1
DEPARTMENT OF HEALTH AND HUMAN SERVICES
FOOD AND DRUG ADMINISTRATION
CENTER FOR DRUG EVALUATION AND RESEARCH
ADVISORY COMMITTEE FOR PHARMACEUTICAL SCIENCE
Tuesday, May 7, 2002
8:30 a.m.
5630 Fishers Lane
Rockville, Maryland
2
P A R T I C I P A N T S
Vincent H.L. Lee, Ph.D. , Acting Chair
Kathleen Reedy, R.D.H., M.S., Executive
Secretary
MEMBERS:
Gloria L. Anderson, Ph.D., Consumer
Representative
Mary J. Berg, Pharm.D.
John Doull, M.D., Ph.D.
Judy P. Boehlert, Ph.D.
William J. Jusko, Ph.D.
Joseph Bloom, Ph.D.
Nair Rodriguez-Hornedo, Ph.D.
Lemuel A. Moye, M.D., Ph.D.
Jurgen Venitz, M.D., Ph.D.
Marvin C. Meyer, Ph.D.
Arthur H. Kibbe, Ph.D.
Patrick P. DeLuca, Ph.D.
GUEST PARTICIPANT:
Ian Wilding, Ph.D.
INDUSTRY REPRESENTATIVES:
Leon Shargel, Ph.D. R.Ph.
Efriam Shek, Ph.D.
INDUSTRY GUEST PARTICIPANTS:
Aziz Karim, Ph.D.
Dr. Jack Cook
SGE PARTICIPANT:
Gordon Amidon, Ph.D.
FDA PARTICIPANTS:
Steven Galson, M.D., M.P.H.
Helen N. Winkle
Ajaz Hussain, M.D.
Larry Lesko, Ph.D.
Ameeta Parekh, Ph.D.
Dale Conner, Pharm.D.
Lawrence Yu, Ph.D.
3
C O N T E N T S
Introductions 4
Conflict of Interest 6
Introduction to Meeting, Dr. Helen Winkle 8
Comments, Steven Galson, M.D., M.P.H. 17
Future Subcommittees:
Introduction and Overview, Ajaz Hussain, Ph.D. 20
Clinical Pharmacology Subcommittee,
Lawrence J. Lesko, Ph.D. 25
Committee Discussion 33
Draft Guidance: Food Effect BE Studies:
Introduction and Overview, Dale Conner, Pharm.D. 49
Science Background and Issues, Ameeta Parekh,
Ph.D. 62
Open Public Hearing:
Brian P. Kearney, Pharm.D. (read by Ms. Reedy) 137
David M. Fox (read by Ms. Reedy) 138
Russell J. Rackley, Generic Pharmaceutical
Association 143
Biopharmaceutics Classification System - Next
Steps:
Introduction and Overview, Lawrence Yu, Ph.D. 160
Presentations:
Gordon Amidon, Ph.D. 161
Jack Cook, Ph.D. 185
Lawrence Yu, Ph.D. 197
Committee Discussion 211
4
1 P R O C E E D I N G S
2 DR. LEE: Good morning. I am calling this
3 meeting to order. I am Vincent Lee, the acting
4 chair of this committee. It is the Advisory
5 Committee for Pharmaceutical Science. I would like
6 to begin by going around the table and letting the
7 members introduce himself or herself, and we will
8 start with my colleague on my left.
9 Introductions
10 DR. ANDERSON: I am Gloria Anderson,
11 Fuller E. Callaway Professor of Chemistry at Morris
12 Brown College in Atlanta.
13 DR. BLOOM: Joseph Bloom, University of
14 Puerto Rico.
15 DR. VENITZ: Jurgen Venitz, Virginia
16 Commonwealth University.
17 DR. MOYE: Lem Moye, University of Texas.
18 DR. BOEHLERT: Judy Boehlert, consultant
19 to the pharmaceutical industry.
20 DR. RODRIGUEZ-HORNEDO: Nair Rodriguez,
21 professor of pharmaceutical sciences, University of
22 Michigan.
23 DR. SHEK: Efriam Shek, Abbott
24 Laboratories.
25 DR. SHARGEL: Leon Shargel, Eon Labs
5
1 Manufacturing.
2 DR. WILDING: Ian Wilding, Pharmaceutical
3 Profiles.
4 DR. KARIM: Aziz Karim, Takeda
5 Pharmaceuticals, in Chicago.
6 DR. CONNER: Dale Conner, FDA.
7 DR. GALSON: Steve Galson, FDA.
8 DR. WINKLE: Helen Winkle, FDA.
9 DR. HUSSAIN: Ajaz Hussain, FDA.
10 DR. LESKO: Larry Lesko, clinical
11 pharmacology at FDA.
12 DR. BERG: Mary Berg, College of Pharmacy,
13 University of Iowa.
14 DR. DOULL: John Doull, KU Medical Center.
15 DR. JUSKO: William Jusko, State
16 University of New York at Buffalo.
17 DR. DELUCA: Pat DeLuca, University of
18 Kentucky.
19 DR. MEYER: Marvin Meyer, emeritus
20 professor, University of Tennessee, College of
21 Pharmacy.
22 DR. KIBBE: Art Kibbe, Wilkes University
23 School of Pharmacy.
24 MS. REEDY: Kathleen Reedy, FDA.
25 DR. LEE: Once again, Vincent Lee,
6
1 University of Southern California. Let me ask the
2 committee members to raise their hand so everybody
3 knows who is on the committee. Thank you very
4 much. I think the committee is wide awake and
5 ready to go. Kathleen, would you please read the
6 conflict of interest?
7 Conflict of Interest
8 MS. REEDY: This is the acknowledgement
9 related to general matters waivers for the Advisory
10 Committee for Pharmaceutical Science for May 7,
11 2002.
12 The Food and Drug Administration has
13 prepared general matters waivers for the following
14 special government employees, Drs. Marvin Meyer,
15 Mary Berg, Judy Boehlert, Vincent Lee, Lemuel Moye,
16 Gordon Amidon and Patrick DeLuca which permit their
17 participation in today's meeting of the Advisory
18 Committee for Pharmaceutical Science.
19 The committee will discuss, one, the
20 current status of, and future plans for the draft
21 FDA guidance entitled guidance for industry,
22 food-effect bioavailability and fed bioequivalence
23 studies: study design, data analysis, and labeling;
24 two, discuss and provide comments on the
25 biopharmaceutics classification system, BCS; and,
7
1 three, discuss and provide direction for future
2 subcommittees.
3 Unlike issues before a committee in which
4 a particular product is discussed, issues of
5 broader applicability, such as the topic of today's
6 meeting, involve many industrial sponsors and
7 academic institutions.
8 The committee members have been screened
9 for their financial interests as they apply to the
10 general topic at hand. Because general topics
11 impact on so many institutions, it is not prudent
12 to recite all potential conflicts of interest as
13 they apply to each member. FDA acknowledges that
14 there may be potential conflicts of interest, but
15 because of the general nature of the discussion
16 before the committee these potential conflicts are
17 mitigated.
18 We would also like to note for the record
19 that Drs. Leon Shargel of Eon Labs Manufacturing,
20 Efriam Shek of Abbott Laboratories, Thomas Garcia
21 of Pfizer, Tobias Massa of Eli Lilly & Company,
22 Aziz Karim of Takeda Pharmaceuticals North America
23 and Jack Cook of Pfizer Global Research and
24 Development are participating in this meeting as
25 industry representatives, acting on behalf of
8
1 regulated industry. As such, they have not been
2 screened for any conflicts of interest.
3 In the event that the discussions involve
4 any other products or firms not already on the
5 agenda for which FDA participants have a financial
6 interest, the participants are aware of the need to
7 exclude themselves from such involvement and their
8 exclusion will be noted for the record.
9 With respect to all other participants, we
10 ask in the interest of fairness that they address
11 any current or previous financial involvement with
12 any firm whose product they may wish to comment
13 upon.
14 DR. LEE: Thank you, Kathy. Now I would
15 like to call Helen Winkle, Acting Director of OPS,
16 to introduce the meeting.
17 Introduction to Meeting
18 DR. WINKLE: Good morning, everyone. It
19 is really nice to see everybody here. I think this
20 is one of the few times everyone has actually been
21 in the room and present because normally we have a
22 lot of people on the telephone. So, it is good to
23 have all our members here.
24 I want to welcome everyone to the meeting
25 today, and I think this is really going to be a
9
1 great opportunity for us to meet with the committee
2 and to discuss what I consider to be a number of
3 really important scientific topics. My job this
4 morning is just basically to give everyone a
5 rundown on the agenda for the next two days, and it
6 is a pretty full agenda but I think there will be a
7 lot of things we can discuss and I think it will be
8 very worthwhile.
9 Today, Dr. Hussain will introduce the
10 Center's proposal for future subcommittees to this
11 advisory committee. As you all know, Dr. Hussain
12 has oversight for the advisory committee, and has
13 been looking at a variety of ways that we might
14 help in making the committee as effective as
15 possible. I think it is very difficult with
16 running this type of committee that is focused on a
17 variety of issues because you have to have a number
18 of different disciplines in the room to discuss the
19 issues, and sometimes it is not as easy to flesh
20 those issues out for presentation to the main
21 committee. So, I think we have been sort of
22 bouncing around ideas internally in OPS for ways in
23 which we can help the committee members in being
24 able to be better prepared to make recommendations.
25 So, Dr. Hussain will talk about our proposal for
10
1 that.
2 Next, following that discussion, we will
3 discuss two biopharm topics, and Dr. Larry Lesko,
4 who has already introduced himself, from the Office
5 of Clinical Pharmacology and Biopharmaceutics, will
6 lead those discussions. The Office of Clinical
7 Pharmacology and Biopharmaceutics, along with the
8 Office of Generic Drugs, has been sort of grappling
9 with these issues in order to finalize several
10 guidances or to actually, in one case, expand on a
11 guidance. So, we will present those issues today
12 and talk about ways that we can move forward in
13 these two really important areas.
14 The first issue that we will talk about in
15 the biopharm area is regulatory recommendations on
16 bioequivalence studies under fed conditions. In
17 order to facilitate getting the guidance out we
18 have basically two questions which need to be
19 addressed today. One is regarding the waivers of
20 in vivo fed studies for ANDAs for BCS Class I drugs
21 and drug products, and the second is the confidence
22 intervals and criteria to claim between fasted and
23 fed states of new drugs and between fed states for
24 generic drugs. This is an issue that I think will
25 have a lot of discussion with it, and I look
11
1 forward to hearing that. We want to listen
2 basically to what can be added to this
3 scientifically, to get your feel on this and then
4 we will go back and regroup internally, and decide
5 where we need to go with this guidance.
6 The second topic we want to discuss under
7 the biopharm area is next steps for the
8 biopharmaceutics classification system. The BCS
9 has been discussed here I think on several
10 occasions. Basically, we have a guidance out which
11 is what I would call conservative in those
12 particular products that we allow to come in with
13 waivers under BCS.
14 So, what we want to do today is talk about
15 expanding the BCS; get your thoughts on the
16 expansion of it, and to get some ideas as far as
17 the next steps for justifying the expansion or
18 extension of BCS. We have already come up against
19 some challenges, and I think we would like to talk
20 about how we can handle these challenges as far as
21 BCS in the future.
22 There is already some work going on in
23 PQRI, the Product Quality Research Institute, on
24 expanding BCS and we will share a little of that
25 information and discuss whether that research is
12
1 actually all that we need to sort of capture where
2 we need to go in our efforts with BCS.
3 As I said, obviously this is a pretty full
4 day. I mean, I think there will be a lot of
5 discussion around these topics. Then, tomorrow we
6 will have several items on the agenda as well. The
7 first thing we are going to talk about is to give
8 you an update on the process analytical
9 technologies, PAT. You all know that we have a
10 subcommittee that was formed. The subcommittee met
11 for the first time in February. I think it was an
12 extremely good meeting and I think a lot came out
13 of that meeting as far as helping us focus on the
14 whole initiative of PAT. Dr. Tom Layloff, who is
15 chairing the subcommittee, will report on that
16 meeting that was held in February. Then, Dr.
17 Hussain will provide a progress report and describe
18 what the next steps are for PAT. Then, we will
19 appreciate your input into those steps and what
20 your thoughts are as far as where we need to go.
21 Of course, this is an extremely exciting
22 subcommittee and the issues I think are really good
23 in helping us focus on what we need to do, and the
24 underlying science for the whole initiative.
25 Also along the same line, at an earlier
13
1 meeting last year we discussed some of the general
2 issues related to rapid microbial testing.
3 Tomorrow we will update you on those issues. Then
4 we will discuss whether the PAT program can
5 adequately address the issues relating to the
6 introduction of rapid microbial testing.
7 After that we will introduce the topic of
8 blend uniformity again. At the last meeting we
9 talked about the PQRI proposal that was coming out
10 on the PQRI research that is being done, and PQRI
11 has now formally submitted that proposal to the
12 agency, and we are finalizing our decision on
13 whether to incorporate their recommendations into
14 our regulatory scheme. So, we will talk a little
15 bit about that final proposal. We still have some
16 questions we need to address as far as that
17 proposal or recommendations and we will discuss
18 that tomorrow as well.
19 Just to mention one thing along this line,
20 as everyone on the committee knows, we did have a
21 draft guidance that was out on blend uniformity for
22 ANDAs and, because of the fact that we felt that
23 guidance really didn't fit into our current
24 regulatory scheme and with the idea that at least
25 the recommendation from PQRI would stimulate our
14
1 thoughts and expand what we believe to be our
2 regulatory position, we have withdrawn the
3 guidance, the draft guidance on blend uniformity.
4 So, that makes it sort of necessary for us to move
5 on getting the new guidance out. So, we would
6 really like to get to our final conclusions with
7 your recommendations today and move forward on that
8 because we have a lot of people who, you know, are
9 sort of waiting to hear what the results of our
10 decision is in this area.
11 The last item on the agenda tomorrow will
12 be a discussion of regulatory issues related to
13 polymorphism. Basically, I consider this to be an
14 awareness topic, just to seek your input on maybe
15 the direction we need to go in, and then we will
16 plan a more in-depth discussion at a subsequent
17 meeting on polymorphism.
18 Again, a very full agenda and I look
19 forward to hearing the discussion. I think these
20 are all very, very stimulating scientific topics
21 and it will be very helpful to us as we move ahead
22 in these areas.
23 There are a number of other topics that
24 will be coming up in future meetings, including a
25 follow-up on DPK. I know you all have been dying
15
1 to hear where we are with DPK. I think what we
2 will talk about the next time we discuss this is
3 basically not only DPK, but to look at other
4 possible methods for determining bioequivalence of
5 topical products. I think at the last advisory
6 committee meeting we talked a lot about DPK and
7 felt that it wasn't completely fleshed out, and
8 that probably we did need to expand our focus as we
9 looked at possibilities for determining
10 bioequivalence. So, I don't think DPK is
11 completely off our agenda for the future, but I
12 think that what we want to focus on is other
13 methodology and discuss that with you. I sort of
14 call it a toolbox of methods that you could use for
15 bioequivalence in this area, and I think it will be
16 important for us to discuss these various methods
17 with the committee in the future. We have put out
18 a Federal Register notice--it should come out any
19 day--which will withdraw the draft guidance on DPK.
20 This is just to touch on future topics,
21 but I would also like to encourage members of the
22 advisory committee to bring possible topics to our
23 attention. I think, obviously, you all are out in
24 the working world every day, dealing with a lot of
25 these scientific issues, and we would be glad to
16
1 hear your recommendations for possible things we
2 can discuss before the committee. So, if you do
3 have any suggestions, please feel free to share
4 those with Dr. Hussain and myself or with Dr. Lee.
5 Last, before I hand over the meeting to
6 Dr. Hussain, I would like to introduce Dr. Steven
7 Galson. Dr. Galson, who is sitting here, on the
8 end, joined the Center last year as the Deputy
9 Director to Dr. Woodcock. We sort of asked him
10 here this morning because we thought it would be
11 helpful to him to meet the committee and get a feel
12 for the types of issues that we do discuss at this
13 meeting. You know, Dr. Galson is already playing a
14 very important role, despite the short time he has
15 been here, in a number of things that are going on
16 in the Center. Mainly he has been what I consider
17 one of the main forces behind risk management
18 implementation. I have asked Steven to say a few
19 words today to sort of introduce himself and some
20 of the things he has focused on, but what I would
21 like to do is bring him back in the future to talk
22 more about risk management. So, before I give it
23 back to Ajaz, I would like to hear from Dr. Galson
24 for a minute.
25 Comments
17
1 DR. GALSON: Good morning, everybody. I
2 am really happy to be here. As you have heard, I
3 have just been with CDER about a year, and I want
4 to start out by really just apologizing that it has
5 taken me a whole year to come and say hello to you
6 as a group. The work of our advisory committees is
7 incredibly important and in the Office of
8 Pharmaceutical Sciences, headed by Helen and Ajaz,
9 we really are on the cutting edge science in how it
10 is applied to drug regulation. Without your advice
11 frequently in the year, telling us what you think
12 about changes that we may be making or other policy
13 issues, we really can't stay on top of cutting edge
14 science nationally and internationally. So, the
15 work that you do is really extremely important and
16 we are very, very grateful for the commitment of
17 your time. We know that you all have lots of other
18 things you could be doing. Also, your commitment
19 to public service. It is really important for the
20 agency and really important for the country to have
21 people like you who are willing to commit to us.
22 The state of the Center for Drugs is very
23 good. We have an excellent working relationship
24 with the new administration. We have a new Deputy
25 Commissioner, as I think you know, Dr. Lester
18
1 Crawford, and we have already been working
2 extremely closely with him and he is very involved
3 in some of our issues, and we have a great
4 relationship.
5 Also, the state of the Center is very good
6 with regards to Congress and our overall funding.
7 I think many of you heard about the Prescription
8 Drug User Fee Act. We have been working hard to
9 negotiate a proposal to extend our user fees with
10 the drug industry over the last few months, the
11 last year really, and this has concluded very
12 successfully. We have sent a proposal to Capitol
13 Hill which we are hoping they are going to act on
14 expeditiously. What this is really going to do is
15 re-authorize and re-fund the user fee program in a
16 way that will help us use our resources in a way to
17 continue to apply the best science in a rapid way
18 to get drugs on the market and to the American
19 people, having a positive impact on public health.
20 So, we are very positive about that. It is a very
21 important thing going on. It will happen in the
22 next year.
23 As Helen said, I would really like to come
24 back at a further meeting and talk to you about
25 many of our initiatives in risk management. This
19
1 is going to be very important to us, as it is now.
2 Congress and outside groups are very, very
3 interested, some of them quite critical, of how we
4 make decisions about approving drugs and how we
5 make decisions about the degree of risk that we
6 allow in our products and in the way our products
7 are used out there in the real world. So, this is
8 an important initiative and I would like to come
9 back and talk to you about it in general when I can
10 and when there is time on the schedule.
11 I have been generally assisting Dr.
12 Woodcock in running the Center for about six
13 months. After September 11 Dr. Woodcock stepped
14 down and worked on a detail on emergency
15 preparedness in the Commissioner's office so I was
16 actually running the Center on an active basis for
17 about six months, and I got an incredibly intense
18 introduction to what everybody was doing and I
19 think I have a good understanding of the Center
20 now, and am going back now, focusing on initiatives
21 and helping in the general management.
22 So, again, I would like to come back later
23 and meet with you more. I will spend a little time
24 here this morning listening to the beginning of
25 your meeting. Again, thank you for all your time
20
1 and commitment to being here with us.
2 DR. LEE: Thank you very much. Dr.
3 Hussain?
4 Future Subcommittees
5 Introduction and Overview
6 DR. HUSSAIN: Good morning. At a previous
7 meeting of the Advisory Committee for
8 Pharmaceutical Science we had sort of briefly
9 discussed the need for creating discipline-specific
10 subcommittees under this committee itself. We
11 perceived the need because of the broad scientific
12 disciplines that are under the oversight of OPS. I
13 think we are all familiar with chemistry and
14 biopharmaceutics as the key area but clinical
15 pharmacology is one of the major areas, and I think
16 its importance is increasing tremendously. Also,
17 microbiology. We have a subcommittee on PAT but I
18 think I want to talk to you about other committees
19 that we want to bring under this advisory
20 committee.
21 The thoughts are to keep the Advisory
22 Committee for Pharmaceutical Science broadly
23 focused and have expertise from various disciplines
24 that we need to address issues in OPS. The
25 subcommittees will then essentially focus on more
21
1 detailed discipline-specific topics for discussion.
2 If I use the example of the PAT
3 subcommittee and what we have learned from that
4 subcommittee, bringing experts with hands-on
5 experience in the areas I think really helps us to
6 identify issues and find solutions quickly and more
7 effectively. In that regard, how do we use the PAT
8 subcommittee? Do we keep the PAT subcommittee or
9 do we do something different?
10 The proposal that I will just discuss
11 briefly, before I call on Dr. Lesko to talk about
12 the clinical pharmacology subcommittee as an
13 example of the new subcommittee structure that we
14 want to present, is to look at PAT as a new
15 technology area but in a sense it addresses issues
16 in manufacturing. Chemistry manufacturing controls
17 is a major part of review activities within the
18 Center for Drugs. But, at the same time, issues
19 related to GMPs, which are equally important, also
20 need to be addressed.
21 Currently, for example, the gaps that
22 exist between review and inspection--there is no
23 mechanism to address some of those gaps. Blend
24 uniformity, that you will talk about tomorrow, is
25 one such example. Was blend uniformity a review
22
1 issue or was it an inspection issue? I think we
2 will discuss that tomorrow.
3 But the frustration that we sometimes feel
4 because of the organization structures and
5 different roles and responsibilities, it is not
6 often feasible, or we don't have a mechanism to
7 bring issues which are on the boundaries of these
8 organization structures or disciplines to address
9 them more effectively.
10 So, the PAT subcommittee right now is
11 focusing on a very specific charter to address
12 process analytical technologies. That committee
13 essentially could sort of be sunset after its
14 initial assignment is over, and be replaced by a
15 manufacturing subcommittee because manufacturing is
16 a general long-term issue and we need a mechanism
17 for addressing issues with respect to GMPs and
18 review in the area of CMC.
19 We currently don't have any mechanism to
20 have discussion or even analysis of issues that are
21 technical in nature, which are in the area of
22 manufacturing, and how do we do that? So, we are
23 thinking probably that as the PAT subcommittee
24 completes its charter of the assigned task, to
25 sunset that committee and put in the place of that
23
1 subcommittee on manufacturing. That will bring the
2 Office of Compliance, Office of Pharmaceutical
3 Science and Office of Regulatory Affairs together.
4 So, essentially it would sort of be a team approach
5 from the FDA to bring issues to the subcommittee
6 related to GMPs, manufacturing and so forth. Most
7 of the time, we hope there will be focus on general
8 technical issues that need to be addressed. This
9 committee could then possibly provide a means for
10 addressing technical issues that are not being
11 addressed today.
12 One way of looking at the current
13 situation is that the Center for Drugs is
14 responsible for developing policies, especially in
15 the area of chemistry, manufacturing and controls,
16 but the field has to enforce that. We have
17 internal mechanisms to address that but, from the
18 industry perspective, we don't have a way to
19 address technical issues or disputes which are
20 technical in nature. The only solution right now
21 is to issue a 483 or a warning letter. We want to
22 see whether we can have a subcommittee that can be
23 a mechanism to address some of those issues. So,
24 that is sort of an example of what we could do with
25 respect to manufacturing.
24
1 Microbiology is a very important
2 discipline. Helen has essentially brought the
3 microbiology review staff to the Office of
4 Pharmaceutical Science level to give them
5 visibility; to give them more recognition in terms
6 of importance; and we are starting to discuss
7 microbiology issues. Would we need a subcommittee
8 on microbiology? I think that is a question that I
9 will leave for now but I think we will have to come
10 back to discuss it.
11 Clinical pharmacology will be the next
12 committee, which probably will be the first
13 subcommittee we will form under this new umbrella.
14 I will ask Larry Lesko to walk you through his
15 proposal of what he thinks the clinical
16 pharmacology subcommittee would do, and how he
17 feels we can constitute that.
18 Following that presentation, I request you
19 to sort of have a general discussion on the concept
20 of this, the subcommittee structure which will be
21 focused on disciplines and what subcommittees do
22 you think would be necessary and what we should
23 move forward with. Our current thought is that the
24 next subcommittee we will form will be the clinical
25 pharmacology, followed by manufacturing by
25
1 sunsetting PAT and moving that into the
2 manufacturing subcommittee.
3 Pharmacology/toxicology is another idea we have;
4 non-clinical studies subsection. I think how we
5 manage that transition to a more general
6 subcommittee on pharmacology/toxicology will be a
7 subject for discussion later on, and so forth.
8 So, with that introduction, I will ask
9 Larry to present his talk on clinical pharmacology
10 and then we can have a general discussion on this
11 concept. Larry?
12 Clinical Pharmacology Subcommittee
13 DR. LESKO: Thanks, Ajaz. Good morning,
14 everybody.
15 [Slide]
16 You should have in front of you two things
17 that are relevant to my remarks this morning. The
18 first is a one-page proposal for a clinical
19 pharmacology subcommittee and the second is a set
20 of slides that I am going to show to walk you
21 through the steps of the formation of this
22 subcommittee.
23 I like what Dr. Galson said in his
24 introductory remarks. He said that OPS is on the
25 verge of cutting edge science. I think this is
26
1 really no more true than in clinical pharmacology
2 where we are seeing many rapid developments that
3 can impact drug development to the regulatory
4 processes, and it is because of this that we feel
5 that there is a need to develop this clinical
6 pharmacology subcommittee.
7 [Slide]
8 What we have in mind is a membership that
9 would consist of external recognized and respected
10 experts in the general field of clinical
11 pharmacology. However, we would like to emphasize
12 three specific areas. The first is
13 pharmacometrics, which has certainly been growing
14 rapidly over the last five years; the field of
15 pharmacogenetics and pharmacogenomics, which is an
16 emerging field; and the field of pediatrics.
17 I want to point out that none of these
18 areas are the sole domain of clinical pharmacology,
19 so we anticipate that any issues that come before
20 the clinical pharmacology subcommittee would be
21 issues that we would work on collaboratively with
22 our medical staff and with our biostatisticians
23 within the Center.
24 [Slide]
25 What would be the responsibilities of the
27
1 subcommittee? Well, we see this as a committee
2 that would advise and counsel us on a broad range
3 of issues and questions from new and emerging areas
4 of clinical pharmacology, specifically to talk
5 about the science and how we might use it or apply
6 it in specific areas relative to regulatory review
7 of INDs or ANDAs and then, further downstream, how
8 we might integrate this new information into
9 research or into regulatory policies that might
10 take the form of, for example, guidances.
11 [Slide]
12 Let's talk about those three areas and
13 explain a little bit more specifically what I mean
14 by those. The first is pharmacometrics.
15 Pharmacometrics encompasses, in our mind, three
16 broad areas. The first is the area of population
17 PK/PD analyses, using samples from clinical trials.
18 The second is modeling of
19 exposure-response relationships, whether they be
20 broadly speaking dose response or more specifically
21 PK/PD. The third is clinical trial simulation.
22 What we see as potential applications of
23 this technology and where we would like to go in
24 working with the subcommittee is to develop
25 standardized approaches using each of these
28
1 technologies in regulatory decision-making. That
2 is to say, what are the best practices given the
3 current state of knowledge?
4 Secondly, in particular we are interested
5 in developing a standardized approach to adjusting
6 doses in special populations when we see an
7 increase or decrease in exposure as defined by area
8 under the curve or Cmax.
9 Third, we would like to apply this
10 knowledge in a more integrated way in the selection
11 of optimal doses for drug approval and, last, to
12 use clinical trial simulation in the design of
13 Phase III trials to try to focus a little bit more
14 on optimized doses.
15 [Slide]
16 The second area is very exciting. It is
17 the area of pharmacogenetics and pharmacogenomics.
18 We are quite interested in this area because of the
19 rapid increase in the number of NDAs and INDs that
20 contain this type of information. In our Office we
21 recently conducted an informal survey and found
22 that over fifty applications have this type of
23 information in them. Two-thirds of those
24 applications utilize genetic information from the
25 polymorphic aspects of drug metabolism. Many of
29
1 these applications have come about in the last two
2 years, even though our informal survey covered five
3 years.
4 But some of the things we would like to
5 bring before the committee for discussion include
6 the role of genotyping in the management of risk of
7 previously approved products. We have some very
8 good examples where prospective trials of TPNT
9 polymorphism, for example, has been shown to
10 influence the toxicity of the purine drugs such as
11 6-mercaptopurine. If you look at the label for
12 those products, there is no indication in the
13 dosage or administration section of the label that
14 a physician should utilize these genotypes, which
15 are now becoming widely available, before
16 prescribing the drug.
17 Secondly, we are beginning to sense a
18 development of drug-device combinations where
19 approvals are based on the measurement of genetic
20 markers, oftentimes linked to clinical outcome,
21 utilizing pharmacodynamic measures of one sort or
22 another. An example might be the
23 haplotype-dependent receptor polymorphism that has
24 been reported publicly in the literature and on the
25 web page of certain companies.
30
1 The third thing we would like to think
2 about in the subcommittee is the study design and
3 analysis of early phase clinical trials. These
4 could be Phase I trials or Phase II trials but
5 basically with the ability to genotype patients as
6 potential entry criteria. It would be worthwhile
7 to talk about enrichment strategies for Phase I and
8 Phase II trials.
9 [Slide]
10 This is a slide of a pediatric study
11 decision tree that we developed in the Center with
12 our other disciplines. I am putting it on here to
13 illustrate a framework which we have used in
14 approving drugs for pediatrics under the
15 exclusivity arrangements that we have.
16 If you look down that tree very carefully
17 you see that many elements of it have to do with
18 clinical pharmacology, whether it is PK studies,
19 whether it is concentration response relationships
20 or PD measurements.
21 [Slide]
22 We have been using this as a general
23 framework but it brings us to the next issue, which
24 is the fact that over the past couple of years we
25 have had a huge number of written requests from
31
1 sponsors to conduct pediatric trials. As of March
2 1 of this year, we have had 241 written requests
3 which embodied 568 studies and over 33,000
4 pediatric patients. That is not to say that all of
5 these studies have been or will be conducted but
6 they represent the intention of sponsors to gain
7 pediatric drug approval.
8 Where we have seen these types of written
9 requests and, in fact, where we have seen studies
10 conducted, the breakdown of those studies is
11 illustrated on this slide. Notice that efficacy
12 studies represent 34 percent of the studies; safety
13 and PK, 30 percent; safety, 17 percent; and PK/PD,
14 10 percent. The point is that many of these
15 studies rely upon clinical pharmacology to provide
16 the evidence of efficacy or safety in the pediatric
17 population. We see this across all medical
18 divisions, the exception being imaging where we
19 have had not much activity, and that slide gives
20 you a range from 0-45 in cardiorenal.
21 Following that, we have had 56 approved
22 active moieties that have been given exclusivity.
23 We have changed about 30 or 40 drug labels with
24 regard to pediatric dosing. But it brings us to
25 the question that we would like to interact with
32
1 the subcommittee on, and that is to say what have
2 we learned from all of this?
3 [Slide]
4 What we would like to do in the upcoming
5 months is to do a retrospective characterization of
6 this database on pediatrics, and look at the
7 magnitude of age and body size dependence of PK and
8 PD of the studied drugs, compare those to the adult
9 population and check whether our assumptions going
10 into these studies were accurate or whether they
11 need to be refined. We have a tremendous database
12 here that needs to be looked at very critically,
13 and I think we would like to do that and bring the
14 information to the clinical pharmacology
15 subcommittee.
16 Why would we like to do that? We want to
17 do that because with this experience in hand we
18 could then discuss the general principles that
19 underpin the types of studies that the agency
20 requests for pediatrics, and begin to look at the
21 role of clinical pharmacology studies and whether
22 we should continue with that role or refine it
23 based on the evidence that these studies have
24 provided.
25 [Slide]
33
1 That is the initial charge of the
2 subcommittee. What we would like to do going
3 forward is to nominate a chair and at least one
4 other member from the current advisory committee,
5 the ACPS; constitute this clinical pharmacology
6 subcommittee with no more than nine members. These
7 would be renewable terms of three years. We hope
8 to meet at least once a year for general briefing
9 on these and other issues. However, we would like
10 to also have the ability to consult on more
11 occasions on specific issues that might relate to
12 the areas I just mentioned. Thank you.
13 Committee Discussion
14 DR. LEE: Thank you, Larry. Ajaz, shall
15 we take questions now or do you have other
16 subcommittees?
17 DR. HUSSAIN: Well, I think the
18 discussion, if you could focus specifically on
19 clinical pharmacology but also broadly on the
20 concept of specific subcommittees.
21 DR. LEE: So, you have no other
22 subcommittees to introduce?
23 DR. HUSSAIN: No.
24 DR. LEE: Any questions for Larry? I
25 think Larry has introduced a very important topic.
34
1 In fact, maybe I can begin and ask you a question.
2 You identified three topics and those three are
3 pretty diverse, and it would seem unreasonable to
4 have one subcommittee to cover the entire
5 waterfront.
6 DR. LESKO: We thought about that and, you
7 know, at the core each of these topics we have
8 basic principles of clinical pharmacology relating
9 exposure to response. You know, response can be a
10 genetic marker; it could be a pharmacodynamic
11 measure in a pediatric population; and, of course,
12 pharmacometrics is the tool that we would use to
13 analyze that data. So it is a lot like three
14 overlapping circles and I think they have some
15 commonality to them that will allow us to nominate
16 a strong subcommittee group.
17 The other aspect of this is that we would
18 like to take, as I mentioned, nine members of the
19 group and try to identify three or four experts in
20 each one of these areas as lead individuals on the
21 subcommittee so that they can take the discussion
22 based on their specific expertise. So, we kind of
23 think the specific expertise of three or more
24 members in a given area, plus the general
25 background of clinical pharmacology would provide
35
1 an excellent committee for input.
2 DR. LEE: Thank you. Dr. Doull, you have
3 comments to make?
4 DR. DOULL: Yes, I am delighted to see
5 that you are going to deal with the pediatrics
6 problem. What you are really dealing with is the
7 issue of sensitive populations. As I am sure you
8 know, EPA in regard to pesticides, has well as
9 Congress, has simply established a dose factor of
10 ten in the Food Quality Protection Act for
11 pesticides. It would be a disaster, I think, if we
12 were to do that in the drug area. So, this makes
13 much more sense. You are going to use science to
14 decide in which cases you do need, in fact, a
15 protective factor.
16 But my question is there are lots of other
17 sensitive populations, and how would you deal with
18 those? Add those on? Old folks, diabetics and
19 what-have-you?
20 DR. LESKO: That is a good point. I think
21 the pediatric population is particularly
22 interesting now because we have so much data
23 in-house that we have gained from the pediatric
24 exclusivity situation. That is not to say our
25 other special populations may not be of interest.
36
1 In fact, we are looking at gender, ethnic origin
2 and other intrinsic factors that define special
3 populations in other settings. But that is not to
4 say this committee's purvey wouldn't include a
5 discussion on, for example, exposure response and
6 dose adjustments in those special populations.
7 I think that is kind of the beauty of the
8 subcommittee. The principles that underlie all
9 these are pretty much the same. How do you bridge
10 data acquired in one setting, for example in an
11 efficacy/safety trial, to a special population
12 whether it be pediatrics, or a population defined
13 by genetics, or a population defined by age or
14 gender. So, I think that is something that we
15 would certainly be open to in the subcommittee. It
16 would depend on the priority and what is going on
17 in other working groups and other committees.
18 DR. LEE: Dr. Berg?
19 DR. BERG: Yes, in regards to gender and
20 the special populations, just so I understand, you
21 would be looking at products already on the market
22 as well as new applications? In other words, what
23 we have on the market and then also new ones in the
24 hopper?
25 DR. LESKO: I think we need to look at
37
1 both. We certainly have a database of products
2 that are on the market for which information, for
3 example in pediatrics, has been obtained. Ideally,
4 I think we want to look at this information in a
5 more prospective fashion to learn as we are moving
6 forward and I think treat it as a continual
7 refinement of the paradigm for assessing pediatric
8 information and drug dosing.
9 DR. BERG: I know just recently FDA
10 received some appropriations for a database for
11 gender--for the globalization through the Office of
12 Women's Health--
13 DR. LESKO: Right.
14 DR. BERG: I think that is very good for
15 the new products.
16 DR. LESKO: Right.
17 DR. BERG: But looking at the products
18 already out on the market, I know we have been
19 looking at this back in Iowa for about three to
20 four years actually with my students, and literally
21 there still is question with regards to looking at
22 gender analysis and then getting into the question
23 of ethnicity analysis for a database. So, those
24 populations are as sensitive as the pediatric group
25 as well.
38
1 DR. LESKO: Yes, a lot of the analyses of
2 databases are focused on the numbers, how many have
3 been in clinical trials, as opposed to the results
4 and what has the result signal in terms of need to
5 look at something differently or reassess the way
6 we interpret the data. So, I would see this
7 initiative as really getting into the data in the
8 population and really analyzing it in a systematic
9 way. We have begun to do this in the Office with
10 some projects that the Center has funded. It is
11 not starting out from scratch but it is starting
12 out with a preliminary assessment of the database
13 that I think will be much more quantitative as we
14 move forward, and then use it in a real-time
15 fashion to provide us feedback on how we are
16 approaching these special populations.
17 DR. BERG: Yes, this is really good
18 because it gets back to the push for the GO reports
19 in regards to gender analysis that came out last
20 year. In other words, industry has been recruiting
21 women into studies but there hasn't been a separate
22 analysis. I know there was quite a big to-do last
23 summer in regards to the report. So, this really
24 helps to really push that issue for that subgroup
25 analysis.
39
1 DR. LESKO: And I think we can go from the
2 specific to the general. I mean, if we look at a
3 class of drugs for which we have had some, say,
4 pediatric approvals or other special populations,
5 what can we say about the class in general so that
6 one might take the next member of that class and
7 perhaps treat it a little bit differently based on
8 what has been learned so far.
9 DR. BERG: Yes, this is a really great
10 start.
11 DR. LEE: Any questions from the other
12 side of the table? Jurgen, any comments?
13 DR. VENITZ: I only want to support that
14 wholeheartedly. I think it is an excellent idea.
15 One of the things I guess I am still unsure about
16 is what is the reporting mechanism in terms of
17 reporting information back from the subcommittee to
18 this committee.
19 DR. LESKO: I don't know if we have a
20 precedent for this or not but, in my mind, what
21 would happen would be that the chair of the
22 clinical pharmacology subcommittee would report
23 back to this committee at least once a year and if
24 this committee met more often and there was a need,
25 more than once a year. But I think the chair of
40
1 this committee will be very important and that
2 would be the connection between the ACPS and the
3 subcommittee.
4 DR. HUSSAIN: I think that the process
5 would be similar to any other subcommittee. Two
6 members of this committee would be members of the
7 subcommittee and essentially the chair reports
8 back, like, tomorrow Tom Layloff reports back to
9 you for the PAT subcommittee. The subcommittee
10 essentially is advisory to this and decisions
11 essentially are made in this committee.
12 DR. LEE: It seems to me that this
13 committee is rather proactive. Is that what you
14 have in mind? A rather proactive committee
15 identifying new issues?
16 DR. LESKO: You know, knowing the members
17 of this community in clinical pharmacology, I
18 expect it will be very proactive and we will be
19 too. We have some issues in mind that we want to
20 start with so I think that is important.
21 DR. LEE: What about the issue of life
22 style?
23 DR. LESKO: Well, that is an interesting
24 issue. I haven't thought of it in the context of
25 this particular subcommittee but I am sure you are
41
1 leading up to another comment.
2 DR. LEE: If you have a global community
3 and all this kind of stuff, I think it is very
4 exciting and I will be very interested to see how
5 this subcommittee will evolve because in my
6 estimation it will probably work rather closely
7 with your Office as well. Isn't that true?
8 DR. LESKO: That is what I expect will
9 happen but, again, there will be other disciplines
10 involved with this as well like, for example, if we
11 start out with the drug safety group there will be
12 multiple disciplines involved.
13 DR. LEE: Dr. Doull?
14 DR. DOULL: I think the only thing that
15 still concerns me is that it seems to me that you
16 are going to be right in the middle of the area of
17 risk management in a sense when you deal with
18 sensitive populations, and somehow the decisions
19 that we make in clinical pharmacology are going to
20 have some really broad implications in terms of
21 risk management. I guess somehow one needs to
22 coordinate so that you don't get crosswise in this
23 subcommittee with, say, a policy that affects risk
24 management for the agency as a whole, food and
25 devices and all that.
42
1 DR. LESKO: That is a good point. I mean,
2 risk management in the Center, as we think about
3 it, is really not one-dimensional by any means.
4 Any risk management strategy has had multiple
5 dimensions and sometimes is pretty complex.
6 I think working with Dr. Galson and others
7 at the Center level on various risk management
8 approaches, this is going to be a piece of the
9 puzzle but I think it is an important piece that we
10 need to look at and integrate with other pieces of
11 information. I can see the information being
12 learned from this exercise going on to become part
13 of other risk management plans that are being put
14 in place. Maybe it will lead to a more systematic
15 approach to risk management that I think the Center
16 would like to get to.
17 DR. GALSON: Just one comment on that, I
18 think that is an excellent point but it shouldn't
19 be a cause of worry really because there isn't any
20 other advisory committee that is working on this
21 particular angle. We do need to put it all
22 together. There aren't any other advisory
23 committees with the expertise of this one that is
24 being discussed that will be dealing with this
25 specific issue. So, we will really count on what
43
1 is coming out of this group in figuring out what
2 direction to go in for the whole Center. But
3 coordination is very important.
4 DR. LEE: Any other comments? Efriam?
5 DR. SHEK: I have more general comments
6 with regard to the characteristics of the
7 subcommittees. If you take the PAT example, it
8 looks like it was a specific task, an assignment to
9 look at that. Now this committee, it looks like it
10 is a more standing committee which will be a
11 permanent, let's say, subcommittee. The same thing
12 may be for toxicology and safety.
13 When we bring up the manufacturing the
14 issue is should we consider broadly if that is
15 going to be permanent for the whole area of CMC? I
16 believe we, in industry, realize that CMC is an
17 umbrella. We cannot just look at drug product
18 manufacturing; we have to look at the drug
19 substance; we have to look at the QC. Everything
20 is tied together, and whether we should consider
21 broadening it to CMC type of a subcommittee.
22 DR. HUSSAIN: That is a good point. What
23 we will plan to do is bring a proposal, like Larry
24 did, on the manufacturing committee and its makeup
25 at the next meeting. The thought process is to not
44
1 only discuss CMC from the review side, but bring
2 and invite Office of Compliance and our Office of
3 Regulatory Affairs to be partners with us on that.
4 So, it will be a whole umbrella of all CMC and
5 manufacturing issues in sort of one direction. So,
6 we will flesh out the proposal and bring that to
7 you next time.
8 DR. LEE: Other comments? Larry, I think
9 you have touched on a topic that is quite
10 interesting so I have another question. What about
11 geriatrics? People like me?
12 DR. LESKO: You have about ten more years
13 before you worry about that! That was probably the
14 first ever "special population" that the agency
15 looked at back in 1983 or '84, '85, and we do have
16 things in place that direct a sponsor to look at
17 age on the high side, specifically within a
18 clinical trial, along with race and gender.
19 Again, I am not excluding that from the
20 domain of this subcommittee but I would say at the
21 moment it is not a high priority, based on where we
22 are with other policies in place with respect to
23 the elderly. We usually have a pretty nice
24 assessment of that within the clinical pharmacology
25 database and look at it quite routinely for any
45
1 need of dose adjustment.
2 DR. LEE: Thank you.
3 DR. MEYER: Would you be more politically
4 correct if you said pediatrics and other special
5 populations?
6 DR. LESKO: I think that would be a good
7 idea. It would really encompass a lot of the
8 comments that the committee members made and
9 signalled that other things can be brought before
10 the committee. So, I would be in favor of that
11 change, sure.
12 DR. LEE: Bill?
13 DR. JUSKO: I have a very strong
14 endorsement of this plan and commend you for doing
15 it. I imagine the committee membership will be
16 somewhat like this one with independent consultants
17 of sorts, as opposed to having representatives of
18 scientific organizations?
19 DR. LESKO: That is correct. I envision
20 the committee as being one of expertise based on
21 the science and the clinical experience as opposed
22 to organizational dependence, for the reasons that
23 we have indicated the reasons for the subcommittee
24 are.
25 DR. LEE: Ajaz?
46
1 DR. HUSSAIN: The plan is to move forward
2 and actually hold the first meeting of the
3 subcommittee to coincide with the next meeting of
4 this committee. I think Larry has already looked
5 at individuals he wants to be on this committee,
6 and I think after this meeting we will be moving
7 forward, contacting them and actually putting the
8 subcommittee together.
9 DR. LEE: I am delighted to see this topic
10 on the agenda. I think it is good to have a
11 somewhat formalized system of subcommittees working
12 with this full committee and also with the Office
13 so that there will be tighter integration and
14 continuity and a sense of progressiveness.
15 Are there other questions before we let
16 Dr. Lesko off the podium? If not, we are doing
17 very well. Thank you, Larry.
18 DR. LESKO: Thank you.
19 DR. LEE: Yes?
20 DR. HUSSAIN: One question would be since
21 the thought process is clinical pharmacology,
22 followed by manufacturing, pharm tox and
23 microbiology are on the tabl, does the committee
24 have any thoughts on what the priority should be
25 with respect to the next few committees? Clinical
47
1 pharmacology, we thought, was the highest priority
2 committee to move forward. What do you thing the
3 other priority should be for the rest of the
4 disciplines?
5 DR. LEE: Shall we turn to the industry
6 representatives?
7 DR. SHARGEL: I would think manufacturing,
8 from my perspective. I don't know if Efriam would
9 agree.
10 DR. SHEK: Yes, I think as you raised the
11 thing with regard to compliance and GMP issues,
12 there are a lot of activities going on there.
13 DR. BOEHLERT: I would agree with the
14 manufacturing, and I also would suggest that you
15 broaden the area to include things like product
16 development because they are all tied together. It
17 is not just manufacturing of a finished product, an
18 active ingredient or the control but product
19 development is definitely tied in, as we found with
20 PAT. That is a very important part of the process.
21 DR. LEE: Well, it looks like the
22 committee is fairly quiet this morning. We are
23 ahead of schedule. Shall we take a break?
24 DR. HUSSAIN: Yes, we could and then we
25 can get started with the next part.
48
1 DR. LEE: All right. Let's come back at
2 about, shall we say, 9:35? Thank you.
3 [Brief recess]
4 DR. LEE: I have been asked about why I
5 didn't get a conversation going before the break
6 because I do know that we have some substantive
7 issues we need to talk about for the rest of the
8 day. Kathy whispered in my ear that she new
9 something about the difference between a
10 subcommittee and a committee, and I thought it
11 would be very useful for us to hear what the
12 regulation has to say.
13 MS. REEDY: The structure is codified in
14 FACA, the Federal Advisory Committee Act, for
15 subcommittees and their relationship to parent
16 committees and 21 CFR Part 14 delineates the report
17 system, and it is as was described. So, it is
18 codified.
19 DR. LEE: In other words, we cannot do
20 whatever we want.
21 [Laughter]
22 Now we are going to the next agenda item,
23 which is on draft guidance, food effect BE studies.
24 You all have the agenda, and i would like to invite
25 Dale Conner to introduce the topic.
49
1 Draft Guidances: Food Effect BE Studies
2 DR. CONNER: Good morning. First off,
3 before I start I would like to thank Drs. Ian
4 Wilding and Aziz Karim who have graciously come
5 here to help us and the committee out. They are
6 both experts who have worked in this area before,
7 and the committee can call on them for opinions in
8 this particular area and I am sure they will have
9 some interesting things to say, perhaps not all
10 agreeing with me but that is what makes it
11 interesting.
12 [Slide]
13 It is my job today to introduce this
14 topic, and then Dr. Ameeta Parekh will do the bulk
15 of the work by actually showing the data and some
16 of the thinking in that regard. I am going to try
17 and give some background on this because one of the
18 issues I found, even among the experts, is when you
19 talk about--most of this topic is about
20 bioequivalence and people often get confused and
21 they sometimes mix up issues that are pertinent to
22 bioavailability to those of bioequivalence.
23 Sometimes the issues and the endpoints in what you
24 are trying to accomplish are quite different. So
25 in the next couple of slides you are going to see
50
1 quite a bit of information comparing BA and BE,
2 bioavailability and bioequivalence, and that is
3 mainly to try and introduce those topics to make
4 sure that we keep each one straight and separate.
5 As my slide says, this is based on
6 discussions of a portion of the new FDA proposed
7 draft guidance. You will note from the slide that
8 this replaces another draft guidance that was out
9 for quite a few years, and has some substantial
10 changes over that original. Larry keeps correcting
11 me but I would say that we have been working on
12 this draft guidance anywhere from about 7 years to
13 12 years, depending on how you count it. When you
14 look at the guidance you are amazed that it took us
15 so long. However, it has proven to be a very
16 difficult enterprise and has gone to a lot of
17 iterations, but I think that we, at least the
18 authors, are content that this is something that
19 was worthy to go out and be discussed in the
20 public.
21 That is, indeed, what we did. The draft
22 guidance was issued in October, 2001 and went
23 through a comment period. We received comments
24 back and basically some of the issues we have
25 before you today are based on those comments. We
51
1 will talk about what those issues are.
2 [Slide]
3 Basically, I have started off by saying
4 why do we do these studies? Why do we do
5 bioavailability studies and why do we do
6 bioequivalence studies, and what is the nature of
7 the studies? Basically, the bioavailability
8 studies are mostly done in NDA type of efforts, IND
9 or NDA. They attempt to be descriptive and to
10 understand how the drug substance and also the drug
11 product, the formulation, behaves; how it is
12 absorbed, over what time course; what factors
13 affect that absorption; and also the interaction of
14 the drug substance with whatever proposed
15 formulation is made. So, the BA part is very much
16 new drugs or an NDA type of question of how does
17 this work. How does the drug behave? And, how do
18 formulations effects affect that knowledge?
19 When we get to bioequivalence it is
20 somewhat different in that, at least if you look at
21 the way we do generic drugs or pharmaceutically
22 equivalent products, the drug substance is the
23 same. So, the BA part is merely a comparison of
24 two formulations. If it is a generic drug type of
25 situation, an NDA type of situation, the
52
1 formulations are pharmaceutically equivalent. So,
2 if you have an immediate release tablet you are
3 comparing it against an immediate release tablet.
4 If it is a solution, it is against a solution. If
5 it is a suppository, it is against a suppository
6 and they contain the exact same amount of drug
7 substance. So, the comparison is entirely on how
8 that formulation performs. That is basically what
9 I have said here.
10 What we are interested in is, is there a
11 differential effect in this particular case, when
12 we talk about food studies, of food on the
13 formulation compared. That is not the same
14 question you would ask early on in the BA, is there
15 a food effect? It is a question of is the food
16 effect different between the two formulations. So,
17 we are looking either for a differential food
18 effect of a lack of a differential food effect. In
19 other words, are they equivalent in the fed state?
20 This can be a direct effect of food on the
21 formulations or it can be based on physiologic
22 effects because, as we all know, food has very
23 significant physiologic effects on the GI tract and
24 a number of other systems as well.
25 So just to keep it in perspective, when we
53
1 are talking about BE, and a lot of these issues and
2 discussions that we are going to talk about are
3 more about bioequivalence issues than
4 bioavailability, keep in mind that it is strictly a
5 formulation question or a comparison of two
6 formulations containing the exact same drug
7 substance.
8 [Slide]
9 I have expanded the first part into a
10 series of questions, and these might be termed
11 questions either the FDA asked, or a sponsor, or
12 someone who is trying to develop a drug or drug
13 product to answer the questions or points that I
14 brought up originally.
15 First I am going to go over the BA or the
16 bioavailability. The first one is does the food
17 affect the drug substance? It is really a question
18 of is there some property of that drug substance
19 whose bioavailability or pharmacokinetics is
20 affected by food? That almost says that that
21 effect is going to occur within reason, no matter
22 what formulation I put it in. It is just simply a
23 property of the drug substance.
24 Furthermore, does food affect the
25 formulation performance? When I use the term
54
1 formulation performance, I mean how that
2 formulation--that tablet, that capsule, that
3 suppository, whatever--releases the drug substance
4 into an available state, usually into solution.
5 So, does the food actually affect, in effect, the
6 tablet or the formulation as a delivery system in a
7 way that delivery system works or functions?
8 Sponsors always ask, well, what food effect
9 bioavailability studies should be done in an NDA?
10 How should they be analyzed? Is it simply a
11 descriptive effect with little statistics, or is it
12 actually a rigorous statistical method that should
13 be applied to make, for an NDA, eventual labeling
14 statements? Are the effects statistically
15 significant if I am doing statistics and,
16 furthermore, beyond the statistical part of it, are
17 those effects clinically relevant? So, I may get a
18 statistically significant effect but, you know,
19 does it really mean anything in a clinical sense?
20 [Slide]
21 For BE the considerations are somewhat
22 different and in some cases significantly different
23 if you read carefully. Does the food affect the
24 formulation to different extents? Again, we get
25 back to what I said originally. This is looking at
55
1 differential effects of two formulations. what we
2 are interested in is perhaps two formulations that
3 are pharmaceutically equivalent and in a fasting
4 state perform exactly the same way but when I give
5 them in the fed state I see a big difference in the
6 way they perform. One is what is dramatically
7 affected by food and the other one perhaps stays
8 the same or goes in the opposite direction.
9 That is what I am interested in
10 discovering with these studies, are these products
11 equivalent and, therefore, interchangeable when I
12 give them with food? Of course, the sponsors and
13 even FDA reviewers often ask what fed BE studies
14 need to be done to determine this. What strengths
15 need to be studied? Do I need to do every single
16 strength in the product line, or is one strength
17 enough? And, we have ways in our regulations that
18 instruct us on how to do that. How should these
19 studies be analyzed, which is part of the questions
20 we are getting into today, and what are the BE
21 acceptance criteria is another part of the issue
22 that you are going to be talking about today.
23 [Slide]
24 Just to briefly discuss, and Ameeta will
25 go into a little bit more detail on what the actual
56
1 comments were from the industry, as I said, we put
2 out the draft guidance for public comment. There
3 was a comment period. We received comments from
4 about 13 sources. Currently only 11 of them were
5 submitted in the official accepted way, which is to
6 the docket where all the public can look at them.
7 Two more were sent in e-mails and we are trying to
8 get those people to also submit to the docket as
9 well, which is the proper method. Just as an
10 aside, if any of you do submit comments to any
11 draft guidance, whether this one or any other,
12 please submit them to the docket because that is
13 the proper way, and instructions are usually
14 included with the draft guidance about how to
15 properly submit those.
16 So, the total number of sources, including
17 two that were not submitted to the docket, are 13.
18 The approximate number of comments was about 130.
19 I say approximately because some of them were text
20 comments and it was very difficult to determine
21 where one comment stopped and the next one began.
22 So, I am saying approximately 130 by our count. It
23 is not 130 different and unique comments. A lot of
24 them were duplicates, either commenting on the same
25 thing or actual identical duplicates of the other.
57
1 So, people obviously collaborated and sent in the
2 same comments under different covers. So, there
3 are really not even 130 unique comments.
4 When we distilled all those down--we
5 actually took a couple of months and read them over
6 very carefully and complied them and what we have
7 come to you today with, based on those comments, is
8 two issues that we felt were very significant to
9 the commentors and very significant to the FDA as
10 far as how the comments came in and the amount of
11 controversy that those particular points raised.
12 [Slide]
13 The first of two issues in the draft
14 guidance provide for a waiver of BE studies under
15 fed conditions based on biopharmaceutics
16 classification system. I think you have all
17 probably heard talks in this committee before about
18 what the BCS, the biopharmaceutics classification
19 system, is but I will give a very brief review, and
20 you will hear plenty about that this afternoon,
21 probably as much as you can handle.
22 Specifically, the guidance tried to allow
23 for the waiver of fed bioequivalence studies for
24 Class I drugs. If you recall, under BCS the Class
25 I status is achieved when a drug substance is
58
1 highly soluble, highly permeable and the drug
2 product is rapidly dissolving. So, one has to have
3 all of those three to be granted a waiver of
4 fasting studies under the current final BCS
5 guidance. As I say down below, when these
6 characteristics are proven about a product or a
7 drug substance through scientific studies, then
8 that is suitable for waiver under Class I status.
9 I think the question comes down to should
10 we also waive fed bioequivalence studies under this
11 same rationale? I mean, if we put the science
12 together that says that we can not only waive the
13 fasting studies but we can also waive for many
14 products the fed studies. My interpretation of
15 this is that a deeper scientific question is when
16 you have a Class I drug that is classified as such,
17 does something that the food does change it into a
18 different category? I think that is the heart of
19 the question really. Do you believe or have any
20 evidence that you would have a Class I drug clearly
21 categorized that you would waive in the fasted
22 state, yet, something about giving it with food
23 changes its characteristics? And, I am talking
24 about the characteristics that I have listed. For
25 example, giving food with a drug substance might
59
1 change its permeability or might change its
2 solubility. Or, giving it with that product may
3 slow down the dissolution of the dosage form to
4 such a degree that it could no longer be considered
5 rapidly dissolving. Therefore, effectively it
6 would essentially transfer that into another class
7 which we wouldn't normally waive. I think that is
8 the basic question.
9 [Slide]
10 This is a study that I have adapted from a
11 talk that Ajaz gave. I think the question is,
12 well, why is it BCS at all? Why is it so
13 important? I think the justifications are that we
14 have a need to decrease or reduce our reliance on
15 in vivo studies as much as possible. A part of the
16 regulations actually instruct us that no
17 unnecessary human research should be done. So,
18 when we get to the point where the science advances
19 to such a state that we consider those studies
20 unnecessary, then the regulations actually instruct
21 us that we shouldn't be doing them anymore, or that
22 we should find some method of decreasing those in
23 vivo studies.
24 The additional factor is that, the more in
25 vivo studies you do, the more the time of drug
60
1 development is extended and the more time on our
2 part to review those studies as well. So, if good
3 science dictates that those studies are unnecessary
4 and that we can make the same decisions effectively
5 with, say, only in vitro information, then the
6 regulations, common sense and good practice force
7 us to go and actually decrease the number of in
8 vivo studies.
9 [Slide]
10 The second issue that came out of the
11 comments, and probably the second significant part
12 of this guidance is a proposed change in how we are
13 going to be analyzing the fed bioequivalence
14 studies. As you may recall, for studies currently
15 that are done in the fed state for bioequivalence
16 the criteria are that the geometric mean of the
17 ratios has to be within 80 to 125. So, there is no
18 real analysis of the variability of the comparison
19 or variability of the products as we do with fasted
20 studies.
21 So, the second issue of the proposal is to
22 change the criteria for those fed bioequivalence
23 studies to true equivalence criteria, identical to
24 what we do with the fasted studies as well. This
25 approach would also be used for NDAs to say that if
61
1 a BA study which is fed against fasted was shown to
2 be not equivalent under this criterion, then it
3 would be labeled as having a food effect.
4 For the fed BE studies it would say that
5 two formulations are truly interchangeable. It is
6 a scientifically and statistically rigorous
7 approach that we already use in other types of
8 studies, especially the fasting studies, to say
9 that two products are interchangeable or
10 switchable.
11 So, the questions that I pose under this
12 issue or the questions that I think this distills
13 down to are in two parts. These reflect what the
14 concerns of the commentors were. A good deal of
15 the comments were from industry. The first is, is
16 an equivalence approach desirable? You know, I am
17 guessing, purely guessing that if you went out to
18 physicians or the public patients and said when you
19 switched from, say, a brand name to a generic, do
20 you want to be assured that when you take this with
21 food that it is truly interchangeable? You know,
22 perhaps the naive answer would be yes, of course, I
23 want that. The second question is how much does
24 this cost?
25 Number one, is it worth it and the second
62
1 one is in doing this are we going to be increasing
2 dramatically the number of subjects that are
3 studied and, therefore, not only the number of
4 people exposed in these trials but also the dollar
5 cost of drug development and eventual dollar cost
6 of the product? Again, it is a benefit versus cost
7 type of equation.
8 I think Ameeta will show you we did a
9 survey of some of the studies, food studies done
10 under ANDAs under current practices and what type
11 of a change we would predict based on the data of
12 studies that were done in the current way.
13 Approximately how many studies would pass under the
14 current power and how many wouldn't need to have an
15 increased power and, therefore, increased subjects?
16 Basically, that is the introduction to the two
17 issues and now I will turn it over to Dr. Ameeta
18 Parekh who will go into a lot more depth and show
19 you some of the data that we have put together to
20 support these issues.
21 Science Background and Issues
22 DR. PAREKH: Thanks, Dale. That was a
23 nice comprehensive overview of the different
24 components of the food effect bioavailability and
25 bioequivalence studies guidance.
63
1 Since Dale started out with a comment on
2 how long we have worked on this guidance, I would
3 like to add a little bit to it because I have been
4 with this guidance throughout. Just to clarify the
5 history, I think we, as the agency, started looking
6 into these since mid-'80's when theophylline issues
7 surfaced and one of our visitors here, Dr. Aziz
8 Karim, was directly involved in that. Since then,
9 we started looking at the science of food effect
10 studies. I would say that for the last ten, twelve
11 years that Dale mentioned we were discussing the
12 science of food effect bioavailability studies.
13 Specific to the guidance though, we have been
14 looking at the guidance for the last five years.
15 That is a reasonable amount of time but, given the
16 complexities, we are trying to make sure that
17 everything is ironed out.
18 I would like to take this opportunity to
19 acknowledge the food effect working group who
20 contributed to the development of the guidance, and
21 also several other people who helped in this
22 effort.
23 [Slide]
24 I will just start with some background.
25 As Dale mentioned, the draft food effect
64
1 bioavailability-bioequivalence studies guidance was
2 published in November of last year and there were
3 public comments that we received. We got comments
4 from 11 sources to the docket but there were two
5 others, as Dale mentioned, that we are trying to
6 get to the docket as well. There was a total of
7 about 130 comments and, as Dale mentioned, several
8 were repetitious. A lot of them were editorial,
9 format type of comments, but there were several
10 that were very good scientific comments and we are
11 looking through these. We have gone through all
12 the comments and we have identified two primary
13 issues that represent a change from our current
14 position. We have taken these two comments for
15 discussion with the advisory committee meeting
16 today.
17 The advisory committee was presented with
18 a background package that contains these two
19 issues. These two issues were identified in the
20 package, and related to these two issues, we also
21 have a list of questions that we will try to focus
22 on today.
23 [Slide]
24 Again, I am going to reiterate something
25 that Dale mentioned already but I think it is
65
1 important to make a distinction between the food
2 effect bioavailability and the fed bioequivalence
3 studies here. The reason I think it is very
4 critical is that the rest of the discussion really
5 hinges on this discussion. Just to emphasize, we
6 are not going to discuss the food effect
7 bioavailability part of the guidance today. We are
8 going to stay focused on the two issues that Dale
9 mentioned that are related to the fed
10 bioequivalence studies.
11 But just to reiterate what the
12 distinctions are, the food effect bioavailability
13 studies, the ones listed on the top, are typically
14 sent with new drug applications, NDAs, and the
15 question here is for companies developing a new
16 product there is one product which is the test
17 product and how does this test product perform
18 under fed conditions as compared to the fasted
19 conditions? When we say "perform" we are really
20 looking for measures of exposure. How is the
21 exposure, the rate and extent, different under fed
22 conditions as compared to the fasted conditions?
23 If there is a difference, how clinically relevant
24 is this difference and how should it be labeled?
25 Basically, as you can sense, the question is that
66
1 of prescribability. Typically, we ask this
2 question of all new chemical entities, of all new
3 products, new formulations.
4 The fed bioequivalence studies, on the
5 other hand, are typically submitted to ANDAs. Here
6 the question is I have two formulations; one is
7 already on the market. Here is an ANDA product
8 that is likely to be switched with this other
9 product. How similar are they under these
10 conditions of use? So, the question here is, is
11 the test product, which is the ANDA product, close
12 enough to the reference product under fed
13 conditions that they could be switched in the
14 patient population? The question here is that of
15 switchability and approval. All modified release
16 formulations for ANDAs typically are expected to do
17 these studies. For immediate release dosage forms,
18 however, whether or not a fed BE study is done, it
19 really is label driven.
20 The current criteria, as Dale mentioned,
21 for approval of these fed BE studies is hinged on
22 acceptance of ratio within a certain range
23 typically or commonly known as point estimates.
24 So, it is basically the geometric mean ratio of the
25 test and the reference product, called point
67
1 estimate, to fall within a certain boundary. In
2 other words, is the test product given under fed
3 conditions within a reasonable distance on average
4 from the reference product given under fed
5 conditions? Note that the acceptance is based on
6 point estimates. The distribution around this is
7 not taken into consideration based on the current
8 criteria.
9 [Slide]
10 The two items that I have listed with an
11 asterisk are the two issues that we are going to
12 discuss today. Issue number relates to immediate
13 release dosage forms, are there some types of
14 products that could be classified as BCS Class I
15 drugs and BCS Class I drug products, rapidly
16 dissolving? Could we comfortably say that we could
17 waive those fed BE studies in vivo provided there
18 is in vitro data to support our comfort level on
19 the equivalence of those products? So, basically
20 using similar dissolution profiles as a surrogate
21 for the measure of in vivo fed bioequivalence, and
22 this is not the first time we are approaching this
23 premise. We have done this in the recent past with
24 the fasted BE studies as well. So, here we are
25 trying to extrapolate this to the fed BE studies.
68
1 The second issue for discussion, again as
2 Dale mentioned, is implementation of true a
3 statistical equivalence approach and the criteria
4 for the fed bioequivalence studies. As I mentioned
5 earlier, right now we use point estimates and we
6 are considering maybe moving to a more statistical
7 approach of confidence intervals within a certain
8 range, and that is what we currently use for the
9 fasted BE studies.
10 [Slide]
11 I will discuss these two issues
12 sequentially. Where possible, I will give a
13 scientific rationale and, where available, I will
14 provide some confirmatory and supportive data.
15 Some justification for waiver of BCS Class I, and
16 Dale has already touched upon that, but the primary
17 supportive data that I am going to provide is from
18 our University of Tennessee studies that were
19 funded by the FDA.
20 [Slide]
21 Just to go into the scientific basis for
22 this, and again we are revisiting this; this is
23 nothing new, we use these for the fasted BE studies
24 waiver and we are really extrapolating that to the
25 fed BE situation now. Just to emphasize, the BCS
69
1 Class I drugs and drug products are defined as
2 those that are rapidly dissolving across a range of
3 pH's, therefore, the formulation effect is
4 minimized. So, we have kind of negated any
5 formulation effect if there is any. Once
6 dissolved, the belief is that once you take this
7 product it is practically in solution very rapidly.
8 So, in solution the drug substance, with it comes
9 from formulation A or B it is the dug substance,
10 and the drug substance is highly soluble and highly
11 permeable and, therefore, well absorbed.
12 So, given that there is minimal
13 formulation effect, given that the drug substance,
14 whether it comes from formulation A or B is well
15 absorbed, there are several examples, and Dr. Aziz
16 Karim has published on this, several BCS Class I
17 drugs have no food effect. They are well absorbed.
18 They are pH independent or, I should say, they are
19 similar between the two formulations and generally
20 there are no food effects unless they are high
21 first-pass drugs or if there is some complexation
22 but both of these are drug substance effects rather
23 than the formulation effect. Therefore, the bottom
24 line is if there are two formulations of the same
25 drug that have minimal formulation effect, BCS
70
1 Class I drugs, rapidly dissolving drug products,
2 they should be bioequivalent and if, in fact, there
3 is some effect it is probably because of the drug
4 substance and, therefore, we could probably waive
5 fed BE studies for the two products.
6 [Slide]
7 To provide some supportive data that we
8 collected from FDA-funded studies at the University
9 of Tennessee, the objective of these studies--there
10 were two studies and the objectives were to
11 investigate the relative bioavailability of two
12 FDA-approved generic products administered under
13 fed conditions. So, the two model drugs that we
14 picked were metoprolol and propranolol. They are
15 BCS Class I and, in fact, metoprolol happens to
16 have high solubility, high permeability boundary
17 but they are, in fact, BCS Class I drugs. The two
18 generic products that we chose for each of these
19 drugs were based on the furthest possible in vitro
20 dissolution. So, we chose the worst possible
21 scenarios that we had for these two formulations
22 for metoprolol and propranolol independently.
23 [Slide]
24 I will share some results with you for
25 these bioequivalence studies that we performed
71
1 under fed conditions. Metoprolol, 18 subjects. As
2 you can see in the last column, it met the
3 confidence interval. The point estimates were
4 reasonably close, three percent for AUC and seven
5 percent for Cmax. Again, note that metoprolol is
6 highly soluble, highly permeable boundary
7 conditions, and note that both these drugs have an
8 increase in bioavailability with food and that is
9 theorized to be partly due to the high first-pass.
10 So, in spite of this big food effect that we see
11 for propranolol and metoprolol, we used those as
12 the challenge drugs for testing this hypothesis of
13 BCS Class I potential waivers and metoprolol shows
14 that, yes, it could meet bioequivalence.
15 [Slide]
16 The same thing was shown for propranolol.
17 Again, propranolol is a high solubility, high
18 permeability drug; much more increase in
19 bioavailability with food. When I say increase in
20 bioavailability wit food, I am talking about
21 fed-fasted comparison and also again for point
22 estimated differences, two percent on average; five
23 percent on average for EC and Cmax.
24 [Slide]
25 Just for completeness, I will show the
72
1 hydrochloric acid. I forgot to mention that. The
2 propranolol that was used was from a combination
3 product, propranolol hydrochlorothiazide. The
4 consideration here is that there was no
5 interaction; there is no pharmacokinetic
6 interaction of propranolol with
7 hydrochlorothiazide. We thought this would be a
8 challenge to propranolol using a drug that doesn't
9 have high solubility, high permeability in
10 combination with propranolol. So, we used a
11 combination product for the test of propranolol as
12 the model for BCS Class I. So just for completion
13 I am showing you the hydrochlorothiazide data as
14 well. You can see that met bioequivalence as well.
15 [Slide]
16 Conclusion: Formulation factors are
17 likely to play a minor role in the bioavailability
18 determination of BCS Class I rapidly dissolving
19 drug products. Studies with metoprolol and
20 propranolol, which are BCS Class I rapidly
21 dissolving drug products, demonstrated
22 bioequivalence under fed conditions and, therefore,
23 the data supports the BCS-based recommendation for
24 the waiver of fed BE studies.
25 [Slide]
73
1 I will move on to the next issue, issue
2 number two, again reiterating what Dale had
3 mentioned, that this is basically saying we are
4 going to try and see if a different approach,
5 implementation of a true statistical equivalence
6 approach for fed BE studies would be a better
7 approach to go with the fed BE assessment. Right
8 now, as I mentioned, we go with the point estimates
9 for the ratio of the test and the reference,
10 geometric mean ratios of the test and the
11 reference. Here we are proposing the same criteria
12 that we used for the fasted BE studies, namely, 90
13 percent confidence intervals for the test and the
14 reference, log transformed ratio to fall within a
15 range which is 80 to 125. This is both for AUC as
16 well as Cmax. With this approach, the question I
17 think we need to ask ourselves--
18 DR. MOYE: Excuse me. I am sorry to
19 interrupt. I have to ask a question just to make
20 sure I understand what this is about. Can you go
21 back for a second, please? When you talk about the
22 criteria for the 90 percent confidence interval,
23 are you saying that the entire confidence interval
24 has to fall within the 80-125? Overlapping is not
25 sufficient? It must lie completely within?
74
1 DR. PAREKH: Right. So, it is a
2 bioequivalence approach and we use the same for the
3 fasted BE studies.
4 DR. MOYE: Thank you. Sorry to interrupt.
5 DR. PAREKH: Does that mean I can start
6 talking?
7 [Laughter]
8 [Slide]
9 All right, the question is what is the
10 purpose of these fed BE studies, and it depends on
11 what your answer is. If your answer is to assure
12 interchangeability of two formulations, and I snuck
13 in another question, how certain do you need to be?
14 then the answer is right there. This is nothing
15 new. We have used these for fasted BE studies. If
16 your answer is, yes, we want to be sure that they
17 are interchangeable products under fed conditions,
18 then we already have these criteria in place. So,
19 the regulated criteria for the BE studies right now
20 for interchangeability assessment is 90 percent
21 confidence intervals for the ratio of population
22 geometric means for the test and the reference
23 treatments to fall within 80 to 125.
24 [Slide]
25 But every good thing I guess comes with a
75
1 price. So the next question relates to what is the
2 price for this, and are these criteria likely to
3 increase the regulatory burden? We are concerned
4 about that too. So, what we did was, rather than
5 just putting it in place, we thought let's go and
6 see what it means if people will consider these
7 criteria for fed BE studies.
8 So we went back and did a retrospective
9 analysis for the ANDA database that we had. It is
10 a partial analysis. We took a subset of 40 ANDAs.
11 I just counted and I think there were about five
12 that were repetition drugs; 35 were independent
13 drugs. We looked at the fed-fed BE aspect of these
14 ANDAs that were turned in and reviewed in the
15 Office of Generic Drugs.
16 So, we looked at the fed BE studies.
17 Remember, these studies right now are not powered
18 for meeting the confidence interval criteria. That
19 is an important thing to keep in mind. Right now
20 the criteria, as I mentioned earlier, is point
21 estimates to fall within a range. With that, we
22 did consider are we looking at a biased piece of
23 data and we thought not really because these
24 studies are not powered for confidence intervals.
25 These are really just assessment of point estimates
76
1 being close enough. So we thought let's go back
2 and recalculate the 90 percent confidence intervals
3 on these fed-fed BE studies. So, we did that with
4 40 ANDAs.
5 [Slide]
6 This slide summarizes the results of this
7 small pilot retrospective analysis that we
8 conducted. Of the 40 ANDAs, as shown in this pie
9 chart, 35 passed the confidence interval. So you
10 could say 87.5 percent of this small subset made it
11 in spite of the fact that these were not powered
12 for confidence intervals. There is a small subset
13 that didn't make it and, again, keep in mind that
14 these studies were not prospectively powered for
15 confidence intervals.
16 For those five ANDAs that failed to meet
17 the 90 percent confidence interval, it doesn't
18 necessarily mean that they were not bioequivalent
19 if they were powered right. If you look at the
20 numbers on the top, that represents the confidence
21 intervals for all of those five that didn't make
22 it. But a small subset did not make the confidence
23 interval criteria. However, it was a small subset
24 and, keep in mind, these studies were not powered.
25 Of the five, there were two that failed on AUC and
77
1 there were three that failed to meet the confidence
2 intervals on Cmax.
3 [Slide]
4 In conclusion, if the current criteria for
5 fed bioequivalence studies, which is point
6 estimate, were to be changed to confidence
7 intervals a retrospective analysis of the existing
8 data suggests that for most studies no increase in
9 number of subjects would be necessary, however,
10 there will be a small subset that may need a larger
11 sample size.
12 With that, I want to summarize and say
13 that there are situations where in vitro
14 dissolution comparisons could suffice or could
15 serve as an acceptable surrogate for in vivo
16 bioequivalence studies, the case being BCS Class I
17 rapidly dissolving drug products. A waiver for in
18 vivo bioequivalence studies, in this case fed
19 conditions, could be considered. However, when the
20 studies are conducted, depending on what the
21 question is, if the question is what is the purpose
22 of these studies, the fed BE studies--is the
23 purpose to address a switchability question, then
24 if so, we need to address the appropriate
25 statistical criteria in that situation. Thanks.
78
1 DR. LEE: Thank you very much, Ameeta.
2 There are two questions put before us, and I have
3 asked Marvin Meyer to digest this information and
4 provide us with some perspective. Before we start,
5 since we have plenty of time, what is the
6 definition of food? This is a half-serious
7 question.
8 DR. PAREKH: That definition of food took
9 us the first twelve years.
10 DR. LEE: I see.
11 [Laughter]
12 DR. PAREKH: We went through a lot of
13 scientific discussion trying to debate what is
14 food. There were papers that said there is no such
15 thing as the right meal. You could be eating
16 something; I could be eating something totally
17 different. Rather than addressing it as a social
18 question, we thought we could address it as what is
19 the regulatory question here. The regulatory
20 question is what happens when I take a drug with
21 meals. Given all the physiology of food
22 effects--gastric emptying time, cholecystokinin,
23 all those things, bile acids, pH changes--we went
24 through a lot of literature. We went through the
25 examples that were tested for theophylline which
79
1 were bench-marking the meals that could be
2 discriminating. We thought let's take a meal that
3 would represent the worst case scenario for maximum
4 perturbation of the gut, and let's use that as the
5 meal. The meal that was chosen was similar to the
6 meal that was shown to be discriminatory in those
7 early theophylline studies.
8 DR. LEE: So, we are asked to think about
9 food that way. Also, I suppose we should think
10 about the subject not as pediatrics or geriatrics
11 but the average population in age. Right?
12 DR. PAREKH: That is right.
13 DR. LEE: And also think about Class I
14 drugs as the average of that range. Right? So,
15 these are the boundary conditions. I am beginning
16 to complicate matters.
17 DR. HUSSAIN: Yes, I am not sure. With
18 respect to bioequivalence, we have always tried to
19 have sort of a general population to study that.
20 The issue essentially is making sure in vivo that
21 the release of the drug from the product is
22 essentially similar. So, that is the question we
23 are asking. With respect to special populations, I
24 think that is more a bioavailability question, not
25 a bioequivalence question. So, if we can keep
80
1 those two separate.
2 DR. LEE: Thank you.
3 DR. CONNER: Just an aside, the meal was
4 very high in fat, the meal that Ameeta was talking
5 about. After a lot of discussion and a lot of
6 research, they came up with a very high fat meal.
7 Now, if you go to different places in the world or
8 even in the United States, that is not necessarily
9 a representative breakfast, hopefully, that most
10 people eat. If they do, their arteries are going
11 to be in very bad shape after a few years. So, in
12 another country, that country may have chosen to do
13 a much more representative meal. For instance, I
14 have reviewed some ANDA food studies for Japan
15 where they took a typical Japanese breakfast which
16 was much, much different than what we are talking
17 about here. It is interesting to look at those
18 side by side. However, we chose something that
19 would have the highest likelihood of being a
20 challenge to the dosage form and the drug
21 substance.
22 DR. LEE: Okay, I wanted to make sure we
23 understand it because now we are looking at version
24 two and pretty soon we will be working on version
25 three.
81
1 DR. HUSSAIN: I think in terms of
2 standardization, the question you raised also goes
3 to the standardization of the meal because this is
4 a quality assurance type of a test. We went to the
5 commercial sources that provide this reproducibly.
6 DR. PAREKH: Yes, we went and picked up
7 things from little fast food places. I remember a
8 few years back Hank Malinowski took a group and we
9 tried out the meal. It is a big meal. I could
10 handle it!
11 [Laughter]
12 Just to get to specifics, Dr. Lee, the
13 meal that is defined in the draft guidance is about
14 800-1000 calories, and we specify the meal as an
15 example meal but 150, 250 and about 500 calories
16 from protein, carbohydrate and fat. You don't have
17 to stick to a certain meal in terms of the
18 components as long as the fat, carbohydrate and
19 protein are similar or close to this, because this
20 is what has been tested in the literature to cause
21 the maximum perturbation. So, we want to know what
22 is the worst case scenario and so go with the meal
23 that represents the worst case scenario.
24 DR. LEE: Very well. Thank you very much.
25 I want to remind the committee that we have two
82
1 consultants, at the other end of the table, to
2 collaborate with. Yes?
3 DR. ANDERSON: On page two of the handout
4 you have something about similar dissolution
5 profiles. Would you comment on how close the
6 dissolution profiles have to be in order to qualify
7 for this?
8 DR. HUSSAIN: In terms of the fasting
9 study where the BCS guidance was first used, the
10 rapid dissolution is defined in terms of a time
11 limit in terms of the rate of dissolution. It has
12 to essentially dissolve within 30 minutes, and it
13 has to dissolve in a pH range of, say, 1 to 6.8 and
14 three different pH conditions. The similarity is
15 that it has to be within about 10 percent. The two
16 profiles should be within plus/minus 10 percent; it
17 is an approximate similarity.
18 DR. ANDERSON: Plus or minus, yes.
19 DR. LEE: Thank you. We do have two
20 questions in front of us. We need to answer these
21 questions and if there is time we can go into other
22 questions. Marv?
23 DR. MEYER: I have a question of your
24 presentation before I get to that, and then I want
25 to make a comment before I get to that. You have
83
1 40 ANDAs that you sampled. Out of how many
2 possible does the 40 represent, and were they a mix
3 of IR and modified release? Thirdly, do you have a
4 recollection of what the point estimates were for
5 the five drugs that failed?
6 DR. PAREKH: I am glad I got up early this
7 morning and checked that. Yes, it was a mix of IR
8 and MR. We didn't select ANDAs based on a certain
9 thing; we just took 40 and there were IR and there
10 were MR. The ones that represent not making the
11 confidence intervals are a mix of IR and MR. So,
12 it is not just all MR or IR. For AUC, there was
13 one that was as high as 151. The point estimate
14 was about 20 off, so 1.2, 120. The other one was
15 also close. It was 118 or somewhere in that range.
16 You can see from the width that that is where it
17 would be.
18 DR. MEYER: So, one could argue that of
19 the five failures, the Cmax failures all could have
20 been taken care of by a few more subjects, and
21 maybe the AUC failures, the 120 and the 118 really
22 shouldn't be approved anyway.
23 DR. CONNER: You know, in looking at that,
24 and obviously I have the ability to know which
25 applies to which product, but I actually just
84
1 looked at the overall and I had the same reaction.
2 You know, when Ameeta and I were going over the
3 results I looked at those five and I said, well,
4 the Cmax, some more subjects, we didn't go through
5 the exercise of calculating how many more subjects
6 would have been required although it is perfectly
7 reasonable to be able to do that. But when I
8 looked at the AUCs I said, oh, these don't look so
9 very good to me because the point estimates,
10 although we don't have them on the slide, are
11 obviously pretty far out. I mean, they are within
12 the 80 to 125 but they are like about 120 or in
13 that range. I don't have the exact numbers. So, I
14 think that simply adding power to that, although
15 theoretically if you added enough power it might
16 squeak by, it is pretty unlikely that adding a
17 reasonable number of subjects to that study would
18 get those to pass the confidence intervals.
19 The open question still is do we really
20 feel comfortable approving those? Now, it is
21 important to say for the record that we are not in
22 any way saying that what we have done in the past
23 or what we are currently doing with the point
24 estimates, that there is anything wrong with that.
25 I don't want anyone to conclude that there is a
85
1 real hazard here. I think we have had some good
2 experience with that. Doing it this way hasn't
3 really created any clinical problems that we are
4 aware of. Our attempt here is, I would say, just
5 to tighten things up and to make a more rigorous
6 equivalence evaluation rather than, you know, what
7 is kind of a "feel good" type of approach but a
8 more rigorous type of approach in what we are doing
9 with point estimates. So, I don't think that what
10 we have been doing in the past is wrong; I think
11 this is just better.
12 DR. MEYER: One point of order, Vince. We
13 have two invited guests and I think a couple of
14 other speakers on this topic. I always wonder why
15 we don't hear from those people before we
16 deliberate.
17 DR. LEE: Because once they start
18 talking--
19 [Laughter]
20 --but I am sure that they will interject
21 at the appropriate time.
22 DR. MOYE: One advantage of moving away
23 from just using the point estimate is that you
24 really don't know what the operational
25 characteristics of it are. You have historical
86
1 information. Sometimes historical information can
2 be very leading and sometimes it can be misleading.
3 If I understand this process correctly,
4 the way it currently is now, and please tell me if
5 I am wrong and I apologize for interrupting you
6 earlier but I was in imminent danger of being badly
7 and irreversibly confused so I really needed to
8 stop and ask you--the way it currently is now, a
9 sponsor will carry out a research effort and come
10 up with an effect size, a point estimate. Even
11 though there is a standard error associated with
12 that and even though the standard error is
13 available, that standard error is set aside and the
14 question is simply asked whether that point
15 estimate is greater than 0.8 or less than 1.25.
16 The suggestion is to replace that with the
17 confidence interval of 90 percent and ask whether
18 the 0.8 to 1.25 range completely encompasses and
19 encloses the 90 percent confidence interval. That
20 is correct?
21 I am not really sure why we need to go
22 through this two-step process, the first step to
23 compute the confidence interval and then, the
24 second step, decide whether the confidence interval
25 falls completely within 0.8 to 1.25. It seems to
87
1 me in order to determine how well that is going to
2 work, again holding historical information aside,
3 it is kind of a complex computation to ask about
4 where the range of a confidence interval is going
5 to fall. So why not, as an alternative, just ask
6 the question how likely is it that the population
7 ratio will fall between 0.8 and 1.25 given the
8 point estimate and given the standard error? That
9 is a fairly easy computation to do, and you can set
10 a value for that probability. That probability
11 must be above some value, and for that the
12 computation is much more direct and, hopefully,
13 much more interpretable.
14 DR. CONNER: It is important to point out
15 that this is not a new method, which is what we are
16 talking about, which is the two one-sided test
17 procedure to determine equivalence. That is
18 something that we have been doing for quite a few
19 years for fasted studies. If you are saying that
20 this, when applied to food studies, may not be
21 totally understood I don't agree with you but I
22 take that criticism. But as far as the properties
23 of this calculation, the properties of the
24 statistics, we understand those very well. We have
25 been doing them for perhaps ten or twelve years
88
1 now, I think, on fasting studies.
2 DR. MOYE: There are two statistics here I
3 think. Are you talking about the one that just
4 uses the point estimate and asks whether that is
5 between 0.8 and 1.25? Is that the one you are
6 talking about?
7 DR. CONNER: No, no--
8 DR. MOYE: Or are you talking about the 90
9 percent CI?
10 DR. CONNER: The fasting studies are done
11 in exactly the way we are proposing to now do fed
12 studies. It was developed by Dr. Sherman and
13 others, the two one-sided test procedure. In other
14 words, what the test essentially does is run two
15 one-sided tests, one in one direction and the other
16 in the other, you know, one test one bound and the
17 other test the other bound. They are run at the
18 alpha equals 0.05 level. So, we have 0.05 on one
19 side--
20 DR. MOYE: Right.
21 DR. CONNER: --and 0.05 on the other. So,
22 the way of actually doing all this in one test, one
23 calculation, is to calculate the 90 percent
24 confidence interval so you get the 5 on one side
25 and 5 on the other, and each one of those has to be
89
1 what we have determined to be a clinically
2 significant difference. The actual operation of
3 this, for the most part the point estimates of
4 fasting studies, when we have done similar types of
5 surveys, for the vast majority of the products we
6 approve based on the fasting results the point
7 estimates don't vary by more than about 4 percent
8 either way from a ratio of 1. We have a few
9 isolated cases where we have as much as 10 or 12
10 percent, but most of them cluster right around the
11 ratio of 1, plus/minus 4 percent for both Cmax and
12 AUC. So, the operational characteristics of
13 controlling that point estimate, the experimental
14 point estimate are actually quite good.
15 DR. MOYE: It sounds like the answer to my
16 question is that this is a procedure that has been
17 well established--
18 DR. CONNER: Yes.
19 DR. MOYE: --and has been used in other
20 analyses looking at bioavailability for fed and
21 fasting. Is that right?
22 DR. CONNER: It is used somewhat in the
23 NDA world but primarily this is used to determine
24 the equivalence or switchability of two
25 pharmaceutically equivalent products. So, the drug
90
1 substance, the amount of drug substance, the type
2 of dosage, all that is held constant and most of
3 the studies we do are crossover so, you know, each
4 individual gets both products. And, we want to
5 make sure that in the end the judgment we make and
6 the generic product we approve, if someone goes
7 into their pharmacy and they are currently taking,
8 say, the brand name, if the doctor switches them to
9 this other pharmaceutically equivalent dosage form
10 they will be getting essentially the same results
11 without any distinguishable difference.
12 DR. LEE: So, you are answering question
13 2.3, what alternative approaches?
14 DR. MOYE: If you say so.
15 DR. CONNER: As an aside, I am not sure we
16 should get much into it today, but if you have
17 suggestions on how we might do this whole thing
18 better--I mean, what we are doing now is simply
19 expanding what we have done for many years to this.
20 If you have some other, you know, just general
21 comments that you might have a better method,
22 perhaps another forum might be the time.
23 DR. MOYE: Well, I wouldn't say it is
24 better at this point; I just say it is an
25 alternative and it may be simpler.
91
1 DR. LEE: Do you have slides?
2 DR. MOYE: Not right now but I can prepare
3 them.
4 DR. LEE: All right. Since the two
5 consultants were mentioned, maybe I will just take
6 the opportunity to see if they have anything to
7 say.
8 DR. KARIM: You mentioned about food
9 effect. I have been talking about food effect for
10 the last thirty years, and one of the most usual
11 and common questions asked is we never have this
12 type of meal so why does FDA do a food effect
13 study? The question here is it is not really the
14 sort of food you would be taking every day. It is
15 really performance of a dosage form under
16 conditions which would produce maximal perturbation
17 of the formulation. So, it is really a quality
18 control test of your formulation, and that is the
19 food which would produce the maximum effect. So,
20 it is not the usual food you take but it is quality
21 control type of food.
22 The second point I want to make is that,
23 in fact, it is correct that I have found that drugs
24 which belong to Class I do not show food effect
25 response in terms of AUC, and in drug development
92
1 the very first study in humans that we do is a food
2 effect study because if there is no food effect
3 response, then we are able to categorize our drug
4 as a Class I drug which, I think, is a new approach
5 of food effect response. We use it a great deal in
6 drug research.
7 One thing which I still feel hasn't been
8 covered is that food will produce, even for Class I
9 drugs, delay in absorption because 50 g of fat will
10 result in stomach emptying time, and if you have a
11 drug which is specifically used for very fast onset
12 of action--an analgesic, antiarrhythmic--you will
13 miss the point because the Tmax is not used in
14 bioequivalency assessment. So, I think the agency
15 needs to look at that before saying that the Class
16 I drugs would not require food effect response
17 because the question of Tmax has not been
18 addressed, what is the effect of a given meal or of
19 food on Tmax.
20 The third point I want to make is that if
21 a drug or formulation is labeled to be taken with
22 food, and if that is how patients take the drug,
23 then it is obvious that the bioequivalency must be
24 shown under fed conditions. I have said that again
25 and again. We should use all the statistical
93
1 criteria used under fasting state to apply to the
2 fed state.
3 I am surprised that the bioequivalency was
4 shown in even 17 to 18 subjects with food because
5 when you give the drug with food you are adding
6 another variable, and that is gastric emptying
7 time. I would be very interested to see whether in
8 a crossover situation the gastric emptying time
9 under fed condition is similar or not. I know
10 under fasting state they are very similar, but I
11 would have expected under fed conditions the
12 gastric emptying time to vary more, and I would
13 have expected that we would need quite a few more
14 subjects to do bioequivalency testing. Thanks.
15 DR. LEE: Thank you.
16 DR. MEYER: Can I ask Aziz a question?
17 DR. WILDING: Can I pick up first because
18 we do a lot of work actually visualizing what fat
19 does to gastric emptying properties in formulation
20 performance. It is certainly true that the current
21 high fat meal as put into regulatory guidance has a
22 maximum effect on the GI tract. That is, it
23 effectively stops the stomach for a couple of hours
24 in most individuals. The reality is that if you
25 put that amount of fat into the stomach, it takes a
94
1 while to realize that it has that large amount of
2 material to deal with and actually sits still for a
3 period of time.
4 What you have to recognize also is that
5 today's population eats less fat than the previous
6 populations. Therefore, what was maximal for them
7 is probably now super-maximal for today's
8 individuals. That is an issue that is worth
9 contemplating. So, I think what we see often is an
10 effect on Tmax associated with significant delays
11 in gastric emptying.
12 Now, the question is, is the CV percent
13 greater in terms of intra-variability fed compared
14 to fasted? Certainly, in our experience there will
15 be no difference between those two that will be
16 noticeable from statistical comparison purposes.
17 Now, unlike Aziz, I don't think that Tmax is an
18 issue because it is a bioequivalence issue or
19 switchability, not prescribability. Therefore, I
20 don't think in this context I could imagine where
21 there will be a Tmax difference associated with a
22 Class I drug that would lead to issues in that
23 particular regard.
24 My final comment, food effects are a
25 generic phrase and we do run risks with the phrase
95
1 food effects because it is, in many respects, an
2 active pharmaceutical ingredient issue, a
3 formulation issue, and there is the combination of
4 the API, the formulation and the food. That is
5 where I think, as Ameeta indicated, it is
6 bioavailability in terms of API alone, formulation
7 alone, but there is also a
8 bioequivalence/bioavailability issue that kicks in
9 when you are contemplating active forms of
10 ingredients of the formulation and drug together,
11 and that is the hardest one to tease out.
12 DR. LEE: Thank you.
13 DR. MEYER: Aziz, you were talking about
14 Class I and saying you have not personally seen any
15 differences in bioequivalence under fed conditions.
16 You said AUC. How about Cmax?
17 DR. KARIM: Yes, what I do is we take AUC
18 ratio fed/fasting and if they fall within 10 or 20
19 percent we categorize it as Class I drug. Now,
20 Cmax I haven't looked at in that detail, but I
21 would say probably it won't be as rigid as AUC.
22 DR. HUSSAIN: Let me sort of go to the
23 issue of Tmax that Aziz raised, and so forth, and
24 let me go through the thought process of the BCS in
25 the fasting state. One of the reasons we designed
96
1 or devised rapid dissolution criteria for the
2 fasting state was because of unpredictability of
3 the gastric residence time and the rapid emptying
4 that occurs under the fasting state, and there were
5 concerns with volume and you will see that in the
6 afternoon discussion also.
7 In fact, the 30 minutes that we have as
8 rapid dissolution criteria was for fasting state.
9 That is overly conservative for a fed state.
10 Although we are not suggesting we change that, we
11 don't believe there will be Tmax differences
12 because of formulation effects. There will
13 definitely be a shift in Tmax because of the
14 gastric emptying time. But if you are going to
15 retain the dosage form in the stomach, which is
16 essentially a reservoir, for a long period of time,
17 then you are giving far more time for dissolution
18 to become peak before it gets emptied out. So, it
19 is less of a concern under the fed condition. We
20 were more sensitive and more conservative in the
21 fasting state.
22 So, that is the reason dissolution-release
23 in vivo under fed conditions, because of the large
24 volume and because of the long gastric residence
25 time, is less of a concern. So, I think our
97
1 proposal will be far more conservative for the fed
2 state.
3 DR. MEYER: Ready?
4 DR. LEE: Yes.
5 DR. MEYER: The questions at hand then are
6 posted there, as well as in the handout we received
7 from Kathleen Reedy on April 22. The questions are
8 really broken into two sections. To what extent
9 can we waive fed bioequivalence studies for Class I
10 drug? Then, secondly, should confidence intervals
11 be applied to fed studies?
12 The first question then, can we waive fed
13 bioequivalence studies for Class I drugs which, of
14 course, are highly soluble, very rapidly dissolving
15 and highly permeable?
16 One question I have, that will come up
17 again this afternoon, is the definition of high
18 permeability. Is propranolol really highly
19 permeable? It is fine to do an intestinal
20 intubation but then what other kinds of
21 measurements can be made? My recollection is that
22 propranolol is not 90 percent systemically
23 available; large first-pass effect. How do we
24 measure high permeability if all we have is bio
25 data? I have no problem with the definition of
98
1 high permeability if it is 90 percent excreted
2 unchanged in the urine or the AUC relative to IV
3 doses is 90 percent. Beyond that, it becomes a
4 little more arbitrary. I see Ajaz is shaking his
5 head.
6 DR. HUSSAIN: No. The BCS guidance that
7 was issued in September of 2000 actually went
8 through and described several methodologies to
9 assess permeability. It also includes a method
10 based on in vitro and HeLa cell culture methods, PK
11 studies, extent of absorption. So, you have a
12 whole host or toolkit for assessing permeability.
13 You are absolutely right, metoprolol and
14 propranolol are both high first-pass effect drugs.
15 If I am not mistaken, the absolute bioavailability
16 of propranolol is 35 percent but its extent of
17 absorption is actually complete and that is the
18 basis for the high permeability class membership.
19 That is the reason we selected propranolol for the
20 challenge studies that we did at the University of
21 Tennessee. The reason is it is so sensitive to
22 food effect. In fact, there is a study from an
23 Australian hospital--I am not able to quote the
24 reference of that, but you can actually induce fed
25 effect studies of propranolol by just smelling
99
1 food; not even eating it. So, that is how
2 sensitive propranolol is to food effects.
3 DR. LESKO: I will address the same
4 question and remind us that the propranolol and
5 metoprolol were two of the drugs that we had in our
6 initial database that defined the BCS. That means
7 the permeability of these drugs was established in
8 human volunteers through intubation of the small
9 intestine. Thus, we have very accurate, gold
10 standard type permeability on those two drugs as
11 opposed to circumstantial data which might have
12 come from CACO 2 or bioavailability studies.
13 As Ajaz said, the reason we picked those
14 two recent studies in Tennessee on fed effects is
15 because we had established previously their
16 membership in the class. Propranolol is highly
17 permeable in terms of passing through the gut wall.
18 Metoprolol was picked because it was more of a
19 borderline between Class I and some other classes
20 based on its permeability characteristics. But
21 they both succeeded in those two studies.
22 DR. LEE: Larry, are you saying that it
23 has taken the metabolism into account, the
24 permeability?
25 DR. LESKO: Well, we have to separate two
100
1 things, absorption from the lumen of the intestinal
2 tract and the bioavailability. The permeability
3 refers to the passage of the drug from the lumen of
4 the intestinal tract into the blood stream. So, it
5 is talking about transversing that border. After
6 it transverses that border there may be some
7 first-pass effects in the liver that will reduce
8 the bioavailability. So, when we talk about
9 permeability we are thinking about absorption as
10 opposed to bioavailability. So, you could have a
11 drug with good absorption characteristics but
12 relatively low bioavailability if the reduction in
13 bioavailability is related to a first-pass effect,
14 say, in the liver.
15 DR. LEE: I think that maybe what Marv was
16 alluding to is the metabolism during passage across
17 the gut wall.
18 DR. LESKO: Well, if it is a 3A4 substrate
19 that is being metabolized in that passage it still
20 has permeated that segment of the wall, as
21 indicated by its high permeability.
22 DR. HUSSAIN: One other way of looking at
23 permeability is that it is essentially the ability
24 of the drug to leave the aqueous compartment that
25 is in contact with the epithelium and get into the
101
1 cell. Essentially, when we went to the BCS, as
2 Larry said, we distinguished between transport and
3 then subsequent metabolism.
4 DR. MEYER: Personally, I think I would
5 feel if the regulation said a product that is 90
6 percent bioavailable relative to IV or maybe even
7 an oral solution, that is something I can hang my
8 hat on and I don't have to worry about gut wall
9 metabolism or metabolism prior to reaching the gut
10 wall. Short of intestinal intubation, let's say,
11 the generic industry--I doubt very many of them are
12 going to do intubation type studies to establish
13 permeability, and CACO 2 and those other surrogates
14 haven't been totally proven, I don't think.
15 DR. HUSSAIN: I think we have.
16 DR. MEYER: Have you?
17 DR. HUSSAIN: Yes. I think those are
18 established.
19 DR. MEYER: Given that then, to what
20 extent does the committee feel that in-house data,
21 which I take it are partially propranolol and
22 metoprolol--
23 DR. HUSSAIN: I think the challenge
24 studies that we did in Tennessee were two products,
25 one metoprolol alone; one containing propranolol
102
1 and hydrochlorothiazide. Hydrochlorothiazide is
2 not a highly permeable drug. So, that was an
3 additional challenge that we had. So, those were
4 prospective studies designed to challenge the
5 system, and we selected two generic products to
6 have a head-to-head comparison. We didn't have
7 such data before because we have looked at
8 historical data that we have in-house and made that
9 conclusion, and we wanted to truly challenge that.
10 DR. LEE: I think the question is very
11 simple, you know, Class I and Class II and so
12 forth, fed state, fasting. I think we all
13 understood that. But I guess Marv was thinking
14 about exceptions. He was thinking beyond the
15 current definition and is not comfortable with the
16 risk.
17 DR. VENITZ: To follow-up on something,
18 Dale, that you mentioned, is there any evidence to
19 suggest that for the Class I and non-Class I drugs
20 there is a differential food effect between the
21 formulations? Because you alluded to the fact that
22 it is unlikely, and I guess based on my
23 understanding of BCS I would agree with that, but
24 do you have any experimental evidence to the
25 contrary?
103
1 DR. CONNER: I am not sure I was trying to
2 imply that it was unlikely. I think that is a
3 question for you.
4 DR. VENITZ: Right.
5 DR. CONNER: You know, how likely you
6 think it is. I posed the question because it
7 seemed to me that the critical thing is do we have
8 any examples, or do we realistically believe that
9 one exists that when we gave a product that was
10 rated as Class I that it would behave differently,
11 that it would behave like it was another class
12 which we wouldn't ordinarily waive? So, I will
13 give you some theoretical examples, and I can't
14 come up with any examples to say the food got in
15 there and this would affect both the formulations
16 equally, but if something in the food complexed
17 with the drug substance and actually formed, say, a
18 permanent or semi-permanent complexation which
19 didn't have the solubility or, more likely, didn't
20 have the permeability that the original drug
21 substance had, I mean, then your resultant effect
22 would be that it wouldn't be permeable anymore; it
23 wouldn't have the bioavailability that it started
24 out with if something in the food complexed with
25 it.
104
1 DR. VENITZ: But it would be a
2 bioavailability not a bioequivalence issue. Right?
3 DR. CONNER: Yes, but it would then mean
4 though that this BCS system that we designed would
5 technically no longer apply to it. It would not
6 necessarily then result in bioinequivalence. It
7 would take it out of the realm of the BCS system
8 into another class and, therefore, even though we
9 would think the likelihood that there would be
10 bioequivalence would not necessarily increase, we
11 would then, based on our BCS system, have to do an
12 in vivo test to confirm that. But the likelihood
13 of a differential effect on the drug substance is
14 small, very small but it would still take it out of
15 the realm of BCS.
16 DR. HUSSAIN: Let me sort of add to that.
17 I think when we were going through this development
18 we had extensive discussion on this. I said I want
19 to have a formulation that would behave differently
20 than the other one. For immediate release
21 formulations it is very difficult to come up with
22 an example, but since Dale raised the issue of
23 complexation, how can I formulate two products, one
24 which will have food effect and one which may not
25 have food effect? If I use complexation as a
105
1 mechanism, then I could include in one of the
2 formulations a chelating agent, sodium EDTA for
3 example, and that could be a trigger for saying, if
4 its a metal complex, you are essentially binding
5 the available metal, and so forth.
6 But those are sort of theoretical
7 assessments and we haven't seen any real examples
8 that actually could be achieved. When we look at a
9 waiver, we also look at the excipients and so
10 forth. So, actually in a BCS waiver we go through
11 an analysis of excipients, and so forth. So, that
12 would sort of come up and be covered under that.
13 So.
14 DR. VENITZ: So, it is correct for me to
15 assume that you haven't seen any evidence either
16 in-house or in the public literature that a Class I
17 drug shows a differential food effect?
18 DR. HUSSAIN: We couldn't find any
19 evidence of that.
20 DR. LESKO: I think I want to qualify that
21 a bit though because in trying to find those kind
22 of differences you described there are two
23 obstacles. One is that frequently you can't
24 identify the BCS class, say, in a new drug
25 application based on the data submitted. So, the
106
1 best we can work on is a suspicion of what the
2 class would be because the company had no reason
3 necessarily to define the solubility at all pH's to
4 measure permeability. So, when we looked at that
5 question to look for the exceptions, we were flying
6 a little bit blind by not knowing for sure whether
7 these were Class I drugs. So, there is that
8 aspect.
9 On the ANDA side, we are sort of a captive
10 audience to what is being submitted to the Office
11 so there are things that may be out there that we
12 don't see or aren't aware of. That may address
13 your question. But recognizing those two
14 limitations, I guess the answer would be no, we
15 don't have any direct knowledge of exceptions.
16 DR. LEE: There is another question about
17 the issue about the mechanism of absorption as
18 well. What if a drug falls in Class I because of
19 an affinity for whatever transport might be in
20 place in the gut?
21 DR. HUSSAIN: With respect to the fasting
22 study, the mechanism of absorption I think came
23 into consideration with respect to the methods of
24 permeability. For example, there is no restriction
25 that a carrier-mediated transport of an active
107
1 transport mechanism would preclude a drug from
2 being a Class I or a highly permeable drug. But
3 the methodology used to assess permeability then
4 has to be looked at more carefully. For example,
5 in the BCS guidance use of CACO 2 or in vitro,
6 essentially we don't recommend using those for
7 actively transported drugs, and so forth. So, that
8 is how we managed that process.
9 DR. VENITZ: But don't you also have a
10 restriction on dose proportionality--
11 DR. HUSSAIN: Yes. Dose linearity was one
12 of the mechanisms to address some of that question.
13 DR. LEE: Other comments from the
14 committee? Yes, Judy?
15 DR. BOEHLERT: I have a question coming
16 back to the dissolution profile when you said it
17 could be plus/minus ten percent. If bioequivalence
18 were waived and then the manufacturers were relying
19 on dissolution to show equivalence and if, indeed,
20 they had test and reference products that were at
21 the extremes of that range and one was plus ten and
22 the other was minus ten, are there any data to say
23 there would be clinical relevancy to that
24 difference?
25 DR. HUSSAIN: I think we looked at that
108
1 quite extensively, and for Class I drugs we don't
2 think there is a reason to believe that. If we
3 were looking at only one pH condition, then I would
4 not be confident with that. That is the reason we
5 request multiple pH conditions. The reason for not
6 relying on one pH condition is, for example, a
7 wheat base. If you just do the dissolution in 0.1
8 normal HCL that may not truly be reflective or
9 discriminating under, say, a less acidic condition,
10 and so forth. That is the reason we went with
11 multiple pH conditions.
12 DR. BOEHLERT: Would that imply that the
13 product would be continually tested at those
14 multiple pH conditions, or would you refer it just
15 to the 0.1 normal HCL and would that be enough to
16 show a difference in physical properties?
17 DR. HUSSAIN: The multiple pH conditions
18 come into play when there is a request for a waiver
19 or there is a substantial formulation change under,
20 say, the SUPAC. For routine quality control or
21 quality assurance you will have the traditional
22 classification.
23 DR. LESKO: I just want to clarify that a
24 bit. With the Class I drugs, when you talk about
25 dissolution it is possible to have a single time
109
1 point. In other words, if the products dissolve
2 within 15 minutes, 85 percent, then we will look at
3 that and say they are the same because that is such
4 a trivial difference. On the other hand, if the
5 dissolution goes to 30 minutes, we then would look
6 at a profile and what we are looking at is
7 basically two profiles, a test product and a
8 reference product. The statistics that are used to
9 differentiate those are called the F-2 statistic.
10 The reality is that to have an F-2 of 50 or
11 greater, which is "passing," you need to have very
12 similar profiles and they can differ by no more
13 than ten percent between the test and the
14 reference. So, you really can't have ten on this
15 side or ten on that side. It is really comparing
16 the two profiles. Generally the differences that
17 cause something to not pass an F-2 statistic occur
18 very early on, say, in the first five minutes or
19 first ten minutes where, clinically speaking, I
20 doubt that they are important but we do have that
21 standard in place to look at that.
22 DR. LEE: Bill?
23 DR. JUSKO: I am in strong agreement with
24 the theoretical and practical arguments pertaining
25 to the Class I type of drugs in relation to
110
1 bioequivalence, but I don't have a very good
2 feeling for the extent of literature that confirms
3 these observations. There were early review
4 articles and now I am hearing that it is rather
5 difficult to determine permeability of these
6 compounds so it is uncertain with a new chemical
7 entity exactly what its permeability is so as to be
8 able to preclassify it in this group.
9 Is there any better evidence for numbers
10 of drugs that have been evaluated to find that
11 there is no problem with bioavailability or
12 bioequivalence for Class I compounds?
13 DR. HUSSAIN: I think the hesitation to
14 say a drug is Class I and Class II has sort of
15 regulatory implications, in a sense. Unless we
16 follow the guidelines that we have provided to
17 classify we hesitate to say this is Class I and
18 Class II. But, clearly, we have a sense of what
19 the likelihood is, and based on that, I think
20 Ameeta did an internal survey and I think Aziz has
21 published extensively on that too. So, maybe they
22 can comment on that. So.
23 DR. KARIM: I think I agree with the
24 theoretical background that if you have a Class I
25 drug, in vitro dissolution specially F-2 tests
111
1 would be appropriate, and you don't even have to do
2 the food effect study. But, believe me, I feel
3 that determining permeability has not been
4 established, and that is a big issue. I mean, you
5 talk about absolute bioavailability of 90 percent.
6 For how many drugs do we have absolute
7 bioavailability or 90 percent? Very few. So, to
8 me, the major unknown is permeability. I think to
9 measure solubility is very easy. To measure
10 dissolution is also reasonable. That is why I use
11 the food effect response as a way of classifying
12 whether the drug is Class I or not and it works
13 very well.
14 So, to answer your question, if you have a
15 Class I drug and truly establish that it is a Class
16 I drug, then I think I am all in favor ofthe
17 guidance that you don't need to do a bio study.
18 DR. HUSSAIN: Again, I would respectfully
19 disagree with that in a sense because folks who are
20 familiar with CACO 2 and other methodologies, and
21 so forth, are very confident of their method. So,
22 our position essentially is that in vitro methods
23 are acceptable under certain conditions once you
24 have established method suitability, and so forth.
25 And, just relying on a food effect study to
112
1 classify a drug was not an acceptable method in our
2 guidance. The reason is that permeability is based
3 on extent of absorption and you do see food effect
4 for highly soluble, highly permeable drugs that
5 have a high first-pass effect, and those are the
6 two drugs we selected for the study. So, that is
7 sort of our position.
8 DR. LEE: I think we are caught in a
9 circular argument. My sense is that question 1.1
10 is premature. Yes?
11 DR. SHEK: Just one comment, looking at
12 the way the question is being phrased--
13 DR. LEE: Yes?
14 DR. SHEK: --it talks about bioequivalence
15 about ANDAs. It doesn't say anything about the
16 existing labeling for the reference, whether that
17 indicates it might be a Class I and indicates
18 specifically food effect. Will that be taken into
19 consideration, or how is that going to be handled?
20 I don't know how many of those 40 ANDAs have
21 something in the labeling about food effect. And,
22 if we don't do the study will the labeling be
23 changed?
24 DR. CONNER: Well, I can tell you our
25 current policy for what triggers us to ask an ANDA
113
1 sponsor for a fed bioequivalence study, and you
2 have to differentiate between a food effect study
3 which asks if there is a food effect on the product
4 or the drug substance versus a fed bioequivalence
5 study where the two products are compared under
6 equivalent or the same fed conditions. The trigger
7 that causes us to ask for a fed bioequivalence
8 study is some mention of food in the innovator
9 labeling, the reference listed drug labeling.
10 People are often confused by saying, well, so it
11 has to be some positive food effect; there is a
12 change. Simply saying, you know, in the labeling
13 we have studied it and there isn't any is enough to
14 cause us to ask an ANDA sponsor for a fed
15 bioequivalence study. So, almost any reasonable
16 mention at the current time of food in the labeling
17 will cause us to ask for a fed bioequivalence
18 determination of an ANDA sponsor. I think that is
19 actually in this guidance. This question simply
20 says, okay, we have gone there; we have determined
21 that we need some kind of decision or determination
22 of fed bioequivalence studies but, further, if it
23 is a Class I drug we could still waive the
24 necessity for that in vivo study based on what we
25 have just described here and discussed. So, that
114
1 is basically our current policy and how we hope or
2 have proposed it to evolve in the this guidance.
3 DR. MEYER: I think we have to remember
4 though that permeability is drug specific. It has
5 nothing to do with the formulation. So, even if we
6 are off a bit in our permeability assessment, the
7 key measurements to me are the solubility that is
8 fairly rigorous, that is fairly reasonably defined,
9 the highest dose in a certain volume; dissolution
10 over a range of pH's, which I think is excellent;
11 and very rapid dissolution for Class I drugs.
12 So, given that scenario, I feel
13 comfortable, I think, with the Class I waiver.
14 Going beyond that I feel much less comfortable.
15 So, I think there is a lot of rationale here. If
16 you don't like what they are presenting, how are
17 they going to fix it is really the 1.2 question.
18 What additional data and what types of experiments,
19 what does the committee need to see next time in
20 order to say, well, they are right?
21 DR. LEE: Yes, Larry?
22 DR. LESKO: I want to get back to the
23 discussion of the permeability issue because it is
24 one that is already established in our guidance.
25 In other words, we can now, today, allow a sponsor
115
1 to identify a drug as a Class I drug based on
2 solubility and permeability in a way that we have
3 indicated in the BCS guidance which came out in
4 2000. So, I think we have established some
5 standards already on how to define permeability,
6 and we can probably better not go back and debate
7 that today but the question is, given that
8 standard, can we then extrapolate it to the fed
9 state?
10 Now, behind that standard, when we put the
11 2000 guidance out on the BCS for fasting studies
12 there was a fairly extensive database of 30 drugs
13 in which we actually measured permeability, extent
14 of absorption, and then correlated the two. That
15 then was built into the guidance in that a company
16 would standardize their CACO 2's using internal
17 controls that represent those drugs in that
18 database. So, there was a continual linkage of
19 human data to CACO 2 and to the other
20 circumstantial evidence such as extent of
21 absorption that gave reliability to characterizing
22 something as permeability.
23 I am not sure how we can do much better
24 with permeability, other than do human studies all
25 the time. But we did get to the point, and we did
116
1 present to the committee here, the ACPS, the
2 fasting BCS guidance and the standards we were
3 going to use for permeability, and that has been in
4 place now for a year and a half. So, I just want
5 to remind people that we are not crossing new
6 ground with this permeability definition.
7 DR. LEE: Art?
8 DR. KIBBE: Just a couple of things, and I
9 love being a devil's advocate so I will probably
10 raise some issues. But to start with, when drugs
11 are marketed, in the labeling they usually have
12 indications as to whether to take them with food or
13 without food. If you have a drug on the market
14 that is clearly indicated to take without food,
15 then the question in my mind is why do we care
16 about a food study if patients are told not to take
17 it with food anyhow? If they follow the
18 instructions, and if their physician and clinician
19 get them to do it correctly, they are not going to
20 even introduce that variable. So, if you have a
21 Class I drug whose labeling from the innovator says
22 take it without food, or take it on an empty
23 stomach, it is almost a moot question to try to
24 look for the other.
25 The second, what we are saying in effect
117
1 by waiving food studies for Class I drugs is that
2 we cannot imagine a formulator formulating
3 something where a formulation would interact with
4 food differently than any other formulation, and I
5 am not prepared to say that. So, I don't know how
6 I respond to that situation because the
7 classification is all about the active ingredient,
8 and the interaction that we care about when we do a
9 bioequivalence study is not about the active
10 ingredient; it is about the formulation. So, at
11 that point I am saying, well, as long as you use
12 spray dry lactose for your direct compressible I
13 don't care if you do a fed study because lactose
14 dissolves so fast that it is out of the way and
15 leaves the drug behind. But if you use a directly
16 compressible product made out of the chick bean
17 grown in Upper Uganda I don't like it. mean, that
18 whole road is kind of difficult for me.
19 DR. HUSSAIN: Just to add to that, that is
20 the reason why a waiver is limited to immediate
21 release dosage forms, not even suggesting it is for
22 modified release. In fact, Ameeta kept mentioning
23 theophylline and the dose dumping situations that
24 we have with theophylline were for modified release
25 only. So, we are talking about immediate release
118
1 dosage forms that dissolve rapidly under different
2 pH conditions. The focus is on formulation
3 similarity from that respective. So, you are
4 talking about pharmaceutical equivalence. You are
5 looking at an excipient database of an acceptable
6 set of excipients and then you are looking at
7 similarity and dissolution as a function of the pH.
8 DR. KIBBE: So, what you are saying is
9 that I could use starch 1500 as a directly
10 compressible excipient, and the agency says it is
11 exactly the same as lactose.
12 DR. HUSSAIN: No, we are not saying that
13 the excipients are the same. The excipients could
14 be different but as long as the product dissolves
15 in a comparatively similar profile under different
16 pH conditions that should be okay. In fact, I will
17 turn that around. I say, all right, now you have a
18 direct compression tablet, say, based on dicalcium
19 phosphate. All right? Then you have a formulation
20 based on starch lactose. So, if you look at it,
21 the dose would still be pharmaceutically equal and
22 they have very different sort of pH behavior.
23 Dicalcium phosphate tends to be fairly highly
24 soluble at pH 1 but the solubility goes down at pH
25 2, and so forth. So, a product containing that
119
1 will not have a similar dissolution profile as that
2 of starch or lactose based formulation. So,
3 actually dissolution is far more discriminating
4 under those conditions for a formulation difference
5 than in vivo. In fact, my concern is that I think
6 the dissolution that we are recommending is far
7 more conservative for the fed state.
8 DR. KIBBE: What you are saying is that
9 the generic which has that is going to have to
10 prove that there is no food effect because a
11 dissolution study isn't going to be similar.
12 DR. HUSSAIN: Unfortunately, yes.
13 DR. WILDING: I would like to echo Ajaz'
14 comments. I mean, that is the key here in the
15 sense that if those two formulations are rapidly
16 dissolving and meet the current requirements under
17 the BCS guidance, then given the fact that they are
18 going to be extended in their residence time in the
19 stomach and they have longer to dissolve in vivo,
20 it is a very conservative approach that we are
21 taking in this particular regard. I think as was
22 indicated by one of your colleagues, if we go
23 outside Class I it is a whole new ball park. In
24 the context of Class I, I think given we have an
25 acceptance of in vitro bioequivalence for Class I
120
1 compounds taking it into the fed domain is actually
2 not a big leap of faith.
3 DR. MEYER: Could I ask just one question?
4 In all the comments that you received, did anyone
5 cite an example that said, well--I like Ajaz'
6 approach of if there are two formulations and I
7 have all the wealth at my command I can make
8 whatever formulations I want, can I make two that
9 will dissolve in 15 minutes; will have similar
10 dissolution profiles but will have a pronounced
11 different food effect? Did anyone comment with an
12 example?
13 DR. PAREKH: No.
14 DR. MEYER: So, we are dealing with a fear
15 of the hypothetical or a fear of the unknown, and
16 the only way to prove the unknown is to do
17 everything which is going to be very expensive.
18 DR. LESKO: But related to that, there is
19 prior information that we can go back to. When we
20 did the original research with the BCS we did make
21 formulations designed specifically to be far apart
22 in their dissolution profile, huge differences in
23 dissolution, probably more so than you would expect
24 to see even with food and fasting. Those
25 dissolution differences for the Class I drugs did
121
1 not translate into bioinequivalence in in vivo
2 studies. They were very close to being
3 superimposable in essence.
4 So, we know that. I mean, that is prior
5 information. We have that document not only for
6 the model drugs, in this case propranolol and
7 metoprolol, but some other drugs as well. I think
8 that is useful information as background to have
9 with regard to differences in dissolution for Class
10 I drugs and what it means in vivo for
11 bioequivalence.
12 I also want to comment on Dr. Kibbe's
13 comment, and maybe Dale can confirm it but I
14 believe if the label says "take on an empty
15 stomach" there is no food effect for an ANDA
16 because, you are right, patients aren't going to
17 take it that way. Is that correct, Dale?
18 DR. CONNER: Yes. I think that is
19 supported by the language in this guidance.
20 However, if you read a lot of labeling, you know,
21 you expect these definitive statements which really
22 aren't there. I mean, a lot of times that type of
23 statement which you mentioned will say, we
24 recommend--you know, I am not literally
25 translating, we recommend that you kind of take
122
1 this with food, leaving the option open to the
2 physician or the patient to say, well, you know, I
3 don't really want to take it with food, or
4 sometimes I want to take it with food and sometimes
5 not. As long as you leave discretion open to the
6 clinician or to the patient you don't have a
7 definite "must take with food." So, I would say
8 that if the labeling is very strong, the
9 instruction saying "do not take this with food," or
10 "take only on an empty stomach," then I agree with
11 you, that should kick into place. But if it is
12 very wishy-washy, giving discretion to the
13 clinician or the patient I would say we have no
14 guarantee that they are not going to instruct the
15 patient, you know, if it will upset your stomach
16 take it with food, or don't.
17 DR. LEE: I think we do need to move on.
18 DR. HUSSAIN: One point that has not been
19 made and I just want to make is that in terms of
20 bioequivalence studies, the fasting studies are far
21 more discriminating than the fed studies. That has
22 been our position. So, if we waive a fasting study
23 it is logical that we would waive a fed study. So,
24 we are actually caught in a logical bind here
25 because when we put the BCS guidance together we
123
1 went for the most difficult part and left the
2 easier part, in my opinion, behind. So, there is
3 an inconsistency in our approach with BCS.
4 DR. LEE: Yes, I think this is the
5 conclusion I want to draw. I am glad that you said
6 it, and I think on that basis we should move on.
7 Sometimes you don't have data in the literature
8 because it can never be published.
9 The question then is what other additional
10 evidence will you need to make yourself feel
11 better? I think that has to be on a case by case
12 basis. It depends on the mechanism, complexation
13 and all that kind of stuff. Isn't that true?
14 DR. DOULL: Wasn't that Marv's suggestion?
15 The question of what additional information would
16 you need, the question is what do you really need
17 to know versus what would be nice to know. The
18 need to know would be additional Class I drugs.
19 You know, we really only have the two just to prove
20 this hypothesis. So, the question is how much more
21 information do you really need in order to be
22 comfortable with accepting that all Class I drugs
23 should not meet that food criteria?
24 DR. LEE: More sponsor studies.
25 DR. DOULL: More drugs, information on
124
1 more drugs.
2 DR. HUSSAIN: I am not sure. Let me sort
3 of summarize. The question is are we willing to
4 agree or make a recommendation that with the
5 guidance, as it is in the draft form right now, we
6 can move ahead and make the recommendation that the
7 waiver for food effect bioequivalence studies for
8 Class I rapidly dissolving drugs is okay. That is
9 the question.
10 DR. KARIM: Just to comment, who puts the
11 rubber stamp that this is a Class I drug?
12 DR. HUSSAIN: It is a review decision.
13 So, FDA.
14 DR. MEYER: Do we need to come to a
15 consensus?
16 DR. LEE: Well, I don't think we need to
17 come to a consensus. I think what is important is
18 for the agency to hear what our individual
19 collective thoughts are. Some issues may not ever
20 come to consensus. It has taken them about seven
21 years to--
22 DR. LESKO: That was the debate about
23 food. But to answer Dr. Karim's question, the
24 specific review division that is looking at the
25 application makes that decision, but a lot of those
125
1 decisions are discussed within the BCS technical
2 committee as well. So, it is really a collective,
3 joint decision between the Office of Generic Drugs
4 and the Office of Clinical Pharmacology.
5 DR. MEYER: In case my individual opinion
6 then wasn't heard, I am in favor of the proposal.
7 DR. LEE: So, what about question number
8 two, the confidence intervals?
9 DR. SHARGEL: I have a question on that,
10 if I may, Vince. My understanding from the agency,
11 as you mentioned, Dr. Conner, is the question of
12 clinical risk. In the past we have only done point
13 estimates. From what I understand, the desire for
14 confidence intervals is to have a more rigorous
15 test. If we use a more rigorous test, the data
16 showed that five studies out of 40 failed. Those
17 would not have been approved on the basis of the
18 new guidance if it were formalized.
19 DR. CONNER: Basically, presumably if you
20 knew the new criteria you would have done those
21 studies all properly powered. You know, I am
22 looking at them again not with a lot of in-depth
23 analysis of those particular studies and probably
24 three of them with somewhat more power would have
25 likely passed. Two of them would have had a great
126
1 deal of difficulty and would probably have failed
2 no matter what the power. But we can't definitely
3 say that. It just looks like to me that the ones
4 that had such extreme AUC values, I am not really
5 sure power would have helped those if the criteria
6 were changed.
7 As you know, when you change the criteria
8 people then adapt to the change and design their
9 studies accordingly with, hopefully, appropriate
10 power calculations. I actually found this even a
11 little surprising, that so many from a randomly
12 selected group like this would have passed using
13 the power that people use to power for point
14 estimates. I was pleasantly surprised. I expected
15 it to be a small difference but the results of the
16 group we picked surprised me. I would have
17 expected a few more to be on the edge but I was
18 pleasantly surprised when we actually looked at the
19 values.
20 DR. SHARGEL: May I just continue on this
21 a little bit because I am just curious in terms of
22 if there is no risk, clinical risk, what the basis
23 is for a more rigorous test. What we are doing is
24 we are using a meal that gives maximal
25 perturbation, as has been mentioned, and this would
127
1 give the largest variability to be observed on Cmax
2 and AUC. Generally in the labeling it would say
3 food effects of the drug but it never really
4 specifically says what kind of food, so that any
5 sort of diet--I prefer a bagel and cream cheese in
6 the morning; that is my preference--we would know
7 if there is a clinical effect of the food. If
8 there is a clinical effect, then you would say take
9 it without food, in the development of the product
10 if there is a big effect in the bioavailability
11 study. Or, if there is reason to take it with
12 food, we already require the 90 percent confidence
13 interval. So, my question here really is, is the
14 requirement here really necessary to have a more
15 rigorous test? And, what does it mean if we fail
16 in terms of safety risk?
17 DR. CONNER: Well, there are a great many
18 products that are labeled out there with simply a
19 descriptive statement of a food effect and, in some
20 cases, how much, the estimate of how much. It
21 doesn't mean that those are unusable products. It
22 doesn't mean that they are automatically restricted
23 from taking with food. A lot of it is in that area
24 of concern where I think the firm and the division
25 that is reviewing it at FDA feel that it is
128
1 important to let clinicians know about that. But,
2 based on the labeling, the physician can still use
3 that drug under those conditions as long as that
4 effect is known.
5 Granted, although there is some variance
6 in the type of meals that people do for NDAs, I
7 think in most modern NDAs we have a very similar
8 meal used. In fact, part of what this guidance
9 does is to bring the ANDA meal and the NDA meal to
10 be the same thing. So, basically what we are
11 saying is, no matter what other food studies are
12 done, the NDA will have a determination of the
13 effect on bioavailability with a virtually
14 identical meal. So, I mean, that will be part of
15 the NDA and part of the labeling.
16 I think from a statistical standpoint,
17 this is really just saying that, you know, we are
18 doing a test here. The meal that we have chosen,
19 as has been said before--you know, we can't really
20 test every conceivable meal. I don't think the
21 generic industry would want to go in for that kind
22 of thing, doing 30 different meals and 30 different
23 studies. So, if we only have one study to do, the
24 meal that we have chosen I think has the most
25 likelihood of being extreme and causing an effect.
129
1 So, if we don't see an effect under those
2 conditions, we are reasonably confident that lesser
3 meals or meals that are less stressful to the
4 dosage form are going to have any effect. I mean,
5 if you only have one chance you use the maximum
6 possibility to obtain an effect.
7 From a statistical standpoint, we would
8 like at the end of the day to say that these are
9 equivalent, that a generic is equivalent to the
10 reference listed drug under reasonable conditions
11 of us. You know, what we have been doing for many
12 years is good but it hasn't really been a true
13 equivalence statement, based on a true statement of
14 equivalence. And, what we are trying to do here is
15 perhaps improve that somewhat so that we can with
16 total confidence say that these two are equivalent
17 under reasonable conditions of use.
18 DR. LEE: Jurgen?
19 DR. VENITZ: I would like to follow-up on
20 that because I am still trying to understand what
21 it is that you are exactly proposing. You are
22 saying for any non-Class I drug, regardless of the
23 label of the reference drug, a generic has to show
24 fasting and fed bioequivalence?
25 DR. CONNER: No, that is not what we are
130
1 saying at all. We are saying that based on the
2 label of the reference listed drug, should that
3 label contain any statement about a food effect and
4 most, if not all, of the modern drugs that were
5 recently approved, within the last few years, will
6 have some type of statement about food effect. If
7 you look at, you know, twenty years ago, NDAs or
8 products that are still out, a lot of them didn't
9 do food studies or they didn't think it was worth
10 putting in the labeling, and so forth, a statement
11 about food in those old products may be totally
12 absent. Those would not trigger us to ask for a
13 food study. But any statement of a food effect in
14 the reference listed drug labeling will trigger a
15 question about whether it is bioequivalent in the
16 fed state as well. And, based on the type of
17 product or the type of drug substance we are
18 proposing dealing with it in different ways. You
19 know, if it is a Class I drug we will deal with,
20 you know, what the first part of the discussion
21 was. If it is not a Class I, then we will do a
22 food study, which we would do today. The only
23 question is how should we power that study, and how
24 should we analyze it, and what kind of conclusion
25 can we come up with based on that approach.
131
1 DR. VENITZ: So, as long as there is any
2 statement but it says there is no food effect, then
3 the official bioequivalence for the fed state--
4 DR. CONNER: Yes.
5 DR. VENITZ: --or if there is a food
6 effect.
7 DR. CONNER: I can tell you during the
8 five, seven, twelve years, whatever, we went
9 through a lot of discussion, a lot of proposals to
10 perhaps not make it such a label-based trigger for
11 having food considerations. We looked at a lot of
12 information on whether the original effect was drug
13 substance related, formulation related and so
14 forth, the assumption being, well, if we can
15 absolutely prove it is drug substance food effect
16 it is going to be the same for a generic versus
17 not. We went through a lot of this and had some
18 proposals to do that, but we finally figured out
19 that 99 percent of the time we don't know or are
20 unable to determine. So, we seldom, if ever, have
21 the data to answer it and we would end up doing
22 food studies virtually for everything anyway.
23 DR. VENITZ: But the consequence then of
24 having done a generic fed study and having failed
25 that study would be the generic would not be
132
1 approved or you would relabel?
2 DR. CONNER: No, the generic would not be
3 approved without a passing study. But that is true
4 today. I mean, with the criteria that we are
5 looking at today, and really the major change here
6 is not doing more studies but simply how we are
7 doing the studies that the generic sponsor would do
8 anyway.
9 DR. VENITZ: And how would that compare to
10 the NDA route?
11 DR. CONNER: I mean, what we are talking
12 about here is a bioequivalence study, which is one
13 of the few studies that is done to get a generic
14 product on the market. The NDA has literally
15 sometimes hundreds of studies of different types,
16 many of them bioavailability, a lot of them
17 clinical studies, studies on a lot of aspects of
18 the drug substance and drug product and how it
19 performs clinically. With a generic you
20 essentially have anywhere from one to perhaps three
21 or four, at the maximum, small in vivo studies to
22 be able to make the decision to approve that and
23 put it on the market.
24 DR. VENITZ: But in terms of assessing a
25 food effect you would use the same approach
133
1 basically?
2 DR. CONNER: We are not assessing a food
3 effect.
4 DR. VENITZ: No, I understand, but I am
5 saying if you are in an NDA situation so you are
6 not talking about generic bioequivalence and you
7 want to assess the food effect you would use the
8 same approach?
9 DR. CONNER: A very similar one.
10 DR. PAREKH: But the final decision is not
11 that of non-approval for NDAs. The final decision
12 is if you fall within this window you can say in
13 the label that there is no food effect.
14 DR. LEE: Let's come back to question
15 number two. Art?
16 DR. KIBBE: Just to go down another
17 wonderful side path, you decided to limit the
18 waiving of a food study to an immediate release
19 because you can get good dissolution data that
20 would overlap on immediate release, as well as the
21 fact that the Class I drug is highly soluble, and
22 what-have-you. But if I make a sustained release
23 product out of a Class I drug and someone else does
24 and we have clearly overlapping dissolution data,
25 and the criteria that we are looking at clearly is
134
1 the effect of food on dosage form, is there
2 evidence that there will be a problem with food
3 when you have delayed release products?
4 DR. HUSSAIN: I think to answer that
5 question, if I look at the example of theophylline
6 controlled release, modified release, the mechanism
7 for dose dumping there was different. Jerry Skelly
8 and others have actually done in vitro work that
9 actually showed that could be predicted. But our
10 confidence in in vitro is not at that level at this
11 point to go in that direction. So.
12 DR. KIBBE: If it is not an effect on the
13 drug moiety itself, the active ingredient, then it
14 is a matter of how confident you are in the
15 formulations being truly similar even if they give
16 the same dissolution profiles.
17 DR. HUSSAIN: The question is can you rely
18 on in vitro dissolution to understand the complex
19 mechanisms. Our answer is no, not at this time.
20 DR. LEE: Marv?
21 DR. MEYER: I tried to jot down the
22 reasons why not to use confidence limits. One, no
23 one takes drugs with a meal of any type. Well,
24 that is obviously not true and since we don't know
25 what type let's use the worst condition, confidence
135
1 limits are not a valid measurement of
2 bioequivalence. I think if they are good enough
3 for fasted, they are good enough for fed. Highly
4 variable drugs will pose a problem, and if they
5 somehow scrape by fasted they may not scrape by
6 fed. Well, that is an economic issue and that is a
7 statistical issue and it may be that we need to
8 change the stats for both fed and fasted to somehow
9 capture a point estimate and the variability of the
10 reference relative to the test, or vice versa but
11 that is a side issue. Too many failures. Well, we
12 have shown here that about five out of 40 would
13 fail marginally. With a proper designed study they
14 wouldn't. There would be like two or three out of
15 40. It would cost too much money; too many
16 subjects. We would have to again change our
17 statistics. I think FDA can't worry about public
18 health in the context of a $50,000 or $10,000
19 bioequivalence studies that some sponsor may have
20 to conduct. Numbers of subjects, we are still only
21 talking 30, 40 subjects. So, I think the reasons
22 why not to have confidence limits aren't
23 substantiated, and I have always felt that if
24 fasted need confidence limits, then fed need
25 confidence limits.
136
1 DR. LEE: Other points of opinion?
2 DR. MOYE: I guess I should say on the
3 record that at the conclusion of this session I
4 will turn over a synopsis of an alternative
5 analysis that would avoid the indirect approach of
6 confidence intervals, and would allow one to now
7 include this measure of variability that has been
8 excluded from the analyses.
9 DR. LEE: So you will have this synopsis
10 as food for thought.
11 DR. MOYE: As an admissible alternative.
12 DR. PAREKH: This is just for the record,
13 Dr. Meyer, you asked a question earlier about the
14 point estimates. For the two products that failed
15 on AUC the point estimates were 1.22 and 1.20. For
16 the three that didn't make it on Cmax, it was 0.86,
17 0.87 and 0.88.
18 DR. LEE: Are you satisfied?
19 DR. MEYER: Yes.
20 DR. LEE: Are there any other ideas or
21 suggestions, opinions? If not, thank you very
22 much. That concludes the agenda item on food
23 effect of BE studies. Now we are into the public
24 hearing. We have three submissions. The first two
25 cannot make it here, and we do have the last person
137
1 here, Russ Rackley. For the record, I have asked
2 Kathy to read the first two, and you all have that
3 in your notes.
4 Open Public Hearing
5 MS. REEDY: Yes, the right side of your
6 red folder has your agenda, your questions and the
7 open public hearing submissions in writing. On the
8 left side are the slides that were submitted in
9 advance. For the slides that were not submitted in
10 advance, they may show up at the time of their
11 presentation.
12 But for the open public hearing, the first
13 submission is from Brian Kearney, senior scientist,
14 clinical pharmacology, Gilead Sciences.
15 Guidance for industry food effect
16 bioavailability and fed bioequivalence studies,
17 commentary on the following issues is not currently
18 included in the draft guidance and FDA
19 Pharmaceutical Advisory Committee perspectives
20 would be much appreciate. One, please comment on
21 the acceptability/utility of parallel study designs
22 and/or secondary statistical analyses of PK data,
23 collected across studies, to evaluate food effects.
24 For example, could pharmacokinetic data derived
25 from fed studies in later stage PK studies b
138
1 compared to fasted, reference data from a previous,
2 formal crossover food effect study?
3 Two, while single dose studies are
4 preferred as they are the most sensitive to food
5 bioavailability effects, please comment on the role
6 and acceptability of steady state comparisons for
7 compounds with a short elimination half-life and/or
8 with predictable, reproducible PK profiles. Those
9 are Brian's comments.
10 The next is David Fox, writing to present
11 the views of Abbott Laboratories on a matter
12 scheduled for discussion at the upcoming meeting of
13 the Food and Drug Administration's Advisory
14 Committee for Pharmacologic Science on May 7th and
15 8th, 2002.
16 Specifically, we wish to comment on the
17 draft guidance document titled, "Food Effect
18 Bioavailability and Fed Bioequivalence Studies:
19 Study Design, Data Analysis and Labeling." We ask
20 that the committee carefully consider our written
21 submission in the course of its deliberations.
22 The food effect guidance recognizes that
23 foods and beverages often have a clinically
24 significant effect on the bioavailability of an
25 active drug ingredient or on the bioequivalence of
139
1 two different formulations of the same active
2 ingredient. Food effect guidance at 2. A growing
3 number of drug products now bear labeling that
4 describes a significant food effect, a trend which
5 Abbott believes is good for patients. Food effect
6 labeling contributes to consistent and more
7 accurate dosing and can help patients adopt a
8 routine set of conditions under which they take
9 their medicines.
10 Second, the food effect guidance
11 recognizes the need for bioequivalence studies
12 under fed conditions, particularly where the
13 reference of the pioneer product bears food effect
14 labeling. Food effect guidance at 4.
15 Food effects may be formulation specific,
16 and two different versions of the same drug may
17 react differently in the presence of food. In
18 fact, two products may react differently depending
19 on the quantity or type of food used. And, he uses
20 a reference discussing an example of two products,
21 each with the same active ingredient and dosage
22 form that had clinically significant
23 bioavailability differences depending on whether
24 the drugs were taken with chocolate milk, apple
25 juice or orange juice. For these reasons, the
140
1 guidance endorses the need for well-controlled and
2 well-designed fed bioequivalence studies where the
3 reference product has a noted food effect. Food
4 effect guidance at 3, noting that the mechanism by
5 which food may affect bioavailability is often
6 unknown and cannot be determined by physical
7 inspection of in vitro study.
8 Abbott agrees and compliments the agency
9 for recognizing these points. Abbott's
10 concern,however, is that the agency has not gone
11 far enough to address the variable bioavailability
12 seen by many drugs under different meal conditions,
13 nor has the agency taken steps to ensure that
14 bioequivalence studies performed by applicants
15 under abbreviated new drug applications follow the
16 same meal conditions used in the study of the
17 reference drug product. Instead, the agency
18 recommends only the use of a high-fat, high-calorie
19 test meal to provide the greatest effects on
20 gastrointestinal physiology so that systemic drug
21 availability is maximally affected, food effect
22 guidance at 6.
23 For a product with a known sensitivity to
24 food, the agency's approach in many instances is
25 likely to mask or obliterate important formulation
141
1 differences. The better approach, we suggest, is
2 to require fed bioequivalence studies under the
3 meal conditions suggested in the labeling or, if
4 the labeling is not specific, under the meal
5 conditions likely to be followed by patients who
6 use the drug. Alternatively, the sponsor of a
7 bioequivalence study should follow the meal
8 conditions that were used to support the efficacy
9 of the reference drug product. Patients on a
10 low-fat diet who are instructed to take their
11 medications with meals should be assured that a
12 generic substitute will behave the same under
13 low-fat conditions as the pioneer.
14 Finally, while the food effect guidance
15 allows for the use of other test meals, food effect
16 guidance at 7, the guidance puts the decision
17 within the discretion of the sponsor. It is the
18 generic drug sponsor's choice, for example, to
19 conduct a bioequivalence study with a test meal
20 other than the maximum 50 percent fat meal
21 described introduction he guidance. Abbott
22 disagrees with this approach. The guidance must
23 recommend the use of a test meal that closely
24 reflects the labeled conditions of use or the
25 conditions under which the reference drug was
142
1 studied. In fact, by allowing the sponsor to
2 select the test meal, FDA invites the real risk
3 that the sponsor may use food selection to drive or
4 optimize the showing of bioequivalence.
5 In short, the agency's thinking on the
6 need for bioequivalence studies is pointed in the
7 right direction but, at this stage, is too general.
8 For products that are food-sensitive, it may be
9 impossible to know in advance whether the product
10 will behave in a linear or predictable way under
11 different meal conditions. Simply comparing two
12 products under fasting and high-fat conditions may
13 be insufficient, especially when the drug is
14 labeled for us under low-fat or other dietary
15 conditions. Food effects are not yes/no
16 propositions. Far too little is known about food
17 effects for FDA to assume the use of one type of
18 meal for all drug products.
19 For these reasons, we respectfully request
20 that the committee consider three related points.
21 The first, the need for fed bioequivalence studies
22 under conditions other than the maximum 50 percent
23 fat meal described in the food effect guidance.
24 Secondly, the need for fed bioequivalence studies
25 under the conditions of use recommended or
143
1 described in the labeling; and, thirdly, the need
2 for fed bioequivalence studies that follow the same
3 study design used in the clinical testing of the
4 pioneer product. We greatly appreciate your
5 attention to this issue.
6 DR. LEE: Thank you very much, Kathy, for
7 reading it, and I don't think we can ask any
8 questions because the presenter is not here. So,
9 next I would like to invite Dr. Rackley, from Mylan
10 Laboratories to give a ten-minute presentation. He
11 is going to be speaking on behalf of the Generic
12 Pharmaceutical Association.
13 DR. RACKLEY: Thank you. It is an honor
14 to be here to speak before you today on behalf of
15 the Generic Pharmaceutical Association.
16 [Slide]
17 ANDAs have been approved and marketed
18 since around 1985 with no documented safety issues.
19 The demonstrated safety and wide acceptance of
20 these products by the general public are indicative
21 of the robustness and adequacy of the current
22 approval process. We propose that the current
23 system for the evaluation of bioequivalent drug
24 products be maintained.
25 [Slide]
144
1 For current fasting bioequivalency studies
2 this represents a standard bioavailability
3 comparison of test and reference drug products.
4 Ninety percent confidence intervals are well
5 accepted as demonstration of bioequivalence.
6 [Slide]
7 For current fed bioequivalency studies,
8 the OGD breakfast represents an extreme food
9 condition. The standard breakfast allows for
10 effect of food on GI motility, the effect of food
11 on the bioavailability of the drug in vivo, the
12 effect of food on the formulation of the drug.
13 [Slide]
14 Point estimate criteria is well-accepted
15 for the fed studies as further confirmation of
16 bioequivalence. The requirement for 90 percent
17 confidence intervals for a food effect study does
18 not improve the safety of the generic drug product.
19 [Slide]
20 Regarding post meal administration,
21 logistically it is difficult for everyone to
22 consume a standardized breakfast in exactly 30
23 minutes and then immediately take the dosage form.
24 Study subjects should be allowed to consume the
25 standard meal within 30 minutes and the dosage form
145
1 will be administered 30 minutes after the start of
2 the meal.
3 [Slide]
4 Pharmacokinetic parameters to assess
5 bioequivalence, AUC and Cmax should remain the
6 primary parameters upon which to assess similarity
7 of rate and extent of absorption. Expectation of
8 Tmax to be comparable is vague and tends to be
9 subjective. Tmax should be provided for
10 information purposes only, and not held to a
11 statistical criteria.
12 [Slide]
13 Regarding sprinkle studies and special
14 foods, if a dosage form is shown to be
15 bioequivalent after a stringent fasting study and
16 similarity is confirmed by a fed study, there is no
17 reason to believe that it will not be bioequivalent
18 when taken with a small amount of food.
19 We acknowledge there are no examples where
20 vehicle has had a significant effect on
21 bioequivalency, and these should be well documented
22 in labeling under dosage and administration.
23 [Slide]
24 However, requirements to demonstrate
25 bioequivalence, when taken with special foods or
146
1 vehicles, will lead to anecdotal stories and open a
2 flood gate for an infinite number of study
3 requirements for generic approval. There is no
4 doubt that this will be taken advantage of to delay
5 generic approvals.
6 [Slide]
7 Standard breakfast, the FDA standard
8 breakfast is adequate for demonstration of food
9 effect on bioavailability. The use of alternate or
10 unusual food studies may be used as a tactic to
11 further delay generic approvals.
12 [Slide]
13 In conclusion, the current approach for
14 performing food effect bioavailability studies
15 using a standardized meal is adequate. Unless the
16 current methods and criteria represent a danger to
17 public safety, we, as responsible scientists and
18 citizens, should challenge unreasonable regulations
19 and requirements. The existing fasting BE and fed
20 BA studies are time-tested methods. Changes to
21 these methods increase the burden to the industry,
22 delays approvals and does not seem to be justified.
23 DR. LEE: Thank you. Are there questions
24 for Dr. Rackley?
25 DR. SHARGEL: Dr. Meyer mentioned about
147
1 variability drugs, where you have a highly variable
2 drug it would seem to me that food effect and
3 trying to match 90 percent confidence intervals
4 would be very tough. How do you feel about that,
5 or widening the intervals past the 90 percent
6 confidence intervals, from 0.8 to 1.25?
7 DR. RACKLEY: Clearly, a highly variable
8 drug product would have had to be powered
9 adequately, probably with large numbers of
10 subjects, in a fasting study. If the same
11 inter-subject CV were to be held or shown for the
12 same drug products in a fed study you would likely
13 be doing, again, huge size studies. So, where
14 there is 10 percent of studies that might not pass
15 confidence intervals, you might also factor in that
16 some of these studies might have to be done with
17 perhaps even over 100 subjects to do a fed study,
18 whereas today they demonstrate or reaffirm what a
19 rigorous, stringent fasting bio study has
20 demonstrated.
21 DR. LEE: Larry?
22 DR. LESKO: Are you aware of any evidence
23 that food can reduce the variability in a highly
24 variable drug case where a drug is highly variable
25 under fasting conditions, but when you give it with
148
1 fed the variability actually is reduced? I mean,
2 as a general assumption the variability is going to
3 go up with food, and I would say we haven't seen
4 that in the analysis of our own data. When Ameeta
5 showed the ANDA data where 35 out of 40
6 applications met confidence intervals, it suggested
7 that the variability did not change compared to the
8 fasting studies, or else not that high number would
9 have passed. So, I am not sure of the assumption
10 that food increases variability, unless we have
11 some evidence to suggest that is one that is
12 necessarily valid. Perhaps in the FDA survey that
13 was done with 40 drug products, or if they want to
14 add more to it, they would provide those point
15 estimates and what the estimates for inter-subject
16 variability were under fasted and fed conditions.
17 That is just a thought. I mean, the data is out
18 there. There is plenty of it that comes in every
19 year.
20 DR. RACKLEY: I guess one question I was
21 going to ask about our own database is what was the
22 size of the fasting studies for the corresponding
23 applications for which you showed fed data. In
24 other words, was the fasting study larger or the
25 same size?
149
1 DR. CONNER: I don't know the exact
2 numbers that correspond to these 40 but generally
3 what we usually see is around a 24-subject study
4 for most products. You know, we might see up to
5 36. The highly variable drugs are, you know,
6 special. Fortunately, in the scheme of things they
7 are a relatively small problem but they are a very
8 special problem which we have to deal with for
9 fasting studies as well. I mean, for most drugs
10 that are very highly variable we are talking about
11 60 or 80 subjects, but there is a very small subset
12 where it is over 100, if not more. So, we are
13 currently thinking or working on ways to do
14 different types of analysis, say, with perhaps the
15 ideas on scaling that came out of the individual
16 bioequivalence efforts, but those things are not
17 ready yet. We still have a lot of work to do on
18 working that out, but we hope to eventually have a
19 way of dealing specifically with highly variable
20 drugs whether we are doing a fasting or a fed study
21 that will, you know, come in with a valid approach
22 at a reasonable sample size.
23 DR. LEE: Very well, thank you. Let me
24 summarize this morning. I think this morning we
25 have witnessed the progressive approach to
150
1 reexamine the guidance as science evolves, as drugs
2 change, and so forth. I think we can come to some
3 cautious conclusions, and I think we are kind of
4 cautious because we, as scientists, always think
5 about exceptions. Also, as a member of the
6 committee I would like to suggest thinking about
7 meals, new composition, as a possibility to see how
8 far that thinking would go. As you can hear from
9 our discussion, what is the intent of the guidance
10 to look at the food effect.
11 On that note, I think we are ahead of
12 schedule but in fear of a long discussion this
13 afternoon--yes, Art?
14 DR. KIBBE: One quick question. Am I
15 right as I read the guidance that you have
16 eliminated now therapeutic index drugs a priori
17 from consideration, or did you just eliminate the
18 ones that don't meet the criteria for high
19 solubility? The therapeutic index is an indication
20 of their interaction with the receptors and not
21 necessarily an indication of the nature of the
22 chemical itself or the dosage form.
23 DR. LESKO: When you say eliminate,
24 eliminate from what?
25 DR. KIBBE: I thought there was a
151
1 statement in there.
2 DR. LESKO: The waiver of NTIs in the food
3 guidance is similar to what we did in the BCS
4 fasting guidance, and I believe they are excluded
5 from bio waivers in both guidances.
6 DR. KIBBE: But my point is that that
7 isn't necessarily necessary. If the therapeutic
8 index is a function of the way the drug behaves in
9 the body and our guidances are a way of helping us
10 determine equivalence between products, then I am
11 having a hard time getting my hand around
12 eliminating a narrow therapeutic index drug from a
13 waiver just because when you give it, no matter who
14 makes it, no matter how it is administered, it is
15 the way that it works in the body that is at issue
16 and not the dosage form.
17 DR. LESKO: I think that is a good
18 question and it is probably an open question. We
19 have discussed it here in this committee and it was
20 related to the level of certainty about the science
21 that you wanted to be careful about expanding this
22 to each and every drug, even those that have narrow
23 therapeutic index. On a scientific basis,
24 mechanistically speaking, you are right in arguing
25 that they should not necessarily be excluded
152
1 because the therapeutic index is related to the
2 pharmacology and not the pharmaceutics of the
3 dosage form. You know, it is something if the
4 committee feels we should revisit, I think we can
5 do that.
6 DR. VENITZ: But I would argue all we are
7 doing is risk management. The stakes are higher.
8 That is what it really comes down to.
9 DR. MEYER: It is okay to continue a
10 little bit with the proposed guidance, or do you
11 want to break?
12 DR. LEE: What would you like to bring up?
13 DR. MEYER: Well, I have a couple of
14 questions. Dr. Rackley raised the issue of
15 sprinkles and special vehicles.
16 DR. LEE: Sure.
17 DR. MEYER: That wasn't one of the
18 questions we should deal with. Can we comment now?
19 DR. LEE: Go ahead.
20 DR. MEYER: I guess my one question about
21 the sprinkles is it seems to make sense if it
22 passes a high-fat meal, why also make people put it
23 on apple sauce and swallow the sprinkles? Is there
24 evidence to suggest that that is a problem?
25 DR. CONNER: I don't view that they are
153
1 studying two totally different things. With the
2 sprinkle it is I think most of the time it pertains
3 to beaded, modified release dosage forms, which
4 depend on their mechanism of release with a coating
5 or some other mechanism that, on direct and perhaps
6 slightly prolonged contact with the food of given
7 properties--pH, fat content and so forth--we are
8 talking about not mixed up in the milieu of the
9 stomach but in actual direct contact, dumped in and
10 mixed into this food, that there is at least a
11 possibility that that coating could be broken down
12 where you wouldn't necessarily see an effect when
13 it is mixed up with stomach contents, and so forth.
14 And, for these type of products often it
15 is stated in the labeling that they are labeled to
16 be given this way. If you have ever worked at
17 hospitals or had small children that had to take
18 this type of dosage form, you know that frequently
19 they are dumped into food and left around perhaps
20 for half an hour, an hour on normal use. So, the
21 worry is that at some point that mechanism that we
22 depend on is disrupted. Now, in a bioequivalence
23 sense what we worry about is not that both products
24 are going to be disrupted in the same way; we are
25 worried that we could have a differential effect.
154
1 If I put the innovator product in apple sauce, it
2 is perfectly stable; no breakdown; you take it
3 after five or ten minutes, and then I put the
4 generic in and it immediately dissolves, you know,
5 I have a real problem with that because those two
6 products are not going to be bioequivalent under
7 those conditions. A lot of people say, well, it is
8 the same thing as the food study we have always
9 done. I think it is a very direct challenge of the
10 coating or mechanism of modified release by direct
11 and very concentrated contact with the food. That
12 is the rationale for doing it.
13 DR. MEYER: It almost seems like that
14 could be studied in vitro with apple sauce mix in a
15 basket, or something.
16 DR. CONNER: I can imagine pouring the
17 apple sauce after the dissolution. You know,
18 theoretically I am not saying that you couldn't
19 develop some kind of in vitro method to get at
20 this. I don't really think that we know enough
21 about it to know what the properties are or how we
22 should approach that. If people have some research
23 or some ideas in mind, we would love to see the
24 data on that. But right now the most direct way of
25 studying this is with an in vivo study. Perhaps
155
1 later on we can develop a system to do it in vitro
2 in a valid way. We are just uncertain of how to
3 approach that with our current knowledge.
4 DR. SHEK: There is at least on case for a
5 liquid where it makes a difference with what type
6 of juice you are using.
7 DR. HUSSAIN: In that case I think it is
8 far more complex. I would rather not discuss that
9 particular case here.
10 DR. LESKO: It is worth mentioning one
11 thing, the problem you were going to bring up is
12 with a fairly old product, I believe. But nowadays
13 any NDA that comes in that wants to make a claim
14 about administering the drug with food, either
15 sprinkles or orange juice, or whatever it is, is
16 going to have to have some evidence to make that
17 claim in the label. Whereas, in the past I don't
18 think we appreciated all the various mechanisms of
19 interactions and we sometimes let some of that go
20 with the vehicles. But I think that has changed
21 today and the label is pretty much going to reflect
22 the evidence that company submits.
23 DR. LEE: Marv, a second point?
24 DR. MEYER: Yes, the one about special
25 vehicles, if the label of the reference listed drug
156
1 says apple juice, orange juice, grapefruit jelly,
2 what-have-you does not affect the absorption, as I
3 read this guidance the generic has to do all of the
4 above to show that they do not affect the generic
5 formulation. Is that a reasonable thing for us to
6 be allowing to happen?
7 DR. LESKO: My sense would be it would
8 have to be case by case. You would have to look at
9 the reference listed product and see what data is
10 available that supported that claim in the label
11 and with there is any mechanistic reason that a
12 study needs to be done. I wouldn't generalize on
13 that issue.
14 DR. MEYER: But the guidance does
15 generalize.
16 DR. LESKO: I think the guidance makes
17 some recommendations rather than exclude it, and
18 you would have to interpret that I think on a case
19 by case basis.
20 DR. KIBBE: Just following up on that,
21 would the generic company then who sees that type
22 of labeling on a product they wish to duplicate do
23 well to talk to you about whether they need to do
24 that study or not before they even go down that
25 road?
157
1 DR. LESKO: I would. I think Dale is
2 going to comment, but I think it might be something
3 we can clarify and deal with because I think we
4 know what the intent is. It is a matter of getting
5 the right words around it.
6 DR. CONNER: It comes up with our recent
7 experience with certain products, which we don't
8 want to talk about today. Fortunately for us,
9 these products that are covered by that are very
10 few and far between. I think we are not dealing
11 with a huge number here. So, we wanted to really
12 leave ourselves the option of dealing with these
13 problems, not only option but the ability to deal
14 with these problems as we saw them. You know,
15 should we see a very complex dosage form or a
16 liquid dosage form or one that needs to be mixed
17 with a beverage, we will have the ability and the
18 sponsors will know that that is a potential problem
19 and they can put that into their thinking as far as
20 how they develop their dosage form, whether it be
21 the original innovator dosage form or a generic,
22 about how to approach that and what to ask us about
23 and what they would like to propose themselves. It
24 really just puts both the FDA and the industry on
25 notice that this is a potential issue and that they
158
1 need to work it out prior to being approved.
2 DR. HUSSAIN: Vince, just to sort of
3 clarify, I think if we discuss that example it
4 brings up the issue of a particular product, and so
5 forth. I think it would be a good question and I
6 think we will go back and consider it maybe at the
7 next meeting. We could actually make that a case
8 study for discussion because for that to happen, I
9 think the key sponsors would need to be present in
10 the room.
11 DR. LEE: Certainly, I think so. As
12 science evolves and we know more about something,
13 you know, what should we do about it? Yes, Leon?
14 DR. SHARGEL: Yes, I agree. You know, for
15 specialized diets the guidance sort of leaves open
16 possibilities of last minute labeling changes,
17 which certainly slows entry of generic products. I
18 think it needs to be clarified a little bit more
19 clearly when a food is required for specialized
20 issues, and I think the innovator who is making the
21 claim when there is an issue should actually show
22 data.
23 DR. LEE: Thank you very much for the
24 discussion. I think that we are going to move on
25 to the afternoon about the BCS and I don't know
159
1 what this discussion is going to lead to. It
2 hopefully won't lead us to come back to revisit the
3 food effect today but maybe in a future session.
4 Kathy does have some announcements to make.
5 MS. REEDY: For those who have contracted
6 for the convenience of having your sandwiches here,
7 in the building, they will be directly across the
8 hall. For those consultants, members and guests
9 who have not yet done so, you may do so by finding
10 Beverly O'Neal and handing her $10.00. For all
11 others, it is a lovely day and there are a number
12 of sandwich shops in the neighborhood.
13 DR. LEE: Thank you. We will come back at
14 1:15.
15 [Whereupon, at 12:05 p.m., the proceedings
16 were recessed for lunch, to reconvene at 1:20 p.m.]
160
1 A F T E R N O O N P R O C E E D I N G S
2 DR. LEE: Welcome back. We heard about
3 BCS all morning. So, this afternoon we will find
4 out what exactly BCS is, for those of you who don't
5 know about it. More importantly, we want to talk
6 about the next steps. These are not baby steps;
7 these might be giant steps. We have Lawrence Yu,
8 Acting Deputy Director of Science, OGD/OPS, to
9 introduce the topic.
10 Biopharmaceutics Classification System - Next Steps
11 Introduction and Overview
12 DR. YU: Good afternoon. Dr. Vincent Lee,
13 Chairman of the FDA Advisory Committee for
14 Pharmaceutical Science, members of the FDA Advisory
15 Committee for Pharmaceutical Science, my FDA
16 colleagues and distinguished guests, this afternoon
17 we will cover the biopharmaceutics classification
18 system - next steps.
19 [Slide]
20 We will have three presentations. Dr.
21 Gordon Amidon, chairman and professor of
22 pharmaceutics at the University of Michigan, will
23 give a talk entitled history and applications of
24 the biopharmaceutics classification system. Dr.
25 Jack Cook, from Pfizer, will give a second talk
161
1 entitled the industrial experience with the BCS. I
2 will give the third talk entitled regulatory
3 implementation and potential extension of the
4 biopharmaceutics classification system.
5 [Slide]
6 Following the three presentations there
7 will be two questions which have been slightly
8 modified. The first question is should the agency
9 consider revising the pH range of the solubility
10 class boundary to be consistent with the
11 dissolution pH range?
12 The second question is should the agency
13 consider expanding the application of the BCS based
14 biowaivers to rapidly dissolving and immediate
15 release products of the BCS Class III drugs,
16 namely, highly soluble and permeable drugs? With
17 that introduction, I will turn the podium to Prof.
18 Gordon Amidon.
19 Presentations
20 DR. AMIDON: Thank you, Lawrence. It is a
21 pleasure to be here, talking about and seeing the
22 evolution of the biopharmaceutics classification
23 system, something that I have worked on I think for
24 almost 15 years. At least if you count the very,
25 very beginnings for an FDA workshop on dissolution
162
1 and absorption, since 1988, I believe, so it has
2 been a long time and I will show some of that
3 history. Then, to see the application of BCS this
4 morning being used as a basis for providing waivers
5 for Class I drugs, waiver of food studies for Class
6 I drugs, I think of that as a superb extension of
7 and use of the BCS concept because how else could
8 you come to that conclusion without having a
9 mechanism for biowaivers? So, I think that is a
10 superb application and I was pleased to see that go
11 so well.
12 [Slide]
13 The process of BCS is based on looking at
14 the systemic availability versus the absorption
15 processes controlling appearance of drug into the
16 plasma, and transitioning from the systemic
17 availability view to the absorption view, and then
18 using that, in turn, to set standards for drugs.
19 Because if we can ensure absorption, we will also
20 ensure systemic availability. The advantage of
21 ensuring absorption is that now we can talk about
22 processes in the gastrointestinal tract and develop
23 scientific hypotheses to formulate and proceed.
24 That process led then to the guidance, the
25 so-called BCS guidance which says waiver of in vivo
163
1 bioavailability and bioequivalence trials. I think
2 that choice of terms I am fairly happy with because
3 it says waiver of in vivo bioavailability and
4 bioequivalence trials. We are not waiving
5 bioequivalence. No one has ever proposed that, and
6 I think bioequivalence, Cmax and AUC is the gold
7 standard and BCS doesn't change that. It provides
8 alternatives to ensuring in vivo bioequivalence.
9 Our goal is to ensure bioequivalence and to meet
10 that standard. In fact, I will argue that I think
11 it is clear that for BCS Class I drugs that
12 dissolve rapidly the in vitro standard is actually
13 a better standard. It is not as good; it is not a
14 substitute; it is actually better because the in
15 vivo test is not very accurate.
16 [Slide]
17 BCS is a scientific framework for
18 classifying drugs based on their aqueous solubility
19 and intestinal permeability. This is fairly
20 straightforward. I will say a little bit about the
21 science today and the extensions. I do want to
22 provide some overview of the process that was
23 involved in moving this guidance along.
24 When I became involved in bioequivalence
25 in the mid to late '80's, it was Cmax and AUC,
164
1 empirical; you do the test and reference and get
2 the result; do the statistics and you pass or fail;
3 and that was kind of the end of the story. When we
4 developed the concept of BCS we also needed a
5 database and scientific support to develop the
6 standard.
7 [Slide]
8 So we began some research with the support
9 of the FDA, at that time the Office of Generic
10 Drugs in 1990 at Michigan and Uppsala and at
11 Maryland. Over the period of the next five years
12 that led to substantial research. The first
13 application of BCS was incorporation actually into
14 one of the SUPAC guidances in 1995. We actually
15 formed a working group at the FDA. I think we made
16 our first presentation to the ACP panel around
17 1996. I can't read that well. In 1996 we made our
18 first presentation and proposal to this committee
19 regarding biowaivers and the BCS approach. It was
20 supported at that time and led to more research.
21 Also, at that time I took leave of absence and
22 spent four months at the FDA, working with Ajaz and
23 Larry.
24 I should say at the very beginning that
25 Larry Lesko was the initiator with me. He referred
165
1 to himself as the grandfather when he passed me
2 this morning after the BCS discussion. If he is
3 the grandfather, what does that make me, Larry? I
4 was trying to think that maybe we could be
5 grandparents but that doesn't work somehow. But we
6 worked on this over about a five- or six-year
7 period, building up the science and the draft
8 guidance.
9 The actual draft guidance was drafter in
10 1995 with Ajaz. So, Ajaz was instrumental. He
11 came in, in 1995 to replace Larry because Larry
12 moved up and took on other responsibilities and
13 Ajaz did a superb job writing the draft guidance.
14 I say that so that if there are any problems with
15 it, it is Ajaz' problem.
16 Many of the extensions, I would say we are
17 talking about today, were discussed at that time.
18 I can't say all of them because I can't remember
19 everything. But in the process of developing the
20 guidance we came up with what we thought were the
21 most conservative and sure-thing in terms of
22 biowaivers because if we were going to change the
23 paradigm of biopharmaceutics we wanted to do it
24 carefully so that it is accepted. We don't want to
25 make a mistake going out there with that first
166
1 application for biowaivers. So, we ended up with a
2 very conservative guidance.
3 [Slide]
4 The actual draft guidance was published in
5 February of 1999 and then the final guidance was
6 published in August of 2000. You can see the
7 number of workshops and scientific discussions we
8 have had--the U.S., Europe and Japan, as well as
9 Latin America, including a workshop at PAHO, the
10 Pan American Health Organization, because this
11 guidance is important in developing countries as
12 they develop or phase in bioequivalence standards
13 throughout the Americas. So, there is a great deal
14 of interest in this approach.
15 [Slide]
16 There was a lot of discussion and I think
17 I can say it is generally accepted. At least we
18 have been out talking about it enough so no one
19 stands up and argues with me anymore. This is kind
20 of the principle of bioequivalence as I think of
21 it, kind of like the central dogma in biology which
22 we now know is wrong because one gene produces more
23 than one protein. At any rate, this is the dogma,
24 similar plasma levels, similar pharmacodynamics;
25 similar in vivo dissolution, similar plasma levels.
167
1 That is similar in vivo dissolution. Then, in
2 vitro dissolution can match in vivo dissolution.
3 Oftentimes when we talk about dissolution, we use
4 that term too generically, like cancer. You know,
5 there are so many different versions of it.
6 Dissolution in what? So, what we want to do is
7 establish a BE or bioequivalence type dissolution
8 methodology which would be more complex and more
9 elaborate perhaps than the usual QC or quality
10 control dissolution methodology that would be used
11 when you make major changes in your formulation
12 that engender a bioequivalence question.
13 [Slide]
14 So, we have changed from systemic view to
15 the fraction absorbed view. Marvin, I think your
16 point was well taken this morning that
17 bioavailability is much easier than fraction
18 absorbed. It can be very hard and sometimes even
19 impossible if your drug is unstable in the
20 gastrointestinal tract and the metabolite or active
21 compound, like an ACE inhibitor, is not well
22 absorbed. So, it can be impossible almost to
23 determine what actually is the fraction absorbed.
24 But in the majority of cases you can determine it
25 by mass balance studies or IV and oral excretion
168
1 studies or bioavailability.
2 Now, the initial rationale for the BCS
3 waiver was the following: If a drug dissolves
4 rapidly like a solution and becomes essentially a
5 solution in the gastrointestinal tract,
6 particularly the stomach, a rapidly dissolving
7 drug, then the rate-determining step for absorption
8 is gastric emptying. It is not a formulation
9 difference; it is gastric emptying. So, on the
10 basis of that rationale, if gastric emptying is a
11 slow step for a high solubility, high permeability,
12 rapidly dissolving drug, plasma levels tell you no
13 information about formulation differences.
14 Consequently, an in vivo test is not the best test
15 for ensuring in vivo bioavailability. In this case
16 then a dissolution test would be more than an
17 adequate surrogate for an in vivo test. And, that
18 is where the waivers are currently allowed for a
19 high solubility, high permeability, rapidly
20 dissolving drug.
21 [Slide]
22 As you think about extensions of BCS, we
23 are going to propose several extensions. We had
24 one workshop on January 31, February 1 on
25 extensions. We have had one meeting at the FDA
169
1 with the internal working group on extensions, and
2 I would say there is a list of about six or eight
3 areas we are considering for extensions, of which
4 the two that we are proposing today represent what
5 we think are the next steps that we should take.
6 [Slide]
7 I will say a few things about other areas
8 of extensions and illustrate them. First is the
9 extension to Class III drugs, which are high
10 solubility but low permeability. Those are drugs
11 like atenolol which are less than about 50 percent
12 absorbed, or maybe 60 or 70 percent absorbed. So,
13 the remainder of the drug is in the intestine the
14 whole time. Fifty percent of the atenolol dose is
15 in the colon all the time, or just about that,
16 because the majority of the residence time is in
17 the colon. That means the colon permeability has
18 to be pretty small.
19 So, there is position-dependent
20 permeability along the gastrointestinal tract.
21 While we think if a drug like cimetidine or
22 ranitidine dissolves very rapidly in the stomach, a
23 waiver should be allowed for those drugs, but they
24 must dissolve in the stomach. So, we think
25 probably a tighter dissolution specification is
170
1 important for low permeability drugs because of the
2 position-dependent permeability, in most cases,
3 along the gastrointestinal
4 tract--position-dependent in the very least. We
5 know some drugs are absorbed in the duodenum
6 jejunum because we have plasma levels, and we know
7 that it is in the colon all the time and it is not
8 completely absorbed. So, there is clearly
9 position-dependent permeability, although evidence
10 for colon permeability is much harder to obtain.
11 It can be obtained but it is much harder.
12 A third area of discussion is low
13 solubility drugs or so-called Class II drugs that
14 dissolve rapidly in the gastrointestinal tract.
15 This is more problematical. Let's say there are
16 more scientific issues here and we are not ready to
17 make a proposal in the area of low solubility
18 drugs, but I will give you one example of my own
19 thinking, and that is if you take salicylic acids
20 like NSAIDs, ibuprofen, ketoprofen, the high
21 permeability drugs, we have measured most of them
22 in humans, all of them in animals and they dissolve
23 very rapidly at pH 6.8 because they ionize. The
24 ionize around pH 4-5. So, the solubility goes up
25 by two orders of magnitude in the intestine. In
171
1 this case dissolution occurs after emptying but it
2 is still a very fast process. So, if we think of
3 it kinetically, yes, there is a small effect of
4 dissolution on absorption but the principal
5 rate-determining step is in gastric emptying. So,
6 I think for Class II drugs, there are some Class II
7 drugs where we can extend biowaivers but that
8 requires more evidence and more debate and
9 discussion and we are not going to propose that
10 today.
11 [Slide]
12 Here is the equation that started my
13 career down this track, for those of you who are
14 interested in it. I am very partial to this triple
15 interval because no one has ever asked me a
16 question on this thing, but that is good. When I
17 had to give my first presentation in 1988, I was a
18 late addition to a program on dissolution and
19 absorption and had to talk about dissolution at an
20 AAPS workshop. I came to the conclusion I was a
21 late addition because it was a workshop on
22 dissolution and no one wanted to stand up and talk
23 about dissolution and absorption and
24 bioavailability and bioequivalence, and I was still
25 young at the time so I didn't know enough to say
172
1 no.
2 So, I wondered how do I handle it and I
3 concluded in the morning before the presentation
4 that if I showed this I would be safe. And, it
5 worked and I have been safe ever since. Basically,
6 it says that the determining factors are
7 permeability and concentration. Absorption is
8 occurring along the gastrointestinal tract. So,
9 you have to add up absorption processes across the
10 whole surface of the intestine. So, this is just a
11 surface integral and then you have to add it up
12 over time as well. But the key factors are
13 permeability and concentration, and in the limiting
14 case the highest concentration is solubility. So,
15 that is very simply Fick's first law. The two
16 critical variables are permeability and solubility.
17 Now, when I was on sabbatical at the FDA
18 in 1990-91, thinking about looking at dissolution,
19 working with Vinod Shah and Jerry Skelly at the
20 time, looking at how dissolution was used to set
21 regulatory standards, we had a regulatory issue
22 regarding carbamazepine at the time. So, I began
23 to think about is there a way--I could see that in
24 the struggle to come up with a guidance for
25 dissolution you would write a guidance that would
173
1 be so general that it was useless and it was a
2 product by product regulatory basis, so I thought
3 is there some way to kind of capture drug products
4 into categories that would be simpler to manage and
5 handle? Over the next couple of years, it took me
6 about two years to realize that the place to start
7 was Fick's first law. My major professor would be
8 appalled at that, Bill Laguchi who taught me Fick's
9 first law, but it took me two years to realize that
10 the starting point for predicting absorption is
11 Fick's first law, and that is P X C, Fick's first
12 law applied to a membrane.
13 [Slide]
14 At any rate, the waiver is applied to high
15 solubility drugs. We take the definition of high
16 solubility of a drug that the highest strength must
17 dissolve in a glass of water. What are you going
18 to use for high solubility? What is your reference
19 point? You have to come up with something
20 practical. This seems very practical to me, the
21 highest dose. Then I learned that sometimes you
22 can dose two of the highest strengths and
23 bioequivalence requirements currently use strength.
24 So, we then used highest dose strength but then
25 that was confusing too. The highest strength must
174
1 dissolve, the highest marketed strength must
2 dissolve in a glass of water. That is a high
3 solubility drug. I think it is a very practical
4 definition.
5 High permeability, we decided to define
6 high permeability and well absorbed as a drug that
7 is absorbed to 90 percent or more. Maybe we drew
8 that bar a little high, and one of the areas of
9 possible extensions is to change that to 85 percent
10 or 80 percent. We are looking at with that is
11 important or not from the point of view of the
12 database within the FDA. Further, if we extend
13 waivers to Class III drugs, which are low
14 permeability drugs, it makes this borderline a
15 little less critical perhaps in terms of drug
16 product regulation.
17 Then, the drug product must dissolve
18 rapidly. Based on theoretical simulation work done
19 at the time, we decided that 30 minutes would be
20 the upper limit for rapid dissolution even though
21 our simulation supported a 60-minute upper limit
22 for Class I drugs, high solubility, high
23 permeability drugs. But we chose 30 minutes, 15
24 minutes as a single point determination; 30
25 minutes, you would have to do a statistical
175
1 comparison using the F-2 metric.
2 [Slide]
3 This shows a partial database. Hussain
4 referred to a data base of about 25 drugs which is
5 being published over the past few years and over
6 the next couple of years, studied under virtually
7 identical conditions in normal subjects. So, we
8 have a permeability database that shows I think
9 around 15 or so of them. The high permeability
10 definition is appropriate metoprolol, approximately
11 where those red arrows are. Unfortunately,
12 metoprolol was mis-plotted on that plot but near
13 the intersection of the fraction absorbed curve and
14 the 90 percent line. So, we have used metoprolol
15 as our main reference compound. It is about at the
16 borderline between high and low permeability and it
17 is about 95 percent absorbed.
18 So, when we do permeability, and this is
19 permeability in humans, we almost always do it with
20 metoprolol being an internal standard. We
21 calculate permeability relative to metoprolol.
22 Yes, there are some potential interactions and they
23 are more theoretical than practical because we
24 rarely see them in vivo in humans or in animals.
25 So, we use metoprolol as a reference compound. If
176
1 the permeability in rat of CACO 2, if the
2 permeability is higher than metoprolol you have a
3 high permeability drug.
4 This allows you to determine the fraction
5 absorbed, the upper limit of the fraction absorbed.
6 The beauty of this is that in 1990 if you said you
7 could predict absorption people would have laughed
8 at you because no one even tried. Now we can
9 predict the upper limit. We just measure
10 permeability. That is the upper limit to systemic
11 availability. Systemic availability is always less
12 than or equal to fraction absorbed. So, from
13 preclinical data now we can predict how well we can
14 do the upper limit. Knowing the upper limit I
15 think is very important. We don't know the lower
16 limit. That is harder and it also includes
17 metabolism. So, the advantage of permeability is
18 that it can be scaled to preclinical animal and
19 even tissue culture methods for predicting
20 absorption.
21 Solubility, I didn't know what to say
22 about low solubility drugs so I put in my best
23 example here. When I think of low solubility and I
24 need a reference point of something that is low
25 solubility everyone would agree that marble is low
177
1 solubility. Right? I calculated the solubility of
2 Venus and she is ten mcg/ml, if I can remember my
3 old physical chemistry. As a reference point, a
4 drug like resiafulvin is about 15 mcg/ml. Some
5 other drugs, like glyburide are around 3 mcg/ml,
6 peroxicam about 7 mcg/ml.
7 So, I take about 10 mcg/ml as our
8 definition of a low solubility drug. But the
9 factors that we need to consider there in the
10 future are drugs like peroxicam which is actually a
11 high solubility drug at pH 6.8, not a pH 3 but pH
12 6. So, we will be looking at potential extensions
13 for drugs that ionize and dissolve in the
14 gastrointestinal tract in the future.
15 [Slide]
16 Just to illustrate kind of the effect of
17 dissolution, I think we have lost sight of the
18 importance of dissolution. So, I calculated the
19 dissolution times here based on the solubilities
20 and assumed particle size. Cimetidine dissolves in
21 one minute at 25 micron particle, typical particle
22 size. Glyburide, which has a thousand times lower
23 solubility, takes 30 hours to dissolve. That is
24 the reason dissolution is critical for glyburide
25 but for cimetidine it is not. This emphasizes
178
1 compartmentalizing the drugs because some are
2 simple and some are hard. Let's not try to
3 regulate everything by the hard rules. Let's try
4 to separate them out and say these are hard and
5 these are simple, and there are some drugs where we
6 may be doing in vivo studies forever because it is
7 too complicated. I also tried to calculate the
8 dissolution time for Venus. I had to use a
9 particle size for Venus so that meant I had to go
10 to the Louvre and see Venus because, you know, you
11 can't tell from pictures. Venus is a big lady, if
12 you have ever gone to the Louvre to see Venus. So,
13 I used a one meter particle size for Venus and I
14 calculated this number. I think it is like
15 million, billion, trillion, and I don't know what
16 the next number is. Does anyone know what the next
17 number is after trillion? One thousand trillion?
18 That is a long time although compared to the age of
19 the earth it is not so long. At any rate, this is
20 the reason solubility is so critical and why for
21 high solubility drugs the dissolution is very rapid
22 and there is not a problem with regard to
23 bioequivalence.
24 [Slide]
25 The waivers of in vivo, so-called
179
1 biowaivers, and I will emphasize this again,
2 biowaivers are not waiving bioequivalence. They
3 are waiving the in vivo test. They are
4 substituting another test which is as good or
5 better. We require bioequivalence. The question
6 is what test. Either a single point of 15 minutes
7 or a minimum of three points if there is 85 percent
8 dissolution at 30 minutes. Then, three pH's,
9 simulated gastric fluid, simulated intestinal fluid
10 and then an intermediate pH of 4.5 because that is
11 a pH which a drug sees as a transition from the
12 stomach to the duodenum and jejunum. In the
13 duodenum you have the mixing of gastric acid from
14 the stomach and the pancreatic bicarbonate secreted
15 from the pancreas through the common bile duct, and
16 also duodenum mucosal secretions. So, there is a
17 tremendous pH fluctuation in the upper duodenum and
18 so we included pH 4.5. So, the drug must dissolve
19 rapidly at those three pH's. We felt that was a
20 very safe criteria for allowing waivers from in
21 vivo bioequivalence.
22 [Slide]
23 Just by way of reference, I included here
24 one slide on the gastric emptying work that we
25 actually did via intubation, where we intubated
180
1 humans and measured gastric emptying of a liquid.
2 Here we used volumes of 50 ml and 200 ml of liquid
3 and then measured the gastric emptying rate. We
4 monitored motility, phase 1, 2 and 3, and then the
5 overall mean. The overall mean for the 50 ml
6 volume was around 22 minutes and the overall mean
7 for gastric emptying for the 200 ml volume was
8 about 12 minutes. So, the gastric half emptying
9 time was typical volume we would administer.
10 Actually a glass of water, the FDA requirement, is
11 8 oz. So, we used 200 ml here because this was a
12 long time ago. The gastric emptying time is about
13 12 minutes.
14 That was the basis for choosing a
15 15-minute, 85 percent dissolution time. Other data
16 from the literature--Ian Wilding has done a lot of
17 that from pharm profiles; and Bob Davis in
18 Nottingham. So, the gastric emptying time is very
19 well established so we felt very confident in the
20 gastric emptying time. We used 200 ml. I have
21 come to realize that that is actually closer to the
22 official Japanese glass of water which is 6 oz.
23 When I realized that I immediately thought of
24 harmonization. Do you think we could ever
25 harmonize a glass of water? This is an example of
181
1 cultural differences. No matter what we, as
2 scientists think might be possible, I doubt that we
3 are going to get cultures to change their official
4 glass of water.
5 [Slide]
6 I think I can summarize by saying there
7 has been strong support or at least very limited
8 resistance. I would like to think of it as strong
9 support but I will take limited resistance for BCS
10 and biowaivers. There have been some concerns
11 expressed at the workshop and commentaries on the
12 BCS guidance. For example, there is some
13 inconsistency between the solubility and
14 dissolution specifications. In particular, for
15 solubility we specify up to pH 7.5 while for
16 dissolution we only require a pH of 6.8. We think
17 we should harmonize those, and one of our proposals
18 is to look at the implications of changing the pH
19 7.5 solubility to pH 6.8.
20 Also, there are many completely absorbed
21 drugs whose systemic availability is less than 90
22 percent. That is kind of a paraphrase. That is
23 like what Marvin was saying this morning.
24 Bioavailability is easy. Fraction absorbed can be
25 hard. So, there is this concern out there that
182
1 fraction absorbed is actually hard to measure.
2 Probably you have to do radiolabeled studies. You
3 can use animal data for radiolabeled studies. You
4 need to do IV and oral because some drugs may be
5 excreted in the feces as well as the urine. You
6 need to measure generally your unchanged drug in
7 the urine, and the ratio IV to oral can be used to
8 estimate fraction absorbed if it is not too highly
9 metabolized. But estimating fraction absorbed is a
10 little tricky. Nevertheless, from the point of
11 view of the scientific approach, focusing on
12 fraction absorbed from the point of view of setting
13 dissolution standards is the correct view, I
14 believe, and fraction absorbed is what we want to
15 regulate.
16 Systemic availability contains absorption
17 plus metabolism. Generally metabolism is not a
18 formulation factors. Yes, you can add some things
19 and that is another factor. So, the systemic
20 availability complicates regulations because of the
21 metabolism variability. So, this allows us to
22 separate out. While we can't solve and simplify
23 all drug products this way, we can simplify I think
24 quite a number of them.
25 The third point is that we are overly
183
1 conservative. I think everyone agrees with that
2 and we should apply waivers to Class III drugs as
3 well.
4 [Slide]
5 More broadly, this kind of summarizes the
6 extension issues that we have been debating for the
7 past--well, I would say it started in 1995 when
8 Ajaz was drafting the guidance. Changing the pH
9 for solubility determination to 6.8 from 7.5;
10 reduce the permeability class boundary from 90 to
11 85 percent. We are not proposing that today
12 because, quite frankly, we are not sure about that.
13 We need a rationale to come to the committee and
14 there are a couple of different ways of doing that
15 using actual compounds and data, but we are not
16 prepared to do that today.
17 Class II, we feel these require extensive
18 research and they, again, are not subjects for
19 extension at this point in time for this
20 intermediate solubility class of drugs that
21 dissolve in the intestine. If there is one
22 solubility you want to know, it has to be the
23 solubility in the intestine for oral delivery
24 because that is where the drug is absorbed. So, pH
25 6.5 or 6.8 to be consistent. So, the solubility of
184
1 pH 6.8 is the single most important solubility for
2 oral delivery. If a drug dissolves rapidly at pH
3 6.8 it may be a candidate for waiver as well but,
4 again, that is going to require more studies.
5 Then you could ask the question about
6 surfactant. What about if it dissolves rapidly in
7 the presence of surfactants? Again, the Class II
8 drugs represent more complicated formulations,
9 perhaps more complicated dissolution
10 methodologies--not perhaps, more complicated
11 dissolution methodologies.
12 Then, for the Class III drugs the high
13 solubility, the low permeability drugs we want to
14 allow waivers if there is 85 percent dissolved in
15 15 minutes. So, again, it is a matter of getting
16 data and evidence to support that.
17 [Slide]
18 To conclude, I think we have established a
19 new paradigm. It has been a long process, starting
20 more than ten years ago with public discussion and
21 debate, including the support of this committee and
22 the FDA and the support of research, external
23 research as well as the many internal meetings in
24 developing the consensus in moving this new
25 paradigm in bioequivalence ahead.
185
1 I think one of the big advantages, of
2 course, is it reduces unnecessary in vivo studies.
3 I didn't realize, this was in the code of the
4 Federal Register, somebody gave me a new reference
5 today that the CFR says we don't want to do
6 unnecessary human studies. I didn't know that that
7 was in there so I have to add that to my slides.
8 But it reduces unnecessary human studies, and it is
9 based on scientific principles that allow us to
10 formulate a hypothesis, do some tests and move
11 ahead.
12 To conclude, I guess it is a great
13 pleasure for me to be here, talking to this
14 committee again and seeing the progress that we
15 have made over the past few years and seeing the
16 interest in extending and in building on it where
17 we can to improve, with our overall goal, of
18 course, of improving public health policy
19 standards. Thank you.
20 DR. COOK: For those that don't know me
21 and probably for those that do, I am Jack Cook,
22 with Pfizer Global R&D. My purpose today is to
23 show you that at least some in industry would
24 welcome additional guidance.
25 [Slide]
186
1 The agenda is that first I want to talk
2 about what I see are the benefits for industry with
3 the current guidance. Second, I want to talk about
4 the barriers because if you talk to Ajaz or
5 Lawrence you will find out that there have only
6 been six, plus or minus one, applications for
7 waivers so far. Finally, I want to talk about what
8 I see are the future benefits for the guidance.
9 [Slide]
10 First the benefits, the BCS guidance
11 allows bioequivalence to be shown by dissolution in
12 lieu of in vivo studies, but the question is will
13 it really save money, and at what cost?
14 [Slide]
15 I looked at the data availability at the
16 FDA web site, and I found over the period from
17 January 1998 to May of 2001 that there were 229
18 different NDA approvals, at the rate of about 67 a
19 year. Over the same time there were 466 ANDA
20 approvals, at a rate of 136 per year. NDAs, I
21 could find data from a recent study by DiMasi, that
22 about 90 percent of those are approved. Also, from
23 the DPQR site, we find that three to six studies
24 per NDA submitted their bioequivalence studies and
25 generics always get it right on the first time so I
187
1 assume that there is one bioequivalence study for
2 an ANDA. When you massage all of that data, you
3 get that industry as a whole performs 350 to 600
4 bioequivalence studies per year. That is probably
5 a little low estimate because it doesn't talk about
6 the drugs that didn't make it to market, and it
7 doesn't talk about studies that aren't submitted.
8 But at least that was a starting idea of how many
9 studies are performed a year.
10 [Slide]
11 The next thing I wanted to look at is what
12 does it cost. At least at Pfizer, Ann Arbor, when
13 you consider the cost for packaging and maintaining
14 samples, the clinical cost to run a study, the
15 bioanalytical cost, the data analysis and report
16 generation or my yearly salary, and then the
17 internalization, it costs us about a quarter
18 million dollars a study to run.
19 [Slide]
20 Again, if you take that number, about 25
21 percent of all drugs are waiver candidates. I
22 don't have a slide on that but that comes from a
23 survey I did over the same period of time, looking
24 at potentially how many drugs are waiver
25 candidates--I should mention that very quickly.
188
1 What I did, I looked at labeling and additional
2 data that were out in the literature, decided that
3 a drug could be classified as highly soluble if I
4 could find that the highest strength was soluble at
5 some pH between 1 and 7.5, but there was no other
6 pH that would preclude it from being a highly
7 soluble drug. So, I didn't have extremely high
8 evidence of it being Class I but I couldn't
9 preclude it from it. So, it could be as many as 25
10 percent.
11 To me, for the permeability classification
12 there was enough data in the literature where it
13 would have to meet one of the BCS requirements.
14 Anyway, if you accept that number of 25 percent you
15 can find that the industry as a whole could save
16 between 22 and 38 million dollars a year.
17 [Slide]
18 If I were to apply that same thing to
19 Pfizer in Ann Arbor, we would find that it is
20 somewhere between half and one million dollars a
21 year at our site, considering that we do about 17
22 bioequivalence studies a year.
23 [Slide]
24 I call that direct savings. There are
25 some direct savings. It is not that unusual for us
189
1 to have bioequivalence studies that are
2 rate-limiting to submission. A typical scenario is
3 that we are changing the site of manufacture and we
4 want to include that bioequivalence study in our
5 submission. So, we, those that would do the in
6 vivo testing, end up being behind the eight ball as
7 rate limiting. Typically, it takes us about six
8 weeks to actually run the study and get the results
9 back. I won't talk about how long it takes us to
10 generate the report, but let's say six weeks to say
11 that we have a product going forward. Assuming
12 that we have peak sales of a drug of one billion
13 dollars, not one trillion dollars, a year, that
14 ends up being that there are 110 million dollars
15 that one can save by doing the in vitro testing
16 rather than the in vivo testing.
17 [Slide]
18 That is all well and good, I want to
19 assure you that there is a cost savings. If you go
20 out and do the formal testing of something to
21 classify something as an in vitro methodology you
22 do, indeed, save money. The characterization cost,
23 depending on how you choose to characterize your
24 compound as highly soluble, highly permeable, ends
25 up being between $10,000 and $60,000 per drug.
190
1 Then, to evaluate a formulation, because that is
2 the second step because not only to you have to
3 have a Class I drug but you have to do the in vitro
4 dissolution for the formulation, is about $15,000
5 per formulation. I have stolen this slide from
6 another talk, but it ends up that that total cost
7 of that $75,000 is far less than the quarter
8 million dollars it costs us to run a study.
9 [Slide]
10 A few years ago I had the opportunity to
11 try this at Pfizer, and I likened it to a favorite
12 poem of mine by Robert Frost, the Road Not Taken,
13 that talks about decision in life and I thought the
14 BCS was the more attractive road and chose to take
15 that less traveled path. I have good news with
16 drug X, which is that we were able to obtain a
17 waiver of in vivo studies and show that it met in
18 vitro bioequivalence requirements. We saved four
19 bioequivalence studies and, like the last line of
20 the Robert Frost poem, that has made all the
21 difference in that it saved Pfizer, Ann Arbor, one
22 million dollars.
23 [Slide]
24 So, why isn't everybody else jumping on
25 the bandwagon? We have seven applications but,
191
1 yet, a quarter of all drugs could potentially meet
2 BCS classification. There are a couple of barriers
3 that actually are not within the agency but within
4 industry itself. One is what I call wrong
5 attitudes, mainly because they don't agree with
6 me--
7 [Laughter]
8 --secondly, about wrong wiring. When I
9 first proposed going this different path within the
10 company, saying I don't want to run a traditional
11 in vivo bioequivalence study; I want to run an in
12 vitro bioequivalence study, it wasn't my decision
13 alone. I needed to take it to the head of my
14 department, the head of regulatory, the head of
15 formulations department.
16 [Slide]
17 To a person, this is the kind of response
18 I get, "you want to do what? Does the agency allow
19 such a thing?" I said, "well, sure they do. Here
20 is the guidance on it." "Has this been done
21 before?" I said, "no." They said, "what, are they
22 crazy?" There is a good scientific rationale
23 behind that.
24 [Slide]
25 So, some of the questions I get are "you
192
1 can't release a new product on the market without
2 testing." That is questioning the science. I do
3 point out that we have been doing this all along
4 with solutions, and the BCS Class I is something
5 that is very similar to solution; it is something
6 that is dissolving very rapidly, behaving very much
7 like a solution.
8 As I mentioned, "the FDA won't allow it."
9 They question the procedure. Actually, what I have
10 been doing to my colleagues in industry is
11 advocating that they get an advocate within the
12 agency to talk to their regulatory people within
13 the company and say that, yes, indeed, it can be
14 done. "Has this been done before?" Fear of the
15 unknown. I go all the time and talk about our
16 success with trying to encourage it.
17 [Slide]
18 There is another thing that kind of stops
19 industry from doing it and that is wrong wiring.
20 This is kind of a diagram of what is needed for BCS
21 classification as far as information flow.
22 Typically within a company, my colleagues in
23 preclinical, there is very good information usually
24 coming to me in the clinic. Chemistry provides
25 decent information with their formulation
193
1 scientists. What is actually needed for the BCS is
2 something like this, there has to be a lot more
3 talk across these inter-departments because we are
4 relying on information generated elsewhere. If I
5 am using preclinical data to help classify a
6 compound as highly permeable, chemical
7 characterization is the one that usually does the
8 full dissolution profile. So, we need to figure
9 out how to have better information flow.
10 The next thing I am doing is bringing
11 across dollar amounts. The size of the dollar sign
12 represents the change in costs for a department.
13 Red means that the costs for a department go up
14 when they decided to classify something this way.
15 For instance, chemical characterization has to do
16 more characterization on a compound than they are
17 used to. Green means where it saves. So, as you
18 can see, I am in clinical pharmacokinetics, I look
19 good and I can claim that we saved our company a
20 million dollars, but other parts of the company are
21 actually spending more. So, this is another
22 barrier that one has to overcome within industry
23 and is why it hasn't been used so much.
24 [Slide]
25 I am going to talk about that a little bit
194
1 when I talk about blue sky, how will industry
2 benefit from the proposals.
3 [Slide]
4 Change within a company is kind of like a
5 chemical reaction. To orient you on the slide, on
6 the Y axis is kind or resources, and going from the
7 old, on the left-hand side, to the new, on the
8 right, you can see that overall if I use the old
9 way, the in vivo bioequivalence, I actually have to
10 spend more resources than the new. But I have to
11 overcome this barrier of activation energy. I have
12 to change how data flows within a company. I have
13 to overcome some mind sets.
14 I submit that if there is benefit and it
15 is only slightly better than the activation energy,
16 that change is going to be slow in a company. They
17 are going to fail to see that for that little good
18 we have to change all these ways that we do things
19 within a company. On the other hand, if through
20 expanding the BCS we can provide a lot broader
21 application of it, those systems will change a lot
22 faster and we will see actually a far greater use
23 of BCS within industry.
24 [Slide]
25 In that same survey I looked at how many
195
1 drugs are potentially future candidates for BCS if
2 we were to include all highly soluble compounds.
3 From that survey we come with something like 45
4 percent of all candidates would be considered
5 highly soluble, with another 25 percent unknown.
6 So, given that some will fall out of that 45
7 percent, they may be replaced by the 25 percent and
8 I submit that that is probably not too
9 unreasonable. So, there is a great potential for
10 the number of candidates that the expansions
11 proposed today would cover.
12 [Slide]
13 I would like to leave you with a few
14 thoughts. First, we feel that the current guidance
15 is useful. Pfizer has saved over a million dollars
16 with it. The barriers right now within company on
17 changing paradigms result in the low rate of use
18 they have so far with the guidance. To overcome
19 that, one thing that will help is expanding the BCS
20 where more candidates will equal a greater savings,
21 and that will be very useful for companies and, as
22 I say, you will see it used a lot more. With that,
23 I will turn it back over to Lawrence.
24 DR. HUSSAIN: Vince, can I make a comment?
25 DR. LEE: Yes.
196
1 DR. HUSSAIN: I think one of the benefits
2 that I think needs to be on the table is the
3 concept of quality by design and I just want to
4 bring a formulator's perspective here. When the
5 work of a formulation development group starts, for
6 initial screening everything is based on in vitro
7 dissolution and we pick a dissolution that we think
8 might work. Actually, we have seen cases where
9 companies may go down the path and actually
10 optimize their formulation before they do the first
11 bio study and in that study the dissolution test
12 was all wrong to start with.
13 So, focusing on the dissolution, relevant
14 dissolution, helps us to do the right thing the
15 first time and I think that is one of the
16 scientific benefits that is not always clear. So,
17 bringing more science to formulation development
18 and linking biopharmaceutics to formulation
19 development is another big benefit here.
20 Also, when I was working on the BCS I saw
21 18 bioequivalent studies in one NDA, and I am not
22 so concerned with the cost at this point. I am
23 more concerned that this is a new drug entity for
24 which the safety and efficacy has not been fully
25 evaluated and you are exposing normal, healthy
197
1 volunteers to a test which may not be adding all
2 the value. I think that is the motivation that
3 sort of drives us here.
4 DR. JUSKO: Could I ask Jack to clarify
5 one thing here?
6 DR. LEE: Certainly.
7 DR. JUSKO: The test compound that you
8 described, I presume you already had oral and IV
9 data for that compound.
10 DR. COOK: Actually, the way we classified
11 it as highly permeable is that this drug is
12 excreted virtually unchanged in the urine. So,
13 just by measuring urinary excretion we were able to
14 show that the bioavailability was above 90 percent.
15 DR. JUSKO: So it was a Class I compound?
16 DR. COOK: Oh, yes. This is a Class I
17 because that is the only way currently that you can
18 get a waiver for in vivo bioavailability. What we
19 are proposing today is to expand that further.
20 DR. JUSKO: Thank you.
21 DR. YU: Thanks, Dr. Amidon for the
22 excellent presentation for an overview and
23 applications of the biopharmaceutics classification
24 system, and Dr. Cook for an excellent presentation
25 on the industrial experience of the BCS.
198
1 I want to emphasize that the driving force
2 for us to have this current guidance and for future
3 extension is the science, the science behind the
4 philosophy driving this change. In the next twenty
5 minutes or so I will talk about two aspects. One
6 is regulatory implementations, and the second is
7 basically potential extensions of the BCS.
8 [Slide]
9 As you can see, this guidance was issued
10 in August, 2000. It is now about 18 months. This
11 guidance basically allows for biowaiver for highly
12 soluble, highly permeable and rapidly dissolving
13 and wide therapeutic window index drugs. There are
14 also characteristics of the drugs to ensure that
15 the solution is not the limiting step in terms of
16 oral drug substance process. Again, the
17 permeability is also not the rate-limiting step.
18 [Slide]
19 So, those characteristics allow them to
20 say that the gastrointestinal emptying is basically
21 the limiting step for these solid oral dose form
22 for BCS Class I drugs.
23 [Slide]
24 In terms of applications, basically this
25 guidance allows applications for BCS for
199
1 investigational drug applications for Phase I to
2 Phase II post-approval changes certainly as ANDA,
3 abbreviated new drug applications.
4 [Slide]
5 So far, we basically have received strong
6 scientific support. As Prof. Gordon Amidon pointed
7 out, there is very little resistance. Some
8 concerns expressed in the public workshops are that
9 we are too conservative or overly conservative with
10 respect to solubility class boundary with respect
11 to BCS Class III drugs, highly soluble and low
12 permeability drugs. Again, the submission activity
13 is relatively low. So far we have received a total
14 of about five NDAs, ANDAs and post-approval
15 changes.
16 [Slide]
17 I want to discuss with you some of the
18 experience we have had with this current BCS
19 guidance. This slide shows you basically the
20 experience with the solubility . The pH range for
21 solubility studies is 1.2, or sometimes we say 1.0
22 HCL to 7.5. Temperature is 37 degrees. The
23 solubility is basically the highest strength
24 divided by 250 at all relevant pH's. For example,
25 for diazepam what you are really looking for is
200
1 lowest solubility, in this case a pH of 7.4, to
2 determine whether this drug belongs to Class I or
3 belongs to another class, Class II or IV. So,
4 there are solubility studies, relevant pH, relevant
5 temperature, and determined by the lowest
6 solubility at all relevant pH's from 1.2 to 7.5.
7 [Slide]
8 I want to discuss with you the experience
9 with permeability. So far, the applications we
10 have received classify permeability based on the
11 following methods: pharmacokinetic studies in
12 humans. For example, bioavailability is basically
13 90 percent or above. To ensure the permeability of
14 this drug, that it is highly permeable.
15 We also received applications using an in
16 vitro cell culture model. We sometimes receive
17 inquiry about the literature method or literature
18 data. I have to point out that the agency has
19 little experience to accept literature data as the
20 sole evidence to support or to classify
21 permeability for the regulatory purpose.
22 [Slide]
23 In these four slides I took advantage of
24 the new technology and I just added them this
25 morning in the hope of addressing the concerns,
201
1 especially Dr. Marvin Meyer's concern about
2 permeability classification. It is not in your
3 handout. First I want to point out that the
4 permeability classification, especially the extent
5 of intestinal absorption, is not bioavailability.
6 Just because bioavailability or extent of
7 absorption includes the extent of drug input into
8 the system added to circulation, so it includes
9 everything, especially for example the solution,
10 metabolism and so on.
11 However, for the purpose of the BCS, you
12 use the extent of intestinal absorption which means
13 extent of drug across the intestinal membrane is
14 not considered a factor of solubility, for example,
15 metabolism is subject to hepatic metabolism. So,
16 we only consider one step here, the extent of drug
17 across membrane. While the bioavailability
18 considers many, many processes involved, including
19 the solution, gastric emptying, GI motility,
20 hepatic metabolism, and so on. So, there is a
21 difference between extent of drug absorption and
22 extent of intestinal absorption for the BCS
23 biopharmaceutics permeability classification
24 purpose, the extent of intestinal absorption.
25 [Slide]
202
1 In the guidance we basically specify a
2 number of methods. You can use any method you
3 would like to classify the drug in terms of
4 permeability class boundary in terms of
5 permeability class membership. So, there is a list
6 of a number of methods availability specified in
7 the guidance, including in vivo intestinal
8 perfusion in humans; including pharmacokinetic
9 studies for example in humans; including in vivo
10 and in situ intestinal perfusion in animals and,
11 certainly, we also include the in vitro cell
12 culture model.
13 [Slide]
14 I just want to elaborate to give you an
15 idea, if you use an in vitro method or an in situ
16 method, in order for this method to qualify to pass
17 the permeability of drugs for the regulatory
18 purpose, the sponsor is required to demonstrate
19 that he has established the so-called system
20 suitability, so basically to show the link or
21 relationship between the permeability, for example,
22 cell culture permeability, and extent of intestinal
23 absorption for 20 representative drugs. For
24 example, you have to have a drug, certainly for
25 these 20 drugs you have to spread from low, medium
203
1 and high. So, you have a certain range from low to
2 medium and high. You also have to show the in
3 vitro method integrity, for example using mannitol
4 or dextran as a marker. In the case of the cell
5 culture models, you have to show that the cell
6 culture model expresses the transporter for
7 example, in this case Pgp, P-glycoprotein
8 transporter.
9 [Slide]
10 In order for this specific model to
11 qualify for regulatory purposes with respect to the
12 permeability classification, you need to establish
13 the correlation between the extent of intestinal
14 absorption and in vitro cell culture permeability
15 in this case. This was done at the FDA lab, and
16 Donna Volpe is the investigator and actually she is
17 sitting in the audience.
18 As you can see here, for these 20 drugs we
19 pretty much get very reasonable correlations
20 between the extent of intestinal absorption and
21 apparent CACO 2 cell permeability. With this
22 establishment, this specific model in a specific
23 sponsor's lab can be utilized for class
24 permeability of drugs. Now, if you use the same
25 principle in a different lab you have to requalify.
204
1 So, we put in relatively conservative criteria in
2 place to make sure the data that come from sponsors
3 does show that the permeability of a specific drug
4 is highly permeable or poorly permeable.
5 [Slide]
6 Again, even with the permeability method,
7 not only do you need to show that the cell culture
8 establishes the system suitability to show that the
9 drug is highly permeable, you are also required to
10 do stability studies to make sure this drug which
11 you are measuring in an in vitro system is stable.
12 These are the recommendations in this slide based
13 on the guidance. For example, you need to show
14 that the drug is stable in simulated intestinal
15 fluid. You need to show that the drug is stable in
16 simulated gastric fluid. Certainly, for stability
17 purposes you need to use stability indicating
18 assay, validated assay. The guidance suggests at
19 this point that the drug is stable if less than
20 five percent is degraded in both small intestinal
21 fluid and the gastric fluid.
22 [Slide]
23 Basically, this is our view in terms of
24 regulatory implementation and some of the
25 challenges and issues we have faced so far.
205
1 Next I want to discuss with you the
2 revisions and extensions with respect to solubility
3 class boundary and with respect to biowaiver
4 extensions, especially for BCS Class III drugs.
5 The objective here, again, is to have a science
6 based in vitro solution to BE standards. Again, I
7 want to emphasize here that the driving force for
8 us to have extensions or to have the current
9 guidance is science. It is the science.
10 Let's talk about the first proposal
11 change, solubility class boundary. Currently, the
12 pH range in defined solubility is 1.2 to 7.5. The
13 potential future direction is for a pH range from
14 1.2 to 6.8 in defined solubility.
15 [Slide]
16 Basically, this is the GI tract here. You
17 have a pH in the stomach, pH in the small
18 intestine; pH in the jejunum. The pH range in the
19 stomach is 1.4 to 2.1 under fasting condition. The
20 pH range for the duodenum is 4.9 to 6.4. The pH
21 range in the jejunum is 4.4 to 6.6. Finally, the
22 pH range in the ilium is 6.5 to 7.4.
23 Let's look at how long it takes for drug
24 solid dosage forms to get into the ilium where the
25 pH is relatively high, as you can see, at 7.5. On
206
1 average, in terms of residence time it is 85
2 minutes for a drug particles to go through the
3 stomach, duodenum, jejunum and to the ilium. So,
4 it takes 85 minutes for a drug solid dosage form or
5 drug particles to get there.
6 Now let's look at what are our in vitro
7 dissolution criteria. Our in vitro dissolution
8 criteria is 85 percent dissolved in 30 minutes.
9 So, by the time the drug gets to the ilium it is
10 likely all the drug is dissolved. Intuitively we
11 would think if all the drug is dissolved, why do we
12 need this criteria? That is first.
13 Second, in our current dissolution testing
14 for BCS, we have a dissolution test at pH 1.2 or
15 0.1 HCL, 4.5 and 6.8. So, in this regard to have
16 consistency between solubility and dissolution
17 class boundary it seems reasonable to reduce the pH
18 requirement from 7.5 to 6.8.
19 [Slide]
20 Now let's move on the next potential
21 extension, which is BCS Class III drugs. Currently
22 we have a biowaiver for BCS Class I, namely highly
23 soluble and highly permeable. One proposal is a
24 wavier to highly soluble and poorly permeable
25 drugs.
207
1 [Slide]
2 So, the question we ask is why do we
3 choose Class III, why not Class II or Class IV?
4 For Class III drugs it is highly soluble and poorly
5 permeable drugs in rapid dissolving dosage forms
6 which essentially behave like a solution if the
7 dissolution of a solid oral dosage form dissolves
8 rapidly. It essentially behaves like a solution.
9 Let's look at the solution requirements
10 here. The FDA policy on oral solutions is
11 basically if bioequivalence is self-evident
12 biowaiver can be granted, and no in vivo
13 demonstration is needed if the test solution
14 contains no inactive ingredients or other changes
15 in formulation from the reference product that may
16 significantly affect the absorption of the active
17 moiety or active ingredients. So, only if the
18 excipients do not affect absorption.
19 [Slide]
20 Now let's look further in terms of
21 mechanistically. Again, you can dose oral dosage
22 forms such as tablets or capsules. A solution is
23 certainly a liquid dosage form. When the solid
24 tablet comes to the stomach or the solution comes
25 to the stomach, what happens for the solution is
208
1 basically gastric emptying, the emptying from the
2 stomach to the small intestine. However, for solid
3 products there is one process which is the
4 dissolution. So, there is a difference in terms of
5 the process in the stomach. But when it comes to
6 the small intestine there is not much difference
7 there. The drugs in solution get absorbed. So,
8 basically in the small intestine or in the colon
9 there is basically a process in terms of
10 mechanistic absorption which is the same for oral
11 solutions or for solid dosage forms.
12 [Slide]
13 Now let's look at the next assumption
14 here. We say if the test product equals a simple
15 solution, if we can show it, and if we have
16 reference products which equal a simple solution
17 then automatically you say the test product equals
18 the reference product if there are two criteria
19 here, they are rapidly dissolving and the second
20 criterion is no excipient effect on oral drug
21 absorption. No excipient effect.
22 [Slide]
23 This is basically a list of potential BCS
24 Class II drugs. I say potential because there is
25 no concrete information to support yes or no and so
209
1 I say potential. This is a list of BCS Class III
2 drugs.
3 [Slide]
4 So the hypothesis here is if two immediate
5 release solid dosage forms dissolve rapidly at all
6 physiologically relevant conditions and contain no
7 excipients that may potentially affect the oral
8 drug absorption of the BCS Class III drugs, then
9 the bioequivalence of these two solid IR products
10 is assured and biowaiver can be granted.
11 [Slide]
12 This is basically the proposal for studies
13 or data collection to test the hypothesis.
14 Certainly we can collect data from human
15 bioequivalence studies to compare a simple solution
16 with two solid dosage forms of at least ten model
17 BCS Class III drugs to show that those data may
18 confirm the literature, the NDA or ANDA or FDA
19 internal studies, maybe unpublished data. We are
20 thinking about going through the PQRI to collect
21 the unpublished data from the sponsors and, if
22 necessary, to do relevant in vitro dissolution and
23 cell culture studies.
24 There are two potential issues here. The
25 first issue is transport which we touched on in the
210
1 morning. As you can see, there is much in vitro
2 evidence to show that transport may affect the
3 absorption of a certain number of drugs. On the
4 other side, we though if dose proportionality is
5 shown over the range from the lowest to the highest
6 strengths, we can conclude that the effect of the
7 transporter may not be significant with respect to
8 the bioequivalence. It may be still significant in
9 terms of drug-drug interaction but with respect to
10 bioequivalence this may not be significant.
11 [Slide]
12 The next question is the potential effect
13 of excipients. Excipients of oral drug absorption
14 can certainly affect GI motility. They can affect
15 permeability. In order to minimize the risk of the
16 bioinequivalence caused by the excipients, we
17 basically have two options.
18 Option number one, we basically identify
19 and exclude excipients that may affect the
20 absorption or pharmacokinetics. In other words, at
21 this point we consider all excipients acceptable;
22 we identify one, we basically exclude it. That is
23 the first option.
24 The second option is we basically exclude
25 every single excipient at this point. We basically
211
1 include them when we find specific excipients have
2 no effect whatsoever on oral drug absorption in
3 vitro and in vivo. So far we have tested a number
4 of products and if they had no effect we included
5 them. So, basically those are the two options we
6 have.
7 [Slide]
8 With that, I conclude my talk and with the
9 following questions we want feasibility and input
10 from you. Thank you very much. Thank you for your
11 attention.
12 DR. LEE: Thank you, Lawrence. Ajaz?
13 Committee Discussion
14 DR. HUSSAIN: Just a perspective that I
15 wanted to share with the committee before we start
16 deliberations. When we put together the first
17 guidance that was published in August of 2000, what
18 were the reasons why we did not include Class III
19 is sort of the one thing which I wanted to point
20 out. The other thing which I wanted to point out
21 which I will address first is our regulations
22 currently allow waiver of in vivo studies when you
23 have in vitro and in vivo correlations. For
24 immediate release dosage forms we don't have that
25 option because correlations are usually not present
212
1 or not apparent because dissolution in many cases
2 tends to be not rate limiting.
3 So, in vitro and in vivo correlations have
4 not actually been very useful for most immediate
5 release dosage forms. There are a few exceptions.
6 So for the BCS based biowaivers, when you think
7 about it, we are making decisions on in vitro
8 dissolution as a source of comparison in absence of
9 such correlations. So the thought process and the
10 justification is based on mechanistic underpinning
11 of that.
12 If I look at Class III drugs, what sort of
13 held us back for recommending waiver in the first
14 instance when we looked at it was the issue of
15 permeability being a mechanism by which you
16 essentially have the same conditions in vivo. So,
17 the volume differences for dissolution in vitro and
18 in are sort of one reason behind that sort of
19 holding back from that recommendation.
20 Also, keep in mind that solution
21 bioequivalence has always been waived, or options
22 have been available for solutions, and some of the
23 work we did suggested that the way we evaluate
24 excipients would have to be tightened up. So, if
25 you look at the bioavailability, bioequivalence
213
1 guidance we actually use a higher standard for
2 solubility forms whereby we limit it to highly
3 permeable drugs because that is sort of protected
4 against some of the excipient effects. In the new
5 guidance that we issues on BA/BE it actually
6 pointed out some of the issues with respect to
7 sorbitol or osmotic ingredients for solution drugs
8 because we have been seeing cases were a solution
9 actually has lower bioavailability than a tablet,
10 and you have one example in your handout. Those
11 are sort of the motivations and thought processes
12 that held us back at that point. So.
13 DR. LEE: Thank you. Are there any
14 questions for the presenters? Yes?
15 DR. RODRIGUEZ-HORNEDO: Yes, maybe a point
16 of clarification, how do you define or how do you
17 classify a compound that is ionizable so that the
18 pH determines its solubility? It is not clear to
19 me from the reading material.
20 DR. YU: Solubility over the pH range is
21 defined as 1.2 to 7.5. So, if it is ionizable, for
22 example as a free base, the solubility will be much
23 higher at the low pH; the solubility will be lower
24 at the high pH. So, actually whether this drug is
25 highly soluble or not is determined by the high pH.
214
1 On the contrary, for acid, for example, the
2 solubility will be lower at the low pH and higher
3 at the high pH so that basically determines whether
4 this compound belong to high solubility or not by
5 the low pH. Essentially in terms of ionizable, we
6 basically ensure that it matches the solubility of
7 all pH's to determine whether it is highly soluble
8 or not.
9 DR. RODRIGUEZ-HORNEDO: So, it is
10 determined by the minimum solubility of the
11 compound at any pH?
12 DR. YU: Correct, absolutely, yes.
13 DR. RODRIGUEZ-HORNEDO: If I may ask a
14 question that is related to something we are going
15 to be discussing tomorrow, I guess then the
16 classification is also dependent on the solid state
17 of the material.
18 DR. YU: Absolutely.
19 DR. RODRIGUEZ-HORNEDO: So, if you have an
20 amorphous compound versus a crystalline compound it
21 will change the solubility. The classification may
22 change depending on solid state structure.
23 DR. HUSSAIN: Well, I think this is sort
24 of an equilibrium solubility.
25 DR. AMIDON: Solid state properties,
215
1 particularly if they can change when the dosage
2 form is introduced into the gastrointestinal tract,
3 are problematical. I think when we use solubility
4 here we think of it as approximate equilibrium
5 solubility. But, in reality, we are only
6 interested if the drug stays in solution for over,
7 you know, four to six, eight hours in the
8 gastrointestinal tract. We don't need to wait
9 days; in days the drug is out of the GI tract. So,
10 in some ways we think of this as kind of a kinetic
11 solubility, but to a physical chemist that is an
12 oxymoron, right, because solubility is equilibrium
13 by definition. So, we think of equilibrium
14 solubility. So, amorphous compounds or compounds
15 like carbamazepine which hydrate and change their
16 physical form in contact with water have to be
17 handled more carefully.
18 DR. LEE: Yes, Gloria?
19 DR. ANDERSON: On page three of your
20 presentation you have solubility equal to greater
21 than highest strength per 250 ml at all pH's. Is
22 there a number that you associate with solubility
23 that is highly soluble, not very soluble, or does
24 this high strength refer to the dosage?
25 DR. AMIDON: That is a good question. We
216
1 are asked that frequently. We use the term high
2 solubility of a drug to be one whose highest
3 strength dissolves in a glass of water. That is
4 not really solubility. That is what we are calling
5 a high solubility drug. You know, if your drug
6 dose is 250 mg and it has to dissolve in 250 ml, 1
7 mg/ml would be a high solubility drug. But if your
8 dose is lower, then you could go to a lower
9 solubility. So, the actual solubility changes with
10 dose. The solubility limit changes with dose.
11 DR. ANDERSON: And from drug to drug.
12 DR. AMIDON: And from drug to drug, yes.
13 DR. LEE: Joe, you have a question?
14 DR. BLOOM: Basically when it is called
15 high solubility it is depending on dose.
16 DR. COOK: It depends on the highest
17 formulation strength one would make. So, it is
18 drug specific and it is the highest strength, and
19 whether that strength will dissolve in 250 ml or
20 not at all relevant pH's. So, you can't think of
21 it as a milligram/ml; it is just a yes or no.
22 DR. KIBBE: And that applies to the
23 highest strength that is available whether or not
24 there are multiple strengths. No one can get a
25 waiver for a 5 mg tablet when a 50 mg won't meet
217
1 that criteria? Is that right?
2 DR. COOK: Currently.
3 DR. LESKO: I think it is important to be
4 clear. The solubility is based on the highest
5 approved strength. If you can imagine a
6 bioequivalent situation where there is a reference
7 product approved and somebody is looking at an
8 abbreviated new drug application, the highest
9 strength that is approved would be the reference
10 for solubility determination. That is different
11 than the highest dose that may be approved if, for
12 example, somebody can administer two tablets or
13 three tablets within the range of an approved dose.
14 That is not what we are talking about. We are
15 talking about the strength of the tablet. We are
16 trying to mimic a bioequivalence study where you
17 compare a tablet of drug that is a test to a tablet
18 of a drug that is a reference, and that is what we
19 want to compare at the highest strength.
20 DR. KIBBE: If four products are
21 commercially available from the innovator, four
22 dosage strengths, 2 mg, 5 mg, 10 mg and 20 mg, then
23 your decision to allow people to get a waiver is
24 going to be based on the highest one whether or not
25 they want to market the highest one or not?
218
1 DR. LESKO: That is correct.
2 DR. KIBBE: Even though they want to
3 market the 2 mg, they can't claim that the 2 mg
4 would meet your criteria and, therefore, it should
5 get a waiver.
6 DR. HUSSAIN: That is the way it is right
7 now.
8 DR. LESKO: You didn't say what the
9 highest approved strength was, but if 20 was the
10 highest approved strength, then that would be the
11 basis for the solubility determination.
12 DR. KIBBE: Regardless of what the company
13 wants to market?
14 DR. LESKO: Well, if they want to market
15 10 mg and they don't market 20 mg, then 10 mg would
16 be the reference.
17 DR. KIBBE: That is my point.
18 DR. LESKO: Yes.
19 DR. KIBBE: That just changed the answer,
20 I think. If there is a company on the market that
21 has four strengths and the highest strength is not
22 a very popular strength but it is on the market as
23 the innovator, and I want to only come in as a
24 generic and market the bottom two strengths, which
25 represent 80 percent of the market, I don't have to
219
1 have, to get a waiver, that the highest strength is
2 soluble at 250 ml. I only have to have the highest
3 strength I want to market that is soluble at 250.
4 DR. LESKO: That is correct.
5 DR. LEE: Has there been any thought about
6 using dose numbers in all these kind of
7 descriptives?
8 DR. AMIDON: Well, yes, actually if the
9 dose number is less than one than you are a high
10 solubility drug. So, really that is the way I
11 think of it.
12 DR. LEE: Yes, Bill?
13 DR. JUSKO: This is a very illuminating
14 set of presentations and I have learned a lot from
15 it. My first, somewhat facetious, comment is,
16 Gordon, I wonder why in your triple integral you
17 didn't include the upper limits of the A variable?
18 [Laughter]
19 We will talk about that later.
20 DR. AMIDON: you are the only one that has
21 ever asked that question. It is not really written
22 right but no one has ever noticed. It really
23 should be a vector integral, quite frankly. It
24 should be a vector integral written over the
25 surface of the intestine, yes.
220
1 DR. JUSKO: That makes everything clear!
2 [Laughter]
3 Speaking computationally, faculty members
4 in our department teach students about Lapinsky's
5 rule of five. I wondered if there is some role in
6 all of what you are doing for a computational
7 approach, structure activity kinds of calculations
8 to address estimations of permeability values.
9 DR. AMIDON: Yes, I actually use them all
10 the time. The question is what evidence would you
11 want to bring to the FDA. I think within industry,
12 if I don't have an experimental partition
13 coefficient I would calculate one just using some
14 software program. I mean, it is one of the first
15 things I do to determine kind of what the
16 permeability of this drug might be. So, I find it
17 a very useful qualitative tool. I know that there
18 has been some interest. Well, Lawrence has actually
19 done some computational work when he was with
20 Glaxo. I think the FDA is very concerned about
21 making decisions based on some computational
22 result, but I personally use them all the time,
23 yes.
24 DR. COOK: As somebody who may work for a
25 company who looks into this, we find it very useful
221
1 for candidate selection, realizing that compounds
2 with the desirable absorption characteristics are
3 ones that likely make to market. If you can have
4 activity plus it is well absorbed, you have
5 something that you should actually fast-track
6 through the system. Our experience is that they
7 have been very useful at that stage. The
8 correlations haven't been precise enough to where
9 we feel comfortable saying for sure it is a Class I
10 compound, and to, you know, absolutely predict it
11 is above 90 and, therefore, do other tests. But
12 some day maybe.
13 DR. JUSKO: In the graphs that I saw
14 showing the non-linear relationship between
15 fraction absorbed and permeability, there was a lot
16 of data on the high side and only three or four
17 points, some complicated by metabolism issues,
18 indicating small fraction absorbed when
19 permeability was low. Plus, the relationship was
20 very steep. So, that makes people wonder how
21 reliable predictions are going to be if the
22 critical information has such a steep profile.
23 DR. COOK: Well, thank goodness, the area
24 of interest is actually the flat part of the curve
25 because if you look at where metoprolol is, that is
222
1 kind of where it starts the flat part of the curve
2 and you have to be there or greater to be
3 considered a highly permeable compound. I think
4 most people agree that that is really hard on that
5 area of the curve where a little bit of
6 insensitivity in your assay measurement could
7 result in a big change. Here, we are on the flat
8 part of the curve and are less susceptible to that.
9 DR. HUSSAIN: I think that is an excellent
10 point.
11 When we were putting in the class boundary that
12 actually came as a decision-making point. The
13 reason we said 90 and above is because of that.
14 Originally I think we thought of 80 and that is the
15 steep part of the curve, and one of the criteria
16 for 90 percent as the boundary was driven by that.
17 At the same time, I think for assessment
18 of permeability one of the recommendations in our
19 guidance is actually to use an internal standard, a
20 known high permeability internal standard so that
21 you can say it is better than that. That is how we
22 addressed that.
23 DR. JUSKO: That is what I didn't quite
24 understand from Dr. Yu's presentation, whether he
25 was indicating that the companies needed to study
223
1 all 20 drugs and establish the profile or could
2 just use the indicator drug as a cut-off.
3 DR. HUSSAIN: The recommendation is to
4 actually establish your own system with all 20
5 drugs; demonstrate suitability, and once you have
6 demonstrated suitability of the method, because lab
7 to lab variability is significant in some of those
8 things so we wanted every lab to define suitability
9 and then, after that you could use one of the
10 internal standards.
11 DR. JUSKO: In these recommendations you
12 are going by cell culture systems. I wonder, is
13 there no room for animal data? Win Chao has shown
14 a very nice correlation between fraction absorbed
15 of a large number of drugs in rats and man.
16 DR. HUSSAIN: I think with respect to
17 extent of absorption, animal data is allowed with
18 respect to perfusion experiments in direct methods
19 of permeability. We stopped short of using extent
20 of absorption in rat. I know we had that
21 discussion with Prof. Win Chao and he had about 100
22 compounds. So, we stopped short of that in our
23 recommendations in the guidance. But animal
24 perfusion experiments truly are okay. They
25 qualify. So.
224
1 DR. YU: In fact, I have a similar plot
2 which is from rat instead of CACO 2, also 18 drugs.
3 DR. LEE: Jurgen?
4 DR. VENITZ: I am very supportive of the
5 approach and I want to congratulate Gordon and the
6 FDA for moving it along as far as you have. Very
7 much like Marvin, I have some concern about the
8 permeability assessment based on in vitro data. I
9 guess I am wondering whether you have any
10 experience with misclassification using the human
11 in vivo as your gold standard. In other words, if
12 you know you have bioavailability of 90 percent or
13 above, you have a high permeability drug. How does
14 that compare to the in vitro predictions based on
15 CACO 2 cells lines?
16 DR. HUSSAIN: I don't have any experience
17 where we have found that problem occur. We are
18 actually in the process of publishing a validation
19 study, our own data, on in vitro studies, and Donna
20 will be here who has done that work. So, I don't
21 have an example.
22 DR. VENITZ: I know of one that was
23 supposed to be a poor permeability drug and it
24 turned out to be a high permeability drug--
25 DR. HUSSAIN: Cimetidine would sort of
225
1 come to my mind as probably an example where I
2 think extent of absorption in vivo in humans, the
3 data would suggest it is either 100 percent or
4 slightly less than that. But under CACO 2 and
5 other perfusion studies, it comes out as low
6 permeability. So, misclassification is on the
7 lower side.
8 DR. COOK: Yes, I would echo that. I did
9 an informal survey of some other companies and that
10 is what their indication was, that more often than
11 not the CACO 2 system was very conservative.
12 DR. VENITZ: With your proposal that
13 wouldn't be a big deal because you are lumping one
14 and three together. So, it doesn't make a
15 difference in terms of the waiver. But as it
16 currently exists, that would make a big difference
17 in terms of with you are waiving or not.
18 DR. AMIDON: it would only make a
19 difference in the dissolution standard you would
20 have to meet.
21 DR. VENITZ: Right. The second question I
22 have for you is about this Class III extension. Do
23 you have any experimental evidence, other than the
24 theoretical considerations that you went through,
25 to suggest that for a Class III compound we can
226
1 safely waive it and still show in vivo
2 bioequivalence.
3 DR. YU: This is basically for
4 information. We are considering those extensions
5 and we will come back with the data next time. We
6 will come back next time to this same committee
7 with data.
8 DR. LEE: So, Lawrence, you understand
9 correctly that probably the high end of the Class
10 III would be more like the low end of the Class I?
11 DR. YU: Yes.
12 DR. LEE: Therefore, you can waive it?
13 DR. YU: Yes.
14 DR. AMIDON: I think there are some drugs
15 where there have bee intubation studies, you know,
16 gastric, duodenal, jejunum. So some of those data
17 sets are availability for at least one or maybe two
18 Class III drugs in published literature. There is
19 more data also in NDAs. I think, for example, that
20 type of data showing site dependence would be one
21 set of data.
22 DR. VENITZ: Since you are going to go out
23 and come back, that would be the kind of data I
24 would like to see to support it experimentally, not
25 just theoretically saying we think Class III is
227
1 fine.
2 DR. YU: Absolutely. We are looking, for
3 example, at the evidence which would show
4 bioequivalence between solid oral dosage forms
5 versus a solution. We have about seven or eight
6 drugs right now. We intend to collect at least ten
7 drugs to deny or confirm the hypothesis we
8 discussed here today.
9 DR. LEE: Larry?
10 DR. LESKO: Yes, I wanted to answer that
11 last question because when we were doing the
12 research at the University of Maryland as part of
13 the scientific basis for the SUPAC guidance we had
14 two drugs from this class, the class that we are
15 talking about today, Class III with the high
16 solubility, low permeability, and Lawrence had them
17 on one of his slides, cimetidine and ranitidine.
18 Those were another two drugs which we tried,
19 through various manufacturing method changes, to
20 sort of ruin the formulations, create big
21 differences in dissolution but in vitro they were
22 very robust in terms of bioequivalence. So, I
23 think that is some evidence that would support what
24 Lawrence is talking about.
25 DR. VENITZ: So, you showed that the two
228
1 different solubility forms were bioequivalent in
2 vivo?
3 DR. LESKO: Yes.
4 DR. VENITZ: What about comparing the
5 solution to a solid dosage form?
6 DR. HUSSAIN: Well, I think that is what
7 Lawrence is proposing now but we don't have
8 prospective data on that now. We have some
9 in-house data but I think Lawrence is proposing to
10 do some studies comparing solution to tablet, and
11 so forth. So, that is one of the sets of
12 experiments that we probably will bring back to
13 this committee.
14 The other experiment that is ongoing right
15 now, we have completed the manufacturing and so
16 forth, and actually the studies have begun at
17 Tennessee, the biostudies. That is to create
18 formulations of a low permeability drug. We took a
19 low solubility, low permeability drug, furosemide,
20 and created dissolution profiles which are very
21 different and actually induced a pH sensitivity in
22 that. I don't know when those studies will be
23 completed but they have already begun. So.
24 The solution as a standard I think is also
25 important because when we were doing the BCS
229
1 guidance we looked at excipients. I think
2 excipients come back as an issue, and we were
3 collecting data with solution, simple solution that
4 was established, and I think from that we
5 identified about 50 excipients which are commonly
6 used which don't seem to have an effect. So, we
7 could build a basis for that and I think that was
8 one of the questions Lawrence posed, should we
9 identify excipients which may be potential
10 problems. That is what we tried to do in the first
11 guidance. I think that is the easier route because
12 for solid dosage forms there are only about 50
13 common excipients and you can make all sorts of
14 dosage forms with those 50 excipients. Of those,
15 the potential problems were surfactants, sodium
16 laurel sulfate, and so forth. And, we have
17 supportive data to say it may not really be an
18 issue in vivo. So, that database also could be
19 brought back.
20 DR. JUSKO: Do you think you can ever
21 really be conclusive about the excipients? Because
22 there could be a very specific interaction between
23 a particular excipient and a drug based on their
24 distinct chemical features.
25 DR. HUSSAIN: We that interaction be a
230
1 chemical or physical interaction, or an interaction
2 at a transport or absorption level? I think our
3 focus is more on the absorption because that is
4 where the concern is. If it is a physical
5 interaction or a chemical interaction, it sort of
6 comes out as a stability issue rather than a bio
7 issue in some cases. So, there would be different
8 ways of addressing chemical interactions.
9 DR. JUSKO: Might one manufacturer use a
10 particular excipient in their product and someone
11 use a different one, and then there would be a
12 potential difference?
13 DR. HUSSAIN: That is possible. For oral
14 products you could have different excipients.
15 DR. LEE: Particularly with the Class IV
16 drugs. Well, shall we keep this conversation
17 going? Marv has a question.
18 DR. MEYER: Yes, one question perhaps to
19 Lawrence. Is there a greater potential for an
20 error to be made for the Class III than Class I? I
21 am asking from the standpoint of your table. If
22 you take a drug, Class I by definition is 90
23 percent fraction absorbed, the most we can go up to
24 is 100 percent. If you take glycinopril, it is 30
25 percent fraction absorbed, and that goes up to 40
231
1 percent. Now, you have a third increase in the
2 available drug. As you get down in FA you have
3 bigger orders of change if you do something to
4 cause a change.
5 DR. YU: That is why the effect of the
6 excipients is kind of critical.
7 DR. MEYER: Whatever. The formulation,
8 whatever you didn't see because you didn't do the
9 biostudy causes it to go from 30 percent to 40
10 percent or 30 percent to 20 percent. That is a
11 bigger change than 90 to 100 or 90 to 80.
12 DR. COOK: If I could jump in, you could
13 have a change the other way and have a drug that is
14 100 percent and all of a sudden it goes down to 10
15 percent. So, it is just depending on whether you
16 are looking at increased chance of adverse events
17 or a loss of therapeutic benefit. But I think the
18 risk is there--
19 DR. HUSSAIN: Jack, sort of a different
20 perspective on that, I think with the rapid
21 dissolution the likelihood is minimized the other
22 way around. I think the excipients with sodium
23 laurel sulfate, and so forth, I think the concern
24 that Marv raised was one of the reasons for holding
25 it back to highly permeable drugs. If you have an
232
1 excipient like sodium laurel sulfate that can
2 enhance permeability what will happen with highly
3 permeable drugs? Very little. But for low
4 permeability drugs the margin of error is high.
5 DR. AMIDON: I just want to caution or
6 advise the committee to not think of excipient
7 effects as yes/no but to think of them as
8 dose-response curves and it is a matter of at what
9 dose and what level they are having an effect. We
10 know that sodium laurel sulfate at a very low
11 concentration has no effect and at a very high
12 concentration dissolves the intestine. So, it is a
13 dose-response curve issue. So, having thought a
14 lot about this excipient issue and interactions
15 with the gastrointestinal track, if we get into the
16 excipient issues we should proceed carefully and
17 mechanistically in evaluating those potential
18 implications.
19 DR. HUSSAIN: A different perspective that
20 I think also is important with excipients is if
21 excipients have significant interactions that
22 alters bioavailability it actually is a much larger
23 issue than bioequivalence. It becomes a label
24 issue because if it is an interaction that changes
25 bioavailability the potential for that interaction
233
1 will be there in the marketplace and I think we try
2 to avoid that, and I think the excipients that are
3 available generally, with a few exceptions, are
4 essentially from that perspective. The famous
5 example is sorbitol.
6 DR. LEE: Then I will just propose that we
7 take a 15-minute break so that we can maybe focus
8 and discuss some of the issues more. Will you
9 please come back at 3:15?
10 [Brief recess]
11 DR. LEE: Lawrence posed two questions to
12 the committee. Actually, I should inform the
13 audience that I began to form study groups in the
14 committee to look at the issues. I have four
15 individuals working this particular topic, Bill
16 Jusko, Leon Shargel, Lemuel Moye and myself. Right
17 after lunch I delegated my responsibilities to Bill
18 and he is going to be the lead correspondent.
19 DR. JUSKO: Are you going to put the
20 questions back up that we are to focus on? We have
21 all heard from this morning's and this afternoon's
22 discussion about the BCS classification system and
23 the guidance that is in place for Class I drugs.
24 It was interesting to learn this afternoon how few
25 companies have actually taken advantage of this
234
1 classification system and proceeded to use it, with
2 only five or six having been indicated.
3 The discussion this afternoon provided
4 much more illumination of the basic scientific
5 ideas and regulatory approaches to dealing with the
6 BCS system, and we were asked to focus on two
7 particular questions. Within the second question,
8 it appears that there is plenty of room for
9 recommendations as to how to proceed with the
10 second question.
11 But let's go to the first one since it is
12 the easier one to deal with. We were shown that
13 there are discrepancies in the pH values used to
14 determine solubility versus dissolution. So, the
15 first question is should the agency consider
16 revising the pH range of the solubility class
17 boundary to be consistent with the dissolution pH
18 range?
19 In my own view, the answer is quite
20 obvious that one should seek consistency. Perhaps
21 other members of the committee would like to
22 provide their comments.
23 DR. MEYER: How about changing the other
24 one to 7.5, have the same range but have 1 to 7.5
25 instead of 1 to 6.8?
235
1 DR. AMIDON: Can we comment?
2 DR. LEE: Sure.
3 DR. AMIDON: I think one element there,
4 Marv, would be the harmonization also with Europe.
5 At a workshop we had at the end of January with
6 European representatives--6.8 is kind of an
7 international standard, U.S., Europe, Japan for
8 dissolution studies, simulated intestinal fluid.
9 So, I think it is partly also that, harmonization
10 to kind of the world standard. I think if we were
11 to go from 6.8 to 7.5 we would have some problems.
12 DR. KIBBE: Yes, I remember when I was a
13 young child my mother always telling me that you
14 don't do things because everybody else did them.
15 So, we have a harmonized number but the question
16 really is, is it missing information or not? That
17 is really the bottom line. Would we really miss
18 out on something important if we left out going up
19 to the physiological pH which exists at the
20 terminal end of the GI tract? If w are clear that
21 we are not going to lose anything, then it is okay.
22 If we are worried that we are, then we should
23 extend the other to 7.5 instead of cutting back to
24 6.8. That is the question I think.
25 DR. COOK: If I can comment on that, I
236
1 think the strongest evidence was when Lawrence put
2 up the slide about transit time, and it is 85
3 minutes to that terminal end but we are requiring
4 dissolution to be essentially complete within 30
5 minutes. So, it will never see the higher pH
6 before it is all dissolved.
7 DR. KIBBE: Your disease requirement is in
8 vitro dissolution and it is predictive of in vivo
9 dissolution, but in vivo dissolution of something
10 in 15 minutes might be 15 minutes and it might be
11 45 minutes. Okay. So, that still isn't a
12 guarantee. I am not saying that 7.5 is where we
13 ought to be, but I think we ought to know whether
14 we are losing any information.
15 DR. RODRIGUEZ-HORNEDO: I was going to
16 comment on that same point, and I struggled with
17 the way that the question is worded until I saw
18 Lawrence's slide with the pH in the different
19 regions of the GI tract. Maybe the question needs
20 to be reworded because it is not really a matter of
21 consistency with the dissolution range which should
22 specify that it is maybe the physiologically
23 relevant dissolution range. It wasn't clear if
24 this was an in vitro dissolution test that you were
25 trying to be consistent with, but what is more
237
1 important is that it is physiologically relevant.
2 So, with that in mind, my reaction is more positive
3 to the recommendation. However, my question still
4 relies on what about acids? Maybe you have weak
5 acids that are very poorly soluble at pH 7. Maybe
6 it is not relevant physiologically but I would like
7 you to address that. Are there any drugs or any
8 properties of drugs that are going to be of
9 concern?
10 DR. AMIDON: For borderline drugs? There
11 are a few NSAIDs. There may be. I think that is a
12 good example. What might this impact? I think it
13 is only a few drugs that it might actually impact.
14 I think that is a good point. I think our goal
15 here is to get the general view. We will come back
16 with the evidence in the future, and we are
17 interested in the type of evidence that the
18 committee thinks would be supportive of a positive
19 answer to this question one. For what types of
20 drugs would this have an impact?
21 I think I would agree. Harmonization is a
22 secondary issue. The question is reflecting the
23 physiological process and having a valid BE type
24 dissolution. So, I agree completely. On the other
25 hand, other things being equal, we would want to
238
1 harmonize rather than disharmonize--other things
2 being equal.
3 Ultimately, it is dissolution that counts,
4 not solubility. Our dissolution standard is based
5 currently on 6.8. So, dissolution is what counts.
6 Solubility is one of the factors determining the
7 dissolution rate but the dissolution rate is what
8 counts.
9 DR. HUSSAIN: One point that I think needs
10 to be kept in mind is the initial introduction of
11 BCS was in SUPAC which covered all drugs. The BCS
12 guidance, though focused on methods for
13 classifying, focused on waivers of highly soluble,
14 highly permeable. So, I think that is the
15 disconnect that we tend to see, that is, the range
16 of 1.2 to 7.5 is because it comes from the SUPAC
17 guidance and the rapid dissolution criteria that we
18 developed were for the BCS waiver guidance only.
19 So, that is how we will have to resolve that.
20 DR. LEE: Okay, so we have answered the
21 first question.
22 DR. JUSKO: I think we have resolved the
23 first question reasonably well. To summarize, I
24 think the answer to that is the inclination is to
25 have them both be determined at pH 6.8 but look
239
1 into the possibility that there may be unusual
2 circumstances where pH 7.4 would be particularly
3 relevant.
4 The second question is should the agency
5 expand the application of BCS-based biowaivers to
6 rapidly dissolving, immediate release products of
7 BCS Class III drugs? That question is a more
8 profound one and appears to be connected directly
9 to the list of proposed studies and data collection
10 efforts to test the hypothesis that this is
11 achievable, and it would be good to look again at
12 one of the slides from Dr. Yu. That one.
13 [Slide]
14 Clearly, it is premature that anyone go
15 directly to implementing this type of policy, and I
16 think we are at a stage where the committee is
17 probably recommending that a number of studies be
18 done to investigate and confirm that this is a
19 reasonable thing to do. This list of studies was
20 proposed and I would welcome comments from other
21 people on the committee.
22 DR. SHARGEL: One, it does strike me as
23 being a reasonable approach. I think, if I
24 understand this correctly, the premise is that
25 these drugs would rapidly dissolve and would be
240
1 very similar to giving it as a solution almost for
2 the time spent in the gastrointestinal tract. So,
3 the issue then becomes if you have a solution of
4 the drug would the excipients in a solid dosage
5 form make any difference in the permeability realm.
6 That is the issue I think as to make this a
7 universal kind of approach.
8 DR. YU: That is correct.
9 DR. HUSSAIN: I just want to make sure
10 that you are not committing to do those studies
11 with our money. We will take this recommendation
12 to PQRI and have industry do those studies.
13 [Laughter]
14 DR. JUSKO: With all the money that Pfizer
15 has saved, I am sure they are going to be the ones
16 to fund it.
17 [Laughter]
18 DR. COOK: That is how I got my salary all
19 the way up to $20,000 a year!
20 DR. LEE: Well, I think it is a serious
21 question and I think underlying this is the meaning
22 of permeability. I think I have heard repeatedly
23 throughout the day that while we are very
24 comfortable with dissolution solubilities being
25 unambiguous, when it comes to permeability that is
241
1 not so. Since someone else is going to pay for it,
2 we may as well address this issue more seriously.
3 What do we mean by permeability?
4 DR. YU: Yes, for BCS Class III drugs we
5 will collect a number of drugs and cover a wider
6 span of permeability. From there we will answer
7 some of the questions and some of the concerns with
8 respect to BCS biowaiver for Class III drugs. For
9 example, with internal studies we are proposing
10 intermediate permeability. Once we have the data,
11 I think the data will tell us which direction we
12 should go in. Thank you.
13 DR. HUSSAIN: I think one sort of point
14 that we would consider, I think is Hans Lennernas
15 has published on water, a glass of water. Water
16 has a permeability value which is fairly close to
17 metoprolol. It so happens that the permeability of
18 water itself is at the boundary. So, that has an
19 implication that when you give a glass of water and
20 a solid drug after an all-night fast, the glass of
21 water might get absorbed more quickly than the drug
22 has time to dissolve. I think we can bring that as
23 a sort of research question and address some of
24 that; some of the work that Gordon has done with
25 perfusion studies, and so forth, and what
242
1 implication that has.
2 DR. LEE: Yes, Larry?
3 DR. LESKO: If we look at that slide as a
4 way forward in anticipation of bringing results
5 back to the committee in the future, to get back to
6 the specific question about biowaivers, I wonder if
7 the committee members would have any thoughts on
8 what they would expect to see from these studies.
9 In other words, let's say I go out and I do a
10 comparative study of a solution versus these dosage
11 forms, would it be important to demonstrate strict
12 bioequivalence based on the 90 percent CI of 80 to
13 125? Would it be satisfactory to deal with the
14 point estimate? These are important considerations
15 in terms of designing and powering these studies to
16 address the question that we have. So, I wonder if
17 anyone has any thought on that.
18 The other part of this question is how we
19 select the solid dosage forms. Is there any advice
20 that committee members could give on the
21 identification of particular excipients that would
22 come to the forefront of people's mind that would
23 be worthwhile considering as part of the selection
24 process for the dosage forms. So, let's say that
25 we do come back in a year or something like that
243
1 and have data, we don't miss something that may be
2 particularly important in terms of potential
3 excipient effects.
4 DR. SHARGEL: Somehow, Larry, I am
5 compelled to talk about 90 confidence intervals and
6 bioequivalence. So, if you do the study I would
7 expect the same criteria would be held up.
8 MR. VENITZ: I would second that.
9 DR. BOEHLERT: I don't have a list of
10 excipients that you should be looking for, but I
11 certainly think that should be one thing you should
12 consider in doing these studies because, you know,
13 you keep repeating that excipients can have an
14 effect on oral absorption and I would like to
15 understand that better, where and how, so we could
16 begin to identify which excipients might be
17 problematic.
18 DR. LEE: Lawrence, have you shown us
19 those ten mono drugs? Did you provide a list?
20 DR. YU: Well, this is just the 12
21 potential BCS class drugs. We will come back with
22 some other drugs which are potentially Class III
23 drugs. That doesn't necessarily mean we will study
24 all ten. Maybe some data is already available from
25 NDAs and ANDAs.
244
1 DR. HUSSAIN: I think we have done two
2 studies, cimetidine and ranitidine, as Larry
3 pointed out. So, we have a good database on that
4 with manufacturing changes and dissolution changes
5 on two of those already. So, one could look at a
6 range of permeability values that could be selected
7 to account for that. At the end of the experiments
8 I think one aspect might be that you might need an
9 intermediate class of permeability because right
10 now you are going from 0-90, and I think when you
11 start going down to 20 and 30 percent, that is
12 where you start having problems. So, a range of
13 permeability values will help us maybe define and
14 intermediate permeability class.
15 DR. KIBBE: Is there less concern for a
16 company who decides to change the site of
17 manufacture from point A to point B and saying,
18 okay, it is a Class III and I am just going to show
19 you that I have the same dissolution
20 characteristics because I have just transferred my
21 process than with a second company who has a new
22 formulation and wants to do a biostudy? Would that
23 delineation help us move Class III's where we could
24 waive it in one case and not necessarily in
25 another?
245
1 DR. HUSSAIN: Well, I think SUPAC scale-up
2 and post-approval change actually did that. It
3 brought a risk-based approach or three-tier
4 approach for that. For example, for site changes
5 alone with no other changes, for a immediate
6 release dosage form it is qualification based on
7 dissolution alone. If you have other types of
8 changes, BCS comes in when there are excipient
9 changes, and so forth.
10 DR. RODRIGUEZ-HORNEDO: My observation is
11 that most of these compounds are weakly basic.
12 Right? Almost all of them?
13 DR. LESKO: Hydrochlorothiazide is a weak
14 acid, I believe.
15 DR. RODRIGUEZ-HORNEDO: Yes. Most of them
16 are weakly basic, and I am coming back to that
17 issue of pH dependence on solubility. I know it is
18 not the main issue here with the permeability but
19 maybe something that hasn't been addressed is the
20 pH dependence of the permeability. Is that of
21 concern?
22 DR. COOK: I don't know if this list was
23 proposed to take the ten drugs from. I think we
24 could take it back. We want to look at acids and
25 bases, and we want to look at a range of
246
1 permeability that probably even exceeds what we
2 have here to provide the best data. So, I don't
3 think I would get too hung up in saying that these
4 are the model compounds that one would use. It is
5 better to use a broader range that encompasses more
6 things so we will have more confidence in the
7 results.
8 DR. AMIDON: That is a good question about
9 pH dependence. The pH 6.5 with the perfusing
10 system that we use in humans provides a reference
11 permeability, kind of like a thermodynamic PK; it
12 is not really what is going on in solution but it
13 is what you use to move ahead. So, we measure this
14 reference PK. We have done permeability studies in
15 humans with alpha methyldopa a long time ago, and
16 that is pH dependent. It parallels that in
17 animals, and there is a variety of reasons for that
18 pH dependence. From the point of view of
19 predicting drug absorption and drug absorption
20 variability, it would be very important. So, I
21 would want to know that as a development scientist.
22 I don't see how it would help in a regulatory
23 classification or decision-making process. We take
24 the mean pH of about 6.5 for the human intestine
25 and say, okay, we are going to use that as our
247
1 reference value and stay with that. It gets to
2 cumbersome otherwise.
3 But for some of these drugs, I know
4 because we have studied hydrochlorothiazide, they
5 are very pH dependent, and we have also done
6 furosemide. So, the actual operative permeability
7 of pH 6.5, the permeability decreases there greatly
8 because it is ionizing. It is probably absorbed.
9 It has a very sharp absorption window because it is
10 the permeability, solubility procedure that counts.
11 Solubility is going up, permeability is going down.
12 I think that is why it is a highly variable drug.
13 It is not bioequivalent to itself, at least in one
14 study, because of the variability so we are getting
15 into problem drugs here--I should say variable
16 drugs. I am interested in the pH dependence, but I
17 can't justify it on the basis of regulatory use.
18 DR. LEE: It seems to me that this is an
19 ideal situation for forming a subcommittee to work
20 with Lawrence to just design a study. Right? The
21 choice of drugs, excipients, in vivo, in vitro,
22 other kind of parameters.
23 DR. YU: That is an excellent suggestion,
24 yes.
25 DR. JUSKO: I think it would also be good
248
1 to keep in mind making maximum use of complementary
2 information, like structure activity types of
3 predictions, as well as the data gathered from
4 animal studies so that one has more than one
5 measurement to base any anticipated results on.
6 DR. YU: This comes to my favorite topic,
7 my true research interest is in the structure
8 activity relationships. As long as my boss says
9 okay, do it, we will do it. Definitely.
10 DR. LEE: I thought you were going to say
11 you would do simulation studies.
12 DR. YU: Yes, we will do simulation
13 studies.
14 DR. LEE: Maybe that is the place to
15 start.
16 DR. LESKO: I want to get to the proposed
17 research because it is such a key to moving
18 forward. One of my concerns, and maybe I will ask
19 Lawrence to comment on this, is what is the
20 possibility or probability that you will be able to
21 find two solid dosage forms of these Class III
22 drugs that meet the rapid dissolution
23 characteristics that are being proposed for it? Is
24 this a study that is sort of Jack Cook's blue sky,
25 or is this a study where you can actually go into
249
1 the marketplace and find these things, or is it a
2 set of studies where you would actually have to
3 formulate the products to meet the rapid
4 dissolution criteria, or all of the above?
5 DR. COOK: Larry, would you consider a
6 solution versus tablet sufficient? That way, I
7 only need to compare those two rather than two
8 solid formulations?
9 DR. LESKO: Well, let's say we are doing
10 two tablets, but as I understand this research, if
11 you are going to go into the marketplace to find
12 those solid dosage forms, tablets, whatever, they
13 aren't necessarily formulated to be rapid
14 dissolution.
15 DR. COOK: That is why I was suggesting a
16 solution which is, for a highly soluble compound, a
17 lot easier to formulate and compare that to a
18 tablet. So, you have one that is extremely rapidly
19 dissolving, the solution, and then the tablet and
20 you can probably look at the excipients in that as
21 well.
22 DR. LESKO: So the tablet would be rapid
23 dissolution as well, 15 minutes?
24 DR. COOK: Well, it would have to be 15 or
25 30 minutes, whatever we propose. So, you would
250
1 have to make one formulation, is what I am saying,
2 rather than two.
3 DR. LESKO: I think actually that would be
4 a good idea because you are talking about ten drugs
5 with a comparative study, which is no less than
6 what we have for the original fasting study,
7 bioequivalence studies. In fact, it would exceed
8 it I think in terms of the total in vivo data to
9 support a biowaiver. But, again, that question
10 about what is the drug and what is the formulation,
11 and whether they are commercially available or not,
12 would be a limiting factor.
13 DR. YU: Certainly, I think we need to be
14 flexible, and we have limited research dollars. If
15 it is available on the market we will supply them
16 for the studies. That is the value of having a
17 subcommittee under the ACPS to get advice from the
18 members to see how best to utilize the money to get
19 the information we can get.
20 Secondly, we certainly want to utilize
21 what is out there in the literature and what is out
22 there in the NDAs and ANDAs. From there, we would
23 design--we only can conduct what is necessary to
24 address issues from those studies in NDAs or ANDAs
25 which we are not able to address.
251
1 DR. KIBBE: Larry, why can't you go to the
2 data the FDA already has and get the dissolution
3 profiles of all these products to start with? I
4 think it might be a little bit better if there were
5 two products out there that would give you relative
6 rapid dissolution. I think you would be better off
7 looking at them, and using as a fall-back a
8 procedure that isn't on the market, a solution.
9 DR. LESKO: Yes, I think the missing link
10 there is the dissolution studies that would not
11 necessarily be available in an application--
12 DR. KIBBE: Why not?
13 DR. LESKO: Well, because we are talking
14 about a very specific set of dissolution test
15 conditions to test a hypothesis of Class III.
16 Those dissolution conditions may not have been
17 studied as part of the normal drug development.
18 So, you can't just go back to the applications and
19 pull that information out. In almost all cases you
20 have to go to a laboratory and redo that to the
21 specifications that you would like to support the
22 hypothesis. But that is doable. I mean, that is
23 just reality; you just have to do it.
24 DR. YU: Absolutely. We actually
25 conducted a food effect study which was presented
252
1 this morning. When we selected a drug we purchased
2 the products and we did a lot of in vitro testing
3 before we selected these two specific products for
4 in vivo studies. It is doable and we have the
5 facility to do that within the agency.
6 DR. MEYER: It seems to me though that one
7 of the pieces of rationale I heard was that Class I
8 and Class III act like solutions. So, if we did
9 studies for low permeability drugs, solution versus
10 a marketed or experimental tablet, what-have-you,
11 just that two-way crossover, you would in a sense
12 prove whether the low permeability--while we know
13 it dissolves rapidly--also is sufficiently
14 permeable or permeability isn't a factor. So, that
15 seems to be a rational way of approaching it given
16 your initial hypothesis, solution versus tablet.
17 DR. YU: You are right. You are
18 absolutely correct, yes.
19 DR. MEYER: Can I raise one other
20 question? Just to kind of support the concept of I
21 think we still need to look at low permeability,
22 and that is that study that Ajaz had in his handout
23 from UT, ranitidine, sorbitol sucrose and
24 metoprolol, sorbitol sucrose, both solutions. And,
25 the metoprolol, which is highly permeable or
253
1 borderline high, had a confidence interval, sucrose
2 solution sorbitol 86-100 for AUC so it was
3 essentially bioequivalent, unchanged by sorbitol.
4 Whereas, ranitidine, which is low permeability,
5 dropped to 62 percent. So, the effect of sorbitol
6 was much greater on the low permeability ranitidine
7 than it was on the high permeability metoprolol.
8 So, we do have to worry about excipient effects.
9 Maybe this is the worst excipient known to man and
10 that is biasing our information, but maybe it isn't
11 so I think we still need to look closely at that.
12 DR. HUSSAIN: I think we would need to but
13 I think I would go back to what Gordon suggested in
14 a sense, for a solid oral dosage form it is the
15 dose of the excipient that is important. When you
16 think of a syrup you are looking at a tablespoonful
17 or two tablespoonfuls so sorbitol in a solution is
18 a much larger dose and a tablet is a much smaller
19 dose. So, that also I think is an issue that
20 should be considered. So. But I think Ian Wilding
21 has done the work with chewable tablets with
22 cimetidine. So. So, two grams of sorbitol with
23 mannitol had a dramatic effect on cimetidine. So.
24 DR. AMIDON: It may relate to the water
25 reabsorption and the absorbable versus not
254
1 absorbable excipients, and it would inhibit water
2 absorption which would slow down cimetidine's
3 absorption and if the transit is also speeded up
4 you can come up with a good rationale for the
5 mechanistic reasons, which suggests that maybe you
6 should classify excipients in some way. I mean, if
7 the excipient is absorbed, it is gone at some
8 point. So, maybe it is low permeability or
9 non-absorbable excipients that may have a problem
10 so you can perhaps reduce the problem that way. I
11 don't know.
12 DR. HUSSAIN: I think we talked about that
13 and actually low permeability, highly soluble
14 excipients are the ones which gave problems. If I
15 go back to Ian's work, and Ian could comment on
16 that, he actually did an experiment--Ian, correct
17 me if I am wrong--where he started with equal
18 osmotic pressure between sucrose, pyrophosphate and
19 sorbitol and mannitol, and showed that initial
20 osmotic pressure essentially.
21 DR. WILDING: We were trying to produce
22 osmotically equivalent concentrations of sodium
23 acid pyrophosphate, mannitol, the intention being
24 to try to work out what the mechanism was. As
25 Gordon indicates, I am sure there are mixed
255
1 mechanisms going on in terms of how the excipients
2 have their effect, but I am sure it is the
3 non-absorbable excipients that will have the key
4 issue in this regard.
5 I was just wondering as you were talking,
6 the choice of excipients that you use in the
7 context of these studies is obviously going to be
8 important. I wonder how much of the work, as Vince
9 indicated, could be done by modeling in advance to
10 create the matrix which is then tested by the human
11 biostudies. So, in looking at drugs for different
12 fraction absorbed in terms of Class III, given the
13 excipients' different release rates, trying to
14 build some form of modeling for that which then
15 forms the basis on which the human biostudies are
16 done. Because what you might find, if you are not
17 careful, is that human biostudies might not provide
18 the answer to the questions, which would be a waste
19 of time, money and effort.
20 DR. HUSSAIN: To that effect in the sense
21 of we worked with Jim Pauley last two years to look
22 at CACO 2 in vitro permeability experiments as a
23 screen to try to identify, hopefully, excipients
24 which might be affecting the permeability of the
25 membrane itself. I think from the literature and
256
1 from what Ian and we have done, we know the
2 osmotics. So, we are essentially looking at
3 several mechanisms by which these excipients can
4 exert an effect. So the studies we do and the
5 models we select, if they are mechanistically based
6 and based so we can actually get a hypothesis and
7 test that, would be far more meaningful than
8 randomly selecting those excipients.
9 DR. YU: Actually, we have done some
10 mathematical modeling work to simulate Ajaz' study
11 done at the University of Tennessee, to look at how
12 excipients in this particular case, sorbitol five
13 grams that one tablet will have, to look at how the
14 sorbitol affects oral drug absorption of
15 ranitidine. We have really nice results.
16 Certainly, we also want to evaluate it in the low
17 dose. I think those study results will all be
18 valuable in the future for how to address some of
19 the concerns expressed here. Thank you.
20 DR. HUSSAIN: One example that you have in
21 your handout is from my presentation. The drug is
22 atenolol, the tablet with a solution, and the
23 tablet has twice the bioavailability than the
24 solution. There is about 750 mg of sorbitol in
25 that. So, you know that even 750 mg in a solution
257
1 can reduce bioavailability by 50 percent compared
2 to a solid tablet. So, I think the thing which is
3 exciting to me is the major mechanisms by which
4 excipients exert their effect. As that happens, we
5 actually happen a means of doing hypothesis-based
6 testing underpinned by mechanistic basis for this.
7 DR. LEE: In other words, the excipients
8 can no longer be considered as inert.
9 DR. HUSSAIN: I don't want to alarm people
10 with that. I think we have to be very pragmatic.
11 I think some excipients have effect but I think
12 overall in a solid dosage form I don't think there
13 is a major concern. So.
14 DR. YU: The majority are inactive and
15 some of them, like sorbitol, may have some
16 concerns, yes.
17 DR. ANDERSON: Aren't you talking about
18 molecular interactions which are pH dependent,
19 particularly with those things that have all those
20 OH groups on them?
21 DR. YU: For solubility or permeability?
22 What aspect?
23 DR. ANDERSON: Well, if the solubility of
24 the drug is pH dependent, that is, if it has the
25 nitrogen or carboxylic acid group in it, and you
258
1 have all the OH's on the other things, whatever you
2 call them, you are talking about molecular
3 interactions which are pH dependent. The pH really
4 affects even those things with the OH groups on
5 them because the OH groups are basic as well.
6 DR. COOK: I guess that is another way of
7 looking at how you are classifying how the active
8 adjuvants, to steal somebody else's classification,
9 interact because not only are we worried about that
10 but things that change the physiology, whether it
11 be something that changes the osmolarity or
12 something that interacts with the membrane itself.
13 I guess the investigation of excipients is even
14 broader than just the molecular interaction.
15 DR. LEE: Bill?
16 DR. JUSKO: It sounds like there has been
17 considerable and very fruitful discussion about the
18 issues relating to these proposed studies. My
19 view, and I believe the committee believes that
20 there is good possible potential for future
21 biowaiver for the Class III agents, but before that
22 is done a very careful assessment of many of these
23 basic questions needs to be done. It appears that
24 an ample data set needs to be collected, and many
25 questions related to the role of excipients remain
259
1 to be resolved. So, there is great encouragement
2 from the committee to continue along this line.
3 DR. LEE: Well put. Maybe a future
4 committee will hear these results. Are there other
5 issues to be brought forth before this group? We
6 have had a very fruitful day.
7 DR. HUSSAIN: One issue, and I don't want
8 to be caught again like with the highly soluble,
9 highly permeable drugs, is the food effect. If we
10 go with a waiver for Class III, I think the logic
11 we be that we have to consider the food effect
12 alongside because otherwise it doesn't make sense.
13 So.
14 DR. LEE: That is for the record.
15 DR. HUSSAIN: So, this should also expand
16 to the food effect too at the same time.
17 DR. YU: You are absolutely right. We
18 will probably begin to collect the coefficient of
19 valence for a number of drugs compared under
20 fasting conditions and under fed conditions to see
21 if the valence becomes bigger or smaller, and how
22 to address this concern that we had this morning.
23 Thank you.
24 DR. LEE: We began the day talking about
25 subcommittees and I think this is an excellent idea
260
1 for clinical pharmacology, and not put a spotlight
2 on clinical pharmacology but also may serve as a
3 catalyst for other changes in the committee. Then
4 we went on to talk about a very interesting issue
5 about food effect on Class I drugs. I think the
6 committee is not that comfortable. Well, the
7 answer seems to be obvious but we don't have enough
8 evidence to support our gut feeling.
9 This afternoon I think we got a very good
10 understanding about the BCS Class I, Class III. I
11 don't want to repeat what Bill Jusko just talked
12 about. He put it very succinctly what needs to be
13 done. I think that we are going to hear about the
14 results of this work in a few years time, but the
15 committee, or at least I would like to see the use
16 of computation as a way to guide the experimental
17 design, and also to think about this permeability
18 more carefully, especially when we are encountering
19 more drugs that require transporters for
20 absorption.
21 DR. HUSSAIN: Let me go back to the issue
22 of the food effect waiver because that is an
23 important issue and I think I want to stress the
24 logic of the situation being such that it doesn't
25 make sense not to give waiver for fed studies for
261
1 Class I rapidly dissolving when we give the waiver
2 for fasting studies. I just want to stress that
3 fact because I heard from Marv that he is in
4 agreement with that. I really would like to have a
5 position of the committee on that one. So.
6 DR. LEE: That is the position.
7 DR. HUSSAIN: What is the position?
8 DR. LEE: What you just said.
9 [Laughter]
10 DR. HUSSAIN: So, the committee agrees
11 with Marv and the logic prevails?
12 DR. LEE: Right. What I have seen today,
13 shall we revise the guidance, reminded me very much
14 about curriculum revision. Tomorrow we can forget
15 about biology more or less, and we will focus on
16 some physical chemical issues. So, we begin
17 tomorrow at 8:30. Please plan on staying the
18 entire day because we have a full agenda, I mean
19 the committee members. You can leave the stuff
20 here because it is safe.
21 [Whereupon, at 4:00 p.m., the proceedings
22 were recessed, to reconvene at 8:30 a.m.,
23 Wednesday, May 8, 2002.]
24 - - -