements to mitigate the
risk, and clearly, acceptability of the decision,
is the decision acceptable to society? This took
We started working on this in '96, and if
you think I was busy with the PAT presentations
around the globe, that is exactly had to do the
same thing for this one, too, because the mind-set,
the paradigm was so entrenched in the old system.
Professor Gordon Amidon spent some time
with us, he and I. I had the luxury of having the
biopharm document room right outside my
the Parklawn Building, so we went through a number
of applications, about 160 applications at that
point, to get a sense of what is happening.
Roughly, what we found was on the new drug
side only, because we have failed studies or we
have all the studies submitted on the generic side,
we couldn't use that database because you just have
the passing studies in there.
So, we looked at the new drug side and
said when does dissolution signal
bio-in-equivalence or does not signal
bio-in-equivalence. What we found there was
generally, you see big differences in dissolution,
no difference in blood levels.
But, on the other hand, there were signals
that dissolution fails to signal bio-in-equivalence
about 30 percent of the time, and we wanted to ask
So, minimizing risk of bio-in-equivalence,
does in vitro dissolution process emulate in vivo
dissolution process in vitro and in vivo? Dosage
form disintegration, dissolution, and
were the concern.
The gastrointestinal fluid volume,
composition, and hydrodynamic conditions were the
concern, and clearly, I think one thing which was
pivotal for the oral discriminating part was the
surface tension, and that could have been picked
Residence time in the stomach and small
intestine were an issue, so we did a lot of
analysis actually of gastric emptying and what
factors affect gastric emptying, and so forth.
Impact of excipient differences on GI
physiology and drug bioavailability were the
The key question was how well this
emulates in vivo, because this is our standard
This was a cartoon that I prepared and to
take a look at typical physiologic parameters in a
single dose fasting BE study. We had fairly good
estimates of the gastric fluid plus the 8 ounces of
We knew what the gastric pH range is
generally in the normal subject.
We had the information on the gastric
emptying time, which is highly variable, but
approximately T50 is 15 percent. The permeability
is low, and that was an advantage in the stomach
compared to small intestine, the surface tension is
lower, and clearly, volumes in the small intestine
were uncertain, and pH, and so forth, and the
permeability was high. Hydrodynamics was a big
question in our minds.
Lawrence summarized to you the debates
that we have had for dissolution for the last 30
years, and that dissolution tests are over
discriminating, on one hand, and in the USP, the
statement that products that dissolve about 70
percent in 45 minutes have no medically relevant
bioequivalence problems, what was the basis of
Dissolution tests are not sufficient to
assure bioequivalence, and demonstration of IVIVC
is necessary, but when you do that, product
specific, so those are two sides of the debate.
I showed you this slide of the problems
with the dissolution tests of false positives and
false negatives, but then we also looked
that we made decisions on.
Here is a product. The generic product
was Product B. We actually withdrew this product
from the market after approval. This is a pre-62
drugs, it was approved on the basis of dissolution,
meeting the USP specification criteria.
This was a pre-62 drug, one of the older
drugs, and we had a challenge from the innovator,
which is Product A, that the generic is not
bioequivalent, and that was the basis at which we
had withdrawn this product, Product B, from the
The Cmax, you can see is clearly high, but
in many ways, Product B was more reliable, less
variable, and it was more modern technology, but
the constraints on us is you have to be equal, if
Here are examples of where there were real
dissolution impacts on in vivo
absorption. Here is
a weak acid where the initial formulation for
clinical was capsule. They went to
wet-granulation, and the to-be-marketed was a
direction compression with dicalcium phosphate, and
the dissolution in this case was Q17 30 minutes in
simulated intestinal fluid.
That's the criss-cross you see if you do
acid in alkaline conditions, you don't distinguish
that, and this had to be reformulated. But this
was I think in my mind a signal that we probably
are designing our products for dissolution rather
than the intended use.
I wanted to walk through this with you,
and that was one of the reasons for the
Here is a Drug X. This is actually a
clinical study, 100 mg dose, so it's a highly
soluble drug by all definition. It's a weak base
exhibits a sharp decline in solubility with
increasing pH above 3.
Now, the clinical trial material in this
case was wet granulation, drug particle
size of D50
of 80 microns, and this was a concern, because our
particle size specification was very crude. I mean
what does D50 percent give me. We had lactose,
microcrystalline cellulose, and so forth. You see
that formulation there.
But the point to focus there is the
dissolution 0.1 normal HCL was 65 percent in 15
minutes and 100 percent in 20 minutes.
Disintegration time was 10 minutes.
The way I had presented this, the
to-be-marketed was the formulation of direct
compression, but actually the wet granulation in
this case was a U.S. formulation, a formulation
using U.S. clinical trials, and the European
clinical trials were done with the to-be-marketed,
and we had to bridge those together.
The to-be-marketed formulation, you can
see what happened here. Direction compression,
drug particle size of 300 microns, dicalcium
phosphate, and so forth, and the dissolution is
more rapid now, 0.1 normal HCL - 85 percent in 15
minutes, about 95 percent in 20 minutes,
initial dissolution burst is very rapid,
disintegration time is 1 minute.
Clinical product exhibits poor dissolution
at 7.4. Can you imagine, I mean this is a BCS Class
I drug. The Cmax or the rate or the exposure of
this in terms of p concentration for the
to-be-marketed formulation in this case was half,
so you see half the blood levels in terms of
height. So, it simply was signal that if you don't
get the physicochemical properties of drug
understood, you will have these problems.
So, in vitro and in vivo dissolution,
dissolution methods evolved over the last 30 years.
The year 2000, I said there were reproducible test
methods for lot-to-lot quality assurance, so you
can imagine my surprise of the calibration, which I
was not aware of at that time.
The dissolution media volume and
composition selected to maintain "sink" conditions.
In vivo dissolution is a complex process, and you
have to consider pH profile, bile concentration,
motility patterns, and so forth.
In vivo, the "sink" condition is created
due to intestinal permeability, and this was a
contention, which Lawrence and others, we
published to show how permeability actually impacts
in vivo dissolution, so we have published on this
I will talk to you about in vitro-in vivo
correlation. The formulation specificity of IVIVC
has been known since 1997. This is drug
spironolactone from a publication in J. Pharm
Science in '97.
So, you can see a change in particle size.
You may have a correlation, but a change in
particle size could be outside that. So, a
correlation itself why this formulation specific
has to be really brought into context.
So, reliance on current dissolution
practice can pose an unacceptable level of risk
from bio-in-equivalence perspective. Compared to
high solubility drugs, risk is higher for low
Products with slow or extended
profiles pose a higher risk. The dissolution is
So, in a sense, we wanted to use
dissolution test only to rule out that dissolution
is not rate limiting. So, that was the basis for
our thought process. So, we constructed a rapid
dissolution criteria for that purpose. We did not
want to use dissolution tests for bioavailable
decisions if there was a hint that dissolution is
Potential for significant differences
between in vivo and in vitro "sink" conditions
higher for low permeability drugs, which we had to
prove later on with a simulation study that
Now, to establish a boundary for rapid
dissolution, we simply postulated that since
gastric mucosa does not have high permeability to
drugs, you have a 15-minute time, so you can take
advantage of that.
So, if the dissolution is rapid, then,
much of it is complete before it empties
small intestine where you have high permeability.
So, in a sense, a very rapidly dissolving drug will
behave essentially like a solution, and it does.
So, here is a snapshot of dissolution
versus AUC and Cmax ratios and the bioequivalence
goalpost of a drug metoprolol. The reason I did
not take the name off here, because this is already
an ACPS presentation, it is already on the website.
So, you can see a solution versus all the
other formulations that we have approved, generic
and innovator, plus there are research
formulations, which we deliberately made to be very
You see that essentially, for the most
part, the slope is zero. We then did extensive
simulation work to establish that if in vivo
dissolution is rapid, as a function of different
gastric emptying as a function of mean intestinal
transit time, you are not likely to see a
difference between solution and a tablet, and that
was the basis for our 85 percent and 30 minute in
vitro criteria for rapid dissolution, but
not apply that to low permeability drugs because of
the risk factors that we felt were coming from
The question we asked was is the current
approach for evaluating excipients for decisions
related to biowaiver for oral solutions sufficient.
That is the database I have.
There were hints that the excipient
effects were not fully appreciated. For example,
there was a study by Ian Wyling's group where you
could show mannitol, 2.3 grams of mannitol clearly
had a big impact on bioavailability of cimetidine,
a low permeability drug, and on the other hand,
Fassihi had shown that a high permeability drug had
minimal impact of sorbitol, even 10 grams of
So, that was a hint, and we actually
conducted a study at the University of Tennessee,
and we have now completed that study, even getting
to a mechanistic basis for generalizing the
permeability effect to other excipients, is to test
Metoprolol was our boundary drug for high
permeability, so we did a study with ranitidine and
metoprolol, and we did a crossover study
replicate design to get an estimate of subject, the
formulation interaction also.
So, it is a very simple formulation. You
have your drug. You have sucrose or sorbitol, and
you have 15 ml of water. That is the simplest
What we found--I will just show you a
picture--just a confirmation for a low permeability
drug like ranitidine, a dramatic effect.
Now we have completed the study of the
dose response, the amount of sorbitol that triggers
this is about 1.2 grams. For metoprolol, the Cmax
was affected, but not to the same extent. AUC was
In addition, there were a tremendous
amount of information coming out of other excipient
effects on transporters, and so forth. So, this
was an evolving issue at that time, and it
continues to be evolving issue, and
were being proposed of in vitro evaluation of this.
I won't get into that.
But we also did an extensive evaluation of
excipients and what we found was I think in most
cases, excipients that are used in solid dosage
form really do not have a significant impact, but
the way we had to evaluate that was comparing the
differences in formulation that we have approved,
like, for example, verapamil, and so forth.
But the risk factor was excipients. Is
the current approach for evaluating excipients for
decisions related to biowaiver of oral solutions
sufficient? Well, I think we left the question
For BCS-based biowaivers, a higher
standard was adopted by limiting biowaivers to
highly permeable drugs. Excipients used in solid
oral products are less likely to impact drug
absorption compared to liquid oral products,
because it was simply the volume and amount in
For example, we had products on
that contains 23 grams of sorbitol in one dose, so
you can see cross interaction possibly.
High permeability attribute reduces the
risk of bio-in-equivalence, decreased small
intestinal residence time by osmotic pressure,
because low permeability generally have a tendency
to be absorbed in small segments of the intestine.
Clearly, on the other hand, the boundary
that we chose for high permeability to be 90,
because there are other surfactants, and so forth,
that could increase permeability, so if you set
your boundary at 90, there is no risk of failing.
There were other examples. There were so
many examples that we had not really paid attention
to. Here is a submission--not a submission--this
is a graph that a student of mine sent me from a
company, and they have seen such effect.
Here is a drug, a tablet and a solution.
The solution has almost half the bioavailability of
a tablet, so you can see that, with sorbitol or
mannitol for a low permeability drug, it can have
So, that was the basis that we came up
with the recommendations and the boundaries for the
BCS classification system that gave the
waiver for new drug applications.
So, Class I drug, you have jejunal
permeability. This was our research that we
classified the number of drugs, and the volume of
water required to dissolve the dose on the x axis.
Class I dissolution in vivo is not likely to be
rate limiting, and well characterized excipients.
So, dissolution itself is likely to be rapid
inherently, and then we can rely on in vitro
dissolution for that purpose.
Class II dissolution is likely to be rate
determining and complex in vivo dissolution, and
solubilization process, so no, not going there.
Class III was where the debate was. Some
hesitation with the use of current dissolution
test, because the site specific absorption was a
concern, and excipients.
Class IV was generally problem drugs with
in vitro dissolution may not be reliable.
So, that was the basis for our
So, wrapping up, in terms of quality
design thinking, what can we do now? If I
summarize my concerns, one major concern was if we
went towards a dissolution-based method,
will design products for dissolution rather than
the intended use was a concern, and that example
sort of illustrates that, the one to-be-marketed
difference, that was a concern.
So, I wanted to feel comfortable having
some formulation assessment as part of this
extension. Clearly, I think excipients and the
transporters, all were evolving issues at that
time. There is a lot more information available
now than we had.
So, in quality-by-design thinking, you
really have focus on what are the critical
variables that affect dissolution, and these are
easily identifiable especially for immediate
release dosage form.
You easily can start thinking
excipients and what their impact might be on
solubilization, and so forth, and Lawrence himself
has done some work in this area, and so forth.
So, with that in mind, what my thoughts
are, in addition to thinking about evaluating the
boundaries themselves, I would like to recommend to
the group that, first of all, BCS should be, and
probably will be, a key, too, in quality-by-design
decision trees that we talked about.
I mean solubility, permeability
characterization has to be a starting point for the
formulation, so clearly, I think we need to build
these concepts in the decision trees we talked to
you about yesterday, but also quality-by-design and
design space with respect to pre- and post-approval
by bridging studies. Waivers for in vivo studies
based on design space concept, sort of is that
connection to extension concept.
The challenge I think is from a generic
industry perspective, there is a lot of hesitation
to seek for approval, the first time approval, a
biowaiver based on that, concerned with
permeability assessment, and so forth.
That concern probably will remain.
It is not a scientific concern, it is a
perceptional concern, it's a regulatory concern,
and so forth, but that doesn't say that you cannot
classify a drug for a generic product after
approval. The rest of the post-approval changes,
they could be based on that, and that could be
quality by design.
I think those are the key connections that
if the Technical Committee sort of starts building
in, their efforts really get connected to the PAT
and the quality-by-design thinking, so you see the
connections sort of evolved.
Clearly, they have already started
developing in FDA's knowledge base, a knowledge
base. Drug-excipient interaction, I think is an
increasing issue from chemistry and clinical
pharmacology aspect, and I think you need to start
connecting those dots.
At the same time, drug substance and
formulation variables and clinical
Mehul alluded to the variability, but with what is
happening on the clinical side now, with focus on
biomarkers, focus on surrogate markers, and so
forth, I think we need to proactively keep an eye
I mean this is an evolving issue, but seek
out some connectivity between quality and clinical,
and be available to what is happening at least, and
that is the point I made also to Jurgen yesterday
when he gave his report.
So, with that, let me stop and open it for
DR. COONEY: Thank you, Ajaz.
Comments and questions from the Committee?
DR. MEYER: While I am formulating my
question, I will ask another one, so I will give
you some time to think.
How confident are you that our knowledge
base on excipients allows a reviewer to sit back
and say, well, it has X, Y, and Z, and therefore,
there is no problem?
DR. HUSSAIN: Well, I think the
traditional excipients for immediate release dosage
forms that we use as formulation aids for process
ability, and so forth, I am fairly confident that
there are very little concern there.
There are other excipients that are
necessary to aid in the dissolution process, such
as surfactants and other aspect. I think a close
grouping of those would be necessary. I think if
we collect this information, it will start making
the case, but if you have properties, such as high
solubility, and so forth, you probably would not
need those anyway.
So, I think you can carve out excipients
that we know are not an issue.
DR. COONEY: If I can just pick up on
Marvin's point for a moment, your initial question
was about knowledge base of the excipients per se,
but it is really the relationship of the excipients
to the drug substance, to the API, that seems to be
the area of uncertainty, and that knowledge, it
seems to me, is much less clear.
DR. HUSSAIN: If I amy sort of put that in
the context, traditional screening experiments that
are done for drug excipient compatibility, may
provide not only information that will be useful
for stability, failure modes of the product, but
also hints about the interactions among excipients
and drugs, how they might have a bearing on
DR. COONEY: Ken.
DR. MORRIS: The point you just raised is
the source of my question, as well, since
drug-excipient interactions are typically for
Is it like particularly for something like
BCS Class III, is it more of the concern that the
interaction with the drug is changing absorption,
or that the interaction of the excipient with the
mucosa or the sites of absorption?
DR. HUSSAIN: I think they are both
concerns in the sense, but the concerns we had when
we were working on this were more on the impact on
the GI membrane, transporters, and so forth, the
concerns with excipient-drug interactions
might be physicochemical were less of a concern,
because we did not really focus on that aspect.
DR. MORRIS: That is sort of where I would
assume it, but I just don't know enough biopharm.
DR. HUSSAIN: The aspect I think is this.
I think the draft guidance that we have issued on
polymorphism, for example, I think is a
concentration there, in the sense what we have said
is you could have a different polymorph, but if you
can design your product well and if it meets the
criteria, it's fine.
So, I think that is the flexibility that
already sort of comes through that, is that ability
to demonstrate that, you can be different, but yet
meet the intended use.
DR. COONEY: Art, then Marvin.
DR. KIBBE: This is more complex than we
can handle in one or two days. The number of
excipients you use exponentially increases the
possibility that one excipient is reacting with
another excipient, that is reacting with the API.
On top of that, the processing
product changes things. I mean we know all sorts
of problems with mag. stearate and overblending,
but we know certain issues. The question that
always sits in the back of my mind is we have been
discovering these issues on a regular basis over
the last 20 years, have we discovered them all, and
how can I be naive enough to think I have
discovered them all, and I don't think I have.
So, that gives me a basic uncomfort level
with just waiving stuff when I really want a part
that works in my patients. So, I am uncomfortable
I have a question, a substantive question.
If I make a soft gelatin capsule which contains a
solution, is it a solution or is it a capsule?
DR. HUSSAIN: Capsule. That's the way it
DR. KIBBE: On your basic I guess third
slide, I could argue it's solution.
DR. HUSSAIN: Yes. That's the aspect, I
think now we can start thinking about those aspects
if you have not done so in the past.
But let me go back to the concern you
raised in the sense, impact of magnesium stearate,
and on dissolution, it is clearly
Impact of magnesium stearate on in vivo absorption
has not been done yet.
All the studies we have done, we had no
impact of magnesium stearate on immediate release,
and so forth, on in vivo absorption. I could not
find a single paper that conclusively tells me that
what we see in vitro, the big difference is
translating in vivo differences.
There are two reasons for that. One, is
that old study that was published in 1967, J. Pharm
Science, by Professor Newton from the University of
London, where he demonstrated for lithium carbonate
that if you include a very, very small amount, 0.01
percent or 0.001 percent of sodium sulfate in your
formulation, you negate the effect of magnesium
stearate that you see in the solution.
So, that was a hint to me that suggested
that the surface tension differences that we see
between in vivo fluids and in vitro
are the reason, because all the studies we did at
the University of Maryland, we actually probed this
for a low solubility drug--I am forgetting the name
of the drug--we didn't see in vivo relevance of
So, now that is the reason we have to
start thinking about is risk-based decision to
really understand the behavior of things in vitro,
because one of the concerns that we had earlier was
you see big differences in vitro, how do you know
this will or will not translate.
If quality-by-design, we are thinking, why
is this assessment, then, that provides a basis to
think about it.
DR. KIBBE: But that argument, I think
would logically lead us to the conclusion that we
have to go to a bioavailability study, we have to
go to a clinical trial. We can't rely on any of
the standard tests that we do that are surrogates,
because they don't work out, because they either
show a problem that isn't real, or they ignore a
problem that is real.
DR. HUSSAIN: Right. That second bullet,
that is what I am really thinking about. If you
have to qualify your design space, your
bioavailability studies, if you are a new drug
applicant, that becomes your test of hypothesis is
to say that we have looked at these are the big
differences we see that has an impact.
So, one category of BCS-based biowaivers
would be SUPAC related where you have demonstrated
this in vivo, and that becomes the basis for that,
and not just rely on in vitro testing and lack of
So, waiver is an extension of SUPAC in
terms of design space is a bigger opportunity
probably in the quality-by-design thinking.
DR. COONEY: Marvin.
DR. MEYER: Ajaz, I think I asked this
question yesterday, and I think you said you were
going to address it and maybe you did and I missed
it, the rigid fixing of the--
DR. HUSSAIN: Boundaries.
--boundaries. I really don't
have any problem with the rigid definition of high
solubility, high permeability, I mean we have
pretty well nailed that down, but then you have, of
course, if it's soluble in greater or less than or
equal to 250, well, if it's only soluble in 300, is
that really poor solubility, and if it's only 89
percent absorbed, is that really low permeability,
and, if so, does it fall in that 30 percent
probability of failing a product?
How do you deal with--you have to draw the
line, but on the other hand, you draw the line, it
becomes somewhat arbitrary and capricious.
DR. HUSSAIN: Very good point. This is I
think an important point because the objective of
this guidance was to make the decision. This is
the decision. You meet this, there is no issue.
If you don't meet, you always have an option to
explain, but nobody uses that option.
So, this, in my opinion, is an approach
that we had before. In the new paradigm, the
decision trees that we developed opens the door in
Here, the decision is pre-made, but
instead of premaking that decision, how can you
allow your science to drive a decision process that
can justify the recommendations that come, but that
then becomes specific to a company. It is not a
general guidance. It is a decision tree to arrive
at a proper decision.
So, that would be an extension concept for
BCS, not a general decision recommendation, which
is what we have been trying to achieve. It has
changed the boundaries, and so forth. But to
incorporate that as a decision process, it becomes
demonstrate this, and the decision can be yours
sort of a thing.
DR. MEYER: So, it's sort of a work in
progress, so to speak.
DR. HUSSAIN: It's not. I mean I think
the group has been busy with a number of things,
but this isn't a thing that they could start
thinking about, we have not done so.
DR. COONEY: Ajaz, if I can get clarity on
that point. I think the point you just made is
that the decisions on class membership
integrated into the thinking about the decision
DR. HUSSAIN: Yes.
DR. COONEY: Is that correct?
DR. HUSSAIN: Definitely. If you are a
Class I drug and you exhibit the rapid dissolution
with the conditions we have outlined, the decision
is okay. Anything else, the guidance does not
recommend a waiver. That's how it is.
Based on what Marvin just suggested, and
what I am formulating that as, this is a decision
for every sponsor right now. Their design and
process understanding would vary from one end to
the other end, but one way of extending this
concept is not a general decision that this is
where you get the waiver, but to define a decision
tree and how you demonstrate to the degree of
confidence that we need, that waiver would be you
have demonstrated an understanding that the waiver
will be granted.
That will be Class I, that will be Class
II, that will be Class III, so you have
levels of complexity in those, but the signs and
the level of understanding could drive you to that,
but that probably will become a post-approval as
part of the design space.
DR. COONEY: Tom.
DR. LAYLOFF: I think I am not confused,
but I don't understand some things like when a drug
goes into the intestinal space, it is bound, not by
water, but probably proteins and various other
things that are present in the medium, and then it
is absorbed through different sites depending on
how it is wrapped in with the rest of the medium,
and that is a drug-specific property, which then
can be affected by an excipient, which might change
the transport site, it may change the structure of
the solution characteristics, is that correct?
DR. HUSSAIN: No, I think the basic
premise is this, yes, you can have binding, you can
have a number of other complexation reactions, and
so forth, that occur. Many of those are ionic and
loose, so you establish equilibrium.
For some drugs, you have
really are almost very tight like with calcium and
tetracycline, and so forth, there are a few such
examples, but in this scenario, what we are talking
about are equivalent behavior of the same drug
molecule in two different formulations.
So, if there are intrinsic properties of
the drug molecule itself that will contribute, but
that molecule is the same, that we are dealing with
two formulations. Now, how do formulations act
with that behavior is a concern.
I will sort of extend that concern to a
paper that we had, a poster that we had, is that of
precipitation. A weak basis will dissolve very
rapidly in acid conditions, but when they get
emptied, there is a potential for precipitation,
and so forth, and that could be a very complex
process, and the size of the particles, not
precipitation, crystallization may differ based on
the excipients you have and the conditions you
There is a potential that excipients could
impact that. So, that is generally Class II drugs,
that's the boundary for Class I for high solubility
was intended to prevent some of those things from
DR. LAYLOFF: Do you think the
complexation and coordination around the drug
substance would actually affect the transport site,
change the site of transport, would change the
properties of the system?
DR. HUSSAIN: Yes, it is clearly possible,
but unlikely for an immediate release dosage form
with the type of excipients we use.
DR. MORRIS: You know when I think about,
it sort of makes my head hurt, but when I think
DR. HUSSAIN: It's complex.
DR. MORRIS: --the amount of time we spent
working on developing design spaces for the
processing end of things, which as Jerry says, may
be a way off, but still in comparison relatively
simple to the larger problem, is there an
opportunity in the context of using development
data to somehow leverage tox studies to
be able to
get early reads on, not the tox itself, but in
terms of some of the dynamics that are going on
with the dosage forms?
DR. HUSSAIN: I have a decision tree,
which I did not present yesterday, but it was part
of the handout. That decision tree came out of our
AAPS workshop on how to leverage that. The paper
is published, Diane Burgess [ph], Eric Duffy from
FDA is on it, so it is there in your handout. I
don't have it in this one. Take a look at it.
That leverage is every bit of information
coming from Pharm Tox, and so forth, to start
building that case for that.
DR. MORRIS: For the design of the dosage
DR. HUSSAIN: Yes, for particle size
dissolution and bioavailability concentration.
DR. COONEY: Marvin.
DR. MEYER: Ajaz, when a generic company
sends in their ANDA, it was my understanding that
the generic group does not go back to the NDA to
review the contents of the NDA, so they
at the excipients and see which excipients are now
different in the ANDA than were in the NDA product?
DR. HUSSAIN: I will let Lawrence answer
that, but we do have a process of inactive
ingredient guide that we consult, and so forth. I
put him on the spot here.
DR. YU: I guess this morning we talk
about excipients, which emphasize how complex the
process is. Yes, with advances in molecular
biology, we discovered I even don't know how many
transporters going on. As far as the PGP, at least
32 and 64 is transporters, however, I want to
emphasize that how those transporters impact
absorption we rarely see in clinical settings.
In other words, we very see excipients
impact on absorption of Drug A, B, C, D, but in
vivo setting, we have very, very few, two or three
publications out there compared to tens of
thousands of publications to show that excipients
impact in vitro.
So, I have to say that we still want to
see more evidence to show the impact of
on absorption of drugs in general.
Secondly, while we see the impact exceeds
absorption, we very open to see the unique of some
of the products out there. The reality for, say,
70 or 60 percent of immediate release products,
people espouse the intensity, use very limited
number of excipients, I would say 10, within 10.
For example, Avacel almost uses the majority of
products. All those excipients impact, and have
not seen in vitro, as well as in vivo.
Certain, because of those common used
excipients, since we have a sufficient knowledge to
judge whether they are going to impact absorption
or not, will generate and not see the formulations,
however, in very few cases, some cases, we suspect
potential impact of excipients absorption, we will
look into it further before we make any scientific
decisions about approvability of any NDAs.
I hope that answers his question.
DR. HUSSAIN: The other aspect, just to
build on what Lawrence said, traditionally, the
composition, especially for immediate
is hardly any different than the quality.
DR. MEYER: My real question was do you go
back and look at the NDA to see if Pfizer used
Avacel, and Teva used who knows what, do you make
that comparison, say, well, wait a minute, they are
putting in two things instead of Pfizer's one
Can that potentially make a difference?
Do you review the NDA product composition?
DR. YU: Well, certainly, we will review
any scientific literature out there and information
available to us to make the best decisions.
DR. MEYER: But do you review the NDA?
DR. HUSSAIN: Marvin, often we don't have
to, because it's in the label.
DR. MEYER: Well, that's true. It didn't
used to be.
DR. HUSSAIN: But definitely, the criteria
there is to look at what has been approved and what
has been used in dosage forms and inactive
ingredient guide, and so forth.
DR. YU: I guess the answer is as long as
are trying to build in the science base or any ANDA
we will use any information which is available,
whether scientific literature or not, to us.
DR. COONEY: Are there any other questions
at the moment? Thank you, Ajaz.
We are running a bit ahead of schedule. I
think this would an appropriate time to take a
break for 15 minutes. We will reconvene at 8 past
10:00 and then begin immediately with Lawrence Yu's
DR. COONEY: I appreciate everyone's
diligence to staying on time. It has worked very
Lawrence Yu will proceed with a
presentation on Using Product Development
Information to Address the Challenge of Highly
Using Product Development Information to Address
the Challenge of Highly Variable Drugs
DR. YU: The assignment to me today, this
morning, is for me to discuss how to use
pharmaceutical development information to
or potentially address the bioequivalence issues of
highly variable drugs.
Before I go on and talk about how to use
or potentially use the pharmaceutical development
information to highly variable drugs by equivalency
issues, I want to give you a very brief overview or
update what has been happening before.
For highly variable drugs, this is not the
first time, it's the second time we present it to
you. In the first presentation on April 14th of
2004, we discussed the challenges and the
opportunities for bioequivalence of highly variable
At this meeting, the objective was to
explore and define bioequivalence issues of highly
variable drugs, for example, what is called highly
variable drugs and discuss potential solutions to
deal with the bioequivalence of highly variable
We invited a number of speakers from
industry, academia to address issues related to
bioequivalence including why the bioequivalence of
highly variable drugs is an issue, highly variable
drugs a source of variability by Gordon Amidon from
the University of Michigan, and the
implications of highly variable drugs by Leslie
Benet, and from bioequivalence method include the
skin method by Laszlo, as well as bioequivalence of
highly variable drugs, we had a good discussion at
I want to highlight some of the things
which have been discussed at this meeting,
particularly the slides by Professor Leslie Benet
from the USCSF. His talk with implications of
highly variable drugs, the argument was why highly
variable drugs are safer.
Specifically, he said for wide therapeutic
index highly variable drugs, we should not have to
study the excessive number of patients to confirm,
to demonstrate that two equivalent products meet
the preset statistical criteria or by
This is because, by definition, highly
variable approved drugs must have a wide
therapeutic index, otherwise, there have been
significant safety issues and lack of efficacy
during Phase III.
Highly variable narrow therapeutic index
drugs are dropped in Phase II since it is
impossible to prove either efficacy or safety.
Now, for the benefit of some new members
for this committee, I have two slides to briefly
review why this issue, why the one-size-fits-all,
what we are using today.
In order to determine bioequivalence, we
normally define as a rate of bioavailability,
defined as a rate and extent of drug absorption.
Bioequivalence is defined as absence of significant
difference in the rate and extent absorption.
In practice, when we give the drugs
orally, for example, to a healthy volunteer, we
will draw the blood. We got the plasma
concentration profile. We are certainly not able
to get exactly how much and how fast drug gets
absorbed, therefore, in practice, we use AUC, area
under the curve, to represent extent of absorption.
We use Cmax as a surrogate for the rate of
absorption, certainly in some cases we also look at
Tmax, because indeed, if you look at Tmax and Cmax
here, it represents the rate of absorption.
So, from that, we will define what the
bioequivalence study is passing or not passing.
Basically, the bioequivalence criteria, either
statistical criteria here is 80 to 125 percent.
At this date, that is the
one-size-fits-all regardless drug, drug product,
regardless of therapeutic class, regardless for
anything, that bioequivalence study, you have to
use preset, so-called bioequivalence criteria,
which is 80 to 125 percent.
Now, let's look and explain why the highly
variable drug is an issue. Let's look at the red
one. If you use a highly variable or intersubject
variability is high. Statistical confidence
interval, if you use the same number of
when variability goes higher, the confidence is
going wider. When confidence gets wider, it
becomes more and more difficult pass the confidence
interval or bioequivalence interval if 80 to 125
So, that explains when the variability
goes higher, it gets more and more difficult to
pass a study.
On the other side, in order to narrow the
confidence interval, for example, here it is fair
to demonstrate bioequivalence for super red one
here in order to make confidence interval narrower,
you have to use a large number of subjects, because
the higher the variability, the higher the
confidence interval, the higher the number of
subjects in general, the narrower the confidence
Therefore, for highly variable drugs, you
will have to use higher number of subjects. Just
to give you example here, for example, normally, we
have a 20 percent or 30 percent intersubject
variability. You maybe use 18, 24, even sometimes
for good product or good drugs, you only need to
use 12, actually, they can pass the bioequivalence
But this is not always true, because when
the variability goes higher, now, this variability
could be because of a drug, or it could be because
of a product, so think about if variability goes
100 percent--some of you think 100 percent, that is
unrealistic, but we do have a drug, we do have
examples--think about with 100 percent variability,
assume test and the reference, there is 5 percent
difference, you end up it could be 500 or more
subjects, or 300 subjects, so this is certainly a
large number of subjects in order to pass the
So, Leslie argued at the previous meeting,
from the clinical perspective, this is not
necessary. To give you a real example, these are
slides from Leslie Benet. Now, you would argue,
you may ask how do we get intersubject variability.
Certainly, you could get this number from
literature or sometimes company conduct a
study, get some kind of estimate how many subjects
need to be used to pass the bioequivalence study.
Of course, in this case, based on
intersubject variability, you need to use 300, now
this is the drug.
So, at the previous meeting, when we
present the issue to you, and also we present some
of the possible potential solutions including
widening the confidence interval. Now, that is
very straightforward. You said by confidence 80 to
125 is too narrow for highly variable drugs, and
your intuitive thinking is just widening the
confidence interval, that is one of the potential
Another solution is a scaling approach, in
other words, based on the variability of reference
list product, reference list drugs, and calculate,
use statistical approaches to calculate the
confidence interval, then, to determine whether the
study is passing or not.
Obviously, I have to say this. The active
approach because the confidence interval
in order for the study to pass, so let's widening
You came in and asked to do that, that
certainly the scaling approach, we ought to
carefully look into, the Committee suggested--a
quote here--"the need to understanding where the
variability originated. The members added that
prior knowledge of all biostudies may help set more
appropriate specifications or criteria to make
So, you suggest that we have to understand
the origin of the variability. Now, to look at the
mechanistic understanding of variability for drug
substance obviously is the same, reference list
product and the generic product, or any other
product, but the potential difference could be
formulation. Certainly, the generic products could
be narrower or could be wider, the variability.
We believe at this point, in order to
understand the origin of the variability, that
pharmaceutical development report, or
pharmaceutical development information
can help us
understand the source of variability, can help us
make rigid scientific evaluation.
Now, in order to see the utility of
pharmaceutical development report to evaluation or
reviews of NDAs, let me go back and review some of
the basic fundamentals or the premises for ANDA
Ajaz has mentioned about therapeutic
equivalence. Basically, the products are considered
to be therapeutic equivalents only if they are
approved as safe and effective, they are
pharmaceutically equivalent, they are
bioequivalent, adequately labeled, and manufactured
according to cGMP.
Now, here, I want to emphasize the
pharmaceutical equivalence. When we define
pharmaceutical equivalence, we basically have the
same active ingredients, obvious. I know we are
managing about pharmaceutical solid polymorphism,
which was presented to you two years ago, has a
drugs guidance out there and published by FDA in
December of 2004.
You have to be same dosage form, same
route of administration, and identical in strength
and concentration, and may differ the
characteristics, such as shape, excipients,
packaging, and so on, and so forth.
Now, under the same dosage form, I think
what the complexity of dosage form is particularly.
Yes, I would say several decades ago, that dosage
form is reasonably simple and in most cases I would
say the immediate release product or solutions.
Certainly, with advances of pharmaceutic
industry and the pharmaceutic technologies,
so-called dosage form gets more and more complex,
and we now have the soft gel capsules, we have
ricin [?] product, we have inhersion [?] product,
presents additional challenge to us in terms to
make scientific decisions, in terms of make
We therefore would believe a
pharmaceutical development report,
quality-by-design, designed to be equivalent,
become more and more significant in this regard.
Why does pharmaceutical equivalence
matter? Because of user experience and
expectations. Then, bioequivalence test is
normally conducted in healthy subjects. Certainly,
we have assumption that equivalence in healthy
subjects equals equivalence in
patients. Now, we
have many, many generic products out there which
are safe, which is a high quality, which are
effective, which is the equivalent to the reference
So, certainly, we have tremendous
experience with that, and certainly the
pharmaceutical equivalence presents more and more,
become because you want to make sure the data from
the healthy volunteer does the equivalent in
patient, and against novel drug delivery systems
presents a challenge.
That is why we want to use more and more
pharmaceutical development approach to make a
judgment, pharmaceutical development information to
make a scientific judgment.
Highly variable drugs very
often have, as
I mentioned, a wide therapeutic index, and the
clinical trials of reference list product have
established the acceptable level of variability,
because I said otherwise, these highly variable
drugs, a big window index, they will be dropped in
So, under an ideal situation, you will
think about variabilities very high, so you will
think it should be easier to pass, easier to design
equivalent product simply because they are so wide,
the target is wide, so it is easier for you to
Obviously, as I said before, if you use
the preset 80 to 125 bioequivalence confidence
interval, it is not the case. While we explore our
tentative approaches to deal with the
bioequivalency issues, certainly, the design issues
So, now, how do we deal with
pharmaceutical development for highly variable
drugs? Obviously, sponsor need to understand what
are reference products supposed to do
of variability, and the purpose of design can be
equivalent, and to evaluate and to verify the
design and hopefully, in the future, use the
bioequivalent study design for highly variable
So, we put more emphasis on design in this
regard to establish pharmaceutical equivalence, in
order to establish therapeutic equivalence, which
will be more appropriate.
While we are looking for shared
information for generics with us, there is a reason
for doing that, not only for evaluation for highly
variable drugs, certainly for pharmaceutical
development is required. It's one of the CTD
format. It is also outlined in ICH Q8, although
they do not apply to us, but I think some
principles should apply to generic industry also.
Also, more significantly, OGD question
based on review. Now, this is still a work in
progress, but I want to share some questions, I
think it is important to ask to share.
What is the formulation
intended to do?
What mechanism does it use to accomplish this?
Were any other formulation alternatives
investigated and how did they perform? Is the
formulation design consistent with the dosage form
classification in the label?
So, those questions will help us get
information about a pharmaceutical product, the
report will help us, pharmaceutical product design
and development, make more sound and appropriate
scientific determination or evaluation.
The question often comes up, why do we
need to provide those things to the FDA? That
again is a quality-by-design paradigm, and
pharmaceutical development report is where you
demonstrate the drug is highly variable.
Now, this demonstrate not necessarily to
study, but you certainly use any source available
to show why this drug or drug product is highly
variable, and may use a different criteria other
than 80 to 125 percent confidence interval.
Also, the pharmaceutical development is
where you justify equivalence of design,
why do you
think the product which you designed is equivalent
to the reference list product.
During the discussion, the members ask
whether it is drug or drug product. Now, for
example, Product A, the variability is the active
ingredient into exceptions, so formulation design
could be rapid release, so demonstrated by
dissolution comparison under physiologically
relevant conditions, if this is BCS Class I drug,
which is highly soluble, highly permeable, even
though they are highly variable, you may still
require biowaiver, otherwise, you will have to
conduct some bioequivalence studies to demonstrate
that they are bioequivalent.
Certainly, the approaches to deal with
highly variable drugs, to deal with the
bioequivalence of highly variable drugs are still
in discussion and still in investigation. I am
hoping in the near future we share with you some
proposal or recommendation we have with respect to
bioequivalence of highly variable drugs.
Another drug could be drug
drug product could be highly variable, even drug
substance is I would say low variability, and
certainly design for equivalence begins with the
characterization of the reference list product, and
generic product should target the mean, and the
current system again would have no reward for
narrow or less variability of generic products.
That is why we need to explore the alternative
approaches or more appropriate approaches to deal
with highly variable drug products.
I just want to give you some examples of
what we talk about here. This is real data. This
is single subject replicate design, in other words,
you give the same product to the same patient
twice. Here is the plasma level, obviously, I am
sure that out of 80 to 125 percent confidence
interval, by any standards, it probably does not
need a statistician to figure this out.
You can see here, this is the first period
or this is the second period. It is not in your
handout or printout because this is in color. If
you look carefully, these two curves are
significantly different, probably different by I
even don't know how many folders.
This is a single-dose study twice,
replicate study design. Sorry, I should do a
better job next time, use red, so you can see it.
DR. KIBBE: I am just looking at the curve
and wondering why we got the hump at the back end
and whether the product is intended to have a
DR. YU: No, it's simply by design, for
whatever reason this peak has come out. Obviously,
a second dose, this peak is no longer there. So,
it's not purpose designed, it's simply because of
This is happening because enteric coated,
this is coated to release at the target pH, so when
the physiological pH in the gastrointestinal tract
may fluctuate, and those curves will change.
Think about, for example, if we have a
product designed will release a pH 7, so then in
the terminal ileum, at one point is pH 6.8, you
will not see the release. But a second day,
because of food or because of other reasons,
terminal ileum pH becomes 7 or 7.2, you do not see
the release. Otherwise, you will not see it.
So, simply pH effect or significant impact
DR. KIBBE: The product had gotten on the
market because it worked clinically?
DR. YU: Yes. Even though we see the
significant variation in pharmacokinetics, but we
have no reason to believe this variation will
impact safety and efficacy.
So, in order to do more appropriate
pharmacokinetic studies, we also look into what
additional information, for example, develop
information will help support those cases or
bioequivalence cases, because you can see the
bioequivalence obviously is very difficult to
conduct variability probably up to 2 or 100
percent, and the number of subjects very high.
So, we want to see can we use any
additional pharmaceutical development information
to help us to make more proper scientific
Again, for example, when we are looking,
in many cases, we do get very consistent, the in
vitro dissolution actually out to say the majority
of cases, those help us out to make a more proper
scientific decision or rational scientific decision
when we recommend any method to demonstrate
bioequivalence, but occasionally, we do get very
strange results, and actually, the variability is
extremely high and does not help you.
I just want to show you another case here
when we conduct the dissolution under physiological
relevant pH condition, and you get dissolution all
over the place.
Now, this is a 6 tablet, same lots, same
bottle, put in 6 vessels, you get a distortion
So, the next question we ask, this is a
large variability because of the operator or
because of other reasons. I think the answer is we
are almost certain those difference is because of
product, not because of other factors.
So, what I can present to you today is we
have challenges to deal with bioequivalence of
highly variable drugs. We use the clinical
evaluation and sometimes we are also facing
challenges when we are trying to use in vitro
information to help us make decisions.
DR. SINGPURWALLA: Lawrence, what is the
difference between each curve, different vessels?
DR. YU: Yes, six vessels.
DR. SINGPURWALLA: Six vessels, so it
could be that the vessels are different.
DR. YU: I think I stated that the
variability because other reasons, for example,
vessel difference, media difference, degassing
difference, operator difference, assay difference.
We do not believe all these reasons can explain.
DR. HUSSAIN: Yesterday, this same figure,
Cindy actually showed you the reason for this
difference was the coating thickness, and so forth,
so this is the same slide.
DR. MORRIS: You wouldn't get 2 1/2 hours
difference in dissolution time from
vessels. That is not the magnitude you would
DR. SINGPURWALLA: How am I supposed to
DR. YU: You have to believe in me. You
don't have any other options.
DR. SINGPURWALLA: I don't believe in
DR. KIBBE: This is a constant pH
throughout, right, we haven't shifted pH during the
process or anything, right?
DR. YU: Correct.
DR. HUSSAIN: The reason to believe that
is I think it was done by Cindy, and with our
stringent mechanical calibration.
DR. LAYLOFF: He demonstrated it with
variable coating. It's variable coating on enteric
coating material, so if there is a crack in the
coating, it disintegrates much more rapidly.
DR. SINGPURWALLA: I think I believe Tom.
DR. YU: Thank you very much.
So, the objective, the case we presented
to you is certainly difficult, I just want to say,
variability issue, whether from clinical
or sometimes for in vitro conditions, in vitro
testing, target main performance question. I am
sure you will ask where is the main performance.
I just want to show you that these are the
challenges which we are facing, and certainly we
are open to any suggestions or input from you.
So, in summary, we believe pharmaceutical
development information will help. I quoted here,
that's the conclusion made by you April 14th of
2004. Understanding what the problem is, as well
as the real fundamentals, for example, physical and
chemical parameters, and make coherent and
scientific science-based decision based on
pharmaceutical development information, I think I
present to you the cases to see hopefully how we
use pharmaceutical development information to help
us in most cases, but in some cases, we still have
challenges and we have opportunities for us to move
Thank you and any comments are welcome.
Thank you very much.
DR. COONEY: Thank you, Lawrence.
We now have time for questions from the
DR. SINGPURWALLA: Lawrence, I have two
kinds of questions. Question No. 1. Is it the
purpose of this presentation of yours to ask the
manufacturers, namely, the industry, to provide
more information to you because there is so much
variability and you are trying to get to the source
of the variability, is that the objective?
DR. YU: Yes, very precise, certainly much
better than I said.
DR. SINGPURWALLA: That is the political
question. The scientific question, and I have heard
this before, what does T/R percent mean in your
Slide No. 8?
DR. MEYER: Test over reference.
DR. SINGPURWALLA: Test over reference.
DR. YU: Yes.
DR. SINGPURWALLA: How was this 80 percent
and 125 percent figure arrived at?
DR. YU: Slide 9. I am trying to get
DR. SINGPURWALLA: That's it, the picture.
So, 80 percent and 125.
DR. YU: T is the test.
DR. SINGPURWALLA: No, forget that. How
did you get 80 and 125?
DR. YU: That's an excellent question, and
we have been asked many times.
DR. SINGPURWALLA: It can't be excellent.
DR. YU: It's back to it was published
when I was in high school, I would say, 20 years
ago, or even more than 20 years ago, when the
pharmacokinetics, the discipline was developed, and
FDA developed the criteria. Actually, this
evolving process and trying to develop what kind of
standards or criteria can we use to judge a
bioequivalence study is okay or is not okay.
I think at that time, the physicians get
together, as we do today, and the physicians
together made the determination that the
percent, the difference between product would not
be considered clinically significant, because the
20 percent will not be considered significant
difference, therefore, when translated into in vivo
setting, you have 80 percent.
So, you would think from 80 to 120 instead
of 25. Now, in the normal processing of
pharmacokinetic data, they used log normal to be
much better to describe the distribution. So, when
you use log normal, 80 is still 80. When you have
the 1 over 80 or 1 divided by 0.80, equals 1.25.
That is why you see 80 to 125.
Now, at the beginning, I would think 20
percent instead of 19 percent or 21 percent, which
is 20 percent, it was decided. Then, the question
come back to us now can we change 20 percent to 25
to 15, 10, 5 percent, and I guess we have to use,
say, over the 20 or 25 years, we approved product,
they are all safe, they are all equivalent, they
are all high quality, because of those experience
or prior knowledge, determining 80 to 125 percent
Now, this does not necessarily mean we
cannot change it, but the criteria we have is very
stringent criteria, we feel confident
Now, with a statistical interplay--
DR. HUSSAIN: Lawrence, if I may.
DR. YU: Yes, please.
DR. HUSSAIN: It's a "feel good" criteria,
we felt good about it.
DR. SINGPURWALLA: I got the answer. I
think I got the answer.
DR. SINGPURWALLA: The answer is
DR. HUSSAIN: No, it's rational science.
DR. YU: It is rational science. I think
I proved it.
DR. SINGPURWALLA: Let me make a
DR. YU: Yes, please.
DR. SINGPURWALLA: That tradition with
some dose of rationality was good 20 years ago when
you were in high school. Times have changed.
These kind of decisions to either prove equivalence
or prove in-equivalence should be based on risk
considerations and should be based on appropriate
So, I think it is time to change, and I
think I said that April 14th, 2004. Has there been
any progress made towards changing?
DR. YU: The answer is yes.
DR. SINGPURWALLA: Oh, good. What?
DR. YU: Certainly, you said you want
suggestion of change, and I think under the
leadership of Gary Buehler, that we are exploring
the confidence interval, for example, the window
index drugs, and also we are exploring confidence
interval for highly variable drugs. In other
words, in the future, I am hoping someday, with
your support and agreement, we will have different
criteria other than 80 to 125 to different class of
drug in consideration of the risk interplay.
Obviously, to make any changes, six months
or one year is not enough.
DR. COONEY: Marvin, then Ken, then Paul.
DR. MEYER: Your talk I believe tried to
marry the quality-by-design to the highly variable
drug and show that you could, in part,
problem by quality-by-design, that's the objective.
DR. YU: Yes.
DR. MEYER: Personally, I think if you
have a competent company, then, your highly
variable drug is biological problem which the
company can't solve. You have to speak directly to
a higher power to get rid of that variability.
So, I think, yes, there is cases where,
for example, I could cite failure by design if you
want to put an enteric coating on something,
because that is, in my view, not a good dosage form
because it is so dependent on gastric pH and
emptying, and all of that, so you are setting
yourself up for failure.
Now, you can say, well, I dealt with
quality by design by not using enteric coated, I
kind of took the reverse of that. A competent
company looking at Slide 24, the 6-vessel graph,
would never go to a biostudy with a
showed dissolution characteristics like that.
So, indeed, some quality built in that
says whoa, let's not spend $100,000 on a biostudy
when our drug is all over the map in dissolution.
So, I think you can deal with some variability, but
that is fairly straightforward I think for a
So, the issue that really faces us is the
physiological variability and do we extend the
confidence limits, do we have point estimate
restrictions or just do we do 600-subject studies.
DR. HUSSAIN: Marv, may I just sort of put
that in context a bit? In some ways, what we are
seeing here is this. Since we are comparing two
formulations of the same drug, the drug is the
same, the variability, the physiologic, the
variability that is coming is the same for the drug
If we can compare formulations and say
that all the conditions that are critical to
exposure are well controlled, and so forth, and get
confidence, what will give us the confidence
that the inherent variability is the physiologic
variability, not the quality variability, then, we
can move forward. I think that is the hope that we
DR. MEYER: Do you think practically, you
can look at the restrictions and the SOPs--
DR. HUSSAIN: No.
DR. MEYER: --and just see how a company
is formulating and designing and developing a
DR. HUSSAIN: Not with the traditional
work we do about formulations, putting things
together, and so forth, no, it has to be a
structured design approach that goes through
identifying the sources of variability in your
materials, and so forth, and putting a convincing
case together to say based on the assessment, in
this case it's a generic product, and based on
characterization of reference material and your
test product, you can make the case that the
variability that you are seeing in your product is
no more different than of the best argument.
That gives you a leverage to now make a
rational decision with respect to what sort of a
biostudy criteria would be necessary.
You can build flexibility, and not go
rigid with, say, the Japanese approach, which was
in your background packet, was to say do we really
need confidence interval criteria here. We just
want to confirm the mean values. It's a
confirmation rather than a complete full-fledged
study. One option could be that.
DR. YU: I think the message we are trying
to convey is when we explore alternative approach,
which could be a wide confidence interval, or your
scaling approach to show or to demonstrate the
bioequivalence is demonstrated with the additional
information, which is pharmaceutical development
information, will help us to make scientific
Right now we don't, we don't have those
informations. In other words, we are not able to
see how dissolution variability here may change it,
for example, in this case, if we change
dissolution is very beautiful, so that is the data
we got. We have now seen this data I showed you on
the screen. Thank you.
DR. COONEY: Ken.
DR. MORRIS: A couple of points. One is I
agree with Marvin in the sense that you wouldn't
expect a company to release dissolution, I mean
going to a biostate with dissolution like that, but
I think those studies were done under different
conditions. These were done here, so they wouldn't
have seen that under normal dissolution conditions.
My more general question is--
DR. YU: You are correct, yes, in normal
conditions, especially, for example, USP
dissolution, maybe you are not able to see.
Actually, dissolutions are beautiful.
DR. MORRIS: Right, so that comes back to
sort of our discussions yesterday in a sense. The
question I have is to what level or to what extent
do the ICH initiatives, I mean including the CTD
and Q8, impact on the ANDA, I mean is there an
intent that they follow suit with NDAs?
DR. YU: Obviously, the basic principles
from ICH and CTD, the CTD cure document for drug
substance and drug products, it is not just
only, for both NDAs and ANDAs. ICH Q8 is, in
principle, a part into NDAs, certainly the basic
principle also apply to ANDAs.
The way I actively look into this to
document and to see what information will help us
to make scientific decisions. Certainly, as I said
before, we are not looking for information which is
nice to know, we are looking for information which
is essential to know.
DR. MORRIS: I guess to that end, because
this is something, of course, we have been
discussing for several years, but the idea that
rather than having checklists of what the companies
have to do, if they can make scientific decisions
based on the intended dosage forms and the
properties of the API, which should be a lower
hurdle, I mean that should be known more by the
time you get to the generic.
DR. YU: Yes.
DR. MORRIS: Instead of having to do a lot
of the other testing that might normally be done,
if they can focus on the identification of the
critical to quality attributes of the product and
capture that in a development report, it seems like
that is a reasonable way forward.
DR. YU: That is correct. In fact,
industry is coming forward and they share some of
the pharmaceutical development report with us, we
are actively looking into this to develop some kind
of review templates which will incorporate
pharmaceutical development information into our
Again, I said we are looking for
information which is essential to know, not nice to
DR. MORRIS: Maybe this is for Paul, is
that a reasonable stance as far as how you look at
DR. FACKLER: I am not sure exactly what
you are asking.
DR. MORRIS: I can clarify if you want,
but basically, if you could, instead of having to
do sort of checkbox testing, if you could do
testing that was largely prescribed by your need to
establish certain scientific issues, rather than
having to do as many, let's say, sort of--what is
the word--statutory testing, if you will, is that a
reasonable stance for you guys?
DR. FACKLER: I don't see a problem with
that. What I didn't hear here was that there are
any different statutory requirements for highly
If a generic company still needs to pass a
bioequivalence study, and we are going to assume
that the pharmaceutical equivalence is simple, I
don't understand what the generic company
understanding the origin of the innovator's
variability has to do with the approvability of a
lot of material that is shown to be
pharmaceutically equivalent and therapeutically
equivalent through a bioequivalence study.
I guess that is the piece I am missing.
Why is the burden on the generic company
understand the variability of the reference listed
drug, and what value does that have if really all
the generic company needs to do still is
demonstrate a bioequivalent product?
DR. MORRIS: You are talking about BE
variability now, not pharmaceutical?
DR. FACKLER: Yes.
DR. YU: Paul, if we use
one-size-fits-all, which is 80 to 125 percent to
some of drugs, you may have difficulty to pass the
confidence interval. So, when we are exploring the
alternative approach including the scaling
approach, you will have to demonstrate this product
is highly variable or not highly variable.
You have to know that because otherwise,
suppose someday in the future, if the scientific is
mature enough, we have a scaling approach, for
example, for highly variable drugs, your submitted
application did not show these are highly variable
drugs, how would we know these are highly variable
So, you have to show, in your
report, that is a highly variable drug before we
DR. FACKLER: Agreed, but wouldn't a
replicate design bioequivalence study inherently
capture the variability of the reference listed
DR. YU: Yes, if you choose to do so, use
replicate design, certainly, you are able to
demonstrate that reference list product is highly
variable or not.
DR. FACKLER: But that is already part of
an ANDA application is my point.
DR. YU: I guess, Paul, we have not
reached a consensus or we have not made a
determination you have to use replicate design.
DR. MORRIS: Is part of that the fact that
you are still struggling with the concepts that are
entailed in that dissolution plot where you can't
factor into the pharmaceutical variability, factor
the pharmaceutical variability from the clinical?
DR. YU: I guess the struggle we have here
is, look, Lawrence, in order for you to
direct for pass, whether you use scaling approach
or you use widen the confidence interval, you
simply widen the confidence interval, let them to
pass. You need to explain why. You need to
explain why you think that is a feasible approach,
you think that is scientifically sound.
So, when you say explain why the
pharmaceutical development report can help us
provide additional information to explain why.
DR. FACKLER: I agree that certainly you
need to understand the variability of the reference
listed drug especially if a generic applicant is
claiming that the variability is an issue for this
DR. YU: Correct.
DR. FACKLER: I am not sure what value the
steps that were taken has to that determination of
variability. Variability sometimes is listed in
the label for a reference listed drug; other times
applicants do replicate design studies or run
reference versus reference to measure that inherent
variability, but that would all be part
application already, as I understand it.
DR. YU: Yes, in many cases actually
lately for some of complex dosage forms. Dosage
form, we very often sent many, many deficiency
letters. Actually, company provide information
during the cycles, and as I said, at the GPA Chair
meeting, we have four or five or six cycles,
provide additional information to us, and
eventually, the product get approval. I am not
saying you not provide that information.
What I am trying to say is with the arena
of pharmaceutical development report in the ICHQ
paradigm, can you provide that information in the
application instead of for us to send many
When we see the OGD list receive 25
percent or more of the applications every year,
where do you want to put resource into those
reviews. Suppose you provide those additional
information, which I believe will help us in our
reviews, and they reduce the cycles, I see it's a
win/win situation for you and for us.
DR. COONEY: Ajaz.
DR. HUSSAIN: I think look at it from this
perspective in the sense the whole aspect
are trying to make a decision and you are trying to
choose the right measurement system here.
Now, the Code of Federal Regulations
essentially has a hierarchy of methods that you
choose for bioequivalence. Our current criteria is
a PK crossover, PK-based, pharmacokinetic-based
study is the most discriminating one.
So, you are looking at, you are trying to
now judge approvability of a generic drug, and for
that you need to establish its pharmaceutical
equivalence and its bioequivalence. The
bioequivalence measurement system that we have has
inherent variability, and much of that variability
is coming from the measurement system, and may not
be coming from the test samples that you are doing.
So, is this measurement system the ideal
measurement system right now or not? That is
really the question.
The dilemma that we have is the in
characterization and in vitro testing with
dissolution often is not reliable enough by itself
to make that call. If it was, you would not be in
So, if you really then look at it, what
are we saying, is we have information generally
that even if I give this drug intravenously or as a
solution, and so forth, the variability is coming
from the subjects, it is coming from physiology,
which is inherently variable. If I sleep on my
lefthand side or righthand side, it will make a
difference, I mean it literally happens.
So, that is the measurement system, but
then you are putting your product into that system
and trying to see is there a difference of 20
percent or not, and to meet that confidence
interval criteria, you need 600 subjects or 300
subjects, and so forth.
Can we utilize the signs of design to say,
to confirm, not necessarily to have a confidence
interval, a confirmation that the new formulation
actually is not contributing to that
is there sufficient science to do that or not.
If it is, then it opens the door for
saying that the bioequivalence assessment then
could be tailored based on that understanding.
DR. COONEY: Before we go on to some other
questions, I would like to see if your question,
Paul, has been addressed.
I think the question was--well, first,
Lawrence is proposing that there be a
pharmaceutical development report added to the
information that is part of the application, and
you are asking what will be the implications of
providing that additional information and
facilitating the next step, which is approval of a
DR. FACKLER: That is part of the
question. The other part was what would be in a
pharmaceutical development report that isn't
already part of an ANDA. That is really what I am
trying to understand, and, of course, then, what
value would that provide.
DR. COONEY: Is there clarity to that
question? So, that is back to Lawrence, to Paul's
question. What would be in that pharmaceutical
development report that is not already part of the
DR. YU: I thought that was a topic of our
next advisory committee meeting.
DR. HUSSAIN: Let me put it this way.
There is nothing there right now. There is nothing
there to even gauge the aspects of. So, what we do
is our decisions are made based on one batch test
results and the biostudy. That is what it is.
DR. MORRIS: Can I just weigh in? I think
part of this is that a lot of what would go in the
development report is stuff that people are already
doing, but doesn't just get included in a summary
fashion, much like we have discussed earlier, that
there is development studies you do, but you don't
That is what we were talking about
yesterday, is that, as a reviewer, if you have to
try to piece together a development rationale from
data here and there, you end up with sort
development rationale that the person filing really
wouldn't want to be there displayed to the world,
you know, sort of a Frankenstein development
So, if the company does it, then, they can
see the logic that you use. Whether they agree or
not is a different question. So, in my sort of
concept of this, which may be flawed, of course, if
the company, let's say, had used Cynthia's
dissolution method, because they said this is what
has really mattered, and they got those curves to
overlay, then, that is a big step forward to say
that the variability that may come out of the BE
studies are not due to our change.
So, if you see the variability of the BE
studies and you have demonstrated that it is not
due to the lack of adherence to a design space, for
lack of a better word, then, that has got to be as
good as the innovator. That is my concept. This
may be down the road, as Jerry said.
DR. YU: I want to make comments that when
we say the pharmaceutical development
I think I emphasize those information that is
essential to know, not just for nice to know.
We are looking into this, what additional
information will help us in making decisions, and I
think we are happy to share with you in the future,
but at this point, we cannot say that for every
single ANDA or for every single product, you need
the pharmaceutical developed, because you have a
prior knowledge, some of the information already
there, so this need clarification when you are
understanding what additional information is
provided. I think we need to discuss and work it
DR. FACKLER: I understand. To Ken's
point, you start over here and the bioequivalent
product is over here, and sometimes you take a
direct approach to it and sometimes you don't. You
are right, oftentimes it is over here and then you
realize you need to be over here, and then finally,
you get where you need to be.
But I am not sure I understand the value
or what it matters what path you took as
you end up in the right place. All this
information does exist, of course, and the field
inspectors have access to it, and we are just
reluctant to expand the content of an ANDA in the
fear that it will slow down an already overburdened
So, where the information is critical to
understanding whether a product is pharmaceutically
or therapeutically equivalent, of course, it ought
to be submitted, but where it is not essential for
that evaluation, I just question whether or not it
ought to be added to the burden of the reviewers.
DR. COONEY: Art, then Marvin.
DR. KIBBE: Let's get back to what we are
trying to determine, and that is whether or not a
clinician who prescribes this medication for its
effect has got a reasonable expectation of a
therapeutically similar outcome when he uses the
innovator or when he uses the generic. That is
where we are.
If a product is inherently variable, as
manufactured by the innovator, then, we
know that early on, and as Les correctly points
out, if that was the case during development and
prior to approval, it wouldn't make it on the
market if it wasn't that that breath of variability
was allowable for clinical outcome, because if the
clinical outcomes wouldn't--there were times when
there were failures and times when they were
toxicities apart, never gets on the market. which
means that we have already historically established
large variability is okay, because we have that
product on the market.
Now, if I am a generic company, all I want
to do is say that I am going to be no more than, or
perhaps less variable, and I am going to get to the
same therapeutic outcome.
If I can test a replicate design that
shows that my level of variability is lower than or
equal to the variability of the innovator, and my
means are on target, then, I can with reasonable
assurance argue that my product used in the
marketplace on patients is going to have the same
efficacy and failure rate as the innovator.
The second thing is we already have agreed
that dissolution is a hammer when we need a
surgical scalpel to figure out what is
and if you make a shift in a dissolution criteria
and all of a sudden you can differentiate tablets
from the same batch, but that batch used in people
isn't differentiatable, then, you are making a
differentiation which is of no value to anyone
except if you want to go back and process improve.
In fact, that is what it should be used
for. The companies ought to be investing time and
energy in process improvement by looking for better
differentiators for their own internal consistency,
and perhaps they could narrow the variability if
they found them.
I think the justification for going to the
study that you said that if they used the USP
numbers, they would all pass, and going to your
numbers, we have this high variability, but that
high variability doesn't relate to clinical
Now, I am coming on the market
generic. If I can establish that I am not more
variable than they are, and my means are the same
as they are in a biostudy, how much more
information does the agency need? I don't think it
needs much more.
DR. HUSSAIN: Art, you are missing the
point in the sense to demonstrate that your
variability is acceptable, you actually have to do
more now through a bioequivalence or replicate
design, and so forth.
What we are saying is in the sense, there
are ways or there should be ways to sort of the
justification that goes into a formulation that you
move forward, could then become a basis to say you
don't have to go through extraordinary means to say
the variability is unacceptable.
So, if we know a drug substance is highly
variable, you mostly have that information that
says you sort of at least definitely will when you
approve the product, then, the signs of formulation
design could provide you a basis for saying there
is no reason your particle sizes, which
critical for your dissolution, your coating
thickness, which are your release mechanism, are
essentially being controlled, and so forth.
So, why should a generic form then have to
do a large study with replicate or with whatever?
Isn't there an option available for something--
DR. KIBBE: So, what you are really
talking about is a waiver of what we would say
would be a standard replicate design to get around
DR. HUSSAIN: Exactly, so that is what we
DR. KIBBE: So, the company then would
come with its own development data and show that a
broad range of dissolution numbers are not highly
variable or something.
DR. HUSSAIN: Yes, the way I would think
about that is in a sense if it's a tablet, I will
go to the basic mechanisms of what the dissolution
will be affected, and here is my assessment of my
particle size, here is my control strategy, here is
the prior knowledge of similar dosage
is no apparent reason for this to be variable from
So, that becomes a basis for a decision
criteria saying that why would we expose normal
healthy subject volunteers, a large number of them,
to simply get our numbers within the confidence
interval criteria, which is somewhat arbitrary.
That is the crux of this.
DR. YU: I don't know if I can clarify,
the point we are trying to make is that if you can
conduct bioequivalence study now to best pass the
confidence interval, this is good enough. I am not
saying this is not good enough. We are not asking
The problem which we are facing is you
will not have difficulty, it is not impossible if
you have recruited 1,000 or 2,000 subjects, it is
almost impossible to do by a current study, and
this is scenario that pharmaceutical variability
information may come into play and to help us out.
That is what we are trying to convey. Thank you.
DR. MEYER: I think part of my problem is
that I believe what you are putting forth is a
concept without any data, which obviously you can't
have yet, because the concept hasn't even been
implemented, it is just a concept.
I think certainly from my perspective, if
you have some ideas that might streamline the whole
system, I would say go for it and then let's see
the meat once the skeleton is exhumed, so to speak.
That is the bottom line, but I think there
are some other ideas in there that are perhaps
easier for me to understand, characterize the
reference listed drug and then presumably, if you
have done that, FDA will take that into
consideration to explain why you have confidence
limits that aren't up to par perhaps.
For example, a simple example, the RLD has
an overage in it of 10 percent. They claim that
isn't released ever, so they just have it in there
because their release mechanism doesn't allow for
except 100 percent.
You have some evidence that says well, in
fact, it is released 110 percent
sometimes, so the
poor generic company is already 10 percent in the
hole when it comes to AUC. If that can be
demonstrated in some reasonable scientific fashion,
that ought to maybe taken into account.
A better example maybe is with the osmotic
pump. We have done studies where you can harvest
the ghost out of the feces, and sometimes it has 50
percent drug in it, sometimes it has 10 percent,
sometimes it has no drug in it. It seems to be a
direct function of intestinal transit time.
Well, if you are a generic trying to match
without using an osmotic pump, you don't have a
snowball's chance in hell of coming across and
matching a product that sometimes is 50, sometimes
is 100, sometimes is 10 percent.
So, I think as long as you hit the means,
and you bring that kind of data to FDA, they ought
to have the latitude of saying yeah, we know that's
a problem with the RLD, and we can therefore adjust
our thinking when it comes to the generic.
Obviously, that is going to take a fair
amount of work, but I think that these
to be thought of, as well as more statistically
based ways of dealing with high variability. That
is kind of a short-term fix which ultimately once
the statisticians get done fighting, then, the rest
of us can agree, but the other is certainly a
concept worth pursuing, I think.
DR. SINGPURWALLA: I would like to respond
DR. HUSSAIN: If I may, there is an aspect
what Marv just said in the sense a practice that
all of us know exists is when you have variability,
then, you pick and choose what your comparator is.
I mean it bothers me in a sense to say that, you
know, you can pick and choose what lot you will
compare to, and so forth.
Why do we have to sort of have those type
of decisions where, you know, I think we can be
better than that, so I think just to build on what
Marv says, to say that I think we can really be
confident in what we are doing, and not to feel a
bit guilty that we are picking and choosing what we
test, and so forth.
DR. MEYER: As you well know I am sure,
there are a number of countries. You do your
dissolution on three lots of the RLD and
pick the one in the middle, not the one that is
closest to what your product happens to be.
DR. YU: I want to make comments about
Marvin's comments. Yes, in the case here, what you
present, actually, those information is not in the
original ANDA submission, but those information
eventually is shared with us.
So, go through many cycles, many, many
months, or even several years to get us that
information. What we are seeing is that we think
if those information, which you eventually shared,
only a couple that go through the five or six or
seven cycles, shared in first place will help us to
make decisions, will help us to reduce cycles, will
help us actually use the resource wisely. That is
what we are trying to say. Thank you.
DR. COONEY: Nozer.
DR. SINGPURWALLA: General comments.
First thing, Ajaz, don't use the word
limits for those two boundaries. Call them control
limits. Confidence limits are completely
The second thing is you are fighting, at
least there is a lot of discussion because there is
a lot of variability. What you seem to have done
is taken reactive approach, have said variability
is there, what shall we do about it.
Well, yesterday, you talked about 6 sigma
in one of your slides. Well, I think wherever you
have these high variability issues, whether they be
in industry or whether they be within your own
system, I would encourage you to put into practice
what you were preaching yesterday about 6 sigma.
I would say, you know, has anybody thought
about that, because 6 sigma came about in industry
because there was a lot of variability, and they
said how do we control it. Well, you just don't
control it by doing statistical methods. You
control it by proper management and proper
procedures, and I would say that you should try to
bring that into the arena.
DR. COONEY: Paul.
DR. FACKLER: The generic industry is just
as interested in minimizing the number of
6 and 7
cycle reviews on products. Clearly, we have the
same goal in mind.
I guess what I would suggest is that for
highly variable drugs, for instance, it would be
useful for the agency to tell industry the kind of
information that is generally lacking, but with 500
applications a year, or 800, whatever the numbers
might be these days, coming into the agency, I
don't think it is wise to require this information
on all of the applications.
I would suggest maybe we clarify the
additional information that is often being left out
of submissions for highly variable products, and
presumably, generic companies in the interest of
having a minimum number of review cycles will
submit it the first time rather than an iterative
process to give you all the information that you
need to make a fair decision.
DR. COONEY: Gary.
DR. BUEHLER: For the development reports
in general, I thank you for not wanting to
overburden us with additional information. We do
have a lot to look at. If we do get additional
information, we will look at it for sure.
I know that we get some amount
information sprinkled through the ANDAs and I would
think Ajaz was a bit draconian when he said all we
get is the batch record and whatever. I mean there
are explanations. We do demand explanations when
there aren't any deviations from what we normally
see, that is in ANDAs and we do look at that.
Lawrence and a group is working on a
question-based review for the Office of Generic
Drugs. It is a very detailed project. He is
working with experienced reviewers in our office,
and he is developing this in a very stepwise
manner, both first by involving both the
supervisors and reviewers in our own office, and
then at a certain point we want to sort of unveil
it to industry.
We want to make sure that when
we do bring
this new review method and these new requirements
or whatever you want to call them with respect to
pharmaceutical development reports, the industry is
very aware of what we want and why we want it, so
that they will feel good about giving us this
information, and like Lawrence said, it will
hopefully reduce the number of cycles we have, it
will not overburden the reviewers, but, in fact,
reduce the burden on the reviewers, because they
won't have to see the same applications four, five,
or six times, and they will understand why we need
It is also a risk-based system, so that
there are some applications that you won't have to
provide this type of information, because there are
some applications obviously that are easier than
other applications, and the applications for
complex dosage forms and unique dosage forms
obviously, we are going to ask for more information
than for the vary standard solid orals that are
fairly easy to manufacture.
But we are doing this over a two-year
period and hopefully, sometime toward the end of
this year, we will be able to begin to tell
industry what we hope to expect in the future
applications and industry will be comfortable with
DR. MORRIS: I just have a quick question
for Gary. I am assuming that development reports,
as you say, depending upon the complexity of the
dosage form, I mean they can be relatively brief if
it's a very simplistic or simple dosage form, so I
am not so sure that it's the burden if the payback
is fewer review cycles or less clinical studies.
Clinical studies are a lot more expensive than
writing a development report and doing a few more
Is that more or less the case, Gary?
DR. BUEHLER: I am not sure it is going to
be able to be submitted in lieu of a study or
DR. MORRIS: No, I meant the extensiveness
of a development report.
DR. BUEHLER: Some development reports
will say we wanted to develop a bioequivalent
formulation, and, you know, here it is, and it
could be a page or two. I mean clearly, it won't
be very long for a generic, because the goal of a
generic is pretty evident, but other development
reports will be more extensive, so yeah, you know.
DR. COONEY: It sounds like there is a
need for clarity on what will be requested and
expected, and also for clarity on what the
implications of that will be. It sounds like that
will be forthcoming.
Ajax, what I would like to do is move on
to the next presentation.
DR. HUSSAIN: Just go back to the original
intent. Our initial thoughts that we wanted to get
the discussion started, so we never intended this
to make a proposal, so these are initial thoughts
and we are moving forward with this.
If industry wants to be proactive, they
had better start thinking about it and how they can
use this opportunity instead of asking us what do
I think it is equally burdensome on
industry to think about how to develop the products
for the intended use, and make the case, and grab
If not, the system as it stays, we are
perfectly happy with it.
DR. COONEY: So, there is an opportunity
here for dialogue and there is no doubt from the
last 45 minutes that there will be dialogue.
I would like to ask Robert Lionberger to
proceed with the next presentation.
Using Product Development Information to Support
Establishing Therapeutic Equivalence of
DR. LIONBERGER: Today, I am going to be
discussing how to apply the concepts of
pharmaceutical equivalence to topical dosage forms
and look at how this is related to quality by
Here, I am going to focus on topical
dosage forms that are in the local delivery, so not
products such as transdermal products that are
trying to deliver drugs systemically.
In the Office of Generic Drugs, as you
have heard several times before this morning, our
mission is to provide therapeutically
products to the public. When someone uses a
generic drug, they should expect the same clinical
effect and safety profile as the branded reference
Just to summarize some things that Ajaz
talked about in his introductions, the preface to
the Orange Book explains how we do that. Products
must be pharmaceutically equivalent and
bioequivalent. But I want to dig a little bit
deeper into this and ask why do we actually require
both, why isn't bioequivalence by itself enough to
determine that the products are the same.
One aspect of that is that consumers have
some expectation about product behavior. If the
reference product is a capsule, you don't want to
replace that with a solution. So, there is some
user experience and expectation.
So, pharmaceutical equivalence
encapsulates concepts related to like the
interface of the product, but then there is another
aspect to it, and I have tried to express it here,
is that pharmaceutical equivalence supports the
determination of therapeutic equivalence based on
We don't say that just because two
products pass a bioequivalence study, they are
therapeutically equivalent products. An example
might be an oral solution and a tablet. There can
be many products for which those two dosage forms
would be bioequivalent, but we wouldn't say that
they are therapeutically equivalent products.
One aspect of that is that our current
determination of bioequivalence is really very
strongly based on in vivo testing. So, again,
there are limitations to testing. We test these
products in a small population and then we
extrapolate that conclusion to all people who are
going to use the products from all batches in the
So, to sort of back up that extrapolation,
there is some other information. Right now that's
the pharmaceutical equivalence between the products
that supports that.
In the occasions when we do equivalence
studies in patients, there are other differences.
Sometimes the clinical endpoints aren't very
sensitive to small differences, bringing in
examples from topical products, you can imagine
there are cases where, say, a cream and an ointment
formulation might have the same therapeutic effect,
but they wouldn't be considered pharmaceutically
equivalent products or therapeutically equivalent
products even though the clinical endpoint study
might show equivalent efficacy.
Again, from the sort of pharmacokinetic
studies for one of the challenges that is often
made to some of our bioequivalent studies for
topical products is since the skin is a barrier,
you say, well, healthy subjects have healthy skin
barriers. There is a question. Sometimes people
will claim in patients, the skin might be diseased
or damaged, so that is a common concern. There is
a common challenge to some of our
So, inside of the pharmaceutical
equivalence concept, there is some idea of other
things we want to know about the products to sort
of generalize this idea of equivalence.
If you think about this and want to sort
of tie this to quality by design, one way that
might be useful for you to think about this is that
our current definition of pharmaceutical
equivalence might be considered a first step toward
a quality by design.
If you were going to design equivalent
products, the first things you would start with
were some of the concepts that are in our current
definition of pharmaceutical equivalence. You
would want to have the same active ingredient. You
would want to have the same strength, the same
So, if we look at sort of a different way
of looking at our paradigm, maybe instead of a
regulatory framework, a more scientific framework,
what we are doing when we review a
is we are looking to see is the product designed to
be equivalent, and then does it demonstrate
bioequivalence in an in vivo study.
So, you can see sort of this combination
sort of parallels our current sort of regulatory
framework of pharmaceutical equivalence and
bioequivalence leading to a determination of
therapeutic equivalence, where we might say that on
sort of a scientific level, what we might want to
be doing in the future might be to say look at the
quality by design, look at the generic product that
is designed to be equivalent to a reference
product, and then based on this design, choose the
appropriate either in vitro or in vivo
bioequivalence testing for this product to complete
the determination of therapeutic equivalence.
So, I want to bring this sort of
conceptual framework and bring it into this sort of
particular example for topical products. Sort of
to motivate that, I just want to outline some of
the complex issues that we deal with that are
related to pharmaceutical equivalence for
Again, we have a lot more experience with
immediate release, oral dosage forms in effective
excipients, what excipients you can change, what
excipients you can't change. For topical products,
a lot of times the excipients may or may not affect
the barrier properties of the skin and drug
We don't have as much experience about
that, so a lot of times we are worried about what
differences in formulation are appropriate for
comparing a test in a reference product - is a
change in solvent appropriate, what if the base of
the formulation in ointment or cream has changed
from being hydrophilic to lipophilic, how much
water content should there be in the product. You
might affect evaporation, the feel of the product.
A lot of these sort of differences in
formulation get wrapped up into the question of are
two products the same dosage form. I will talk a
little bit more in detail about that in the rest of
We also have questions, when we don't have
good bioequivalence methods for use for topical
products, what indications should be used
clinical equivalence studies. Perhaps the product
has multiple indications, which one is the most
appropriate one to use.
These are the kinds of issues that we deal
with in generic topical products. Some of the
implications of these for the ANDA sponsors are
that the approval times for these products can be
longer. If there are these issues that we don't
have a good understanding internally, we have to
schedule meetings with the appropriate people, have
to have internal discussions.
When the sort of standards aren't clear,
this is an opportunity for the reference listed
drug sponsors to challenge correspondence to OGD or
through the citizen petition process that we have
to address the scientific issues there that aren't
sort of clearly defined.
A lot of times, at the end of these
discussions, we will end up going back to
sponsors and asking them for more information to
help us resolve these issues, and then usually that
is done through deficiency letters to them, and it
ends up with sort of multiple review cycles.
So, as we heard in Lawrence's talk, there
is the question of more product development
information in the ANDA itself may help OGD deal
with these issues more efficiently.
This is sort of very similar to some of
the things that Lawrence talked about, that there
are harmonization efforts underway that describe a
product development report, but I think it is clear
that these are mainly aimed at new drug
applications, so it is not sort of obvious or clear
how these should apply to ANDA sponsors.
I think the theme of this talk to see this
as an opportunity, these development reports, as an
opportunity to provide information that will help
the agency set rational specifications for products
that are complex, for immediate release oral dosage
forms we have various standard systems set in
place, but for topical products, where we
experience, the more information that is provided
about, say, why was an excipient changed, and why
do you know that it is not going to have any effect
could be very helpful to us in making efficient
Again, the product development reports are
the place in the application to emphasize the
quality by design, that the product is designed to
be equivalent. That will help us set the right
requirements for the bioequivalence testing for
This is just a few examples of what some
of these harmonization documents say about a
pharmaceutical development report.
In this case, again, the key part here
might be to establish that the dosage form and the
formulation are appropriate for the purpose
specified in the application, or in the Q8
document, it talks about an opportunity to present
the knowledge gained through the application of
Here, it is talking
sort of formulation and development for the topical
products, that there is information that the
company that is developing the generic product
knows about why they made certain choices in the
formulation. It would be very helpful to us in
deciding that that is acceptable, where the agency
itself has less experience with particular dosage
I have emphasized this concept of quality
by design or, in the case of the generic products,
quality by design means you are designing the
product to be equivalent to the reference product.
So, I want to try to be a little bit more
specific about what that means. There are two
cases. One, the mechanism of release. Clearly,
the mechanism of release between a generic product
and the reference product can be different, but the
intent of those different mechanisms ought to be to
produce the same rate and extent of absorption.
This is the bioequivalence criteria.
Again, we also recognize that depending on
the particular product, that the release
the product might not be the rate controlling step
at absorption. So, the determination of how close
release rates might have to match would depend on
the absorption process involved and what is the
rate-limiting step in the absorption process.
Again, between generic products and
reference products, the excipients can be
different. Again, it is a good thing to understand
the differences between the excipients.
The IIG limits are a starting point. They
tell you that this excipient has been used in this
dosage form up to a certain amount, and that really
addresses, specifically in the case of topical
products, safety-related exposures, so you know
that level of exposure.
The thing that complicates the topical
products is when you change the excipients, the
real question that we often deal with is do the
changes in the excipients to the products affect
the permeation of the drug through the skin. I
think that is the sort of challenging question
there for the topical products that we
have to deal with.
Again, as I said, the purpose of quality
by design is to design the equivalent product. I
want to just give sort of three sort of examples of
this process here.
The first is talking about Q1 and Q2
equivalent products for topical products, and then
look at what happens when you make changes to the
formulation, they become Q1 and Q2 different, and
then this leads into the discussion of issues
related to the dosage form classification and how
product development information might help us make
a better decision or more scientific based
decisions on dosage form classifications.
First, I want to start off with the
definition of Q1, Q2, Q3. So, products that are Q1
have the same components, so both the generic and
the reference product would have the same
If products are Q2, they would have the
same components, but they would also have the same
amount of each ingredient.
The Q3 concept is same components, same
concentration, but here I am saying same
arrangement of material or
microstructure, and this
is particularly important for topical products that
are semisolid dosage forms, so non-equilibrium
dosage forms, where you might have, say, an
emulsion with exactly the same components, exactly
the same concentrations, but say, for example, the
droplet size might depend on how you have
manufactured that product.
So, there is potential differences for
semisolid dosage forms depending on how they are
produced even if overall the composition is exactly
A contrary example would be a solution.
If a solution is Q1 and Q2, because the solution is
at thermodynamic equilibrium, you would be able to
say we know that this product has exactly the same
arrangement of material in the product.
The importance of the Q3 concept is when
you know that the products have the same
arrangement of material, you know that
going to be bioequivalent, there is no question
about that. An example of that is again a solution
where you know that the products are in
thermodynamic equilibrium if the compositions are
the same, the structure and arrangement of the
material is the same.
Unfortunately, for most topical semisolid
dosage forms, they are not necessarily equilibrium
arrangements of material, and so a direct
measurement of Q3 level equivalence is challenging.
So, if we have the topical products where
Q1 and Q2 are identical, again, the only potential
differences are differences in this Q3 parameter,
which can come from differences in manufacturing
processes, because they are not going to be
manufactured by exactly the same process.
We know for particular semisolid dosage
forms, such as emulsions, that rheology and in
vitro release rates can be very sensitive
measurements of microstructure and are related to
So, the sort of idea that sort
advancing here from a scientific point of view is
when the products are Q1 and Q2, that in vitro
tests should be equivalent to ensure bioequivalence
of the two products, because again here, the issues
are detecting differences due to differences in
manufacturing processes, and the argument would be
that in vitro tests are the best evaluation method
to detect whether any differences in manufacturing
process have significant differences in the product
formulation or performance.
Now, things get more complex when a
generic product and a reference product have
different compositions, and this connects with the
dosage form classification, and these differences
occasionally could be barriers to generic
A generic company might want to formulate
products that are Q1/Q2 different because the
innovator has formulation patterns, so there might
be either legal reasons or perhaps manufacturing
process reasons why you might want to formulate a
product that is not exactly identical in
composition to the reference product.
One of the products, again, one of the
members of the Committee mentioned the sort of
economic effect of uncertainty on product
development if you don't know what dosage form the
product is going to be classified as. That adds
cost to the development process because of
uncertainty of what is going to happen to the
In particular, if we think about methods
by which we would classify the dosage form of
topical products, here, I have generated a list of
four possible ways that you could approach this.
One is we would just use whatever the
sponsor says their product is as long as it is
consistent with some of the traditional definitions
that are available in various sources, and we will
look in sort of detail at some of those traditional
You might say, well, the generic product
is the same dosage form if it feels the same to me,
so I will just try it out and see if it
same, if it passes. You know, if the look and feel
of the products are the same. That is getting to
the aspect of pharmaceutical equivalence, a sort of
patient experience rather than sort of scientific
issues related to product performance.
Then, I am going to sort of discuss recent
work that the FDA group led by Cindy has done on
looking at a whole bunch of products and coming up
with a quantitative decision tree to classify
Then, sort of the fourth aspect of that is
looking at whether or not issues about dosage form
classification for complex issues would be
something that you would want to include in a
product development report, so justifying the
formulation development as being the same as the
reference product. That sort of might be a more
scientific way to look at these issues.
First, if we look a some of the
traditional definitions. Here, I will just focus
on the difference between a cream and an ointment.
One source is the CDER's data standards
definitions. These are sort of similar to USP
definitions of these products.
The cream is a semisolid dosage form
containing one or more drug substances dissolved in
a suitable base, and then it says more recently the
term has been restricted to products consisting of
oil-in-water emulsions. That is obvious what a
company should do - does a cream have to be an
oil-in-water emulsion or not.
Then, it talks about products that are
cosmetically and aesthetically acceptable, is part
of the definition of the cream, so that is not very
quantitative. It is hard to say is this product
aesthetically acceptable. That is really opinion
An ointment is a semisolid preparation
intended for external application. It seems to me
that a cream could be a semisolid preparation and
fit under the ointment definition. So, it doesn't
seem that those two definitions are really
In another FDA guidance, this
is the SUPAC
semisolid guidance. It has a glossary with
definitions of dosage forms, but these aren't the
same as the previous ones. A cream is a semisolid
emulsion, and an ointment is an unctuous semisolid
and typically based. So, typically based is not
sort of a definition, it doesn't have to be based
This definition talks about an ointment
being one phase, and not having sufficient water.
Again, the USP definition is sort of similar to the
one in the CDER data standards, but it is not
word-for-word identical, and talks about four
different classes of ointments.
So, again, the problem with the
traditional classifications is they are not really
consistent, and not very quantitative. So, a lot
of the sort of decision process would depend on
what your opinion was of a particular product, and
they might be overlapping, like you might be able
to call--under a particular definition of a
particular product, you might be able to call it a
cream or an ointment.
So, the result of this Topical Working
Group led by Cindy has been presented to previous
advisory committee meetings, and they
published a paper outlining this classification
What they did was they surveyed existing
products and devised a classification scheme, and I
just included the classification scheme here just
for reference in the presentation.
These are just some slides from their
previous presentations to give you the general idea
of what they did. They measured particular aspects
of products, say, creams and ointments, they
They looked at the loss on drying, and
then based on these products that were either on
the market or manufactured for them, they came up
with a classification scheme that sort of put the
products in the right category based on existing
The real advantage of this is it is
quantitative. If you take a product and
through and you measure the things outlined in
their decision tree, you will always end up in the
same place, and it will always be consistent.
In addition to that, this is a very data
driven approach. They looked at the products and
then drew the lines. It wasn't said here is sort
of a mechanistic definition of what a cream or an
ointment should do.
So, the question is, could this be overly
restrictive. If you follow this classification
scheme, you would be restricting products to
essentially what has been done before, and then
there is a question.
They didn't survey every product that is
on the market now, so there is a question, if a
reference listed drug falls into a different part
of this classification scheme, then, it's labeling.
So, it might be labeled as a cream, but by the
definition, it would be classified as an ointment.
What should a sponsor do in that case?
So, the final sort of approach to dosage
form classification might be to look at a
scientific view of the formulation design. I just
want to point out that sort of the legal aspects,
referring to topical use, sort of point toward this
here from the CFR.
It talks about inactive ingredient changes
for topical products. It says again that
abbreviated applications can use different
ingredients if they identify and characterize the
difference and provide information demonstrating
that the differences do not affect the safety or
efficacy of the proposed drug products.
So, a current way of looking at is a
change in formulation acceptable, you should check
the new excipients against the IIG. As I said
before, this looks at the safety of the individual
We also really consider that passing
bioequivalence tests are evidence that the
formulation change is acceptable. That is one
strong piece of evidence against that. But again
the product development report is an opportunity
for sponsors to characterize the differences.
Again, this could be important, you know,
if you are formulating a product and you are on the
boundaries of these, we have this
decision tree, what happens if you are on the
boundaries, how do you explain that this product
should be considered the same as the reference
product, you know, from a scientific point of view
rather than empirical classification scheme, of if
someone says, well, no, your product is not really
an ointment because it doesn't meet a particular
So, again, the product development report
is the opportunity for a sponsor to characterize
the difference that is sort of requested in the
Again, also, in the statutes, they list
reasons to reject ANDAs, and they talk about drug
products for topical administration where there is
a change in lipophilic properties of the vehicle.
Again, in this case, a product development
report is an opportunity for sponsors to explain
why the changes, Q1 and Q2 differences
appropriate for this particular product.
If this issue comes up, a lot of times we
will actually have to go from the review process,
you will have to go back and ask sponsors for more
information about these particular issues. So, the
development report is sort of an up-front way to
explain the reasons for doing that.
Just to sort of conclude the discussion
here, the first concept is the importance of Q1,
Q2, Q3 classification to identify appropriate
bioequivalent studies for the level of difference
in the product design. If a few products have
exactly the same active and inactive ingredients,
you might want to request different in vivo
bioequivalence studies than for a product where
there has been a change in inactive ingredient that
may affect the absorption of the drug product.
So, here again, we are looking at the
second concept, we are looking at the evolution of
the concept of pharmaceutical equivalence where we
have these traditional dosage form definitions,
maybe now backed up by empirical decision
but in the future, looking for a quality-by-design
aspect where the determination of whether a product
should be considered equivalent would depend on the
mechanistic understanding and the formulation
design rather than some traditional definitions,
and that the ideal state would be that this
understanding would reduce the need or allow us to
set the appropriate in vitro testing for a
particular product, and also to expand sort of the
formulation design space beyond past experience.
If you want to formulate a product that
goes beyond, say, an empirical dosage form
classification, this is the sort of way that you
would approach it, by providing the scientific
information to show that the formulation you have
chosen gives equivalent performance in the key
attributes as the reference product.
With that, I will conclude my
DR. COONEY: Thank you. I believe the
purpose of your presentation today is to bring us
up to date on the current thinking where
and where you are going as opposed to requesting a
specific action on our part, is that correct?
DR. LIONBERGER: Yes, that's right.
DR. COONEY: I would like to invite
questions and comments from the Committee. Yes,
DR. SELASSIE: This is a very general
question. With all these product development
reports that you get, obviously, there is going to
be a lot of information that is extraneous and
won't be useful for that particular application,
but will you all retain this information like in a
database, so that it could have use down the line?
DR. LIONBERGER: I don't know if we would
retain it in a database, but I would say that like
as Lawrence said, we are looking at our review
process, and in that, had the opportunity to read
several product development reports.
I find that they are a very useful way to
get an overview of what is going on with a
particular product. You know, an hour of reading
the development report, it seems like a
way to start the review of the application in more
detail, so I think it can be very valuable.
We don't have much experience with using
them yet, so in that sense, it could be valuable,
but we don't know how we would use that information
or store that information in the future.
DR. HUSSAIN: If you are suggesting that
there is a need to capture and create databases, I
think we do want to move in that direction, and we
tried to do that. Currently, our systems does
capture some of the key aspects. The inactive
ingredient guide is a process that we capture every
inactive ingredient that comes, but developing a
formal knowledge base would really be helpful, and
I think we have been thinking about it.
I tried to do that with immediate release
dosage forms and actually did some modeling with
that data that we have, and so forth, so we will
look into that.
DR. COONEY: Are there any other questions
from the Committee? Marv.
Just a quick comment. You
have my sympathies. I thought it was difficult to
determine how to do the BE studies on topicals, and
now I have been reinforced, you don't even know
what slot to put them in the Orange Book if they
are bioequivalent, so you have a big job ahead of
DR. COONEY: If there are no further
comments, Robert, thank you very much.
There have been no requests for
participation in the open public hearing at 1
o'clock, so we will proceed with the continuation
of the discussion on quality by design precisely at
1 o'clock when we come back from lunch.
We will begin that by a presentation of a
summary of the plan by Ajaz, and then we will
continue the discussion that we began earlier this
So, enjoy lunch and we will see you back
at 1 o'clock.
[Whereupon, at 11:59 a.m., the proceedings
were recessed, to be resumed at 1:00
A F T E R N O O N P R O C E E D I N G S
DR. COONEY: I would like to welcome
everyone back from lunch.
We will proceed with the afternoon
schedule. The first topic this afternoon will be
Ajaz Hussain, who will provide a summary
description of the plan to go forward.
Quality-by-Design and Pharmaceutical
Summary of Plan
DR. HUSSAIN: I am going to go back to the
slides I used in the morning instead of the ones I
had for this session. That was based on the
discussion that occurred.
Just on reflection, I just want to make a
couple of points. Yesterday, in a sense, as part
of the tactical plan to start our journey towards
the desired state, in a sense what we have done at
this meeting is to take a look back last 10 years
or so to see how our policies have evolved and how
they could evolve with two tools that we
introduced, the PAT guidance and ICH Q8.
Tom Layloff reminded me that in many ways,
some of the topics we have discussed, we have been
discussing for 30 years, and we keep discussing
those topics again and again, and the difference
that we have tried, at the training session that
some of you attended, I am clearly cognizant of the
fact that we are discussing topics that we have
been discussing for 30 years, and the quote I had
was the thing that if you tried to approach the
problem with the same tools and the same approach
again, we are bound to find the same solutions, so
we need something new.
What is new at the issue of this problem
is the science of formulation design, of science of
product design. The key aspect, much of that has
always been considered as an art, and as the
complexity of products is increasing, that art will
not be sufficient to really achieve the performance
we are trying to achieve.
So, it is a reflection back of saying all
right, 30 years of pharmaceutical
particular pharmaceutics, industry, pharmacy, and
so forth, what have we learned and what we need to
learn more to do things differently.
In some sense, that is the heart of the
debate. I also sort of mentioned to you, and this
is my original starting point in the thought
process was that you really need at FDA more people
with that background to really make that happen. I
changed my thoughts over the last several years.
What we have at FDA is scientists from
many, many different disciplines who sort of work
together. The reason I changed my mind was I think
looking at some of the practices and formulation