RTF - FDA by UTfR05p7

VIEWS: 14 PAGES: 281

									                                                     1
              UNITED STATES OF AMERICA

      DEPARTMENT OF HEALTH AND HUMAN SERVICES

                PUBLIC HEALTH SERVICE

            FOOD AND DRUG ADMINISTRATION

      CENTER FOR DRUG EVALUATION AND RESEARCH

         ENDOCRINOLOGIC AND METABOLIC DRUGS
                 ADVISORY COMMITTEE

                     MEETING # 68

                      WEDNESDAY,
                  NOVEMBER 19, 1997

                      + + + + +


            The meeting took place in the Embassy Rooms
I, II, and III, Bethesda Ramada Inn, 8400 Wisconsin
Avenue, Bethesda, MD, 20814 at 8:00 a.m., Robert S.
Sherwin, MD, Acting Chair, presiding.


                   MEMBERS PRESENT:

        Robert S. Sherwin, MD, Acting Chair
        Kathleen Reedy, Executive Secretary
              José Francisco Cara, MD
              Cathy W. Critchlow, PhD
                  Jules Hirsch, MD
           D. Roger Illingworth, MD, PhD
              Robert A. Kreisberg, MD
                 Robert Marcus, MD
                Mark E. Molitch, MD
                  Maria I. New, MD

          CONSUMER REPRESENTATIVE PRESENT:

                Jaime A. Davidson, MD

            FDA REPRESENTATIVES PRESENT:

              G. Alexander Fleming, MD
                  Solomon Sobel, MD
2
                                   3
SPONSOR REPRESENTATIVES PRESENT:

       Martin Edwards, MD
    Jannie Fuhlendorff, PhD
        Barry Reit, PhD
      Frederick Reno, PhD
     Poul Strange, MD, PhD

         ALSO PRESENT:

        Richard Carr, PhD
  Wayman Wendell Cheatham, MD
         Peter Damsbo, MD
    Gerald A. Faich, MD, MPH
Michael J. Fossler, PharmD, PhD
         Kurt Furberg, MD
      Kristian Hansen, PhD
            Dr. Nisben
        John Whisnant, MD
                                                         4
                       A G E N D A

MORNING SESSION            PAGE

Call to Order              Robert Sherwin 4
Meeting Statement          Kathleen Reedy 4
Introductions              7
Introductory Remarks       Alexander Fleming    8

NOVO NORDISK PRESENTATION

         Introduction:          Barry Reit 10
         Pharmacology:          Jannie Fuhlendorff
         14
         Preclinical Safety:    Frederick Reno 19
         Clinical Pharmacology and Efficacy:
                                Poul Strange    26
         Clinical Safety:       Martin Edwards 53

Committee Questions        67

FDA and Industry Interactive Discussion

    1. What are the theoretical or realized benefits
       or risks of this oral hypoglycemic agent's
                rapid onset and offset of activity?
                             Wendell Cheatham
      142
                             Peter Damsbo     147
AFTERNOON SESSION

FDA and Industry Interactive Discussion (cont)
    2. Do any significant efficacy issues
        remain?         175

    3.    Could the recommended dosing regimen be
          further optimized or tailored? 219

    4.     What is the significance of myocardial
           ischemic events observed in the comparative
           trials, and will the proposed Phase IV
           cardiovascular safety study adequately
           resolve this issue?   243
                                Dr. Gerry Faich 247
                                Dr. Kurt Furberg
         258

    5.    What are the demonstrated and potential
                                                     5
        roles of repaglinide therapy in the treat-
        ment of type 2 diabetes patients?    282

Discussion and Questions 285
Meeting Adjourned
                                                              6


 1                     P-R-O-C-E-E-D-I-N-G-S

 2                                                   8:04 a.m.

 3                  ACTING CHAIRMAN SHERWIN:   I'd like to

 4   welcome everyone.     I think we have a quorum here and

 5   we'll begin.     Hopefully, we have a few members that

 6   will be coming in probably in a few minutes.   We'll move

 7   along and introduce them when they come.
 8                  I'd like to welcome you to this meeting of

 9   the Endocrinologic and Metabolic Drugs Advisory

10   Committee of the FDA.    The drug that we will be focusing

11   on today is repaglinide from Novo Nordisk.

12                  I'd like to begin by asking Kathleen Reedy

13   to read the meeting statement which will be somewhat

14   long, I suspect, today.

15                  MS. REEDY:   Conflict of interest

16   statement for the Endocrinologic and Metabolic Drugs

17   Advisory Committee, November 19, 1997.

18                  The following announcement addresses the
19   issue of conflict of interest with regard to this

20   meeting and is made a part of the record to preclude

21   even the appearance of such at this meeting.

22                  Based on the submitted agenda for the

23   meeting and all financial interests reported by the

24   Committee participants, it has been determined that all
25   interests in firms regulated by the Center for Drug
                                                           7

 1   Evaluation and Research present no potential for a

 2   conflict of interest at this meeting with the following

 3   exceptions.

 4                 In accordance with 18 United States Code

 5   208(b)(3), full waivers have been granted to Dr. Mark

 6   Molitch and Dr. Robert Sherwin.   A copy of these waiver

 7   statements may be obtained by submitting a written

 8   request to the Agency's Freedom of Information Office,

 9   Room 12A30 of the Parklawn Building.

10                 We would also like to note that Dr. Jaime

11   Davidson is excluded from participating in the

12   meeting's discussions and vote regarding Prandin.

13   Further, we would like to disclose that Dr. Robert

14   Marcus' employer, Stanford University, has an interest

15   in Eli Lilly, the manufacturer of several competing

16   products to Prandin which is unrelated to the firm's

17   competing products.    Although this interest does not

18   constitute a financial interest in the particular

19   matter within the meaning of 18 United States Code 208,

20   it could create an appearance of a conflict.   However,

21   it has been determined notwithstanding this interest,

22   that it is in the Agency's best interest to have Dr.

23   Marcus participate in all official matters concerning

24   Prandin.

25                 In the event that the discussions involve
26   any other products or firms not already on the agenda
                                                              8

 1   for which an FDA participant has a financial interest,

 2   the participants are aware of the need to exclude

 3   themselves from such involvement and their exclusion

 4   will be noted for the record.

 5                 With respect to all other participants, we

 6   ask in the interest in fairness that they address any

 7   current or previous financial involvement with any firm

 8   whose products they may wish to comment upon.

 9                 ACTING CHAIRMAN SHERWIN:   Now, normally

10   at this point we have an open hearing and we entertain

11   any statement from the audience regarding the product

12   we're dealing with.    Now, no one has come forth today

13   for any statement from the public and I would entertain

14   any right now.

15                 If not, we will move ahead.   Thank you.

16                 Well, although I've stalled enough, I

17   think what we'll do is begin by introducing the

18   Committee.    What I'll do is when those individuals who

19   will join us in a few minutes, I assume, I'll introduce

20   them as the time permits.   So, perhaps we should begin

21   with Dr. Hirsch who's behind the machine that I can

22   hardly see.

23                 Why don't you introduce yourself?

24                 DR. HIRSCH:   Jules Hirsch, Rockefeller

25   University, New York.
26                 ACTING CHAIRMAN SHERWIN:   Yes, please use
                                                                 9

 1   the microphone.

 2                 DR. HIRSCH:     Jules Hirsch, Rockefeller

 3   University, New York.

 4                 DR. ILLINGWORTH:     Good morning.    Roger

 5   Illingworth, Oregon Health Sciences University,

 6   Portland, Oregon.

 7                 DR. MARCUS:     Robert Marcus, Stanford

 8   University.

 9                 DR. CARA:    José Cara, Henry Ford Hospital,

10   Detroit.

11                 DR. CRITCHLOW:     Cathy Critchlow,

12   University of Washington, Seattle.

13                 ACTING CHAIRMAN SHERWIN:    Robert Sherwin,

14   Yale University.

15                 MS. REEDY:    Kathleen Reedy, Food and Drug

16   Administration.

17                 DR. KREISBERG:     Robert Kreisberg,

18   Birmingham, Alabama.

19                 DR. NEW:     Maria New, Cornell University

20   Medical College.

21                 DR. FLEMING:     Alexander Fleming in the

22   Division of Metabolic and Endocrine Drugs, FDA.

23                 ACTING CHAIRMAN SHERWIN:    Okay.     What I'd

24   like to do is begin with Dr. Fleming.       We'd like him

25   to begin with some introductory remarks.
26                 DR. FLEMING:     Good morning, ladies and
                                                           10

 1   gentlemen.   On behalf of Dr. Sobel and my colleagues

 2   at the FDA, we welcome you to this very important

 3   Advisory Committee meeting.   Today, we will discuss

 4   repaglinide, a very promising oral therapy for type 2

 5   diabetes.

 6                Repaglinide, like sulfonylureas causes

 7   insulin to be released from the beta cell.   But unlike

 8   sulfonylurea therapies, repaglinide has a very rapid

 9   onset and offset of action.   When taken immediately

10   before meals, repaglinide therefore results in insulin

11   secretory profiles that are more physiologic than

12   sustained insulin released induced by sulfonylureas.

13                Repaglinide's major promise is that it may

14   result in less serious hypoglycemia than longer-acting

15   agents.   Hypoglycemia, of course, is one of the major

16   limitations of therapy with the currently available

17   insulin secretagogues.   Because of the potential of

18   this drug to provide glycemic control with the

19   reduction in hypoglycemia, we ended a vision,

20   identified the repaglinide NDA for expedited review.

21   Since this is a new therapeutic approach, it is

22   important that the Committee examine the available data

23   as well as explore the ramifications of this approach.

24                I want to acknowledge the hard work,

25   particularly because this did involve a priority
26   review, of our primary reviewers:   Mike Fossler, John
                                                            11

 1   Gueriguian, Herman Rhee, Baldeo Taneja and Xavier

 2   Ysern, and our consultant from the cardiorenal

 3   division, Mary Ann Gordon and her colleagues; and

 4   finally, our project manager, Mike Johnston who is a

 5   very important force in managing our effort.

 6               I also want to thank the members of the

 7   Committee who continue to serve with distinction.

 8   Your willingness to add a third day to this meeting,

 9   occasioned by this expedited review, is an example of

10   your dedication.   Your participation is an extremely

11   important part in FDA's drug evaluation process.

12               After the company presents an overview and

13   we have an opportunity for general questions from the

14   Committee, all of us -- that is, the Committee, the

15   company and the Agency -- will then engage in

16   interactive discussions of several important issues.

17   Committee members, of course, are invited to ask

18   questions at any time, but they may want to defer

19   questions that pertain to one of the interactive

20   discussion points until we arrive at that point in the

21   agenda.

22               Once again, I want to thank you very much

23   for being here.    We look forward to the discussion

24   today.

25               ACTING CHAIRMAN SHERWIN:    Thank you.
26               I'd like to now introduce Mark Molitch from
                                                             12

 1   Northwestern who's joined us.

 2               I think we can go on with the presentation.

 3               Oh, Dr. Sobel, I almost forgot about you.

 4   Do you have anything you'd like to say?      You weren't

 5   on my schedule, so I wasn't excluding you.

 6               Okay.    Dr. Sobel just joined us.

 7               I'd like to begin then.    We are a half-hour

 8   ahead of time and perhaps we can keep moving along at

 9   that rapid pace.    So, I'd like to begin by having Barry

10   Reit from Novo Nordisk begin his presentation.

11               DR. REIT:     Good morning Dr. Sobel, Dr.

12   Fleming, FDA members, Dr. Sherwin, Advisory Committee

13   members, members of the press, colleagues and guests.

14   My name is Barry Reit.     I am vice president of

15   regulatory affairs at Novo Nordisk and I am here to open

16   the presentation of Prandin tablets, the first of a new

17   chemical class of compounds designed to lower prandial

18   glucose loads.

19               In 1984 and 1988, the ADA identified a need

20   in the choices of oral hypoglycemic agents.      They

21   stated in the Physicians' Guide to Type 2 Diabetes that

22   in general, older patients have more renal failure and

23   cardiovascular and hepatic problems as well as a

24   tendency to skip meals and snacks.     For this reason,

25   it is best to choose an agent with relatively short
26   duration of action which is less likely to cause
                                                            13

 1   profound hypoglycemia.     Again in 1994, the ADA

 2   expressed the fact that this need continued by stating

 3   that severe hypoglycemia is the major complication of

 4   sulfonylurea therapy.    Elderly patients as one

 5   subgroup are more susceptible to hypoglycemia,

 6   particularly when they have a tendency to skip meals

 7   or when renal function is impaired.    It is within this

 8   context that repaglinide    was developed for treatment

 9   of type 2 diabetes.

10                 Repaglinide, the active drug substance, is

11   the pure S enantiomer of a highly substituted benzoic

12   acid derivative, a new chemical entity.     It has

13   strongly pH dependent solubility and is highly

14   lipophilic.    It was discovered in 1986 by Dr. Karl

15   Thomae, a subsidiary of Beringer Ingolheim.     The drug

16   product is formulated from a spray dried granulate with

17   solubilizing agent and compressed into tablets of 0.5

18   one and two milligram strengths.      The tablets have a

19   pH independent dissolution profile at pH 1 to 7 with

20   a rapid disintegration and dissolution rate.

21                 The proposed indication and usage for

22   Prandin tablets is as an adjunct to diet and exercise

23   to lower blood glucose in patients with type 2 diabetes

24   mellitus whose hyperglycemia can not be controlled

25   satisfactorily by diet and exercise alone.     The dose
26   ranges from 0.5 to four milligrams taken with meals to
                                                            14

 1   regulate meal related prandial glucose load.

 2                 The US Clinical Development Program began

 3   in 1992 following submission of an IND.   An end of Phase

 4   II meeting was held in December 1994 at which time

 5   initial demonstration of efficacy was presented in US

 6   placebo control study 033.    Five one-year active

 7   control studies including US study 049 were initiated

 8   worldwide and a six month US placebo control safety

 9   study, 065, and definitive US dose response trial 064

10   were planned.    A pre-NDA meeting was held this past

11   January, followed by the NDA submission at the end of

12   June and granting of priority review in August.      The

13   safety update was submitted in October leading to

14   today's Advisory Committee meeting presentation.

15                 The remainder of our overview presentation

16   this morning will be made by Dr. Jannie Fuhlendorff who

17   will discuss pharmacology.    Dr. Frederick Reno will

18   present the preclinical safety section.     Dr. Poul

19   Strange will address clinical pharmacology and

20   efficacy.   Finally, Dr. Martin Edwards will discuss

21   clinical safety.

22                 Before I turn the presentation over to Dr.

23   Fuhlendorff, I want to take a minute on behalf of my

24   colleagues at Novo Nordisk and Beringer Ingolheim to

25   thank the Agency for inviting us here to discuss
26   Prandin.    Additionally, I want to thank Dr. Fleming,
                                                             15

 1   Michael Johnston, the CSO, and all the FDA reviewers

 2   for their expeditious, supportive, and interactive

 3   participation throughout the review of the Prandin NDA.

 4   Open communication between Novo Nordisk and the FDA has

 5   been an essential part of the timely review of this NDA

 6   and preparation for this meeting.   We wish to thank you

 7   for all of your efforts.

 8                DR. FUHLENDORFF:    Thank you, Dr. Reit.

 9                Ladies and gentlemen, I'm pleased to

10   present to you this morning, the pharmacology of this

11   new drug for type 2 diabetes.

12                Various preclinical pharmacology studies

13   over several years have shown that repaglinide is a

14   potent insulin secretagogue.    Its mechanism of action

15   is via the ATP sensitive potassium channel and it does

16   not cause thymic exocytosis of insulin.     It has

17   distinct binding sites.    The insulin secretion is

18   glucose dependent and there's no secretion of insulin

19   at sera millimolar glucose.     In contrast to known

20   effects of sulfonylureas, repaglinide does not inhibit

21   proinsulin biosynthesis.

22                This first data slide shows the in vivo

23   potency by blood glucose lowering in normal fed rats

24   after oral dosing.   The y axis is the change in blood

25   glucose.   Repaglinide in blue is 13-fold more potent
26   than glyburide compared at the half maximal dose.      The
                                                          16

 1   clinical dose of repaglinide is indicated with a blue

 2   bar in the bottom and represents 2,320 micrograms per

 3   kilogram.

 4                The blood glucose lowering effect was

 5   studied further in fasted glucose loaded rats.    The y

 6   axis is the plasmic glucose in millimoles per liter.

 7   The glucose loads are three, two, one, and .5 grams per

 8   kilo glucose.   Please note that the repaglinide dose

 9   response is found over the range that includes clinical

10   doses and note that the glucose level plateau at about

11   three millimoles per liter or 45 milligrams per

12   deciliter in rats.

13                The dose response for repaglinide was also

14   demonstrated in fasted dogs as shown in this slide.

15   Again, the y axis is the blood glucose in millimoles

16   per liter.   There's a dose dependence decrease in blood

17   glucose in dogs from 10 to 1,000 micrograms per kilo

18   or one milligram per kilo.

19                The corresponding plasma insulin response

20   is shown in this slide and the doses are the same as

21   in the slides shown before.    Please note the shape of

22   the curve.   There's a fast onset and decay over two to

23   four hours after dosing.   The most efficacious dose for

24   insulin release is 300 micrograms per kilogram in this

25   model.
26                The pharmacological effects of our insulin
                                                                17

 1   release is also shown in a rat model of type 2 diabetes,

 2   the low dose streptosodazin model.     The y axis is the

 3   blood glucose in millimoles per liter.    Repaglinide in

 4   a dose of one milligram per kilogram is given a time

 5   serial and repaglinide decreased the blood glucose

 6   level from seven to four millimoles per liter 60 minutes

 7   after administration.     The right panel show the plasma

 8   insulin in picamoles per liter.    At the same time, the

 9   insulin level doubles.

10                  The next series of slides shows the in

11   vitro studies with this new chemical entity.        First

12   here, glucose dependent insulin secretion in perifused

13   mouse islets.    We compare equally potent doses at five

14   millimolar glucose and that is 14 nanomole of

15   repaglinide and 200 nanomoles of glyburide.         This

16   shows a glucose dependent response with repaglinide and

17   no secretion at sera millimolar glucose.

18                  Repaglinide has a distinct binding profile

19   in receptor binding studies.      We were able to

20   differentiate the sites in whole beta TC3 cells using

21   two radioligands:      first, radiolabeled repaglinide

22   and second, radiolabeled glyburide.      Further, we used

23   four compounds as pharmacological tools and they are

24   listed here.      In order to differentiate the sites,

25   three binding sites were identified.     First, a higher
26   phenesticized for repaglinide with KD of 3.6 nanomolar
                                                             18

 1   and lower affinity for glyburide.      This site is PPP

 2   insensitive.     This site corresponds to the in vivo

 3   potency.   The next two sites were PPP sensitive.      The

 4   functional significance of these two PPP sites is not

 5   known.

 6                  The next slide more clearly demonstrates

 7   the differences.     The IC50's, again in beta TC3 cells,

 8   are listed here.     Please notice this value.   The IC50

 9   for repaglinide on glyburide binding site is very high,

10   equal to low affinity.    We saw before that repaglinide

11   was 13-fold more potent than glyburide in vivo in rats.

12   Instead, the IC50 for the repaglinide binding sites

13   reflect the in vivo rank order of potency as seen here.

14   Sulfonylureas tolbutamide, gliclazide, and glipizide

15   inhibits biosynthesis of insulin at low glucose

16   concentration and might therefore exhaust the beta

17   cell.    There's no suppression biosynthetic activity

18   with repaglinide at low glucose concentrations.       This

19   is the inhibition of proinsulin biosynthesis with

20   sulfonylureas tolbutamide, gliclazide and glipizide

21   and there's no inhibition with repaglinide.

22                  The direct exocytosis insulin was examined

23   in patch clamped mouse beta cells.      Under these

24   conditions, there's no transport of potassium through

25   the ATP sensitive potassium channels and there's no
26   increase in intracellular calcium.     Sulfonylureas are
                                                              19

 1   able to release insulin by stimulation of direct

 2   exocytosis.     Two-thirds of the insulin is estimated to

 3   come from this route.     So, clinical relevant

 4   concentrations of glyburide, glipizide and tolbutamide

 5   cause direct exocytosis and that is contrary to what

 6   is found with repaglinide for which there's no

 7   exocytosis with hundred nanomolar to 5,000 nanomolar.

 8                  The direct exocytosis with the repaglinide

 9   and glyburide is indicated here and please focus on the

10   right panel.     On the y axis, the increase in

11   capacitance is indicated.     The glybentlomide or the

12   glyburide curve is this one and repaglinide      is in the

13   bottom here.     So, no direct exocytosis with

14   repaglinide.

15                  To conclude, on the preclinical

16   pharmacology, repaglinide is a potent insulin

17   secretagogue compared to OHAs in fasted dogs, normal

18   rats, fed, fasted or glucose loaded rats.      It acts

19   exclusively via the ATP sensitive potassium channel in

20   a tissue selective manner and does not cause direct

21   exocytosis of insulin.    Distinct binding sites exists.

22   Repaglinide caused glucose dependent insulin secretion

23   with no secretion at sera millimolar glucose.

24   Repaglinide acts without inhibition of proinsulin

25   biosynthesis.     Finally, it is without peripheral
26   effects or insulin synthesizing effects for which I did
                                                           20

 1   not show any data.

 2                 So, the pharmacological profile of

 3   repaglinide is a new chemical entity of benzoic acid

 4   derivative.   It's an oral insulin secretagogue with

 5   distinct binding sites in the beta cells.      There's no

 6   direct exocytosis and no suppressant of protein

 7   synthesis.

 8                 It's my pleasure now to turn the program

 9   to Dr. Fred Reno.    Please?

10                 DR. RENO:   Thank you, Jannie.

11                 Good morning.    I'd like to summarize for

12   you the rather extensive preclinical safety program

13   that's been conducted on repaglinide that involves both

14   safety pharmacology and toxicology.

15                 With regard to safety pharmacology,

16   approximately 16 studies have been performed to

17   evaluate the potential for repaglinide to have

18   unanticipated pharmacological effects in other organ

19   systems.   At clinically relevant exposure levels,

20   repaglinide failed to elicit any significant effects

21   on central nervous system, cardiovascular,

22   respiratory, gastrointestinal or smooth muscle

23   systems.

24                 Lagan binding assays such as possible

25   effects on the N and L calcium channels and potassium
26   channels revealed no inhibitory activity except for the
                                                            21

 1   effects on the ATP sensitive channels described by Dr.

 2   Fuhlendorff.     Increases were seen in diuresis and

 3   sodium excretion at single doses that are 100 times the

 4   proposed clinical regimen.     The multiple

 5   cardiovascular evaluations indicated that adverse

 6   effects have not been seen at intravenous doses of 1,000

 7   micrograms per kilogram.

 8                  An extensive program of acute and chronic

 9   toxicity studies has been performed including

10   carcinogenicity evaluation in two species and studies

11   evaluating the potential effects on all aspects of the

12   reproductive process.     Teratology studies have been

13   carried out in two species and a complete ICH compliant

14   genotoxicity evaluation was performed as well as

15   immunogenicity evaluations.

16                  Chronic toxicological evaluations in rats

17   and dogs have been performed at duration treatments of

18   up to one year.    In the rat, the no effect dose is 16

19   milligrams per kilogram which results in plasma

20   concentrations that are 38 to 85 times the human

21   exposure level.    At higher doses, alkaline phosphatase

22   levels are increased without histopathological

23   effects.   Dogs are sensitive to the hypoglycemic

24   effects of repaglinide which is responsible for most

25   of the effects in this species.    At 50 milligrams per
26   kilogram there were elevated hepatic enzymes with
                                                            22

 1   histological evidence of periportal enlargement with

 2   no evidence of hepatocyte degeneration.      Thus,

 3   compared to the human dose of 0.32 milligrams per

 4   kilogram per day, there are no clinically relevant

 5   laboratory or histopathological changes.

 6                The drug is not mutagenic in a battery of

 7   six genotoxicity studies.     Four immunogenicity

 8   studies have revealed no evidence of immunologic

 9   responses or allergic reactions.     In reproduction

10   studies, repaglinide failed to produce an effect on

11   fertility.   It is not teratogenic when administered to

12   rats and rabbits during the first trimester period of

13   organogenesis.   There is a developmental effect which

14   is seen when the drug is administered in late gestation

15   and early lactation.    I'll describe that in more detail

16   later.

17                Carcinogenicity studies have shown no

18   tumorigenic responses at doses that are more than 50

19   and 100 times the clinical exposure level in males and

20   females respectively.    I'll discuss that more later,

21   also.

22                In the reproduction findings, there are

23   limb deformations that are developed in the offspring

24   of females that are treated later, beginning with the

25   third trimester of gestation.     This was initially
26   observed in animals that were eight to ten weeks of age
                                                             23

 1   with an observation of altered ability to walk

 2   correctly.   It came about as a result of the behavioral

 3   evaluations that have been performed in these animals,

 4   an evaluation that is relatively new in preclinical

 5   development.

 6                  Subsequent studies have revealed that this

 7   effect is due to an altered structure of the limbs.

 8   Mechanistic studies that have been performed that have

 9   been designed to identify the specific period of effect

10   have shown that this effect does not occur if the

11   animals are treated in the first or second trimester,

12   and is limited to the third trimester of gestation and

13   the early period of nursing.     There is histological

14   evidence of chondromalacia and an inhibition of the end

15   growth of osteogenic buds.

16                  Glucose levels are significantly reduced

17   in maternal animals during this period of gestation and

18   studies have identified that the offspring also have

19   decreased glucose levels.     Studies have identified

20   that repaglinide can be transferred to the offspring

21   via milk as evidenced by the fact that cross-fostering

22   of offspring with untreated mothers also elicits this

23   effect.

24                  In summary, these are developmental

25   changes as opposed to teratogenic effects and they've
26   only been seen at doses that are significant multiples
                                                           24

 1   of the human exposure level and have not been seen at

 2   doses that are six times the human exposure level.    In

 3   the carcinogenicity evaluation, repaglinide was not

 4   tumorigenic in the mouse at exposure levels that run

 5   from 71 to 160 times, 169 times the human AUC in males

 6   and females.

 7                  This is a bar graph in the rat

 8   carcinogenicity study that describes the exposure

 9   margins for the four treatment groups of males and

10   females.   I call your attention here.     These numbers

11   at the top of the bar graph represent the multiples in

12   excess of the human AUC that resulted from the exposure

13   of animals at these four doses.   I point out to you that

14   in this study at these two doses here, the two lowest

15   doses, which represent 51 and in excess of 100 times

16   the human exposure level, there are no tumorigenic

17   effects.

18                  There is at the doses that result in 90 to

19   200 times the human AUC, an increase in benign thyroid

20   tumors in the males.     It's interesting to note that

21   these benign thyroid tumors were not seen in the females

22   even though the females' plasma concentrations were

23   significantly higher than those of the males.     At the

24   very highest dose only that results in a 200 fold margin

25   of the human AUC, there is an increase in the
26   spontaneous rate of benign liver tumors in these male
                                                          25

 1   animals.   It is again interesting to note that females

 2   who were exposed to higher plasma concentrations of

 3   repaglinide at that same dose, these tumors did not

 4   develop.   This tumor type spontaneously occurs in rats

 5   and in this study were only seen at an increased

 6   incidence.

 7                A study was done to elucidate the mechanism

 8   for the development of the thyroid tumors in the male

 9   rats.   It was identified through these studies is that

10   animals that are treated at those two higher dosage

11   levels develop a decrease in plasma T3 levels.     The

12   decrease in the plasma T3 levels result in increased

13   levels of TSH and that results in an enhanced

14   proliferation within the thyroid gland.     That

15   phenomenon of the increased TSH resulting in an

16   increase in proliferation is a known phenomenon that

17   has been seen with other drugs such as phenobarbital

18   and some of the phenothiazine antidepressants.     The

19   current state of knowledge would suggest that that

20   mechanism is not comparable to anything that is seen

21   in humans.   In the clinical program that will be

22   described later, there were no changes in T3 uptake,

23   T4 or TSH levels during the clinical program.

24                So, with regard to the conclusions from the

25   carcinogenicity evaluation, we can say that
26   repaglinide is not genotoxic.    That there is a high
                                                             26

 1   exposure safety margin within these studies.     That the

 2   development of the thyroid tumors is a mechanism that

 3   is specific for rats.    That the mouse carcinogenicity

 4   study is negative and the conclusion would be that there

 5   is no clinical risk as a result of this information.

 6                  With regard to non-clinical

 7   pharmacokinetics, repaglinide in all of the animal

 8   species study is rapidly absorbed with peak

 9   concentrations achieved in less than one hour.        The

10   drug is highly bound to plasma proteins exceeding 95

11   percent in all species examined.    That in rodents,

12   plasma levels in females are two to three times higher

13   than those seen in males and that is a situation that

14   is frequently seen in rodent studies.     The drug is

15   highly excreted by the bile with only eight percent of

16   radiolabeled repaglinide excreted in the urine.       The

17   drug is metabolized by glucuronidation and/or

18   oxidative pathways within the liver.    The metabolite

19   profile in the preclinical species are similar to those

20   seen in man.

21                  In conclusion, the preclinical safety

22   assessment of repaglinide has shown a favorable safety

23   profile with no suggestion of potential adverse

24   toxicity at clinically relevant doses.       That's

25   described on the enhancement of the slide that was shown
26   to you by Dr. Fuhlendorff.
                                                               27

 1                 Now I'd like to introduce to you Dr. Poul

 2   Strange who will discuss with you the clinical

 3   pharmacology and the clinical efficacy of repaglinide.

 4                 DR. STRANGE:   Thank you.

 5                 As in animals, repaglinide is rapidly

 6   absorbed and eliminated in man.   Depicted on this slide

 7   is a pharmacokinetic profile comparing oral solution

 8   with a tablet.    Note the Tmax at 45 minutes and the

 9   rapid elimination.    Note also that the tablet is

10   virtually identical to the oral solution profile

11   demonstrating the in vivo correlate of the rapid

12   dissolution of the repaglinide tablets.      The levels of

13   drug and plasma after these things are generally in the

14   level of 10 to 15 nanograms per ml.       So, it's rapidly

15   absorbed from the gastrointestinal tract.       Tmax is

16   unchanged by food.   There's a marginal decrease in AUC

17   with food.   It is rapidly eliminated from the

18   bloodstream with a half-life of one hour.       High

19   clearance, 38 liters per hour and other PK parameters

20   are listed below.

21                 Sixty percent of the plasma concentration

22   at any time point is parent compound.       There are no

23   chiral conversion in vivo.    Repaglinide is primarily

24   metabolized by a cytochrome P450, isoform 3A4.         None

25   of the metabolites contribute any significant
26   activity.    Ninety percent of the dose is excreted in
                                                            28

 1   the feces via biliary secretion.    The major metabolite

 2   found in feces is a dicarboxylic acid which is inactive.

 3   Eight percent is excreted in the urine as metabolites.

 4   Of those eight percent, less than one percent is parent

 5   compound.

 6                Already in early clinical studies in type

 7   2 diabetes patients, the rapid absorption and

 8   elimination of repaglinide was confirmed with clinical

 9   use.   In this study, patients were given meals at 8:00,

10   12:00 and 6:00 p.m.   This is really 8:00 in the morning.

11   What is seen is that the repaglinide profile shows a

12   rapid increase and a rapid decrease down to almost

13   baseline levels.

14                On the next slide, the simultaneous

15   insulin profiles are demonstrated.     In the left panel

16   for your reference is repeated the previous

17   pharmacokinetics slide.    On the right panel, the

18   insulin concentrations in plasma are shown.      The

19   dashed line here is the baseline value in response to

20   the three standardized meals.      The solid line is the

21   insulin response to the same standardized meals given

22   with the dose of repaglinide.    Note here that the peak

23   is increased and that the insulin secretion declines

24   down to the levels seen with the normal insulin response

25   to the meals in these type 2 patients.
26                On the next slide, I'll show you the
                                                          29

 1   simultaneous glucose profile.    And again, I've

 2   repeated the repaglinide PK profile and the insulin

 3   profiles for reference.   The glucose curves are shown

 4   here with the dashed line being the baseline value

 5   without repaglinide treatment and the solid line being

 6   the glucose concentrations with time with repaglinide

 7   treatment.   Note the substantial decrease in glucoses.

 8   The average 24 hour glucose in this trial decreased from

 9   about 190 milligrams per deciliter to about 130

10   milligrams per deciliter.

11                Subsequent to this, dose ranging and dose

12   tolerance trials investigating the dose range from .125

13   all the way to 20 milligrams preprandially to each meal

14   was investigated.   Based on those data, those response

15   trials was designed and I'll describe that in some

16   detail.   Patients with type 2 diabetes, either naive

17   to oral hypoglycemic therapy or previously treated with

18   oral hypoglycemic therapy went through a screening and

19   went through a two to three week stabilization period

20   without drug.   Patients that after that stabilization

21   period had fasting plasma morning glucoses between 180

22   and 300 were then randomized to either placebo or five

23   dose levels of repaglinide given preprandially with

24   each meal.   Doses were taken 15 minutes before each of

25   the three main meals.
26                Patients were confined to a hospital unit
                                                            30

 1   in weekly 58 hour stays, during which they received the

 2   same set of standardized meals at every of those two

 3   day visits.   The last 24 hours of the visit, 20-point

 4   repaglinide insulin and glucose profiles were

 5   determined.   The schematic outline of the slide can be

 6   viewed like this.   There is a screening and

 7   stabilization phase.   The hatched areas demonstrate

 8   the two days that patients were confined in the hospital

 9   every week.   During the second day of which the 24 hour

10   20-point profiles were determined.    Altogether, the

11   treatment went for 28 days or four weeks and patients

12   treated themselves in the periods between the hospital

13   visits.

14                 Now, on the next slide the repaglinide

15   profiles obtained in the study at week four is depicted.

16   Again, 8:00 in the morning -- the meals in this trial

17   were given at 8:00, at 1:00, and at 6:00 in the evening.

18   Note here the repaglinide -- again, the rapid

19   absorption with the peak and the rapid decline to almost

20   baseline levels for all doses except the four milligram

21   dose.   Specifically note that the nighttime values of

22   repaglinide almost zero for all patients except some

23   patients in the high dose.

24                 On the next slide, I'm going to show you

25   the simultaneous repaglinide profile, insulin profile
26   and glucose profile from one of the repaglinide doses.
                                                           31

 1   The dose chosen is the .5 doses depicted here in the

 2   dashed blue line.   So, just for reference here, again,

 3   the time axis here was 8:00, 12:00, 6:00.   In the solid

 4   black line is repaglinide levels.    All -- both

 5   repaglinide insulins and glucoses are plotted on the

 6   same numerical axis but obviously, with different

 7   units.

 8                The placebo control group is depicted in

 9   dashed lines with the glucose in blue and the insulin

10   in red.   Note to this, a small dose of repaglinide of

11   .5 milligrams preprandially to the meal, that they are

12   barely distinguishable in decreases in insulin

13   secretion over the placebo group.    Also note the

14   relatively slow decline of insulin to baseline levels

15   in the placebo group.

16   In contrast, we see substantial decreases in blood

17   glucoses of about 40 milligrams per deciliter or 50

18   milligrams per deciliter in the peaks.      Also note on

19   the glucose curve here, this hump which represents 100

20   kilocalorie evening snack which was not covered by a

21   repaglinide dose.

22                On the next slide, I'm going to show you

23   the same plot but for the full dose of four milligram

24   preprandially to three meals.    Repaglinide

25   concentrations are higher.   At this dose, the
26   increases in insulin secretion as measured in
                                                              32

 1   peripheral blood is more visible with the enhancement

 2   and the decrease down almost to the placebo group,

 3   control group levels, and the very substantial

 4   decreases in blood glucose of an average 80 milligrams

 5   per deciliter.

 6                On the next slide I'm going to show an

 7   average 24-hour glucoses by dose group as a function

 8   of time.   Here's another busy slide.    So, the axis on

 9   this slide is time on the X axis, baseline one, two,

10   three, four weeks treatment, and on the y axis, the

11   average 24-hour blood glucose.    Note first the placebo

12   group that remains stable throughout the trial.        Then

13   note that we see a dramatic and significant effect for

14   all doses tested already in one week.      It's also

15   evident that we see the vast majority of the total

16   effect within the first week for all doses except the

17   .25 milligram dose.     After three weeks, we hardly see

18   any increased response at all.

19                On the next slide I'm showing the classical

20   dose response curve at four weeks which is these data

21   in a different representation as a function of dose.

22   Placebo level at 240 milligrams per deciliter on

23   average for 24 hours.    We see the dose response through

24   all the doses tested and the magnitude of the effect

25   is about 80 milligrams per deciliter for the four
26   milligram dose.
                                                           33

 1                The exposures of repaglinide as measured

 2   AUC repaglinide observed in this study is highly

 3   variable.   On this slide is depicted the dose on the

 4   x axis and the AUCs on the y axis.   We've plotted the

 5   minimum and the maximum values and the first and the

 6   third quartile with each dose given.   Note the highly

 7   variable plasma levels that spans 100-fold range.

 8   Also note, the highest exposures attained with this

 9   expected normal clinical use of three doses

10   preprandially are of about 830.

11                On the next slide I have repeated this

12   panel on the left for reference.   On the right, I have

13   plotted the exposures observed in a dose tolerance

14   trial where patients were dosed all the way up to 20

15   milligrams preprandially to four meals.   The line here

16   is, again, the line of the highest exposures expected

17   in normal clinical use or most widespread clinical use.

18   We have experience with exposures of repaglinide ten

19   times higher than that dose all the way up to 11,000.

20   Notably, the treatment was safe at these doses.

21                We have done special population

22   investigations comparing 12 young, healthy patients to

23   12 elderly, healthy patients.   The mean and the range

24   of the AUC is attained essentially the same,

25   demonstrating that age as an independent factor does
26   not influence repaglinide pharmacokinetics.    We have
                                                            34

 1   done trials with liver dysfunction comparing 12 healthy

 2   subjects to 12 patients with severe liver disease of

 3   Child Pugh Scale B and C -- grade B and C it's called

 4   -- and as expected from the metabolism and the biliary

 5   secretion of the drug, we do see increases in the

 6   exposures observed in these patients suggesting that

 7   careful titration in patients with liver disease may

 8   be warranted.

 9                  We have also done renal dysfunction study

10   comparing six healthy subjects to six patients with

11   mild/moderate disease and six patients with severe

12   renal disease with creatinine clearances less than 25.

13   Please note first here that the levels attained in the

14   normal control group for some reason turned out lower

15   than we've seen in other trials with normal controls.

16   Irrespective of that, a little unexpected thing that

17   happened in this trial, we do see increases in AUC both

18   in the mean and the range of AUCs with renal

19   dysfunction.

20                  I'm going to show you a little more detail

21   on those things showing the correlation between the

22   creatinine clearance and the exposures attained in

23   these patients.     So, on this slide on the x axis,

24   creatinine clearance is depicted and on the y axis the

25   AUC is repaglinized.     The normal group has high
26   creatinine clearances and low levels of drug in their
                                                            35

 1   blood.   The mild/moderate renal dysfunction have for

 2   five out of the six patients essentially decreased

 3   creatinine clearance, obviously.    Essentially, the

 4   same levels of repaglinide with the exception of one

 5   outlier.

 6                Now, upon scrutiny, this outlier turned

 7   out to have a history of hepatic disease suggesting that

 8   this patient may more fit in the hepatic impairment

 9   group than really, primarily the renal impairment

10   group.   For patients with severe renal dysfunction

11   with creatinine clearances less than 25, we do see

12   increases in AUCs, but they're well within the level

13   or the range of exposures that we're experienced with

14   and that appears to be safe.

15                We've done drug interaction studies with

16   free compounds with very a low safety margin with

17   digoxin, warfarin and theophylline.    Repaglinide did

18   not influence any of these three drugs' pharmacokinetic

19   profile indicating that those adjustments of these

20   drugs are not necessary when instituting repaglinide

21   therapy.   For warfarin, we've also looked at the

22   dynamics of warfarin and there was no effect on the

23   dynamics either.   We've done an interaction trial with

24   cimetidine because of its inhibition of gastric acid

25   secretion may interfere with repaglinide absorption.
26   Also because cimetidine is known to inhibit several
                                                             36

 1   liver enzyme systems.     There were no influences of

 2   repaglinide on the repaglinide pharmacokinetic

 3   profile.

 4                  So, before we turn to efficacy, I'd like

 5   to summarize the drug profile so far.      We have a new

 6   chemical entity.     It's a potent oral insulin

 7   secretagogue with a distinct beta cell binding profile.

 8   It does not induce direct exocytosis of insulin from

 9   beta cells.    It does not suppress protein biosynthesis

10   in beta cells.     It's not neurogenic, photogenic or

11   carcinogenic, and there are no clinically relevant

12   preclinical safety findings.                For the

13   clinical pharmacology profile, we have a rapid onset

14   with a Tmax of .7 hours, rapid plasmic clearance.       It

15   enhances insulin responses to meals.      It results in

16   clinically significant blood glucose responses.       It is

17   effective in doses from .5 milligrams preprandially

18   with meals.    It is highly variable.   It is excreted by

19   the bile.     There are no significant interactions with

20   either digoxin, warfarin, theophylline, or cimetidine

21   and no dose adjustments appears to be required, only

22   for patients with liver dysfunction.    With this, I will

23   turn to demonstration of efficacy.

24                  Efficacy is best demonstrated in three US

25   trials, placebo controlled US trials, summarized on
26   this slide.    We have titration format trial.   Patients
                                                           37

 1   were titrated from .25 to eight milligrams, or

 2   titration range .25 to eight milligrams --    obviously,

 3   some patients start below that -- of 18 weeks' duration

 4   with 66 patients on repaglinide.    As I just described,

 5   we have a dose response trial with fixed dosing

 6   exploring the range .25 to four milligrams

 7   preprandially, four weeks' duration with 120 patients

 8   treated with repaglinide.   We have the largest placebo

 9   control trial, 65 which explored doses one and four

10   milligram with three meals of half a year's duration

11   representing 289 patients on repaglinide, totalling

12   almost 500 patients on repaglinide in placebo

13   controlled trials.

14                The first trial I'll demonstrate is the

15   titration format trial, the design of which were that

16   patients were screened either OHA naive patients or

17   patients previously treated with oral hyperglycemic

18   agents.   Went through a stabilization period without

19   any drug, after which they were randomized to receive

20   either placebo or repaglinide.     They went through a

21   titration period in which they were titrated through

22   doses .25, .5, one, two, four and eight milligrams

23   preprandially to three meals.      Patients who did not

24   achieve an increased effect of eight milligrams over

25   the four milligram dose were back titrated to four
26   milligrams before the start of the maintenance phase
                                                             38

 1   of the study.

 2               At this point, patients were not titrated

 3   further and remained on that determined optimal dose

 4   for the rest of the study.   As seen on the trial, there

 5   is good effect of repaglinide and the placebo groups,

 6   glucose controlled, deteriorates as expected when

 7   patients on therapy basically stop that therapy.       It

 8   results in a difference between the two treatments at

 9   the end of the study of 1.7 percentage points, HbA1c,

10   a very significant and clinically relevant effect of

11   repaglinide therapy.

12               I'll repeat the dose response curve to

13   demonstrate that these results are consistent with the

14   average BG mean decreases we saw through the doses

15   tested in the dose response curves of bringing the blood

16   glucoses from about 240 to 160 on average for the

17   highest dose.

18               This slide demonstrates the HbA1c response

19   over time for the largest placebo control trial with

20   almost 300 patients on repaglinide.     Patients with

21   type 2 diabetes, if they were OHA or all hypoglycemic

22   agent naive, the requirement was that their HbA1c should

23   be above 6.5.   If they had been previously treated with

24   oral hypoglycemic agents, the requirement was their

25   HbA1c should be less than 12.    Those patients went
26   through a stabilization phase of two to three weeks'
                                                              39

 1   duration during which they didn't take oral

 2   hypoglycemic agents.      They were then randomized to

 3   receive either placebo in the black solid line, or

 4   repaglinide one milligram in the green dashed line, or

 5   repaglinide four milligram in the red dashed line.      The

 6   resulting sustained effect on glucose control after six

 7   months or the separation after six months of therapy

 8   here is 1.8 percentage point, HbA1c, again demonstrating

 9   the efficacy of repaglinide on glucose control.

10               On the next slide, I'll show you the subset

11   of patients in this trial that were naive to oral

12   hypoglycemic therapy.      The placebo group of naive

13   patients deteriorated somewhat while the active

14   treatment arms had substantial effects on glucose

15   control.   With separation between the difference

16   between the placebo, no treatment group and the four

17   millimeter group was 2.9 percentage points on average

18   for the patients.

19               On the next slide, I will show in absolute

20   numbers where patients in this trial ended up by dose

21   as a function of HbA1c.    So, on the x axis here, we see

22   the HbA1c at the end of the trial.     On the y axis, we

23   see the cumulative frequency by dose group.    Note here

24   that if we look at this point, in the placebo group,

25   half the patients ended up with HbA1c's above 10.        In
26   the one milligram group, half the patients ended up with
                                                          40

 1   HbA1c's less than 8.3.    In the four milligram group,

 2   half the patients ended up with glucose control of HbA1c

 3   less than 7.8.

 4                We saw in fasting morning plasma glucoses

 5   were consistent with these data with somewhat

 6   deterioration in the placebo group -- I'm now back to

 7   talking about all patients, I should say -- somewhat

 8   deterioration in placebo group of 20 milligrams per

 9   deciliter.   The effect on the axle arms of the study

10   are a decrease of 50 milligrams per deciliter in fasting

11   morning plasma glucose.

12                To study safety, one-year comparator

13   trials were done.   I'll just go through the trial

14   design of these safety studies in some detail.

15   Patients were either naive or previously treated with

16   oral hypoglycemic agents.    They were screened and went

17   directly from their therapy if they had one before into

18   either repaglinide or the comparitor.   This particular

19   trial I've shown you is the US trial in which glyburide

20   was used as a comparitor.

21                Patients then went through a titration to

22   fixed glucose control, meaning that we basically

23   titrated the drugs to equivalent glucose control.

24   When the beginning of the maintenance phase of 12th

25   month began no dose adjustments were possible should
26   glucose control deteriorate at that point.     So,
                                                             41

 1   importantly, they were titrated to fixed targets,

 2   titrated to equivalence, and then the dose was

 3   maintained.    The results of this study was, as expected

 4   from the design, that we see equivalent responses in

 5   HbA1c over time between repaglinide and the comparitor.

 6   Repaglinide is the solid line and that should have

 7   interest.

 8                 On the next slide is shown a summary of the

 9   glucose response data in all those five comparitor

10   trials.   As shown here on the y axis is difference

11   between repaglinide and the comparitor in the change

12   from baseline over time.   Now that means that if we have

13   a negative number here, it means that it is in favor

14   of repaglinide and a positive number is in favor of the

15   comparitor.    Just to summarize again, the 49 trial, as

16   I just showed you on the previous slide, has a

17   confidence interval that symmetrically around zero

18   demonstrating the equivalent effect in this particular

19   trial design on glucose control.     We also obtained

20   equivalents in other trials and one of the trials, the

21   glipizide comparison, turned out in favor of

22   repaglinide in terms of glucose control.

23                 The efficacy of repaglinide was confirmed

24   in a combination study with metformin.    In this trial,

25   metformin monotherapy failures inadequately treated
26   with metformin were randomized to either receive
                                                               42

 1   repaglinide as monotherapy, metformin as monotherapy,

 2   or the combination between the two drugs.       Patients

 3   went through a titration period and then were on fixed

 4   dosing for three months thereafter.        The effects on

 5   blood glucose control as measured by HbA1c is depicted

 6   on the next slide, showing as expected that in these

 7   metformin therapy failures, the metformin monotherapy

 8   remained essentially constant as does repaglinide

 9   monotherapy.     But in the combination arm of the study,

10   there are very substantial decreases from 8.7 down to

11   6.9 in glucose control demonstrating synergistic

12   effect of the two compounds.

13                  Note the end result here:    the average

14   glucose control less than seven and more than half the

15   patients -- and that's not shown on this slide -- had

16   glucose less than seven at this point.       Essentially,

17   those metformin failures were rescued by the add-on of

18   repaglinide therapy.

19                  So, in summary, the clinical pharmacology

20   profile of repaglinide is as follows:    rapid onset Tmax

21   within an hour; rapid plasma clearance; enhances

22   insulin responses to meals.     Repaglinide results in

23   clinically significant glucose responses.       It's

24   effective from .5 milligrams.     It's highly variable.

25   It is excreted by the bile.     There are no significant
26   drug interactions with either digoxin, warfarin,
                                                            43

 1   theophylline, or cimetidine.   Dose adjustments seem to

 2   be required only for liver dysfunction patients.

 3                The summary of the efficacy profile -- turn

 4   the discussion over to safety, the vast majority of the

 5   response within one week, 40 to 80 milligrams per

 6   deciliter on average.    Those response through the

 7   doses .5 to four milligrams preprandially with three

 8   meals.   Significant difference versus placebo

 9   repeated in both titration and fixed dose formats.

10   Very consistent results with overall same effects on

11   blood glucose.   It improves blood glucose, depending

12   on the trial and the subset, anywhere from 1.6 to 1.9

13   HbA1c on average.   There's a maintenance of glycemic

14   control for at least one year and there's a substantial

15   additive effect to metformin in the trial where the

16   metformin failures were actually rescued back into good

17   glucose control with repaglinide.

18                With this, I'd like to turn over the

19   presentation to Dr. Edwards who will present safety of

20   the compound.

21                ACTING CHAIRMAN SHERWIN:   I'm sorry.    Dr.

22   Cara would like to ask a couple of questions.     I had

23   hoped that we would go through, but that's okay.

24                DR. CARA:   Just while it's still fresh on

25   my mind, a couple of questions.
26                Do you know whether the binding sites for
                                                               44

 1   the repaglinide actually get down-regulated?       Does

 2   that have any clinical significance in terms of

 3   development of tolerance?

 4               DR. STRANGE:     Whether repaglinide

 5   receptors get down-regulated?

 6               DR. CARA:   Binding sites.

 7               DR. STRANGE:    I'm not the right person to

 8   answer that question.

 9               Dr. Carr, do you have an opinion about

10   this?

11               DR. CARR:   Yes, my name is Richard Carr.

12   I work in research at -- of Copenhagen.       We conducted

13   experiments in vivo over three weeks and we see no

14   evidence of down-regulation of these receptors.

15               DR. CARA:   Does that mean then that

16   there's no evidence of tolerance to the doses that

17   you're talking about?

18               DR. STRANGE:     The best clinical evidence

19   we have of tolerance or no tolerance is the yearlong

20   trials where we've seen sustained effect --

21               DR. CARA:   Sustained, good.

22               DR. STRANGE:     -- for a year.    We do see,

23   as is seen in I think all diabetes trials, that there

24   is a little decrease in the beginning, probably as a

25   function of participating in the trial.    We do see that
26   and then a little rebound.    But we do see sustained
                                                             45

 1   effect.   In the naive subsets in the yearlong trials,

 2   we do see a sustained effect of a decrease of HbA1c of

 3   more than one or maintained for more than the 14 months

 4   of trial duration.

 5                  DR. CARA:   You show that there's quite a

 6   bit of variation despite equal dosing.       You show that

 7   there's quite a bit of variation in terms of plasma

 8   levels of repaglinide.        Even though you showed mean

 9   data for each dose group, do you have any evidence

10   showing that the doses of repaglinide or the plasma

11   levels, if you will, correlate with blood sugar control

12   on an individual patient basis?

13                  DR. STRANGE:    If you look at the whole 64

14   trial, there is a correlation between the attained

15   levels in plasma and the blood glucose control but there

16   is variability.     So, the prediction for any given

17   patient is not very good.       But I mean, there is the

18   exposure response through all the exposures that we

19   have tested.     But because of the scatter, the

20   prediction for any given patient is not very good.          I

21   mean, it's tough to predict from the onset.

22                  Page 23 of the briefing document.    I don't

23   have that right here.      That's right.   We put a figure

24   of that in the briefing document at page 23.       There it

25   is.   Dr. Sherwin has it now.
26                  What you see in this figure is the
                                                             46

 1   repaglinide exposures observed following doses of

 2   repaglinide.     If you have very, very good eyes, you can

 3   actually see that each of the doses has a symbol on here.

 4   Also shown is the regression line showing that in this

 5   regression which is basically a model, the decrease in

 6   average blood glucose as a function of exposure.       So,

 7   with one decade here, we see roughly, in the model data,

 8   a decrease of average blood glucose of 40 milligrams

 9   per deciliter.

10                  But to answer your question, because of the

11   scatter, the prediction for any given patient at the

12   onset is not very good.

13                  ACTING CHAIRMAN SHERWIN:   Is that related

14   to binding proteins?

15                  DR. STRANGE:   Could you repeat that

16   question?    Sorry.

17                  ACTING CHAIRMAN SHERWIN:   I was just

18   curious.    I don't want to belabor -- binding proteins.

19   Obviously, the drug is tightly bound to protein.       Does

20   the amount or affinity of the drug to binding proteins

21   account for the variable responses?

22                  DR. STRANGE:   The drug is found with 98

23   percent of plasma protein, the vast majority of which

24   is albumen.     That correlation has not been made but

25   it's very highly unlikely because of the very high
26   protein binding.      The free drug available will be
                                                              47

 1   relatively independent of the absolute protein level

 2   because of the very, very high protein binding, percent

 3   of protein binding.

 4               DR. FOSSLER:      Hi.   Mike Fossler, FDA.

 5               This is probably expected for a drug that's

 6   metabolized by 3A4.   There's a lot of free A4 in the

 7   gut and so you're going to see day-to-day fairly wide

 8   fluctuations in bioavailability.       That's probably the

 9   most likely explanation.

10               ACTING CHAIRMAN SHERWIN:       Thank you.

11               DR. CARA:    Yes, but when I look at this

12   graph on page 23 in looking at the data, I mean, you're

13   looking at a graph that is really comparing log area

14   under the curve versus change in blood glucose.         It

15   strikes me that it's awfully flat.

16               DR. STRANGE:     Well, I mean, if you look at

17   the y axis -- can you turn that back on?       If you look

18   at the y axis on that specific plot, you'll see that

19   this difference here is 100 milligrams per deciliter

20   in average blood glucose.    Now, that's a very, very big

21   difference, 100 milligrams in average blood glucose.

22   I mean, if you have one patient who is on average, let's

23   say, 230 which is pro-control and you bring that patient

24   with 100 here down to 130 or 140 or 150, that's a decent

25   control, very decent control.       So, I mean, this curve
26   is a little deceptive.      It understates the effect of
                                                                48

 1   the drug.

 2                 DR. MOLITCH:     That's not the point.     The

 3   point is that you have such wide variations in the area

 4   under the curve with such a little change in blood

 5   glucose.    You've got 100-fold change in concentration

 6   for a relatively small change.      Also, even for the same

 7   very large amounts, you have such a wide change in

 8   bioactivity for even very huge amounts of drug.

 9                 What's the explanation for the lack of the

10   dose response, essentially?

11                 DR. STRANGE:     The lack of dose response --

12                 DR. MOLITCH:     Or the minimal dose

13   response.

14                 DR. STRANGE:     There is a dose response.

15   Let's state that first.      Then we'll say the explanation

16   for the variation is that it is probably the state of

17   the patients when the patient is treated that's more

18   important than the absolute.       I mean, it's the

19   responsivity of the patients.      The patients ability to

20   respond, their sensitivity to this therapy that's more.

21                 I'd like to call in one other point that

22   seems to be a little disturbing here.     What you see here

23   is intersubject variability, right?        You see one

24   patient, to the next patient, to the next patient which,

25   admittedly, there is a large variability.        For the
26   intraindividual variability, when you measure the
                                                              49

 1   exposures attained at week one, two, three, four in this

 2   trial, that is very substantially smaller.    I think the

 3   intersubject variability is 95 percent while the

 4   intersubject variability is 35 percent.

 5                  So, the prediction you have in any given

 6   patient you treat over time is going to be the same.

 7   You don't have this large variability once you treat

 8   one patient.

 9                  DR. MOLITCH:   I mean, are we dealing with

10   a salability type of phenomena to binding site for this

11   drug so that everything that everything that's in

12   excess really isn't doing very much?      I mean, are we

13   really having a variability of effect at the cellular

14   level?

15                  DR. STRANGE:   That would be pure

16   speculation so I'd rather not venture into that.

17                  DR. CARA:   Do you have data on individual

18   patients that have been treated with progressively

19   higher dosages to see if there is, in fact, a dose

20   response on an individual basis?

21                  DR. STRANGE:   There has been done dose

22   escalation trials in individual patients in which

23   patients have been receiving -- it was before we really

24   got the dose range nailed down, but they received first,

25   a half milligram, then two milligrams, and eight
26   milligrams.     We do see increased response in the same
                                                           50

 1   patients with the higher dose.    Yes, that is we do see

 2   increased response.

 3               DR. CARA:   Is it as shallow as this?    How

 4   does it compare to the data that you've got up here?

 5               DR. STRANGE:   This is a very long time ago

 6   so I'd rather not answer that straight.

 7               Yes?

 8               DR. HIRSCH:    Just to understand this

 9   curve, what is the ordinant?   It says mean glucose over

10   -- what's measured?

11               DR. STRANGE:    Yes, I'm sorry about that.

12   BG mean is the average 24 hour glucose.    What has been

13   done is that you've taken the AUC of the glucose over

14   the 24 hour period and divided by the time which is

15   essentially 24 hours.   So, I mean, if you took the

16   response over time and then you made it one flat line,

17   what is the average?

18               DR. HIRSCH:    So, food intake variability

19   could be a big factor in this as well, which must vary

20   enormously in these people, or not?

21               DR. STRANGE:    No.

22               DR. CARA:   But even so, you're talking

23   about a blood glucose change of maybe 50 for a dose that

24   varies by 1,000-fold.

25               DR. HIRSCH:    I understand.
26               What happens, by the way, when the drug is
                                                              51

 1   given and they don't eat anything?        Someone must get

 2   sick sometime or something, whatever.

 3                DR. STRANGE:    We've done a large amount of

 4   healthy volunteer studies that have been done fasting

 5   and I think the very first curve -- not I think.        The

 6   very first curve I showed you was done fasting.

 7   Healthy individuals have no problems.       Some of them

 8   experienced hypoglycemia, not unexpectedly, but

 9   otherwise, we don't see any --

10                DR. HIRSCH:    The diabetic subjects?      Has

11   that been studied in them as well?

12                DR. STRANGE:     I don't think we've ever

13   given this drug fasting to diabetes patients.       I get

14   confirmatory nods.   No, we haven't done that.

15                DR. KREISBERG:     Robert?

16                ACTING CHAIRMAN SHERWIN:       Okay.

17                DR. KREISBERG:    Can I ask, are we going to

18   revisit this particular issue?

19                ACTING CHAIRMAN SHERWIN:       We're going to

20   go back and go over everything.      This is just the

21   beginning.

22                My view would be let's let the company

23   finish their presentation.     We'll take a break and then

24   we'll begin the real questioning.

25                DR. EDWARDS:     Okay, good morning, ladies
26   and gentlemen.   You heard earlier from Dr. Reno our
                                                             52

 1   encouraging preclinical safety profile.       I'd like to

 2   continue now by describing the clinical safety features

 3   of repaglinide.

 4                  Let's start by looking at the duration of

 5   exposure in the control trials.      This data shows you

 6   here the number of subjects exposed to repaglinide

 7   versus the time intervals of their exposure.        Let me

 8   draw your attention to this column which indicates that

 9   we had 831 patients treated for more than a year with

10   repaglinide.     The total exposure exceeds 1,000 patient

11   years of repaglinide which we think is fairly

12   substantial for this stage in the drug's development.

13                  Now, Poul Strange made some observations

14   specific to the study 049 which is the US trial of this

15   type.    Just let me extend those observations.    Most of

16   the safety exposure comes from a long-term active

17   control trials including all those 831 patients, as you

18   saw.    So, just let me try and orientate you to the type

19   of patients that we are talking about here.

20                  These trials, the basic design of which Dr.

21   Strange explained, you can see there's a full year

22   treatment with repaglinide after the titration phase.

23   To get you a feel for the type of patients, if we cull

24   all our data across the -- trial that I'll describe to

25   you, the typical patient was 60 years old.        They had
26   had their diabetes about eight years and had an HbA1c
                                                           53

 1   of just under eight, although that describes an

 2   enormous variation in that variable.

 3                Let's take a look now at the comparitor

 4   trials.   This is the trial 049 that Dr. Strange showed

 5   you with 576 patients.   There were two other trials of

 6   this type, 046 and 050, where glyburide was the

 7   comparitor agent.   There was one trial 048 where the

 8   comparitor agent was glipizide, showing 81 patients

 9   exposed and one trial 047 in which glipizide was the

10   comparitor agent.   I want to draw your attention to the

11   fact that all of the trials featured a two to one

12   randomization of repaglinide versus comparitor agent.

13                This slide summarizes for you the exposure

14   by age, by gender and by race with the US studies on

15   the left of the slide as you look and the European

16   studies on the right.    If we look across the age line,

17   we can see that approximately 25 percent of patients

18   in this population were above 65 years of age.

19   One-third were women and within the United States'

20   trials, there was a reasonable racial mix of the

21   expected population, while in Europe almost all the

22   patients were Caucasian.

23                This slide deals with discontinuations.

24   On the left-hand side, you see the placebo controlled

25   trials.   Those trials were the focus of Dr. Strange's
26   presentation.   On the right-hand side, you see the
                                                            54

 1   one-year comparitor trials which I will be focusing on.

 2   At the top of the slide, you can see the number of

 3   patients exposed.    If we go down to the proportion

 4   completing, you will see that more patients on

 5   repaglinide than placebo actually completed the trial

 6   period.    Now, the main reason for this difference is

 7   seen here which is the large number of placebo patients

 8   were withdrawing because of hypoglycemia ineffective

 9   therapy.   However, if we look at the adverse event

10   line, we can still see that placebo patients actually

11   suffered more AEs than the repaglinide.

12                 Now, if we look at the right-hand panel in

13   the slide where we look at repaglinide versus all the

14   comparitor agents culled, you can see that the

15   proportions completing repaglinide and comparitor

16   agents were the same.    The discontinuation rates for

17   adverse events were the same.    I'd like to draw your

18   attention to this line here which we'll return to later

19   which is that twice as many patients in the comparitors

20   withdrew with hypoglycemia than with repaglinide.

21                 On this slide, we're looking at adverse

22   events overall.    For an event -- here to appear on the

23   slide, it had to be experienced by five percent more

24   patients in any one treatment category.     Now, if we

25   look first at the repaglinide versus placebo and just
26   look down those two lines, the only thing that seemed
                                                            55

 1   to stand out was this here, something in the respiratory

 2   area.   But when we look across the repaglinide versus

 3   active comparison trials, we really don't see any

 4   difference.     So, we think that the safety profile is

 5   good and it's very comparable to active comparitors

 6   tested.

 7                  Now, we looked for evidence of dose

 8   response in our adverse events and really didn't see

 9   anything.    I just want to revert for a minute to this

10   trial, trial 036, described earlier by Dr. Strange,

11   which was the ascending tolerance trial in type 2

12   diabetics.     To get you orientated here, this was a dose

13   escalation trial starting at 16 milligrams per day and

14   going up to 80 milligrams a day.      That is five times

15   the maximum recommended dose.      You can see there were

16   very few adverse events reported in 15 patients on

17   repaglinide and five on placebo.     I'd like to point out

18   at this point that we didn't detect any changes in liver

19   enzymes of note, nor any changes in ECG intervals, or

20   indeed, any evidence of ischemia.

21                  I'd like now to turn to hypoglycemic events

22   in one-year trials.     Just let me explain the

23   nomenclature on the slide for you.     The top line is the

24   number of patients exposed which are the numbers we've

25   seen before.     The next line is the number of patients
26   who experienced one or more hypoglycemic events.      The
                                                            56

 1   next line is the percentage of patients who d/c

 2   discontinued because of hypoglycemia.    The next line

 3   is the percentage of patients who had a hypoglycemic

 4   episode where blood glucose was measured.   So, in this

 5   case, 50 percent of the episodes reported there was

 6   actually a blood glucose measurement made.    The next

 7   line is the percentage of patients who had a blood

 8   glucose when it was measured which was less than the

 9   threshold value of 45 milligrams per deciliter.

10   Finally, the mean blood glucose measured in this

11   population of patients.

12               Now, if we start at this line and we look

13   across, we can see that the overall frequency of

14   hypoglycemia appears very comparable.    However, this

15   difference, apparent small difference, between

16   glyburide and repaglinide in repaglinide's favor, it

17   becomes more apparent when you look down the slide.    If

18   we look at the proportions discontinuing, you can see

19   it was lowest in repaglinide.   If we look at those

20   patients who had a blood glucose value less than four

21   to five, it was half the number of patients with

22   repaglinide than with glyburide.    So, we find that

23   information on hypoglycemia very encouraging and we'd

24   like to return to that later in our presentation in

25   discussion section.
26               I'd now like to turn to cardiovascular
                                                          57

 1   events and start with a simple slide.    These are the

 2   numbers we've seen before.   The number of patients in

 3   the one-year, long-term active comparitor trials 1,228

 4   with Prandin, repaglinide, 417 with glyburide, and 81

 5   with glipizide.   This slide shows you the crude event

 6   rates and the percentage of patients experiencing those

 7   events for three types of events:    serious

 8   cardiovascular events, cardiac ischemic events, and

 9   death due to cardiovascular events.

10               If you look across the serious

11   cardiovascular event line first, you see four percent

12   of patients had such events with repaglinide, two

13   percent with glyburide and six percent with glipizide.

14   The cardiac ischemic events, anginas, in total

15   myocardial infarctions and so on in two percent of

16   repaglinide patients, one percent of glyburide

17   patients, and five percent glipizide patients.    If we

18   look at death due to any cardiovascular event, in this

19   case, within capillaries 10/10 and 10/40 of the ARD

20   system, we see there were six such deaths now returned

21   with repaglinide and two deaths with glyburide.

22   Please bear in mind when you look at these numbers, you

23   need to look at the relative exposure rates.

24               These were the deaths.    You can see here

25   the treatment years for repaglinide, as I said about
26   1,000, and this shows you the data for all the
                                                          58

 1   comparitors pulled.   We're five infarcts with

 2   repaglinide while with one comparitor, one death due

 3   to cardiac failure with repaglinide.   One heart block,

 4   as it is recorded, with a comparitor agent and one event

 5   reported as a cardiac arrest meaning a total of six

 6   versus a total of three such events.

 7                Now, we're going to look at the data that

 8   we have on cardiovascular events in a somewhat more

 9   sophisticated way.    This slide shows you

10   cardiovascular serious adverse events.     I have three

11   thoughts of this type, so just let me explain the first

12   one carefully.   What is shown here is the time to first

13   event with each little blip in the line representing

14   the event.   The exposure time in these long-term --

15   trials here and the cumulative incidence of such events

16   on the y axis.

17                The solid blue line here represents the

18   repaglinide values.   The fainter blue lines at each

19   side, the 95 percent confidence intervals for these

20   events.   The light underneath shows you the value for

21   all the comparitors pulled with their confidence

22   intervals.   As you can see, while in the absolute

23   numbers the repaglinide are larger, that the confidence

24   intervals clearly overlap.

25                Now, we can cut this data in many different
26   ways, and indeed, we have cut it in many different ways.
                                                            59

 1   What this shows you is the cumulative instance

 2   unadjusted for the same events seen on the previous

 3   slide.    Now, the data is difficult to interpret for a

 4   couple of reasons.     Firstly, let's deal with the

 5   placebo.    As Dr. Strange showed you, we had asymmetric

 6   randomizations with placebo.      So that in placebo

 7   control trials, far more patients were actually treated

 8   with active agent than placebo.      As I showed you

 9   earlier, a lot of the patients withdrew because of

10   ineffective therapy.

11                 With glipizide, while the percentage of

12   patients with events was high, the absolute number of

13   events is small.    You can see if we just look at it in

14   a simple way, repaglinide here appears in the middle

15   between glipizide, gliclizide and glyburide shown

16   here.    Now, on the same data set if we look at all acute

17   ischemic cardiovascular events -- so this would include

18   myocardial infarctions and all anginal episodes.     What

19   you can see here is essentially the same pattern with

20   glipizide sticking out up here, but now the data for

21   repaglinide, glyburide, gliclizide appears much closer

22   together.

23                 Now, in a Cox Regression model which has

24   looked at this data, we see some fairly straightforward

25   things which we would expect, I think, which gives us
26   some confidence in the model.     Firstly, if we look at
                                                            60

 1   cardiovascular events overall, as one would expect, the

 2   older was the patient, the greater was the risk a year.

 3   The same was true with patients who had a previous

 4   medical history of cardiovascular risk.      This, I

 5   think, is somewhat important in the sense that when you

 6   analyze the baseline covariates in some of the trials

 7   that the patients are not equally distributed for

 8   previous history of cardiovascular risk.    I would like

 9   to return to that later.     Also, as you would expect,

10   patients with very baseline ECG abnormalities also had

11   a high risk.     And if we look from the time of the

12   subject's first hypoglycemic event, it's also to some

13   extent an important covariate in determining the time

14   at first event in the model.

15                  So, I've shown you some absolute numbers

16   and I've shown you some cumulative incident plots.     I

17   now want to just return briefly to this famous landmark

18   study, the UGDP.    Many people here, I'm sure are very

19   familiar with this trial.    It has been very significant

20   in determining labeling of oral hypoglycemic agents for

21   a long time -- first off in 1961 and the results are

22   still with us.    What you can see here is the cumulative

23   mortality rate in percentage versus the years in the

24   trial.   To remind those of you who are unfamiliar, the

25   chart is best known for having shown the success risk
26   of cardiovascular events with tolbutamide versus the
                                                              61

 1   other groups tested insulin here or placebo.

 2               Now, notice the duration of the trial.       If

 3   we had looked at this trial data in a shorter period

 4   of time, we might not have drawn the same conclusion.

 5   Because if we just look at the first couple of years,

 6   the actual event rates were higher with insulin than

 7   with either tolbutamide or placebo.     Just to put this

 8   in context and clear that the data is not drawn

 9   contemporaneously.   What we've just shown here to put

10   this in perspective is the actual event rates that we're

11   talking about, this is repaglinide, this is

12   glubanclumide and of course if I put glipizide on it,

13   it would have been up here.   You can see that the actual

14   event rates for cardiovascular mortality are low.       You

15   also need to bear in mind that with UGDP we were talking

16   about patients within one year of diagnosis.      The

17   average patient in the trials we're talking about has

18   had their type 2 diabetes eight years.

19               As I try and summarize the safety profile,

20   overall mortality versus comparitor agents when we look

21   at them together is the same.     That is true whatever

22   we look at, whether we look at malignancies, whether

23   we look at cardiovascular disease.    When we look at the

24   overall safety profile -- and here, I'm thinking of the

25   slide where I showed you all events with a frequency
26   of more than five percent -- the profile appeared very
                                                             62

 1   comparable to approved OHAs.    We believe the high --

 2   profile is acceptable and maybe good.      We feel very

 3   encouraged about that area with repaglinide.

 4                I've shown you the overall cardiovascular

 5   profile is comparable to sulfonylureas.     When we look

 6   at our data in isolation in comparison with glyburide,

 7   we do see a small increase in non-fatal cardiovascular

 8   events.   We believe other than as Dr. Strange told you,

 9   that in patients with liver impairment -- other than

10   those patients, we do not believe special precautions

11   are required regarding dose adjustment.    We think that

12   we have really quite a wide therapeutic index.    If you

13   think about the 036 trial where I mentioned that

14   patients had received up to 80 milligrams a day, the

15   average area of the cadre you see there was on the order

16   of 5,000 nanograms per mil per hour.    Dr. Strange told

17   you that he expected the upper limit in patients within

18   the therapeutic range recommended was about 800.        As

19   you recall, we saw very few events in that trial.

20                To try and summarize our formal

21   presentation for you, if we look at the preclinical

22   profile, repaglinide, a new chemical entity, benzoic

23   acid derivative -- an oral insulin secretagogue with

24   distinct beta cell binding sites.      Insulin is not

25   released by direct exocytosis.    The compound is not
26   mutagenic, teratogenic or carcinogenic and we saw no
                                                            63

 1   clinically relevant preclinical safety changes.

 2                 The pharmacology is a rapid onset of

 3   action.    The Tmax of .7 hours and rapid plasma

 4   clearance.    We saw it enhanced insulin response to

 5   meals, a clinically important blood glucose response.

 6   The drug was potent, effective in doses of five

 7   milligrams and above.    We saw a wide variation AUC

 8   repaglinide, which has already been briefly discussed.

 9   Excretion more than 90 percent by the bile.     We saw

10   informal drug interaction studies.    No effective

11   repaglinide on the pharmacokinetics, digoxin,

12   warfarin, theophylline and no effect on cimetidine of

13   the kind that exhibit -- We feel that dose adjustment

14   will only be required specifically for patients with

15   liver dysfunction where careful titration is advised.

16                 In the efficacy profile, we saw a prompt

17   blood glucose response within one week of therapy.     We

18   saw in 064 a dose response in the range of .5 t four

19   milligrams given preprandially three times a day.      In

20   both titration 033 and fixed dose 044, 064 and 065

21   studies, we saw a significant benefit to repaglinide

22   versus placebo.    We've absolute reductions depending

23   on the study between 1.6 and 2.9 percentage points in

24   HbA1c.    We saw that in the long-term study, glycemic

25   control was maintained as well as with comparitor to
26   agents, and we saw a substantial additional effect when
                                                            64

 1   repaglinide was added to metformin failures.

 2                Dr. Sherwin, Dr. Fleming, that concludes

 3   Novo Nordisk's formal presentation.

 4                ACTING CHAIRMAN SHERWIN:     Thank you.

 5                I would like to thank Novo Nordisk for a

 6   succinct presentation.     It was really one of the first

 7   times we were really on schedule.    I really appreciate

 8   that.

 9                What I'd like to do is take advantage of

10   the break time now and give people a chance to sort of

11   digest the presentation.     My watch says 12 minutes to

12   10:00.   I would suggest that we begin at five after

13   10:00.

14                (Whereupon, off the record at 9:48 a.m.,

15   until 10:13 a.m.)

16                ACTING CHAIRMAN SHERWIN:     Okay.

17   Hopefully, we can reconvene.

18                We've conferred about, you know, how to

19   proceed.   My feeling is that the best way to proceed

20   at this point is to open up the forum to the Committee

21   and have them ask general questions that have arisen

22   as a result of the presentation.    Then we'll get to the

23   discussion of the specific questions that are raised

24   by Dr. Fleming.

25                So, I'd like to open it up to the panel.
26   I know Dr. Kreisberg had some questions.
                                                                65

 1                  DR. KREISBERG:    I defer to Dr. Marcus.

 2                  ACTING CHAIRMAN SHERWIN:    Oh, Dr. Marcus

 3   has more burning questions?      Okay.

 4                  DR. MARCUS:    I have one large question,

 5   but a couple of very small ones that perhaps you could

 6   just address first.

 7                  Is this drug approved elsewhere -- that is,

 8   other countries -- and in particular, is there any

 9   evidence from any work you may have done in Asia which

10   might give some insights as to how Asian-Americans

11   might respond?     Your representation by the Asian

12   community is extremely small that you presented.

13                  DR. EDWARDS:    Martin Edwards.   I think I

14   can address that for you.

15                  The current situation in Japan is we do

16   have clinical trial programs active in Japan.       We are

17   currently in phase III in Japan and we have

18   approximately a couple of hundred patients treated with

19   repaglinide.     So, that's the actual situation.     We

20   haven't seen anything untoward in terms of safety.         It

21   looks like the patients will probably have somewhat

22   lower doses in Japan.     That's not unusual.

23                  DR. MARCUS:    Okay, thank you.

24                  So, because this drug leads to some

25   interaction with calcium channels, I just wonder among
26   drug interactions, is there any interaction that you've
                                                               66

 1   seen with people taking calcium channel blockers?

 2                 DR. EDWARDS:     The simple answer is no.

 3                 DR. MARCUS:     Okay, good.

 4                 Now, my major concern has to do with an area

 5   that has really been very little discussed in your

 6   submitted documents and that has to do with weight loss.

 7   I think we all know that any intervention involved with

 8   management of patients with type 2 diabetes, whether

 9   it be an exercise in nutritional or a pharmacological

10   intervention, is highly interactive with any changes

11   in body weight.    I'd like to know whether there were

12   on average any changes in weight during your studies?

13   And whether there's any relationship even if there were

14   no change in the average weight between change of weight

15   of an individual person and the response to a drug?

16                 Also, in line with that, there's the whole

17   panoply of other cardiovascular risk factors which are

18   certainly paramount to patients with diabetes.

19   Triglycerides, high density lipoprotein, low density

20   lipoprotein, cholesterols, fibrinogen, plasminogen

21   activator inhibitor, and lipoprotein a, and there was

22   really vanishingly little of that in your formal

23   documents and nothing of that in your presentation.       I

24   would ask that somebody address those.

25                 DR. WHISNANT:    Thanks for the opportunity
26   to comment.   All of those are important questions and
                                                            67

 1   we're happy to respond.    I'm trying to find you a slide

 2   on the weight changes.

 3                  Let me summarize briefly by saying that in

 4   the one-year comparitor trials, that the patients on

 5   average experienced virtually nil change in weight, but

 6   that's an average phenomenon.      That is, patients on

 7   repaglinide in a table that I will show you and a figure

 8   I'll show you lost about .6 kilos over the one-year

 9   period of time.     It's different for patients who were

10   completers versus patients who dropped out of the trial

11   early-on.     That compares to, for various trials,

12   somewhere between .6 and one kilos of weight gain on

13   average for the comparitor drugs.

14                  The slide in front of you summarizes weight

15   change for all patients in the US comparitor trial, 049.

16   Note the distribution of patients within five percent

17   and below nil change, and within five percent above a

18   nil change.     Notice that for repaglinide in the gray

19   bars or pale blue bars, that there is a slightly higher

20   columns minus for the less than five percent and for

21   the zero to five percent below the mean compared to

22   glyburide.    The other trials look about the same.

23                  The second part of your question is do, for

24   instance, highly responsive patients or naive patients

25   gain more weight during their response to repaglinide
26   as is very often seen with other OHA drugs?    The answer
                                                          68

 1   is basically yes.    We find that the distribution of

 2   patients -- it's all right.    I think the point is well

 3   made.    The distribution of patients for naive patients

 4   in the 49 trial is somewhat above no change at the end

 5   of the year, with most of those being between zero and

 6   five percent weight gain and some 25, a quarter of the

 7   patients would gain even more than that.     The

 8   variability of weight change for some patients is very

 9   significant, either very significantly lost or very

10   significantly gained.

11                 DR. MARCUS:   Was an attempt rate made to

12   control dietary intake?     What sort of dietary advice

13   was given to these patients during the course of the

14   trial?

15                 DR. WHISNANT:   Patients who entered the

16   comparitor trials were given standard dietary, if you

17   will, advice during the run-in period of the trial, but

18   there was no long-term, for instance, intensified

19   program like is being planned for the DPP trial or for

20   other long-term intensive management kinds of trials.

21   These were conventional diabetes care trials with the

22   addition of repaglinide versus comparitor.

23                 DR. MARCUS:   And the other cardiovascular

24   risk factors?

25                 DR. WHISNANT:   The other cardiovascular
26   risk factors actually, in response to a question that
                                                              69

 1   we're going to address in a little bit about

 2   cardiovascular risk itself as an outcome, I can tell

 3   you that there was an imbalance in the 49 trial of

 4   patients who had had prior MIs, angina, baseline EKG

 5   changes, et cetera, and I'll show you that slide in a

 6   little while.

 7               The other cardiovascular risk factors that

 8   you asked about and it's, in effect, the population risk

 9   factors, I'd like to ask Dr. Edwards to address lipid

10   changes, et cetera.

11               DR. EDWARDS:     Thank you.

12               I think the most straightforward answer I

13   can give is that we did not see changes in the lipids

14   in our long-term one-year trials.       We did look, as I

15   briefly mentioned about important covariates, our

16   focus was to look for baseline imbalances between the

17   groups in the long-term trials.     With respect to

18   lipids, there were no imbalances.

19               ACTING CHAIRMAN SHERWIN:      Are there other

20   data that we could see?

21               DR. EDWARDS:     On lipids?    We have not

22   prepared slides on lipids.    We can.     We have got our

23   integrated summary of efficacy with us.      If you would

24   like, we could prepare a couple of slides --

25               ACTING CHAIRMAN SHERWIN:      Well, you know,
26   depending on what you can do, perhaps after the lunch
                                                             70

 1   break --

 2                DR. EDWARDS:    Yes.

 3                ACTING CHAIRMAN SHERWIN:     -- whatever

 4   data you might have on cardiovascular risk factors

 5   would be helpful.

 6                DR. KREISBERG:    I didn't understand his

 7   response, Bob.

 8                Are you saying that there was no difference

 9   in cardiovascular risk factors such as lipid levels

10   when you evaluated your drug versus the comparitor?     Is

11   that what you're saying?

12                DR. EDWARDS:    Yes.

13                DR. KREISBERG:    But there must have been

14   changes in lipids that occurred as a consequence of the

15   use of the drug compared to placebo, or is that not true?

16                DR. EDWARDS:    Yes.   Well, what I was

17   focusing on, I was thinking of cardiovascular risks in

18   terms of our long-term active comparitor trials because

19   that is where most of the exposure is.     The answer I

20   gave was that if we look at the change in lipids in those

21   trials, we don't see any difference between our drugs

22   and the comparitors we tested.              When we

23   looked to the baseline imbalances at randomization in

24   terms of risk programs, we saw some important

25   differences.     We saw some important differences with
26   respect to previous cardiovascular history, with
                                                                71

 1   respect to ECG differences, but we did not see

 2   differences in respect to lipids at baseline.

 3                  DR. KREISBERG:    But there must have been

 4   some --

 5                  I'm sorry, go ahead, Bob.

 6                  DR. MARCUS:    No, my question was -- I mean,

 7   it was simply, if you did a repeated measures analysis

 8   of variance on any one of these lipoproteins or other

 9   biochemical risk factors, was there a difference across

10   time in response to your drug?

11                  DR. EDWARDS:     No, we don't believe so.

12   We'll check in the ISE for you.     We don't believe that's

13   the case.

14                  DR. MOLITCH:    We all want that

15   information.

16                  DR. EDWARDS:    Okay.

17                  ACTING CHAIRMAN SHERWIN:     Yes, including

18   the placebo data because a lot of these changes will

19   occur earlier than one year.       So, it would make sense

20   to me, you have placebo control trials this would be

21   a critical element as well.       So, we're interested in

22   comparison with placebo and changes over time.

23                  Maria?

24                  DR. NEW:   I just would like to address my

25   question to Dr. Edwards.      It must have been a shock to
26   you to see no change in lipids in view of the rise in
                                                           72

 1   insulin?

 2               DR. EDWARDS:    May I suggest that we put

 3   together two or three slides which summarize all of the

 4   data?

 5               ACTING CHAIRMAN SHERWIN:    Yes, I think we

 6   should give you a chance to respond.    We'll take this

 7   up after lunch.

 8               Bob?

 9               DR. KREISBERG:    I have several questions

10   and then maybe we could cover some of them and let me

11   come back to the others.

12               It seems to me that this issue of

13   cardiovascular risk is important, even though the

14   comparative data suggests that it is as good as other

15   sulfonylureas.    That doesn't necessarily mean that it

16   is safe, just that it is as safe as other drugs that

17   are currently approved.    Because the drug involves the

18   ATP sensitive potassium channel which influences a

19   phenomenon called conditioning or pre-conditioning

20   response to ischemia, the question is do you have any

21   studies showing whether your drug influences the

22   response to ischemia, either in experimental animals

23   or in any other setting?

24               DR. FUHLENDORFF:    We have not studied

25   ischemia, but we have studied the affinity on the ATP
26   sensitive potassium channel, both in heart and in beta
                                                             73

 1   cell.   There was 100 to 400-fold potency difference or

 2   affinity difference.    So, the affinity was much higher

 3   to the beta cell than the heart.    That's called tissue

 4   selectivity.     We have no studies of ischemia.

 5                  ACTING CHAIRMAN SHERWIN:   By the way, have

 6   you looked at brain as well?

 7                  DR. FUHLENDORFF:   Yes, we have looked at

 8   brain as well.     The affinity of repaglinide in brain

 9   is the same as it is in the beta cell.

10                  DR. KREISBERG:   The studies were not

11   designed really to look at cardiovascular endpoints and

12   I don't think that there are sufficient patients

13   enrolled, considering what the projected frequency of

14   a clinical endpoint would be to come to any meaningful

15   conclusion about whether this drug does or does not

16   predispose to cardiovascular endpoints.      I think that

17   this is an important issue to be addressed in ongoing

18   studies by the firm should this drug receive approval

19   and actually be used in patients because I don't think

20   that's addressed here.

21                  The other thing that I would like to talk

22   to you about is actually getting back to this issue of

23   the variability in the plasma levels of the drug,

24   vis-a-vis the biologic response in the patient.        It

25   suggests to me that while 99 percent of this drug is
26   protein bound, that may be in the aggregate what it is.
                                                            74

 1   But have you ever looked in individual patients to see

 2   what the variability is in protein binding, to see

 3   whether some of the variation in the response has to

 4   do with a greater free fraction of the drug.     I assume

 5   that free is what is biologically active in this drug.

 6   Whether there is a varying free fraction of the drug

 7   that accounts for a variation in biologic

 8   responsiveness.              A follow-up on that is simply

 9   to say that in your drug interaction studies, it looks

10   to me like of four drugs that you evaluated, three of

11   them were competing for a common hepatic pathway and

12   only one, warfarin, might have been a drug in which you

13   were looking at competition regard to binding on

14   albumen.   It seems to me that there must be other drugs

15   that you could evaluate that would compete with

16   repaglinide for binding sites on albumen.

17                DR. WHISNANT:     I think we can only take

18   those as good suggestions.      We have not seen any

19   evidence -- just to return to the question that Dr. Cara

20   originated, we've not seen any evidence that the highly

21   variable plasma levels of this drug, whether bound or

22   -- well, total plasma levels measured correlate with

23   clinical toxicity events.     In fact, we reanalyzed the

24   information from the US long-term safety trial, study

25   49, looking for whether or not variability in steady
26   state levels correlates with cardiovascular events.
                                                            75

 1   The fact is, it doesn't.

 2                So, we are perfectly happy to pursue the

 3   suggestions that you're offering on a mechanistic basis

 4   in order, perhaps, you know in the future, to develop

 5   some more rational bases.    But the clinical correlate,

 6   however, is going to be very difficult for us because

 7   therapeutic drug monitoring for hypoglycemic agents is

 8   certainly a new field, to say the least.    It's not the

 9   norm and it has not been the way this drug or other drugs

10   has been developed.    So, we take your suggestions as

11   a future development, but I'm not sure we have specific

12   therapeutic drug monitoring or specific albumen

13   binding fraction data to help with the dosing of the

14   drug at the present time.

15                ACTING CHAIRMAN SHERWIN:    In that regard,

16   are you able to measure free drug?

17                DR. WHISNANT:    What is the specificity of

18   the new LCMS assay with regard to -- can we answer that

19   question?

20                DR. HANSEN:     Kristian Hansen, Novo

21   Nordisk.

22                We are not able currently to measure the

23   free fraction.    What we measure is total drop in

24   plasma, okay?    Obviously, we have a bound fraction

25   which is pretty high.
26                What I'd like to point out -- it's a very
                                                              76

 1   good point you make, but this drug is actually a highly

 2   clearance drug.    Obviously, if there is variations in

 3   the free fraction, that will be rapidly clear for the

 4   bloodstream.    So, that, perhaps, de-emphasizes a bit

 5   the protein interactions you're referring to.

 6                  ACTING CHAIRMAN SHERWIN:   Roger?

 7                  DR. ILLINGWORTH:   To extend the drug

 8   interaction question, since the drug is metabolized by

 9   a cytosol 3A4 system, have you looked at drug

10   interactions with cyclosporin, erythromycin,

11   ketaconazole -- drugs that metabolize by the same

12   system?

13                  DR. STRANGE:   The short answer is that

14   there has been no specific drug interaction trials

15   performed.   We have done a very detailed post hoc

16   analysis with 3A4 inhibitors, ketaconazole, that

17   patients in the long-term trial who happen to be on

18   those agents.    We have seven treatment exposure years

19   concomitantly with repaglinide in 18 patients.         Only

20   two of those patients had any events of hypoglycemia

21   which is the only dose related event we have been able

22   to find.   Two out of 18 is actually less than our

23   baseline rate in all patients.

24                  So, it's not the specific interaction

25   trial, but it is evidence that we don't have any
26   clinically relevant interactions.
                                                             77

 1                  DR. MOLITCH:    And conversely -- the effect

 2   of those drugs?    How about the effects on those other

 3   drugs?   The converse of that, not the effect on

 4   repaglinide but the effect on these other drugs,

 5   including estrogens?

 6                  DR. STRANGE:    That hasn't been looked at.

 7   What we've looked at is adverse events in those patients

 8   and we haven't seen anything that's different from what

 9   we expect in all the treatment experience we have with

10   repaglinide.

11                  We have also in the 64 trial, which we

12   mentioned before that we measured drug exposures in a

13   reasonable number of patients, 120 patients, we found

14   that the exposures attained concomitantly with 3A4

15   substrates did not differ from the rest of the patients.

16                  ACTING CHAIRMAN SHERWIN:     Mark?

17                  DR. MOLITCH:    I guess coming back to this

18   issue of the free drug.       Although we realize that this

19   is a somewhat new experience for oral hypoglycemic

20   agents if, in fact, you did show a correlation of free

21   levels with the biological activity, it might actually

22   be not only interesting but a therapeutically useful

23   target just like we measure other drug levels.       Given

24   the wide dose range that we're seeing here -- at least

25   the wide area under the curve and the wide dose range
26   in safety that you're providing for us, that in fact,
                                                              78

 1   perhaps in some patients may need to go into higher

 2   doses and so much lower doses depending on the free drug

 3   level for that patient.     Maybe that's the way it should

 4   be titrated rather than a fixed oral dose for the

 5   patient.

 6                DR. WHISNANT:    Certainly, the fact is that

 7   response to drugs like this are highly variable.       In

 8   fact, response to insulin is highly variable, as you

 9   all know.   So, any way in the future that we can dissect

10   the variability of that response, we would certainly

11   see as a future study of the mechanistic aspects of this

12   drug.

13                Dr. Fossler has some help for us about

14   this?

15                DR. FOSSLER:     I think you're forgetting

16   your in vitro work, which I think is the most important

17   aspect of your drug interaction studies.       You know,

18   they did some fairly good in vitro work which showed

19   quite clearly, the interaction with 3A4 substrates and

20   that's in the labeling.      So, you know, the current

21   state-of-the-art right now and the Agency's opinion

22   based on just recent guidances that have been issued

23   is that if you show an interaction in vitro, we can use

24   that in the labeling and that's fairly predictive.

25                DR. MOLITCH:    Part of my concern also with
26   the total drug levels that we're seeing and the great
                                                              79

 1   variability is in your statement that you think it's

 2   safe in patients with renal insufficiency.       There

 3   clearly was a rising level at the lowest creatinine

 4   clearance levels on 20 or so moles per minute, or 7.3

 5   meters squared.    I think that given that you only had

 6   six subjects, given the degree of variability, that I

 7   would really -- I don't think that you can say this

 8   without a larger number of subjects being looked at to

 9   look at the cumulation of drug, a considerably larger

10   number of subjects.

11                 DR. KREISBERG:    If I could make a

12   suggestion that's sort of a corollary of what has been

13   going on.   One of the things I think the firm could do

14   is actually look at indices of insulin sensitivity in

15   these patients to see whether that determines what the

16   response is to a particular dose of the drug.       That

17   would correlate what we know to exist in type 2 diabetic

18   patients and that is a varying degree of insulin

19   resistance.

20                 DR. WHISNANT:    The only piece of data that

21   I can give you for sure is that C peptide levels, fasting

22   C peptide levels at entry and during study do not

23   correlate with hypoglycemia frequency in this trial and

24   do not predict response, actually.

25                 ACTING CHAIRMAN SHERWIN:     Maria?
26                 DR. NEW:   I just would like to suggest that
                                                                 80

 1   a possible explanation for variability is in a drug

 2   which is metabolized by a cytochrome p450 system and

 3   which has a very large clearance.         Since the

 4   cytochrome p450s are genetically programmed and very

 5   variable among individuals, that that's probably the

 6   explanation.    You could test that, actually, by giving

 7   atypical cytochrome p450 metabolized drug in

 8   individuals that are widely varied in their drug levels

 9   and see if that's the difference.

10                  DR. WHISNANT:      Variability both in the

11   liver and the gut and there's a well known system now.

12   So, that's a future trial that perhaps would help us.

13                  DR. NEW:    Yes.

14                  DR. CARA:    But would that explain the

15   individual variability?

16                  DR. NEW:    You know --

17                  DR. CARA:    I mean, the CV is over 60

18   percent.

19                  DR. NEW:    Yes.    José, the thing I'm most

20   acquainted with is the difference in the cytochrome

21   p450s that metabolize cortisol.        They're extremely

22   variable.    We delude ourselves when we think that we

23   give a program dose and pediatricians per kilo per meter

24   square.    You really have to measure the clearance and

25   the degradation to know what dose to give.
26                  DR. CARA:    Sure.
                                                            81

 1                DR. NEW:    The point I keep making about

 2   children, you know, we keep saying that we have to scale

 3   down the dose in children because they're smaller.    But

 4   in fact, they metabolize faster and they need a bigger

 5   dose.

 6                DR. CARA:    But these issues bring up

 7   important other questions which relate to -- you know,

 8   you talk about titration.    What is your definition or

 9   your proposed definition as the endpoint of dose

10   titration?

11                DR. WHISNANT:    It was defined

12   experimentally in the clinical trials.     As designed,

13   titration means begin at .5 milligram dose with each

14   meal.   Assess morning fasting plasma glucose response

15   after 10 days to 14 days, and adjust upward by doubling

16   if the patient has not achieved a target fasting plasma

17   glucose of 160 or 140.    The target fasting plasma

18   glucose was the same during the dose titrations for both

19   repaglinide and for the comparitor drug in those

20   trials.   So, it's an empiric definition within the

21   context of the experimental design and the dose

22   titration steps were .5, one, two, and four.

23                DR. CARA:    I guess my concern is that

24   there's no stopping the titration point.       Because if

25   you say, "gee, you know, I got down to a fasting blood
26   sugar of 150", what's to say that doubling the dose is
                                                             82

 1   not going to decrease it to 120?   And that quadrupling

 2   it will lead to a further fasting blood sugar level?

 3   I mean, everybody is going to end up on the maximum dose

 4   in view of the fact that in diabetes, we typically try

 5   to get the blood sugar down to as normal a level as

 6   possible.

 7                 We admit that one of the weaknesses in the

 8   comparitor trial design as carried out was that there

 9   was, if you will, a stopping rule in those trials that

10   said "titrate to this level and stop, and do not change

11   the dose during the maintenance phase of the trial."

12   We admit that the trial design was carried out that way

13   primarily for regulatory purposes in order to show that

14   we are not statistically worse than the comparitor

15   drug.

16                 ACTING CHAIRMAN SHERWIN:   Now, what did

17   you use as your basis for that comparitor?     I'm just

18   curious about that.    In other words, was it fasting

19   glucose or --

20                 DR. WHISNANT:   Morning fasting plasma

21   glucose.    That's correct.

22                 ACTING CHAIRMAN SHERWIN:   Yes, because

23   that would trouble me a little bit because you're

24   comparing now drugs that have a longer duration of

25   action which will affect the fasting.     Whereas, you
26   might be giving more drug here to reduce the fasting
                                                             83

 1   since it doesn't theoretically last quite as long.     So,

 2   in other words, you would think that this drug would

 3   work more in the interim during the day and a little

 4   less during fasting, and using your fasting as your

 5   comparitor.    I think that's a little bit of a problem.

 6                 DR. WHISNANT:   Dr. Sherwin, we admit that

 7   the data are what they are.     The study showed that we

 8   were equivalent within the definition of not being

 9   worse than .6 grams percent of HbA1c at 12 months.

10   That's the reason the trials were carried out and

11   therefore, the data have to speak to only that

12   conclusion.

13                 A variety of other things could be done now

14   that we understand a lot more about this new dosing

15   paradigm which really is a part of the discussion that

16   Dr. Fleming is going to lead us to.

17                 ACTING CHAIRMAN SHERWIN:   Sure.

18                 DR. FLEMING:    I do think it would be

19   helpful to show the outcome of dose adjustments that

20   occurred in the comparative studies.       If you could

21   pull those slides up showing what happened with

22   repaglinide and with the comparitor drugs?

23                 ACTING CHAIRMAN SHERWIN:   Do we really

24   know -- I'm just curious how long this drug actually

25   works?
26                 DR. WHISNANT:   It works for over a year.
                                                              84

 1                ACTING CHAIRMAN SHERWIN:   No, no, no, no.

 2   I didn't mean it that way.    I'm sorry.    I'm sorry.

 3                What I meant was in terms of giving a dose

 4   and then seeing how long the effect lasts.      I'm

 5   confused.   Is this a drug that's over at 10:00 at night

 6   and doesn't work anymore?    You know, what's the

 7   duration of action of the drug?   Because I didn't see

 8   any data that really told me that.

 9                DR. WHISNANT:   Well, that is also built

10   into the points for discussion that Dr. Fleming has

11   designed.   Let me just refer you back to the three

12   paneled slide that Dr. Strange showed in the clinical

13   pharmacology studies.

14                If you accept plasma insulin as the kinetic

15   endpoint, if you will, for this drug then plasma insulin

16   curves follow the same as the meal related physiologic,

17   if you will, plasma insulin curves that are normally

18   seen without drug.   Therefore, that establishes a

19   kinetic for the drug.

20   If you ask for the kinetic of the blood glucose

21   response, remember that was also shown on the three

22   paneled slide that Dr. Strange showed you.      In fact,

23   the plasma glucose response also follows the, if you

24   will, meal related, normal physiologic pattern.

25                ACTING CHAIRMAN SHERWIN:      Well, I'm not
26   sure.
                                                            85

 1                  DR. WHISNANT:   Okay.

 2                  ACTING CHAIRMAN SHERWIN:   In other words,

 3   because the glucose levels are different in the two --

 4   you know, they're not the same, I'm not exactly sure

 5   that that proves that duration is short or long or

 6   whatever.   I'm not sure that you wouldn't see where I

 7   would -- you know, if you could show me the same pattern

 8   with a comparitor with lower glucoses after the

 9   comparitor and then show me the profiles and show me

10   differences between the drugs, maybe then I could

11   accept that.    But I'm not sure that I can accept that

12   under the -- I saw no data personally that really

13   totally convinced me what the duration of action of the

14   drug was.

15                  DR. WHISNANT:   Maybe we should go to the

16   issues discussion that's been --

17                  ACTING CHAIRMAN SHERWIN:   Well, I can stop

18   here and we can go back to that issue later on.    That's

19   my impression when I looked at the data that I've seen

20   so far.

21                  DR. FLEMING:    Well, I do think it's very

22   important to understand how these comparative studies

23   were conducted.     It's important in understanding the

24   comparability of dosing that occurred between the two

25   groups in each trial.    Now I received from the company
26   some data which showed at what doses patients ended up
                                                           86

 1   in the comparative studies and that allows us to look

 2   at both the comparitor and repaglinide itself in each

 3   study.

 4                I don't know if you've been able to put your

 5   hands on that particular transparency, but it is very

 6   interesting that apparently, there is almost a

 7   superimposability of the distribution of doses used

 8   during the observation period with respect to both

 9   repaglinide and the comparitor.     In other words,

10   approximately half the patients ended up at the high

11   dose of repaglinide and the comparitor and it went down

12   the line.   About 25 percent ended up at the -- it would

13   be the two milligram repaglinide dose or the second

14   highest comparitor group.    Each group seemed to

15   correspond very closely.

16                Now, I think it would be worth explaining

17   what the dosage rules were in the comparative studies.

18   This will, I think, reflect some approximation of

19   clinical practice going to Dr. Cara's excellent

20   question about whether one could perhaps go overboard

21   and drive patients into the ground by over-prescribing.

22   I don't think there is an indication that that happened

23   in these comparative studies.

24                DR. WHISNANT:   Certainly, to the contrary

25   actually, Dr. Fleming.    Thanks very much.    The dose
26   titration paradigm, let me repeat carefully.    Patients
                                                             87

 1   were started on .5 milligrams after a basal evaluation

 2   period, either a run-in drug free period or a crossover

 3   -- patients were started on .5 milligrams.     After an

 4   average of ten days, patients had a fasting plasma

 5   glucose assessment.     If their target had not been

 6   achieved -- that is, 160 or 140 -- then they were dose

 7   escalated.    After another week to 10 days, they were

 8   reassessed and they were dose escalated again.     So,

 9   those patients who achieved the target in either drug

10   stopped at the dose where they achieved that target and

11   that was their, if you will, "maintenance dose" from

12   then to the end of the trial without changing dosing.

13                 So, in effect, for both treatments, we got

14   what we set out to get.     We titrated to equivalent

15   endpoints and proved equivalence for these kinds of

16   patients.    Which patients ended up out of the

17   randomized, double-blind design -- which patients

18   ended up at the lowest dose level were those that were

19   obviously more responsive to the lowest dose.

20                 Yes?

21                 DR. MARCUS:   How many fasting plasma

22   glucose determinations went in after the ten days, just

23   one?    Or did you get daily for a few days?

24                 DR. WHISNANT:   A single one.

25                 DR. MARCUS:   One fasting glucose made --
26   okay.
                                                               88

 1                  DR. WHISNANT:     It's the normal dosing

 2   paradigm, titration paradigm of diabetics.

 3                  DR. MARCUS:     No, it's not.    I mean, when

 4   we see patients in clinic, we ask him to bring in a book

 5   documenting the last several weeks of home glucose

 6   monitoring.      And we can see that, you know, on one day,

 7   they may have been 90 and on other days they may have

 8   been 340.    So, I mean, that's --

 9                  DR. WHISNANT:    And we would have done the

10   trial that way if the Agency would accept BGMs as

11   endpoints.

12                  DR. CARA:   But you're not really

13   describing a dose response study then.

14                  DR. WHISNANT:     This is not a dose

15   response, sir.     This is a dose titration study to

16   equivalence.

17                  ACTING CHAIRMAN SHERWIN:       Correct.

18   That's right.

19                  DR. CARA:   Well, do you have any dose

20   response studies?

21                  DR. WHISNANT:    Yes, we do.    We showed you

22   two dose response studies.

23                  DR. CARA:   The non-comparitor studies?

24                  DR. WHISNANT:     One was a placebo

25   controlled phase II dose comparison over .25 milligrams
26   to four milligrams --
                                                              89

 1               DR. CARA:      Right.

 2               DR. WHISNANT:     -- fixed dose for patients

 3   who were randomized to fixed dose for four weeks --

 4               DR. CARA:      Right.

 5               DR. WHISNANT:      -- with primarily a blood

 6   glucose endpoint.   Then we showed you a six month trial

 7   randomizing patients to placebo, one milligram and four

 8   milligrams, where patients were evaluated with

 9   glycemic control, HbA1c.    In those studies, we believe

10   we've showed you an adequate response over that dose

11   range.

12               DR. KREISBERG:     I don't think so.   If you

13   would look on your page 22 and your figure 5.6, and on

14   page 36 your figure of 6.3 -- which actually is the same

15   study just displayed a little bit differently -- it

16   seems to me that the response kind of plateaus as you

17   go from .5 to two, you do have the sense and maybe you

18   get a little bit better response when you're up at four

19   milligram dosing but the numbers of patients in these

20   studies are relatively small.       I really wonder if

21   you're not on a plateau somewhere between 0.5 and two.

22   If you look at your figure 6.3, the mean glucose change

23   for the half, one and two milligram doses are the same.

24               DR. WHISNANT:      If you turn this figure

25   upside down and subject it to an Emax modeling exercise,
26   I believe the statisticians, clinical pharmacologists
                                                                90

 1   in the room will agree that this more-or-less fits an

 2   Emax model dose relationship for this drug.        We do

 3   admit that you've got a lot of response out of the lower

 4   doses.   You might ask what's the rationale for having

 5   chosen .5 milligrams as the lowest marketed dose of the

 6   drug, if you will.     That is because a substantial

 7   portion of patients, actually 25 percent of patients

 8   in the titration trials at first step, achieved

 9   response.   But in this trial and others, the .25

10   milligram dose was not nearly as effective in terms of

11   a responder analysis.

12                  You might also conclude that two

13   milligrams might be enough for a very large percentage

14   of patients.     We would not disagree with that.     In

15   fact, we've identified a subset of patients in which

16   two milligrams seems to be an ideal starting dose.

17                  DR. KREISBERG:    Well, do you think that

18   the mean reduction in glucose for the four milligram

19   dose, taking into consideration the number of patients

20   that have been studied, is significantly -- and I mean

21   statistically significantly different than the values

22   attained with .5 to 2.0?

23                  DR. WHISNANT:    Actually, the study was not

24   designed to test statistically the difference between

25   two milligrams and four milligrams.        It was designed
26   to test statistically the dose response over the dose
                                                              91

 1   range.    Those statistics are clear and included in the

 2   report.

 3                 DR. MOLITCH:    I'd like to come back to this

 4   fasting glucose business and also your mean blood

 5   glucose levels.    Figure 5.4 looks at the average blood

 6   glucose profiles.    Are these the blood glucose

 7   samplings that made up the mean glucose profiles that

 8   we're talking about?    Because the sampling -- then

 9   dividing by 24 or whatever?

10                 ACTING CHAIRMAN SHERWIN:     Where is this?

11                 DR. MOLITCH:    Figure 54 on page 19.    Are

12   those the sampling times for that glucose profile?

13                 DR. WHISNANT:    Sir, that's a different

14   trial.

15                 DR. MOLITCH:    But is that the nature of the

16   sampling profile?

17                 DR. WHISNANT:    What you would do --

18                 DR. MOLITCH:    What made up the mean

19   glucose levels?    What sampling times?

20                 DR. WHISNANT:    You would do a 20-point

21   profile on each patient.

22                 DR. MOLITCH:    And when were they done?

23                 DR. WHISNANT:    The time intervals for the

24   20-point profiles?

25                 DR. MOLITCH:    Yes, exactly.
26                 DR. WHISNANT:    What are the time
                                                               92

 1   intervals?

 2                DR. STRANGE:   It's most easily

 3   demonstrated on the curves.

 4                DR. MOLITCH:   Which curve?

 5                DR. STRANGE:   Well, I demonstrated them.

 6   It's at 8:00, 8:30, 9:00, 9:30, 10:00, and then 11:00,

 7   12:00, 1:00, 1:30 --

 8                ACTING CHAIRMAN SHERWIN:   Maybe you could

 9   put that slide up?

10                DR. STRANGE:   -- around the meals.

11                DR. MOLITCH:   Yes, but see, that's my

12   point that your mean blood glucose of the level is

13   essentially around the time of the meals when we're

14   seeing the action of this drug.    We're then ignoring

15   about ten hours during the night when this drug is

16   probably not having very much activity when blood

17   glucose levels could be considerably higher.

18                DR. STRANGE:   Was a value determined at

19   12:00 in the evening?

20                DR. MOLITCH:   Yes, a single value.      But

21   we're talking about taking all of these values and then

22   dividing by 24 or whatever the number is.    So that, it's

23   not a true 24-hour day curve.

24                DR. STRANGE:   It's not this curve.      It's

25   the 64 trial.
26                Forward, forward, there.       Oh, you can
                                                              93

 1   actually not see the dots.    But you can see the

 2   indentations in the curve exactly where the sampling

 3   times are.

 4                DR. MOLITCH:    So that, it's very much a

 5   weighted curve towards the daytime rather than the

 6   nighttime where you probably are having much less drug

 7   action?

 8                DR. STRANGE:    It is.   It is a weighted

 9   curve centered around the meal peaks, but there are

10   values one hour before the beginning of a meal.   If you

11   note the very far left peak of them, you'll see -- if

12   you follow the -- you have a 30 minute value, a one hour

13   value, a one-and-a-half hour value, two hour value, and

14   a four hour value.   I think there's also a three hour

15   value actually, but a four hour value which is one hour

16   before the next dose at five hours following the first

17   dose.   So, you have a fairly good representation of the

18   whole profile.

19                I agree with you that you don't have a good

20   representation of the profile from 12:00 in the night

21   until 8:00 in the morning.    But if you look at the

22   curve, the numbers are so low on the repaglinide

23   profiles that, you know, it doesn't really matter.

24                DR. CARA:   But what if you look at the

25   glucose profiles?
26                DR. STRANGE:    Could you go forward one
                                                               94

 1   slide?   Next one.

 2                There are the glucoses in the night.     The

 3   change if you follow the green line -- the area we're

 4   now arguing about is this area from this data point to

 5   this data point which is only a data point there and

 6   a data point there.   If you look at the difference

 7   between that value and that value, either this way or

 8   that way is not going to influence the average of that

 9   curve over 24 hours to any big degree.

10                DR. MOLITCH:   Well, if you had the same

11   sampling interval over the course of that time that you

12   did in the morning, sure it will.   Then divide by the

13   total number of points, of course it will.       So, it

14   greatly influences --

15                DR. STRANGE:   Just let me understand what

16   you're saying.   You say that these eight hours of the

17   curve where we see a decrease of 20 milligrams per

18   deciliter or something -- you say that because we don't

19   have many values in there where people did not take

20   meals but slept in their bed, so that's going to

21   influence the average over 24 hours --

22                DR. MOLITCH:   Certainly.

23                DR. STRANGE:   -- to a great degree?

24                DR. MOLITCH:   Absolutely.

25                DR. STRANGE:   Okay.
26                ACTING CHAIRMAN SHERWIN:     It would give us
                                                            95

 1   the mean of all numbers, yes.

 2                DR. MOLITCH:    Part of the problem is that,

 3   we're getting back at this fasting glucose as using it

 4   as a measure of efficacy of this drug,    since the drug

 5   doesn't last overnight, presumably.      There was a

 6   suggestion in one of these slides earlier that maybe,

 7   in fact, with the four milligram dose that there may

 8   still be some drug levels out by the time of that fasting

 9   -- in the morning.

10                How do you think that this drug is working

11   to lower the fasting glucose levels?     What mechanism?

12                DR. WHISNANT:   Well, by the same mechanism

13   that for many, many years, people whose glycemic

14   control was provided by a single dose of insulin a day.

15   When patients get better glycemic control, their

16   fasting plasma glucoses in the morning are decreased.

17   We're not, you know, trying to say that we understand

18   the natural history of islet cell function enough to

19   know why that occurs.    That's an issue for a very

20   sophisticated analysis of diabetes biology.

21                The fact is that when you dose this agent

22   with a very small dose, the .5 milligram dose, where

23   within a few hours the drug is completely gone and the

24   insulin profile is back to meal related insulin

25   profile, that over four weeks you do get better glycemic
26   control of those patients as reflected in the fasting
                                                            96

 1   plasma glucose.

 2                ACTING CHAIRMAN SHERWIN:   Yes, I mean, I

 3   agree with you.    Clearly, the glucose levels are

 4   better.   The question was just to get a better

 5   understanding of how.    Because if you look at the

 6   fluctuations during the meal, depending on the graph,

 7   almost in each case the fasting glucose is

 8   substantially lower.    Then if you look at the

 9   fluctuations with the mean, the fluctuations are not

10   grossly different as compared to placebo.

11                So, it looks as if -- I mean, almost that

12   the drug is not -- it's supposed to be working during

13   the day to diminish meal induced fluctuations and yet,

14   the changes are not that dramatic during that period

15   when the drug is supposedly doing its major work.     So,

16   the question was sort of whether this drug might have

17   other effects that we don't totally appreciate?      Or

18   that its effects over time diminish with respect to meal

19   induced changes.

20                DR. WHISNANT:   We certainly hope that the

21   drug has other effects besides enhancing the

22   physiologic profile of insulin and therefore, has other

23   potential for, you know, long-term modifications of the

24   natural history of diabetes.    Obviously, the obvious

25   answer to the question of does this drug provide
26   glycemic control is that it does provide glycemic
                                                               97

 1   control over a year.    Unless we were doing something

 2   as a consequence of this enhancing physiologic insulin

 3   profile -- that is reflected in morning fasting plasma

 4   glucose -- we wouldn't get glycemic control over --

 5                ACTING CHAIRMAN SHERWIN:    No.    I guess

 6   the point is the uniqueness of the medication.       You

 7   know, it's not that it doesn't work.     You've

 8   demonstrated a change.    It's just is this different

 9   than any other drug, or does all these drugs improve

10   glucose control a little bit?      Fasting glucose

11   diminishes because glucose production has changed, and

12   then everything else we see is much the same.    No matter

13   how you get to that point, ultimately, the drug is

14   working like every other drug, or is there something

15   unique about this drug?   That's what I would -- that's

16   the point.

17                DR. WHISNANT:   We appreciate your

18   offering us futuristic advice about that.       If we

19   weren't here on an accelerated approval basis, maybe

20   we might have some more patients to show you.

21                ACTING CHAIRMAN SHERWIN:    Sure.    No, I

22   understand that.    No, I understand that point too.

23                Have you looked at, let's say, two days,

24   or one or two days when things are starting out much

25   the same way?   You know, if you don't look at, you know,
26   long-term effects, short-term effects in people with
                                                              98

 1   diabetes, are there differences then that are much more

 2   obvious?

 3                DR. WHISNANT:    Differences compared to

 4   other drugs, you mean?

 5                ACTING CHAIRMAN SHERWIN:     To placebo.

 6                DR. WHISNANT:    In short-term studies, the

 7   enhancement of the physiologic insulin profile is very

 8   clear over placebo, including an alteration of the

 9   morning fasting plasma glucose a day or even two days

10   after the drug has already been discontinued.      So,

11   there is effect on the biology of the pathology, if you

12   will, of diabetes but those kinds of studies have not

13   been included in this submission.

14                ACTING CHAIRMAN SHERWIN:     Okay.

15                DR. MOLITCH:    It seems that if part of the

16   unique action of the drug is its short activity, then

17   it clearly would lend itself to some sort of combination

18   with other agents that would have a longer action

19   overnight, such as a long acting sulfonylurea at night

20   with this drug during the day or long acting insulin

21   at that time with this drug in the morning.   Those might

22   be future studies that could be done unless you have

23   anything like that ongoing.

24                DR. WHISNANT:    I thank you very much.     If

25   you're available as a consultant, we'd be happy to
26   include you in the design of those studies.   I actually
                                                            99

 1   have with me, a variety of plans for future questions

 2   of this type, you know.    Because of the differential

 3   binding and because of the potency of this drug, does

 4   it actually rescue patients who are inadequately

 5   treated with sulfonylureas, for instance?      Does it

 6   rescue patients like the metformin patients?       I mean,

 7   that's one case that we've demonstrated already.     Will

 8   it rescue trivitisone failure patients?

 9                Those studies are in our designs, you know,

10   and at some point during this discussion we can talk

11   specifically about what suggestions you have for our

12   future with this drug.   We are also planning to present

13   here this morning, a three to four year study primarily

14   at least targeted toward providing additional safety

15   information regarding cardiovascular events, but

16   during which we will collect long-term natural history

17   data about this drug.    So, I mean, there are a number

18   of these kinds of questions that will be very helpful

19   to us.

20                ACTING CHAIRMAN SHERWIN:    With regard to

21   the issue you just spread out, you showed us data of

22   the total group and you showed us data, I think, of naive

23   patients which showed perhaps -- it looked to me like

24   a greater response relative to placebo.

25                DR. WHISNANT:   Yes?
26                ACTING CHAIRMAN SHERWIN:    I don't
                                                            100

 1   remember seeing sulfonylurea failure patients as a

 2   subgroup.   The question is, in people that fail in

 3   sulfonylureas, how do they respond to this drug?

 4                DR. WHISNANT:    Actually, the trial

 5   database includes a lot of those patients that you, as

 6   clinicians, might say are sulfonylurea failure

 7   patients.   I mean, we haven't used that word in our

 8   entry criteria for the trials.      But in the entry

 9   criteria particularly for the comparitor trials, we

10   have said patients who are inadequately treated on

11   other therapies, including sulfonylureas, many, many

12   of those patients, Dr. Sherwin, come in with HbA1c's of

13   nine or 10, or 11, or 12.   If you remember Dr. Strange's

14   distribution plot of the endpoints --

15                ACTING CHAIRMAN SHERWIN:     Right.

16                DR. WHISNANT:    -- of those HbA1c's, we're

17   talking about, in effect, failure patients.        In that

18   population of patients, we get a 1.6 to 2.9 delta HbA1c.

19   Admittedly, depending on the inherent responsivity of

20   the patient or where that patient is in the natural

21   history of the disease, we can show you very clearly,

22   data that say that not only did not do the naive -- that

23   is, not previously treated patients -- respond better,

24   but the patients who were previously treated still

25   respond although the mean response is not as great.
26                ACTING CHAIRMAN SHERWIN:    Do you think you
                                                          101

 1   could, at some point, just show us the actual data in

 2   terms of the failure patients?   I mean, the problem I

 3   guess is, a failure patient is not quite a failure

 4   patient.   Once you take the patient off the drug --

 5               DR. WHISNANT:   Right.

 6               ACTING CHAIRMAN SHERWIN:    -- usually they

 7   get a little worse.   So, if you're comparing that to

 8   a placebo, you'll see a difference.

 9               The question, I guess, is in patients who

10   are inadequately controlled who you then just compare

11   that to continuing on the sulfonylurea versus using

12   this drug instead, which is what you would do

13   clinically, do you have any data with regard to that?

14               DR. WHISNANT:   Well, actually, that's

15   inherent in the design --

16               ACTING CHAIRMAN SHERWIN:    In the

17   comparitor study, right?

18               DR. WHISNANT:   It's inherent in the design

19   of the comparitor trials.   Remember that the patient

20   population was recruited.   That over 80 percent of them

21   in those trials were previously treated patients.

22   Then they were randomized to either continuation of,

23   in many cases, a sulfonylurea or glyburide in a random

24   double-blind fashion.   So, we have data to show that

25   transferring patients to repaglinide who are those
26   kinds of previously treated, high HbA1c, long history
                                                               102

 1   patients were actually completely satisfactorily

 2   maintained on this drug for over a year.

 3                ACTING CHAIRMAN SHERWIN:      Right.   But

 4   there's not a difference between them and the

 5   sulfonylurea group, right?     So, it's hard to tell from

 6   the data.   That's why I'm trying to say clinically, if

 7   you have a patient who is not responding to a drug, often

 8   what you'll do is switch them to another drug and see

 9   if they do better.    I can't eke out from the data thus

10   far, you know, how that would sort out.

11                DR. HIRSCH:     That's really another

12   question.   I mean, if you're saying -- well, let me go

13   back a minute.

14                I assume that what we're talking about here

15   is a short acting oral agent, and that's a very

16   desirable thing to have presumably for the treatment

17   of, or prevention of hypoglycemia.     That's one set of

18   analyses which has --

19                DR. WHISNANT:     Yes, sir.

20                DR. HIRSCH:     -- been the brunt of what

21   you've presented --

22                DR. WHISNANT:    We'll show you some more of

23   that.

24                DR. HIRSCH:     -- and the major thrust of

25   that.   But the issue of whether this is now a kind of
26   rescue drug for those who have failed from other
                                                            103

 1   treatments, that would have to be presented differently

 2   so that we could analyze that specific issue and see

 3   how often that's the case and how useful that is.

 4                  Now, I'm assuming the thing works because

 5   A1c hemoglobins go down the same with this drug, unless

 6   this is a protein bound drug.      Unless it's some really

 7   weird thing that this affects the glycocilation of

 8   hemoglobin and I'm sure you've thought about that --

 9                  DR. WHISNANT:    Yes.

10                  DR. HIRSCH:     -- and ruled it out so that

11   the drug doesn't do anything of that sort.       But what

12   I'm most interested in now is the spontaneous

13   hypoglycemia kind of people, or the induced

14   hypoglycemia.     In their nature, do they have lower A1c

15   hemoglobins than the others?       Are these the group who

16   are trying very hard to manage themselves?     What is the

17   cost psychologically and in terms of compliance of

18   taking a drug three times a day rather than once a day?

19                  DR. WHISNANT:    Excellent questions.   Let

20   me try to address very briefly the first part of your

21   question.   Thank you for your support.

22                  The fact is, the comparitor trials were not

23   designed or carried out in such a way to show

24   superiority.     They were titrated to equivalence.

25   Therefore, the chance of showing that this physiologic
26   insulin dosing profile compared to a sustained long
                                                           104

 1   acting insulin secretagogue is different.      They

 2   weren't designed to show that.

 3                We didn't do comparitor trials randomized

 4   to a low dose versus a high dose of our drug, and a low

 5   dose versus a high dose of a comparitor drug, say,

 6   glucatrol XL.   We also didn't yet do trials to say if

 7   you take patients who are titrated to some submaximal

 8   dose of another secretagogue.      Say in that profile

 9   shown up there on that slide, two-thirds sulfonylurea

10   maximum dose and then you titrate in repaglinide as an

11   addition, what additional response will you get by this

12   different binding site, more potent if you will,

13   secretagogue?   And then as a final phase, to further

14   titrate those patients to maximum dose to see what

15   maximum effect we can really get.     That hasn't been

16   done.   We admit that.   So, we're happy to pursue those

17   kinds of second line, third line product expansion

18   kinds of questions based on what we've learned about

19   the drug.

20                With regard to the subpopulation that

21   might be sensitive to hypoglycemia, we actually have

22   some slides for the next session here.

23                Do you want me to go to those at the present

24   time?

25                Can I go to hypoglycemia section of the
26   presentation, toward the end?    My staff is doing a
                                                            105

 1   better job of finding slides than I am of explaining

 2   them, I think.

 3                  To answer your question specifically, Dr.

 4   Hirsch, in the 065 trial where we compared placebo one

 5   milligram and four milligrams, we actually looked at

 6   the frequency of hypoglycemia, percent of patients

 7   having hypoglycemia based upon their baseline HbA1c.

 8   Most diabetologists have seen this information and say

 9   "what's new?    We knew that all along."   Because the way

10   it turns out is that previously treated patients who

11   have a low HbA1c, say below seven or below eight, have

12   a pretty low frequency of hypoglycemia naturally and

13   not much of a hypoglycemia problem.        You know, we're

14   talking about 10 maybe 20 percent, one out of five

15   patients on the average of those patients would have

16   hypoglycemia.     A little bit of dose response in this

17   low HbA1c subset, but essentially no response in the high

18   HbA1c subset.

19                  On the next slide, I'll show you the same

20   information for naive patients where the percentage of

21   patients developing hypoglycemia is not only dose

22   related, but occurs much more in the low HbA1c patients

23   than in the high baseline HbA1c's.     So, one could

24   rationally design therapy, customize therapy, if you

25   will, for patients based upon their risk of developing
26   hypoglycemia which, by the way, is probably the same
                                                                106

 1   set of variables that predicts response.

 2                 DR. MOLITCH:    Show me that -- slide,

 3   please?

 4                 DR. WHISNANT:      Go back.

 5                 DR. MOLITCH:    Why is there a 15 percent

 6   risk of hypoglycemia in those with baseline hemoglobin

 7   A1c over ten with placebo therapy?          I'm not sure I

 8   understand that.

 9                 DR. WHISNANT:      That's a very strong

10   effort.

11                 DR. MOLITCH:    Why should it be any?     Even

12   in   the   nine   to 10 group,    you've     got substantial

13   --

14                 DR. WHISNANT:      Because if you follow

15   patients with diabetes on placebo, they report

16   hypoglycemia.

17                 DR. MOLITCH:    Which makes me wonder about

18   your criteria for hypoglycemia.

19                 ACTING CHAIRMAN SHERWIN:        Excuse me.

20   Would you be able to talk in the microphone?

21                 DR. WHISNANT:    I mean, I don't mean to be

22   simplistic but that's the only answer I know.        That if

23   you follow patients on placebo, they report

24   hypoglycemia.

25                 ACTING CHAIRMAN SHERWIN:        Let's get back
26   to the --
                                                            107

 1                  DR. MOLITCH:    That's getting to be

 2   worrisome.

 3                  ACTING CHAIRMAN SHERWIN:     -- definition.

 4                  DR. WHISNANT:    Right.   What's your

 5   definition.

 6                  ACTING CHAIRMAN SHERWIN:    That's one of

 7   the questions I wanted to get to is how do you define

 8   hypoglycemia?

 9                  DR. WHISNANT:    Hypoglycemia is reported

10   as any symptom where the patient interprets the symptom

11   as hypoglycemia or where the doctor synthesizes what

12   the patient has reported and checks hypoglycemia on an

13   adverse reaction form.    All hypoglycemias are reported

14   as mild to moderate unless the patient required

15   assistance for managing that event, in which case it's

16   reported as severe.

17                  ACTING CHAIRMAN SHERWIN:    Yes.   So, the

18   key to the question -- because obviously, the mild and

19   moderate hypoglycemia is extremely difficult to

20   quantify.    Consequently, in fact, the DCCT try to avoid

21   looking at those issues because of the difficulties in

22   quantifying.    What about the instance of severe

23   hypoglycemia requiring help?      Do you have data to look

24   at that specific issue?

25                  DR. WHISNANT:    Severe and --
26                  ACTING CHAIRMAN SHERWIN:    Severe
                                                           108

 1   hypoglycemia as defined by the DCCT was a patient

 2   requiring help by another person, glucagon, or a

 3   hospital admission.

 4                DR. WHISNANT:    Right.   Assistance

 5   required hypoglycemia did not occur with repaglinide

 6   in these trials.

 7                ACTING CHAIRMAN SHERWIN:     And how about

 8   the comparitors?

 9                DR. WHISNANT:    It occurred a number of

10   times that you can count on one or two hands, but it

11   was not a frequent event.    I think it was not a frequent

12   event in part because these were not intensification

13   kinds of trials.   They were trials where patients'

14   drugs were not pushed.

15                ACTING CHAIRMAN SHERWIN:     Now, the next

16   issue somewhat related to that is the data that we saw,

17   my impression was that most of the patients we saw had

18   fasting glucoses that were quite high and       generally

19   in a range, in fact, where beta cell responses are not

20   that good.

21                So, were there differences in response to

22   drug depending on the level of fasting glucose or

23   glycohemoglobin -- we're looking at percentages here,

24   but not absolute numbers.     So, were the majority of

25   patients here, do they have fasting glucoses over 150?
26   You know, I'm trying to get a sense of whether there's
                                                             109

 1   a difference in response depending on the level of

 2   glucose, or we're dealing with a homogeneous group with

 3   most of their fasting glucoses above 180, for example,

 4   to start out.

 5               DR. WHISNANT:     You're dealing, first of

 6   all, with a heterogenous group of patients who had

 7   relatively high blood glucoses based upon the inclusion

 8   criteria for the trials.    The patients were picked

 9   greater than 160 --

10               DR. STRANGE:    For an HbA1c less than 12.

11   For sulfonylurea treated patients HbA1c less than 12.

12               DR. WHISNANT:     We're talking about the

13   minimum fasting plasma glucose was 160, right?

14               DR. STRANGE:    What trial are we talking

15   about?

16               ACTING CHAIRMAN SHERWIN:      Well, I'm just

17   trying to get a sense because most of the data I saw

18   had mean fasting glucoses over 200.    So, it looked like

19   the majority of patients that were treated had very

20   poorly controlled diabetes.    We know that from other

21   drugs that that's the group that seems to have the best

22   responses for some ungodly reason, perhaps because

23   there's more play in the system.      I just wondered

24   whether people that were a little better controlled,

25   whether their response was somewhat different?
26               DR. WHISNANT:     So, you'd like to see a
                                                           110

 1   delta FPG versus baseline FPG in order to look at the

 2   correlation --

 3                 ACTING CHAIRMAN SHERWIN:    I'm just

 4   curious.    Yes.   In other words, I just don't have a

 5   good feel for, you know, the group.      My sense is that

 6   we're dealing with people that are very poorly

 7   controlled and we're seeing an improvement.

 8                 DR. WHISNANT:   Well, actually, the range

 9   of HbA1c's for the trial population, for instance in the

10   65 trial which is the trial we've already showed you

11   the distribution -- the range of HbA1c's was all the way

12   from seven, which is the lower limit allowed in the

13   trial, up to 12 which was the upper limit allowed in

14   the trial.

15                 ACTING CHAIRMAN SHERWIN:    Right.

16                 DR. WHISNANT:   The mean in most of the

17   trials turned out to be above nine.   So, we're dealing

18   with a relatively poorly controlled population of

19   diabetics.

20                 ACTING CHAIRMAN SHERWIN:    Right.   Now,

21   those patients have much poorer beta cell function in

22   general.    So, my question is, does the drug have

23   different effects when your beta cells are working

24   better?    You know, it could go either way.    I mean, I

25   just don't have a good sense of -- you know, I looked
26   at curves before with glucose levels extremely high and
                                                            111

 1   I saw some insulin responses to mixed meals.        But it

 2   may be that the responses are much greater and therefore

 3   -- I mean, then the question of hypoglycemic risk, you

 4   know, may be different depending on the status of

 5   disease you're dealing with and the starting out levels

 6   of glucose.    You know, in other words, in patients who

 7   are naive --

 8                  Now the diagnosis of diabetes has gone down

 9   so that now we're talking about a fasting glucose of

10   126.   Now, if you use this drug in a population with

11   a fasting glucose of 130 that has retained beta cell

12   function, is the response of the beta cells going to

13   be different to this secretagogue.       I mean, those are

14   the issues --

15                  DR. WHISNANT:    I'll help you out as much

16   as I can, Dr. Sherwin.     The facts that we know are --

17                  ACTING CHAIRMAN SHERWIN:     And I realize

18   that you have moved ahead quickly and I --

19                  DR. WHISNANT:    That's okay.

20                  ACTING CHAIRMAN SHERWIN:    -- realize, you

21   know, that -- it's not a criticism.     It's just that I'm

22   raising questions.

23                  DR. FLEMING:    Well, I think there are some

24   data to answer that question.

25                  DR. WHISNANT:    Yes, there are.
26                  DR. FLEMING:    In terms of the hemoglobin
                                                            112

 1   A1c, they have formally looked at an interaction with

 2   baseline A1c.    Now, can you pull this figure up?     I'm

 3   not sure how to tell you which one it is.

 4                ACTING CHAIRMAN SHERWIN:    What page is it?

 5                DR. FLEMING:    Oh, this is in part of their

 6   NDA submission.    I don't know whether you have it.

 7   It's figure 7.2 in your --

 8                DR. WHISNANT:    Well, actually, I can tell

 9   you -- I don't have the figure even in the back of

10   slides.   But I can tell you that the correlation

11   between HbA1c at baseline and HbA1c at conclusion of the

12   trial is almost one because really bad patients who have

13   very high HbA1c's get about a one to one-and-a-half delta

14   improvement with this drug.     Low HbA1c patients get

15   proportionately a little more than that but the

16   correlation coefficient is still very high.      And as you

17   would suspect.    I mean, that's why you're asking the

18   question.

19                That is, in part, reflected in the baseline

20   HbA1c data that I showed you for hypoglycemia.    It's the

21   naive patients with low HbA1c's who tend to be the most

22   responsive and therefore, have either relatively low

23   blood glucoses measured by meter readings, or who have

24   relatively rapid decrement in blood glucose, either of

25   which gives you symptomatology.
26                ACTING CHAIRMAN SHERWIN:    Yes, that would
                                                           113

 1   have been my guess.      So, the one thing that, you know

 2   -- if this drug is more potent in terms of the beta cell,

 3   in terms of that, I mean one of the issues will be to

 4   look carefully at the hypoglycemic risk in that

 5   subpopulation of patients.      Compared to some other

 6   drugs, you know, in terms of what the relative risks

 7   might be because it looked fairly high for the patients

 8   that were well controlled to start with.

 9                So, whether there should be a warning for

10   people with regard to type who are well controlled

11   already in terms of hypoglycemic risk.     I mean, I guess

12   that's sort of where I'm heading.

13                DR. CARA:     As a follow-up to that

14   question, you took patients that were in suboptimal

15   control, put them on therapy, and the goals of therapy

16   were still suboptimal.      No, really, okay?

17                DR. WHISNANT:     Okay.

18                DR. CARA:    I mean, do you have any evidence

19   to show what happens when you push those patients

20   further in terms of trying to get reasonable control,

21   i.e., fasting blood sugars in the 80 to 150 range in

22   terms of the incidence of hypoglycemia?

23                DR. WHISNANT:     What we do have is indirect

24   evidence to answer that question because we did

25   randomize placebo control blinded trial of a lower dose
26   versus a full dose.   So, for that heterogeneous patient
                                                           114

 1   population that includes some of those naive, low HbA1c

 2   patients, we know what the relative risk of

 3   hypoglycemia is relative to HbA1c and to dose.   You're

 4   absolutely correct that we've identified a subset of

 5   patients for relatively more responsive and have

 6   relatively more hypoglycemia.    It is those patients

 7   that we would suggest should be titrated starting at

 8   .5 milligrams.

 9                 Frankly, the rest of the patients who have

10   higher HbA1c's and who previously treated whether

11   controlled or not, have perfectly acceptable, normal,

12   almost placebo controlled levels of hypoglycemia.

13   That was on the two stack bar charts that I showed you.

14   So, we're dealing with a subset of patients who have

15   relatively high response, are sensitive to the drug,

16   have low HbA1c's at baseline and therefore should be

17   titrated carefully.    That's in the labeling.

18                 DR. CARA:   Another question related to

19   just the opposite phenomena maybe.    That is, have you

20   looked at the percent of patients that are not

21   responders?

22                 DR. WHISNANT:   Have we looked for --

23                 DR. CARA:   I mean, what percent of

24   patients do not respond to therapy?

25                 DR. WHISNANT:   We know -- yes, the shift
26   curve that we showed you for 65 actually shows you that.
                                                            115

 1   It shows you the distribution of final HbA1c's around

 2   50 percent of the patients.       It shows you what

 3   percentage of patients achieve seven or eight or nine

 4   at the end of a six month trial.

 5                Can we go back to that?

 6                DR. CARA:    But that's not really my

 7   question because you started out with patients with

 8   high glycohemoglobins to begin with.     It's hard to tell

 9   from that whether they ended up better or worse.

10                DR. WHISNANT:    Well, it's obviously hard

11   to take a patient with a 10 and make them a seven.    Most

12   drugs don't do that.

13                DR. CARA:   But if you take a patient with

14   a 14 and make them an 11, that's still a fairly good

15   response.   Whereas, if you look at it in the absolute

16   value, it's still high.

17                DR. WHISNANT:    Oh, we understand that.

18   The mean change in that study is about 1.6 to 1.8, and

19   actually goes up to 2.9 for the naive, highly responsive

20   patients, right?

21                DR. CARA:    Okay.

22                DR. WHISNANT:   So, on that shift curve we

23   showed you, if we can find that --

24                DR. CARA:    I just want a clear answer.

25                DR. WHISNANT:    Well, the answer is that
26   the decrement in people's glycemic control -- the
                                                               116

 1   endpoint in people's glycemic control does depend on

 2   where they start.     That's for sure.

 3                 DR. MOLITCH:    What percentage of people

 4   are non-responders stratified by baseline hemoglobin

 5   A1c?    At each hemoglobin A1c level, what percentage of

 6   people are non-responders?

 7                 DR. WHISNANT:    Non-responders as measured

 8   by --

 9                 DR. MOLITCH:    A change in hemoglobin A1c of

10   greater than a half percent, or less than half percent.

11                 DR. WHISNANT:    Oh, virtually everybody

12   changes by that much.

13                 DR. MOLITCH:    What's the number?

14                 DR. WHISNANT:    We'll get you the

15   distribution of numbers over lunch but --

16                 DR. MOLITCH:    Thank you.

17                 DR. WHISNANT:    -- that's an achievable

18   target.

19                 DR. MOLITCH:    Well, I'd like to see that.

20                 DR. WHISNANT:    Thank you.   That's an easy

21   one.

22                 ACTING CHAIRMAN SHERWIN:      Yes.   What

23   percentage of patients overall -- I mean, I know that

24   most of them start out high and therefore, didn't reach

25   that point -- reach the point of less than seven percent
26   hemoglobin A1c?
                                                             117

 1                  DR. WHISNANT:   We showed you that on the

 2   shift curve.

 3                  ACTING CHAIRMAN SHERWIN:      Percentage of

 4   the overall group?

 5                  DR. WHISNANT:   I can only show you by

 6   study.   I don't know for everybody.     But here's the 50

 7   percent -- this is cumulative frequency of patients.

 8   Fifty percent of the patients at the end of the study

 9   in the high dose were below eight -- below 7.9 actually

10   -- and for the lower dose group, it was a slightly larger

11   endpoint for 50 percent of the patients.        If you want

12   to know what fraction of patients achieved seven, it's

13   small.

14                  ACTING CHAIRMAN SHERWIN:      It looks about

15   20, 25 percent though.

16                  DR. WHISNANT:   I mean, it's a number but

17   remember that this is the distribution of patients --

18   this is the end distribution for the placebo patients,

19   but it really represents the distribution of all

20   patients at the beginning of the study because the

21   placebo patients didn't change.     Right?    A little bit,

22   but I mean, this is approximately the distribution of

23   the HbA1c's at the beginning.

24                  We can actually plot it that way if you want

25   to see that so that you can see, not on an individual
26   patient basis, but statistically as a group, that's the
                                                          118

 1   kind of magnitude that occurs at various levels of

 2   baseline HbA1c.

 3               ACTING CHAIRMAN SHERWIN:    From what you

 4   said before, if 20, 25 percent reach a level less than

 5   seven percent, my impression was that the percent of

 6   people that had hypoglycemia was about 50 percent or

 7   something like that from the bars that you showed?

 8               DR. WHISNANT:    No, it depends on the

 9   subset of patients that you're looking at.     Actually,

10   for naive patients who are sensitive to the drug and

11   have low baseline HbA1c's, patients who would start down

12   there somewhere below eight, those are responsive

13   patients and over half of them would develop

14   hypoglycemia to any drug.    So, you have to be careful

15   of those patients.   But for patients above an HbA1c of

16   eight and who were previously treated, probably later

17   in the stage of their disease, then the frequency of

18   hypoglycemia in that group is down on the order of one

19   out of four, one out of five.

20               ACTING CHAIRMAN SHERWIN:    Bob?

21               DR. KREISBERG:    This gets back to a point

22   that I was trying to make previously about the

23   differences in doses.   If you look at the cumulative

24   distribution in response to the one milligram and the

25   four milligram dose, I'm not impressed that there is
26   a substantial difference in the response, which gets
                                                           119

 1   into the issue of what particular dose the firm is going

 2   to recommend for the treatment of patients with type

 3   2 diabetes.

 4                  I believe that in your material, that the

 5   maximum dose is going to be four milligrams, four times

 6   per day.   But it looks to me like that you get most of

 7   what you're going to need without going to four

 8   milligrams four times a day.      Do you want to continue

 9   to recommend the four milligram dose?

10                  DR. WHISNANT:   The four milligram dose has

11   been shown, first of all, to be a safe dose given four

12   times a day.     In fact, one-fifth of the maximum dose

13   that we've studied in a group of patients, the four

14   milligram dose does give you an increment of

15   responsivity both for an individual patient and for

16   groups of patients.     I mean, you're already probably

17   up at ED80, ED90 or something like that when you pass

18   two.

19                  As a matter of fact, some of the dosing

20   discussions that we've been involved in would indicate

21   that for previously treated patients, particularly

22   those that are relatively lower risk of hypoglycemia,

23   that we probably should recommend those patients just

24   start on two milligrams and that's their dose.    Because

25   as you say, we're probably up somewhere on the dose
26   response curve.    It is true that the difference between
                                                         120

 1   one milligram and four milligrams is bigger for the

 2   naive patients, the more responsive patients.   But for

 3   previously treated patients, it could be that four

 4   milligrams is just an increment of both efficacy in

 5   terms of percent of responder patients as well as an

 6   increment for an individual patient.    We agree with

 7   that.

 8               DR. EDWARDS:    And the F90s are shown here

 9   --

10               DR. WHISNANT:    That's the difference in

11   dose for HbA1c over six months for the 65 trial.    So,

12   there's a difference.   But when you look at all treated

13   patients or when you take out this subset and look at

14   the previously treated patients, then you know, once

15   you pass one milligram, you've got a lot of the effect.

16   It's a very potent drug.

17               DR. HIRSCH:    I just think we shouldn't

18   factor out of the equation, the compliance issue and

19   assume automatically that if the A1c hemoglobin is low

20   that means that they're in earlier stage of the illness

21   or more sensitive or whatever.   There may also be the

22   issue of how compliant people are which hits to the

23   central issue of whether you want to try to solve the

24   problem by making a greater puzzle.   They have to take

25   something four times a day instead of once or twice a
26   day.
                                                         121

 1                So, you have no way, I suppose, of

 2   monitoring compliance or couldn't during these

 3   studies?

 4                DR. WHISNANT:   Well, we certainly agree --

 5   I mean, we do have a compliance number and we'd be happy

 6   to share that with you as we have with the Agency.

 7   These studies were based on a more than 80 percent

 8   compliance of doses based on pill counts, based on

 9   histories.   So, we know that probably if you don't take

10   the drug, it probably won't work as well and we give

11   you that.

12                We also give you the fact that in this

13   country, the standard care still reflects that most

14   patients have an HbA1c above nine -- most patients of

15   diabetes -- and that most patients don't have screening

16   eye exams and yearly HbA1c's and so forth.   So, I mean,

17   we actually understand that part of the reason patients

18   go into our trials with poor control is because of poor

19   care.   You know, we're not up to standard yet.    Our

20   company would hope that if we can -- 90 percent of

21   patients to more than 80 percent of the doses in the

22   65 trial.

23                We hope that we can actually turn around

24   this business of, you know, compliance being worse with

25   multiple doses of drug per day because if we tie a dose
26   of a diabetic drug to a glucose load at a meal, maybe
                                                               122

 1   the doctors will use that to teach patients about their

 2   disease and teach patients to do something about that

 3   peak of glucose with each meal.    That would be sort of

 4   our hope that we would actually contribute something

 5   to the natural history and care of patients with this

 6   disease.   It's not an approvability issue.

 7                DR. MARCUS:    Well, I think it is.   I wish,

 8   actually, there would be a little more emphasis on this

 9   issue because we actually approved -- or we recommended

10   approval of an altered insulin about a year ago,

11   specifically for the reason -- in large part for the

12   reason that it offered a degree of flexibility to

13   patients in terms of their timing of meals.        That is

14   a feature, I think a cardinal feature, of this agent

15   that has not even been mentioned today.        So, I think

16   it is worth a mention.     I think if a person is going

17   to skip lunch, or if a person is going to have an

18   additional meal, it does give you an opportunity to

19   interact with the effects of that meal, or lack of that

20   meal more efficiently than you could --

21                DR. HIRSCH:     But there's no data on that

22   at all.

23                DR. CARA:     But you haven't actually

24   determined whether you can actually do that with this

25   drug.
26                DR. HIRSCH:     That's correct.    We have
                                                           123

 1   data on normals, but no data on diabetics skipping

 2   meals.

 3                  DR. WHISNANT:   Actually, we are going to

 4   show you some data.

 5                  DR. HIRSCH:   Okay.

 6                  DR. WHISNANT:   We have a short session

 7   designed on, if you will, the dosing paradigm and the

 8   implications of dosing paradigm for skip a meal,

 9   different flexibility, and what the implications of

10   that are.   I will respond and say thank you very much.

11   We've actually been accused of being something called

12   the "oral humalogue".     You know, if that paradigm makes

13   sense to people who try to take care of patients with

14   diabetes, then we're happy to, you know, help with that.

15                  DR. NEW:   Actually, that was going to be

16   my question.    In the use of the rapid acting insulin,

17   in children at any rate, there's been a great

18   satisfaction reported by parents and children.       I'm

19   wondering whether your patients in the clinical trials

20   have reported a satisfaction of knowing that, you know,

21   if they're going to go out to dinner, they take the

22   Prandin 15 minutes before they're going to eat.      What

23   kind of a response do you get psychologically and

24   satisfaction-wise from that?

25                  DR. WHISNANT:   I can only give you
26   indirect evidence and a promise.      The indirect
                                                          124

 1   evidence is that the dropout rates in our trials,

 2   certainly versus placebo, were very satisfactory.    The

 3   promise is that we're building in quality of life

 4   assessments into the phase III BM4 trials that are being

 5   done.

 6                I'm sorry I can't answer anymore detail

 7   than that, but it's obviously a very important question

 8   in terms of the impact of a new therapy like this on

 9   care.

10                ACTING CHAIRMAN SHERWIN:    José?

11                DR. CARA:   You recommend giving the

12   Prandin 15 minutes before the meal.    What's that based

13   on?

14                DR. WHISNANT:   It's based on a study that

15   was done showing that there's no difference in the

16   plasma profiles if you dose 30 minutes before, 15

17   minutes before, or at the time of the beginning of the

18   meal.   I mean, obviously, we realize that a meal event

19   is not a one minute event --

20                DR. CARA:   Right.

21                DR. WHISNANT:   -- but you can mark in a

22   trial, the initiation of the meal.    Then you can do the

23   dose then, or 15 minutes before, or 15 minutes before

24   that.   So, we suggest that there's a 30 minute window

25   based on PK, not based on a therapeutic endpoint.
26                DR. CARA:   So, I mean, there's no basis to
                                                             125

 1   say that the patient can't take the medication

 2   immediately before eating?

 3               DR. DAMSBO:     It can be taken at the same

 4   time --

 5               DR. WHISNANT:     Yes, it can.   I said that.

 6               DR. DAMSBO:    -- show that taking it at time

 7   zero and time minus 15 and --

 8               DR. WHISNANT:     Is the same thing.

 9               DR. DAMSBO:     -- gave the same profile.   It

10   can be taken within 15 --

11               DR. CARA:     Same profile, okay.   I mean,

12   just in terms of compliance issues, I think that you

13   will get a lot more compliance if you say you can take

14   this immediately before eating versus taking it 15

15   minutes before and then having to wait 15 minutes.

16               DR. WHISNANT:     Oh, we understand that, Dr.

17   Cara, but you also very well -- I know you understand

18   that when you're carrying out clinical trials, you have

19   to specify how you want things done in order to get

20   consistency of data.

21               DR. CARA:     Sure.

22               DR. WHISNANT:     So, our answer to the

23   question is the vast bulk of the data were generated

24   on a minus 15 schedule for purposes of consistency and

25   clinical trials, but there is a PK trial that Dr. Damsbo
26   might be able to find the data on to show you that minus
                                                             126

 1   30, minus 15, and zero are the same.

 2               DR. CARA:      If you could find that, that

 3   would be nice.

 4               DR. WHISNANT:      We'll show you that.

 5               ACTING CHAIRMAN SHERWIN:       Mark, I'm

 6   sorry.

 7               DR. MOLITCH:     I have a question that's not

 8   really been addressed previously.      Viewing from a

 9   relatively simple clinician's point of view, one of the

10   concerns about using sulfonylureas in the past, over

11   the years, that's never been proven or disproven that

12   maybe their use helps to further exhaust the islet cell.

13   I don't know if that's true or not true.

14               You've made a lot about this lack of

15   insulin exocytosis.    What does this mean clinically to

16   us if we're going to be giving this to patients from

17   a mechanistic point of view?    Does this have any effect

18   on that controversy?    I mean, so what?    Why do I need

19   to know about that?

20               DR. WHISNANT:      Well, what we hope it means

21   is that a drug whose mechanism is as described would

22   not -- the correlation in pharmacology is dump.    We all

23   know about drug dumping.    You take a dose of drug.   You

24   know, because of a lipid meal or whatever, you know,

25   sometime later you get this big surge of drug and you
26   get a problem.
                                                         127

 1               What we hope is, because this drug is

 2   carefully modulated and as it approaches very low

 3   levels of glucose, we actually don't get any further

 4   release and you don't get any release that's

 5   independent of the channel mechanism.     Whereas, with

 6   the other drugs if, for instance, you keep pushing and

 7   pushing and pushing the glyburide dose, then at some

 8   concentration that Dr. Fuhlendorff actually showed

 9   you, the insulin will just release without regard to

10   the channel modulating the thing at all.

11               So, the goal for us is to have a drug that's

12   carefully modulated and to some extent, is modulated

13   based on the glucose profile itself.   So, what we hope

14   we can show you is a safer profile relative to

15   hypoglycemia frequencies, both in the elderly at night

16   and for the population as a whole.     We can also show

17   you in that analysis that there is substantially fewer

18   patients who have very low blood glucoses.     If you

19   count the number with BGMs below 45, the number is very

20   low compared to the comparitor population.

21               Have we proven that this lack of exocytosis

22   is directly related to a lower frequency or a lesser

23   severity of hypoglycemia?   No, we haven't proven that,

24   but it's consistent with what we know about the

25   mechanism of the drug.
26               ACTING CHAIRMAN SHERWIN:    Could you speak
                                                             128

 1   into the microphone please?       I'm sorry.   I apologize

 2   for the setup.   It's not ideal.      I realize that.

 3                DR. DAMSBO:     This is a study performed in

 4   type 2 diabetic patients given one milligram of

 5   repaglinide at the time zero, minus 15 or minus 30

 6   before a standardized meal.     As you can see, the area

 7   under the curves is equivalent, as well as the Cmax.

 8                DR. CARA:   But what happens with the blood

 9   sugar?

10                DR. DAMSBO:     The blood sugar is equally

11   reduced.   It's the same as the --

12                DR. CARA:   Do you have that data?     Do you

13   have the blood sugar profiles?

14                DR. WHISNANT:     Not here.

15                DR. DAMSBO:     Not right here, no.

16                DR. WHISNANT:    I mean, the curves in this,

17   minus 30, minus 15, and zero study of drug and of insulin

18   response, in response to that variability.

19                DR. CARA:   Right.    I mean, it makes sense

20   that the degree of insulin response is going to be the

21   same.

22                DR. WHISNANT:     Right.

23                DR. CARA:   The issue is whether the timing

24   of the insulin response is such that you need to take

25   the medication 15 minutes before the meal versus, you
26   know, zero minutes before the meal.
                                                           129

 1                DR. WHISNANT:     I understand.

 2                DR. CARA:   That's the issue.

 3                DR. WHISNANT:    Because of the meal related

 4   effect on insulin --

 5                DR. CARA:   Right.

 6                DR. WHISNANT:     -- added to our drug

 7   related effect on insulin, do the two added up change

 8   the overall therapeutic effect in those patients?

 9                DR. CARA:   Right.   And is there a greater

10   incidence of hypoglycemia if you take it just before

11   the meal versus 15 minutes?    I mean, those are the sorts

12   of issues that --

13                DR. WHISNANT:     We understand that, Dr.

14   Cara.   Unfortunately, this is not a therapeutic trial,

15   but we do, somewhere, have the BG response in those

16   patients.   Not for this trial.

17                For this trial?

18                DR. DAMSBO:     Not for this trial.

19                DR. WHISNANT:     Not for this trial.

20                ACTING CHAIRMAN SHERWIN:     Okay.

21   Hopefully, we've grilled you enough right now, I think.

22   Maybe you could take a break.

23                DR. WHISNANT:    We thank you very much for

24   your help, for your ideas.     We're prepared to move on

25   with the rest of the presentation, depending on what
26   Dr. Fleming would like to do.
                                                              130

 1                 ACTING CHAIRMAN SHERWIN:     Dr. Fleming,

 2   would you like to address the first question and then

 3   we'll take a break for lunch?

 4                 DR. FLEMING:    All right, very good.

 5                 ACTING CHAIRMAN SHERWIN:     Or do you think

 6   that it's going to be --

 7                 DR. FLEMING:    Yes, I think we might as well

 8   use the 15 minutes that we had counted on.

 9                 DR. WHISNANT:    Unfortunately, our

10   response to the first question is more than a 15 minute

11   response.

12                 DR. FLEMING:    Okay.

13                 ACTING CHAIRMAN SHERWIN:     Okay, that's

14   important.    What would you estimate your response to

15   the first question?

16                 DR. WHISNANT:    I don't know.    Probably

17   2:30, 3:00.

18                 ACTING CHAIRMAN SHERWIN:     You made me

19   nervous.

20                 DR. WHISNANT:    Actually, a good bit of

21   your -- I started to say questions, but maybe

22   interrogation is the right word.

23                 ACTING CHAIRMAN SHERWIN:     Right, right.

24                 DR. WHISNANT:    A good bit of this

25   discussion relates to this question and maybe we could
26   go faster based on that.
                                                           131

 1               ACTING CHAIRMAN SHERWIN:     Well, let's try

 2   to get it done in a half-hour.

 3               DR. WHISNANT:      Okay.

 4               DR. FLEMING:     Yes, I believe we have

 5   substantially dealt with many of the issues that could

 6   be covered under this point.    We're starting with this

 7   as the first discussion point because, after all, this

 8   was the probably most attractive feature of the drug.

 9   That is, the potential for reducing hypoglycemia while

10   achieving equivalent glycemic control.

11               Now, as you know, the studies were not

12   specifically designed to demonstrate a difference in

13   hypoglycemic potential or outcome, but we do have some

14   data that are encouraging.     It's clear that we do not

15   have definitive proof that in clinical practice,

16   particularly as patients are being more aggressively

17   managed, that they will also experience a reduction in

18   significant hypoglycemic episodes.     So, we would

19   propose that we look particularly at this question.

20   The company obviously has some data, some they've

21   already shown.   Perhaps we could deal with it fairly

22   expeditiously.

23               DR. WHISNANT:      Would you like us to move

24   on?

25               ACTING CHAIRMAN SHERWIN:     Yes, sure.
26               Would you prefer to come up here?     Maybe
                                                        132

 1   it would be better.   It's up to you.

 2               DR. WHISNANT:   Actually, I'm going to ask

 3   two of our staff to give a couple of prepared -- an

 4   introductory perspective about this question and then

 5   to provide some specific information with regard to the

 6   kinetic profile, as well as to the further data on the

 7   hypoglycemia consequences of that kinetic profile.

 8   So, I'll just introduce and let them go to the podium.

 9   Then it will be easier for them to see their slides and

10   so forth.

11               Dr. Wendell Cheatham is the medical

12   director for Novo Nordisk in Princeton in the American

13   affiliate office.   Dr. Cheatham is at home in

14   Washington where he was an endocrinologist for many

15   years until we stole him away.   He's going to provide

16   the introductory perspective on this question relative

17   to diabetes care.

18               Then Dr. Peter Damsbo who is the director

19   of clinical research in our Copenhagen office will

20   provide some data to help answer the question.

21               Dr. Cheatham?

22               DR. CHEATHAM:   Thank you, Dr. Whisnant.

23               Dr. Sobel, Dr. Fleming, Dr. Sherwin and

24   distinguished members of the Advisory Board, what I'd

25   like to do at this point is to set the stage for a
26   discussion of the clinical relevance of repaglinide to
                                                          133

 1   the questions that are being asked at this point.

 2                  I don't need to belabor the point that

 3   we've recently added two million additional

 4   individuals to the roles of people who have diabetes

 5   in this country.    Eighteen million individuals now and

 6   virtually all of those individuals who have been added

 7   have type 2 diabetes.     That gives us approximately 16

 8   million individuals with type 2 diabetes, but the most

 9   important point that we need to pay attention to, as

10   most of you who are educators and also have patients

11   with diabetes, is that less than half of these

12   individuals with diabetes are under any form of

13   therapy.   Beyond that, of the half that are under

14   therapy, less than half of those are under appropriate

15   therapy.   So, less than one-quarter of the patients in

16   this country with diabetes are being appropriately

17   treated for their diabetes.

18                  Another important point to keep in mind is

19   that more than half of the individuals with diabetes

20   in this country are over the age of 60, 58 percent, in

21   fact, or some 10 million individuals.      In another 12

22   years, for those of us who are part of the baby boom

23   population that will swell the ranks of those who are

24   in that age group, we're going to have some 24 million

25   individuals with diabetes particularly because of the
26   first point.     Some 64 percent of individuals by that
                                                           134

 1   time because of the growth in that segment of the

 2   population will be over the age of 60, some 15 million

 3   individuals in this country with diabetes.

 4                We know about the several year history of

 5   a goal for control of diabetes being at a hemoglobin

 6   A1c of seven percent or below.   The recommendation that

 7   intervention is definitely indicated when the

 8   hemoglobin A1c is at eight percent or above.

 9   Unfortunately, as our studies have indicated, just

10   based on our recruitment of individuals for trials and

11   also on population surveys and studies that have been

12   published, the best results that we can get for

13   hemoglobin A1c in this country, in a broad ranging

14   population of individuals with diabetes, is no less

15   than 9.1 percent which translates into an average blood

16   sugar of at least 200 milligrams per cent or above.   So,

17   although we talk about individuals in our trials being

18   poorly controlled, unfortunately, that's a profile of

19   diabetes in the United States.

20                Possible reasons for inadequate therapy?

21   Well, delayed diagnoses.    We know that that takes

22   place.   We know, in fact, that the average person with

23   type 2 diabetes is diagnosed some five to eight years

24   after they truly have developed the biochemical markers

25   of the disorder.   There's a low sensitivity to the
26   seriousness of the disorder not only in the patient
                                                         135

 1   population, but also in the practitioners to a large

 2   extent.   There's a fear of hypoglycemia with effective

 3   therapy and perhaps to some extent, we shouldn't

 4   necessarily stigmatize the practitioners for having a

 5   low sensitivity.

 6                But after all, if you can raise the level

 7   of blood sugar by 100 percent to 200 milligrams per

 8   deciliter and not impact symptomatology very much, but

 9   lower it by 20 percent below the given norm and

10   individuals have significant problems with the

11   symptomatology, then you would understand why

12   individuals perhaps aren't willing to necessarily jump

13   off and start treatment right away, especially with the

14   treatment armamentarium that we have available to us

15   at this point in time for individuals who are just

16   crossing the threshold into diabetes.

17                Fear of hypoglycemia is real because those

18   of us who are clinicians recognize that you always enter

19   a period of limbo between the time point when you've

20   diagnosed a person with diabetes, you've attempted diet

21   and exercise therapy and those have failed.     Their

22   target hemoglobin A1c is not being met but you know

23   through experience that if you start oral agents, and

24   in the case of our traditional oral agents, the

25   sulfonylureas, you're bound to have a high frequency
26   of hypoglycemia and severe hypoglycemia at that.
                                                         136

 1                  We have non-compliance which, of course,

 2   is a problem that we are attempting to impact.     I'll

 3   say something more about that a little later.    We have

 4   primary failure of medications.     Often, individuals

 5   have started on medications and there's a psychological

 6   comfort in putting a person on a medication and perhaps

 7   even lowering the hemoglobin A1c by a half or one

 8   percentage point, but still not achieving true control.

 9   Whether you want to call that primary failure, or

10   partial primary failure of course is a discussion of

11   verbiage.   We have secondary failure with individuals

12   who go on oral hypoglycemic agents, the sulfonylureas

13   traditionally, but we recognize that those drugs have

14   a duration of activity, or at least usefulness, if you

15   look at them critically for no more than approximately

16   eight years.

17                  The clinical dilemma then is one where we

18   know that we are to achieve near normalization of blood

19   glucoses.   That is the clinical aim to prevent the late

20   diabetic complications.     Indeed, the goal of seven

21   percent or below is a translation from the diabetes

22   control and complications trial, which although it

23   dealt with type 1 diabetes, we know at least that the

24   microvascular complications of that trial, we believe

25   -- scientific observation translates itself into
26   hemoglobin A1c.    In fact, population studies again
                                                           137

 1   relate reginopathy, neuropathy, nephropathy, and limb

 2   amputation directly to surveys of hemoglobin A1c.

 3                We've experienced an inability to reach

 4   these near normal blood glucoses with oral hypoglycemic

 5   agents, not necessarily due only to primary or

 6   secondary failure.   Often, it's a chosen under-dosing

 7   to avoid hypoglycemia because as was previously stated.

 8   Thus, we actually may be trading, in some cases,

 9   long-term complications because we want to avoid

10   hypoglycemia or we're dealing as well with a

11   psychological resistance to moving forward to the next

12   stage of therapy which would be insulin therapy.

13                At this point, I'm gong to turn the

14   discussion over to my colleague, Dr. Peter Damsbo, who

15   will discuss for you the clinical dilemma and the

16   application of our studies in practical terms to a

17   potential answer to the clinical dilemma at least at

18   one particular level.

19                DR. DAMSBO:   Thank you, Dr. Cheatham.

20                Ladies and gentlemen, I would like to start

21   out with the slide that Dr. Reit started out with this

22   morning which is an ADA statement saying "severe

23   hypoglycemia is the major complication of sulfonylurea

24   therapy.   Elderly patients are more susceptible to

25   hypoglycemia particularly when they have tendency to
26   skip meals or when renal function is impaired."
                                                           138

 1                Let me just after this, shortly summarize

 2   some of the pharmacology and pharmacokinetic resource

 3   we have as of now.   Dr. Fuhlendorff told us this morning

 4   about the reduced effect on insulin release at low blood

 5   glucose levels and at the same time, there was no direct

 6   exocytosis contrary to existing sulfonylureas.      From

 7   the pharmacokinetic beta percented by Dr. Strange, we

 8   had a short action of drug and an incident that reverts

 9   to the control levels.     Meal related dosing is the

10   concept that came out of it.     This with a

11   tablet-a-meal; no meal, no tablet.

12                I'll try to dig a little further into this

13   to illustrate the short action here on the next slide.

14   As you can see here, we have on the left lower panel

15   repaglinide profile.    This is the four milligram dose

16   given as a single dose.    You have the rapid absorption

17   and you have the rapid elimination of the drug.      So,

18   after two or three hours, there's hardly any drug left.

19   When it comes out to the lunchtime here -- these columns

20   here shows the breakfast and lunchtime -- there's very

21   little drug left.

22                This gives together with a meal rise to

23   insulin profiles like this.    The red one is the insulin

24   profile.   The green one is the placebo control.    It's

25   across all studies in the same patients.       As you can
26   see, there is a rise in the insulin and the insulin comes
                                                           139

 1   all the way back to the control level before the next

 2   meal.   That's a very good indication of a similar

 3   insulin profile at the time where you enter to the next

 4   meal.   That is that you have a short action on the beta

 5   cell.

 6                This is translated into a glucose profile

 7   that you see on the lower panel here at the right.   This

 8   is the placebo control.    When you have given four

 9   milligrams of repaglinide, you have a dose profile like

10   this.   As you can see, the following meal has this same

11   profile, so to speak.    It's just shifted downwards.

12   Actually, as you might also be able to see from this,

13   the increment in glucose is higher after the treatment,

14   indicating that the drug has stopped its action on

15   insulin.

16                The idea here is that you need insulin when

17   you eat.   To dip further into that one on the next

18   slide, we conducted a study where we looked at the fixed

19   and the mixed meal concept.    Could you give the drug

20   in a fixed way three times a day with three meals and

21   compared it to the group that has shifting meals, going

22   from two, to three, to four?    Will they be able to

23   obtain the same glycemic control over time?     It was a

24   three-week study in-house.     At the end of the study,

25   the patients were followed in a tight -- blood glucose
26   profile.   As you see here,t he red line is the two meals
                                                               140

 1   and two tablets.    The green line, three meals, three

 2   tablets, and four meals and four tablets.        As you can

 3   see, the profiles follow, so to speak, the dosing and

 4   the meals nicely.    The green one with the three here

 5   and the four meals was an extra snack given out here

 6   in the evening.

 7                 DR. CARA:     What's the dose?

 8                 DR. DAMSBO:    The dose here was -- the same

 9   dose for all patients was one milligram.

10                 DR. KREISBERG:     Are these normal people?

11                 DR. DAMSBO:    Those are diabetic patients.

12                 ACTING CHAIRMAN SHERWIN:      Was the earlier

13   data diabetes also?

14                 DR. DAMSBO:     These have a combination of

15   naive and sulfonylurea.       But    it's   a fairly mild

16   type --

17                 ACTING CHAIRMAN SHERWIN:      No, I meant the

18   previous slide.

19                 DR. DAMSBO:     Yes.   That was in type 2

20   diabetic --

21                 ACTING CHAIRMAN SHERWIN:       Also?

22                 DR. DAMSBO:     Yes.

23                 ACTING CHAIRMAN SHERWIN:       Okay.

24                 DR. DAMSBO:     So, when looking at this,

25   it's quite obvious that you can give the same.       You can
26   dose the drug either two, three, or four times with the
                                                            141

 1   meals and obtain the same glycemic control.

 2                 To then go a little further into the

 3   possibilities of using this drug in the setting when

 4   you omit a meal, we conducted a study in comparison to

 5   glyburide.    These were patients who were taken in.

 6   They were treated, titrated to the maximum effective

 7   dose.   After three weeks, they entered a stabilization

 8   period.    Those patients who had a blood glucose below

 9   145 or 140 entered the last phase.    Then the day that

10   this profile was made, lunch was omitted.     In the

11   repaglinide group, only two doses were given with the

12   two meals, and in the glyburide group two doses were

13   given as well before breakfast and before dinner.

14                 But now the patients omitted lunch and the

15   impact, as you can see here, is that the glyburide

16   group, which is the red one here, comes down.     Just

17   before lunchtime, the blood glucose keeps on going down

18   actually.    It goes further down here and it stays down

19   until the next meal comes and increase the blood

20   glucose.    In comparison to repaglinide, you have the

21   blood glucose that comes up, goes down, and stays at

22   the level of normal, around 90 milligrams per

23   deciliter, and remains down there.    The effect of this

24   on the safety side was that there was in the glyburide

25   group, six patients who experienced hypoglycemic
26   events during that afternoon, and no patients in the
                                                           142

 1   repaglinide group.

 2                 One other little detail I would like to

 3   draw your attention to is that as was mentioned earlier,

 4   I think by Dr. Molitch, was that what happens during

 5   night?   As you can see here, we have an increase during

 6   the night in the blood glucose of both the glyburide,

 7   but also even more pronounced by the repaglinide group.

 8   So, during the night, the blood glucose levels are

 9   higher with the repaglinide group than it was with

10   glybenlimide, although the glycemic control -- the area

11   under the curve, hemoglobin A1c -- was the same in these

12   two groups.

13                 So, then we dug down into the results from

14   the Phase III studies and tried to identify those

15   patients who had nocturnal hypoglycemia.    Out of those

16   who reported events, there was approximately 50 percent

17   who also -- no, these are only the reported events.

18   When we put them into the categories of either being

19   from 6:00 pm to midnight or midnight to 8:00 am, it

20   became pretty clear that the dark blue one, which is

21   the repaglinide, had more hypoglycemic events --

22   slightly more hypoglycemic events during the evening

23   time, but a lot fewer hypoglycemic events during the

24   nighttime.    This is 4.5 percent and the difference was

25   up to 16.5 percent with the glyburide.     So, this
26   reflects very well the action profile of the drug and
                                                          143

 1   the clinical outcome of it, namely the glycemia.

 2                We go to the next slide.    We also looked

 3   into the patients who actually measured their blood

 4   glucose when they had a hypoglycemic event.   As you can

 5   see here, it is divided into those who had a level which

 6   was less than a measured blood glucose less than 30,

 7   between 30 and 40, 40 and 50, 50 and 60, and so forth.

 8   The blue, again, is the repaglinide patients.      Those

 9   patients who then experienced a hypoglycemic event went

10   and took a blood glucose measurement.   You can see that

11   the glyburide curve, which would be something like this

12   -- there is a lot more reports on very low blood glucose

13   values.   The repaglinide curve is, so to speak, shifted

14   to the right.   This, again, is an indication that the

15   drug has shorter action and furthermore, that it

16   results in fewer low hypoglycemic events.

17                Let me shortly reiterate the slide that Dr.

18   Edwards showed earlier.    When we looked at all

19   hypoglycemic events, there was a tendency to a slightly

20   lower percentage of hypoglycemia.    Discontinuation

21   was fewer with repaglinide.   The blood glucose levels

22   with very low blood glucose levels was half that of the

23   other groups.   If we do make the same slide for the

24   elderly -- that is, the patients over 65 years -- you

25   can see there is hardly any difference to the existing
26   drugs out there.   Still, again, you see the marked
                                                             144

 1   difference on the patients who discontinued too to

 2   hypoglycemia and you also see the very marked

 3   difference on the patients who measured a blood glucose

 4   value which was lower than 45 milligrams per deciliter.

 5                  So, in order to summarize on the next

 6   slide, no reported hospitalizations, coma, or death,

 7   as Dr. Edwards reported earlier, was seen with

 8   repaglinide.     Severe reactions, assistance required,

 9   less often than comparitors.     We had fewer nocturnal

10   hypoglycemic events; discontinuations less often than

11   comparitors, and no increased frequency in the elderly

12   patients over 65 years.

13                  That would lead you to the conclusion on

14   the last slide here which says that "preprandial

15   treatment with repaglinide leads to significant

16   improved glycemic control, yet the risk of low blood

17   glucose values and severe hypoglycemic events is low."

18                  Thank you.

19                  DR. CHEATHAM:   Thank you, Dr. Damsbo.

20                  If I could just bring the clinical dilemma

21   now full circle, and Dr. Marcus, if I could just take

22   your question at that point?     Thank you.

23                  So, now, we have a situation in which

24   earlier diagnosis has been addressed at many levels.

25   Of course, we now have had new guidelines established
26   for the diagnosis of diabetes, hoping that by doing so
                                                            145

 1   we actually will shift the curve to a degree in regard

 2   to when diabetes is diagnosed.    Greater sensitivities

 3   for the need of diagnosis and treatment is being

 4   accomplished through patient education and also

 5   recognition programs that are being administered, both

 6   by governmental agencies and also by professional

 7   organizations.    But the search and continued

 8   improvement upon the idea of a potent and effective

 9   therapy with minimal impact on significant

10   hypoglycemic events continues.

11                  There's a suggestion, at least, by this

12   data that this particular agent, being one that does

13   admittedly result in some degree of hypoglycemia, but

14   it appears that when hypoglycemia occurs, it's a

15   forgiving hypoglycemia.    It may very well help to add

16   to the armamentarium that we have.    It's no doubt a

17   potent drug.     In dealing with primary failures and

18   secondary failures?    Well, that's another issue.     We

19   can't present any data today necessarily dealing with

20   that, although the suggestions that you have been

21   giving certainly would be ones that the company would

22   be very interested in pursuing because there may be some

23   observation there.

24                  At the end of the day, our concern in

25   diabetes as diabetologists is that we add to the
26   armamentarium to give more effective therapy.     That
                                                          146

 1   we're able to, even if it's bit by bit -- but in this

 2   situation, we believe as a company that we have an agent

 3   that expansively adds to the ability to treat

 4   individuals with type 2 diabetes with a potent drug with

 5   minimal risk of hypoglycemia, and hypoglycemia that

 6   when it occurs is minimal and individuals recover from

 7   it, at least presumably, without significant problems.

 8                  We also have anecdotal information.   A

 9   question was asked whether or not patients have

10   concerns about having to take the medication three

11   times a day.    Through our studies, we do have anecdotal

12   information that comes back from those individuals that

13   tells us that they like the idea of designing their day

14   and their meals to this particular dosing.     That data

15   has been collected and can be alluded to, although it's

16   not as hard as the clinical research data that you have

17   seen.

18                  In regard to the scattergram that you saw

19   in regard to responsiveness compared to the factor of

20   dosing level, I think it's important to keep in mind

21   that as I've pointed out, we deal with primary and

22   secondary failures.     Indeed, our patient populations

23   for that particular slide in regard to responsiveness

24   of glycemic control or change in glucose with

25   increasing dosage was not specified for a distinct
26   group of people.    In diabetes, we deal with people who
                                                           147

 1   exist all along the continuum of beta cell

 2   responsivity.   The natural history of type 2 diabetes

 3   is one in that the longer individuals have diabetes,

 4   the less responsive the beta cell becomes.       So, when

 5   you add any one particular dose and use that in a broad

 6   population of patients, you're bound to see varying

 7   responsiveness if you have not controlled for the

 8   duration of diabetes, or at least somehow predetermined

 9   the beta cell responsivity from the very beginning.

10               So, I think in addition to perhaps some of

11   the other explanations in regard to protein binding and

12   others, that also needs to be borne in mind in regard

13   to the responsiveness from dose finding.      And indeed,

14   again, in type 2 diabetes with the use of oral

15   sulfonylureas, we've trained to start with low doses

16   and work our way up, titrating individuals patients to

17   where we find the responsivity.       Because this drug

18   undoubtedly in low doses gives greater that 50 percent

19   responsivity at low doses.   What we reach for as we

20   increase the dose beyond that are the few people that

21   will respond to higher dosages because their beta cells

22   perhaps, if you will accept a simplistic finical

23   endocrinologist's suggestion, their beta cells need a

24   little bit more kick in the butt in order to release,

25   perhaps, a little bit more insulin.    But that certainly
26   isn't seen in the broad spectrum.
                                                             148

 1                 I'll stop here and answer questions.

 2                 Dr. Marcus?

 3                 DR. MARCUS:    Yes.    Thank you.

 4                 One of the worst examples of hypoglycemia

 5   that most of us have seen I think is the patient who

 6   has come in with profound glycemia because it has been

 7   mixed with an oral agent, or insulin has been mixed with

 8   alcohol.    One would not predict, since this drug

 9   apparently does not inhibit gluconeogenesis, that that

10   would be a particularly bad combination or at least it

11   would be better with this drug than it would be with

12   other oral agents.            Do you have any experimental

13   evidence in your preclinical data to look at an

14   interaction between alcohol or aspirin, for example,

15   and this drug?

16                 DR. CHEATHAM:     I just looked back to my

17   basic scientists and pharmacologists, but the answer

18   is no.    I am not aware of any and they tell me no.   That

19   certainly is very, very important and it again, becomes

20   something that a drug of this type lends itself to study

21   within.

22                 DR. WHISNANT:     But the clinical

23   correlate, sir, is that we've had no reports of coma,

24   loss of consciousness, hospitalizations, for that kind

25   of problem.    That's a serious kind of complication.
26                 DR. MARCUS:    Sure.
                                                            149

 1                 ACTING CHAIRMAN SHERWIN:     José?

 2                 DR. CARA:   Yes, you know, I need to

 3   reiterate the fact that a lot of the patients that you

 4   put in your trials were not in optimal control either

 5   at the beginning or at the end.      My concern is that as

 6   you pushed the envelope, so to speak, the incidence of

 7   hypoglycemia may, in fact, increase.

 8                 Have you looked at patients, for example,

 9   with glycohemoglobins of less than seven-and-a-half?

10   And looked at the incidence of hypoglycemia in those

11   patients versus patients with higher glycohemoglobins,

12   or at least done some sort of a scannergram where you

13   look at incidents of hypoglycemia versus

14   glycohemoglobin levels, for example?       I'm sure you

15   must have that data.

16                 DR. CHEATHAM:   Yes.    Yes, indeed, we do

17   have that data and I think Dr. Whisnant can speak to

18   that data directly.

19                 DR. WHISNANT:   It's in that set somewhere.

20   Keep going.    Keep going.

21                 There it is, Dr. Cara.     I think I showed

22   those two slides earlier in another context.

23                 DR. CARA:   But this is baseline

24   glycohemoglobin.

25                 DR. WHISNANT:   That is correct.     You want
26   it at the time of the hypoglycemic episode?
                                                               150

 1                 ACTING CHAIRMAN SHERWIN:      Yes, right.

 2                 DR. CARA:    While on therapy.

 3                 DR. WHISNANT:    I do not have that data.

 4   We could do you    an analysis of the nearest HbA1c

 5   proximate to the event, okay?      It wouldn't be exactly

 6   at the time of.

 7                 DR. CARA:    Sure.   No, but, let me make

 8   sure I understand this slide correctly.

 9                 DR. WHISNANT:    Okay.

10                 DR. CARA:    When you say baseline

11   glycohemoglobin, it's glycohemoglobin before entering

12   the study?

13                 DR. WHISNANT:    At the time of

14   randomization.

15                 DR. CARA:    So, it doesn't tell you

16   anything about the incidence of hypoglycemia while on

17   study drug.

18                 DR. WHISNANT:    It only uses a baseline

19   predictor.    That's all it does.      It tells the doctor

20   if the patient starts out with this relative level of

21   HbA1c, they are more or less likely to be one of those

22   patients who will be trouble.

23                 ACTING CHAIRMAN SHERWIN:      Maria?

24                 DR. NEW:    I just want to comment that the

25   changes in hemoglobin A1c from the beginning to the end
26   are not very great.      Therefore, this probably does
                                                            151

 1   relate to your question.       If somebody starts out with

 2   good control and a hemoglobin A1c below seven, do they

 3   have more or less hypoglycemia?      That's your question.

 4   And since --

 5                  DR. CARA:   No, my question is in the

 6   patient that is on therapy --

 7                  DR. NEW:    But the changed the hemoglobin

 8   A1c --

 9                  DR. CARA:    -- and responds with a

10   glycohemoglobin level to where the glycohemoglobin

11   gets more in a suitable target range of approximately

12   seven-and-a-half or below, do they have a higher

13   incidence of hypoglycemia?

14                  DR. WHISNANT:    What I can give you, Dr.

15   Cara, after lunch is the delta HbA1c in each one of those

16   subsets.   Because we know as some measure of response

17   within those subsets whether -- and in fact, it's pretty

18   much as you would predict, as I recall the table.          I

19   don't have it on a slide.      But as I recall the table,

20   it's these higher dose patients with lower HbA1c's --

21   and particularly in the next slide, please, -- the naive

22   subset of patients who have relatively lower HbA1c's.

23   Those are the patients who need to be titrated.      Those

24   are the patients where, you know, while we've subsetted

25   down to a relatively small number of patients but still
26   on a percentage basis, we're talking about, you know,
                                                         152

 1   a very substantial percentage of those patients need

 2   to be started on drug carefully at .5 milligram where,

 3   in fact, a lot of those patients are going to respond.

 4               DR. CHEATHAM:   And I would just add to that

 5   that after over 1,200 patients being studied with this

 6   particular agent, we have absolutely no evidence of

 7   severe hypoglycemia with the use of this agent.

 8   Although that's just a statement.    It's a statement

 9   from a clinical endocrinologist who would look at

10   something like that and say "maybe that means

11   something", and you experts, of course, would request

12   the data to look at that question further.

13               ACTING CHAIRMAN SHERWIN:    Yes, there have

14   been no patients that have required assistance during

15   all the trials?

16               DR. WHISNANT:   That is correct.

17               DR. CHEATHAM:   That is correct.

18               DR. NEW:   And the overall change in

19   hemoglobin A1c, if I took my notes correctly, is this

20   1.6 to 2.9 percent?

21               DR. WHISNANT:   That is correct depending

22   on --

23               DR. NEW:   That's from the beginning to the

24   end of the study for all your patients?

25               DR. WHISNANT:   Well, that's a range of
26   delta HbA1c's depending on the trial, depending on the
                                                             153

 1   subset of patients, depending on the dose.       It's a

 2   total range.    The least HbA1c delta that we saw was 1.6.

 3   As you would predict, that would be in previously

 4   treated patients, relatively resistant patients,

 5   patients, you know, with less responsiveness.     The 2.9

 6   number comes from the 065 trial looking at naive

 7   patients only.

 8                  ACTING CHAIRMAN SHERWIN:    Dr. Kreisberg?

 9                  DR. KREISBERG:    I wonder, will you

10   recommend this drug for all type 2 diabetic patients?

11   Let me explain why I've asked that question.

12                  DR. WHISNANT:    Okay.

13                  DR. KREISBERG:    There seems to be an

14   evolving concept, because of the continuum of type 2

15   diabetes as already referred to, is that when patients

16   are early in their disease or have relatively mild

17   hyperglycemia, insulin deficiency is not the primary

18   problem.   It is only as they get further in their

19   disease that that becomes a more important problem.

20                  Consequently, the use of drugs to treat

21   patients with type 2 diabetes indicates that we might

22   be more selective in targeting our drugs to the patients

23   using either insulin sensitizers or other types of

24   drugs that do not augment insulin secretion at the

25   beginning and other drugs that either augment insulin
26   secretion or actually involve the use of insulin later
                                                          154

 1   on in the disorder.   Where do you see your drug fitting

 2   in in this continuum?   Is that going to be part of your

 3   recommendation, or are you going to let the clinicians

 4   slide it in where he or she wants it?

 5                DR. WHISNANT:   The first objective is to

 6   assure you and the FDA that the drug is safe and

 7   efficacious based on the data that we submitted and

 8   therefore, approvable for market.    The next step in

 9   that process is to provide an expansion of data which

10   is consistent with what is going on in the diabetes

11   community with regard to using this drug as monotherapy

12   for patients who are unresponsive to diet and exercise

13   alone.   That's class labeling for these kinds of drugs,

14   although not yet approved for this drug.

15                But in the next stage, obviously, we'll be

16   looking at different ways of modulating therapy.     I

17   think the other tendency that's going on as opposed to

18   the early disease, you know, "let's manage it.   Prevent

19   as much as possible."   The other tendency that's going

20   on in the community is the addition of one therapy to

21   another.   We understand that our drug, you know, is

22   only one part of the armamentarium which primarily

23   relates to providing insulin on a, hopefully,

24   physiologic basis.

25                We have a study designed to compare this
26   drug alone to a sensitizer alone, versus a combination.
                                                            155

 1   We've already shown you data that says that this drug

 2   can be added to metformin and do a pretty good job of

 3   salvaging metformin unsatisfactorily-treated

 4   patients.   So, we're moving on to try to develop

 5   rational guidelines that will be consistent with

 6   exactly the direction that you're talking about.

 7                Do we have data in glucose intolerant

 8   patients showing that you can modify the natural -- we

 9   are not included in DPP.     Wish we were because we think

10   that's probably a logical kind of next step for

11   understanding whether or not our drug and its not new

12   mechanism, but very old mechanism -- I mean, you know,

13   this is physiologic insulin, enhancement of insulin.

14   We believe that our drug has a chance of being effective

15   in modulating the natural history conceivably, beta

16   cell sparing, in this disease and we don't have those

17   databases yet.

18                DR. CHEATHAM:     Right.   If I could just add

19   on to that, Dr. Kreisberg?      I think your question is

20   extremely appropriate in this day and time.

21                There is no question, undoubtedly, type 2

22   diabetes is a condition that usually coexists with

23   insulin resistance and relative insulin deficiency.

24   There are lots of clinical models, however, that would

25   suggest that the insulin resistance itself although it
26   may occur, does not become clinically apparent or at
                                                          156

 1   least does not cause clinical elevation of glucoses

 2   until there is relative decline in the ability of the

 3   beta cells to release insulin.              I think the

 4   question still is, which avenue do we need to effect?

 5   If we take care of one particular side of the scale,

 6   what are we missing out on on the other side of the

 7   scale?   I don't think there needs to be an argument back

 8   and forth in regard to whether we deal with insulin

 9   sensitizers or beta cell stimulators.    The bottom line

10   is that both of those defects exist in type 2 diabetes

11   and present themselves as the clinical problem.

12                Additionally, we still know that most

13   people with type 2 diabetes require some form of insulin

14   augmentation in order to achieve optimum control.

15   There is evidence that there is luring of hemoglobin

16   A1c's and people do better.   But if we look at the broad

17   spectrum of individuals who are treated with type 2

18   diabetes today, with all agents that are available, the

19   vast majority of people still require some degree of

20   insulin stimulation or insulin supplementation to

21   achieve the guidelines that we're looking for.

22                DR. WHISNANT:    Dr. Cara, an answer to an

23   earlier question.    It's an approximate answer if

24   you'll allow that.

25                The difference in delta HbA1c in patients
26   who have hypoglycemia versus those who do not have
                                                            157

 1   hypoglycemia is approximately a half-a-percent

 2   difference.

 3                 DR. CARA:   So, what does that mean?   The

 4   patients that dropped their glycohemoglobin more than

 5   half-a-percent have higher incidence of hypoglycemia?

 6                 DR. WHISNANT:   No.   It means that in this

 7   analysis of the data, that patients who demonstrated

 8   hypoglycemia on average, have a better HbA1c response

 9   than those --

10                 DR. CARA:   Oh, by half-a-percent.

11                 DR. WHISNANT:   By half-a-percent than

12   those who do not report hypoglycemia.

13                 DR. CARA:   And what sort of incidents are

14   we talking about?

15                 DR. WHISNANT:   Well, the kinds of

16   incidences that are on those --

17                 DR. CARA:   Okay.

18                 DR. WHISNANT:   For the population as a

19   whole, the number to remember is a fourth to a fifth

20   of the patients are going to have hypoglycemia.      Why

21   we've been through all this subsetting process is to

22   try to identify the contaminating part of that number

23   which is much higher, and therefore, requires special

24   management.    That's the goal of this subsetting

25   process.
26                 ACTING CHAIRMAN SHERWIN:    I have a
                                                             158

 1   question about the meals, skipping lunch study.       We're

 2   going to eat.    I promise.    Just a quicky.

 3                 First of all, the statement was six events

 4   occurred with glipizide or one of the other agents, or

 5   glyburide and none occurred.      I didn't know, first of

 6   all, what the n was overall.      Secondly, what's an

 7   event?

 8                 Third, my concern with that study is that

 9   the fasting glucose is about 15 to 20 milligrams per

10   deciliter lower in the comparitor group.     The level of

11   glucose is about 15 to 20 milligrams per deciliter lower

12   in the comparitor group during the interval with the

13   meal that's skipped.    Consequently, you could argue

14   that results are very similar.     So, you'd like to start

15   them off at the same level of glucose if you're going

16   to look at hypoglycemia in a well controlled

17   population.

18                 DR. DAMSBO:     That's absolutely correct.

19   It would have been ideal if they had started out at the

20   same level, but this is, again, one of the defects.    The

21   comparison was to glyburide.

22                 ACTING CHAIRMAN SHERWIN:    My concern is --

23   I mean, obviously, this is the key element to the

24   advantage.    Not that the drug isn't efficacious, and

25   the key study to show that really is hard to interpret.
26                 DR. DAMSBO:     It's hard to interpret from
                                                               159

 1   the point of view that they do not start out on the

 2   fasting blood glucose.      I agree with you that makes it

 3   a little -- it confounds the whole thing a little.        But

 4   if you look at the actual profile and the curve is up

 5   there again now -- if you look at the actual profile

 6   and if you look at the increment at the breakfast meal

 7   -- that is, if you move the red curve up these 15

 8   milligrams, right?

 9                 ACTING CHAIRMAN SHERWIN:       Right.

10                 DR. DAMSBO:    You would have a higher

11   increment, correct?

12                 ACTING CHAIRMAN SHERWIN:       Right.

13                 DR. DAMSBO:    And again, you will have a

14   higher increment at the dinner time and you will have

15   a lower value throughout the night.

16                 So, we can not overcome.       We can not do

17   both.    We can not have a short acting drug that, at the

18   same time, lowers the fasting blood glucose to the

19   level.    It's very difficult.      I would say that fasting

20   blood glucose is one split second of the diabetic's life

21   and does not really reflect the dynamics of the glucose

22   and insulin curves of the patients.

23                 ACTING CHAIRMAN SHERWIN:      And the events?

24   The event is level of glucose below a certain point or

25   is it symptoms?
26                 DR. DAMSBO:    Yes.    It was symptoms of
                                                             160

 1   which three of them were -- all of them were

 2   biochemically measured because it was an in-house

 3   study.   These events occurred, all of them, in the time

 4   period mentioned.    Three of them were below 45

 5   milligrams.    The rest of them were between 45

 6   milligrams and 55 milligrams.

 7                 DR. WHISNANT:    Some below 45, some --

 8                 DR. DAMSBO:   Yes.

 9                 DR. WHISNANT:    This might help the

10   perspective of the discussion a little bit in terms of

11   the kinetic difference.     Admittedly, this is not a

12   dynamic study, Dr. Cara.      This is a comparison of drug

13   levels, drug profiles, if you will, for our drug versus

14   two of the comparitors that we've studied.

15                 Just as a point of reference, this is how

16   those drugs look in terms oral dosing based upon the

17   recommended dosing.    Our drug would be given

18   three-times-a-day with those three curves down there,

19   and glyburide would be given once-a-day like that.

20   Glipizide would be given like that.     So, we're talking

21   about a very substantial contrast in the kinetic

22   profile of the different kinds of therapy.      I think it

23   is that contrast that Dr. Fleming is trying to get us

24   to address in terms of the implications of what that

25   means.   Perhaps not so much in terms of what happens
26   at the time of our peak, but what happens between our
                                                           161

 1   peaks and at night.

 2                ACTING CHAIRMAN SHERWIN:   Do you know

 3   anything about the binding characteristics of the drug

 4   to the K channel?   In other words, it's not just the

 5   drug level but how tightly the drug binds to its

 6   protein.   Because if it stays on the molecule, it's

 7   going to have a longer duration of action.

 8                So, my question is, how does the binding

 9   characteristics between this drug and the

10   sulfonylureas -- are they the same at different sites,

11   or obviously different -- or I think they're different.

12   Presumably, that might affect how long the molecule

13   stays on the protein.

14                DR. FUHLENDORFF:   It's a very difficult

15   question to answer because we tried to make the

16   experiment and we couldn't keep it on the receptor.

17   So, that was one thing.   So that tells me, at least,

18   that it's not a very long-lasting effect.

19                ACTING CHAIRMAN SHERWIN:   That's helpful.

20                Are we ready to eat before we all get

21   hypoglycemic?

22                MS. REEDY:   There is a table reserved in

23   Chatters for the Committee.

24                ACTING CHAIRMAN SHERWIN:   Okay.

25                (Whereupon, the meeting was recessed at
26   12:31 p.m., to reconvene later this same day.)
     162

 1

 2

 3

 4

 5

 6

 7

 8

 9

10

11

12

13

14

15

16

17

18

19

20
                                                             163

 1               A-F-T-E-R-N-O-O-N    S-E-S-S-I-O-N

 2                                                    1:35 p.m.

 3                 ACTING CHAIRMAN SHERWIN:     I would

 4   recommend that we begin.     I'd like to begin by thanking

 5   Dr. Marcus for his generous contribution to this

 6   meeting.    The lovely plate of brownies that we have on

 7   the back.   Because of ethical issues and the fact that

 8   we may be provided with too much nourishment during this

 9   meeting, I think, the rules are that we can only have

10   coffee for this meeting.      So, it was very nice of Dr.

11   Marcus to donate the brownies.       We appreciate that.

12   At least it keeps us from hypoglycemia during the

13   remainder of the meeting.

14                 Well, I think we're back and we should go

15   to question number two.      Dr. Fleming, I think you have

16   the podium.

17                 DR. FLEMING:    Thank you, Dr. Sherwin.

18                 Again, thank you, Committee members, for

19   your very helpful discussion.       I think that we have

20   touched on many of the discussion points already, so

21   we will be able probably to move through these in a

22   fairly expeditious fashion.

23                 Now, under this particular issue, I think

24   it is good to summarize exactly what the company has

25   done in its approach to develop this drug.       We are
26   basing our estimate of efficacy basically on three
                                                            164

 1   placebo controlled studies.    That's to make a

 2   distinction between the placebo controlled studies and

 3   those that involved an active comparitor.     We can

 4   derive some interesting information from those

 5   one-year studies involving active controls, but they

 6   are not meant to, in themselves, demonstrate efficacy.

 7                Now, we have the results of, for example,

 8   the next overhead from study 065.   This is showing you

 9   the mean hemoglobin changes from baseline at each week

10   visit.   Then the study 033, I don't have an overhead

11   for, but I think we should talk just a moment about the

12   one imbalance that was seen in the patients entering

13   the study.   There was a greater preponderance of

14   patients who were naive in the repaglinide group.       I

15   think, let's see, the comparison was 13 versus -- or

16   let's see, 23 percent versus nine percent in the placebo

17   group.   That might tend to make the patients treated

18   with repaglinide perform better.    I think it would be

19   useful if we looked at adjusted analysis where we look

20   only at the patients who were not naive to therapy.

21                I wonder, Dr. Whisnant, if you could

22   respond on that particular point?

23                DR. WHISNANT:    I'm happy to.

24                If we could just replace the slide that's

25   up with a slide from the projector?
26                Dr. Strange showed you this morning,
                                                         165

 1   ladies and gentlemen, this placebo controlled trial 33.

 2   The slide that he showed you was a comparison of a

 3   two-to-one randomization for approximately 100

 4   patients randomized in this trial.   This is the overall

 5   analysis that Dr. Fleming has referred to comparing the

 6   drug to placebo.    Remember that this was a

 7   double-blind, placebo controlled, randomized

 8   assignment of patients.    So, it's as controlled an

 9   observation as you can get.   Whether you think that the

10   increase in HbA1c for placebo controlled patients is as

11   it should be, or is more than it should be or whatever,

12   it is what it is.   The delta between the two groups is

13   as we reported, 1.8 percent HbA1c.

14                The next slide shows the corrected

15   analysis, if you will, adjusted analysis for this trial

16   removing the naive patients.    Sometimes

17   randomizations are not balanced and in this particular

18   case, the randomization came out to be 23 percent to

19   nine percent, as Dr. Fleming has indicated.     So, the

20   red line indicates that for 26 patients in the early

21   phase, we ended up with 14 patients up there who were

22   in that naive removed analysis -- the analysis with this

23   group of patients

24   -- versus an end study of 43 patients in the treated

25   patients.   As you can see with this previously treated
26   subset of patients, while this line doesn't drop quite
                                                           166

 1   as far as when you included the -- it turns out there

 2   were only three naive patients in that original subset,

 3   but the placebo patients still have their

 4   characteristic loss of glycemic control because

 5   they've been taken off of their therapy at the beginning

 6   of the trial.

 7                So, our answer to the question,

 8   respectfully, is that we still have placebo control

 9   studies demonstrating efficacy.

10                DR. FLEMING:    All right.    We can

11   certainly come back to placebo controlled trial

12   evaluation at any point, but then we might go on if there

13   are not questions about these particular studies

14   offhand.

15                DR. NEW:   I'm confused.     The difference

16   between the slide you showed before and this one is

17   what, exactly?

18                DR. WHISNANT:    Could we go back, please?

19                You'll notice that in the placebo group of

20   patients, the red patients, that there were 29 patients

21   at baseline and 17 patients at the end of the trial.

22                DR. NEW:   Right.

23                DR. WHISNANT:    Next slide.

24                You'll notice in this analysis, there were

25   26 patients at baseline and 14 patients at the end of
26   the trial.   What that means is that we have removed from
                                                            167

 1   the analysis, those patients who had not been

 2   previously treated with OHA drugs.       That is, those

 3   patients who are -- we've removed and what are remaining

 4   are the previously treated patients.       The naive

 5   patients have been pulled out and we're left now with

 6   a comparison of patients who had been previously

 7   treated with oral hypoglycemic agents and are now taken

 8   off therapy for this trial.

 9                 DR. NEW:   Okay.    So here's my question

10   then.    You only took three patients out that were naive

11   and the difference --

12                 DR. WHISNANT:     We took three patients --

13   sorry.

14                 DR. NEW:   Then the difference in the

15   treatment group -- never mind the placebo group -- is

16   what?

17                 DR. WHISNANT:     We took three patients out

18   of this group down here --

19                 DR. NEW:   Yes.

20                 DR. WHISNANT:     -- and we took 13 patients

21   out of that group up there because of the unequal

22   randomization.    Sorry, the other way around.

23                 DR. NEW:   So, the removal of 13 patients

24   gave you such a profound difference in the response?

25                 DR. WHISNANT:     Well, actually, the delta
26   for this subset analysis is in the same magnitude as
                                                          168

 1   the delta in the all-patient analysis.    If you just

 2   flip back and forth between the two slides, you'll see

 3   that --

 4               DR. NEW:    The ordinate is changed?

 5               DR. WHISNANT:    Well, we're talking about

 6   a difference between 1.1 down to minus .6, which is

 7   about 1.8, right?

 8               Next slide, we're talking about 1.7 down

 9   to about minus .1 which is 1.8.   It's in a different

10   position but the delta is the same.

11               DR. KREISBERG:    But there must be a

12   different significance.   I mean, if you look at the

13   treatment group, it's not different than zero.      So,

14   what you're basically saying now is that the drug

15   prevents any deterioration in glucose control, whereas

16   before, with the other previous figure, it showed that

17   there was actually a glucose lowering effect by

18   lowering the hemoglobin A1c below the baseline.

19               DR. WHISNANT:    We've actually shown that

20   in several studies, Dr. Kreisberg.    That's the nature

21   of this beast that we're dealing with.

22               DR. KREISBERG:    I understand that.    But

23   in this analysis, it basically shows no deterioration.

24               DR. WHISNANT:    Well, it shows rather

25   dramatic deterioration on placebo versus control of
26   that deterioration --
                                                            169

 1                DR. KREISBERG:    Right.

 2                DR. WHISNANT:    -- right?    So, whether you

 3   call that a maintenance effect or a therapeutic effect,

 4   it is still a difference between patients who are not

 5   receiving the drug.   I think it's an important word

 6   "distinction".    For relatively responsive naive

 7   patients, you see a much nicer reduction, or what you

 8   might call clinically a treatment, right, as opposed

 9   to for previously treated patients with higher HbA1c's,

10   you see more a control of or maintenance of the control.

11   We've seen that repeatedly, not just with this drug but

12   with a whole variety of other drugs.

13                DR. KREISBERG:    Right.     But because we

14   treat patients and not groups of patients, the effect

15   that a patient realizes from the drug is in their mind,

16   and I think in the physician's mind, does their glucose

17   concentration get better, not whether it stays the

18   same?   Whether it actually gets better is an important

19   distinction, even though I know the nature of the beast.

20                DR. WHISNANT:    Their glucose

21   concentrations actually get better than their overall

22   hypoglycemic control improves.     Their delta glucose

23   around mealtime is a measurable effect, even in

24   previously treated patients.    So that, the short-term

25   management on a day-to-day basis with this drug gives
26   you a different feeling when you're on the drug.
                                                           170

 1   Whereas, the long-term maintenance of glycemic control

 2   is obviously very different for these previously

 3   treated patients as opposed to good prognosis, highly

 4   responsive, naive patients.

 5                 DR. KREISBERG:    Except that the objective

 6   of all therapy is to get the hemoglobin A1c as low as

 7   you possibly can without producing hypoglycemic

 8   symptoms.   So, whether it stays the same is really not

 9   an important issue because the improvement in the

10   patient has to be reflected in the hemoglobin A1c which

11   reflects the mean glucose concentration.      So, keeping

12   it the same is not an advantage.

13                 DR. MOLITCH:     No, but it's just equally

14   potent to the prior sulfonylurea that they were using.

15   That's all.

16                 DR. KREISBERG:     Right.

17                 DR. MOLITCH:     That's all.

18                 DR. CARA:   So, does that mean that at time

19   zero, these patients were on treatment?

20                 DR. MOLITCH:     It's a two week wash-out.

21   That's all, isn't that correct?

22                 DR. WHISNANT:     Just a two week wash-out.

23                 DR. MOLITCH:     Which is really too short a

24   wash-out period.

25                 DR. CARA:   And then they just stopped the
26   ongoing therapy and their glycos went --
                                                            171

 1                DR. WHISNANT:    Right.

 2                DR. CARA:   Do you know what the blood

 3   glycos were at baseline?

 4                DR. WHISNANT:    8.5 mean.

 5                DR. CARA:   Mean, for both?     That's the

 6   mean for the whole group, right?       The whole group.

 7                DR. WHISNANT:    What do you mean the whole

 8   group?

 9                DR. CARA:   Before randomization.      If you

10   look at all the patients together, they had a mean of

11   8.5.   Then they're randomized so presumably, you get

12   the same number.    But in this case, we got an unequal

13   randomization and that was the reason for asking for

14   the repeat analysis.

15                DR. WHISNANT:    We may be, respectfully,

16   Dr. Kreisberg, back to your earlier question about how

17   we see this drug.   You know, for previously treated

18   patients who need more therapy, then maybe they either

19   need more of this drug or some other drug added to it.

20   So, we're into, you know, future studies kinds of

21   questions.

22                ACTING CHAIRMAN SHERWIN:      Cathy?

23                DR. CRITCHLOW:    Could I ask for an

24   interpretation or just a further interpretation of

25   figure 6.4 in our briefing document?        Which is
26   essentially the same phenomenon of Prandin versus the
                                                           172

 1   comparitors showing no decrease and, in fact, maybe

 2   slight increases in HbA1c.    Page 40.

 3                DR. FLEMING:    Yes.   That is actually on

 4   the next overhead.

 5                Mike, if you can put that up?

 6                This sort of leads into that point.      The

 7   same general idea that we are, if anything,

 8   deteriorating a little bit in control.    Again, I think

 9   we have to acknowledge that the particular design of

10   the study where the dose was fixed could explain part

11   of the fact that the glycemic control actually did

12   deteriorate over that period of time.      But this is

13   exactly what I wanted to bring up at this point.

14                Perhaps, Dr. Whisnant, if you'd like to

15   comment about why we have this result?

16                DR. WHISNANT:    Does this answer your

17   question?

18                DR. CRITCHLOW:    Yes.

19                DR. WHISNANT:    I think we have this result

20   because that's what we set out to show.      Not to be

21   clever about it, but let me, you know, really comment

22   on that.

23                If you look at the dotted line which is 047

24   trial, for instance, this is a compilation of the

25   European trials indicated down at the bottom of the
26   slide.   So, in effect, it represents 1,228 repaglinide
                                                         173

 1   patients and half that many comparitor patients.    So,

 2   if you look, for instance, at this group of patients,

 3   follow the dotted line, what you see is the study effect

 4   within a month or two because when you take better care

 5   of patients, you know, their glycemic control gets

 6   better.   This happens to be fasting plasma glucose so

 7   this is the ongoing monitoring of the study, monitoring

 8   of the patients.

 9                So, their fasting plasma glucose gets a

10   little better as you put them on study.   Then gradually

11   over time, this drifts either toward or slightly above

12   the number that they started with.    Whether it drifts

13   above the number that they started with depends on --

14   just can I stop one second?

15                Wong Chin, is this the attrition adjusted

16   data or is this all patients?    This is all European

17   trials?

18                Kristian, do you know?   It's all?

19                This is not attrition adjusted data.   So,

20   what you're seeing is all the patients were evaluated

21   at that point in time.   So, part of what you're seeing

22   here is a slightly different population of patients as

23   we go forward, right, and part of what you're seeing

24   is the natural history of this disease.     If you take

25   a group of patients who have a mean HbA1c of
26   eight-and-a-half or nine and whose FBGs are not very
                                                             174

 1   well controlled, then their natural history if you

 2   follow them -- well, certainly, their natural history

 3   if you follow them on placebo is that they go up like

 4   that.    They go way up.   That's the problem in doing

 5   placebo control trials, as you can imagine.       But their

 6   natural history on any insulin or insulin-like, or

 7   insulin secretion therapy drifts just like that.

 8                 DR. CRITCHLOW:    So, if a patient required

 9   insulin during the course of the trial, are they in the

10   ones that are withdrawn due to inadequate --

11                 DR. WHISNANT:    They dropped out of the

12   trial.

13                 DR. CRITCHLOW:    Okay.

14                 DR. WHISNANT:    And the dropout rate was on

15   the order of 30 percent for both arms.

16                 DR. KREISBERG:    There's just a slight

17   discrepancy in the slide.      It says hemoglobin A1c, but

18   you talk in terms of millimoles of glucose, right?

19                 DR. WHISNANT:    Sorry.    Well, that's the

20   wrong -- let's see.

21                 DR. KREISBERG:    They could be the same,

22   actually.

23                 DR. WHISNANT:    This is fasting plasma

24   glucose.    It's the wrong heading.

25                 DR. KREISBERG:    Okay.
26                 DR. WHISNANT:    Okay?    I know that the mean
                                                          175

 1   HbA1c would be down there, right, nine -- a little more

 2   than nine.   So, this is millimoles.

 3                DR. MARCUS:   Well, we're told that this is

 4   a problem with this being a fixed dose study, but how

 5   many of these patients or in how many of these studies

 6   was the dose fixed at four milligrams three times-a-day

 7   which meant that sure, it's fixed dose but you can't

 8   go higher?   It's a max dose.

 9                DR. WHISNANT:   The answer is about half.

10   About half the patients failed to achieve a

11   satisfactory FBG at one of the lower doses and

12   therefore, completed the titration scheme all the way

13   to four milligrams.

14                DR. MARCUS:   I wish I felt more

15   comfortable with the assertion that this is the nature

16   of the beast.   I'm not a diabetologist, but I do see

17   them in our clinic and I have to say we have people who

18   maintain adequate hemoglobin A1c's that is seven or

19   below long-term without a sign of a drift.   Now, I don't

20   know.   I'd like to get maybe Bob Sherwin or Kreisberg

21   or someone has --

22                DR. HIRSCH:   Let me just ask a question

23   before.   Can you just help me to understand the slide

24   that I'm looking at?   I can't see that one, but this

25   is page 40 which I think is the one that was referred
26   to.
                                                              176

 1                DR. MARCUS:    Yes, that's it.

 2                DR. HIRSCH:    What that shows is that

 3   everybody on the drug seems to be getting a slow drift

 4   upward of A1c hemoglobin as compared to the slide we saw

 5   before in which the treated group seemed to be drifting

 6   a little bit downward from a zero position.          This is

 7   a percent change.   I'm sorry, it must be a different

 8   study or something.    I'm confounded now.

 9                DR. FLEMING:    You know, I think I am the

10   culprit here in that I have taken the wrong figure,

11   which may not actually be in your book, and put the right

12   title over it.   If you'll excuse me for that.

13                Can you pull up figure 64.

14                DR. HIRSCH:    Is that the one that's on page

15   40?

16                ACTING CHAIRMAN SHERWIN:       That's

17   glycohemoglobins.

18                DR. FLEMING:    Okay, well, it's the same

19   thing.

20                DR. HIRSCH:    There you go.    But whichever

21   it is, I don't understand it.      I mean, I see what it

22   says, but I don't understand.    The figure we saw before

23   this five or ten minutes ago showed everybody sort of

24   drifting downward in A1c hemoglobin.

25                DR. WHISNANT:    Over three months, sir.
26                DR. HIRSCH:    Over three months?
                                                              177

 1                DR. WHISNANT:     Right.

 2                DR. HIRSCH:   I see.      So, in all instances

 3   if you keep it going a year, it looks like in this

 4   situation, you're worse off than when you started with

 5   A1c hemoglobin.

 6                DR. WHISNANT:     I'm sorry.     We showed you

 7   two sets of data.   The initial data that I showed you

 8   the corrected analysis for was the 33 data which was

 9   a 12 week study -- sorry, 16 week study.

10                DR. HIRSCH:     So, this is the best one for

11   time.

12                DR. WHISNANT:     And that's the year data

13   that you're looking at now.    You see a couple of things

14   from the slide that's in front of you.

15                DR. HIRSCH:     Well, I can't.    I mean, I

16   apologize.   You'll have to tell me because I can't see.

17                DR. WHISNANT:     Okay, all right.

18                DR. HIRSCH:     I'd have to look in the book

19   because I can't see the slides at all.        So, go ahead,

20   just tell me about it and I'll try to understand it.

21   They're invisible to me.

22                DR. WHISNANT:     Okay.    From the slide

23   that's in your book --

24                DR. HIRSCH:     Good, page 40.

25                DR. WHISNANT:     It's also the slide that's
26   in front of the group.
                                                           178

 1               DR. HIRSCH:     Good.

 2               DR. WHISNANT:    The top line you see is the

 3   US trial, the little dotted line up on top?

 4               DR. HIRSCH:     Yes.

 5               DR. WHISNANT:    What that shows is that in

 6   the United States, we start with a patient population

 7   that's in relatively poor control.      Their HbA1c's are

 8   approaching nine.    As Dr. Cheatham showed you before

 9   lunch, that's where we are in this country with treating

10   type 2 diabetes.

11               DR. CRITCHLOW:    Excuse me.   These are all

12   repaglinide treated patients, everybody on this slide?

13               DR. WHISNANT:     Those are repaglinide

14   treated patients.    That is correct.

15               DR. CRITCHLOW:     What I didn't understand

16   when I originally asked the question was, it said

17   something to the effect of differences between that and

18   the comparitor.    So, maybe you could just explain what

19   -- so this is just over time, the HbA1c levels?

20               DR. WHISNANT:    This is HbA1c for each of the

21   five trials and the European patients tend to be better

22   taken care of or better treated or whatever, or they

23   don't eat as much or whatever.

24               DR. CRITCHLOW:    Then what is going on with

25   the patients in the comparitor arms?
26               DR. WHISNANT:     The same.
                                                           179

 1               DR. HIRSCH:     All we see here are the

 2   repaglinide, is that correct?

 3               DR. WHISNANT:     It's the same.

 4               Give me primary 49 because the lines are

 5   right on top of each other.    On a per study basis, the

 6   lines are right on top of each other, yes, sir.

 7               DR. ILLINGWORTH:     Do you have data on the

 8   same slide of body weight?     Did this progressively

 9   increase?

10               DR. WHISNANT:     I commented on body weight

11   in the comparitor trials this morning.      For the

12   repaglinide patients in the comparitor trials, on

13   average, it's about half-a-kilo weight loss over the

14   year versus the comparitor.   For some studies, for some

15   comparitors, it's a kilo of weight gain; for others,

16   it's half a kilo weight loss.     So, it's in the same

17   order of magnitude.

18               Why did you ask the question?

19               DR. ILLINGWORTH:     It's a confounding

20   variable if the glycemic control varies.

21               DR. WHISNANT:     Yes, okay.   I did not do a

22   covariant on weight for glycemic control.

23               There's the two drugs.     At least in the US

24   trial, the repaglinide -- and here, again, is that three

25   month study effect.   Then gradually, over time, they
26   go back to about where they started.   I'm reminded that
                                                            180

 1   there's a number of natural history studies that some

 2   of us in the room are aware of that show the

 3   deterioration is on the order of a few tenths of an HbA1c

 4   percent per year in natural history studies that have

 5   been done, including UKPDS, et cetera.

 6                DR. ILLINGWORTH:    One other question

 7   related to that issue is do you have data on compliance

 8   to the drugs where the patient is more compliant in the

 9   first three months and then became more complacent as

10   they were on therapy longer, based on pill counts or

11   --

12                DR. WHISNANT:    The compliance rate

13   overall was a number that Poul gave me this morning,

14   80-some percent --

15                DR. STRANGE:    In the 49 trial, 93 percent

16   of the patients took at least 80 percent of the drug.

17                DR. WHISNANT:   Eighty percent of the drug

18   was taken by 93 percent of patients overall, but that's

19   not the question you're asking.    You want it by month

20   or something on that order?

21                DR. ILLINGWORTH:    By time, yes.

22                DR. WHISNANT:    We don't have it here, but

23   we can go into a database and look it up.

24                DR. ILLINGWORTH:    Because that may

25   explain some of your variations in control.
26                DR. WHISNANT:    It's worth looking at.
                                                              181

 1                  DR. KREISBERG:    Dr. Whisnant?

 2                  DR. WHISNANT:    Sir?

 3                  DR. KREISBERG:    You have similar type of

 4   data on your patients who were naive.         I believe that

 5   they lost --

 6                  DR. WHISNANT:    Yes, sir.

 7                  DR. KREISBERG:    -- their hemoglobin A1c

 8   dropped by over two percent.       Do I recall that right?

 9                  DR. WHISNANT:    The largest decrement was

10   2.9 percent in a naive subset in protocol 65.

11                  DR. KREISBERG:    Do we have a follow-up of

12   them over a 12 month period of time?

13                  DR. WHISNANT:    Yes, sir.    We have a slide,

14   right, for the naive subset in 49?          We have six month

15   data, but we don't have the 12 month data.

16                  In the six month trial, I think actually

17   we showed you that subset --

18                  DR. KREISBERG:    You may have.

19                  DR. WHISNANT:    -- analysis this morning.

20   What happens is that the naive subset of patients get

21   a nice what you would call "therapeutic response."

22   Their HbA1c's go down over three to six months.         Then

23   if you follow that naive subset over the subsequent

24   year, then they drift to about half as high as they were.

25   They lose about half of what you gained when you were
26   starting them on therapy.       I believe that is also --
                                                            182

 1   sorry, this is not an excuse.   It is my perception that

 2   that is the general experience with, if you'd want to

 3   call it the natural history.    But that's the general

 4   therapeutic experience with these kinds of drugs.

 5               DR. FLEMING:    Well, just summarizing, we

 6   can see that there is a subacute effect on the order

 7   of 1.6 or more percent hemoglobin units.   But depending

 8   on the patient population, this can be considerably

 9   less, particularly over a period of time.     We do not

10   have a study that formally demonstrates durability,

11   though these active comparitor studies at least give

12   us a look at a single fixed dose over a one-year period

13   of time.

14               I wonder if there are any other points to

15   be made about efficacy before we move on?

16               DR. HIRSCH:    Yes, I had -- I wonder if you

17   or any member of the industrial group could just

18   succinctly give me the best evidence for a difference

19   in hypoglycemia.   I'm confounded now by all the

20   different kinds of things that are going on.     But

21   what's the single best piece of information that with

22   this drug, there's going to be less hypoglycemia in one

23   year than with any other drug?

24               DR. WHISNANT:    Meaningful hypoglycemia

25   difference, I believe is best demonstrated by patients
26   who have meaningful hypoglycemia.   What we've shown is
                                                           183

 1   that there's one-third -- one-half to one-third

 2   depending on the trial as many patients discontinue

 3   therapy for reasons of hypoglycemia, number one.

 4   We've shown that in the repaglinide treated patients

 5   versus glyburide treated patients in long-term

 6   treatment trials, that one-fourth as many patients have

 7   blood glucoses lower than 45.   We've also shown that

 8   there were no serious events, including

 9   hospitalizations and coma, with our drug compared to

10   a countable number, not a statistical, you know, huge

11   number with the comparitor drug.

12               So, we would --

13               ACTING CHAIRMAN SHERWIN:    What is the

14   countable number?   Because you haven't given us a

15   countable number.

16               DR. WHISNANT:   It's a few.    I don't

17   remember.

18               DR. HIRSCH:   The one-fourth who were lower

19   that 45 -- there was a ratio of one to four you said.

20               DR. WHISNANT:   One to four.

21               DR. HIRSCH:   Okay, and the total number

22   that we're talking about who would achieve this less

23   than 45 -- how many?

24               DR. WHISNANT:   Eight percent of 1,200

25   versus 33 percent of --
26               DR. HIRSCH:   Eight percent versus --
                                                            184

 1                DR. WHISNANT:     No, sorry, that's not quite

 2   right.   It's eight percent of those who had BGM

 3   measurements which is 50 percent of the 1,228 people

 4   had BGM -- 50 percent of people with hypoglycemia

 5   reported BGM measurements.     Of those, eight percent of

 6   those versus 33 percent of those had blood glucoses

 7   lower than 45.

 8                DR. HIRSCH:     Oh, so it's eight percent, 33

 9   percent versus what percent?

10                ACTING CHAIRMAN SHERWIN:      Could you get

11   that table up?   Maybe that would help us a little bit?

12                DR. WHISNANT:     You want to see that table

13   again?

14                ACTING CHAIRMAN SHERWIN:      Yes.   We have

15   many times, but --

16                DR. WHISNANT:     Okay.

17                ACTING CHAIRMAN SHERWIN:      The other

18   point, you know, regarding these issues is I haven't

19   heard anything about statistics throughout.       I mean,

20   had there been a statistical analyses done by the FDA

21   of these data in terms of --

22                DR. FLEMING:     We certainly have examined

23   the efficacy data formally, but we've not applied a

24   biostatistical analysis of the safety data as of yet.

25                ACTING CHAIRMAN SHERWIN:      Yes, because I
26   think that's critical in terms of actual, you know, what
                                                           185

 1   we're saying here.    I mean, clearly, we need to know

 2   what's statistically significant and what isn't.

 3               DR. WHISNANT:     Dr. Sherwin, here's some

 4   hypoglycemia for you.     This is the elderly subset of

 5   patients, okay?   The subset where it's clinically more

 6   of a problem, if you will.     Of 343 patients exposed,

 7   16 percent had hypoglycemia and one-and-a-half percent

 8   discontinued for reasons of hypoglycemia.

 9   One-and-a-half percent of 343 as opposed to 4.6 percent

10   of 131.

11               DR. HIRSCH:     As opposed to, let's see --

12   16 percent as opposed to --

13               DR. WHISNANT:     Sixteen percent of total

14   versus 18 percent of total.

15               DR. HIRSCH:     So, they had the same amount

16   or whatever, the same reported hypoglycemia in the two

17   groups.

18               DR. WHISNANT:     These two numbers are

19   probably not different.

20               DR. HIRSCH:     Okay.

21               DR. WHISNANT:    These two numbers probably

22   are different.    Okay?

23               Of the patients with hypoglycemia, 45

24   percent of them had meter readings.    Eight percent of

25   the meter readings were below 45 milligrams percent as
26   opposed to 33 percent of the 63 percent who had meter
                                                              186

 1   readings in the glyburide treated patients.

 2                  Now, let me hasten to add --

 3                  ACTING CHAIRMAN SHERWIN:   This is very --

 4   I must say it's very confusing to sort out actual

 5   numbers.

 6                  DR. HIRSCH:   It's even worse if you have

 7   to look around this way.

 8                  ACTING CHAIRMAN SHERWIN:    Yes, Jules,

 9   would you like to -- I mean, we could put it --

10                  DR. NEW:   Jules, you can put a chair here.

11                  ACTING CHAIRMAN SHERWIN:    Yes.   I

12   apologize for the room.      I don't think the government

13   could afford a larger room.

14                  DR. WHISNANT:   Sixteen percent of 343

15   reported hypoglycemias.

16                  ACTING CHAIRMAN SHERWIN:   So, it's like 60

17   patients or something like that.     Less than 60.    About

18   50 something.

19                  DR. WHISNANT:   Forty-five percent of

20   those had blood glucose monitoring.

21                  ACTING CHAIRMAN SHERWIN:    So, it's about

22   25 patients.

23                  DR. WHISNANT:   Eight percent of the 25

24   reported very low blood glucoses.

25                  ACTING CHAIRMAN SHERWIN:    It's about
26   three patients.
                                                            187

 1                 DR. WHISNANT:    Eighteen percent of 131

 2   patients reported hypoglycemia.      4.6 percent of 131

 3   discontinued for reasons of hypoglycemia.

 4                 ACTING CHAIRMAN SHERWIN:    So, it looks

 5   like about four or five patients in the glyburide group

 6   versus three patients in the --

 7                 DR. MARCUS:    It's about three versus

 8   seven.

 9                 ACTING CHAIRMAN SHERWIN:    Are you sure?

10                 DR. MARCUS:    Well, 131 -- it's about one

11   out of five --

12                 ACTING CHAIRMAN SHERWIN:    That's about 25

13   or so.

14                 DR. MARCUS:    Okay, yes.

15                 ACTING CHAIRMAN SHERWIN:    Twenty-five, so

16   it's about 15 and a third of those is five.    That's what

17   I'm saying.   So, I think it's three versus five, with

18   the caveat that you have twice as many --

19                 DR. WHISNANT:    Three versus 15 then.

20   There are three times as many patients.

21                 DR. MOLITCH:    Somehow, it would seem that

22   this is one of the major reasons why this drug is being

23   brought forth to us to look at on an accelerated

24   fashion, that some sort of statistics would have been

25   done on this data to see if these are either clinically
26   meaningful or statistically meaningful?       You know, we
                                                           188

 1   need some help with this and why wasn't it done, if

 2   that's the reason it's being brought forward?

 3                DR. FLEMING:    Well, I actually felt that

 4   we did not have much of a signal for a difference.   That

 5   these results did not make the case that there was a

 6   major difference in outcome.    So, my position has been

 7   that theoretically, the drug offers this potential, but

 8   it has not been conclusively demonstrated.     We have

 9   some signals here, but I'm not sure that they would

10   convince anybody.

11                DR. WHISNANT:   And we would hasten to add

12   that there's been no purpose designed hypoglycemia

13   study.   We're talking about monitoring of adverse

14   events out of efficacy trials.

15                ACTING CHAIRMAN SHERWIN:    Right.

16                DR. WHISNANT:   So, unless you've done a

17   trial which we actually have designed now to look at

18   hypoglycemia in the elderly, where every patient has

19   been monitored preferably under observation like in a

20   retirement home, nursing home community, et cetera,

21   then there's no purpose designed study where analyses

22   of this kind of magnitude of change are appropriate.

23   So, we're reporting what we're reporting when you look

24   at the absence of serious events, the absence of severe

25   events, a lower frequency of discontinuation of events,
26   a lower number of patients who reported nighttime
                                                           189

 1   events.                   Out of those who were

 2   monitoring their blood glucose and out of those who were

 3   reporting,   the profile is consistent with the

 4   theoretical advantages of this drug.    We are committed

 5   to do -- actually, already planning -- a purpose design

 6   study to look at hypoglycemia including quality of

 7   life, et cetera.

 8                ACTING CHAIRMAN SHERWIN:    José?

 9                DR. CARA:   I get the gist that based on

10   your focus on the comparitor studies this morning, and

11   yet Dr. Fleming's comments that they're going to look

12   strictly at the placebo controlled studies, that

13   there's some miscommunication here in terms of the data

14   that we should be looking at.    Is that, in fact, the

15   case?

16                DR. WHISNANT:   No miscommunication at all

17   except, perhaps, what we've miscommunicated to you.

18   Let me clarify specifically.

19                The substantial evidence required for the

20   demonstration of efficacy is two adequate and well

21   controlled trials.   The best adequate and well

22   controlled trial design is a placebo controlled, double

23   blind trial and we've done three of those.       So, we

24   believe that the efficacy of this drug in type 2

25   diabetes has been unequivocally demonstrated.
26                On the other hand, the other part of the
                                                             190

 1   equation is the safety part of the equation and the

 2   numbers of patients and the durations of trials done

 3   in placebo control trials does not allow us an adequate

 4   full evaluation of the safety profile of the drug.

 5   Therefore, the company carried out five relatively

 6   large comparitor studies, four of them in Europe where

 7   you can't do placebo control trials.    It is that safety

 8   database that Dr. Edwards concentrated on primarily

 9   this morning.

10                DR. CARA:   Okay.   Let me ask you this.    I

11   was quite intrigued, surprised and concerned, about

12   what happens to the data when you take out the naive

13   patients.   In your comparitor studies, what was the

14   percent of naive patients there?

15                DR. WHISNANT:   Twelve to 15 percent per

16   trial.

17                DR. CARA:   And what happens to the data

18   when you take out naive patients?

19                DR. WHISNANT:   You get the same

20   equivalence when you used all patients.   It's a subset.

21   It's a minor subset.

22                ACTING CHAIRMAN SHERWIN:     Dr. Marcus?

23                DR. MARCUS:   We are on the subject of

24   efficacy and I expressed an opinion earlier in this

25   proceeding that I think efficacy includes a variety of
26   other endpoints.   We were going to be shown the results
                                                             191

 1   of those.    I think we can't leave this topic without

 2   having the company have an opportunity to show us those

 3   endpoints.

 4                 DR. WHISNANT:    Would you like to see the

 5   lipid data?

 6                 DR. MARCUS:    Absolutely.

 7                 DR. HIRSCH:    While they're doing that

 8   though, I did want to clarify my confusion which I think

 9   has been straightened out now.      I misinterpreted

10   efficacy to mean that this was a drug that was going

11   to be efficacious in the prevention of hypoglycemia,

12   which was sort of the thrust of my thinking about it

13   this morning.    Obviously, it was an unfair thing on my

14   part.

15                 I've been convinced, I think, about the

16   efficacy in terms of comparison as a hypoglycemic agent

17   with other available oral agents, but we seem to be a

18   little at sea.   Although theoretically, it might be a

19   good thing for spontaneous or other hypoglycemia.       We

20   don't have evidence at hand.

21                 Am I now straightened out?

22                 DR. FLEMING:    I may have contributed to

23   that confusion by saying that the attractive feature

24   was the potential for reducing hypoglycemia.

25   Certainly the company is not making a claim about
26   reducing hypoglycemia.
                                                          192

 1               DR. CARA:     So, your primary endpoint, in

 2   other words, is still efficacy in terms of blood glucose

 3   and glycohemoglobin?

 4               DR. FLEMING:    That's right.

 5               DR. EDWARDS:    We said before lunch that

 6   we'd try and respond to the questions about the lipid

 7   profile with some data.    I'm not sure whether I

 8   misunderstood the question before lunch or not.     I

 9   believe I gave an accurate answer and we've tried to

10   scramble together some actual results to show you.

11   We don't have complete data on every trial, but we've

12   got what we've got.

13               The data shown here is from the placebo

14   controlled United States' study 065 in which

15   repaglinide one milligram, repaglinide four

16   milligrams, and placebo were compared.      I've simply

17   tried to show you on this slide what the baseline values

18   for the cholesterol were in terms of mean and standard

19   deviations and those at the end of the trial, in this

20   case six months.   As I tried to explain this morning,

21   we do not believe they show any differences.

22               In terms of triglyceride values, same

23   presentation:   number of patients, mean and standard

24   deviation, the end of trial.    The only thing you may

25   be able to see -- and I haven't brought all the
26   statistical analysis with us -- is that if you look at
                                                               193

 1   the triglyceride values, you can see some tendency

 2   perhaps in the higher dose group, but the standard

 3   deviation is so broad that you're not likely to be able

 4   to distinguish it.

 5                  DR. MARCUS:    Well, did you put that on a

 6   repeated measures -- that could easily be statistically

 7   significant.

 8                  DR. EDWARDS:    We have --

 9                  DR. MARCUS:    Or even just a paired to

10   aspect.   It's about a reduction of ten percent or

11   thereabouts and that could certainly --

12                  DR. EDWARDS:    Okay.

13                  Wong Chin, can you help me with that?

14                  DR. EDWARDS:    Okay.   Our statistician

15   says we didn't analyze it.

16                  Okay, if we look at the long-term active

17   control trials, the answer I tried to give this morning

18   was that we did not see any difference when we compared

19   repaglinide with sulfonylurea agents.       What this data

20   shows you, expressed slightly differently, the number

21   of patients, the mean and the standard deviations, and

22   the change from baseline to the last visit which in this

23   case is a year study, and the confidence intervals

24   around those changes.

25                  DR. CARA:   Do the asterisks mean
26   significance?
                                                               194

 1                  DR. EDWARDS:     Yes, they do.

 2                  DR. KREISBERG:    What kind of units are you

 3   using there?    221 is a milligram per deciliter unit and

 4   then your change from baseline is 0.16 and 0.20.       That

 5   must be millimolar.

 6                  DR. EDWARDS:     I beg your pardon.

 7                  DR. MARCUS:    Yes, the top label says

 8   millimoles per liter.

 9                  DR. EDWARDS:     Okay.

10                  DR. MARCUS:    But it's a significant drop.

11                  DR. EDWARDS:     There was a significant

12   drop, yes.

13                  In those groups, same thing.

14                  AA:   So, the mean values are in millimoles

15   -- or the changes are millimoles per liter and the

16   baseline mean guides are going back to the deciliters?

17                  DR. MARCUS:    Now, these are fasting

18   triglycerides, presumably.       Overnight fasting?

19                  DR. EDWARDS:     Yes, these are fasting.

20                  DR. MARCUS:    Given what I think we've

21   learned in the last few years about the potential

22   importance of meal induced glypenia, if you were to

23   follow-up these data -- which I think you need to do,

24   follow up this whole area of concern -- it would be

25   important to do more than just fasting TGs and
26   cholesterol fractions, but to actually look -- I'm not
                                                               195

 1   sure when, but at some point, after meals in the

 2   immediate post-Prandial --

 3                  DR. EDWARDS:    Thank you.   We have planned

 4   to do exactly the study you describe because we hope

 5   that the prompt reduction of blood sugar after meals

 6   seen with Prandin may hope to control post-prandial

 7   hypolipidemia.     So, we have plans to do exactly what

 8   you describe.

 9                  DR. KREISBERG:     The change that you show

10   there for triglyceride is generously .1 millimolar

11   which is about nine milligrams per deciliter, from a

12   baseline of 220 with a standard deviation of 200.          I

13   would personally be shocked if that was statistically

14   significant.     But what I'd like to suggest to you is

15   --

16                  DR. MARCUS:     It's not.

17                  DR. KREISBERG:     Oh, it's not.   I thought

18   there was an asterisk.        Oh, that was the cholesterol

19   that had the asterisk.

20                  DR. EDWARDS:     No, it's just a spot on the

21   slide, I'm afraid.

22                  DR. KREISBERG:     You ought to also analyze

23   your data, vis-a-vis the response in glycemic control.

24   Because if you give a drug and there's no glycemic

25   control, I personally would doubt that there would be
26   any improvement in the triglyceride.        If there is
                                                          196

 1   glycemic control, then you might reasonably expect

 2   there to be a relationship.

 3               So, it may be that within this, you have

 4   a subgroup that would actually show some improvement

 5   in the dyslipidemia that is shown in these types of

 6   patients, but it would have to be correlated with the

 7   improvement in glycemic control.

 8               DR. EDWARDS:    I understand your question.

 9   I'm not sure I can answer it, spot on.   We thought you

10   might go in that general direction, and so what we've

11   got here from the integrated summary of the long-term

12   trials, integrated summary of efficacy, is the change

13   in baseline in HbA1c seen on the point

14   patient-by-patient basis.     Females shown in the

15   circle, males shown in the crosses showing the change

16   from baseline in the total cholesterol.    We really

17   don't feel that it shows anything.

18               Now, I was also asked did we have any other

19   information about cardiovascular type risk factors.

20   We don't have -- we have some very simple information

21   on fibrinogen we've been able to gather over lunch.

22   Again, just a very simple presentation of it.    This is

23   the data from the USA long-term active comparitor

24   trial, repaglinide/glyburide, with mean and standard

25   deviations for fibrinogen at the beginning and the end
26   of the trial.   We don't believe they show any
                                                              197

 1   differences.

 2                  ACTING CHAIRMAN SHERWIN:    Mark?

 3                  DR. MOLITCH:    Now, I presume since you

 4   didn't show it, that we don't have any data on HDL and

 5   LDL?   In these -- I know, for example, metformin will

 6   raise HDL, lower LDL, and the total cholesterol stays

 7   about the same.

 8                  DR. EDWARDS:    The same trail repaglinide

 9   versus -- again change from baseline very small, so we

10   don't believe there's any differences, but yes, we --

11                  DR. CARA:   Do you have the information

12   regarding the percent of patients that did not respond

13   to therapy, percent non-responders?

14                  DR. WHISNANT:    We know the percent of

15   patients who did not achieve a FPG target in the

16   comparitor trials and that's about 43.

17                  DR. CARA:   Total patients?

18                  DR. WHISNANT:   Yes, who do not achieve the

19   target in the titration period.

20                  DR. CARA:   And that's at the highest dose,

21   four milligrams?

22                  DR. WHISNANT:    They're titrated all the

23   way and they still don't achieve the target.

24                  DR. CARA:   And what percent is that of the

25   total patient base?
26                  DR. WHISNANT:    It's about 43 percent.
                                                             198

 1                 DR. CARA:   Oh, so it's not 43 patients.

 2   It's about 43 percent.

 3                 DR. WHISNANT:    Oh, yes.

 4                 DR. CARA:   So, 43 percent --

 5                 DR. WHISNANT:    You're targeting to 140.

 6                 DR. CARA:   So wait a minute.   Forty-three

 7   of patients treated with Prandin did not achieve a

 8   target blood sugar of --

 9                 DR. WHISNANT:    140.

10                 DR. CARA:   -- 140.

11                 DR. MOLITCH:    But that doesn't mean

12   non-responder.

13                 DR. CARA:   I'm sorry?

14                 DR. MOLITCH:    They can go from 280 to 180

15   and still not meet the target, but be a responder.

16                 DR. CARA:   Well, the way that you defined

17   responder was based on glycohemoglobin level.      Do you

18   have that?

19                 DR. MOLITCH:    I just said it was

20   something, maybe half-a-percent.

21                 DR. CARA:   Half-a-percent.

22                 DR. MOLITCH:    Or something like that.

23                 DR. CARA:   I mean, that's something that's

24   reasonable.

25                 DR. MOLITCH:    Actually, if we look at a
26   year's worth of data -- if you're looking at an
                                                            199

 1   improvement of a half percent.

 2                 DR. CARA:   If you exclude the naive

 3   patients --

 4                 DR. MOLITCH:    -- talking about naive

 5   patients.

 6                 If you look at naive patients, how many

 7   have the half percent -- response?

 8                 DR. WHISNANT:   A half percent of HbA1c, is

 9   that what you're asking?

10                 DR. MOLITCH:    Or greater.

11                 DR. WHISNANT:    Or less.

12                 DR. MOLITCH:    Whichever way you look at

13   it.

14                 DR. WHISNANT:    Right.   The percent

15   responders defined by greater than .5 percent HbA1c, for

16   instance, in the 24 week trial, placebo control trial.

17                 DR. MOLITCH:    Naive patients.

18                 DR. WHISNANT:    Right.   For patients who

19   started out greater than ten, then you get between 67

20   and 72 percent of those having more than a half a point

21   decrease.   For patients who started out between seven

22   and eight, then the half-a-point decrease is obviously

23   harder to achieve because you've only got between 38

24   and 52 percent of those.

25                 DR. CARA:   Thirty-eight to 52 percent?
26                 DR. WHISNANT:    Right.
                                                          200

 1               DR. CARA:   And how about in-between eight

 2   to ten, thereabouts?

 3               DR. WHISNANT:   In-between.

 4               DR. CARA:   Fifty-two to 67, somewhere

 5   around there?

 6               DR. WHISNANT:   Between the numbers,

 7   right.

 8               So, using .5 as a reasonable measure of

 9   some value achieved, then you're talking about

10   two-thirds of patients at high baselines and a third

11   of patients at --

12               DR. CARA:   And this was, again, titrating

13   up to a maximum dose of four --

14               DR. WHISNANT:   No, these were fixed dose

15   trials.

16               DR. CARA:   Oh, these were the fixed dose

17   trials.

18               DR. WHISNANT:   This was a comparison of

19   one, versus four, versus placebo.

20               DR. CARA:   Okay.

21               DR. WHISNANT:   That's the best data we

22   have.

23               DR. CARA:   Okay, that's the best data

24   you've got so you can't really separate out based on

25   the dose?
26               DR. WHISNANT:   Well, in the titration
                                                              201

 1   trials, I mean, you have all those confounding

 2   variables.

 3                 DR. CARA:   Right.

 4                 DR. WHISNANT:    So, it's harder to answer

 5   your question with any reasonable certainty using the

 6   titration format.    Now, if you require a greater than

 7   one point decrement, then you get 50 to 64 percent of

 8   patients at high baselines and only 28 percent patients

 9   at low baselines.    So, it's in that range.

10                 ACTING CHAIRMAN SHERWIN:     Dr. Marcus?

11                 DR. MARCUS:    Was there any change in

12   average systolic or diastolic blood pressure or resting

13   pulse rate?

14                 DR. WHISNANT:    On average, all the blood

15   pressures that we have showed no changes, right?

16                 DR. STRANGE:    That's correct.

17                 ACTING CHAIRMAN SHERWIN:     Okay.   I think

18   we can go on to the next question.

19                 Dr. Fleming, do you have any --

20                 DR. FLEMING:    No, I think that will do it.

21                 ACTING CHAIRMAN SHERWIN:     José?

22                 DR. CARA:   I'm sorry, just one last

23   question.    Was there any specific characteristic that

24   you could kind of pinpoint or did you look at anything

25   that would tell you which patients were likely to be
26   responders versus non-responders based on the
                                                            202

 1   glycohemoglobin criteria?     For example, degree of

 2   obesity, length of time of diabetes, age, et cetera,

 3   et cetera.

 4                 DR. WHISNANT:   Right.    I mean, there was

 5   an effort to determine correlates of response, the

 6   response rates by age -- greater than 65, less than 65

 7   are the same.    There's a considerably higher response

 8   rate in naive patients than in previously treated

 9   patients.    That's a song I've been singing all day.

10   The final HbA1c is governed by the initial HbA1c, but the

11   delta HbA1c is only to the extent that I've described

12   for you, governed by the previous HbA1c.     The response

13   is not governed by C peptides.

14                 What else did we look at?     Duration of

15   treatment.

16                 DR. HIRSCH:   I just want to make sure I

17   understand the efficacy finally.       You're saying, in

18   fact, if we treated a lot of people now at the four

19   milligram level, it would do all of the same things in

20   terms of sugar lowering as glyburide or other oral

21   agents do.    But still 46 percent or thereabouts of the

22   patients would still be sub-optimal in treatment, is

23   that correct?   Or not achieve the ideal goals?   I mean,

24   can we expect that half the people would not be as well

25   treated as we might like after one year of this?       Is
26   that right?
                                                             203

 1                DR. WHISNANT:     That is the correct

 2   conclusion within the limits of the design and conduct

 3   of this trial.    Obviously, if you want to optimize

 4   therapy of type 2 diabetes, you have to do a lot of

 5   things, perhaps including adding other drugs.

 6                DR. HIRSCH:     Or increasing the doses.

 7                DR. WHISNANT:     Or increasing the doses.

 8                DR. HIRSCH:     Which brings us to the next

 9   point, I guess.

10                DR. WHISNANT:     Right.

11                DR. MOLITCH:    But Jules, that's also their

12   target of 140 and optimally, we might want to be even

13   considerably below that, so that the numbers would be

14   even less.

15                DR. HIRSCH:     But these people were kept on

16   it for a year or thereabouts.      So we know --

17                DR. WHISNANT:     Same dose without being

18   allowed to change doses.

19                DR. HIRSCH:     Right.

20                DR. CARA:   Now, on comparitor studies, you

21   were allowed to titrate, is that not correct?

22                DR. WHISNANT:     The titration period was up

23   to month zero which was the period of maintenance.

24   After they entered the 12 month maintenance period,

25   there was no dose adjustment.
26                DR. FLEMING:     Dr. Cara's question
                                                          204

 1   certainly leads us to this next discussion point.   It's

 2   realizing, of course, that the response of individual

 3   patients is highly variable depending on their current

 4   level of glycemic control, their exposure to previous

 5   treatment.   There may be a slight even gender

 6   difference between women and men.

 7                  At this point, we would like to explore

 8   what the state of your data would allow us to conclude

 9   about more refined dose regimen recommendations, and

10   what studies might be needed to pursue other

11   refinements.

12                  DR. WHISNANT:   Thank you very much, Dr.

13   Fleming.

14                  Let me go through this introduction very

15   fast because I think so much of this has been discussed

16   already in one way or the other, and then show you a

17   tiny bit of some analysis, some of which you've seen

18   already.   So, this will be a fast response to the

19   question, Dr. Fleming.

20                  Obviously, the ideal in this setting is to

21   give a drug which builds upon the natural intact

22   feedback mechanism whereby insulin controls the

23   glucose load and remains sensitive to the glucose load,

24   and therefore would presumably not encounter the

25   exogenous, idiopathic if you will, physician induced
26   hypoglycemia because of these drugs.
                                                         205

 1               Our hypothesis is that therefore, this

 2   insulin profile in red -- which looks very much like

 3   the dotted line insulin profile which is the natural

 4   meal related insulin profile induced by in this case,

 5   four milligrams of our drug -- should be a better

 6   approach to not only short-term FBG response, long-term

 7   glycemic control but hopefully, more physiologic in its

 8   patrol mechanism.

 9               What we have therefore provided, what we

10   believe we are providing with this drug is in effect,

11   an individualized or patient controlled flexible

12   dosing almost based upon the kind of paradigm that's

13   been developed for analgesics and potentially other

14   drugs where patients learn to take care of themselves

15   by taking a dose before each meal.    They actually can

16   take it anywhere up to 30 minutes before beginning each

17   meal without changing the PK profile or they will show

18   you the blood sugar data, Dr. Cara.    That dose should

19   be managed according to the patient's eating schedule

20   and that smaller doses give lower insulin responses and

21   we know that from dose response studies.

22               So, theoretically as we go forward with

23   this drug, we ought to be able to customize the amount

24   of insulin response needed based upon how much the

25   patient is taking in at that particular time.   I hasten
26   to tell you that study has not been done, but we look
                                                           206

 1   forward to doing it.   And as Dr. Damsbo showed you that

 2   doses can be taken two, three, or four times a day

 3   without compromising either the glycemic control or the

 4   threat of hypoglycemia.

 5               Then the final point on this slide is one

 6   that I'd like to comment on, again, with hopefully a

 7   little more clarity than I have this morning.      The

 8   question of whether or not dose titration is required

 9   relates, we believe, to the question of does the

10   hypoglycemia of this drug profile suggest less severe,

11   less frequent, or certainly less frequent, less severe

12   hypoglycemias than with other therapies?     If that is

13   the case, then dose titration perhaps can be omitted,

14   dosing can be simplified at least for certain patients

15   where their risk of hypoglycemia is not as high.    So,

16   I return to the slides and a simplified version of those

17   slides to formulate my request to you.

18               What I showed you this morning was that the

19   percentage of patients having hypoglycemia is

20   determined in part by their baseline HbA1c numbers

21   relative to the dose group.    These are previously

22   treated patients where, at a four milligram dose with

23   patients who have low HbA1c's -- let's look at, say, the

24   two lower dose groups between six and eight, all the

25   way up to eight.   What you can see is that the
26   percentage of those patients reporting any
                                                        207

 1   hypoglycemic episode comes up -- approaches half the

 2   patients, as opposed to patients with higher HbA1c's or

 3   patients who are -- for those previously treated

 4   patients.   In contrast, the naive patients that we all

 5   know now are the more responsive patients have an even

 6   higher percentage of hypoglycemia, particularly those

 7   subsets who are eight and below, and those patients who

 8   are given, if you will, full dose.

 9                So that, our conclusions of this subset

10   analysis of the data are that hypoglycemia risk is

11   related both to the variable of previous treatment and

12   to the variable of baseline HbA1c.   So, if I put those

13   variables on a slide for the four milligram dose group,

14   what you can see is that the naive patients, these two

15   bars, as well as the patients with a low HbA1c, have

16   relatively higher hypoglycemia rates.    In this case,

17   probably an unacceptable rate of hypoglycemia compared

18   to the previously treated patients and the previously

19   treated patients who have higher contrast of that one

20   milligram group, which I guess I don't have in there.

21   It also shows this relationship, but with the bars

22   proportionately lower.

23                What we're suggesting, therefore, is that

24   it is this subset of patients, in fact, who need more

25   therapy, if you will.    Their response rates are going
26   to be less and probably need more intensification of
                                                          208

 1   therapy.   It's this subset of patients that perhaps we

 2   don't need to wait a month or six weeks to titrate

 3   because they're going to have a very acceptable rate

 4   of hypoglycemia even with full four milligram dose.

 5                Now, it turns out that we tested the

 6   hypothesis in a European trial.     This is a

 7   retrospective review of a piece of data.      Because in

 8   a European trial, one of the comparitor trials, the

 9   doctors who were carrying out this trial were allowed

10   "at the discretion of the investigators, patients with

11   greater than 160s while on previous SU were allowed to

12   start on one milligram."     The patients in one trial

13   were allowed to start on titration at two milligrams

14   three times-a-day with meals if they had blood plasma

15   glucoses of greater than 180.    So, we actually tested

16   the percentage of hypoglycemia -- rather than test the

17   titration design, these doctors actually tested the

18   other answer without knowing it.     So, they gave us a

19   hypoglycemia rate for those patients that were started

20   on .5, 1 and 2.   It's 54 patients and it's four percent.

21   So, it's not a big number, but what it says is that out

22   of patients that were not in a purpose designed

23   randomized trial but in patients who were, in fact,

24   started on that dose, that the hypoglycemia rate is

25   relatively low.
26                So, our request for your consideration is
                                                             209

 1   that since we know these things about the rapid insulin

 2   response with this drug, since we know that

 3   hypoglycemia events are not severe or serious, since

 4   we know the efficacy response is not only prompt but

 5   sustained, but is also dose related, we would ask for

 6   your consideration that in addition to a standard

 7   titration format of labeling that we consider whether

 8   or not it's rational to give those patients who meet

 9   certain clinical laboratory criteria the benefit of

10   full dose of drug at the outset.

11                DR. HIRSCH:     What is full dose at the

12   outset?

13                DR. WHISNANT:    Well, Dr. Kreisberg thinks

14   it's two milligrams with each meal.     In our clinical

15   trials, the full dose was four milligrams with each

16   meal.   The delta between two milligrams and four

17   milligrams is measurable on a dose response curve, but

18   you know -- and is safe, clearly, four milligrams even

19   four times a day and 20 milligrams four times a day did

20   not have associated toxicities, either laboratory, EKG

21   or symptomatic.   And so, the four milligrams is well

22   within the safety margin for dosing and it would be,

23   you know, a doctor's decision about whether or not to

24   give the full four milligrams.

25                I guess what we would prefer is that the
26   starting dose be two milligrams with each meal with the
                                                            210

 1   option of doubling that dose to determine if a patient

 2   has additional response.

 3                  DR. HIRSCH:   I was wondering what your

 4   best guess is, how much more than four milligrams?

 5   Because at four milligrams, you said half or not fully

 6   controlled, so it obviously has to be more than that.

 7                  DR. WHISNANT:    I just don't know because

 8   we've not done efficacy trials actually testing the

 9   difference between, say, four and eight four

10   times-a-day.    What we know is above four milligrams

11   with each of four meals that the drug is safe, so we're

12   not concerned about the margins.       I believe our

13   recommendation would be that for those patients that

14   I've showed you on this subset analysis slide -- that

15   is, the previously treated patients with HbA1c's below

16   eight -- that two milligrams with each meal is an

17   appropriate starting point and that the safety margin

18   allows testing of at least twice that much for an

19   individual patients based on response.

20                  DR. MARCUS:   I'd like to ask a question

21   that is directed to Dr. Fleming or Dr. Sobel.        Many

22   lacunae in our knowledge about this drug have come up

23   and without an answer except to say that this is a

24   subject for future study.      We're in an unusual position

25   -- at least I feel in an unusual position, a little bit
26   awkward about trying to make a recommendation to you
                                                            211

 1   as to whether this drug is ready to appear before the

 2   American public or not with all these lacunae in the

 3   background.

 4                 Now, I have not seen a guidance from the

 5   Agency about diabetes drugs.    If one has been

 6   disseminated since my appearance, since my joining this

 7   panel, I'm not aware of it.    Is it necessary and

 8   sufficient that a drug be shown to be indistinguishable

 9   in terms of the specific efficacy point that you've

10   focused on, that is hemoglobin A1c and fasting plasma

11   glucose, and that the safety be as the sorts of things

12   we've heard about today?     Or are we supposed to be

13   considering that in all the rest of the material that

14   many of us have expressed concerns about?

15                 ACTING CHAIRMAN SHERWIN:   This is the

16   question I think all of us have.

17                 DR. FLEMING:   Well, by the way, we are

18   working on a guidance for development of oral diabetic

19   agents.   I think the basic principle is in terms of

20   demonstrating efficacy, that a clinically significant

21   change in glycemic control needs to be demonstrated.

22   That is necessary and sufficient for demonstrating

23   efficacy.

24                 Then it comes to evaluating safety.

25   Everything is germane to that, the theoretical and
26   known risk.   Though we may compare a therapy with
                                                             212

 1   currently available therapies, that does not

 2   necessarily mean that -- or we're not actually formally

 3   involved in that kind of comparison, but it's hard to

 4   avoid.   You certainly are going to be comparing a new

 5   therapy with what is out there.

 6                  So, ultimately, it is the risk benefit that

 7   determines the approvability of the drug.         You would

 8   start by showing a minimal level of glycemic

 9   improvement.     Of course, glycated hemoglobin is the

10   endpoint of choice for that.        Certainly, the company

11   has gone far beyond what would be considered a minimally

12   acceptable change.     Then it's a matter of making an

13   estimate of the risk involved.        At this point in the

14   drug development, we certainly don't expect to answer

15   all the questions.     There will be significant

16   questions that have to be answered as a larger

17   population is exposed.       But certainly, you need to have

18   most of your safety issues at a reasonably defined

19   state.

20                  DR. MARCUS:    Thank you.

21                  ACTING CHAIRMAN SHERWIN:     You showed us

22   pictures of insulin going up and down and the

23   implication is that the drug is causing those kinds of

24   fluctuations to allow physiologic responses.         Do you

25   have any data with other sulfonylureas?         I mean, you
26   showed us the drug levels which I agree would be a higher
                                                          213

 1   with, let's say, glyburide as an example.   Clearly, the

 2   drug levels are much higher and sustained for a longer

 3   period of time.

 4                What's the impact on the profiles of those

 5   sustained higher levels?     Because my impression is

 6   that you see the same sort of fluctuations in insulin

 7   with glyburide and glipizide and the other

 8   sulfonylureas, even though they have a longer duration

 9   of action.   Is that sufficient to say that the drug is

10   just working at that meal time?

11                DR. WHISNANT:    Far be it from us to

12   contradict your impression, sir.

13                ACTING CHAIRMAN SHERWIN:   No, no, no, no,

14   I -- something out of it.

15                DR. WHISNANT:    Our understanding is that

16   the normal meal related fluctuations in insulin, which

17   we have demonstrated over and over again because we have

18   placebo comparisons for each of those curves.     So, the

19   normal meal related fluctuation is still intact when

20   you give this drug, or when you give any drug.

21                ACTING CHAIRMAN SHERWIN:    Right.

22                DR. WHISNANT:   The question really is, is

23   there a difference at nadir and is there a difference

24   at night?

25                ACTING CHAIRMAN SHERWIN:    Right.
26                DR. WHISNANT:    Because that's the time
                                                            214

 1   when the long acting drugs cause trouble.

 2                ACTING CHAIRMAN SHERWIN:     Right.   And my

 3   question is that I didn't see that data, or did I see

 4   it between --

 5                DR. WHISNANT:    We haven't studied other

 6   people's drugs.

 7                ACTING CHAIRMAN SHERWIN:     You've not

 8   compared the other drugs?

 9                DR. WHISNANT:    Well, Dr. Damsbo showed you

10   a small study looking at the skip-a-meal hypothesis.

11                ACTING CHAIRMAN SHERWIN:    Right, but that

12   has some problems with it.    But in terms of other kinds

13   of studies, looking at 24 hour profiles, we don't have

14   that data.   Is that right?

15                DR. WHISNANT:    We do not have that data in

16   our database.

17                ACTING CHAIRMAN SHERWIN:     Right.

18                DR. WHISNANT:    We did not --

19                ACTING CHAIRMAN SHERWIN:     Not that I'm

20   saying -- you may not have been asked to provide that

21   data.

22                DR. WHISNANT:    That's okay.

23                ACTING CHAIRMAN SHERWIN:    You know, so I'm

24   not trying to say that, you know, this is your fault

25   for not showing that data.    I'm just curious about it
26   because the implication is that the insulin levels will
                                                           215

 1   be higher with some of the other sulfonylureas at night.

 2   I think theoretically, that's so.    The issue is, you

 3   know, it would be nice to know that in a more defined

 4   way.

 5               DR. WHISNANT:    The theoretical concerns

 6   based upon the kinetic profiles of the drugs I suspect

 7   people can dig out that information from various

 8   studies that have been done previously.     But let me

 9   clarify that we did not come to the Agency to make a

10   superiority claim.   We came to the Agency for efficacy

11   and safety of this drug.   If we were going to go to the

12   next level, if you will -- sorry, Dr. Marcus, but this

13   would be a future question that we would have to address

14   in order to make a superiority claim.

15               ACTING CHAIRMAN SHERWIN:     Sure.    Right.

16   And the Agency is not looking for superiority decision

17   in terms of efficacy.

18               DR. FLEMING:    Oh, absolutely not.    That's

19   a very important point.     The company is only obliged

20   to show safety and efficacy.    If they want to make a

21   superiority claim, they would have to have the data that

22   demonstrate that the drug is actually superior to a

23   given therapy.   Now, it wouldn't have to be necessarily

24   confined to efficacy.   They could formally demonstrate

25   that hypoglycemia, for example, was reduced with
26   comparable efficacy achieved.    That would be another
                                                              216

 1   means of getting a superiority claim in effect.        But

 2   the company is not making that case here.

 3                 ACTING CHAIRMAN SHERWIN:     Right.

 4                 DR. WHISNANT:    I would add though that we

 5   believe we have the basis, theoretically,

 6   scientifically, with clinical signal to do that.       So,

 7   I mean, while that purpose design study is not a part

 8   of this application, we believe that it is now rational

 9   to do that.

10                 ACTING CHAIRMAN SHERWIN:    Right.    I guess

11   it relates to claims when drugs are released and

12   advertising is --

13                 Dr. Nisben, welcome.

14                 DR. NISBEN:     I'd just like briefly to

15   respond to what Dr. Marcus had asked.      I am actually

16   working on a guidance for development of diabetes

17   drugs.   Also, I think with respect to this particular

18   product, we are supposed to have two adequate and well

19   controlled trials demonstrating efficacy.     But I think

20   it should be pointed out that that's adequate and well

21   controlled trials demonstrating efficacy in the

22   population which is intended to be treated.         I think

23   that's something which has not been adequately

24   discussed in my opinion.

25                 Most of the studies that have been
26   presented have been in patients who have been poorly
                                                           217

 1   responsive to sulfonylureas, the comparative trials --

 2   and I think as Dr. Cara has said very, very well, they

 3   were poorly controlled when they began.    They were even

 4   worse controlled at the end.   So, really, one can't say

 5   anything about efficacy in those trials.

 6                When you look at the other trials to

 7   placebo control trials, although it appears that there

 8   are three adequate, well controlled trials, I really

 9   don't think that's the case.    The number of naive

10   patients is extremely small.    The total exposure is

11   only about 100 patients.   We've already, I think, come

12   to the conclusion that this is not a drug that you're

13   going to take patients off of glyburide and put them

14   on repaglinide because we know that they don't do any

15   better.

16                This is really the intent, it seems to me,

17   is to use this drug in naive patients.    But where's the

18   data?   Where's the database to show that it's effective

19   in naive patients?   Well, it probably is, but the

20   number of patients exposed is very small.      Also, I

21   think the intent is to decrease hypoglycemia and I think

22   as Dr. New pointed out, it would be very, very, very,

23   very nice to be able to take a drug before each meal.

24   And if you skip a meal, you wouldn't take it and you

25   would theoretically prevent hypoglycemia.      But from
26   the data we showed, it looked to me like if anything,
                                                             218

 1   the risk of hypoglycemia might actually be worse in the

 2   naive patients.    Those were the ones that had the most

 3   robust responses.

 4                It seems to me we do not have long-term data

 5   on those patients.     The total database is only about

 6   100.   It seems to me that it is a perhaps a bit premature

 7   to be releasing this on the American public.      I don't

 8   know of any other oral hypoglycemic agent where the

 9   total database for the intended population is really

10   so very, very small.    This, I think, would be a first.

11                DR. FLEMING:   I think it's good that we can

12   have different opinions within the Agency.    Obviously,

13   Dr. Nisben has stated his opinion.

14                I actually take a different slant from what

15   you just heard.    I think the intended population is

16   what was tested.    Basically, this treatment will be

17   offered to more patients that have been on other agents

18   than not.   This is just a reality.     So, there was

19   nothing wrong in the population that was selected.

20   That's simply a reflection of the current situation

21   today.

22                Now, you could say that we don't have

23   enough naive patients period, in an absolute sense, to

24   say what the degree of efficacy will be.    I don't think

25   anyone would believe that naive patients would respond
26   any less than patients who have been treated with other
                                                               219

 1   agents before.    So, I'm not sure that we have a concern

 2   that the drug would be less efficacious in naive

 3   patients.   I'm not sure what the value of a huge study

 4   that simply was confined to naive patients would be.

 5   We would probably find that these patients responded

 6   somewhat better on average than this mixed population

 7   that has been studied.

 8                  But in terms of my opinion about the

 9   company's placebo controlled studies, I do think that

10   they would represent studies that are adequate to

11   support efficacy.

12                  DR. CARA:    You were going to show us some

13   blood glucose data to support the statement regarding

14   timing of the dosing.

15                  DR. NEW:    They showed it when you were out

16   of the room.

17                  DR. CARA:    Oh.   Could somebody fill me in

18   on what it showed?

19                  DR. NEW:    No difference.

20                  DR. CARA:    No difference.

21                  ACTING CHAIRMAN SHERWIN:      Cathy and then

22   Maria.

23                  DR. CRITCHLOW:     If I could just get a

24   clarification on the placebo controlled trials?

25                  Since the percentage of naive patients was
26   very small, were these then patients who were being
                                                            220

 1   treated with something and then taken off of that

 2   treatment?   And they were the controls?

 3                DR. WHISNANT:    Naive in our definition of

 4   inclusion criteria for the clinical trials means that

 5   they have by ADA criteria type 2 diabetes mellitus, but

 6   have not been previously treated with an oral

 7   hypoglycemic agent.

 8                DR. CRITCHLOW:    Right.   And that

 9   percentage was small in the --

10                DR. WHISNANT:    Well, it's in the --

11                DR. CRITCHLOW:    I mean, was it 23 percent?

12                DR. WHISNANT:    -- comparitor trials, it's

13   on the order of 12 to 15 percent of --

14                DR. CRITCHLOW:    But in the placebo

15   controlled --

16                DR. WHISNANT:    -- 2,000 patients, and in

17   the placebo controlled trials it's --

18                DR. CRITCHLOW:    Well, it's nine or 23.

19                DR. WHISNANT:    What's the percentage in 65

20   and --

21                257 naive patients treated with

22   repaglinide in various trials.

23                ACTING CHAIRMAN SHERWIN:     Okay.    Maria

24   and then we'll go on to the fourth question.

25                DR. WHISNANT:    Can I just offer one more
26   comment about that question, please, Dr. Sherwin?
                                                         221

 1               I'd just like to offer in response that the

 2   number of patients required, to some extent, depends

 3   on the delta because after all, it's statistical

 4   difference that we're asked to demonstrate a clinically

 5   meaningful and statistically significant difference.

 6   So, because response in naive patients is

 7   quantitatively so much different, then I could

 8   logically conclude that perhaps not as many patients

 9   required in order to satisfactorily demonstrate that

10   therapeutic effect.

11               DR. CRITCHLOW:    No, I agree.   But in the

12   non-naive patients and if they were in the placebo arm,

13   were they currently on therapy when they were recruited

14   for the trial and then taken off of therapy for the

15   duration of the trial?    So, we basically saw no

16   difference in control with the treated patients and,

17   like you said before, an obvious worsening of control

18   in the placebo patients among basically a large group

19   -- the subset of patients that were previously being

20   treated who were in the control arm who were essentially

21   not being treated for the purposes of the trial.    So,

22   the difference that we're seeing is essentially taking

23   people who were being controlled to some extent, and

24   then being taken off --

25               DR. WHISNANT:    Take them off therapy, and
26   their disease gets worse.    You compare them to
                                                             222

 1   continuing to maintain that on another therapy.       That

 2   is correct.

 3                 DR. FLEMING:    And I think that's an

 4   important point.     These are not burned out patients.

 5   They, because of the protocol, had to demonstrate that

 6   they were receiving some benefit from the previous

 7   therapy.

 8                 ACTING CHAIRMAN SHERWIN:    Maria, this is

 9   the last question.

10                 DR. NEW:   Cathy, are you saying that you

11   would like to see a comparison of placebo versus treated

12   in cohorts that have never been treated?

13                 DR. CRITCHLOW:    I think we saw --

14                 DR. NEW:   You never saw that because I

15   think most of the placebo patients are withdraw

16   patients.   They are not never treated patients.        In

17   contrast to those that are treated with repaglinide,

18   those patients who are considered naive are never

19   treated patients.

20                 Am I correct in my conclusion that your

21   placebo group does not mean never treated?     Only your

22   treated group needs never treated in the naive group.

23                 ACTING CHAIRMAN SHERWIN:   No.   You've got

24   me confused now.

25                 DR. WHISNANT:    We're using words in a
26   slightly different way, Dr. New.
                                                            223

 1                 DR. NEW:   Okay.    Let me just ask my

 2   question and then I'll be clear, and I think that that

 3   will clear Cathy.    Because then I want to ask my own

 4   question.

 5                 When you call a study naive, I understand

 6   that those that you treat have never been treated

 7   before.

 8                 DR. WHISNANT:   That is correct.

 9                 DR. NEW:   Now, what about the placebo

10   group?

11                 DR. WHISNANT:   They've never been treated

12   before either because it's a randomized, controlled,

13   double-blind trial.

14                 DR. NEW:   Okay, then I'm wrong.

15                 DR. WHISNANT:   You take the population,

16   whatever the population is --

17                 DR. NEW:   I got it.

18                 DR. WHISNANT:      -- it could be previously

19   treated or naive, and you randomize them blind.

20                 DR. NEW:   Okay.    Here's my question, my

21   question.   It seems to me that your persuasive powers

22   to say that this drug is a good drug to give to the

23   American public resides in three slides that you've

24   shown.    The first two slides of those that are your

25   annual trials, patients treated for one year, those
26   that are naive and those that are already treated.     You
                                                            224

 1   showed two slides.   The third is a table in which you

 2   demonstrated the hypoglycemic complication of the drug

 3   in that slide that has a table with 343 patients treated

 4   with Prandin and 131 treated with glyburide.

 5                ACTING CHAIRMAN SHERWIN:     That's the

 6   elderly, yes.

 7                DR. NEW:    Of the Prandin one, 65 percent

 8   reported symptoms and of that 65 percent that reported

 9   symptoms, eight percent actually measured their blood

10   sugars.   Am I right?

11                ACTING CHAIRMAN SHERWIN:     No, no.

12                DR. WHISNANT:   Eighty percent had very low

13   --

14                ACTING CHAIRMAN SHERWIN:     Documented

15   glucohypoglycemia.

16                DR. NEW:    That's right,   had documented

17   -- measured -- that's what I said, actually measured

18   the blood sugars that were less than 45.

19                ACTING CHAIRMAN SHERWIN:     Right.

20                DR. NEW:    Okay.   So, those numbers when

21   you calculate them --

22                ACTING CHAIRMAN SHERWIN:    Three patients.

23                DR. NEW:    No, I don't get that.     Out of

24   343, 45 percent reported hypoglycemic symptoms.      That

25   comes to 154 patients.    Of that, if you take eight
26   percent of those that reported symptoms and then
                                                              225

 1   measured their blood sugars or documented their

 2   hypoglycemia biochemically, that's 12 patients.

 3                 ACTING CHAIRMAN SHERWIN:     Somehow, we'd

 4   have to go back to the table.     I'm not sure now.

 5                 DR. NEW:   Okay, anyway, but can you pull

 6   those three slides and then I'll feel like I can make

 7   a decision?

 8                 ACTING CHAIRMAN SHERWIN:     Okay.

 9                 While you're pulling those slides --

10                 DR. NEW:   That's the two annual slides and

11   the table on hypoglycemia.

12                 ACTING CHAIRMAN SHERWIN:     Dr. Fleming,

13   can we move on to the fourth because we're going to be

14   here until tomorrow.

15                 DR. FLEMING:    Okay.     Issue   number

16   four --

17                 ACTING CHAIRMAN SHERWIN:    We'll come back

18   to Maria's question at the end.       I promise.

19                 DR. FLEMING:   Well, as the company has

20   already pointed out, there was the observation of an

21   imbalance in the number of myocardial ischemic events

22   that were observed in not just one trial, but probably

23   two.

24                 Let's have the next slide.     Perhaps it's

25   the one before that or the one after.       Let's go back
26   -- yes, that's fine.
                                                         226

 1                Now, this is the result from the US study

 2   with glyburide as the comparitor.     You can see that

 3   particularly when you get down to acute ischemic events

 4   as a subcategory of cardiovascular, that there is a

 5   fairly high relative risk.   This is the risk compared

 6   to the comparitor group, glyburide.   On the other hand,

 7   the statistical significance is trending, but it is

 8   certainly not reaching the point that we would conclude

 9   that it has reached statistical significance.

10                Now, in the next slide this simply sums up

11   the unadjusted and adjusted relative risk with respect

12   to these various comparitors, including placebo.    You

13   can see that with respect to glyburide in particular,

14   there is an adjusted relative risk of about two times.

15   Even in the case of placebo, there is an adjusted

16   relative risk that approaches two.    So, this is our

17   signal.   This is unexpected in the sense that we did

18   not expect to see a difference between these treatment

19   groups.   They work in basically the same general way.

20   Obviously, everyone knows about the story of UGDP and

21   the cloud that that study has cast on sulfonylurea

22   therapy as well by guanides by the way.   Phenformin was

23   involved in that trial, I'll remind you.

24                At any rate, these are what we have.    We

25   do not have, certainly, statistical significance.
26   When you meta-analyze the entire trials, basically,
                                                         227

 1   these effects wash out.

 2                Now, the question is what do we do at this

 3   point.   I think it would be now time for the company

 4   to respond with how they would view these data, and more

 5   importantly, what you would do to resolve the issue.

 6                DR. WHISNANT:   Thank you very much, Dr.

 7   Fleming.   We're happy to respond to this.

 8                It's a serious signal that we've listened

 9   to, watched for very carefully.    Let me remind you

10   first that Dr. Edwards spoke to the cardiovascular risk

11   profile for this drug when he concluded that our risk

12   profile for cardiovascular events is similar to

13   comparable to sulfonylureas.   But that we have a small

14   increase of non-fatal events, a count phenomenon.   The

15   number is increased, non-fatal events in comparison to

16   glyburide.

17                So, having seen that signal, we went

18   through the detailed analysis that Dr. Edwards showed

19   you a couple of slides on.     I'd just like to remind

20   you that this is the cumulative incidence curve that

21   he showed you for all trials, for all ischemic events,

22   normalized to the 049 study in the United States.   The

23   reason why that's important is that the 049 study

24   actually has in it another unequal randomization -- not

25   an imbalance in the randomization that was simply a fact
26   of the kinds of patients who are recruited into this
                                                               228

 1   trial.     I remind you that this was a randomized,

 2   controlled, double-blind trial so there was no control

 3   as to stratification over these kinds of baseline

 4   events.

 5                  There were 12 patients in the repaglinide

 6   group who had had prior MIs, six patients who had had

 7   congestive heart failure, five patients had at baseline

 8   ischemic changes on their EKGs, and 17 patients with

 9   a medical history of coronary artery disease.        As

10   opposed to in half as many patients now in the glyburide

11   group, two with prior MIs, one with EKG ischemia and

12   four with coronary artery disease.

13                  I do not show you this slide in order to

14   deny the signal of cardiovascular disease that we have

15   seen in our trials.    Whatever the imbalance was in the

16   trial, we have taken the signal and taken it very

17   seriously.

18                  DR. MARCUS:     Is that, by the way, a post

19   hoc analysis?

20                  DR. WHISNANT:    That's a post hoc analysis.

21   All the histories were reviewed independently, I might

22   add.     All the EKGs were reread independently and this

23   analysis was constructed because we saw the signal.

24   All the data had been collected prospectively, but then

25   we went back and reviewed the case report forms in order
26   to look at these kinds of numbers.
                                                          229

 1                Now, I also remind you that developing

 2   drugs in this arena means that we're working in this

 3   kind of environment.   We recognize the environment in

 4   which we work where if we treat enough patients -- if

 5   we treat 100 patients for a year, 5.5 percent of them

 6   on average are going to die.    This is cumulative

 7   mortality figures from the ADA, from the national

 8   follow-up study that's reported in the ADA manual.    I

 9   also remind you that depending on how many patients we

10   have in the old rates group, then our mortality will

11   be higher.

12                The range of cardiovascular events that

13   we're dealing with in these kinds of trials is

14   represented on this sort of survey, list, compilation.

15   It runs from 1.6 percent total mortality in the 50 age

16   decade up to as high as six percent in another

17   prospective microalbuminemia study.

18                Now the company has taken this challenge,

19   this signal, this worry, if you will, very seriously.

20   We've met with the Agency to discuss the analysis of

21   the data and we come here today to show you a proposal

22   that we believe adequately addresses the ongoing need

23   for assuring the safety profile of this drug.    I know

24   of no better person to address this issue than Dr. Gerry

25   Faich.
26                Sir?
                                                           230

 1                DR. FAICH:    Mr. Chairman, ladies and

 2   gentlemen, what I would like to do is describe for you

 3   how this expert committee that you see listed here which

 4   I chaired approached this issue of cardiovascular risk

 5   in type 2 diabetes, and approached the issue of

 6   designing a study, at least in outline form.   I'm going

 7   to be very brief because we deliberated through a number

 8   of things and I'd like to at least share that process

 9   with you as opposed to sharing with you a completely

10   finalized study.

11                I need to point out at the outset that our

12   group, having seen these same data, was reasonably

13   ambivalent about whether this was a meaningful signal

14   to begin with.   So, what I'm going to do is assume that

15   there is a possible problem and that what one is

16   involved with here is, in a sense, proving a negative.

17                We also knew at the outset that when one

18   talks about oral hypoglycemic agents and type 2

19   diabetes, the UGDP remains with us as you all know.    I

20   might just remind you that was a study with 200 patients

21   per arm, total of five arms depending in part how you

22   counted.   The study went on for six or eight years,

23   again, depending at which point you thought the study

24   was actually terminated.    So, in that instance, we're

25   talking about 8,000 person years to get to the
26   conclusions that were raised, and you well know those
                                                          231

 1   conclusions suggesting that tolbutamide had twice the

 2   cardiovascular mortality.    That now appears as class

 3   labeling in oral hypoglycemics.

 4                The other thing that we recognized as a

 5   background issue is that in the face of UGDP and around

 6   some of the other issues that you all have been

 7   discussing, one of the really epidemiologic and medical

 8   and therapeutic issues is what is the natural history

 9   of treated type 2 diabetes.    We all know that that's

10   the critical issue.   We didn't think for a moment that

11   we were going to be able to get answers to all those

12   questions in the design of this study.   We targeted the

13   study to ask the question of what is the cardiovascular

14   risk of repaglinide and appropriate comparitors?    Let

15   me show you how that process went.

16                I might just say, the expert committee was

17   made up of Sean Dinneen from the Mayo Clinic who brought

18   some epidemiologic background around the expected

19   background rates because that was an issue powering the

20   study.   Saul Genuth has previously served on this

21   Advisory Committee and was on the board for the DCCT.

22   Bob Makuch is chairman of biostatistics at Yale, and

23   Jamie Rosenzweig has participated in many clinical

24   trials of diabetes at the Joslin Clinic.

25                Before we started, we did ask ourselves
26   what other Phase IV approaches are available to looking
                                                           232

 1   at the performance of a compound in the marketplace.

 2   Of course, the usual epidemiologic observational

 3   methods including passive surveillance, prescription

 4   event monitoring, and other registry and cohort

 5   approaches and case control studies are out there.

 6   Just to cut to the quick on that, the issues in all of

 7   them is selection bias.     Without a randomized control

 8   process, we felt that it would be hopeless to use

 9   epidemiologic methods around the issue of

10   cardiovascular associated events in type 2 diabetes

11   long-term.

12                  That took us quickly to talking about and

13   discussing a randomized, but simplified, clinical

14   trial.   We recognized, as I've already said, that this

15   was destined to be a very large undertaking.      It would

16   demand, by definition, internal comparitor

17   comparitors.     That there were some very real

18   feasibility issues but, in fact, at the end of the day

19   if one wants to ask these kinds of questions, that's

20   probably the only real option.

21                  Bruce Stadel here at FDA as well as Patrick

22   Waller and others in thinking about and talking about

23   Phase IV studies, when they were discussed, pointed out

24   -- and I like these points very much.     I think they're

25   most appropriate -- that one wants to be mindful that
26   the study be conducted in representative populations,
                                                           233

 1   that you choose the right endpoints that are

 2   meaningful.    More often than not, that means hard

 3   endpoints that are less subject to observer bias and

 4   other sorts of biases.    That the studies be

 5   sufficiently powered and yes, that the results come in

 6   in your lifetime and mine and that they not be

 7   historical undertakings.    So that timeliness of the

 8   study as well as duration, appropriate duration, so

 9   that one can look at long-term effects are important

10   points to consider in the design of such studies.

11                  The one point that's not on here is that

12   one also ought to choose appropriate -- and that means

13   clinically, real world practice appropriate

14   comparitors.    So, our group met and we began discussing

15   what comparitors do we feel are appropriate and

16   essential.     We discussed at length in terms of entry

17   criteria, would one want to restrict this to if not

18   naive type 2 diabetics, naive to oral hypoglycemics,

19   then indeed, patients who did not have a history of

20   cardiovascular disease.    We, in fact, concluded that

21   in the spirit of being representative as well as the

22   simplification process, that one ought to take, in

23   effect, all comers and not use a restrictive approach

24   to entry criteria.

25                  We felt that the endpoints of critical
26   interest were cardiac hospitalizations and all-cause
                                                           234

 1   mortality.   We discussed at some length what stopping

 2   rules might look like and some of the ethical issues,

 3   not unlike some of the discussions about placebo.     We

 4   feel -- and just to mention it here -- that conducting

 5   this with a no treatment placebo arm would be, at this

 6   point of the state-of-the-art and science of treating

 7   type 2 diabetes, would be unethical.    I'm glad to see

 8   a couple of you nodding because theoretically, that

 9   would be ideal and we don't question that.   But we think

10   it's not appropriate and not on.

11                We recognized that one would have to target

12   the several arms toward achieving reasonably

13   comparable therapeutic goals in terms of blood glucose

14   and hemoglobin A1c.   There are issues there about how

15   tightly you run the trial and how tightly you measure

16   those endpoints.   We would view that as a variable to

17   enter into the analysis as opposed to one that one wants

18   to target to a fixed level.    We discussed secondary

19   endpoints particularly including therapeutic failure

20   rates and serious hypoglycemia, not least around some

21   of the issues that you all have been discussing here

22   and recognize what we're talking about here would be,

23   if you will, a population, a very large population,

24   followed over time in actual practice, as it were.

25                So, that was all by way of background.   The
26   next issue that we grappled with at length, and I've
                                                            235

 1   been talking around it here in these few moments, was

 2   how large does such a study have to be?     The size is

 3   going to be driven by the number of arms of the study,

 4   the level of statistical power.   We basically said this

 5   ought to have 80 percent power with a p of .05.     The

 6   relative risk issue here is that this is, in effect if

 7   we're talking cardiovascular endpoints in equivalence

 8   trial -- that is, we would be targeting to demonstrate

 9   a relative risk of one with, yes, a confidence limit

10   around it that we selected as .07 to 1.3.     As one

11   narrows that confidence limit, by the way, the size of

12   the study goes up logarithmically.

13               We discussed what would be a practical,

14   reasonable, appropriate and useful length of follow-up

15   and we chose a period of three years of patient

16   observation, recognizing there would probably be a

17   one-year period of enrollment.    So, on average, you

18   would have 42 months or three-and-a-half years of time,

19   person time, per each individual in the study.   I might

20   just say that one of the reason we came to that is we

21   felt that after three or four years of following

22   patients, patient crossover to other drugs, patient

23   migration, issues of loss to follow-up and the like

24   would become very real.   Also, in the spirit of

25   timeliness, we would rather have more patients for a
26   relatively shorter period of time than a smaller number
                                                            236

 1   of patients for a much longer period of time.

 2   Obviously, a set of compromises.    I don't think there

 3   are any hard and fast rules in that.

 4                We also talked about what would be in the

 5   one of the determinants, maybe one of the main

 6   determinants of sample size is what's the expected

 7   background rate of cardiac hospitalizations in

 8   all-cause death?   We took as four percent to power the

 9   study.   We did examine a range actually down from three

10   to up to five percent of rates.    You saw in the slide

11   that John Whisnant presented a moment ago where we

12   looked at the data sources to get those estimates.     It

13   included the UK prospective diabetes trial, the WISTAR,

14   Wisconsin epidemiologic ophthalmic study, et cetera.

15                Lastly, we discussed at some length

16   practical limitations of conducting the study in terms

17   of everything ranging from patient eligibility to what

18   kinds of materials would be provided to patients?      How

19   would drug be provided?    Where would the sites come

20   from and the like?    I don't intend to go into all of

21   those details here.   Having said all of that, this is

22   the study that we propose at this point.     We have

23   discussed this with FDA and let me just walk you through

24   this because this ought to be, perhaps, a one slide

25   presentation and this is it.
26                The study we would propose would enroll
                                                          237

 1   type 2 naive or previously treated type 2 diabetic

 2   patients previously treated with oral hypoglycemics

 3   who have a hemoglobin A1c equal to or greater than eight.

 4   The only age restriction we would put on it would be

 5   greater than or equal to 45, simply because the events

 6   of interest are going to be much less frequent below

 7   that age group.   Ideally, we would like to have this

 8   population be representative of the type 2 diabetic

 9   population in the US.

10                We proposed a three arm study which would

11   be achieved through randomization,     repaglinide and

12   equal size of arms, insulin and glipizide.     The study

13   would be, as I've said, three years in duration in terms

14   of patient observation.    That is, each patient but

15   since there would be a patient enrollment here, that

16   would translate into three-and-a-half years on average

17   of patient observation.    So, that would give us

18   something on the order of 20,000 patient years of

19   exposure.   Again, this is an enormously large

20   undertaking, needless to say.

21                In part, as a consequence of that, the data

22   variables collected at baseline -- we would collect all

23   of the appropriate and important covariates.    But then

24   along the way, we would restrict data collection to

25   those endpoints of critical interest.     We would get
26   baseline hemoglobin A1c's and do that annually.   EKG at
                                                         238

 1   baseline, again for allowing for analysis by that

 2   covariate.    We would collect the endpoints of interest

 3   meaning cardiac hospitalizations, any changes in drug

 4   therapy, any episodes of hypoglycemia and death.     We

 5   propose that the Drug Safety Board would meet at the

 6   end of the first year and then at six month intervals,

 7   and would be armed with the usual kinds of interim

 8   analyses and stop rules.

 9                 So, that's a very quick overview of what's

10   proposed for study.    We feel this study would indeed

11   meet the requirements of representative population,

12   timeliness.    We feel it is feasible as described and

13   it has the appropriate comparitors.    Needless to say,

14   we discussed these comparitor arms at length in terms

15   of what the several options might be and we're prepared

16   to discuss that if that seems appropriate here.

17                 Well, just to summarize what I've

18   described here in very outline fashion is a proposal

19   for a randomized, simplified clinical trial.   It would

20   be obviously multicentric and perhaps multinational.

21   Exposure would be, on average, three-and-a-half years.

22   The comparitors are, as mentioned, repaglinide,

23   insulin, glipizide.    Primary endpoints cardiac death,

24   hospitalization for acute cardiac disease, all-cause

25   mortality.    Secondary endpoints would be hypoglycemia
26   and treatment failures.
                                                         239

 1                Let me, before I take questions, invite Dr.

 2   Kurt Furberg who is known to many of you and is a most

 3   prominent clinical trialist and cardiac epidemiologist

 4   to make some comments.   Then both of us would entertain

 5   your questions.

 6                DR. FURBERG:    Thank you, Gerry.

 7                Mr. Chairman, colleagues, I was asked by

 8   the sponsor to take a look at the cardiovascular event

 9   data.   I think the charge to me was to give some advice

10   as to whether the observed findings represented noise

11   of random variation or a true signal.   The limitations

12   of that are quite obvious:    small numbers, different

13   trials.

14                My first approach was to look at the

15   totality of the evidence, to look at the three outcomes,

16   all cardiovascular events, the serious cardiovascular

17   events -- which is a subset of the all cardiovascular

18   events -- and then trimming it down even more to the

19   acute ischemic events.    You already heard that there

20   were differences between the trials and within the

21   trials, so I think the proper way of looking at that

22   is to consider the adjusted analyses.

23                If I look at the all cardiovascular events,

24   look at the repaglinide versus all comparitors pooled

25   in adjusted analyses and I focus on all cardiovascular
26   events, I get a risk ratio of 1.14 which is very, very
                                                         240

 1   close to unity.   But in doing so, I realize that all

 2   cardiovascular events is a mixed bag, including

 3   symptoms like palpitations, findings from physical

 4   exam like murmur, and then serious events.     So, it

 5   makes sense to try to focus the analyses on the more

 6   important events.

 7                So, if we move down then to the serious

 8   cardiovascular events, again, pooled analyses, the

 9   risk ratio goes up.    It's 1.55, but the confidence

10   interval includes unity.    So, the difference is not

11   statistically significant.    Still in that group, I

12   defined events like peripheral ischemia, thrombotic

13   events, cerebrovascular events, arrhythmias, atrial

14   fibrillation.   Again, a fairly mixed bag.    It's hard

15   to think about a mechanism by which the drug could cause

16   these problems.

17                So, looking at the acute ischemic events,

18   again, pooled analyses, risk ratios, almost at unity,

19   1.02.   That composite outcome includes angina --

20   particularly angina leading to hospitalization, acute

21   myocardial infarction, coronary artery disease, and

22   myocardial ischemia.    So, in commenting on the

23   totality of the evidence, I would say I don't see any

24   significant overall increase in adverse cardiovascular

25   events in the completed trials.
26                I agree with the approach taken that you
                                                           241

 1   also need to look at the individual comparitors

 2   including placebo.   You may want to look at the

 3   individual trials, but limitations methodologically

 4   are even more apparent.    The numbers that were small

 5   to start with are getting smaller, and the findings are

 6   more susceptible to imbalances that we've heard about.

 7   They increased what here is the multiple comparisons.

 8   I don't know how many we have, but probably 30 or 40.

 9   To assign appropriate level of significance is

10   apparent.   When I look at the findings, I don't see

11   anything that is consistent.   So, my conclusion is that

12   the findings of the individual trials should be

13   interpreted cautiously.                In conclusion, I

14   do not believe that the available cardiovascular event

15   data should be a reason for concern.   I've always been,

16   for now 25 years, a proponent of large long-term trials

17   and I really welcome the commitment by the sponsor to

18   support the big trial.    I think it's important from a

19   clinical point of view, public health point of view,

20   to get data on cardiovascular mortality and morbidity

21   from a comparison of repaglinide and the other standard

22   treatments available today.    Ideally, like you, I

23   would have liked to see a placebo control trial but that

24   is not feasible.   That was underscored at a meeting

25   that I attended about a month or so ago, a special
26   emphasis panel sponsored by the National Heart, Lung
                                                           242

 1   and Blood Institute.

 2                 So, I think the trial is recommended.       I

 3   think it's the best we can do.    It's a major commitment,

 4   major advance, as I see it.      If we see a difference

 5   favoring one treatment or another, that could have

 6   major implications for the treatment of type 2

 7   diabetes.    I think what the company can do is pray that

 8   they come out ahead.   They should be satisfied if they

 9   are equal and take the consequences if they come out

10   as losers.    Thank you.

11                 ACTING CHAIRMAN SHERWIN:     Dr. Kreisberg

12   and then Dr. Marcus.

13                 DR. MARCUS:   I need to go to the airport,

14   so can I play through?

15                 DR. KREISBERG:     Yes.

16                 DR. MARCUS:   I liked that study very much,

17   but I'm worried about one issue related to recruitment

18   and retention.    That is, it seems a little bit

19   inflexible for patients who might be uncontrolled on

20   any of these drugs you're assigning them except insulin

21   where, of course, a dose could be very flexible.

22                 I wish there were some way you could build

23   into it that a person who is assigned to one of the oral

24   agents could have added to their regimen, metformin or

25   troglidazone, some sensitizer.      I assume you've
26   considered it.    Is there any way to work it in?
                                                           243

 1                DR. FAICH:   The answer is we discussed

 2   that and I don't think we've come to final conclusions

 3   on it.   But there's little doubt that one will have to

 4   have some kind of rescue therapy -- that's not quite

 5   the right word, but augmentation therapy for many of

 6   these patients and we recognize that.   That makes the

 7   analysis difficult, but on the other hand, if you look

 8   at this as an intent-to-treat from the start across the

 9   three arms, that's what we would propose to do.

10                The other thorny issue in this is how do

11   you control for level of hemoglobin A1c achieved as the

12   critical confounder without, in fact, driving the

13   treatment?   The place we came to on that is, we said

14   set treatment goals, encourage the achievement of those

15   goals, and then deal with it in the analysis.   But it's

16   an extraordinarily difficult -- both those questions

17   were difficult for us.

18                DR. KREISBERG:   It's an interesting

19   undertaking because the baseline risk of the patients

20   is high.   Rather than introducing a therapy that is

21   expected to lower the risk, you're actually introducing

22   a therapy that may increase the risk further.   So, it's

23   different from a lot of previous trials.

24                But the thing that strikes me is, how do

25   you randomize these patients?    Because if they're
26   randomized simply as first come, first served into the
                                                           244

 1   various treatment arms, how do you guarantee that the

 2   coronary heart disease risk factors, which probably are

 3   more important than the diabetes or even the therapy

 4   in determining a risk, are equal among the groups?

 5                DR. FAICH:    The ideal way to do that, of

 6   course, would be do a block stratified randomization

 7   on the front end to be sure that if someone has risk

 8   factor, they have equal opportunity.    We talked about

 9   that and I don't think we've completely ruled it out.

10   We felt that this was an issue, given 2,000 patients

11   per arm, it was highly unlikely in contrast perhaps to

12   the much smaller trials we've been hearing about today

13   that we would get a maldistribution at the end of the

14   day.

15                I mean, those are the two choices, in fact.

16   Either way, you have to handle it in the analysis.   What

17   we certainly did not want to do, and we talked about

18   it, would be to stratify to ensure sufficient numbers

19   so that we could analyze independently patients with

20   prior cardiac risk versus patients without because we

21   knew that would blow the sample size through the ceiling

22   again.

23                DR. HIRSCH:    I'm sure you don't need

24   reminding that the UGDP study didn't show significant

25   results until about the fourth year, I think.   That may
26   not be rectifiable by numbers.    I mean, if you have
                                                           245

 1   another model that there's an incubation period of

 2   whatever the effect is, it doesn't matter how many

 3   people you put in the study, you may have to wait four

 4   years to see it.     It would be a shame to make a cutoff

 5   point of three or four years at this point, but perhaps

 6   an analysis at that point and an extension if needed.

 7                 DR. FURBERG:    Dr. Hirsch, I think you have

 8   a very good point.    There could be a lag time to benefit

 9   or harm if that is what we are dealing with.    It's very

10   common now in many of the big trials to extend follow-up

11   in a lot of examples particularly from the

12   NIH-sponsored programs.      So, I think that, we can deal

13   with.

14                 DR. FAICH:    The only other comment I would

15   make is you know it may well be that UGDP didn't see

16   those events because it enrolled relatively younger new

17   onset diabetics in the initial enrollment cohort.

18   That was one of the reasons why we said we didn't want

19   to do that.   I would expect that there would be probably

20   a relatively linear accumulation of cardiac events over

21   time because of that in this kind of study.    But you're

22   right.

23                 DR. HIRSCH:     Well, it's a very important

24   consideration.

25                 DR. FAICH:     The other thing is, is this
26   biology?   Are we really talking about an induction
                                                           246

 1   period or an incubation period or a latency period?

 2   And I think our answer is, we don't know the answer to

 3   that.

 4                DR. HIRSCH:    That's my point.

 5                DR. FAICH:    Sure.

 6                DR. HIRSCH:    Therefore, the

 7   interpretation of the one-year data that we have, it's

 8   almost impossible to interpret them in any way.

 9                DR. ILLINGWORTH:      Yes, a couple of

10   questions.   One is, I wonder why you didn't consider

11   including troglidazone instead of insulin since you

12   basically have three regimens that are going to raise

13   insulin levels?   Hyperinsulinism itself may be a risk

14   factor and troglidazone would potentially give you a

15   positive control looking at that other mechanism.

16                DR. FAICH:    Yes.

17                DR. ILLINGWORTH:      And the second question

18   concerns stratification for lipid lowering drugs based

19   upon the data now of clear benefit from treatment.

20                DR. FAICH:    Right, right.   Let me take the

21   troglidazone, well, we did discuss it.       One way to

22   think about that would be at a fourth arm.    I don't have

23   to tell you what that does to both sample size and

24   complexities of running it, and maybe loss of a

25   potential augmenting therapy.      So, that was one reason
26   why we said no fourth arm.
                                                           247

 1                 The issue about insulin versus let's say

 2   troglidazone is historically we know that insulin from

 3   UGDP to some considerable degree is the question.      In

 4   terms of being able to titrate down in dose, insulin

 5   probably -- if you have to make a choice between those

 6   two, has much to recommend it.     Maybe the negative side

 7   of thinking about an insulin arm is how will that affect

 8   patient recruitment on the front end?       Will you then

 9   end up with a less than fully representative

10   population?    We did say we would want a pilot list to,

11   in fact, look at that issue.    That was the way the logic

12   went on it.

13                 DR. FURBERG:   And regarding the lipids, I

14   think you're right.   There are other factors that will

15   have to be considered also:      treatment of blood

16   pressure, aspirin and so on.     I think the trial you're

17   talking about would be on top of good medical treatment

18   with assurances that you have balance between the

19   groups.

20                 DR. MOLITCH:     Microalbuminurea is

21   probably as important as cholesterol as a risk factor

22   as well.   But whether you treat that, we don't know

23   makes any difference in cardiovascular disease.       So,

24   that's yet another unanswered question.       My guess is

25   that you're going to have so many variables that are
26   going to be treated in so many different ways that
                                                          248

 1   you're going to end up like the UKPDS and not find

 2   anything at the end.

 3                DR. CARA:   You know, I can appreciate the

 4   time and effort that you put into designing a study like

 5   this, but my concern is somewhat similar to Dr.

 6   Molitch's in the sense that I don't know how feasible

 7   a study like the one you described actually is.   I mean,

 8   you're talking about a study that involves at least

 9   5,000 to 6,000 patients assuming a dropout rate of

10   somewhere between 30 to 40 percent.     You'd have to

11   recruit at least 10,000 patients for a study of this

12   sort, and that's only with the three arms that you

13   described.   I don't see how feasible that actually is.

14                DR. FURBERG:   Let me tell you that I'm

15   involved with Women's Health Initiative, NIH

16   sponsored, looking at three different interventions in

17   a two by two by three factorial design.      If you're

18   looking at complexity, that is one when it comes both

19   to enrolling and interpreting and dealing with that

20   issue.   I'm also involved in the NHLBI-sponsored

21   ALLHAT study.   Forty thousand patients with

22   hypertension comparing four different regimens for

23   treatment.

24                So, I think there is a collective wisdom

25   and experience out there to deal with these issues.
26   You're absolutely right.    We need to be fully cognizant
                                                          249

 1   of the issues that you have raised and others, and be

 2   sure that at the end, we can interpret our findings.

 3                 DR. CARA:    Well, I appreciate your

 4   comments.    Thanks.

 5                 My concern is whether the sponsor would

 6   want to commit to a study of that sort.   I mean, you're

 7   talking about studies that have been essentially

 8   proposed and supported by the government.     That's one

 9   thing.   Here, we're talking about something completely

10   different.

11                 DR. FAICH:   Let me try to respond to that

12   to the best of my ability and maybe turn it back to the

13   sponsor.

14                 I think, as Dr. Furberg has pointed out,

15   that one of the reasons why we can move into and it's

16   something of a paradigm -- thinking about these very

17   large trials is you also have to think about doing them

18   not in phase III heavily monitored, 80-page, 100-page

19   case report forms.     You really have to get down to the

20   critical covariates and think about them as perhaps

21   more of an epidemiologic undertaking.      There's

22   randomization on the front end.      There's no question

23   you're assigning therapy, so that makes them a trial.

24   But they have a different flavor in their aconda.     We

25   did, by the way, power the study allowing for a 20
26   percent dropout per year.     That was yet another reason
                                                         250

 1   why we said there's no way this study can go on much

 2   beyond three years.   There will be few patients left

 3   and that was part of this kind of we took this at a

 4   realistic approach.

 5               The other thing in terms of the cost of the

 6   study is, we don't think we're going to answer

 7   everybody's questions.   I don't think we're going to

 8   answer the microalbuminurea question related to

 9   glucose control.   This was very targeted toward the

10   endpoints that we mentioned.   While there are 100 other

11   natural history questions that one would love to

12   answer, each time you start doing that, that's where

13   the price really starts to go up as well.   So, that was

14   the other issue.   We think that this is actually cost

15   feasible and we did some preliminary costing on it.

16   Again, I can't speak for the sponsor around that issue.

17               ACTING CHAIRMAN SHERWIN:    Dr. Critchlow?

18               DR. CRITCHLOW:     Well, clearly, the

19   feasibility has something to do with whether you think

20   the relative risk is somewhere in the nature of 1.1 as

21   you said your meta-analysis might have shown versus the

22   1.5 to 2 full risk that we saw based on the preliminary

23   data.

24               A couple of questions.    On the current

25   safety data -- I know the numbers are small but is there
26   any evidence that the risk varies by whether or not
                                                            251

 1   there was previous cardiovascular disease?      Was it

 2   essentially comparable, two-fold increase, among those

 3   with and without disease or did that vary?

 4               DR. FAICH:     John Whisnant showed that,

 5   that if you have prior cardiovascular disease, that in

 6   and of itself increases your risks some threefold, not

 7   surprisingly.

 8               DR. CRITCHLOW:     Okay, so the difference

 9   between the repaglinide versus whichever was three or

10   four fold among those with -- cardiovascular disease?

11               DR. FAICH:     Oh, no, that was in the

12   unadjusted analysis which you saw --

13               DR. CRITCHLOW:    No, but if you stratified

14   by previous cardiovascular disease, what was the

15   relative risk in each of those, among those with prior

16   disease and then among those without prior disease?

17               DR. FAICH:    Oh, yes.   The numbers go away.

18   You can only do that in a multivariate approach.

19   That's where you saw the curve that was higher --

20               DR. CRITCHLOW:     I saw the adjusted --

21               DR. FAICH:     -- then come down to be

22   comparable to glipizide.

23               DR. CRITCHLOW:    I mean, I would think it's

24   relevant whether or not they had prior disease or not.

25   I guess there's no data to assess at this point, whether
26   the risks would vary --
                                                             252

 1                  DR. FAICH:   Yes, the one study that's

 2   critical in that regard was 049.     John, you'll recall,

 3   showed the data of the imbalance in the randomization

 4   relative to prior cardiac disease.       But going beyond

 5   that, the numbers just disappear and you really don't

 6   have much of an analytic opportunity.

 7                  DR. CRITCHLOW:   So, in the current study,

 8   you powered it to do whatever, your stopping rule is

 9   based on what?    That if you see something in excess of

10   1.5 or --

11                  DR. FAICH:   We didn't work through all the

12   stop rules.     It's clear that you have to allow for a

13   wider confidence limit early-on.      Then as you get more

14   and more power, you narrow that down and that's not

15   unprecedented.     So, I don't know what that first year

16   would be and it is risky, obviously, because you don't

17   want to stop the study prematurely at the same time you

18   want to discharge the ethical responsibility.       It's

19   probably two or greater than two at year one for the

20   relative risk stop rule issue, and then it would begin

21   to decline after that.      Bob Makish, actually, has had

22   a lot of experience with that and brought that to our

23   discussions.

24                  I would point out again, to some extent,

25   this is a work-in-progress and I'm presenting it --
26                  DR. CRITCHLOW:   No, I understand that.
                                                               253

 1                  DR. FAICH:     -- conceptually.

 2                  DR. CRITCHLOW:     And your total projected

 3   incidence rate of disease is on the nature of what, ten

 4   percent?

 5                  DR. FAICH:     Well, no, it's actually more.

 6                  DR. CRITCHLOW:     It's four percent per

 7   year.

 8                  DR. FAICH:     We're estimating four percent

 9   a year so it's more like 12 percent, and then you have

10   the dropout.    So, it is ten percent.    That gets you 200

11   events in each arm.    So, you know, we reckon we'd have

12   a lot of power, relatively speaking, including to allow

13   us to do some stratification.

14                  ACTING CHAIRMAN SHERWIN:     Dr. Kreisberg?

15                  DR. KREISBERG:     Did your committee

16   consider the possibility that there would be

17   confounding factors associated with therapy?        That is,

18   improvement of glycemic control in the hemoglobin A1c

19   potentially reducing the risk in the drug or a drug,

20   potentially increasing the risk?

21                  DR. FAICH:    Yes, I think that's absolutely

22   right.   That is the most difficult thing about

23   designing the study.        You have a choice.   You can try

24   to push everybody to a targeted control level that

25   actually means central laboratory.        It means a lot of
26   feedback.   It means a lot of work at the patient/doctor
                                                             254

 1   level.    So, that implies a lot of cost as well, to say

 2   nothing of whether you can really achieve it or not.

 3   Or you can, in fact, say no, we'll give those proposed

 4   targets and we'll analyze that, recognizing it may well

 5   be a confounder for the outcomes of interest.   But it's

 6   a confounder you probably can analyze for to some

 7   extent.   I would say that from a design and analytic

 8   viewpoint, that's the toughest issue in the whole

 9   study, without question.

10                DR. CARA:    But you would really have to do

11   some form of dose escalation study with some target

12   endpoints because otherwise, you really can't evaluate

13   the potential risks and the potential benefits.

14                DR. FAICH:   I agree.   We would handle that

15   probably by putting that in the protocol of suggesting

16   when therapy changes, et cetera, within certain limits.

17   See, the issue isn't whether you can provide that

18   guidance.   It's how much you enforce it in terms of the

19   cost of the study and what you're doing.

20                The other way to answer that question to

21   some extent is, is this to answer a biologic question

22   or a actual care question?     Is this a study of

23   effectiveness or idealized efficacy?      I mean, one

24   issue is if you start to push it very hard in terms of

25   protocol driven study in terms of the outcomes, you
26   answer a better scientific question but it may be less
                                                                 255

 1   generalizable.    Then there's a dilemma in that as

 2   you'll recognize as well.

 3                 DR. CARA:    But    you    didn't   mention

 4   safety --

 5                 DR. FAICH:   Well, we would collect adverse

 6   events.   We would collect all SAEs, of course, and

 7   follow up on them appropriately, submit them

 8   appropriately and the like.

 9                 ACTING CHAIRMAN SHERWIN:        Just a quicky.

10   Your choice of glipizide, was that based on the

11   frequency of -- the reason I bring it up at all is in

12   your preliminary data, you had very few patients on

13   glipizide and they seemed to do worse than the drug.

14   Whereas, you have a lot more data with glyburide and

15   they seem to have less events.          So, you know, it just

16   seemed to me that the logical choice would have been

17   glyburide.

18                 DR. FAICH:   Yes, I would say we talked

19   about it.    John wants to go ahead.

20                 DR. WHISNANT:      I wouldn't conclude from

21   the earlier data that --

22                 ACTING CHAIRMAN SHERWIN:        No, there were

23   not enough patients to say anything.         No, I understand

24   that.

25                 DR. WHISNANT:      And at least from our
26   perspective as a diabetes company, glucatrol XL,
                                                           256

 1   glipizide, long-acting glipizide in this country is the

 2   largest, most available therapy.      It is also

 3   pharmacokinetically the most different drug.       So, if

 4   we're looking at a hypothesis that our therapy is new

 5   and that this dosing PK mode needs longer term testing,

 6   then that gives us the maximum delta difference in

 7   looking at that hypothesis.

 8                 ACTING CHAIRMAN SHERWIN:     Okay.

 9                 DR. CARA:    I might be asking you the $20

10   million question here.     I don't know.   Given the fact

11   that typically, phase IV studies are fairly poorly

12   monitored and poorly controlled, are you proposing that

13   these studies be done prior to approval of the drug?

14                 DR. FAICH:    No, that is the $20 million

15   question.   No, this is designed and proposed as a

16   post-market -- I think you probably feasibly couldn't

17   do this under IND rules.     I think there needs to be a

18   study done with some rigor.     I think it has to be done

19   in a credible manner, but I don't think you can talk

20   about monitoring each site and validating each data

21   bit.    One would want to think about doing that on a

22   sampling basis, looking for systemic error and the

23   like.   That's yet another reason why I would see this

24   as a post-marketing strategy.

25                 DR. CARA:    I mean, do you think in all
26   honesty that a study of this sort could be done as a
                                                           257

 1   phase IV?

 2                 DR. FAICH:   Oh, I absolutely think it can.

 3   You know, actually, Richard Pieto and other people have

 4   talked about when you think about feasibility criteria

 5   for these kinds of studies.     It has to be relatively

 6   common disease.    Therapy has to be relatively easy to

 7   apply.    You can't have complex diagnostic

 8   requirements.     The outcomes have to be objective and

 9   hard.    You have to be able to have sufficient power and

10   so on.   I think this study fits that.   I think that this

11   is very much a feasible study.

12                 Again, I think one has to approach it quite

13   differently than the usual phase III study.       That's

14   why, for some of us, this is -- I mean, there are

15   challenges at many levels but some of that is

16   philosophical change as well in terms of how to approach

17   these trials.

18                 DR. MOLITCH:   I'll just say one more time

19   for the record that I think that in a study, if it's

20   going to be as loose as you make it sound to be with

21   the major focus being on the drug, that is probably of

22   five risk factors for cardiovascular disease.       It's

23   the fifth one on the list that will actually affect

24   cardiovascular disease -- after LDL cholesterol, after

25   microalbuminurea, after glycemic control, after blood
26   pressure control and the type of blood pressure
                                                           258

 1   control, that this three drug analysis is at the bottom

 2   of that list as the thing that will affect cardiac

 3   outcome.

 4                  So, my guess is that we're really not going

 5   to see anything here because of all the other more

 6   powerful, confounding events.      I would suggest that

 7   this study not be done in this design.

 8                  DR. FURBERG:   I'm part of another large

 9   NHLBI-sponsored study, a part of -- health study.    It's

10   a study looking at risk factors of coronary heart

11   disease and stroke and they are really something that

12   takes on more of a meaning as you get some gray hair.

13                  In that study, the two strongest

14   predictors of cardiovascular events are hypertension

15   and that is lipids lose their predictive power at a

16   certain age.    It's much less.   So, the important thing

17   according to our data is to deal with those.    The trial

18   will deal with the glycemia, diabetes, and the

19   hypertension we need to control.     I think in addition,

20   we may want to add in some other factors but I don't

21   think it's number five in the age groups that we are

22   looking at, the old.

23                  DR. MOLITCH:   Doing that 45 and older?

24                  DR. FURBERG:   Well, that is still under

25   discussion.     My recommendation is that we go up to at
26   least 55.   That's where the events are.    If you really
                                                             259

 1   want to do a study and get events, you don't start off

 2   at 45.

 3                DR. MOLITCH:     That was your design.     I'm

 4   sorry.

 5                DR. FURBERG:     Well, it's one that's

 6   proposed.   We haven't discussed all the issues, but at

 7   least that's one issue I agree with you on.         Go up in

 8   age and get the events up and focus on the two most

 9   important risk factors.

10                DR. NEW:   Are you going to use both men and

11   women?

12                DR. FURBERG:     Absolutely.

13                DR. NEW:     Then how will you control for

14   estrogen use and the cardiovascular risk -- pardon?

15                DR. MOLITCH:     Or raloxifine.

16                DR. NEW:     Well, that's marvelous.     But in

17   the meantime, I think that's --

18                DR. FURBERG:     I don't know whether it's

19   coincidence or not, but I'm involved in another study,

20   the HER Study, hormone, estrogen replacement study.

21   We're going to have results first half of next year.

22   I think that would guide us as to how to deal with that

23   issue.

24                ACTING CHAIRMAN SHERWIN:       Okay.

25                DR. FAICH:    I can't help it.    I just would
26   add one other thing.
                                                            260

 1                 ACTING CHAIRMAN SHERWIN:     I can't stop

 2   this Committee.

 3                 Roger, go ahead.

 4                 DR. FAICH:   Obviously, all the risk

 5   factors you can, and should, and need to collect at

 6   baseline.    So, that's one part of the answer.      The

 7   other part of the answer is that in this kind of study,

 8   I certainly think you have to, in fact, collect data

 9   on medications used.    So, you have the opportunity to

10   enter that into the analysis.

11                 By the way, I would do that as opposed to

12   measuring blood pressures because it seems to me that

13   medications as indicators, in many cases, are a more

14   accurate measure.    Because once you have to start

15   specifying how you measure other clinical variables,

16   which is not to say that they're not important, the

17   feasibility does get compromised.       Thank you.

18                 John, I should probably turn it back to you

19   to wrap --

20                 DR. WHISNANT:    Roger?

21                 DR. ILLINGWORTH:    Just one question

22   concerning the lipid stratification, just based on the

23   data we have available and NZPT guidelines.     Presuming

24   you're going to have an upper level of LDL in which you

25   can not be ethically untreated?
26                 DR. FURBERG:    I agree with that
                                                            261

 1   wholeheartedly.

 2               ACTING CHAIRMAN SHERWIN:     Okay.

 3               DR. FAICH:   I'll turn it back to you, John.

 4   Do you have a closing comment?

 5               DR. FLEMING:     Dr. Sherwin.

 6               DR. FAICH:     Dr. Sherwin, you have it.

 7               ACTING CHAIRMAN SHERWIN:     I just wonder,

 8   do we need to go to number five or can we skip number

 9   five?

10               DR. FLEMING:    I think the point of number

11   five would be to allow the company to very quickly --

12               ACTING CHAIRMAN SHERWIN:     Okay.

13               DR. FLEMING:     -- summarize their hard

14   findings and to give us an understanding of studies that

15   are currently in progress or immediately anticipated

16   starting.

17               DR. WHISNANT:     I take the signal that we

18   want this to be brief.

19               ACTING CHAIRMAN SHERWIN:     Very brief.

20               DR. WHISNANT:    Let's see if I can rise to

21   that challenge.

22               ACTING CHAIRMAN SHERWIN:     Okay.

23               DR. WHISNANT:     I should speak into a

24   microphone for the reporter in the back.

25               Novo Nordisk has submitted an NDA which has
26   been reviewed.    We believe the NDA includes more than
                                                            262

 1   an adequate basis demonstrating efficacy of this drug

 2   in the treatment of type 2 diabetes mellitus.

 3                  We believe that the safety profile of this

 4   drug has been demonstrated both by a safety margin of

 5   dose and by exposure of 1,228 patients over a year's

 6   trial -- actually 834 patients, fully exposed for more

 7   than a year.     We believe there's an adequate

 8   representation of patients in that database in order

 9   to assure the safety of this product.

10         Because of a number of cardiovascular events in

11   one trial which we believe has some considerable

12   question about randomization bias, we've made a major

13   commitment to at least propose for your consideration

14   on a post-approval basis, that we will carry out a study

15   that will include not only cardiovascular risk

16   monitoring, but will teach us more about the use of a

17   different kind of secretagogue therapy versus, I must

18   admit, our insulin in a comparitor long-term trial.

19                  There are a number of questions which

20   remain unanswered.    We are not asking for an indication

21   which addresses the natural history question, but we

22   believe that our long-term comparitor trial will assist

23   in generating that information.     We are not asking for

24   indications of combination therapy except for

25   combination with the drug that we've demonstrated a
26   significant benefit with, that is metformin.      We
                                                            263

 1   believe within these limitations that this new product

 2   should be added to the armamentarium available for

 3   treatment of patients with type 2 diabetes.

 4                I will be happy to answer any remaining

 5   questions.   I have some slides to indicate what our

 6   continuing program of studies is relative to the

 7   combination of this drug with a dione with regard to

 8   a purpose design study to get more data regarding the

 9   use of this drug relative to an advantage of

10   hypoglycemia severity, frequency, and relationship to

11   dose.   That study is also well along in its design phase

12   and ready to implement as soon as the Agency gives us

13   the go ahead.

14                We thank you very much.    I'll be happy to

15   address any further questions.

16                ACTING CHAIRMAN SHERWIN:    Okay.   The

17   group is suddenly silent.    That's terrific.    Okay.

18                Okay, I think we're about ready to address

19   the four questions that are posed to us.      We'll go

20   around the room from my right to left and then we'll

21   go backwards.   The first is, "are the various study

22   designs and efficacy endpoints adequate to assess the

23   effectiveness and safety of this drug?"      The various

24   study designs that have already been produced.

25                Joe?
26                DR. HIRSCH:    Well, let me just say as a
                                                           264

 1   prelude to my answer, if I may, in one sentence or two.

 2   This is an extremely interesting drug and a very, very

 3   promising one.   I would hope that some further animal

 4   studies which haven't been done utilizing genetically

 5   obese animals and animals otherwise having pancreatic

 6   dysfunction would make use of this fascinating drug to

 7   probe those abnormalities.

 8                My answer to one is going to sound awfully

 9   bad, but it's a very hardy no.   I do not feel that the

10   designs and efficacy endpoints are adequate to assess

11   the effectiveness and safety of the drug.    My specific

12   reason for that is that we haven't seen clear studies

13   as to how, in fact, the drug would be used.       I assume

14   it will have to be used in much higher dose or with other

15   drugs to be fully clinically effective.     There are not

16   adequate studies in my mind of that to permit me to say

17   that this is now -- I now understand all the ins and

18   outs of the efficacy and effectiveness and safety of

19   the drug, et cetera.    So, my answer is no.

20                ACTING CHAIRMAN SHERWIN:     Mark?

21                DR. MOLITCH:    My answer is yes.    I think

22   we have sufficient data to show that it is as effective

23   as other sulfonylureas and is at least as safe.    You're

24   right.   We don't know all the ins and outs of this and

25   I think a lot of those still need to be learned.      But
26   I think at this point, it does fit all those criteria.
                                                              265

 1                ACTING CHAIRMAN SHERWIN:    Roger?

 2                DR. ILLINGWORTH:   I would say yes.     I

 3   think the efficacy data that Dr. Fleming described has

 4   been demonstrated in placebo control trials and

 5   comparative trials.   I would echo my colleague's

 6   comments that we clearly need more data about the use

 7   of this drug in combination therapy with other oral

 8   agents.   I think we also need to look at potential drug

 9   interactions with Asians that haven't been looked at.

10   Things, in particular, that go with the cytochrome P450

11   system in the liver, how they're going to increase the

12   risk of being hypoglycemic if you're on erythromycin

13   or something like that.    But my answer is yes.

14                ACTING CHAIRMAN SHERWIN:    Okay.

15   Kathleen will read Dr. Marcus'.

16                MS. REEDY:   Dr. Marcus says yes but only

17   if the question is specifically whether this drug is

18   as good as the comparitor drugs that are already

19   approved.

20                ACTING CHAIRMAN SHERWIN:    José?

21                DR. CARA:    My answer is yes, but echoing

22   some of the comments that have been raised.       That is

23   that I think there are still a variety of questions that

24   remain that will hopefully tap into the true potential

25   of this drug.   I think there are a variety of
26   theoretical benefits to this drug that have been
                                                           266

 1   alluded to by the sponsor and by other members that are

 2   here.    Unfortunately, they have not been borne out by

 3   some of the clinical data that's been presented and I'm

 4   hopeful that additional studies will really address

 5   some of those issues.

 6                 I think the safety issues are still a

 7   question.    The data that's been presented so far

 8   certainly show that the drug is no worse than currently

 9   available therapy, but I've got to really make a point

10   about the fact that these are very short-term studies

11   in what is a very chronic type of disease.   I hope that

12   the sponsor will agree to and will, in fact, carry out

13   some additional studies, longer-term studies that will

14   address some of the safety issues that have been raised.

15                 DR. CRITCHLOW:   Well, I say yes and

16   seconding the comments of Dr. Molitch and Marcus and

17   Cara.    Just one additional comment is the efficacy

18   data, again, are not consistent with what might be

19   heralded as a breakthrough in terms of the action of

20   the drug.    It's clearly no better than what's out

21   there.

22                 ACTING CHAIRMAN SHERWIN:   I'll vote yes as

23   well, even though it's I believe barely adequate.    But

24   if I have to say between yes and no, I would come down

25   at yes even though I think all of the Committee has
26   serious concerns about safety issues and the lack of
                                                            267

 1   the full profile of its efficacy.

 2                DR. KREISBERG:    I'd like to compliment the

 3   sponsor on maintaining their composure during all this

 4   badgering.   It's been fairly remarkable to me that you

 5   could stay in such good humor.

 6                I vote yes.    My comment is that it is

 7   probably as good as, but I have not seen any evidence

 8   that it's any better than the sulfonylureas and it's

 9   probably as safe.   I think it's a very interesting and

10   exciting and new drug.     If the sponsor follows through

11   on all of the suggestions that have been made, that we

12   ought to know a lot more about it the next time around.

13   I think that that would be very gratifying.

14                The one thing that continues to bother me

15   -- and I guess doctors will be doctors and that's why

16   it bothers me -- is that the naive patients seem to be

17   more susceptible to the hypoglycemic effects of the

18   drug.   I would hope that the labeling would carry some

19   clear instructions on how to utilize this drug,

20   particularly in those types of patients.

21                ACTING CHAIRMAN SHERWIN:     Maria?

22                DR. NEW:    I'm having trouble deciding how

23   to vote and have finally decided I'll vote this way.

24   I think it's yes on the basis of the short-term data

25   which have been presented.     The drug is as at least as
26   good as any of the other glucose lowering drugs.
                                                              268

 1                The long-term data that's been presented

 2   to me -- after all, diabetes is not something you're

 3   going to treat short-term -- I find that the clinical

 4   studies don't address the actual practical practice of

 5   a diabetologist in regulating glucose control because

 6   the annual study -- those were the slides I wanted to

 7   show -- did not indicate to me that there was any benefit

 8   at the end of 12 months.      Now, I don't know whether

 9   there will be benefit at the end of two or three years,

10   so I really have a bifid vote.

11                ACTING CHAIRMAN SHERWIN:      Is there such a

12   thing as a bifid vote?

13                DR. KREISBERG:      If you're a pediatric

14   endocrinologist there is.

15                ACTING CHAIRMAN SHERWIN:      So, we've got a

16   half-a-vote on either side, is that it?

17                DR. NEW:    Yes.

18                ACTING CHAIRMAN SHERWIN:       Yes, okay.

19                Okay, the next question is "are there any

20   issues specifically related to the use of the

21   short-acting preprandial oral therapy that have not

22   been addressed by the sponsor?"

23                Maria?

24                DR. NEW:   No.     In fact, I think that's the

25   most exciting and valuable aspect of this new drug which
26   is that I'm convinced on the basis of data that taking
                                                            269

 1   the drug before the start of meals does improve the

 2   glucose control post-prandially.    I think they've done

 3   this study very well.

 4                DR. KREISBERG:    Well, I would agree with

 5   that.   I think it's an exciting new concept.    I would

 6   like for the sponsor to develop more information on the

 7   marked discordancy or variability between plasma

 8   levels of the drug and the acute response in terms of

 9   glucose disappearance, and to look into issues that

10   have to do with nutrient drug interactions.      Because

11   we didn't talk about that, but I think that's a

12   potentially important problem in things relating to the

13   bioavailability of the drug.

14                ACTING CHAIRMAN SHERWIN:    Okay.   Any

15   issues that have not been addressed?

16                Yes, I would like to see free levels of the

17   drug measured so I could better interpret the results.

18   I would urge the sponsor to develop an assay if at all

19   possible, put effort into that.

20                I'd also feel that the kinetics of the

21   drug, in terms of its biological action, have not been

22   adequately answered.    I believe that measuring insulin

23   levels with differing glucose levels makes it

24   uninterpretable to me to figure out what the duration

25   of action is.   That's critical.     Although the drug
26   levels are impressively up and down, I still don't know
                                                            270

 1   about the biology, kinetics of the biological response.

 2   So, I think that some fundamental studies should be done

 3   at least in animals.   If not, it could be done in humans

 4   to look at this.

 5                DR. CRITCHLOW:    I agree with Dr. Sherwin.

 6                DR. CARA:   I think there are several

 7   issues that really need to be addressed, or at least

 8   that I would like to see addressed.      One of the

 9   principal ones relates to mechanism of action.        It

10   seems that we've got just a black box where Prandin does

11   something that we can evaluate clinically, but we have

12   a very little insight into what's actually within that

13   big, black box.    Finding what's in there I think would

14   be very important, especially in terms of evaluating

15   efficacy and perhaps drug combinations or alternative

16   therapies that may work better.

17                What I would like to see is also some sort

18   of dose escalation study.     I feel like even though the

19   sponsor did a fairly good job of presenting efficacy

20   data, I would have liked to have known what that baby

21   could do in terms of really using optimal doses to get

22   optimal effect.    I don't think that was actually done

23   in many of the studies that were described by the

24   sponsor.   So, I think having more legitimate targets

25   of efficacy and pushing the drug a little bit would make
26   a lot of sense.
                                                            271

 1                  I'd like to see some data on convenience

 2   and some of the issues that were addressed before in

 3   terms of whether this is really as convenient as we

 4   think it is.     My impression is that it probably will

 5   be, but it would be nice to have corroboration of that

 6   impression.

 7                  Those are my major questions, if you will,

 8   that I'd like to see the sponsor take on.

 9                  ACTING CHAIRMAN SHERWIN:    Roger?

10                  Oh, what did Dr. Marcus do?

11                  MS. REEDY:   Okay, Bob Marcus.    Thorough

12   analysis of lipoprotein changes, both fasting and

13   post-prandial.     True incidence of hypoglycemia

14   measured blood glucose based on average and variation

15   on FPG on therapy.

16                  ACTING CHAIRMAN SHERWIN:   So the answer is

17   yes?

18                  MS. REEDY:   That's are there any issues?

19                  ACTING CHAIRMAN SHERWIN:    Yes, the answer

20   is yes.   Yes.

21                  DR. ILLINGWORTH:   Yes, I would echo the

22   utility looking at post-prandial lipemia as a potential

23   beneficial effect.     I think I'd also like to see some

24   other patient populations looked at.      The drug is bound

25   to albumen.    Patients with diabetes frequently develop
26   nephrotic syndrome.     What about patients with
                                                          272

 1   nephrotic syndrome?   Is the pharmacokinetics the same?

 2                 Then finally, just looking more at

 3   drug/drug interactions within the intestinal tract.

 4   Do things that delay gastric emptying affect the

 5   absorption?   We know about cimetidine.     What about

 6   some of the other proton pump inhibitors?      Do they

 7   affect absorption?

 8                 DR. MOLITCH:    I think the answer is yes.

 9   I think there is a major shortfall here on the part of

10   the sponsor as far as trying to really capitalize on

11   this short acting drug.   I don't know how fasting blood

12   sugar levels are decreased with this drug.      I don't

13   know what's happening to hepatic glucose output.      I

14   haven't seen any clamp studies looking at glucose

15   production versus glucose disposal.      I'm trying to

16   figure out what's going on first thing in the morning

17   here.

18                 I haven't seen them capitalize on this

19   short acting drug to increase insulin secretion acutely

20   controlling post-prandial blood sugar levels and then

21   using that in combination with a longer acting agent

22   overnight or using it with an injection of insulin

23   overnight which might be a really very, very nice

24   combination of therapy.      I think that those things

25   really should be done to help exploit the benefits that
26   this drug might give us.
                                                            273

 1                I'm concerned, as I mentioned previously,

 2   about patients who do have decreasing clearance.       I

 3   thought I detected a buildup of drug.    Given the large

 4   variability that we have seen before in the area under

 5   the curve with the dosing, that we need much more data

 6   on that to be able to say that we really can use it in

 7   patients who have decreased clearance.     I would not

 8   approve the drug for use in patients with decreased

 9   renal function at this point in time until we have such

10   data in addition to not using it in the patients with

11   hepatic disease.

12                So, I think that a lot of work needs to be

13   done in this area.   I think that work will prove of

14   great benefit to the sponsor for expanded use of this

15   drug.

16                ACTING CHAIRMAN SHERWIN:    Jules?

17                DR. HIRSCH:   I'm not sure we know the right

18   dose.   I'm not sure I know the best combination if it's

19   to be used with something else.     I don't know the

20   relationship to spontaneous hypoglycemia and all the

21   other things.   So, my answer is yes.

22                ACTING CHAIRMAN SHERWIN:    Another point

23   that I forgot to mention is that the trials really were

24   biased against numbers of minorities.     We should be

25   sure that all the various minorities that are
26   represented in the US are adequately studied to look
                                                              274

 1   at the risk benefit ratio in those populations.

 2                  Okay.   Number three:    "Is the excess in

 3   cardiac events reported for Prandin-treated patients

 4   compared to those treated with other therapies

 5   significant.     If so, how should this issue be

 6   resolved?"

 7                  DR. HIRSCH:    Didn't we just deal with

 8   that?

 9                  ACTING CHAIRMAN SHERWIN:     We did.

10                  DR. HIRSCH:    Well, we'll start here.    The

11   answer to that is I don't know.        I would hope that if

12   it ever is marketed, we'd find out.        Many of these

13   questions are sort of ambiguously worded, you now?         I

14   don't know what addressed by the sponsor means, et

15   cetera.   My answer to this is I don't know.

16                  DR. MOLITCH:    I'm not worried about the

17   excess in the cardiac events in this particular drug

18   compared to other drugs.      Nor am I particularly worried

19   about hypoglycemia which, to me, is not a big deal in

20   patients with type 2 diabetes that I treat with

21   sulfonylureas.     It's just not a major problem for me

22   with these patients.     I think that the study that was

23   outlined to address the cardiovascular issues is going

24   to be an Emperor's New Clothes where we're going to

25   spend millions and millions of dollars, pretend that
26   we know what we're doing and not get an answer.
                                                               275

 1                 DR. HIRSCH:     So, what is it, yes or no?

 2                 DR. MOLITCH:    I'm not concerned about it.

 3                 ACTING CHAIRMAN SHERWIN:      So, the answer

 4   is --

 5                 MS. REEDY:     No.   It's not significant.

 6                 ACTING CHAIRMAN SHERWIN:      Correct.

 7                 DR. ILLINGWORTH:     My answer would also be

 8   no.   It's a higher risk patient population.       Just

 9   looking at the cardiovascular events that occurred,

10   there wasn't anyone that predominated or that suggested

11   a red flag for patients with a history of a certain

12   arrhythmia or on certain other agents.

13                 I think it would be worthwhile monitoring

14   in any post-marketing surveys, is there any particular

15   patient group who is at higher risk for developing some

16   kind of arrhythmia or -- population in something like

17   that.    I think the data available -- I'd say no.

18                 ACTING CHAIRMAN SHERWIN:      Dr. Marcus?

19                 MS. REEDY:     Dr. Marcus says no.

20   "Post-marketing phase IV follow-up study is needed.

21   The proposed study seems reasonable but would prefer

22   to see a mechanism to add metformin for patients who

23   do not respond adequately to the assigned study drug

24   alone.   Without that, there may be a problem with

25   retention of subjects for three years."
26                 He already expressed that to you.
                                                           276

 1                  DR. CARA:    I think there are issues that

 2   suggest that the cardiac risks and other potential side

 3   effects may be significant.      I think what's going to

 4   happen as this drug becomes available -- presumably,

 5   it will be although we'll obviously see that in just

 6   a little bit -- I think people will start using it fairly

 7   liberally.   It's a fairly decent drug and I think that

 8   people will start using it for different sorts of

 9   populations.     I think that having a better sense of

10   what the potential risks and benefits are is important.

11                  Unfortunately, I don't think that a phase

12   IV study is the way to go about doing that.       I think

13   it will only be the test of time that will really tell

14   us about the long-term efficacy and safety of this drug.

15                  DR. CRITCHLOW:    The excess may be

16   significant.     I don't think there's anyway based on the

17   data we have to adequately address that.      I also agree

18   with Dr. Cara that without additional exposure data,

19   that would be the only way we would get that

20   information.

21                  MS. REEDY:    Is that a yes?

22                  DR. CRITCHLOW:    Yes.

23                  ACTING CHAIRMAN SHERWIN:    I have no idea

24   based upon the data.       I mean, it depends also on the

25   comparitors.     But you know, we just don't have enough
26   data, I don't think, to answer yes or no to that
                                                            277

 1   question.    I definitely think if the drug is released

 2   that a careful study should be done regarding

 3   cardiovascular risk.    Then the other issue is given the

 4   fact that we don't know about the risk, should we have

 5   anything -- you know, an insert regarding the potential

 6   risk since it is not established at this time, whether

 7   patients with a history of cardiovascular problems

 8   should use the drug with caution.

 9                  DR. KREISBERG:   That means no diabetic

10   will get it.

11                  ACTING CHAIRMAN SHERWIN:     Well, no, no.

12   No, that's not what I said.     I'm talking about people

13   who have had a previous MI who are currently being

14   treated for arrhythmia.     All patients with type 2

15   diabetes have a higher risk of cardiovascular disease,

16   but then there's a subgroup of patients who have active

17   ongoing cardiac risk that's significant.        I think in

18   that group of patients, given the lack of knowledge and

19   the preliminary data that we have, that we should be

20   a little bit cautious in terms of the prescribing

21   community.

22                  DR. KREISBERG:   Okay.   I'd like to answer

23   that I also don't know.     I'm not sure that the trial

24   that has been described will be successful.      I do think

25   that it's important to keep track of all of the adverse
26   events that occur with this drug.       As best I can tell,
                                                             278

 1   I can't see that the risk with this drug is any greater

 2   than it is with any of the other sulfonylureas, and I

 3   don't think it should receive preferential labeling.

 4                DR. MOLITCH:     Your answer is no then?

 5                DR. KREISBERG:    My answer is I don't know.

 6                MS. REEDY:   I don't know.    I have a

 7   special category for those.

 8                DR. NEW:   My answer is probably no because

 9   there isn't a significant difference from other drugs,

10   but it's very difficult for me to decide.

11                I would recommend that rather than this

12   elaborate study, that a careful post-marketing

13   monitoring study be done in which complications over

14   time are reported and tabulated, and then a

15   reassessment made.

16                ACTING CHAIRMAN SHERWIN:     Okay.   Now, we

17   get to the final.   Based on the efficacy and safety data

18   presented and your assessment of the overall benefits

19   compared to the risk of Prandin therapy, do you

20   recommend that this drug be approved for marketing?

21                DR. NEW:   My answer is yes.     I say yes

22   because I don't want to deprive my patients and my

23   family members of type 2 diabetes on an excellent,

24   short, rapid-acting drug.     I think that this drug, even

25   if used short-term, has been shown to be efficacious.
26   As I said in my previous vote, I don't see much
                                                               279

 1   difference from the complications of other

 2   glucose-lowering drugs.

 3                  DR. KREISBERG:     My answer is yes and with

 4   the same proviso that I think the sponsor develop some

 5   clear guidelines for physicians on how to use it.

 6                  ACTING CHAIRMAN SHERWIN:     I will vote yes

 7   also, even though I have a lot of concerns about all

 8   the problems we've discussed, particularly the numbers

 9   of patients studied who were virgin, naive patients.

10   I think that we need to have more information on that

11   group of patients.

12                  DR. CRITCHLOW:     I also say yes, basically

13   because it is not significantly worse than what is out

14   there.

15                  DR. CARA:    I vote yes, although I too have

16   reservations.     Unfortunately, I think it's only

17   through approval of this drug, that at this time appears

18   to be relatively safe and efficacious, that we will

19   learn more about its long-term effects and its true

20   safety and efficacy.

21                  ACTING CHAIRMAN SHERWIN:     Dr. Marcus?

22                  MS. REEDY:    Marcus says yes.

23                  ACTING CHAIRMAN SHERWIN:     Dr.

24   Illingworth?

25                  DR. ILLINGWORTH:    My vote is also yes with
26   the hope that they will conduct further studies looking
                                                             280

 1   at other hypoglycemic drugs in combination therapy to

 2   extend what's already been done.

 3                ACTING CHAIRMAN SHERWIN:     Dr. Molitch?

 4                DR. MOLITCH:    Yes.

 5                ACTING CHAIRMAN SHERWIN:     Dr. Hirsch?

 6                DR. HIRSCH:    No.

 7                ACTING CHAIRMAN SHERWIN:     Okay.   Well,

 8   I'm glad they were not unanimous, you know, after this

 9   session.

10                So, the final vote on the last question is

11   8:1, obviously.

12                I'd like to thank the sponsor for their

13   efforts today, the FDA, and all of you for spending the

14   whole day with us, a day that we thought would end very

15   quickly.   Thank you.

16                MS. REEDY:    I would like to ask the

17   Committee to please take all of your materials with you.

18   We are meeting in a different hotel tomorrow.      Your

19   blue folder contains tomorrow's agenda and questions,

20   so please take the blue folder.

21                If you would like to leave your materials

22   to be shredded, you may.    We'll have somebody else pick

23   that up.

24                (Whereupon, the meeting was concluded at

25   4:28 p.m.)
26
     281

 1

 2

 3

 4

 5

 6

 7

 8

 9

10

11

12

13

14

15

16

17

18

19

								
To top