DEPARTMENT OF HEALTH AND HUMAN SERVICES
FOOD AND DRUG ADMINISTRATION
CENTER FOR DRUG EVALUATION AND RESEARCH
ANESTHETIC AND LIFE SUPPORT DRUGS
Thursday, May 16, 2002
Holiday Inn Gaithersburg
Two Montgomery Village Avenue
Chair: Nathaniel P. Katz, M.D.
Executive Secretary: Kimberly Topper, M.S.
Janice Bitetti, M.D.
Charles H. McLesky, M.D.
Thomas Foster, Pharm D.
Solomon Aronson, M.D.
Michael Ashburn, M.D., M.P.H.
Vera Bril, M.D.
Robert H. Dworkin, Ph.D.
David Cornblath, M.D.
Eva Feldman, M.D., Ph.D.
Michael Polydefkis, M.D.
Michael Rowbothom, M.D.
Peter Dyck, M.D.
John Farrar, M.D.
Mark Rendell, M.D.
Steven Shafer, M.D.
David J. Wlody, M.D.
Clifford Woolf, M.D., Ph.D.
Cynthia McCormick, M.D.
Gerald Dal Pan, M.D., M.H.S.
Sharon Hertz, M.D.
Bob Rappaport, M.D.
C O N T E N T S
Opening Remarks: Nathaniel P. Katz, M.D. 4
Conflict of interest Statement:
Kimberly L. Topper, M.S. 7
Welcome: Cynthia McCormick, M.D. 14
Open Public Hearing:
Najib Babul, Pharm.D. 20
General Clinical/Regulatory Issues in Development
of Drugs Intended for Treatment of a Chronic
Sharon Hertz, M.D. 39
Specific Clinical/Regulatory Issues:
Gerald Dal Pan, M.D., M.H.S. 49
Electrophysiologic Tests Used in the Evaluation
of Peripheral Neuropathy and Neuropathic Pain:
David Cornblath, M.D. 70
Scales Used in the Evaluation of Peripheral
Eva Feldman, M.D. 97
Skin Biopsy in the Evaluation of Peripheral
Neuropathy and Neuropathic Pain:
Michael Polydefkis, M.D. 151
Charge to the Committee:
Cynthia McCormick, M.D. 182
Entry Criteria 190
Outcome Measures 235
Point-Counterpoint: Extrapolation of Findings
from One Type of Neuropathy Pain to Another
Robert Dworkin, M.D.
and Michael Rowbothom 258
Patient Populations 321
Primary Endpoints 331
1 P R O C E E D I N G S
2 Opening Remarks
3 DR. KATZ: Good morning. This is the
4 meeting of the Anesthetic and Life Support Drugs
5 Advisory Committee. We will be speaking today
6 about neuropathy, clinical trials and neuropathic
7 pain. So, if that is the meeting you are
8 interested in, you are in the right place.
9 Otherwise, they can help you find the right meeting
11 My name is Nathaniel Katz. I will be
12 chairing the meeting this morning.
13 What we will do now is I will just make a
14 few brief introductory comments and set out some
15 ground rules for everybody. We will do
16 introductions and then we will have a welcome and
17 introductions from Dr. McCormick.
18 First of all, the topic, again, that we
19 will be speaking about today is clinical-trial
20 issues in patients with peripheral neuropathy or
21 neuropathic pain. I would like first to extend my
22 welcome to our invited guests. We have managed to
23 assemble a great group of individuals here who
24 really are the true thought leaders in this area so
25 I am sure we will have a very productive discussion
2 In terms of some concrete ground rules for
3 the people around the table, there are a few things
4 that you have to know that will make the meeting
5 work. First of all, when you speak, you have to
6 speak into the microphone because everything is
7 being recorded, so don't forget that. I will be
8 sort of obnoxious. When you forget the first few
9 times, I will cut in and remind you and then would
10 should cruise after that.
11 You do have to press your "speak" button
12 on the microphone which sets up this little red
13 light. So don't forget to do that and, unless you
14 want people to hear all the little whispered
15 comments that you make during the rest of the
16 meeting, don't forget to hit the button and turn it
18 Secondly, the way that I will know who
19 wants to talk is if you could just raise your hand.
20 Then Kimberly Topper, our Executive Secretary, will
21 take your names down and we will try to get to you
22 in order. It is not a pure first-come-first-served
23 basis in that we may call on people first who maybe
24 have to leave or may not have expressed their
25 viewpoint prior to that. So don't be upset if it
1 seems like we are not calling on you in the exact
2 order that you raised your hand.
3 That being said, there are sometimes
4 visibility problems. If you find that I am
5 persistently not recognizing you, then say
6 something at some point because, last meeting, for
7 example, we had somebody over there who kept
8 raising his hand. I couldn't see him and that was
9 a problem that I had to correct about halfway
10 through the meeting. So let me know if that seems
11 to be the case.
12 In terms of the nature of our discussion
13 today, for the people, again, around the table, I
14 want to emphasize a few aspects of our goals for
15 today. What we are trying to do today is to try to
16 define some of these problems, shed light on some
17 of the issues that have been raised and bring to
18 bear some of the scientific and clinical knowledge
19 and experience that will help illuminate these
21 What we are not trying to necessarily do
22 today is come to any consensus about anything.
23 That would seem to be premature before we have
24 fully defined the problem and I wouldn't want to
25 stifle discussion by any efforts to reach a
1 premature consensus.
2 So disagreements are fine. I will
3 encourage minority points of view. We want to,
4 again, bring out all the relevant points for
5 discussion here before we seek towards achieving
6 consensus. Of course, if we achieve consensus,
7 that is fine but that is not the primary goal so
8 don't be afraid to bring out countervailing points
9 of view.
10 So, with that, I will introduce Kimberly
11 Topper, our Executive Secretary, who will read the
12 conflict of interest statement.
13 Conflict of Interest Statement
14 MS. TOPPER: The Food and Drug
15 Administration has prepared general matters waivers
16 for the following special government employees who
17 are participating in today's meeting of the
18 Anesthetic and Life Support Drugs Advisory
19 Committee Meeting being held by the Center for Drug
20 Evaluation and Research for Dr. Nathaniel Katz, Dr.
21 Vera Bril, Dr. Michael Ashburn, Dr. Solomon Aronson
22 and Dr. Robert Dworkin.
23 The waivers permit them to participate in
24 the committee's discussion of specific issues in
25 the development of pharmaceuticals for the
1 treatment of neuropathy and neuropathic pain.
2 Areas for discussion will include the duration of
3 clinical trials, evaluation of nerve function,
4 evaluation of electrophysiological endpoints,
5 appropriate clinical endpoints and appropriateness
6 of general and specific claims.
7 A copy of these waiver statements may be
8 obtained by submitting a written request to the
9 FDA's Freedom of Information Office located in Room
10 12A30 of the Parklawn Building.
11 Unlike issues before a committee in which
12 a particular product is being discussed, issues of
13 broader applicability such as today's meeting
14 involve many industrial sponsors and academic
15 institutions. The committee members have been
16 screened for their financial interests as they
17 apply to the general topic at hand. However,
18 because general topics impact so many institutions,
19 it is not prudent to recite all potential conflicts
20 as they apply to each member.
21 FDA acknowledges that there may be
22 potential conflicts of interest but, because of the
23 general nature of the discussion before the
24 committee, these potential conflicts are mitigated.
25 With respect to FDA's invited guests, we
1 would like to disclose that Drs. Peter Dyck, David
2 Cornblath, John Farrar, Thomas Foster, Michael
3 Polydefkis, Mark Rendell, Michael Rowbothom,
4 Stephen Shafer and Clifford Woolf have reported
5 financial interest in firms which may be affected
6 by the committee's discussion.
7 Dr. Dyke reported that he has received
8 honoraria and grant support from Asta Medica and
9 Eli Lilly over the past three years. Dr. Cornblath
10 reports that he has been involved in clinical
11 trials supported by Pfizer and Wyeth-Ayerst. He
12 has been a consultant to Asta Medica, Vertex
13 Pharmaceuticals, R. W. Johnson and Pfizer. He has
14 also been a member of the Schwarz Biosciences Data
15 Safety Monitor Board.
16 Dr. Farrar reports that he has been a
17 consultant to Endo Pharmaceuticals and has been
18 involved in Pfizer-supported research. Dr. Foster
19 reports that he owns stock in Johnson & Johnson and
20 Pfizer. Dr. Polydefkis reports that he has
21 received research support from Pfizer
22 Pharmaceuticals and Johnson & Johnson. He has also
23 received consulting fees from Johnson & Johnson.
24 Dr. Rendell reports that he is a principal
25 investigator on many studies and does studies on
1 many neuropathic drugs. Dr. Rowbothom reports that
2 he is a researcher on Pfizer and Johnson &
3 Johnson-supported studies and has an Endo
4 Pharmaceuticals study pending. He also receives
5 consulting fees from End Pharmaceuticals.
6 Dr. Safer reports that he does consulting
7 for Ethicon-Endo Surgical Division of Johnson &
8 Johnson. Dr. Woolf reports that he is the
9 principal investigator on Pfizer and
10 Pharmacia-sponsored studies and he receives
11 consulting fees from Pfizer, Pharmacia, Endo
12 Pharmaceuticals and Wyeth. In addition, Dr. Woolf
13 receives speaker fees from Pfizer and Pharmacia.
14 In addition, we would like to note for the
15 record that Dr. Charlie McLesky is participating in
16 this meeting as an industry representative acting
17 on behalf of regulated industry. As such, he has
18 not been screened for any conflicts of interest.
19 In the event the discussions involve any
20 other products or firms not already on the agenda
21 for which FDA participants have a financial
22 interest, the participants are aware of the need to
23 exclude themselves from such involvement and their
24 exclusion will be noted for the record.
25 With respect to all other participants,
1 we ask, in the interest of fairness, that they
2 address any current and previous involvement with
3 any firm whose products could be affected by the
4 committee's decision.
5 Thank you.
6 DR. KATZ: Thank you.
8 What I would like to do now is to go
9 around the table and do introductions just so we
10 can get to know each other and to help facilitate
11 our efforts together today. So if we could just go
12 around the table and if everybody could take 30
13 seconds and let us know who you are, where you are
14 from, what you do and what your role is with
15 respect to neuropathy and neuropathic pain.
16 Why don't we start at that end of the
17 table, please.
18 DR. McCORMICK: Hi. I'm Cynthia
19 McCormick, FDA. I am the Director of the Division
20 of Anesthetic, Critical Care and Addiction Drug
22 DR. RAPPAPORT: Good morning. I am Bob
23 Rappaport. I am the Deputy Director of the
24 Division of Anesthetic, Critical Care and Addiction
25 Drug Products at the FDA.
1 DR. HERTZ: Hi. I'm Sharon Hertz. I am
2 also with the FDA, the same division. I am a
3 medical reviewer.
4 DR. DAL PAN: I am Gerald Dal Pan. I am a
5 medical reviewer in the same division at FDA.
6 DR. McLESKY: I am Charlie McLesky. I
7 work for Abbott Labs today representing industry.
8 DR. FOSTER: Thomas Foster, Professor of
9 Pharmacy and Anesthesiology at the Colleges of
10 Pharmacy and Medicine, the University of Kentucky
11 Medical Center, Lexington, Kentucky. I am the
12 consumer representative.
13 MS. DELPH: Yvette Delph. I am patient
14 representative from the HIV community, Silver
15 Spring, Maryland.
16 DR. ASHBURN: I am Michael Ashburn. I am
17 Professor of Anesthesiology at the University of
18 Utah. I am Medical Director of Pain Programs at
19 Primary Children's Medical Center and at the
20 University of Utah.
21 DR. BITETTI: I am Janice Bitetti. I am
22 with the Department of Anesthesia and Critical Care
23 at George Washington University and I am one of the
24 committee members.
25 DR. SHAFER: Steve Shafer. Despite what
1 it says here, my primary appointment is Professor
2 of Anesthesia at Stanford University, Adjunct
3 Professor of Biopharmaceutical Science at UCSF and
4 I am here for both anesthesia and clinical
6 DR. BRIL: I am Vera Bril. I am a
7 neurologist from Toronto. I am a consultant to the
8 FDA. I am interested in clinical trials of
9 diabetic neuropathy and various other neuropathies
10 and neuromuscular disorders.
11 DR. DWORKIN: I am Bob Dworkin, Professor
12 of Anesthesiology and Neurology at the University
13 of Rochester School of Medicine.
14 DR. ROWBOTHOM: Michael Rowbothom,
15 Professor of Clinical Neurology and Anesthesia,
16 University of California, San Francisco.
17 DR. POLYDEFKIS: Michael Polydefkis. I am
18 a neurologist at Johns Hopkins and I am interested
19 in the use of skin biopsy in diabetic neuropathy
20 and in clinical trials.
21 DR. RENDELL: Dr. Rendell. Mark Rendell.
22 I am Director of the Diabetes Center at Creighton
23 University. I am interested in diabetic
25 DR. WLODY: I am David Wlody. I am an
1 Associate Professor of Anesthesiology at the State
2 University of New York, Downstate Medical Center.
3 DR. FARRAR: I am John Farrar. I am a
4 neurologist with appointments in the Department of
5 Neurology, Anesthesia and Epidemiology at the
6 University of Pennsylvania. My interest is in the
7 design and methodology of analysis for clinical
8 trials of pain, in particular neuropathic but also
9 somatic pain.
10 DR. CORNBLATH: Hi. I'm David Cornblath.
11 I am a neurologist at Johns Hopkins. I have been
12 interested in electrophysiology and nerve
13 conduction in clinical trials.
14 DR. WOOLF: I am Clifford Woolf, Professor
15 of Anesthesia Research at Harvard Medical School
16 and Massachusetts General Hospital. I am
17 interested in pain mechanisms and its application
18 to new clinical outcome measures.
19 DR. KATZ: Thank you.
20 With that, let's have introductory
21 comments from Dr. McCormick.
23 DR. McCORMICK: Thank you. Dr. Chairman,
24 committee members, invited guests, members of the
25 FDA and members of the public, welcome to today's
1 meeting of the Anesthetic and Life Support Drugs
2 Advisory Committee to discuss issues surrounding
3 the development of drugs for peripheral neuropathy
4 and to treat neuropathic pain.
5 This meeting has been convened to provide
6 an opportunity for the FDA to gain advice from its
7 distinguished advisors and experts in the area of
8 neuropathy and neuropathic pain on issues that will
9 enable the FDA to provide guidance for industry to
10 develop solid programs that will ultimately support
11 the approval of new pharmacotherapies for these
13 There are currently over forty agents in
14 various stages of development for the treatment of
15 neuropathy and neuropathic pain. Along with the
16 pharmaceutical industry, we face many challenges in
17 the development of drugs for these conditions. For
18 example, there is little history or precedent of
19 drugs demonstrated to be successful to treat
20 peripheral neuropathy.
21 The course of many neuropathies such as
22 diabetic polyneuropathy is slow and others variable
23 and this must be factored into the duration of
24 trials, particularly if the agent under evaluation
25 is anticipated to slow the course of the
2 To perform clinical trials of several
3 years duration may be a huge undertaking for
4 industry and should be embarked upon with the best
5 information on the most relevant outcomes and best
6 analysis methods in hand to deal with the
7 inevitable problems that we will see; for example,
8 high dropout rates.
9 The definition of an outcome that is
10 clinically meaningful to patients may be disputed.
11 The tools used to measure outcomes are abundant and
12 choosing the most appropriate is a challenge. The
13 role of objective measures of nerve structure and
14 function such as biopsies, electrophysiologic
15 testing and quantitative sensory testing may have a
16 role but should be placed in an appropriate context
17 relative to clinical outcome, either as a
18 supportive role or potentially as a surrogate
19 marker if appropriate validation exists. We will
20 be discussing some of these today.
21 As in any rational drug-development
22 program, attention should be given to the projected
23 target population or populations and should neither
24 be too broad nor too narrow as this will ultimately
25 be reflected back in the labeling for the product
1 once it is approved.
2 Ideally, the characteristics of that
3 population should be described in the label.
4 Attempts to acquire broad marketing claims from
5 large open-label safety studies gained in
6 populations not relevant to the identified target
7 population will likely not gain inclusion in the
9 The populations studied in Phase III
10 efficacy trials is too narrow. Labeling that is
11 overly narrow may result. While that may not
12 affect how the drug is used in real practice, it
13 will affect how it can be advertised, something of
14 importance to industry. In that context, there is
15 also the potential that the important safety
16 information is not collected in the most relevant
18 Turning to neuropathic pain, today's focus
19 will solely be on pharmacologic therapy for
20 neuropathic pain recognizing that there is a also a
21 role for non-pure-pharmacologic approaches such as
22 nerve block, dorsal-horn stimulation and so on.
23 There are only two drugs that are
24 currently approved for pain associated with
25 neuropathy, carbamazepine, initially approved in
1 1968 for epilepsy and later gained an indication
2 for trigeminal neuralgia and Lidoderm patch
3 approved in 1999 for postherpetic neuralgia.
4 Quite a large number of medications are
5 currently under development for the treatment of
6 the symptoms associated with postherpetic neuralgia
7 as well as for the treatment of pain of neuropathic
8 origin associated with many diverse etiologies.
9 For these agents, we need to understand whether
10 there is consensus on what outcomes are clinically
11 meaningful, what measures are best to describe
13 To what extent should specific
14 characteristics of neuropathic pain such as static
15 and dynamic allodynia, pain descriptors,
16 spontaneous pain and so forth be assessed.
17 One of the most challenging questions from
18 a regulatory standpoint is the whole issue of the
19 extent to which the success of a new agent in one
20 neuropathy or disorder manifested by neuropathic
21 pain can be extrapolated to a second or a third or,
22 even more generally, is the state of knowledge
23 advanced sufficiently to be able to consider a
24 general claim for neuropathic pain. If so, what
25 should be the criteria; common mechanisms of drug,
1 common underlying mechanisms of disease, PK-PD
2 modeling considerations, some other thoughtful or
3 reproducible criterion or some have proposed simply
4 an arbitrary number of replicated trials.
5 These are the things that we are
6 struggling with on a daily basis. It is our hope
7 today that we may hear the thoughts from the
8 committee on some of these areas. The questions
9 that have been formally submitted to us from
10 industry have been incorporated into the questions
11 that we have brought forth for the committee or, in
12 other cases, you will hear from the FDA speakers.
13 It is important to have adequate consideration for
15 Today, you will be hearing from the FDA
16 staff of the Division of Anesthetic, Critical Care
17 and Addiction Drug Products to give you the
18 regulatory context for today's discussion. We have
19 asked several of the guest speakers to speak on
20 selected topics that will, hopefully, stimulate
21 discussion surrounding questions about quantitative
22 measurements, of nerve function, confirmatory
23 measures in clinical trials, discussion of
24 neuropathy scales which are most appropriate for
25 clinical drug trials.
1 This afternoon, we will hear a
2 point-counterpoint discussion on the issue of
3 general versus individual claims for pain
4 associated with neuropathy, the lumping versus
5 splitting debate.
6 We hope to gain new insights from the
7 discussions of the committee today viewing it as a
8 starting point, applying what we learn from today's
9 meeting to the first steps of developing a guidance
10 for industry.
11 Thank you and welcome.
12 DR. KATZ: Thank you, Dr. McCormick.
13 What we will go to next is the open public
14 hearing. As most of you know, members of the
15 general public are invited to share their thoughts
16 and comments with us as part of these committee
17 meetings. One member of the general public has
18 requested time and that is Dr. Najib Babul. Dr.
19 Babul, you could step to the podium, please. You
20 have got ten minutes to share your thoughts with
22 Open Public Hearing
23 DR. BABUL: Good morning, Dr. McCormick,
24 Dr. Katz, FDA and members of the advisory
2 My name is Najib Babul. I am the Chief
3 Scientific Officer of TheraQuest Biosciences based
4 in Blue Bell, Pennsylvania. I am here because of a
5 keen interest in analgesic drug development
6 including neuropathic pain. I would like to
7 address the committee on the issue of analgesic
8 drug development for neuropathic pain specifically
9 some of the methodologic issues that we have been
10 struggling with.
12 At the present time, the regulatory
13 framework for development of analgesics is actually
14 fairly limited. We have the 1992 guidelines.
15 These guidelines are directed primarily at
16 single-dose evaluation of analgesics in acute pain.
17 They say virtually nothing with respect to the
18 evaluation of drugs for chronic pain or with
19 respect to the evaluation of drugs for neuropathic
21 More recently, the CPMP has issued a draft
22 guidance document on evaluation of analgesics for
23 pain. These guidelines, too, although more recent,
24 don't provide substantive support and direction to
25 drug developers and, in my opinion, to regulators
1 for chronic pain and for neuropathic pain as well.
3 We also have a number of supportive
4 guidelines, both from the CPMP and from the FDA. I
5 would argue that if we look at the osteoarthritis
6 guidance document, while directed at a more mature
7 discipline, may represent a basis for some
8 long-term approach by the agency for guidelines
9 development in neuropathic pain.
11 What is the regulatory framework for
12 approval of drugs for neuropathic pain? Put
13 another way, should a sponsor be able to obtain a
14 broad indication for neuropathic pain or is it
15 necessary to replicate evidence of efficacy for
16 each neuropathic-pain state. This is an issue that
17 a number of us have been struggling with and I know
18 that the division, likewise, has been considering
19 this issue.
21 Let's look at the pros and cons on this
22 issue. I certainly will not be able to do a kind
23 of justice that speakers later on who have a bit
24 more time will be able to do, but let me just
25 review this issue by saying that proponents of a
1 broad indication for approval for neuropathic pain
2 would argue that the response is often
3 generalizable, that pivotal studies in several pain
4 states should be adequate for a broad claim, that
5 if we require a sponsor to replicate evidence in
6 every neuropathic-pain state that this will push
7 developers to a minimalist approach to development
8 getting a very narrow indication with the attendant
9 off-label use of the drug.
10 Consequently, some would argue that many
11 painful neuropathies may remain orphaned. People
12 who support the view that we ought to look at this
13 on a subindication, if you will by a subindication
14 basis, would argue that the etiology, presentation
15 and natural course of these neuropathies is
16 different, that the mechanisms of pain are
17 frequently different, that replication is, indeed,
18 essential in order to avoid erroneous chance
19 findings, and we have seen some in the literature,
20 to be sure, and that, quite to the contrary,
21 failure to require studies in each painful
22 neuropathy may, itself, result in orphaning of
23 specific neuropathies
25 I think it will come as no surprise to Dr.
1 McCormick and Dr. Rappaport that I would make a
2 case for a broad neuropathic claims structure.
4 But, before we do that, we need to make
5 sure that we have our operational definitions in
6 order because when we are talking about neuropathic
7 pain, it conjures up different things to different
9 Are we talking about peripheral
10 neuropathies? Are we talking about phantom pain?
11 Are we talking about complex regional-pain syndrome
12 I or type II. Are we talking about nerve-root
13 disorders, central pain or spinal-cord-injury pain.
14 These are all different issues, different
15 presentations and natural histories and we need to
16 be certain that we are using the same terminology.
18 If we drill it down further, just looking
19 at peripheral neuropathic pain and, again, to
20 buttress the point that a
21 subindication-by-subindication claim would be very
22 difficult, we have a wide variety of clinical
23 presentations. We have patients with traumatic
24 mononeuropathies which could range from entrapment
25 neuropathies to transection to causalgia to stump
1 pain and post-thoracotomy pain to other
2 mononeuropathies and multiple mononeuropathies
3 including diabetic and postherpetic neuralgia and
4 trigeminal neuralgia and, of course, a series of
5 polyneuropathies of varying etiology from
6 nutritional and metabolic to drug-induced, each one
7 with a somewhat different mechanism, to hereditary
8 polyneuropathies and neuropathies secondary to
11 I hope this is not a rhetorical question,
12 but the question I would have is will we ever get
13 drugs approved for neuropathic pain or at least a
14 broad indication of neuropathic pain if there is a
15 requirement for replicate evidence in each painful
18 To compound the issue further, when we are
19 talking about neuropathic pain, we are not just
20 dealing with neuropathic pain of noncancer origin.
21 Indeed, in a series of randomized clinical trials
22 that we have been doing for the last fifteen years,
23 we have attempted to systematically stage the
24 patient's pain characteristics. This slide shows
25 data from four specific studies where anywhere from
1 2 to 12 percent of patients had solely neuropathic
2 pain or primarily neuropathic pain as their
3 reporting symptom.
4 In terms of contributory neuropathic pain,
5 anywhere from 9 to 45 percent of patients had some
6 contributory neuropathic-pain component. So it is
7 certainly a complex challenge for drug developers.
9 One of the questions that we ask ourselves
10 is whether there is a wide divergence in efficacy
11 response to various pharmacologic agents in painful
12 neuropathies. I would suggest that if the answer
13 is yes, that there is wide divergence, then a broad
14 claim may not be possible. If the answer is no,
15 then, clearly, a broad claim may be possible.
16 What is the evidence for a comparable
17 response across painful neuropathies?
19 We recently completed a retrospective
20 evaluation of the literature looking at randomized
21 double-blind placebo-controlled studies, looking
22 only at orally administered drugs that were given
23 for at least four weeks duration. We restricted
24 our evaluation only to studies in the public domain
25 involving postherpetic neuralgia and diabetic
1 neuropathy given that there is a fair bit of
2 evidence in those two neuropathies.
3 We looked at baseline and final endpoint
4 scores and attempted to calculate an overall
5 response by subtracting the placebo response which,
6 in general, was anywhere from 30 to 50 percent of
7 the overall response from the drug response.
9 What I have here is a slide with the data
10 on diabetic neuropathy. As you can see, a series
11 of agents including amitriptyline, desipramine,
12 gabapentin, pregabalin, limotrigine, mexiletine,
13 tramadol, oxycodone and dextromethorphan show a
14 fairly robust response in diabetic neuropathy.
15 There are some missing data here because
16 we were unable to obtain baseline data in some
17 cases and there was a carryover effect in a number
18 of crossover studies. In the case of limotrigine,
19 there is also data in HIV neuropathy and in central
20 pain although there is inconsistent data in
21 spinal-cord-injury pain and mixed polyneuropathy.
23 If we look at postherpetic neuralgia, we
24 find that, at least for a number of commonly used
25 drugs including amitriptyline, desipramine,
1 gabapentin, pregabalin and oxycodone, there is also
2 a similar robust pharmacologic response almost of
3 comparable effect size within the variability we
4 expect from study to study.
5 These data would suggest, at least to me,
6 that it should be possible, within a preponderance
7 of evidence, to generalize and obtain a broad
8 neuropathic-pain claim.
10 One of the other issues that we have been
11 struggling with is what it is that we need to
12 measure in neuropathic-pain studies.
14 In a study that Peter Watson and I did in
15 and published in Neurology in 1998, we
16 systematically looked at this issue. Mitchell Max
17 and others have done this as well.
18 Almost all patients, 97 percent of the patients,
19 had ongoing or steady pain and about 90 percent of
20 patients had brief pain and evoked pain described
21 by a variety of different descriptors.
23 If you look at the specific pain
24 characteristics, certainly in terms of peripheral
25 neuropathies, steady pain, paroxysmal pain and
1 allodynia are fairly common features. These
2 patients often have some sensory impairment as
3 well. Certainly these are some of the things we
4 ought to look at in all randomized clinical trials
5 in neuropathic pain.
7 These are data from a randomized
8 placebo-controlled clinical trial we did with
9 oxycodone, in this case, OxiContin, looking at
10 these three dimensions of pain, steady pain,
11 paroxysmal pain and allodynia. On all three
12 dimensions, we found a fairly robust pharmacologic
13 response for oxycodone.
14 These data are not unique to oxycodone or
15 to opioids. The have been shown with meprotalin,
16 amitriptyline, desipramine and a number of other
17 pharmacologic agents.
19 The other issue is what else should we be
20 measuring. Clearly, as Dr. McCormick suggested,
21 the durability of the response needs to be
22 measured. My presentation here largely deals with
23 symptom relief. I am not here to speak to the
24 issue of disease progression and the subset of
25 agents that are being looked at in terms of
1 disease-modifying agents, but the durability of
2 efficacy response is an important issue given that
3 these patients are going to be on treatment for a
4 long period of time.
5 Quality of life and function are also
6 important issues. The role of quantitative sensory
7 testing certainly is something that is the subject
8 of some debate. One of the issues that I would put
9 to the division and to the advisory board is if you
10 find a significant difference or a positive finding
11 on electrophysiologic testing and find no actual
12 subjective benefit, what does that mean?
13 If, on the other hand, you find a negative
14 finding on objective electrophysiologic testing and
15 find a positive finding on the subjective findings,
16 what does that mean? In other words, I am not
17 entirely certain that, other than in an exploratory
18 or mechanistic sense, that this adds much to the
19 labeling, itself.
20 Finally, if we are looking at centrally
21 acting drugs, as we often are, we need to consider
22 neuropsychological and cognitive effects of these
25 This is my last slide. I would like to
1 just briefly suggest to you, at the cost of being
2 somewhat prescriptive because I think this is where
3 the rubber meets the road, as to what a core
4 development program could look like for a 505(b)(1)
5 drug for a broad neuropathic-pain indication.
6 I would suggest that one of the things
7 that is lacking uniformly with a range of
8 pharmacologic agents across therapeutic agents and
9 divisions is proper dose-finding studies. So I
10 think it is important that dose-finding and
11 dose-frequency-finding studies be conducted in at
12 least two painful neuropathies. However, these
13 studies probably can be incorporated into pivotal
14 clinical trials.
15 In addition, I would suggest that
16 replicate evidence of twelve-week efficacy, which
17 is a standard that I think most of us, including
18 the division, have accepted in chronic pain of
19 noncancer origin, replicate evidence of twelve-week
20 efficacy in postherpetic neuralgia combined with
21 replicate evidence of twelve-week efficacy in
22 diabetic neuropathy ought to be a sufficient basis
23 for a broad neuropathic claim.
24 I think, however, if the division should
25 take such an approach, sponsors should be given
1 some latitude in terms of drug development.
2 Perhaps robust response in twelve-week efficacy
3 studies in two separate painful peripheral
4 neuropathies plus one or two other models such as
5 central pain, spinal-cord pain, complex
6 regional-pain syndrome, nerve-root pain, et cetera,
7 might be adequate as a basis for a broad
9 I think cognitive impairment, both acutely
10 and chronically, need to be evaluated. Obviously,
11 there is a need for long-term safety data.
12 Finally, the clinical pharmacologic section of the
13 label should reflect the efficacy data, the precise
14 studies in which the drug has been found to be
15 effective, ineffective, the magnitude of the
16 pharmacologic response and, indeed, the specific
17 pain dimensions that have shown a positive
19 Thank you.
20 DR. KATZ: Thank you, Dr. Babul. Stay
21 there for one second.
22 Does anybody around the table have any
23 questions for Dr. Babul based on the information he
24 has just presented?
25 DR. KATZ: Dr. Farrar?
1 DR. FARRAR: I was interested in knowing,
2 with the effect-size slide that you showed, you had
3 subtracted out the placebo rates. I am not quite
4 sure how you calculated an effect size. Was it the
5 remaining effect size?
6 DR. BABUL: That's correct. What we did
7 is we took the baseline value, subtracted the final
8 endpoint value from that to come up with the effect
9 of the test drug, did the same thing for the
10 reference drug and then subtracted one from the
12 In general, what we found is the placebo
13 response was about the same as what we see in
14 osteoarthritis, for instance.
15 DR. FARRAR: If I could follow up. The
16 effect size was presented as a percent. I am
17 wondering, a percent of what?
18 DR. BABUL: That was a percent of the
19 baseline value in terms of percent reduction of
20 baseline value, probably more appropriately labeled
21 as response rather than effect size.
22 DR. KATZ: Dr. Woolf?
23 DR. WOOLF: You used that same slide to
24 argue the case that different drugs had similar
25 degrees of efficacy. But your desipramine had
1 about a 10 percent effect in diabetic neuropathy
2 and over 30 percent in postherpetic neuralgia.
3 That, obviously, could be by chance but it does
4 raise the issue that there may be differences in
5 efficacy between different conditions.
6 DR. BABUL: You are quite correct. Let me
7 make a couple of points in that respect. The first
8 is that I think most of us have accepted, although
9 not all, that a minimum clinically perceptible
10 difference is about 10 percent and some have argued
11 perhaps 15 percent.
12 So, in that sense, I think that most
13 clinicians agree that desipramine provides a
14 reasonable response in postherpetic neuralgia and
15 in diabetic neuropathy. I think part of the
16 challenge here is that a number of studies did not
17 lend themselves to calculating a pharmacologic
18 response because of the absence of baseline values.
19 Without a doubt, there are some
20 differences which, perhaps, would argue for
21 replication. My point is that replication may be
22 reasonable. Certainly, there is a sound foundation
23 for replication at the agency although arguments
24 have been made for large single studies as well.
25 But replication in all neuropathies may be
2 The other point I would make is that
3 mechanistically, within a given neuropathy, there
4 are substantial differences. So, if we start
5 looking at diabetic neuropathy, there are
6 mechanistic differences in terms of presentation of
7 patients within a given neuropathy so where,
8 exactly, does this process end?
9 There are also other differences. I
10 talked about lamotrigine in terms of some
11 variability where in certain states, like HIV
12 neuropathy, the findings are positive. In central
13 pain, they are positive. There are no data on
14 postherpetic neuralgia, unfortunately, that I am
15 aware of but we know that in a recent study
16 published in Pain, in spinal-cord injury pain, the
17 results were negative and in mixed neuropathy the
18 results were negative.
19 So it always hard to know whether it is
20 the design, a function of dose, whether it is a
21 question of polypharmacy, appropriateness or
22 washout, the instruments that are being used and I
23 think there is probably a need for standardization.
24 DR. DAL PAN: Any other questions? Dr.
1 DR. SHAFER: Our pain group at Stanford
2 feels fairly strongly that VAS scores for chronic
3 pain can be very hard to interpret and primarily
4 push for quality-of-life indicators. But, in your
5 presentation here, talking about postherpetic
6 neuralgia, at least what I am inferring from your
7 presentation is you see VAS as being more the
8 primary endpoint and things like quality of life
9 being potentially secondary endpoints on the
11 Is that a correct interpretation of your
12 experience and where you are directing this?
13 DR. BABUL: In the literature, a majority
14 of investigators have used either a visual-analogue
15 scale or a categorical scale for evaluating pain as
16 a cardinal feature. Most studies have not looked
17 at various dimensions of pain. To be sure, people
18 have--Mike Rowbothom and others have employed the
19 McGill Pain Questionnaire with the various
20 descriptors that that provides, but most people
21 have not specifically targeted at each visit
22 specific dimensions of pain.
23 But a majority of people have used the
24 visual-analogue scale. There is this separate
25 issue about what constitutes a win. This is an
1 ongoing struggle. Drug developers concerned about
2 coprimaries--in other words, a requirement that a
3 win be based not just on pain but on quality of
4 life. Some would argue function or return to work
5 which is a rather daunting task.
6 I think many of us who are involved with
7 pain management feel that pain relief alone is a
8 reasonable endpoint. Certainly, we hope that that
9 translates into quality of life. There is not a
10 huge amount of work done in terms of
11 quality-of-life instruments in neuropathic pain
12 although there is some literature out there.
13 DR. SHAFER: Just to quickly follow up,
14 part of the distinction was acute- versus
15 chronic-pain syndromes. Do you see any bifurcation
16 between the measures for acute and the measures for
18 DR. BABUL: In both acute pain and in
19 chronic pain, in chronic pain as it relates to,
20 say, osteoarthritis, myofascial pain, cancer pain,
21 any neuropathic pain, both categorical and
22 visual-analogue scales have shown validity and
23 actually fairly good reliability. Unfortunately,
24 VAS seems to be something that most investigators
25 and academics seem to prefer and I think most
1 patients probably prefer some sort of a numerical
2 or categorical scale and there is this challenge.
3 But both in acute and chronic pain, we have used
4 VAS successfully.
5 DR. KATZ: Thank you.
6 Dr. Farrar?
7 DR. FARRAR: Just two quick comments. One
8 is the minimal perceptible difference is clearly a
9 different measure than a clinically important
10 difference and the second is that, to try and
11 conclude something from the graphs that you have
12 here, it is very important to remember that these
13 measures are looking at the mean value and that the
14 mean value is not a unique answer to the question
15 of how many people actually got better.
16 You can come up with any of a number of
17 different interpretations and I would be interested
18 if any of these studies actually published
19 something about the number of patients who actually
20 got better to try and look at some of that data as
22 DR. BABUL: Dr. Farrar, I would certainly
23 approach this issue with some trepidation in your
24 presence, but let me suggest that, from a
25 number-needed-to-treat basis, there are generally
1 consistent findings as well for most of these
2 pharmacologic agents with some discrepancy that you
3 would expect across clinical trials.
4 DR. KATZ: Thank you, Dr. Babul. We
5 appreciate your comments.
6 We do have a little bit of time left in
7 the Open Public Forum so if there is anybody in the
8 room who would care to come up and share some
9 thoughts with us about these issues, you are
10 welcome to do so at this time. Just approach the
11 center mike right up front.
12 I feel like I have a clean conscience that
13 everyone has been offered an opportunity. We will
14 go on with the rest of the program, then.
15 Next, we will have a number of
16 presentations from the FDA folks on some of the
17 regulatory issues in this area beginning with Dr.
18 Sharon Hertz.
19 FDA Presentations
20 General Clinical/Regulatory Issues in
21 Development of Drugs
22 Intended for Treatment of a Chronic Illness
23 DR. HERTZ: Good morning.
25 I am going to discuss the general
1 regulatory issues that are involved in drug
2 development in general so that we can think of them
3 as we discuss neuropathies specifically. The
4 general regulatory framework in which we work here
5 at the agency compels us to keep the entire
6 drug-development process in mind when we review all
7 submissions. This extends from the time of the
8 initial application to study the drug in humans,
9 the IND submission, to the time when the product
10 will be considered for marketing at the submission
11 of the New Drug Application, or NDA.
12 Clinical drug development plans and NDAs
13 are reviewed for efficacy in the context of the
14 drug safety profile. At the same time, the choice
15 of clinical-trial design and study populations are
16 considered for the future promotional and marketing
18 The clinical trials used to support an NDA
19 are the basis for the drug's indication and will be
20 reflected in the language of the product label.
21 Marketing and promotional claims are based on the
22 information in that label. This last point is
23 important and I will refer to it later at the end
24 of my talk.
1 Basically, a company has a hypothesis that
2 Drug A is capable of treating a symptom or a
3 disease in a safe and effective manner. The proof
4 is at least two adequate and well-controlled trials
5 demonstrating this hypothesis to be true with
6 additional safety information as needed. The
7 results, hopefully, are approval of the product and
8 a label. Then the product will be promoted based
9 on the findings of efficacy.
11 So what is the regulatory basis for
12 studies in support of efficacy? What is the
13 regulatory basis for the requirements of the safety
14 database? And how are these findings, the product
15 label and promotion related?
17 The legal standard requiring the
18 demonstration of effectiveness was added to the
19 Food, Drug and Cosmetic Act in 1962. It states
20 that no person shall introduce, deliver for
21 introduction, into interstate commerce any new drug
22 which basically hasn't been shown to be effective.
24 The regulations also state that full
25 reports of these investigations which support the
1 demonstration of efficacy must be submitted to the
2 application and that a finding of substantial
3 evidence that the drug will have the purported
4 effect in the intended conditions of use must also
5 be provided to support approval for the
8 The regulations also describe the term
9 substantial evidence that is necessary in support
10 of a finding of efficacy. Substantial evidence is
11 defined as evidence consisting of adequate and
12 well-controlled studies by experts qualified to
13 perform those studies so that the studies can be
14 the basis to conclude the drug will have the effect
16 The term "adequate and well-controlled
17 investigations" was taken by the agency to mean at
18 least two adequate and well-controlled trials.
20 The Code of Federal Regulations describes
21 the essential characteristics of an adequate and
22 well-controlled trial. This includes the required
23 documentation of planning, conduct, data handling
24 and record keeping. The purpose of conducting
25 these clinical investigations is to distinguish the
1 effect of the drug from other influences such as
2 spontaneous change within the course of the
3 disease, placebo effect or biased observation.
4 Additional, the Regulations describe the
5 types of study designs that permit what is
6 considered a valid comparison using a control to
7 provide quantitative assessment of drug effect.
8 This section also describes the use of concurrent
9 placebo control or dose-comparison controls or the
10 use of objective measures when available and a
11 placebo effect is expected to be negligible.
12 Concurrent acting controls are described
13 along with the potential pit fall for a lack of
14 assay sensitivity if not used with other types of
17 There is some flexibility with respect to
18 the number of trials required for approval based on
19 the situation and the availability of other
20 supportive data according to the FDA Modernization
22 The legal and scientific bases for the
23 quality and quantity of evidence necessary to
24 support effectiveness are summarized in a guidance.
25 I just want to say that the requirement for more
1 than one adequate and well-controlled study doesn't
2 reflect so much the need to replicate findings in
3 the same type of study but more the need to provide
4 independent substantiation of experimental results.
5 The intent is to avoid unanticipated bias
6 or chance results and to demonstrate the findings
7 are generalizable to patients under different
10 The finding of safety is more accurately
11 the finding of acceptable risk in the context of
12 the efficacy of the drug. The requirements for the
13 safety database for drugs intended for chronic
14 administration are also described in a guidance.
16 The finding of effectiveness is then
17 reflected in the product label in pertinent
18 sections, particularly indications and usage
19 material must be supported by substantial evidence
20 of effectiveness. Comparative statements about
21 other products must also be supported by
22 substantial evidence.
24 Findings referable to safety are reflected
25 in several sections of the label according to the
1 regulations and postmarketing information can be
2 added as needed.
4 Once the wording in the label is agreed
5 upon and approved, the sponsor may advertise and
6 promote the product in accordance with the
7 regulations. The advertisements must be accurate
8 and balanced and limited to the indications
9 included in the label. This is a point that has
10 been mentioned already and it is an important point
11 for the following reasons.
12 First of all, a product that is effective
13 for more than one indication may be effective under
14 different conditions of use, different dosing
15 regimens, so it is important that findings of
16 efficacy be supported by data for that indication.
18 It is also particularly important because
19 a product that is used in different populations may
20 have different safety profiles based on the
21 characteristics of those populations so age,
22 comorbidity, concomitant medications with potential
23 for drug-drug interactions are all important
24 features that need to be explored in an adequate
25 safety database.
1 The one other feature why this is
2 important is because it is necessary to set a level
3 playing field where all companies are held to a
4 comparable standard. So, for a company to promote
5 their product for a specific indication, it is
6 incumbent on them to demonstrate the effectiveness
7 and safety for that indication.
8 That is not to say that a product cannot
9 be used in a manner according to clinical judgment
10 by any given physician, but the approval and
11 promotion of drugs are regulated processes and the
12 FDA is responsible for implementing those
15 So as we discuss the approach to drug
16 development for products to treat neuropathic pain
17 and underlying neuropathies, please keep in mind
18 how these different pieces, the clinical trials,
19 the safety data, the product label and product
20 promotion fit together.
21 Thank you.
22 DR. KATZ: Thank you, Dr. Hertz.
23 Any questions from around the table for
24 Dr. Hertz? Dr. Farrar?
25 DR. FARRAR: The one area that the
1 guidelines don't really speak to is with regards to
2 the size of the beneficial effect. I wonder if you
3 could just comment on that.
4 DR. HERTZ: I hope we cover that somewhat
5 today in the discussions. We struggle with
6 statistically significant differences in effect
7 size between the placebo group and the active
8 treatment groups versus the concept of a clinically
9 meaningful difference. That is going to be on the
10 roster for discussion today, so we don't have an
11 answer yet specifically in this area.
12 DR. KATZ: Other questions for Dr. Hertz?
13 I have a question. It sounded like, and correct me
14 if I am wrong, you were making the point that, in
15 meeting this criterion of two adequate and
16 well-controlled trials for a specific indication
17 that the agency is more impressed by a pair of
18 trials where one actually differs from the other in
19 terms of details of study design, location where
20 the trial was conducted, et cetera, et cetera, as
21 opposed to what we sometimes see which is two
22 replicate trials that truly are replicated, where
23 the trial is exactly identical and you could
24 combine them or split them and it is the same
1 Am I hearing you correct? Is that how
2 that issue is perceived?
3 DR. HERTZ: Yes, short answer, for the
4 reason that you want to have a little bit more
5 generalizability. Otherwise, it is basically one
6 big trial separated by some other divider.
7 DR. KATZ: Thank you.
8 Dr. Woolf, please?
9 DR. WOOLF: In terms of indications, it
10 wasn't clear whether you were talking about, in the
11 context of this meeting, symptom, let's say acute
12 versus chronic pain, or neuropathic pain or
13 postherpetic neuralgia.
14 Is there a difference between indication
15 as a symptom or as a disease syndrome?
16 DR. HERTZ: The indication is basically
17 what the claim for efficacy is based on. So, if
18 you are going to say that a product is capable of
19 relieving the pain of diabetic neuropathy, then
20 that is your indication, symptom relief. It could
21 also be that your product is intended to slow the
22 progression or reverse the changes associated with
23 diabetic neuropathy and then that would be the
25 So it is really defined by what you see
1 the product, what the company sees the product,
2 capable of doing and capable of proving efficacious
4 DR. KATZ: Other questions for Dr. Hertz?
5 Thank you very much. Next we will have
6 Dr. Dal Pan from the FDA who will be speaking
7 further about specific clinical and regulatory
8 issues that arise.
9 Specific Clinical/Regulatory Issues
10 DR. DAL PAN: Good morning.
12 We have just heard from Dr. Hertz about
13 the clinical requirements for the development and
14 regulatory approval of drugs to treat chronic
15 disease. The basis of this is embodied in the
16 substantial evidence requirement which states that
17 the drug will have the effect it purports or is
18 represented to have under the conditions of use
19 prescribed, recommended or suggested in the
20 proposed labeling thereof. In other words, the
21 drug has to do what the label says it does.
22 What does this mean, then, for drugs for
23 peripheral neuropathy and for chronic neuropathic
24 pain. The basic challenge for the agency, for the
25 industry and for researchers is to operationalize
1 the substantial-evidence requirement into
2 clinical-trial design and clinical-development
3 planning for drugs to treat peripheral neuropathy
4 and chronic neuropathic pain.
5 So I would like to take a little bit of
6 time today and just present to you some of the
7 specific examples in clinical-trial design and
8 clinical-development planning that confront the
9 industry and confront us when we meet with industry
10 to go over trial design and development planning.
11 The examples are not so much today to get
12 specific answers to specific questions or specific
13 plans but rather to present to you the scope of the
14 important issues that are facing us and to be
15 followed later today by a discussion of what the
16 scientific and clinical issues are and how we can
17 best be informed about these issues so we can carry
18 that into sound decision-making in the future.
20 So let's start with the example of Company
21 A. The company wants to develop a drug to slow or
22 reverse the progression of diabetic polyneuropathy.
23 So several issues come up here with regard to
24 clinical-trial design.
25 One of the first issues is what is the
1 appropriate outcome measure or measures. Some of
2 the challenges here are there is no regulatory
3 precedent. No drugs have been approved for this
4 indication and there aren't many large-scale trials
5 to guide us or to inform us as to what the best
6 outcome measures are.
7 Because diabetic polyneuropathy is a
8 complex disease, the issue of a composite outcome
9 versus a single-measure outcome comes up. There
10 are many composite-measure outcomes in the
11 literature and we have seen a lot of proposals to
12 use such composite outcome measures.
13 An example of such a measure would be the
14 Neuropathy Impairment Score, or NIS, of the lower
15 limbs known as NIS(LL)+7. This is a composite
16 clinical measure that looks at weakness, sensory
17 loss, reflexes and electrophysiologic studies of
18 motor and sensory nerves, heart rate variability
19 and vibratory-detection threshold.
20 One of the challenges is defining the
21 degree to which this composite measure or any
22 composite measure, or any single measure, for that
23 matter, really reflects what the clinically
24 important effect of a drug to treat diabetic
25 neuropathy really is. Closely related to what the
1 outcome measure is is something we have heard in
2 some of the discussion already this morning; what
3 is the magnitude of the effect size.
4 We are translating clinical issues into
5 quantitative measures, be they measures of
6 percentage of patients who respond by a given
7 criteria or mean values on some numeric outcome.
8 What is the scientific and clinical basis for
9 determining how big an effect size should be? That
10 is important because that, then, becomes the
11 measure of the effectiveness of the drug and, from
12 a practical point of view, it is important in trial
13 design because it forms part of the basis for
14 sample-size determination.
15 When we also look at this class of drugs,
16 we want to distinguish between slowing progression
17 versus arresting progression of disease versus
18 actually reversing disease. This may have
19 implications for what the outcome measure is. It
20 may also have implications for the duration of the
21 trial as well as the sample size.
22 We want to also consider what is the role
23 of other testing such as electrophysiologic
24 testing. Measures of nerve-conduction studies have
25 been well documented in diabetic polyneuropathy as
1 measures of extent and severity of disease as well
2 as change over time. To what degree can these
3 measures serve as markers or surrogate markers of
4 the important clinical effects we want the drug to
5 be able to have.
6 If a drug is going to reverse or slow the
7 progressive neuropathy, it may also have a
8 beneficial effect on symptoms during the course of
9 the disease and how can we capture this in the
10 trial as well. So these are some of the challenges
11 involved in drugs for slowing the progression of
12 diabetic polyneuropathy.
14 Let's turn now to a different scenario.
15 Company B wants to develop a drug to treat chronic
16 neuropathic pain due to diabetes. Several of the
17 previous issues are important here as well. Again,
18 we come back to the appropriate outcome measure or
20 What is the role of pain intensity
21 reduction? What is the role of pain relief. What
22 is the role of function as an outcome. What is the
23 role of quality of life as an outcome? Because
24 neuropathic pain can vary from person to person,
25 what is the role of characterizing different
1 symptoms such as allodynia, lancinating pain,
2 burning pain and, again, for both composite
3 measures and single effect measures, what is the
4 magnitude of an effect that is clinically important
5 and what is the basis for determining what that
6 effect size is?
7 Because chronic diabetic neuropathic pain
8 is a complication of a systemic disease, we want to
9 also consider how to account for the role of
10 potential confounders; for example, the severity of
11 nerve dysfunction and the level of diabetic control
12 during the trial, especially since those may
13 actually impact the outcome of the trial. Finally,
14 because it is chronic disease, we want to be able
15 to assess the durability of the effect.
17 My last example is a sponsor that wants to
18 have a drug to treat both chronic painful diabetic
19 neuropathy and postherpetic neuralgia. The central
20 issue here is the degree to which data from one
21 etiology of neuropathic pain can support data from
22 another etiology of neuropathic pain and, more
23 broadly, can results from these studies be
24 generalizable to types of neuropathic pain not
1 So I have tried to give you an overview
2 here of some of the important issues that are
3 facing us today. We have more talks on the agenda
4 to address some of these issues in particular, and
5 we have put forth a variety of questions to spark
6 some discussion.
7 Thank you.
8 DR. KATZ: Thank you.
9 First, we have a new arrival at the table.
10 Everyone else had to introduce themselves, so, in
11 the interest of equal treatment, please introduce
13 DR. FELDMAN: My name is Eva Feldman. I
14 am a Professor of Neurology at the University of
15 Michigan and I also direct a juvenile diabetes
16 research foundation center where we study
17 complications of diabetes.
18 DR. KATZ: Thank you. For logistical
19 reasons, what we will do now is have Dr. Cornblath
20 speak on electrophysiologic tests used in the
21 evaluation of peripheral neuropathy and neuropathy
23 Oh; I'm sorry. My mistake. Any questions
24 for Dr. Dal Pan before he steps down? Dr. Shafer?
25 DR. SHAFER: Just quickly one thought, or
1 question, rather. There are a number of issues
2 that you allude to including things like
3 sensitivity to covariate effects. These kinds of
4 trials have other complications. Commonly, the
5 data are right sensors. People drop out of the
6 trials. Trying to separate out the inter- and
7 intra-individual variability which you were
8 referring to would try to distinguish effect size
9 from the number of people who actually have any
10 effect at all.
11 To what extent do you expect to see
12 population approaches brought into the analysis of
13 data in pain trials?
14 DR. DAL PAN: Population approaches; you
15 mean by percent responders?
16 DR. SHAFER: Population approach is really
17 where you have a model of intra- and
18 inter-individual variability and are modeling those
19 effects simultaneous with an overall model of
20 effect including, actually, survival in the trial
21 which allows you to account for right censoring of
22 your data.
23 DR. DAL PAN: The issue of censoring has
24 come up in a lot of pain trials. I would actually
25 like the committee maybe just to address that later
1 this afternoon. I think that one of the issues
2 that concerns us is differential dropout rates.
3 People in placebo groups drop out because they are
4 not getting pain relief and people in active
5 treatment groups drop out because they are getting
6 toxicity from the drug or can't tolerate it, even
7 if they had, say, pain relief in a pain trial.
8 So I think that might be something
9 interesting for the committee to address, how to
10 handle that. It is something we have dealt with.
11 DR. SHAFER: For acute pain, there has
12 been a lot of good work with population modeling.
13 I haven't seen much in chronic pain.
14 DR. DAL PAN: I am not very familiar with
15 that, either.
16 DR. KATZ: Other questions for Dr. Dal
17 Pan? Dr. Bril?
18 DR. BRIL: Hi. One of the basic issues
19 that I find confusing is in trials in diabetic
20 neuropathy when we are trying to prevent
21 progression. They are very difficult. And we know
22 that the rate of progression really varies very
23 much with glycemic control. And we know that we
24 can improve control in a lot of people but we know
25 we don't improve it in many people.
1 We know a lot of people are out there with
2 poor control and those are the people who have more
3 complications. Yet, in some of our long-term
4 studies now we are designing, we are selecting for
5 people whose control is as good as we can make it
6 but we kind of exclude the population who may be at
7 highest risk for the complication.
8 I am just wondering what the agency thinks
9 about broadening the study population to include
10 people who might benefit most from the
11 interventions you may want to be using. It is a
12 real problem, I think, and has implications for the
13 generalizability of use if a drug ever was found
14 effectiveness for diabetic neuropathy.
15 You would be saying it is in those who
16 have fairly good control. This is something that
17 really exercises my mind. I wonder what the agency
19 DR. DAL PAN: I think it is a good point.
20 I think that it is important that the drug be
21 studied in the patients who could benefit from it.
22 At the same time, I think your point is also right
23 that control of diabetes during the trial can
24 confound the outcome. So that is why we have
25 wanted some criterion in the beginning as to who
1 can enter.
2 It is not necessary to include only people
3 with the best diabetic control. I think that is
4 actually one of the questions we have for the
5 committee later is about the entrance criteria for
6 diabetics. So I think maybe we can have some
7 discussion on that later by the committee.
8 DR. KATZ: Dr. Farrar?
9 DR. FARRAR: I think, actually, the
10 question was targeted more at the issue of efficacy
11 versus effectiveness. I think the question was
12 that if you use a very selective population and are
13 able to show an effect size of some magnitude, the
14 question then becomes what about people who are
15 likely, or perhaps even more likely, to benefit
16 from them but because of other issues may have a
17 different set of problems.
18 I think you are referring to a population
19 that is not generally studied which are the people
20 who have highly variable glucose control.
21 DR. BRIL: I am referring to the
22 population where a lot of studies now have
23 upper-limit cutoffs for glycosylated hemoglobins.
24 Yet, there are still people who are out there with
25 these levels in spite of all efforts to improve
1 their control and then the argument is said, well,
2 these are noncompliant people anyway.
3 But, actually, they are not. They would
4 be happy to be in a study. I don't think they
5 should be dismissed. So the question is how do we
6 incorporate them into long-term trials and not
7 exclude them?
8 DR. DAL PAN: I think that is something
9 that we would like the committee to discuss this
10 afternoon, actually.
11 DR. KATZ: Dr. Foster next.
12 DR. FOSTER: A question along the same
13 vein. In the introduction this morning, we learned
14 that there are multiple agents in development now.
15 I think if you parse them into disease-modifying
16 agents versus palliative agents, the question comes
17 in Dr. Hertz' presentation at the end, in
18 advertising, as we fast forward to the end, does
19 the agency consider plans for polypharmacy in this
20 area where drugs would be, say, in a diabetic who
21 is developing neuropathy where initially palliative
22 agents would be placed, then prescribed with
23 disease-modifying agents. Is there a plan to
24 incorporate this type of multiple drug use into the
25 design of clinical trials?
1 DR. DAL PAN: I am not aware of any plan
2 for that right now. The disease is to slow or to
3 reverse the progression of diabetic neuropathy.
4 Studies are generally entering patients with
5 earlier-stage disease so who haven't developed a
6 lot of the severe complications such as this
7 chronic neuropathic pain.
8 So, usually patients with severe chronic
9 neuropathic pain are not entered into those
10 studies. They are entered more into studies for
12 DR. KATZ: Dr. McCormick, did you care to
13 amplify on that?
14 DR. McCORMICK: Sure. I think, in so far
15 as these many drugs that are under development are
16 all being developed by different sponsors, each may
17 have its own intent. I think that there certainly
18 is a precedent for having approval for adjunction
19 therapy. That is something that would have to be
20 studied but could potentially make it into a
21 product label if it had been studied.
22 DR. KATZ: Dr. Woolf.
23 DR. WOOLF: You mentioned the complexity
24 inherent in studying the progression of a chronic
25 disease that may be changing. Some of those
1 changes may be associated with the mechanisms that
2 may be responsible for the pain so that, early in
3 the disease, the pain may be responsive to a
4 particular pharmacological mechanism and later it
5 may not be.
6 That needs the mechanisms to separate out
7 the response of different patients according to
8 where they are along the natural history of that
10 DR. DAL PAN: I think you are right. I
11 think we are going to have some discussion later
12 today about mechanism-based selection of agents.
13 DR. KATZ: Are you suggesting, Dr. Woolf,
14 that it may be important in clinical trials of
15 neuropathic pain to categorize patients up front
16 based on duration of disease among other things in
17 order to, later on, look at subgroups of patients
18 who may be more or less responsive based on their
19 position in the natural history?
20 DR. WOOLF: We all recognize that some
21 patients respond and others don't to the treatment.
22 I think, certainly, one of the explanations would
23 be that the symptoms that are being generated are
24 reflecting different mechanisms which occur at
25 different times in the disease course.
1 So, rather than always doing that post
2 hoc, I think one of the ways is to try and define
3 that up front.
4 DR. KATZ: Dr. Dworkin, then Dr. Rendell.
5 DR. DWORKIN: I was wondering, with
6 respect to this issue of a broad indication versus
7 specific indications, are there any precedents
8 where the FDA has approved a drug in other areas of
9 medicine for a broad indication based on controlled
10 trials in several more specific diseases?
11 DR. DAL PAN: I frankly have to admit
12 ignorance to answer that question. I can't answer
13 you yes or no because I just don't have an example.
14 Maybe one of my colleagues does.
15 DR. KATZ: Anybody else from FDA?
16 DR. McCORMICK: Actually, I think in the
17 area of pain, there has been that precedent. We
18 are currently examining that issue but that has
19 been the precedent since about 1992.
20 DR. KATZ: Anyone else from FDA have any
21 comments about the areas of medicine perhaps
22 outside of pain where there is a precedent for
23 providing broad labels after studies are done in
24 specific subcategories? I wonder if acute pain,
25 itself, might be an example of that where trials
1 are typically done, and correct me if I am wrong,
2 in usually dental pain and some post-surgical model
3 with a pair of controlled trials in each one and
4 then the label is given for acute pain broadly
5 despite the fact that there may be different types
6 of acute pain that were not addressed in the
8 DR. McCORMICK: Right. That is what I was
9 referring to.
10 DR. KATZ: Dr. Rendell?
11 DR. RENDELL: One of the greatest concerns
12 that I have, having looked at most of the
13 diabetic-neuropathy agents and having seen them
14 fail on statistical grounds time and time again is
15 that we are dealing with diabetic neuropathy as if
16 it were a single disease as opposed to a condition
17 with multiple different etiologies, the
18 recognition that there may be certain subgroups of
19 patients who may respond to a given agent and that
20 subgroup of patients is not enough to sway the
21 overall statistic in the favor of significance.
22 I have no answers, but I would like to
23 throw out the consideration that we need to start
24 making an effort to identify responders, subgroup
25 responders, and try to decide what it is about them
1 that makes them respond to a given drug so that we
2 might be able to offer these subgroups meaningful
3 treatment although the overall response of a given
4 drug, as David and I both know, having done this
5 for years, is going to be negative when we look at
6 the overall statistic.
7 So I throw that out as a challenge and
8 certainly I have no ideas on how to do that.
9 DR. KATZ: It sounds like there are at
10 least two implications from your comments. One is
11 that the trials may be false negative in the sense
12 that, while overall negative, they may fail to
13 identify, indeed, an important effect in a subgroup
14 that otherwise there is no specific technology for
16 Secondly, the splitting issue may become
17 even more complicated than that. Even a medication
18 that works for painful diabetic neuropathy in
19 general may, in fact, indeed only works for a
20 subgroup of yet those patients which makes the
21 splitting debate even more complicated.
22 Dr. Farrar is first and then Dr. Shafer.
23 DR. FARRAR: Just a couple of comments.
24 There are, actually, some design methods of getting
25 at what you are talking about, one of which is
1 using an enriched population and there is,
2 obviously, great concern about how one does that.
3 But, for instance, if you are interested in
4 studying if a tricyclic is effective in a
5 particular group, you could take patients who were
6 responsive already to a previous tricyclic, take
7 them off an put them back on.
8 There are a lot of design problems with
9 that and we don't need to get into it. The second
10 thing is that there are some statistical issues one
11 can look at to enhance the ability to find small
12 populations that, in fact, respond. We can talk
13 about those at some point later, too.
14 DR. DYKE: Dr. Shafer?
15 DR. SHAFER: Again, just in follow up, I
16 am wondering if there is a role in the study
17 design, potentially in the labeling, too, for
18 exactly that kind of enrichment that Dr. Farrar is
19 referring to. We often do things--like if we are
20 interested in trying a sodium channel blocker,
21 mexiletine, we will bring patients in and
22 essentially give them a total-body beer block. We
23 give them lidocaine and examine their acute
24 response to it to see if they have an analgesic
25 response and then, if they do, consider them a
1 reasonable candidate for sodium-channel blockade.
2 Or they could just be responsive to
3 opioids, an acute trial in the clinic of I.V.
4 opioids to see if they are going to respond before
5 trying them long-term of opioid maintenance. Is
6 there a role in the process and, potentially, in
7 the labeling as well for enriching it on a
8 mechanistic basis, to say that the patients will
9 first be shown responsive to this class of
11 DR. KATZ: Dr. Dal Pan, any comments on
12 the issue of the regulatory issues for enriched
13 enrollment trials?
14 DR. DAL PAN: First I would like to say
15 that some of the things that have been brought up
16 here about identifying who the drug is affective,
17 and which subgroups may respond, a lot of that is
18 what Phase II of drug development is about. It is
19 about defining and characterizing the effect of the
21 Then we traditionally call, then, Phase
22 III, the confirmation of that finding. So I think
23 what some of the committee members here have really
24 done is distinguish between what should be done in
25 Phase II and what should be done in Phase III. You
1 don't just start with an hypothesis and jump into a
2 confirmatory trial. There is some not only
3 dose-finding but also some hypothesis-testing of
4 what the range of what the range of effects of the
5 drug could be including in specific subpopulations.
6 So I think that is a lot of what is going
7 on here. With regard to specific labeling, maybe
8 one of my colleagues could answer. If we could
9 actually put something in the label about what Dr.
10 Shafer was mentioning, the patient may be
11 responsive to Drug X if they respond to an I.V.
12 opioid, for example.
13 DR. McCORMICK: First let me just say that
14 what you have just described as the ideal in Phase
15 II development is an ideal. It is something that
16 we often don't see bear fruit in Phase III so
17 frequently we aren't able to really identify the
18 real responders and parse them out of the clinical
20 But, if we were, if we had a mechanism to
21 identify responders and if it was adequately
22 studied, then we certainly would consider how that
23 would find its way to the label.
24 DR. KATZ: Other comments from FDA folks
25 on the regulatory implications of enriched
1 enrollment designs?
2 Thank you, Dr. Dal Pan, very much. Why
3 don't we then go on to Dr. Cornblath. I'm sorry;
4 one question, Dr. Rowbothom?
5 DR. ROWBOTHOM: I was just going to make
6 one comment about study designs using some kind of
7 a potentially predictive test. I have had a number
8 of discussions with various pharmaceutical
9 companies about Phase II studies that use things
10 like I.V. lidocaine infusion or I.V. opioid
11 infusions. Generally, there has been hesitancy to
12 adopt those designs because of potential risks of
13 the I.V. infusion, what do you do with patients who
14 don't respond to the I.V. infusion, a number of
15 other methodologic questions, plus there is very
16 little published literature in that area.
17 So, although it is a very intriguing idea
18 and the evidence that is available suggests that it
19 would be a valid and successful approach, there is
20 still very, very little data actually in the public
21 domain that is available on that.
22 DR. KATZ: Thank you.
23 Without further ado, Dr. Cornblath.
24 Electrophysiologic Tests Used in the Evaluation
25 of Peripheral Neuropathy and Neuropathic Pain
1 DR. CORNBLATH: Thank you.
3 I would like to make three sort of opening
4 comments. One, I would like to thank the
5 organizers for asking me to come. It is a pleasure
6 to be here. Two, I notice the chair next to me,
7 Dr. Dyke, is not here. I think a lot of us in the
8 room owe him a great gratitude of thanks for all
9 the work that he has done over at the Mayo Clinic
10 over many, many years. I will be quoting liberally
11 from that.
12 The third is that I think there are still,
13 and we will hear this from Michael and Eva, a lot
14 of unresolved issues from the scientific standpoint
15 here that are, if you will, separate from the
16 industry issues but tie in very closely. Eva, I
17 know, will be bringing up a number of these talking
18 about these composite measures and particularly
19 their use over time.
20 There is a document currently in
21 preparation coming from the NIH to the Congress, I
22 believe, on issues related to diabetic neuropathy
23 and unresolved scientific issues that, if I am
24 correct, Eva, should be available in the next
25 months, should be out, and will highlight a number
1 of the issues that all of us are bringing up that
2 are still ripe for funding from the NIH.
3 So, with that brief introduction, let me
4 just say I am going to talk briefly on this topic.
5 There is a lot written and what I have tried to do,
6 basically, is boil it down to sort of a summary
7 essence without a lot of data. Gerald and I talked
8 about sort of what I was supposed to say.
10 This is sort of the outline of what he
11 told me I was supposed to say which is I was
12 supposed to talk briefly about electrophysiologic
13 tests, their natural history in diabetes, the
14 correlation with outcomes, their use in clinical
15 trials, a few practical issues and then I could
16 give my own summary.
18 So there are a number of
19 electrophysiological tests available. I have
20 changed the term briefly, as you will see here, to
21 neurophysiological tests and, in fact, I saw Joe
22 Arezzo, who is in the audience in the back, who is
23 really a world-class expert in this. I hope he
24 will correct me when I am wrong.
25 But there are a number of tests available
1 that can be used. I think one of the issues we
2 keep hearing about is nerve conduction, nerve
3 conduction, nerve conduction and, although that is
4 the most studies and what I will spend most of the
5 time talking about, you should be aware that there
6 are a number of other testing modalities available.
7 Not all have been as well studied but all are out
8 there, all have been looked at to some extent in
9 terms of reliability, validity and, in some cases,
10 change over time.
11 The main ones that you hear about are
12 sensory-motor-conduction studies and, in
13 particular, as I will mention later from the
14 Japanese, the use of F-waves in monitoring
15 long-term electrophysiologic change in diabetic
16 neuropathy, electromyography--that is, the actual
17 placing of a needle in the muscle because that is
18 viewed as minimally invasive, hasn't really been
19 used much--although it is possible to do it, it
20 hasn't been used much--quantitative sensory
21 testing, and there are number of devices out there
22 that can be used.
23 They are part of many of the composite
24 measures that you will hear about from Eva and is a
25 very nice and, i some cases, very simple highly
1 reproducible test that we shouldn't forget about.
2 Autonomic-function testing and QSART are,
3 in my view, much more advanced. They require a
4 degree of sophistication and expertise and don't
5 yet have the longitudinal multicenter experience
6 that I think we would like to bring these into
7 clinical trials currently.
9 The most comprehensive data we have in
10 that we in the natural history of EDx studies in
11 diabetes is longitudinal studies of a large number
12 of diabetics who were tested very carefully using
13 the Mayo measures which, again, are highly
14 reproducible within their centers. They have
15 published and studied over a long time.
16 As Eva will tell you, this is what is
17 needed very dramatically with other measures and in
18 other centers and in other populations. Some of
19 that work is being proposed today. There is an
20 enormous need to look at other measures in other
21 populations over time. But this is the best data
22 that we have and I won't read the little numbers up
23 there. You can read them for yourselves. They are
25 But the data is very solid that if you do
1 the NIS(LL)+7, you have highly competent people to
2 do it, you are doing it at a center where
3 essentially it was invented, you can show that
4 there are these very precise changes over time and
5 everybody from industry knows that you can then use
6 these to say whether you want to, as was proposed
7 earlier, show that you can slow the rate of
8 progression, you can stop a disease of, in fact,
9 you can improve a disease.
10 There are a lot of other measures that
11 have been used. They all show the same thing; that
12 is, a worsening over time. But none have the sort
13 of extensive precision that the NIS(LL)+7 has.
15 Again, the best data comes from Peter Dyke
16 and his colleagues at Mayo. It essentially shows
17 that nerve conductions, and, again, I am going back
18 to nerve conductions, are clinically meaningful if
19 you accept the statement, and it is hidden in
20 there, that a two-point change in the
21 neuropathy-impairment score is a clinically
22 meaningful measure. Again, I don't know how many
23 here are neurologists and have done this measure.
24 Two points is, in my view, sort of right at the
25 border of what probably two of us could get when we
1 are doing it based, side to side, the same
3 But at least, when it is done by Mayo
4 physicians at the Mayo Clinic, this is a very
5 reliable number and it is equal to a precise change
6 in nerve-conduction velocity of either a composite
7 number of nerves or a single nerve or a change in
8 the amplitude for either the composite nerves or
9 single nerves.
10 So, if you can get the nerve conductions
11 done, you can both look at amplitude and velocity
12 in these motor nerves and you can show that they
13 are equivalent to a change in the NIS score and two
14 points on the NIS score is a significant clinical
17 So where do we sort of stand? Again, this
18 is summarizing a lot of data that is out there in
20 nerve-conduction studies. They have been used
21 forever. Probably the first one where it was used
22 was, in fact, Eliason's study of diabetic rats
23 where he made them diabetic and he could show that,
24 in the diabetic rats, nerve conduction worsened
25 compared to the controls. That was, I guess, in
1 the 50s. Since that time, nerve conductions have
2 been used time and time again, either primarily or
3 secondarily in this.
4 They clearly have shown in diabetes an
5 improvement when the change in the diabetic case is
6 very dramatic; the introduction in insulin therapy,
7 the introduction of pumps, or dramatic treatment in
9 The third one is the one that has bothered
10 everybody. Mark has already mentioned it. All the
11 drugs have failed. Therefore, "all the composite
12 measures have failed." One of the difficult
13 questions that I think all of us around the table
14 are asked constantly from industry which is, is it
15 the drug or is it the measure.
16 I think that, for the moment, we can't be
17 certain except to know that both have failed. We
18 can say it is the drug and, therefore, the measures
19 couldn't have worked or we could say actually we
20 thought the drug was pretty good, but the measures
21 were not very good. It is sort of a cart and horse
24 So there are a number of practical issues
25 to consider when looking at these. The first is
1 what is the outcome that you are actually looking
2 for and what is the fiber population that you are
4 So these nerve-conduction studies, as the
5 neurologists know, are predominantly large-fiber
6 measures. If you are looking for a drug that is
7 going to affect a small-fiber function, then you
8 wouldn't do nerve conductions because it is not
9 going to get at it. But you might do either skin
10 biopsies, which you will hear about, or
11 quantitative sensory-testing measure to look at
12 small-fiber function.
13 So this is an issue that comes up time and
14 time again. Think about the fiber population that
15 you want to affect, and then pick the endpoint
16 measure that you are interested in. What parameter
17 is going to get better? Is it a velocity parameter
18 which happens very quickly if you improve diabetic
19 control or is it an amplitude measure which is
20 going to be most likely to take a long period of
21 time and have a slower change because it is
22 fundamental property of nerve regrowth and
23 collateral reinnervation?
24 Last, as you can see, you fast will the
25 intervention work? If you improve glycemic
1 control, nerve-conduction will change very quickly
2 but then, after that, it is going to stay very
3 stable while amplitude won't change except very
4 late in the study.
5 That comes into the second issue here
6 which is what, really, will your drug do? What is
7 it going to affect? Is it going to affect
8 velocity? Is it going to affect large fiber, small
9 fiber, autonomic function and then you need to go
10 into the top issue to pick the outcome choice that
11 you want.
12 I think the last question comes up quite
13 frequently. The answer is an unequivocal yes. All
14 of these techniques can be done. With training,
15 you can get away from this issue of the test is too
16 complicated or the measure is so complex and there
17 is such variability that nobody could ever do it
18 and we have got to do something stupidly simple.
19 The answer is it has been done time and time again.
20 You can do nerve conductions. You can do
21 quantitative sensory testing in multiple sites.
22 You just need a little bit of training like you do
23 for a neurologic exam. I said here in the note
24 that there have been some multicenter Japanese work
25 that has been done looking at nerve conduction and
1 they have shown that F-wave is an extremely robust
2 measure and probably, in their hands, the best
3 measure in terms of reliability.
4 But, again, before accepting that, you
5 need to decide, is the F-wave going to change in
6 your trial and is that what you are interested in.
8 Let me try and summarize because I think
9 we ought to leave more of the time for discussion,
10 clearly nerve conductions are the best studies and
11 the most accepted tests. They correlate with
12 measures. A change in time is real and that they
13 can look at both worsening and improvement.
14 The other electrophysiologic tests are
15 there. They are good, but a lot of them we need
16 more data. That is what this NIH report to
17 Congress is going to say in some respect. We have
18 got to figure out can these others be done and can
19 they be done in large populations over time.
20 That the nerve conductions are
21 particularly important in my view as we think about
22 disease-modifying agents, and, again, we will hear
23 more of this from Eva, I hope, in these composite
24 measures. The Peter Dyke one is the NIS(LL)+7. We
25 have done TNS and Eva has done her own. But they
1 are all useful because they look at a variety of
3 You can then begin to look the subdomains
4 essentially suggesting a little bit of what Mark
5 said, that there may be subpopulations or
6 submeasures of these larger domains that improve at
7 a time when the main domain may, in fact, not
10 I have not really talked about the issue
11 as regards to symptom of neuropathic pain because I
12 view that as symptomatic treatment. The
13 electrophysiologic tests shouldn't be forgotten,
14 either nerve conductions or quantitative sensory
15 testing. Both we and Joe Arezzo and others have
16 shown that these are extremely valuable in toxicity
18 So, if you think your drug is going to
19 cause a problem, even though it may help symptoms,
20 these are very reliable measures to look at but
21 they really don't have a use in outcome criteria
22 for these kinds of pain studies because they look
23 at large fibers which are not going to be affected
24 and they are fundamentally not altering the
1 Thank you.
2 DR. KATZ: Thank you, Dr. Cornblath.
3 Any questions from around the table for
4 Dr. Cornblath?
5 DR. BRIL: I have a question.
6 DR. KATZ: Yes. Dr. Bril?
7 DR. BRIL: Thank you for that reminder of
8 the importance of nerve-conduction studies. I
9 guess my question had to do with the magnitude of
10 change which is the essential question because the
11 thing that we all see changing is conduction
13 One of the problems with using nerve
14 conductions as a surrogate is what does it mean.
15 So I would challenge you to just tell us and share
16 with us the magnitude of change after
17 transplantation, the magnitude of change in
18 velocity or amplitude after a year or two after
19 transplantation or after the insertion of an
20 insulin pump because, although Peter Dyke has
21 developed those quantitative measures that say you
22 have to have 2 meters per second in order to detect
23 a clinical change, I would be surprised if you can
24 obtain that degree of change very easily in a
25 chronic disorder such as diabetic neuropathy.
1 So could you just clarify that?
2 DR. CORNBLATH: Yes. The data, and this
3 is one of these unfortunate things, that the kind
4 of comparative data that you would like, Navarro
5 has the best data from Minnesota on the degree of
6 change in nerve conduction but they are not doing
7 it in extent with NIS scores or NIS(LL) scores so
8 it is a little bit of apples and oranges.
9 But these kinds of values are very easy to
10 see after the several meter per second, after
11 implantation of pumps or the beginning of insulin
12 therapy. It is very common to see multimeter
13 changes in their hands.
14 Now, they didn't go back and look at the
15 change in terms of NIS(LL) or in terms of other
16 quantitative measures.
17 DR. BRIL: But I think if you follow them
18 out for five years, it may be a meter per second
19 but it is not that quickly, that rapidly. The
20 magnitude isn't that great in a short time after
22 DR. CORNBLATH: It can be when the
23 diabetic control goes to normal.
24 DR. BRIL: Well, perhaps, in a few. But,
25 over the long term, I think the mean changes are
1 not that great. If you see the non-transplant
2 versus the transplant, they do separate, but
3 slowly. The magnitude is not that great in mean
4 numbers. Yes, in selected patients, you may have
5 large shifts but you do that in almost any study.
6 DR. KATZ: Other questions for Dr.
7 Cornblath? Go ahead.
8 DR. DWORKIN: It sounded like you were
9 suggesting that the NIS(LL)+7 has considerable
10 reliability and validity but all the data are from
11 the Mayo Clinic. Is that the case?
12 DR. CORNBLATH: Eva can speak to that.
13 DR. DWORKIN: So that will be--thanks.
14 DR. KATZ: I have a question if nobody
15 else does about the NIS(LL)+7. My understanding
16 from the literature is that when the folks at Mayo
17 were trying to figure out what degree of change in
18 this composite disability score is clinically
19 meaningful, they decided to focus on what the
20 minimum change was that a physician, a neurologist,
21 could detect in that exam.
22 So the two-point change in the NIS was
23 arrived at based on the conjecture that that was
24 the minimum number of points a physician could
25 detect and then that, somehow, got translated into
1 that must be what is the minimal change that is
2 clinically meaningful for patients.
3 My question is what is the evidence,
4 actually, that that two-point change in the NIS is
5 clinically meaningful for patients.
6 DR. CORNBLATH: Go ahead. This is a tag
8 DR. FELDMAN: Actually, the history of
9 that is that the Peripheral Nerve Society met. Dr.
10 Dyke chaired the meeting and there were probably
11 about 100 of us there. A consensus was reached
12 that two points was a meaningful change in the NIS.
13 So that was done somewhat prospectively by a group.
14 I am fairly sure you are referring to Dr.
15 Dyke's paper in Neurology, I think 1997 or 1998,
16 where he, then, looks at the NIS(LL)+7, and I will
17 be discussing this when I speak, and looks at the
18 change in the NIS(LL)+7 over time, which David
19 mentioned, and then, separately in that paper,
20 says, but if we wanted to look at two points in the
21 NIS, which is very different than the NIS(LL), and
22 I will also explain that to you, then this is what
23 we would supposedly need to see in terms of numbers
24 of patients and time.
25 So that was very arbitrarily chosen.
1 There are data from Peter Dyke, though, looking,
2 for example, at sural-nerve biopsies and comparing
3 them to nerve-conduction velocities and degree of
4 clinical impairment, as there are from other
5 individuals, and I will also discuss that.
6 But this two points on the NIS was kind of
7 grabbed from the sky.
8 DR. KATZ: So, if I am hearing you
9 correctly, there really is no evidence that that
10 is, in fact, the change that is meaningful to
12 DR. FELDMAN: What I will do is show you
13 composite scores where the NIS is a part of the
14 composite score but whether or not--the NIS,
15 itself, is a total neurologic exam so two points--I
16 mean, you could have a cranial-nerve abnormality
17 and that could give you two points. Or you could
18 have shoulder weakness.
19 So it may not necessarily be relevant, the
20 entire NIS. Now, the NIS(LL), which I will show
21 you, is more targeted but still has a large motor
22 component to it.
23 DR. CORNBLATH: If I could comment. I
24 think you are absolutely right. I said "a
25 clinically meaningful." I didn't use the words,
1 and you have added them, "to patients." So there
2 is a thing that will come around which is can we
3 take that and put it with some symptom score or
4 some giant quality-of-life event. As far as I
5 know, that has not been done.
6 Do you know that, that is a change in the
7 NIS at the same time in a study looking at the NIS
8 change with a QOL measure? I don't know that that
9 has been done.
10 DR. FELDMAN: I don't think so, either.
11 DR. CORNBLATH: No; I don't think any of
12 us know of that.
13 DR. KATZ: Dr. Shafer, I think you were
14 actually on deck first. Did you still have a
16 DR. SHAFER: That was it.
17 DR. KATZ: Dr. Woolf and then Dr. Bril.
18 DR. WOOLF: You highlighted the
19 difficulty, or the impossibility, with
20 electrophysiology of looking at small-fiber
21 function. I just wanted your views, the difference
22 in susceptibility in terms of large-fiber between
23 sensory and motor and you didn't mention
24 sympathetic small fibers at all.
25 DR. CORNBLATH: As you know, the
1 techniques that are required are highly specialized
2 and very difficult in terms of patient cooperation.
3 As far as I know, they have not been used in trial,
4 unless I am mistaken. So that is why I didn't
5 bring them up. But, theoretically, one could look
6 at these at C-fiber conduction, C-fiber spontaneous
7 firing, but they are technically very demanding.
8 Is that correct? Yes? Thanks.
9 DR. KATZ: Dr. Bril?
10 DR. BRIL: I guess I had a couple of
11 comments as well. I know we will hear more about
12 the NIS. I think it is limited for diabetic
13 neuropathy for various reasons which I am sure Eva
14 will discuss when she is discussing the scales,
15 partly because it is so heavily weighted to motor
17 But I guess I would ask Dr. Cornblath what
18 he thinks. That scale just takes a couple of
19 nerve-conduction parameters that seem to fit with
20 the group. Should we just be doing one
21 nerve-conduction parameter or do you think that if
22 we are going to do nerve conductions we should do a
23 full assessment? Should we do summary scales of
24 nerve conductions?
25 How should we handle this large volume of
1 data and look at it, not combining it with clinical
2 scales, but just by itself as a large-fiber
3 measure, as the most accurate large-fiber measure
4 we have. How should we actually handle it?
5 DR. CORNBLATH: Again, I think the issue
6 is really what you are looking for if you are doing
7 a trial in terms of drug development; that is, do
8 you expect that your measure is going to improve
9 conduction velocity, as you might in a demyelating
10 neuropathy and as has been shown in CIDP for
11 example, or do you think you are going to affect
12 nerve function, itself, with connections at the end
13 in either motor or sensory fibers over a longer
14 period of time in which you would prefer to do
16 So I think, in my view, it is what you
17 want to ask. You are going to get, as you point
18 out correctly, a large number of measures and most
19 of the either composite measures or, when it is
20 done singly, have only selected out one or two of
22 Since you and I do these every day, there
23 are ten or fifteen or twenty individual parameters
24 that we get. I think what has happened is that the
25 composite people, when we developed ours and when
1 Peter developed NIS(LL)+7, picked out those that
2 either we thought were going to be helpful, so we
3 picked out a motor and a sensory amplitude for TNS
4 and he has picked out a number of other things for
6 But I think it ought to be prospectively
7 thought based on what you think the effect is.
8 DR. KATZ: Actually, Dr. Farrar, you were
9 first and we will keep going from there. Did you
10 have a question, John?
11 DR. FARRAR: I really wanted to point out
12 and would ask for your comment on the following
13 which is that, ultimately, the real issue is what
14 is the question. That is probably the first of
15 many times that you will hear that over the course
16 of the day.
17 I am not overly familiar with this
18 particular scale, but the fact that there are motor
19 components to it clearly is asking a different
20 question than if there was a strict sensory
21 neuropathy. You wouldn't be able to, perhaps,
22 detect it with that.
23 I think the other issue I wanted to point
24 out is that EMG and even quantitative sensory
25 testing to a degree depend on a generalized
1 disease. Diabetes and postherpetic neuralgia are
2 clearly very different. It would be very hard, I
3 think, to find an EMG abnormality in someone with
4 postherpetic neuralgia. I don't know whether that
5 has been done.
6 I think it is important to keep in mind
7 that this discussion is targeted at two very
8 different issues, one of which is diabetic
9 neuropathy and the other is nerve-induced pain. It
10 is clearly reasonable to consider the two together
11 because diabetic neuropathy is one of the causes of
12 neuropathic pain.
13 But I would just like the committee and
14 would ask your opinion about whether, in fact, EMG
15 abnormalities or even quantitative sensory testing
16 abnormalities are necessary for a patient to
17 experience pain.
18 DR. CORNBLATH: That is why I broke that
19 up in the summary. So most of what we are talking
20 about in the use of neurophysiological
21 electrodiagnostic tests is, absolutely you are
22 correct, applicable to the so-called
23 disease-modifying issue here. I don't think they
24 play much of a role, if any role, in the other
25 state. Mike probably has the best QST data in PHN
1 if you want to speak to that.
2 DR. BRIL: Could I just answer one thing
3 to that. There are two areas, symptomatic
4 improvement and then specific modifying disease.
5 However, there are recent studies. There was a
6 study I was involved in--I know it was
7 retrospective but it showed that by selecting
8 patients, depending on the electrophysiological
9 severity, those who responded were those who still
10 had residual nerve function that worked.
11 The role of electrophysiologic studies in
12 a disease such as diabetic neuropathy in treating
13 painful symptoms may be to stratify the patients
14 and help determine or predict who would respond and
15 that would be the role. I have seen that. There
16 are some posters at the ADA that are going to say
17 something similar.
18 So if you have sural response left, it
19 predicts a response to the intervention as opposed
20 to if you don't have a sural-nerve response left.
21 So, clearly, the number of surviving large fibers
22 really does have an indication to, perhaps,
23 small-fiber function or response to pain. Now, not
24 in postherpetic neuralgia or trigeminal neuralgia.
25 Those are totally different disorders with
1 different endpoints.
2 DR. KATZ: Dr. Rowbothom, I think the
3 specific question is what the role is for
4 electrophysiology or quantitative sensory testing
5 in clinical trials in postherpetic neuralgia, if
6 you wanted to comment on that.
7 DR. ROWBOTHOM: Thanks. A few things.
8 One is quantitative thermal sensory testing in
9 postherpetic neuralgias can be readily performed.
10 There are some difficulties in interpretation just
11 because some patients have such a hyperalgesic
12 response to heat stimuli and they fatigue very
13 quickly. So it is difficult to do those studies.
14 What we have evolved towards is using that
15 plus things like targeted application of capsaicin
16 in the area of pain and evaluating the response to
17 that and skin-biopsy assessments rather than
18 relying on a single tool such as quantitative
19 thermal sensory testing.
20 For most patients with postherpetic
21 neuralgia, the great majority are going to have it
22 on the trunk or on the face which are places that
23 are just completely impossible to do conventional
24 nerve-conduction studies.
25 DR. CORNBLATH: We wouldn't use then, in
1 any case.
2 DR. ROWBOTHOM: Exactly.
3 DR. KATZ: A specific comment about that
4 issue? Dr. Dworkin.
5 DR. DWORKIN: When you are talking about
6 QST and PHN, you are referring to it as a way of
7 selecting patients and, perhaps, predicting
8 treatment response or do you also mean with respect
9 to an evaluation of treatment response as an
10 outcome measure?
11 DR. ROWBOTHOM: None of our studies have
12 actually used QST as an outcome measure over time.
13 We did some work with looking at acute changes in
14 it but not exactly what you are referring to that
15 would be more analogous to the diabetic-neuropathy
17 DR. KATZ: Dr. Rendell?
18 DR. RENDELL: With respect to Vera Bril's
19 comments, it does raise an important issue because,
20 in our diabetic-neuropathy trials, clearly David
21 and Peter make a big issue of how well
22 nerve-conduction tests are done. Yet, in pain
23 trials, nerve-conduction tests are not done very
24 well. They are not standardized in many trials and
25 the question is should we be applying the same
1 rigor to nerve-conduction trials and pain trials
2 that we are doing in functional trials of diabetic
4 DR. KATZ: Dr. Bril?
5 DR. BRIL: Absolutely. The trial that I
6 spoke about initially was a symptomatic trial and
7 the nerve conductions in that trial were done with
8 the same rigor as some of the more specific trials.
9 It was a post hoc analysis so it is weak, and who
10 knows, and development has not proceeded with that
11 particular agent.
12 But, looking at them, there was a clear
13 separation with and without surals. Then there is
14 more recent work that is being present at the ADA
15 that showed changes in a composite symptom score or
16 positive symptoms of neuropathy and those
17 determined somewhat by the presence or absence of
19 So I would say yes, definitely. In the
20 studies of diabetic neuropathy. Now, I know this
21 isn't very popular in a lot of pain clinics because
22 a lot of patients have advanced disease and lack
23 surals and there is always the wish to include
24 these patients in trials as well, and so maybe they
25 should be, but a stratification done with respect
1 to who has surals and who doesn't, and since surals
2 can be technically challenging, yes; they have to
3 be done with the same rigor.
4 DR. KATZ: Dr. Shafer?
5 DR. SHAFER: Bucking the trend, I am going
6 to direct this question to the speaker.
7 DR. CORNBLATH: He prefers not, but--
8 DR. SHAFER: What I have heard is that
9 these tests on nerve conduction with proper
10 training and guidance can be objective and
11 reproducible, although that is obviously a source
12 of some debate here. Also a source of some debate,
13 but, in your opinion, is that they can show changes
14 in a tractable time course for a clinical trial.
15 If that is the case, if a company wished
16 to make a claim that preservation of large-nerve
17 function was a good thing and that they had a drug
18 that would help to preserve large-nerve function in
19 diabetic patients, would neuropathic-pain studies
20 be appropriate as a primary endpoint for a clinical
22 DR. CORNBLATH: I think they would
23 because, as we heard before, you could say it is
24 the proper question driving the choice of the
25 endpoint. If the endpoint, you believe, is that
1 you can save sensory-nerve function and one measure
2 of doing that is to look at the amplitude of the
3 sural response done by trained people in the same
4 way where, again, I really do believe that the
5 issues of reliability, variability, et cetera,
6 inter- and intra-rater reliability are all put to
7 rest, then you are asking the right question.
8 You are asking for the right measure. But
9 all you need now is some knowledge of the magnitude
10 of change over time in that measure in the target
11 population. That is, I am sure Eva will say, one
12 of the things that we are missing because that kind
13 of information is either out there for the
14 Rochester study or hidden proprietarily in many of
15 the companies who have done negative studies.
16 Some of it is published but a lot of it is
17 hidden within centers. But I think you could ask
18 that question and it would be appropriate. We have
19 to get to the issues that the Chair raised about
20 what is meaningful. Is it okay to have your
21 amplitude be 1 microvolt better than the other
23 DR. KATZ: I am going to take the
24 prerogative of calling for a break now. These are
25 all questions that will fill the rest of our day's
1 discussion and I am sure we won't lack them. So
2 let's resume in fifteen minutes.
4 DR. KATZ: Dr. Eva Feldman now will speak
5 to us on scales used for the evaluation of
6 peripheral neuropathy. Dr. Feldman?
7 Scales Used in the Evaluation of
8 Peripheral Neuropathy
9 DR. FELDMAN: Thank you very much. I am
10 really very pleased to be here today.
11 What I think I am going to do is maybe
12 take one step back and define diabetic neuropathy
13 as we see it as neurologists and I believe as
14 probably most clinicians see it and then tell you
15 about scales and really an historical manner and
16 how they developed over time, and really highlight
17 some of the major trials that have already occurred
18 that have, unfortunately, not been successful as we
19 have heard, and then end by trying to pull together
20 what I think are the best composite scales that are
21 currently available.
23 So, as you can see here, the definition of
24 diabetic neuropathy--it has been defined by the
25 World Health Organization as a disease
1 characterized as a progressive loss of nerve fibers
2 eventually leading to sensation loss, foot
3 ulceration and amputation.
5 Here is, I will say, is the star of the
6 hour and that is the myelinated nerve. I just
7 really wanted to remind you, the nerve cell body
8 that we are interested in lies either in the
9 dorsal-root ganglion neuron for the sensory nerve
10 or the spinal cord for the motor nerve. It gives
11 out this large axon that has to transverse down the
12 length of the arm or the leg.
13 Then there are these nerve terminals. In
14 a sensory nerve, as you know, these nerve terminals
15 then bring afferent input into the spinal cord and,
16 in a motor nerve, there is efferent output that
17 goes out.
18 Now, the terminology in the peripheral
19 nervous system is actually a little confusing.
20 Many people refer to this as a nerve.
22 But, as you can see in the next slide,
23 this nerve really lies in what is also known as a
24 large nerve fiber or a nerve bundle. So there are
25 multiple individual nerves in these individual
1 fascicles and theses individual fascicles of nerves
2 together make up either a pure sensory nerve, a
3 pure motor nerve or, more commonly, a mixed nerve.
4 What is important is I have shown you an
5 example of the myelinated nerve but, as you have
6 heard earlier, it is not just myelinated nerves
7 that we are interested in but in this mixed nerve
8 bundle, in this fascia, there are also unmyelinated
9 nerves and thinly myelinated nerves. These nerve
10 fibers carry distinct types of information.
12 In the peripheral nervous system, damage
13 due to diabetes is thought to primarily be axonal
14 in nature, at least initially, although there
15 likely is some demyelinating component, some attack
16 in the myelin in the peripheral nervous system.
17 Here is just an example, a diagram,
18 showing distal axonal loss of a neuron. We also
19 believe that there could be primary insults to the
20 dorsal-root ganglion neuron. But what one then
21 sees, though, is distal loss of nerve function
22 really mimicking, then, the pathology. Again,
23 depending on what nerve fiber type is involved,
24 that would, of course, then, depend on what type of
25 symptoms and signs you would find as the clinician.
2 So neuropathic systems and signs, to
3 summarize, are going to reflect the type of
4 nerve-fiber damage. I think, very importantly, and
5 it has been alluded to today, but diabetic
6 peripheral neuropathy is primarily a sensory
7 neuropathy. These thinly myelinated or
8 unmyelinated fibers that we have been discussing,
9 they mediate pain, alter cold, heat and light
11 These are the fibers that are difficult to
12 measure on standard nerve-conduction studies and
13 really require more sophisticated techniques that
14 are not routinely done in clinical trials. In
15 contrast, the large myelinated fibers, these carry
16 vibration, proprioception, your position sense from
17 the mechanoreceptors. These are easily measured on
18 nerve-conduction studies.
19 Most frequently likely both fiber types
20 are involved in diabetic neuropathy but it is very
21 important to understand that the pain component of
22 neuropathy is more likely mediated by the
23 small-fiber component although there are people who
24 believe that joint pain is a component of
25 neuropathic pain and we could discuss that this
2 So it isn't just one simple disease.
3 There are some patients who have very painful
4 neuropathy and when you examine them, they have
5 normal nerve conductions, normal vibration and
6 normal proprioception. Then there are some
7 patients who have little pain and when you examine
8 them, what you see is sometimes light touch is
9 moderately intact but they have a large loss of
10 vibration and proprioception.
11 So this disease can selectively affect
12 different fiber populations although most commonly
13 it does affect both, although we don't understand
14 why some people have more pain than others and we
15 will discuss that later also.
17 To now put things in context, you can
18 imagine a patient that has selective disease of,
19 say, a group of large myelinated fibers and small
20 myelinated fibers in a distal to proximal gradient.
21 That will then cause the symptoms that the patient
22 most notes.
24 These symptoms can be acute in onset or
25 very insidious. The course of the symptoms that
1 patients complain of could be monophasic, meaning
2 they start and they just kind of keep on going, or
3 they can be fluctuating with or without drug
4 intervention sometimes dependent, of course, on
5 glycemic control.
6 Now, the sensory symptoms; really, as a
7 neurologist, we talk of two types of sensory
8 symptoms. You can have what we call negative
9 symptoms. That is the patient is numb, and they
10 really have loss of sensation. They are not going
11 to come to you as the clinician and say, I'll come
12 to you to enter a trial in pain, because they
13 really have just what is called the insensate or
14 numb foot. That is believed to account for about
15 80 percent of the patient population that has
16 diabetic neuropathy at any one time.
17 In contrast, those patients who have
18 tingling, prickling, burning pain, those are called
19 positive symptoms. Those are believed to account
20 for approximately 20 percent of the patients at any
21 one time.
22 Importantly, and this will be very
23 important when we talk about the scales, you know,
24 real motor symptoms are rare in diabetic
25 neuropathy. Certainly, there is a subset of
1 patients that have motor involvement, but this is
2 primarily a sensory neuropathy and we need to keep
3 that in mind as we are looking at scales.
5 Now, the signs that you see when you
6 examine a patient and this becomes, again,
7 important as we design our clinical tools for our
8 trials, is you will see a dry, atrophic skin in the
9 feet, loss of hair and sweating and, in more
10 advanced cases, distal muscle atrophy. Sensory,
11 again, findings are the most common and we have
12 already talked about the large-fiber findings, the
13 vibration and proprioception and then the
14 small-fiber findings of light touch and pin prick.
15 Motor would be distal muscle weakness.
16 Let me emphasize distal because when I talk about
17 the NIS(LL), that has a large component of proximal
18 motor examination which then makes it not really
19 relevant to us. And then reflexes are either
20 absent or depressed.
22 So, in summary, anatomic changes that I
23 have discussed, leads to these signs and symptoms
24 giving you this class of diabetic peripheral
25 neuropathy, this stocking-glove pattern that
1 everyone discusses and that we have all seen as
4 Up until I would say approximately the
5 mid-1980s, the diagnosis of neuropathy and the
6 epidemiological studies were really somewhat--oh,
7 they weren't particularly prospective. They were
8 mainly retrospective. There were some prospective
9 trials but they were done as the clinician so
10 decided to do it, like Peral looked at 4,400
11 patients and he used vibration as the way to
12 determine whether or not they had neuropathy.
13 It wasn't until the San Antonio consensus
14 statement occurred in 1988, and this was formed by
15 a consensus statement from the American Diabetes
16 Association and the neurologic community led, in
17 part, by Peter Dyke and also Jack Griffin from
18 Johns Hopkins, that it was said that if you are
19 going to look at diabetic neuropathy in a
20 quantitative fashion for a clinical trial, you
21 should look at some sort of clinical scale.
22 At that time, the Neurologic Disability
23 Score, which is the mother of the NIS, or the
24 father, considered the quantitative sensory
25 testing, autonomic function testing and nerve
1 conductions. Based on abnormalities in these,
2 patients were actually staged as Stage 1A through C
3 if they had no symptoms, and these would be
4 positive symptoms, or Stage 2A through C if they
5 had positive symptoms.
6 One of reasons this all happened is that,
7 at the same time historically, the DCCT was being
8 designed and occurred. As you recall, in the DCCT,
9 neuropathy was examined.
11 The way the DCCT was designed, and
12 remember, it is occurring in parallel with San
13 Antonio consensus criteria, is they decided to
14 define diabetic neuropathy by simply a clinical
15 exam by a neurologist in nerve-conduction studies.
16 The clinical exam was very simple. You looked at
17 sensation, small and large fiber, and ankle
18 reflexes. There was a symptom score. If two of
19 the three were positive, meaning sensation, if you
20 had abnormalities in sensation, reflexes or the
21 symptom score, you had probable diabetic
23 If you had all three, then you had definite
24 neuropathy. Nerve-conduction studies
25 were also performed on the DCCT patients. This, I
1 think, was useful and could prove to be useful to
2 us in the future on 1,243 patients at baseline and
3 five years later. The perineal motor nerve
4 conduction was 3.5 meters per second faster in the
5 intensive versus the conventional treatment group
6 after five years.
7 This is one of many studies that shows
8 that the perineal nerve conduction, while it is a
9 motor nerve and I have told you that this is
10 primarily a sensory disorder, this particular motor
11 nerve, as in some studies, the medium motor nerve,
12 the conduction velocity does appear to be possibly
13 a good surrogate marker for disease progression.
15 Here is, actually, the neurologic outcome
16 of the DCCT. I think this is interesting to see,
17 if you want to look. The black are the intensive
18 patients and the hatched the conventional. You can
19 see the difference in terms of better outcome in
20 the intensive group when it comes to nerve
21 conductions in neurologic examination, and also
22 autonomic-function studies were done. Our
23 variability was done in the DCCT.
25 I am going to show you this slide because
1 I would like to make a pitch. The DCCT patient
2 population is currently being followed as the EDIC
3 population, so they continued the DCCT, as you all
4 know, for another ten years. The only part of the
5 neurologic examination that is being done as
6 something is a tool that we developed, that David
7 mentioned, called the MNSI, the Michigan Neuropathy
8 Screening Instrument.
9 It is a very simple tool. What it shows,
10 and I am not going to go over it in any detail, but
11 just it does show that if you look at the percent
12 of patients with neuropathy by the MNSI, those
13 people who were on intensive therapy remained with
14 a lower percentage of neuropathy than those
15 patients on conventional therapy in both the
16 primary and the secondary cohort.
17 I show you this not to tout the scale that
18 we developed but rather to say that these patients
19 would be available for nerve-conduction studies and
20 that would give us, over time, nerve-conduction
21 studies in a well-classified patient population at
22 Time 0, or say, Year 0, Year 5 and then later on.
23 They are now entering approximately Year 9, maybe
24 Year 10.
1 About the early '90's--so the DCCT is
2 ongoing. We don't have those results. We have had
3 the San Antonio consensus criteria and now people
4 begin to look at drugs in the treatment of diabetic
5 neuropathy and, really, what primary endpoint to
6 use, what should be the primary efficacy point.
7 This was a really hot topic of discussion
8 as it continues to be. Let me show you the results
9 of three trials, two of which are aldose-reductase
10 inhibitors. The tolrestat was a drug, and
11 aldose-reductase inhibitor, and was brought into
12 clinical trial, a fairly well-designed large-scale
13 placebo blind clinical trial.
14 Some of the trial design emanated from two
15 previous ARIs, one known as sorbinil where nerve
16 biopsies had been done before treatment, then after
17 twelve months of treatment and then a second nerve
18 biopsy was done. Actually, morphometry on the
19 sural nerve biopsies were looked at in the sorbinil
20 and a probable positive effect--well, actually, the
21 sorbinil trial did not show a probable positive
22 effect, did it? I am getting that one confused.
23 That is so long ago.
24 DR. BRIL: The morphology was done in a
25 single site. The morphology did show a positive
1 effect that was published.
2 DR. FELDMAN: That was right.
3 DR. BRIL: But the electrophysiology in
4 the multicenter trial did not.
5 DR. FELDMAN: Did not. That's right.
6 That's exactly right. There was a positive effect
7 not confirmed by electrophysiology. That, however,
8 led to the development of using actually paired
9 sural-nerve biopsies as the primary efficacy point
10 in clinical trials with diabetic neuropathy.
11 This was frequently paired with
12 nerve-conduction studies. But, in the trials I am
13 going to discuss, nerve-conduction studies were a
14 secondary endpoint. So, in the tolrestat trial,
15 the primary efficacy was nerve morphometry and
16 sorbitol content, aldose-reductase inhibitors.
17 What they do is they decrease the conversion of
18 glucose to sorbitol. So if you use an ARI, you
19 should measure less sorbitol in the nerve.
20 The secondary parameters were
21 nerve-conduction velocities, a clinical exam and a
22 clinical exam. After a twelve-month analysis,
23 there appeared to be no effect on sural-nerve
24 biopsy on the morphometry and I am going to show
25 you some pictures.
1 However, there was, probably, a mild
2 effect on motor-nerve-conduction velocity which was
3 discussed but the study, itself, was terminated
4 although, in a small subset of patients that
5 continued to get the drug for a longer period of
6 time and actually had a nerve biopsy at a later
7 point, it appeared that potentially the drug was
8 efficacious in those patients.
9 But that drug also had some mild toxicity.
10 So a second aldose-reductase inhibitor,
11 zopolrestat, was then brought to a Phase III
12 clinical trial. Actually, Dr. Arezzo was very
13 involved in this particular compound and might
14 speak to it.
15 The endpoints with zopolrestat were
16 similar to tolrestat in the Phase III clinical
17 trial. So, again, they used nerve morphometry.
18 Unfortunately, they used half the dose they used in
19 the Phase II clinical trial which showed a very
20 robust effect on nerve-conduction studies, and they
21 did an 18-month interim analysis. They did elect
22 to do this trial for three years and there was no
23 effect on sural-nerve morphometry. So the trial
24 was discontinued.
25 Alcar, which Vera mentioned earlier, an
1 excellent trial done by Hoffman LaRoche, used as a
2 primary endpoint nerve morphometry and
3 nerve-conduction studies and there was no effect.
4 So, at this point, there was some
5 discussion in the neurologic world as well, I
6 think, in the FDA whether actually doing bilateral
7 paired sural-nerve biopsies on patients was a
8 necessary primary endpoint.
10 Here is an example of sural-nerve
11 biopsies, and I will go back to that diagram I
12 showed you earlier. Here is a cross section of a
13 nerve and here, on high power, you see multiple
14 large myelinated nerves and you see some small
15 thinly myelinated nerves.
16 These patient samples are still available
17 for study. They are under the care of the
18 University of Michigan and we are glad to give
19 those out with certain requests. There is a
20 protocol that needs to go through with ourselves
21 and the companies.
23 What happened is then these individual
24 nerve biopsies were then quantitated on the
25 computer. The red boxes, for example, are large
1 myelinated fibers and the blue boxes are a
2 different fiber class. So there was very elegant
3 nerve morphometry done on these biopsies.
5 Here is an example of a nerve histogram
6 that was generated from one of the sural-nerve
7 biopsies from the trial. You can see, in this
8 particular fascicle, the definite axon loss
9 compared to what I showed you before. And here is
10 a typical myelinated fiber histogram that was
11 generated. So it is important to know that this
12 has been done.
13 In some ways, maybe the most sensitive
14 measure, it showed no effect but was it the measure
15 or was it more likely, as most people believe, the
16 compound or the small time in which the compound
17 was administered.
19 Now let's go to another aldose-reductase
20 inhibitor that I think will teach us even more and
21 that is the aldose-reductase inhibitor Zenarestat.
22 It really began, interest in it, after tolrestat
23 had failed and zopolrestat was working. It was
24 under clinical trial. Zenarestat occurred and the
25 entry criteria for Zenarestat to enter this study
1 and this is important in relation to what Vera was
2 saying is you needed to have two of three of either
3 symptoms, signs, abnormal nerve conductions in two
4 nerves or abnormal vibratory perception threshold,
6 However, importantly, you had to have both
7 surals present. They could be abnormal but they
8 had to be present and your vibratory perception
9 threshold had to be recordable. They did a Phase
10 II 52-week trial, double-blind placebo-controlled,
11 reported the results in Neurology. It was very
13 I'm sorry; I should tell you also that, in
14 these patients, they also did biopsies. But, along
15 with the biopsies, they did quantitative sensory
16 testing which was very good and they did a quantitative
17 neurologic exam.
19 Here are some of the results. If you look
20 at the sural sensory-nerve-conduction velocity, and
21 this is meters per second, they saw a dose response
22 from their drug in the sensory-nerve-conduction
23 velocity in change at baseline to final. They did
24 a nerve-conduction composite which I will tell you
25 a little bit more about but, again, they saw an
1 improvement in, if you looked at from placebo to
2 their highest drug, baseline to final.
3 That correlated with nerve-fiber-density
4 changes. So here are fibers per meter squared.
5 Here they had a loss. A loss was seen in the
6 placebo group but there was a dose-dependent
7 response with actually a positive effect on fibers
8 per meters squared by morphometry. So this was
9 very exciting.
11 These are the actual data from the paper.
12 I think it is important to see--the change is here
13 in the black. What is important to see about that
14 is the changes are all in the very thinly
15 myelinated or small myelinated fibers. Those would
16 be in the small myelinated fibers, the pain and
17 temperature fibers.
19 So, from that, they decided to do a Phase
20 III clinical trial and this was really somewhat of
21 a breakthrough and to not propose to use
22 morphometry but, rather, say, look, we have seen
23 very good surrogate markers in terms of
24 nerve-conduction studies and in terms of
25 quantitative sensory testing so let's use that in
1 lieu of doing paired bilateral sural-nerve
3 This is the first trial, to my knowledge,
4 that also suggested to use a composite score. So
5 what they suggested to use is a composite rank
6 score for the median four-arm sensory, perineal,
7 motor and sural sensory-conduction velocities, so
8 three sensory-conduction velocities, plus they had
9 a composite rank score of the QST, of quantitative
10 sensory testing, for vibratory and cool perception.
11 You can see what their secondary endpoints
12 were; nerve-conduction velocities, F-waves,
13 amplitudes. This is the Michigan Diabetic
14 Neuropathy score which I actually think has too
15 much motor in it now. When we developed it six or
16 seven years ago, I think we were more naive. A
17 health-related quality of life.
18 Unfortunately, this study, where an
19 interim analysis looked promising, was discontinued
20 because the patients developed probable renal
21 toxicity and increase in creatinine. It is
22 unfortunate because this could have been a trial
23 that could have given us the answer about true
24 composite endpoints as a compositive score for a
25 primary endpoint.
2 The next and final trial that I am going
3 to discuss is the nerve-growth-factor trial. Nerve
4 growth factor was administered subcutaneously in
5 the Phase II clinical trial to 250 patients. I
6 want to make this point because it is so important
7 as we talk about trial design, and that is nerve
8 growth factor is certainly going to be efficacious
9 in small fibers and you are going to be able to
10 measure its efficacy by seeing changes in heat,
11 cold, probably light touch.
12 You would not see its efficacy if you
13 measured motor function, if you measured any
14 large-fiber function, if you measured ankle
15 reflexes. So, it is important. You are really
16 looking here at a drug that should primarily have
17 only a small-fiber function.
18 Here are some of the details of the study.
19 The Phase II clinical study was given
20 subcutaneously and small improvements were seen in
21 sensory symptoms and QST. Unfortunately, this is
22 the neuropathy-impairment score. To show you maybe
23 a little bit of my naivete, I thought there were
24 just going to be a few of us around the table so I
25 brought five copies of this.
1 So I am going to just tell you because we
2 are going to talk about the NIS in a few minutes.
3 But what it is, and I am sure you can't see it, is
4 this part is a cranial-nerve exam. These are all
5 measures of muscle strength. These are reflexes.
6 This is measure of sensation in the hand. Just
7 these last four measures are sensation in the foot.
8 That is the entire NIS. Let me just tell
9 you for a slide in a couple of minutes, five minute
10 from now or three minutes from now, that NIS(LL)
11 that David was referring to, that has kind of
12 somewhat become much the standard right now. What
13 it is is 17 through 24 here on the NIS. That is
14 looking at hip flexion. That is looking at knee
15 extension, ankle dorsiflexion, plantar flexion, so
16 it is a lot of proximal muscle strength. That is
17 important to know.
18 The NIS(LL) looks at the knee reflex and
19 the ankle jerk and then it does the large- and
20 small-fiber function in the foot.
22 But the NIS was used in the NGF trial.
23 What they found was a change in the lower-limb NIS
24 with NGF and it appeared to be dose-dependent
25 change looking at the placebo. However, we now
1 believe, because of the Phase III clinical
2 trial--this is because these patients were
3 unblinded--when you got the NGF, it stung. When
4 you got the placebo, it didn't.
5 It is generally held by both the people
6 who devised the trial as well as the principal
7 investigator that that is likely what happened.
9 There were potentially changes in
10 cold-detection threshold and heat-perception
11 threshold in the NGF study although they were not
12 necessarily as dose-related. But there was a
13 definite difference when compared to the placebo.
14 Now, importantly when you do quantitative sensory
15 testing, you know the patient has to be able to
16 cooperate with you. It is unlike nerve conductions
17 where they can just lay there and you do it to
19 With QST, they have to be able to
20 cooperate. So there was also a question of
23 The Phase III clinical trial, 1,119
24 patients. The primary was a change in the NIS(LL).
25 So I have just told you that that is probably not
1 the best measure to choose because this is a
2 small-fiber--NGF would be a small-fiber function
3 and, on the NISLL, there are only two points that
4 are looking at light touch and pain with a pin that
5 potentially have changed. You see the secondary
6 endpoints were the QST, a symptom and change
7 questionnaire which I also brought, if anyone is
8 interested in seeing, nerve-conduction studies
9 which shouldn't have changed and also using a
10 monofilament which possible could have changed
11 because that is light touch. That was an
12 unsuccessful trial.
13 So what have we learned from all of this
14 and where, really, do we stand?
16 Here are the measures that are currently
17 in clinical trials because I think, as you well
18 know, we are currently doing clinical trials and
19 also these are measures that are being proposed to
20 use in clinical trials. I should say that these
21 are clinical trials looking at the drug that really
22 going to affect the pathogenesis of diabetic
23 neuropathy. This is not talking about, obviously,
24 a drug for pain.
25 The current test is the NIS(LL)+7. What
1 is the +7? It is vibratory-perception threshold.
2 It is RR variability with deep breathing, so it is
3 actually a measure of autonomic function and five
4 nerve conductions; perineal--that is the motor
5 nerve in the leg, looking at its size, its
6 conduction velocity and its distal latency; the
7 tibial nerve, another nerve in the leg, looking at
8 its distal latency; and then the sural, which is
9 the sensory nerve in the leg, looking actually at
10 its amplitude.
11 Importantly, what this +7 means is if you
12 have an abnormality in one of those tests and it is
13 between the 95th and the 99th percentile abnormal,
14 you get one point. If you are greater than 99, you
15 get two points. Then what happens is you really
16 get an added composite score.
17 As David told us so nicely, in the
18 Rochester diabetic cohort, there was, in one year,
19 a change of 0.35 in those patients that they
20 believe do not have diabetic neuropathy while, in
21 those patients who do have neuropathy in this
22 composite score yearly, they saw a change of 0.85.
23 They also published a very nice paper that
24 shows that this NIS(LL)+7 correlates with other
25 microvascular complications, particularly--well,
1 obviously, the two others, retinopathy and
2 nephropathy. So it is a composite score looking at
3 motor-nerve-conduction function, autonomic
4 function, motor strength, reflexes and sensory
5 examination in the lower extremity.
6 Here are some of the salient references.
8 I think that, importantly, as I have
9 pointed out, the NIS(LL) is primarily a motor test
10 and when you actually go into Dr. Dyke's excellent
11 papers, he says the same thing, that when you look
12 at the NIS(LL) and see what is actually abnormal in
13 patients with diabetic neuropathy, what you find is
14 that reflexes and vibratory sensation are what is
15 abnormal and there are essentially no motor
16 abnormalities on the NIS(LL).
17 So you can imagine, because of the
18 multiple points it has, if you are just primarily
19 looking at sensory, how you can get a confounding
20 effect in a clinical trial because in inter- and
21 intraobserver variability and testing so many
22 points that really are not relevant for your
24 This idea that reflexes and vibration
25 sensation are most frequently abnormal has been
1 corroborated by a large study by Fedele and
2 colleagues in 2,300 patients. What Dr. Dyke also
3 has reported from the Rochester Diabetic cohort is
4 the motor-nerve conductions of the lower extremity,
5 the perineal nerve and the sural snap, are the most
6 frequent abnormal nerve conduction.
7 So if you are just going to do two, you
8 would do the perineal and the sural. Fedele also
9 showed that. Dr. Dyke has also shown, in the
10 Rochester Diabetic cohort, that vibration
11 perception threshold is easier to measure, more
12 reliable, and usually more often abnormal than
13 cold-perception threshold.
14 I believe that is a question whether RR
15 variability is a viable clinical endpoint. It
16 seems like, if you are really--you know, in some
17 ways, if you are primarily looking at somatic
18 sensory-motor peripheral neuropathy, I am not sure
19 you want to confound your measurement by looking at
20 the RR interval, which the NIS(LL)+7 does.
22 So I decided I would propose a clinical
23 composite score. This is, again, the--these are
24 the last two slides of my talk. What I would like
25 to propose based on, really, having done many of
1 these studies and, primarily, though having
2 reviewed the literature is a NIS(LL) but minus
3 Questions 17 through 22 or maybe 14 through 24.
4 That is really getting rid of testing hip strength
5 and quadriceps strength, all this proximal
6 strength, so that the clinical exam, really, then
7 becomes focused on what one sees in the disease
8 and that sensory loss in the lower extremities,
9 ankle-reflex loss.
10 Potentially, we could look at very distal
11 weakness. If we did, we would keep in Questions 23
12 and 24. That is toe extensor and flexor. I
13 believe a composite nerve-conduction-velocity score
14 is a good idea. I think the perineal motor nerve
15 appears to be the one that has been used the most,
16 most reliable in multiple trials, and also the
17 sural-nerve amplitude, although there is more
18 variability in measuring that, as David and Vera
19 and Dr. Arezzo also will tell us, that also appears
20 to be a reliable measure.
21 Quantitative sensory testing? BPT is more
22 reliable and reproducible than CPT but we also need
23 both because we need a measure of large and small
24 fibers. Secondary endpoints, I think, should be a
25 symptom questionnaire, maybe a quality-of-life.
2 I think that I would like to summarize by
3 saying that our experience over the last twelve
4 years, in my mind, clearly shows that drug efficacy
5 in DPN and diabetic polyneuropathy cannot be judged
6 by just one single parameter. It is just really
7 too complicated a disorder, as I have tried to
8 portray for you today. I believe what we need to do
9 is develop a good composite score.
10 I am happy to take any questions.
11 DR. KATZ: Thank you very much, Dr.
13 Dr. Dworkin, you are first.
14 DR. DWORKIN: It seems to me that
15 treatment responsiveness is one aspect of
16 establishing validity.
17 DR. FELDMAN: Right.
18 DR. DWORKIN: But to go back to your
19 original definition of diabetic neuropathy, you
20 emphasized foot ulceration and amputation.
21 DR. FELDMAN: Right.
22 DR. DWORKIN: So my question is do any of
23 the measures that we have heard discussed this
24 morning in prospective studies establish themselves
25 as risk factors for either foot ulceration or
1 amputation, which I might want to propose is the
2 gold standard for a validity of one of these
3 surrogate endpoints.
4 DR. FELDMAN: There is some very nice work
5 from Andrew Bolton in England who has looked at
6 vibratory-perception threshold over time and then
7 the development of foot ulceration. He has shown a
8 correlation between decreased vibratory-threshold
9 sensation, VPT, over--this was a very long
10 study--until, essentially, VPT is absent. And then
11 the patients develop foot ulceration.
12 As you probably know, a diabetic has about
13 a 15 percent chance in his or her lifetime to
14 develop foot ulceration. So the problem, of
15 course, with using foot ulceration as an outcome is
16 that we are talking ten, twenty, twenty-five years
17 into the disease. That is really end stage.
18 I do think that a feeling that we all have
19 in this area is what we want to do, and I threw
20 this on as it would be nice to treat patients early
21 in their disease. So I really do think that we do
22 need entry criteria which I didn't think I have
23 time to talk about.
24 But, in my mind, our aim ought to be to
25 halt progression. I am less likely to think we are
1 actually going to show improvement. I know that is
2 not necessarily a popular view to take, but I
3 think, hopefully, if we could just halt the
4 progression of what is really kind of a relentless
5 progressive neurodegenerative disease.
6 What I would propose is we enter patients
7 who are very early in their disease but using a
8 fairly protracted time course. I would say we
9 probably need at least a three-year study.
10 DR. KATZ: Dr. Bril next please.
11 DR. BRIL: Andrew has also extended that
12 work using some electrophysiology to look at
13 prediction of foot ulceration. So these surrogates
14 are now being tied more and more strongly to
15 long-term neuropathy outcomes.
16 DR. KATZ: Dr. Farrar, did you have a
18 DR. FARRAR: Since clinical trials are so
19 dependent on selecting the right population, I
20 wonder if you could comment on your sense about
21 whether the criteria used to decide whether or not
22 people had sensory-reflex or symptom scores where
23 appropriate and how, actually, that was decided.
24 DR. FELDMAN: So this would be in the
25 Zenarestat trial or any trial--whichever trial I
1 would like to talk about?
2 DR. FARRAR: Are they different?
3 DR. FELDMAN: Oh; they are different.
4 They are very different. Currently, there is a
5 drug, alphalipoic acid, where patients can have
6 actually a relatively neuropathy, no obtainable
7 sural responses and very poor vibration-perception
8 threshold and be entered into the trial.
9 I think that most trial, however, and,
10 again, I want to defer to my colleagues if they
11 would like to add anything, most trials have tried
12 to use patients who have what we would say mild
13 neuropathy, maybe at the extreme moderate, so that
14 sural amplitudes needed to be present and they
15 needed to be measurable reproducibly.
16 But if the surals were normal, then you
17 needed to have another abnormal measure to go with
18 it. So patients who had mild abnormalities in
19 their sural nerves and a mild decrease
20 quantitatively in vibration-perception threshold or
21 cold-perception threshold, in my mind, would be the
22 ideal patients to enter.
23 So the idea is that you if have got nerves
24 of wood, if all the nerves are dead, there is not
25 going to be a Lazarus effect which is what was
1 discussed in the mid-'80's with the ARIs. So I
2 think we need to see early patients because the
3 disease is going to progress. If you want to halt
4 the progression, you have got to be able to monitor
5 the progression. So you have got to be able to see
6 the surals go down, the perineals get slower, the
7 vibration-perception threshold change.
8 DR. FARRAR: Let me just follow up with a
9 quick question.
10 DR. FELDMAN: Yes; please.
11 DR. FARRAR: Very specifically, how did
12 they decide if the reflexes were less or not? As a
13 neurologist, I have trouble doing that in most of
14 my patients.
15 DR. FELDMAN: So reflexes were graded. In
16 the NIS, and in most of these scores, the reflexes
17 are graded simply as present, present with a
18 gendracic maneuver or absent. So it is a very
19 straightforward thing.
20 DR. CORNBLATH: That's not right.
21 DR. FELDMAN: What; in the NIS. The most
22 recent NIS? OC?
23 DR. CORNBLATH: No; the NIS was always
24 normal, reduced or absent.
25 DR. FELDMAN: Oh; I'm sorry. The NIS was
1 normal, reduced or absent.
2 DR. CORNBLATH: So that was a choice and
3 that was determined that those three that
4 neurologists could rely upon determine--
5 DR. FELDMAN: Thank you, David. That's
7 DR. CORNBLATH: The gendracic has nothing
8 to do with it.
9 DR. FELDMAN: That was in the Zenarestat
10 study, I'm sorry, that they used that. But one
11 thing that Peter Dyke did evolve over time, which I
12 think is important, is that NDS, the Neurologic
13 Disability Score, that had, for example, in
14 sensation, I think five choices and, in reflexes,
15 four or five choices. What he did is he simplified
17 When he did do that, then, within the Mayo
18 Clinic, several individual physicians would examine
19 the same patient and he found a great deal less
20 variability between examinations when he simplified
21 his scores. And we would all agree, of course.
22 DR. KATZ: Dr. Cornblath.
23 DR. CORNBLATH: At the risk of touting my
24 own horn, Eva, we have developed a scale, as you
25 know. It is called the Total Neuropathy Scale. I
1 think one of the things Cynthia and I were talking
2 about in between was what are some of the
3 difficulties with the NIS.
4 DR. FELDMAN: Right.
5 DR. CORNBLATH: I think one that you
6 alluded to but didn't directly mention is this very
7 important issue that these are length-dependent
8 neuropathies. As a result, if you have neuropathy
9 up to your ankle, the likelihood, as you suggested,
10 of showing a drug that will change sensory function
11 at the great toe, which is what the NIS looks at,
12 is highly unlikely. It would be, as you said, a
13 Lazarus effect.
14 So what we did, in designing ours, was to
15 us this opportunity of length-dependent to
16 essentially assign points from a 0 to 4 scale
17 depending on the length. So one of the very
18 serious criticisms of NIS is this dependence upon
19 the great toe and subsequently, then, the
20 opportunity to change function at the great toe
21 which you and I think is highly unlikely during the
22 course of a clinical trial.
23 That includes changing the vibratory
24 threshold or cooling threshold which is also
25 measured at the great toe during a trial. So what
1 we did in TNS, as you know, is to change the great
2 toe to a length issue saying it is either bad up to
3 knee, up to the ankle, to the toe or normal. So
4 this is another method to get away from one of the
5 many criticisms of NIS.
6 So there are other composite measures
7 around. You have a composite measure.
8 DR. FELDMAN: Yes; I did not talk about
9 mine, either.
10 DR. CORNBLATH: I think that one of the
11 issues for discussion is are we going to be left,
12 at the end of the day, fooling with the NIS and
13 trying to alter it to fit what we want or, in fact,
14 does the NIS have such severe limitations that, in
15 fact, it can't be used in this disease, again
16 because of the biology of our understanding of what
17 is possible.
18 DR. FELDMAN: Those are excellent points,
19 David, and I apologize, really, for not discussing
20 your scale or my scale or other scales. The scales
21 I chose to discuss were those that are currently in
22 clinical trial for diabetic neuropathy. And Vera
23 has a scale also. So we all have scales.
24 The thing that is common about our
25 scales--the scale team--is that motor strength is
1 deemphasized and it is emphasized in the NIS. I
2 have now pounded this to death but, also, as David
3 so nicely pointed out, these scales also look at a
4 length-dependent sensory loss which I think is very
6 David's composite score also has a
7 component of--you also have nerve conductions in
9 DR. CORNBLATH: We have nerve conductions,
10 a simple vibratory threshold. We have large- and
11 small-fiber function and we don't have yet, but we
12 could easily take out one of those and put in
13 something else for, again, a specific biological
15 DR. FELDMAN: So it is a good composite.
16 I think your score is a very good composite score.
17 You have used it in a trial of suramin toxicity,
18 haven't you?
19 DR. CORNBLATH: So far we have used it in
20 monitoring in three chemotherapy things for
21 toxicity. We have not had the opportunity to use
22 in longitudinally.
23 DR. FELDMAN: The idea, though, that I
24 think you are hearing is that prospectively, when a
25 pharma comes to you, what we would suggest is a
1 composite trial that emphasizes sensory loss, that
2 has a quantitative component and has a
3 motor-nerve-conduction component.
4 DR. KATZ: Dr. Rendell, you were next, if
5 you still have a question. Oh; sorry. Dr. Shafer?
6 DR. SHAFER: You mentioned, in passing,
7 the work done by DCCT. I happen to have the DCCT
8 database in front of me. They did a ton of testing
9 and it appears that they actually did it yearly,
10 but perhaps it was not done as frequently as
11 yearly, on sural, perineal, median nerve, both
12 motor and sensory amplitudes and conduction
14 Of course, it is such a huge study and so
15 well controlled, would there be any point in going
16 back to that database and trying to ask whether or
17 not one can develop yet another scale from it?
18 DR. FELDMAN: I brought, actually the nice
19 person who let me in the room--I actually brought a
20 suitcase full of papers in case. Again, I thought
21 there were going to be five of us sitting around a
22 table. So I have all the DCCT papers and there was
23 a paper done by the DCCT working group published in
24 Neurology in 2000 where they looked at all the
25 nerve conductions in detail and made associations,
1 et cetera.
2 I have got that paper there. What I think
3 would be more useful would be for us to restudy
4 those people now. We would actually have a really
5 good well-defined population and really understand
6 over a ten-year period what happens to nerve
7 conduction in a group that is still relatively
8 well-controlled that is interesting and a group
9 that is less well-controlled. That is really the
10 way EDIC has fallen out. But I will give you those
12 DR. KATZ: Dr. Woolf, you were actually
14 DR. FELDMAN: But not much was out of it,
15 though, except for perineal motor-nerve conduction,
16 I should say.
17 DR. WOOLF: A key concern for us all here
18 is why do trials fail. We have heard either the
19 drug or the outcome measure. I think the nerve
20 growth-factor trial is a classic in that case
21 because the outcome measures did not measure the
22 time to--conduction velocity does not measure
24 The testing, with the greatest
25 respect--light touch is not a small-fiber test. It
1 is an a-fiber. It may not be the large
2 proprioceptors, but they are large myelinated
3 fibers and they are not NGF-responsive.
4 DR. FELDMAN: No; I stand corrected.
5 DR. WOOLF: The morphometry, again, is
6 large fiber. You need electron microscopy.
7 DR. FELDMAN: Right.
8 DR. WOOLF: So, by all those three
9 standards, the composites exclude the very fibers
10 that are being targeted by the drug and so that
11 trial will fail before you even start it.
12 DR. FELDMAN: Well, it did fail, as you
14 DR. WOOLF: I know. But you could predict
16 DR. FELDMAN: I know. And it was
17 predicted by many. As many of the neurologists in
18 this room know, it was predicted to fail. So that
19 was a frustrating point. But you are right.
20 DR. WOOLF: So you have made a very
21 convincing case how the composite studies were
22 geared towards motor weakness which is no longer
23 relevant. I would say that any composite measure
24 has to include small-fiber measures, however
25 difficult they are, because, otherwise, you are
1 going to have the same problem.
2 DR. FELDMAN: Would you be happy with
3 cold-perception threshold, then, from quantitative
4 sensory testing which appears to be--you know, in
5 these large-scale clinical trials when we want to
6 enter 1000 patients, that is probably truly doable
7 and relatively reproducible. I think some of the
8 more sophisticated electrophysiology probably
9 really isn't doable.
10 DR. WOOLF: It is certainly better than
11 nothing but I would like at least one other
12 objective measure.
13 DR. KATZ: Dr. Woolf, just pursuing that
14 further, do you have a specific thought about what
15 would be the best objective measure for small-fiber
16 function in such clinical trials?
17 DR. WOOLF: Function is very difficult, I
18 accept. But morphometry, you can do electron
19 microscopy. You can, actually, count the number of
20 unmyelinated fibers and there are now unmyelinated
21 fiber markers as well.
22 DR. FELDMAN: I think that the neurologic
23 community really came out somewhat in force and I
24 think backed by our endocrinology colleagues, kind
25 of together, that probably bilateral sural-nerve
1 biopsies were not necessary in this disorder and
2 too invasive
3 Also out in my suitcase of papers are two
4 papers looking at the outcome of patients with
5 bilateral sural-nerve biopsies comparing diabetic
6 with nondiabetic patients. There does appear to be
7 more likely to have persistent pain. So there is a
8 morbidity to a sural-nerve biopsy in a diabetic
10 Interesting, though. We do have all those
11 samples and no one has any interest in looking,
12 doing EM, on the small fibers. We have over 1000
13 pairs of sural-nerve biopsies.
14 DR. KATZ: Dr. Hertz, did you have a
16 DR. HERTZ: I just wanted to ask if
17 somebody could address, maybe at this point, the
18 use of F-waves.
19 DR. FELDMAN: I am happy to, or David, do
20 you want to? Or I can. It doesn't matter.
21 DR. CORNBLATH: For this question, none.
22 There is no value. And, in general, they are just
23 going to be another measure of long latency nerve
24 function. They will parallel, or they should
25 parallel, what is seen in perineal motor
1 conduction. They are a little bit more reliable in
2 terms of a multicenter trial but, in terms of the
3 kind of information that they give you
4 intellectually, it is no different.
5 DR. KATZ: Just to be clear, you are
6 saying that because they are not relevant to
7 small-fiber function.
8 DR. CORNBLATH: That's correct. Can I try
9 to answer this question and this is something I
10 know Michael is going to talk about, I think one of
11 the issues when we think about looking at
12 small-fiber change, and I will use it as a global
13 sense of the small-fiber change, I think we have to
14 go back to this issue of what do you expect to
16 If you look at, as Eva suggested,
17 cooling-detection thresholds, again, you are
18 talking about the great toe and the question is, do
19 you have a drug that could do this. If you don't
20 have a drug that could do it, change at the great
21 toe, it is a waste of time.
22 If you are talking about morphometry, the
23 neurologic community is not going to allow
24 bilateral sural-nerve biopsies. I believe they are
25 unethical at this point in time. But the
1 skin-biopsy technique or a technique that would
2 allow you to do quantitative sensory testing at
3 another site with, for example, the Medoc device or
4 another device that could be moved to a level on
5 the skin where you would like to see
6 sensory-function change, are going to be the wave
7 of the future.
8 That is what we can do with skin biopsy
9 doing morphometry, but you could do the same thing
10 with a QST device if you could move it along the
11 skin. A number of these are either available or in
13 DR. FELDMAN: You know, that is an
14 excellent point and I just echo it a thousand
15 times. I don't know, and Michael is going to
16 educate us, if we can do--I know that Hopkins has
17 done nice skin biopsy, using skin biopsies, in HIV
18 drugs. But if we are at the point where we could
19 use it in a large-scale trial for diabetes, it
20 would be superb. It would be a superb measure for
21 small-fiber function.
22 DR. KATZ: Just to push the QST point a
23 little bit further, is there a validated procedure
24 or any experience with using QST in such a way as
25 to float upwards from the big toe?
1 DR. FELDMAN: No--oh, I'm sorry David. I
2 was going to say no. All the validation with the
3 Case IV QST has been done on the great toe or the
4 forefinger. David probably knows more about the
5 Medoc than I do, though.
6 DR. CORNBLATH: There are devices
7 available that you can move. You could move Case
8 IV. It would require a little bit of change in the
9 sort of device, itself.
10 DR. FELDMAN: The design.
11 DR. CORNBLATH: You could move one of the
12 devices anywhere and, as long as you did some
13 studies that would show that you do it this way in
14 every person, the same stuff you do for regular
15 QST, you would have no trouble.
16 But it could be done easily. Is that
17 right, Michael?
18 DR. FELDMAN: I don't think it has been
19 done; is that correct?
20 DR. CORNBLATH: It has not been done on a
21 giant scale. Individuals have done it.
22 DR. FELDMAN: I think it is an excellent
24 DR. CORNBLATH: I think Mike has data on
25 this point. He talked earlier about doing it on
1 the areas of postherpetic neuralgia.
2 Michael--well, the two Michaels, Michael Polydefkis
3 and Michael Rowbothom have both done it.
4 DR. FELDMAN: I would just say that,
5 again, we have got the DCCT patient population.
6 They want to do more for us. If we had the
7 funding, we could do this on that population and
8 couple it with nerve-conduction studies.
9 DR. KATZ: Ms. Delph, you were next.
10 MS. DELPH: You have suggested a number of
11 primary endpoints which are basically surrogate
12 markers. I think that it is important for us to
13 have a good idea of how useful those surrogate
14 markers are in terms of translating into clinical
15 benefit. How useful are the ones that you have
16 suggested, because if you don't have a good idea of
17 what kind of clinical benefit you are looking at,
18 then it is really difficult to weigh the
19 risk-benefit ratio and efficacy versus toxicity.
20 DR. FELDMAN: So the clinical benefit that
21 I would aim for in a clinical trial would be the
22 halt the progression of the disease. So we can go
23 back to the Rochester diabetic neuropathy study
24 which is shown in the group of patients with
25 neuropathy, just progression on a yearly basis, of
1 the NIS(LL)+7 and all other parameters they have
2 looked at of nerve function and go to the DCCT
3 which shows a progression--or, well, you can look
4 at it two ways but essentially those patients with
5 conventional therapy had a loss of
6 motor-nerve-conduction velocity within a five-year
8 Those measures, those surrogate measures,
9 do correlate at least with disease severity or
10 intensity as monitored by clinical examination and
11 more extensive electrophysiology testing, not by
12 symptoms. So notice I haven't been talking about
13 symptoms. I would be glad to talk about symptoms,
14 but not by symptoms in these measures. So the
15 measures I chose, one was a measure of where the
16 patient stands clinically, so clinical efficacy,
17 what is their sensation now in the great toe.
18 David makes a good point that that may not
19 be ideal, but what is their sensation, let's say,
20 in the foot or ankle region. What is a
21 quantitative measure of their sensation,
22 vibration-perception threshold. What is a
23 motor-nerve conduction velocity, for example, and
24 maybe a sural-nerve-evoked amplitude.
25 What I would maintain is that you take
1 that composite measure at Time 0, and if you have a
2 successful intervention, I would maintain that that
3 composite would stay the same. Maybe, if you hit
4 the home run, you would get some improvement. In
5 those patients treated with placebo, as we know
6 this disease does, you would see the relentless
7 progression of increased abnormalities in the
8 components of that composite measure over time.
9 The reason to use more than one measure
10 again is this isn't just a simple disease. You
11 have got large fibers and small fibers so you need
12 to be able to measure both.
13 MS. DELPH: I don't think you have
14 answered my question.
15 DR. FELDMAN: Sorry.
16 MS. DELPH: At the end of the day, it is
17 very nice to show improvement in conduction
18 velocity and so on, and to show improvements in
19 different size fibers and so on. But when you are
20 looking at approving a drug, what is important is
21 how does that drug, for example, affect your
22 likelihood of developing ulceration. How much pain
23 relief are you likely to get from it?
24 DR. FELDMAN: I understand.
25 MS. DELPH: From those endpoints, you are
1 talking about, it is going to be very hard, I
2 think, without objective data to weigh the
4 DR. FELDMAN: Let me take a step back. I
5 understand better. What these data reflect, these
6 abnormalities in nerve conduction and quantitative
7 sensory testing is a loss of axonal fibers. I
8 showed you that pathology at the very beginning.
9 So there is data that correlates axonal fiber
10 density with motor-nerve conduction, amplitude and
11 vibration-perception threshold. Again, that is
12 work by James Russell and Peter Dyke in, actually,
13 the late 1980s.
14 There is a very nice correlation between
15 loss of myelinated fibers and loss of these
16 parameters. These parameters are simply our way of
17 seeing how many nerve fibers there are. Then step
18 2 is, we know that, as you lose nerve fibers in the
19 foot, those are the feet that are going to develop
21 So these are surrogate markers to look for
22 nerve-fiber loss, and it is nerve-fiber loss that
23 eventually is going to cause ulceration. Does that
24 answer your question? No? I'm so sorry.
25 MS. DELPH: What I am asking basically how
1 do you actually translate all of these surrogate
2 markers into an adequate measure of clinical
4 DR. FELDMAN: Okay. The adequate measure,
5 in my mind, of clinical benefit is halt of
6 progression of the disease because if you halt the
7 progression of the disease and you have measurable
8 electrophysiologic parameters and measurable
9 sensation, then you are not going to develop an
10 ulcer. Those patients ulcers who lose all those
11 parameters as they lose axons.
12 DR. KATZ: I think the question is that
13 there is a philosophy that there should be a search
14 for a patient-centered outcome at the end of the
15 day and that nerve conduction to the patient, they
16 don't know what their nerve conduction is. They
17 know if they got symptoms, if they have trouble
18 walking, if they are developing an ulcer, that sort
19 of thing.
20 So I think the question is how does one,
21 in a clinical-trial program eventually connect
22 surrogate marker to the patient-centered clinical
23 outcome or is there a need to make such a
24 connection? Is that a fair translation?
25 MS. DELPH: I think, very simply, if I can
1 rephrase it, is if you get an improvement in
2 conduction velocity or amplitude of X amount, that
3 can translate into a decrease in your likelihood of
4 getting an ulceration or this level of--on average.
5 DR. FELDMAN: So the data would be
6 available to look at if you have a preserved
7 nerve-conduction velocity, that means you have got
8 this many myelinated fibers and you are very
9 unlikely to get an ulcer. You can translate that,
10 then, to having really no recordable nerve
11 function, and on having no recordable or no
12 visualized axons in a nerve biopsy and developing
13 an ulcer.
14 But there is that jump there because they
15 are surrogate markers of axonal function.
16 DR. KATZ: Dr. McCormick, a comment from
17 you on this?
18 DR. McCORMICK: I think it may be helpful
19 to think of your question in the context of other
20 kinds of drugs that prevent disease even though, in
21 this case, we are not preventing disease, we are
22 preventing the ultimate course of the disease, for
23 example, cholesterol-lowering agents or
24 antihypertensive medications where you may not
25 directly see the long-term effects of the change
1 but there is an anticipation and, in fact, data to
2 suggest that certain complications of the disease
3 will be prevented.
4 So I think that is what we are looking at
5 here. The patients may not notice that they are
6 not getting worse but we are trying to collect
7 evidence that will allow us to draw that
9 MS. DELPH: I understand that. But if you
10 are looking at a cholesterol-lowering agent and an
11 agent can lower your cholesterol by X amount, you
12 have an idea of how much it is likely to lower your
13 risk of a cerebral-vascular accident or a
14 myocardial infarction.
15 What I am saying is if you have X change
16 or if you have a quantifiable changes in these
17 surrogate markers, in order to adequately weigh
18 risk-benefit, the drug may produce nephropathy or
19 different complications. What do you weigh, a
20 complication and likelihood of developing
21 nephropathy or something versus X amount of
22 improvement in conduction velocity?
23 DR. FELDMAN: I think that, as we said--I
24 understand now better what you are trying to ask
25 and it is a very good question. There is a large
1 study out of the Veterans Hospital looking at the
2 morbidity of diabetic neuropathy and also
3 development of ulcers. It is actually a very
4 highly morbid condition, so it is a high degree of
5 patient morbidity.
6 And then it is the most frequent cause of
7 hospital admission for a diabetic patient. That is
8 an interesting and well-established fact. It is
9 neuropathy and a nonhealing ulcer. As I mentioned
10 earlier, 15, to some people say, 20 percent of all
11 patients require amputations. So those are really
12 the very end markers for all nerve-fiber loss.
13 I think what you would have to do is then
14 measure the risk-benefit and the benefit would be
15 if you could halt nerve progression. These are all
16 the consequences of relentless nerve progression.
17 So that would be what you would want to weigh,
18 those two things. What we don't know is why some
19 patients don't develop microvascular complications
20 although they are poorly controlled.
21 DR. KATZ: One more comment on this issue
22 from Dr. Bril and then we will go on to the next
24 DR. BRIL: There is work with the
25 surrogates that show that if you have a certain
1 level of function in the surrogates, you are less
2 likely to have foot ulcers and then you are much
3 more likely to have. So if you have something that
4 holds you in a low level, then the projection is
5 that it will work in the long level.
6 That is not to say that a new drug
7 shouldn't be tested once you have the effect on the
8 surrogates, that it shouldn't be tested in
9 longer-term studies. But the investment needed to
10 show prevention of foot ulceration is a five-year
11 trial. A lot of companies won't commit that kind
12 of resource until they have some promising evidence
13 in shorter studies.
14 I guess my comment on the scale question,
15 clinical scales, too--I mean, even the clinical
16 scales are made up of how patients perceive
17 sensation. My own scale has symptoms in it, for
18 good or bad. So these scales are based directly on
19 the patient. They are not nerve conductions. They
20 are not QST. They are how the patient perceives
21 sensation and symptoms.
22 I think the basic thing we all agree on,
23 although we may not agree on the right scale, we
24 agree that scales summarizing clinical findings are
25 important plus or minus other endpoints that we may
1 want to put in there. But, even the NIS, and Peter
2 Dyke who started it all--I mean, we agree that they
3 are all valuable.
4 The question is exactly what you should
5 have in a particular scale. I am not sure we can
6 determine that but there is a consensus that I can
7 see that we feel that the clinical examination
8 needs to be reduced to some kind of number that you
9 can follow even though you might follow elements.
10 So, at the end of the day, we are not
11 saying that this drug will reduce numbness or this
12 drug will reduce pain, but it will reduce a
13 composite score, a composite clinical score plus or
14 minus other stuff, and we all seem to be convinced
15 of that from what I have heard.
16 DR. KATZ: I am going to make a few
17 enemies around the table who still have persistent
18 questions and, despite that, go on to the next
19 speaker. We have about an hour this afternoon
20 devoted primarily to trying to better understand
21 the meaningfulness of clinical outcome measures.
22 So please accept that your comments in this issue
23 will not be lost.
24 So Dr. Michael Polydefkis will speak to us
25 about the use of skin biopsies in the evaluation of
1 peripheral neuropathy and neuropathic pain.
2 Skin Biopsies in the Evaluation of Peripheral
3 Neuropathy and Neuropathic Pain
4 DR. POLYDEFKIS: Good morning.
6 There has been a lot of talk about
7 small-caliper nerve fibers and I am going to talk a
8 little bit about skin biopsy which is a technique
9 that has evolved over the past decade or so to look
10 at this class of nerve fibers.
12 So, as an outline, I am going to give some
13 background as to where this technique came from. I
14 will review the technique, itself, and how it has
15 evolved into a clinical diagnostic test. I will
16 review some of its use in clinical trials and how
17 it has been used to study diabetic neuropathy and,
18 potentially, to develop some novel outcome
21 So, first, small-fiber neuropathy sort of
22 came to light of most prominence actually in a
23 cohort of HIV patients who had prominent symptoms,
24 most notably pain, in their feet and distal legs
25 yet there was this paradox in that they were
1 relatively normal on exam. They had normal
2 strength, reflexes, normal proprioception and
3 normal nerve conduction and EMG test results.
4 They were clinically felt to have a
5 small-fiber neuropathy but there was a relative
6 absence of clinical tests to evaluate them. So
7 that is where the beginning of looking into the
8 skin for nerve fibers evolved and actually has its
9 roots, again, in the Mayo Clinic where they
10 investigated nerve fibers in the skin.
12 This is a sural nerve. I show it to you
13 just to emphasize that the nerves that I am going
14 to be talking about, you actually can't even see
15 here. The red arrow depicts a large myelinated
16 fiber, the yellow a small myelinated fiber. But
17 the class of fibers that I will be talking about
18 are predominantly the small unmyelinated fibers
19 which are C and A delta fibers which, again, you
20 need electron microscopy to see.
22 So, again, there is this size dichotomy
23 but also a functional correlate. As we said,
24 large-fiber nerves convey information about balance
25 and pressure while small fibers convey information
1 related to temperature, heat, pain and pain
2 sensation. Their loss or dysfunction really
3 correlates with pain. So pain is the hallmark of
4 patients with a predominantly small-fiber
7 In terms of clinical tests, we have heard
8 a lot about nerve-conduction tests. They are
9 really a large-fiber test. You also have
10 sural-nerve biopsy which, as we have heard, can
11 also be used to measure large-fiber nerve
12 morphometry. It is important to emphasize that
13 small-caliper nerve fibers are invisible to
14 nerve-conduction velocity testing. That is a point
15 that has been made several times. Quantitative
16 sensory testing can be used to measure
17 small-caliper nerve-fiber function but, again, it
18 is a psychophysical measure and it is important
19 that the stress--I believe that it really requires
20 vigilant patient cooperation and attention.
21 So, in the battery of QST tests they have
22 done using the Case IV device, it is at least a
23 forty-five minute procedure.
24 QSART, Quantitative Sudomotor Autonomic
25 Reflex Testing, is a measure of autonomic
1 small-fiber nerve function and is a sophisticated
2 device which I personally don't have much
3 experience with. Also, sural-nerve biopsy can be
4 used to evaluate small-fiber nerves but, as has
5 been pointed out, you have to go to electron
6 microscopy, another level, and it is quite
8 And now we have skin biopsy which I am
9 going to talk about.
11 We have learned that epidermal nerve
12 fibers are predominantly sensory and they represent
13 free nerve endings without Schwann-cell
14 ensheathment. There are multiple neurological
15 conditions with prominent small-fiber nerve
16 involvement and many of these have been studied
17 with skin biopsy.
19 The technique, itself, is pretty
20 straightforward. We use a 3-millimeter punch
21 biopsy. This is what a typical biopsy, or four
22 biopsies, look like. Typically, we shave the area
23 but the biopsies heal by a process of granulation.
24 There are on sutures involved. The risk of
25 infection is nominal, on the order of one-half of
1 one percent including many diabetics.
3 This is what biopsies can look like at two
4 months. There is a mild scar.
6 It is not uncommon at eight months to
7 really be hard pressed to see any evidence of a
8 biopsy although, in fairness, many people do have a
9 mild scar that persists.
11 I know it is close to lunch but if you
12 think of skin biopsy as a loaf of bread, what we do
13 is we section it and, from each biopsy, we get
14 fifty-five sections, on average. We use 50-micron
15 sections so you should get sixty sections.
16 Clinically, at random, we select four slices, and
17 that has been shown to give a representative sample
18 of the whole biopsy.
20 So, if this were raisin bread, by getting
21 four sections, we get a representative number of
24 Then we look at individual sections.
1 This is an example. This is the surface
2 of the skin and this is a section that is stained
3 with a Panex solo marker PGP9.5 and these are the
4 nerve fibers. The red line depicts the
5 dermal-epidermal junction. So when a fiber crosses
6 this junction, it is designated an epidermal nerve
9 So, as we have said, most neuropathies,
10 including diabetic neuropathy, is a
11 length-dependent process and so we typically take
12 biopsies from three locations; proximal thigh,
13 distal thigh, distal leg. I will give you the
14 example of how this is used to define
15 life-dependent small-caliper neuropathy.
17 This is a very proximal site, actually the
18 back. But this is the normal, nonneuropathic
19 individual. You can see there are plenty of nerve
20 fibers. It is very well innervated which is what
21 we would expect at a proximal site. Even the
22 person with neuropathy, the epidermis is well
23 innervated although, qualitatively, there are some
25 Again, this is what we would expect to
1 see. At a proximal site, we would expect to see
2 innervation even in the neuropathy individual at a
3 proximal site.
5 Going distally, we are now at the thigh,
6 we still have plenty of nerve fibers in the healthy
7 person. In the neuropathic individual, there are
8 still fibers but, again, morphologically, I think
9 they are more abnormal, that we have swellings or
10 fragmentations here. That is what we have taken
11 over the years to be a predegenerative change.
13 Now, at the most distal site, the ankle,
14 and, again, the normal individual has preservation
15 of innervation but, in the neuropathic person,
16 there is a complete absence of epidermal
17 innervation and the single fiber we see in the
18 dermis, again, is fragmented. It appears to be
19 degenerating. So that is how we have used this
20 technique to define a life-dependent small-fiber
23 Again, one of the strong suits of this
24 technique is that it is quantifiable so, with a
25 computer algorithm, we can measure the precise
1 distance and we can counts these fibers
2 specifically to arrive at a density of nerve fibers
3 per millimeter.
5 The normative range has been established.
6 So these are densities of normal people. I think
7 it is important to point out that there is a
8 healthy range of what is normal, but using the
9 fifth percentile as the definition of abnormal, it
10 is useful clinically with a diagnostic efficiency
11 and specificity of 88 and 97 percent.
13 I think it is also important to point out
14 that if you biopsy many biopsies within one region,
15 the measurement is very consistent.
17 With training, you can have very high
18 inter-rater and intra-rater reliability. Also, if
19 you measure healthy individuals over time, if you
20 serially biopsy one site over time, it is a very
21 stable measure in a healthy population.
23 So the Hopkins experience now totals over
24 7,000 biopsies including many diabetics. We
25 typically do the three standard sites, as I pointed
1 out, and it has shown good correlation with QSART
2 as well as sural-nerve biopsies.
4 So a skin-biopsy technique has been used
5 in a study of nerve-growth factor in HIV-associated
6 painful sensitive neuropathy. This slide depicts
7 some of those results. So I will focus at the
8 distal-leg site, again, because this is a
9 life-dependent process. The dark bars represent
10 patients with severe or extremely severe pain while
11 the dark gray is low to moderate pain.
12 It is the patients with more severe pain
13 that are lower at distal-leg epidermal nerve-fiber
14 density as measured by both the physician and the
15 patient pain assessment. So this is consistent
16 with the idea that loss of these fibers is
17 associated with neuropathic pain.
19 Also consistent with clinical observations
20 is the fact that distal-leg nerve-fiber densities
21 tended to be lower in patients with more severe
22 immunosuppression. Again, that is consistent with
23 the fact that HIV neuropathy is typically a disease
24 of advanced HIV disease.
1 So skin biopsy has been used by several
2 groups to study diabetic neuropathy. This is a
3 slide from Bill Kennedy in which he demonstrated
4 that patients with increasingly severe diabetes
5 have lower epidermal nerve-fiber staining.
7 This is echoed in another study by Levy et
8 al. in which they quantified epidermal PGP 9.5
9 staining in three populations; a normal control
10 population, a population of diabetic patients who
11 were normal by exam, symptoms, electrophysiology
12 and quantitative sensory testing and the
13 neuropathic diabetic population, and there seems to
14 be linear relationship.
16 Recently, there have been results of
17 several studies looking at a precursor to diabetes
18 in impaired glucose tolerance. Those studies have
19 looked at impaired glucose tolerance in patients
20 who otherwise have no known cause for their
21 neuropathy. These patients, for the most part,
22 prominently had pain as a feature of their
24 These reports are published at the
25 University of Utah and Yale report roughly a 35
1 percent prevalence of impaired glucose tolerance in
2 this population and that contrasts with a 15.8
3 percent IGT prevalence from the National Health and
4 Nutrition study.
6 We performed a similar study at Hopkins
7 which is not yet published, but I will summarize it
8 briefly. The results were very similar in that we
9 found 36 percent of our patients with neuropathy of
10 unknown cause were found to have impaired glucose
11 tolerance and 20 percent were frankly diabetic.
12 Again, that represents at two- to threefold
13 increased prevalence above NHANES.
14 We also found that there was relation--the
15 patient with the IGT-associated neuropathy had a
16 less severe neuropathy than those with
17 diabetes-associated neuropathy. So there was an
18 implication that there is a dose-response
19 relationship between the degree of glucose
20 dysmetabolism and the degree of neuropathy.
22 So the natural history of glucose
23 dysmetabolism has been addressed by several large
24 studies which have shown that impaired glucose
25 tolerance is a risk factor for diabetes and
1 precedes diabetes and, based upon that, we
2 hypothesize that the neuropathy associated with
3 impaired glucose tolerance could be a precursor to
4 diabetic neuropathy. Consistent with that was the
5 observation that our patients' duration of symptoms
6 in the IGT group was shorter than the diabetic
8 When we stratified patients by their fiber
9 type, there seemed to be a sequential progression
10 from small-fiber sensory involvement to combined
11 small-fiber and large-fiber sensory involvement to
12 sensory-motor involvement. So this argues that, at
13 least in the population which we looked at, which
14 is arguably a tertiary neuropathic population, that
15 skin biopsy may be the earliest detectable sign of
16 abnormality in these patients and have oral glucose
17 testing might be a more sensitive marker of glucose
20 We have also done some studies using skin
21 biopsy to look at nerve regeneration in humans. I
22 think the technique has several advantages in that
23 it uses skin, which is easily accessible, it is
24 easily biopsied and, as have heard, a sural-nerve
25 biopsy is not trivial. Also, skin can easily be
1 rebiopsied. You can only biopsy sural nerves
2 twice, one on each side. Skin is naturally
3 regenerative and, as we have said, it is
6 So we have developed two models to look at
7 two forms of regeneration, regenerative collateral
8 sprouting. I may touch upon that.
10 So this is a measure of regenerative
11 sprouting. This is a confocal micrograph which
12 showed baseline epidermal nerve fibers. After
13 injury, these fibers are completely eliminated from
14 the epidermis.
16 After recovery, this is 56 days, we see
17 nerve fibers growing back. I believe this
18 represents actual nerve growth and not an artifact
19 of staining because we get the same results whether
20 we stain with different Panex solo markers. It is
21 also correlated with heat-pain thresholds.
23 Collateral sprouting is another measure
24 which we can measure nerve sprouting into a
25 denervated zone. That is a different form of nerve
1 growth which has different neurotrophic
4 So, conclusions; I believe that
5 small-caliper nerve fibers are prominently affected
6 in diabetes, or they may be, at least in some
7 populations, the first class of nerve fibers to be
8 affected. They have been relatively unstudied or
9 understudied. I think that just points to the fact
10 that the tools we have had to look at them have not
11 been developed until relatively recently.
12 Their loss appears to be important in
13 neuropathic pain and this approach offers the
14 potential for an efficient way to measure nerve
15 growth in nerve-regeneration trials.
16 Thank you.
17 DR. KATZ: Thank you, Dr. Polydefkis.
18 Before we proceed with questions for Dr.
19 Polydefkis, we have a new arrival at the table.
20 Dr. Dyke, would you care to introduce yourself to
21 the group?
22 DR. DYKE: Peter Dyke, Mayo Clinic.
23 DR. KATZ: Thank you.
24 Questions for Dr. Polydefkis about skin
25 biopsies? Dr. Dworkin?
1 DR. DWORKIN: Setting aside the issue of
2 regeneration, I guess my question involves whether
3 you think it is possible to use biopsies as an
4 endpoint in the clinical trial. In other words,
5 would one, in an early intervention designed to
6 retard the progression of diabetic neuropathy,
7 predict, with active effective treatment, less loss
8 of epidermal nerve fibers in the treated group
9 versus the placebo group, or my concern, based on
10 the data you present is that this loss of epidermal
11 nerve fibers occurs so early in patients with
12 impaired glucose tolerance that it has not
13 potential as an endpoint because it has already
14 occurred before you would ever get these patients
15 into a clinical trial.
16 DR. POLYDEFKIS: I think it is fair to
17 potentially use it in a clinical trial. It was
18 used in HIV although that trial didn't last very
19 long. I think you can also vary the site. Like
20 David Cornblath said, if you focus on the toe, you
21 might be missing what is happening at the site of
22 the neuropathy or the junction of the neuropathy
23 and so, potentially, you could look at a more
24 proximal site.
25 DR. KATZ: Dr. Farrar, you were next.
1 DR. FARRAR: With relation to Bob was just
2 asking about, but also in terms of thinking about
3 how to look at the data, I was struck by the slide
4 you showed from Dr. Kennedy's work in the overlap
5 between those three sets of figures.
6 I wonder, in correlation, then, with
7 another slide that you showed which showed
8 proximally there was no difference between the two
9 biopsies between the two groups and distally there
10 was. I wonder whether you, in fact, looked at the
11 ratio between the number of nerve fibers in a
12 relatively normal area versus a relatively abnormal
13 area, whether that, in fact, helps to differentiate
14 the groups to a larger degree.
15 DR. POLYDEFKIS: Right. So that has been
16 done mostly notably by Chester MacArthur. That
17 ratio can be helpful although, in general, we use
18 absolute cutoffs. But if it is sort of on the
19 border, obviously by looking at where it fits into
20 that patient, you can put that number in
22 So if a person is borderline at the
23 distal-leg site but proximally they have an
24 abundance of fibers, that puts you toward saying it
25 was more abnormal.
1 DR. KATZ: Dr. Bril, you were next.
2 DR. BRIL: Thank you. I think this is
3 really an exciting field now. But I have a few
4 questions that maybe you can clarify.
5 This is useful on those patients who have
6 prominently small-fiber disease and, in the
7 diabetic group, this would be early neuropathy in
8 whom nerve conductions would be normal and a lot of
9 the other clinical measures would be normal.
10 I guess my question is informational. So
11 you have someone with burning feet, yet you have
12 biopsied their ankle and more proximally. What
13 does that mean to the burning feet? Which
14 fibers--if you are losing fibers, what is
15 signalling your pain, what is really carrying your
16 pain forward? What is the relevance of the loss of
17 these fibers in the skin at the ankle and more
18 proximally to the burning-feet syndrome that we are
19 dealing with mostly?
20 Usually, when the pain comes up higher,
21 they usually have large-fiber involvement as well.
22 So that is the thing. The Kennedy data, when I
23 looked at that paper, and you look at the
24 correlation with epidermal nerve-fiber density, it
25 drops and then it is just at the bottom.
1 So, from a fairly--I mean, the mild
2 patients, there may be a correlation, but you get
3 moderate to severe, that is lost. I mean, it is
4 just the fibers are all gone. They are not
5 detectable anymore. So the reflection of the
6 clinical state is a little, still, I think, early.
7 So I would like your comments on that.
8 DR. POLYDEFKIS: First, the pain. I guess
9 the question is what is causing the pain. That is
10 an unknown. That is not known. But you are losing
11 fibers from the epidermis but they are still there.
12 The distal end is probably in the dermis.
13 DR. BRIL: But you would think in the
14 feet, they would probably have more loss because
15 you have got the gradient. You are not even doing
16 the feet where they have the burning pain.
17 DR. POLYDEFKIS: Right. So just
18 practically, we didn't biopsy feet because it is
19 logistically complicated, increased risk of
20 infection. People wear shoes. But I suspect you
21 may be right. If you biopsy them more distally,
22 you would see more severe loss.
23 The other question is that you are right.
24 Once you get to 0, you can't go lower than 0. But
25 you can biopsy more proximal sites. So, in more
1 neuropathic individuals, even though a distal-leg
2 biopsy may not give you that much information, a
3 distal-thigh biopsy might.
4 DR. BRIL: I guess the question, then, is
5 if we are looking at nerve dysfunction up in the
6 thigh and it is not related to the pain in the
7 feet, how are we going to relate those two in a
8 study, in an endpoint study, because we are going
9 to have the same comment that we have had about
10 other surrogates.
11 We are going to say, your nerve-fiber
12 density is better in the skin and the thigh. But
13 if the thigh is not even bothering you, if your
14 thigh is perfectly normal, you can't detect a
15 sensory deficit, there is no pain, the burning pain
16 is all in the feet, you are going to have to answer
17 the same comment.
18 DR. POLYDEFKIS: That is a fair point.
19 That is why I kind of tried to point out some of
20 the morphologic abnormalities. Even though
21 patients won't have symptoms in their thigh, there
22 is evidence of nerve injury in the thigh by the
23 swellings and segmentations of the nerve fiber.
24 So I think you could argue that if you are
25 improving a site, even though it might not be
1 symptomatically neuropathic, you are having an
2 effect on nerves.
3 DR. KATZ: Dr. Rowbothom?
4 DR. ROWBOTHOM: Let me just make a few
5 comments on that because we have been using skin
6 biopsies to study postherpetic neuralgia and have
7 published in this area since 1996. There you have
8 a different situation in that you have a
9 contralateral side that doesn't have clinical
10 symptoms. It is not a perfect control because
11 Zoster does produce some bilateral changes and so
12 there may be some change in nerve fibers
13 contralateral to the area of pain. But it
14 certainly gets around the problem that you have in
15 diabetic neuropathy where you have two feet that
16 are deafferented.
17 What our studies show is that in the
18 center of the area of greatest pain, that is where
19 the nerve-fiber dropout is usually the greatest, if
20 there is nerve-fiber dropout. As you biopsy
21 towards the edge of the area that is affected, you
22 get nerve-fiber counts that are closer and closer
23 to what you see on the contralateral side.
24 The relationship between pain and
25 allodynia, thermal-sensory function and the number
1 of fibers in the skin is quite complicated. In
2 some earlier studies that we did, we found,
3 actually, an inverse correlation between
4 thermal-sensory impairment, thermal-sensory
5 detection impairment, and pain so that it was the
6 patients who had the best ability to detect thermal
7 stimuli that actually had the most pain and the
8 most allodynia which would suggest that it is not a
9 complete loss of all the fibers that is necessary
10 but that there is an important intermediate point
11 where there are fibers there.
12 They are functioning, but they are not
13 normal fibers. They are sick in some way. They
14 are damaged and they can't fully recover. So the
15 other point I just want to make is that patients
16 who have no fibers left in their skin generally
17 don't have allodynia to touch in postherpetic
18 neuralgia. The ones who have allodynia, especially
19 severe allodynia, actually do have a fairly--either
20 a normal or near normal number of fibers in the
21 skin in their area of greatest pain.
22 So that is a disorder where we can analyze
23 the problem a little differently than diabetic
24 neuropathy, but I just want to echo what David was
25 saying earlier and also what Michael was saying is
1 that this is a technique that you can do serially
2 and patients tolerate it well.
3 It may be a surrogate marker as far as
4 quality of life or pain or other things, but it is
5 a hard marker in that you actually are visualizing
6 and characterizing the nerves. So if your agent is
7 designed to be neuroprotective, you are actually
8 getting real anatomical data about the physical
9 state of the nerves that you are interested in.
10 DR. KATZ: Dr. Shafer, you were next.
11 DR. SHAFER: Two things. One is certainly
12 what I have heard so far has been very positive in
13 the sense that this is something that we have seen
14 data now for diabetes, we have seen data for
15 HIV-associated pain. You just reported data in
16 postherpetic neuralgia, with the obvious exception
17 of phantom-limb pain.
18 Is this something that, in fact, could be
19 considered to be a broadly applicable surrogate for
20 neuropathic pain?
21 DR. ROWBOTHOM: Yes and no in that there
22 is not a perfect correlation between the number of
23 nerve fibers and pain. That is really a major
24 issue because the biopsy tells you how many fibers
25 there are and, as Michael showed, you can make a
1 lot of inferences about morphology. But I think
2 our state of knowledge about what we are seeing in
3 the skin is still crude enough that we can't say
4 that this biopsy picture guarantees pain and if you
5 reverse that abnormality, then you have alleviated
6 pain. We are not there yet.
7 DR. SHAFER: That actually goes right to
8 the other thing I wanted to ask, then, was have you
9 looked at counts of nerves versus pain as opposed
10 to the morphological indices that we saw on the
11 slides, and counts of abnormal nerves, dilated
12 nerves, things like this, versus pain.
13 DR. POLYDEFKIS: It has not been looked at
14 systematically. It is very challenging to look at
15 that so you have to quantify what is swelling, what
16 is a morphologic abnormality. So we have global
17 impressions but, beyond that, it has not been
18 systematically looked at.
19 DR. KATZ: Dr. Cornblath?
20 DR. CORNBLATH: We have done thousands of
21 these biopsies in patients in our own place and I
22 would echo what Mike said. It is not perfect. It
23 is not going to be a correlate for this symptom.
24 Some of that reflects the fact, as Mike said, that
25 he has already shown that people have good numbers,
1 can have more allodynia, and it also reflects the
2 fact, as we have talked about here multiple times,
3 this proximal-distal gradient.
4 Our biopsies are done 10 centimeters above
5 the lateral malleolus. That is our standard site.
6 You can have your toes on fire and be really in a
7 lot of discomfort and you can have a normal biopsy
8 at that site because it reflects a morphologic
9 change, then. So I don't think it would be useful
10 unless you, again, started moving it all around
11 and, even then, based on Michael's data, I am not
12 sure it would work as a correlate of the symptom of
14 It is a correlate of morphologic abnormality of the
16 DR. KATZ: Dr. Feldman?
17 DR. FELDMAN: Michael, could you refresh
18 my memory on the David Hermann paper that
19 showed--you mentioned in your talk, that actually
20 shows that this is a good surrogate for sural-nerve
21 biopsy. I mean, that, in a way, is very exciting
22 if we could do these types of biopsies in lieu of
23 sural-nerve biopsies and be able to do them
24 separated in time and get similar or meaningful
1 DR. POLYDEFKIS: I can't recall the exact
2 numbers but every patient who, on sural-nerve
3 biopsy was felt to have small and myelinated
4 nerve-fiber loss, that was in agreement with the
5 skin biopsy and there were a few patients who had
6 normal sural-nerve unmyelinated nerve-fiber counts
7 who had abnormal skin biopsies.
8 It just spoke to the point that skin
9 biopsy might be a more sensitive measure of that
10 population than sural-nerve biopsy and that would
11 make some sense because, again, skin is a more
12 distal structure and so it may be consistent with
13 skin being infected first.
14 DR. CORNBLATH: Again, we ought to be very
15 careful because I don't think it is really a
16 surrogate for the unmyelinated counts in sural
17 nerve, and it isn't because there are people, as
18 Michael said, in both the Holland paper and the
19 Hermann paper, who have normal unmyelinated fiber
20 counts in the trunk of sural nerve as we take it in
21 the mid calf who have abnormal skin. That is
22 perfectly predicable on the length-dependent nature
23 of this disease. So it won't be a surrogate.
24 DR. KATZ: Are you saying, then, that it
25 may, in some cases, be more sensitive than the
1 sural-nerve biopsy or is just a matter of--
2 DR. CORNBLATH: I believe so. But, again,
3 that is all predicted on the basis that this is a
4 length-dependent dying-back neuropathy and the
5 sural biopsy looks at it like in the upper arm and
6 Michael's technique and others look at it down in
7 the fingertips where the action starts.
8 DR. KATZ: Dr. Woolf, you were next.
9 DR. WOOLF: I think you need to be
10 extremely cautious about this because it is not a
11 marker of nerve fibers. PGP stains are unbiquinase
12 so it is not staining the nerve fibers. If that
13 enzyme is downregulated, which it may be, or its
14 transport is affected, which it may be, by the
15 disease state, you will have an apparent
16 disappearance of nerve fibers but the nerve fibers
17 may be there or atrophic.
18 So I think we have got to be a little bit
19 cautious about that in the same way that I think we
20 have got to be extremely cautious about correlating
21 the entire experience of pain with peripheral-nerve
22 endings where so much of pain is centrally
23 generated by altered processing in the CNS.
24 DR. POLYDEFKIS: I think that is a good
25 point. I should say if you use other Panex solo
1 markers we see the same thing. So I suspect the
2 conclusions may well be correct.
3 DR. CORNBLATH: We have looked at EMs in
4 skin when the PGP 9.5 is not there and the fibers
5 are not there.
6 DR. WOOLF: I have no difficulty with
7 that, but that doesn't mean that every time--I am
8 sure that if there are no fibers there, you will
9 have no PGP. What I am saying is can you do the
10 other way around, just because PGP is gone, can you
11 be always confident fibers aren't there.
12 DR. CORNBLATH: In the cases, and we
13 haven't done thousands of them because, as you can
14 imagine, they are technically difficult, when there
15 isn't PGP 9.5 staining, there are not nerve fibers,
16 if that answers the question. That is, we have not
17 seen cases where the PGP stain is absent--
18 DR. WOOLF: I think Frank Rice has an
19 experience where the fibers can get so thin and
20 atrophic that, on thick sections like 50 micron
21 sections, you may not get staining but, in fact,
22 when you reduce the size--this is getting into
23 technical issues, but if you increase the
24 sensitivity, you can start to see very thin
25 atrophic fibers.
1 DR. CORNBLATH: I would be interested to
2 see that material.
3 DR. KATZ: Dr. Farrar, you were next.
4 DR. FARRAR: With regards to the comment
5 about pain and the central processing, I think it
6 is important to point out two features of diabetic
7 neuropathy that are paramount in terms of thinking
8 about how to treat the discomfort.
9 The first of the features is that the
10 improvement that we are looking at is in the
11 peripheral nerve. I think it was commented earlier
12 that if the peripheral nerve becomes enough, you
13 actually get damage to or potentially death of the
14 cell body at the dorsal-root ganglion.
15 My guess is that if the cell body dies
16 that the nerve doesn't come back, in general,
17 anyway. Once that has happened, any amount of
18 trying to control the process that caused the nerve
19 to die originally is not going to help. I guess
20 the analogy is once the car has crashed into the
21 tree, fixing the brakes doesn't help very much.
22 The second issue I think is that the
23 process we are talking about in terms of this
24 disease is really a peripheral process and we know,
25 as Clifford was alluding to, that some of the
1 treatments that we use don't work peripherally at
2 all but, in fact, work centrally to increase the
3 downregulation of changes that occur at the spinal
4 cord. So you might see a drug that works very
5 effectively in a symptom--i.e., control of the
6 pain--that has no effect or no benefit on the
7 peripheral system.
8 I think that is the problem in trying to
9 look at these two things together. Clearly,
10 preventing the progress of the disease is a good
11 thing and probably, ultimately, results in changes
12 in sensation and/or pain discomfort, although I
13 think I agree that it needs to be clearly
14 demonstrated. Some of it has been.
15 But, in addition, there are going to be
16 agents that don't work at all peripherally that
17 would be clearly beneficial for the symptoms.
18 Would you agree?
19 DR. POLYDEFKIS: Yes.
20 DR. KATZ: Dr. Dworkin, you have the last
22 DR. DWORKIN: We have talked a lot about
23 using skin-punch biopsies as endpoints. I was
24 wondering if you could comment on their potential
25 use as part of the inclusion criteria for a study.
1 In other words, could you imagine a study of
2 idiopathic small-fiber neuropathy where that would
3 be an inclusion criterion, that the patient has
4 small-fiber loss? Is there a role there?
5 DR. POLYDEFKIS: I believe so. I think it
6 is also potentially would support some of the
7 scales that have been discussed.
8 DR. DWORKIN: As part of the composite.
9 DR. POLYDEFKIS: Right.
10 DR. KATZ: I would like to end with just
11 one final question. One thing that I may have
12 missed in your talk is that has the skin-punch
13 biopsy neurofibrodensity been followed
14 longitudinally in a patient population to look at
15 what magnitude of change one sees?
16 DR. POLYDEFKIS: We are doing that but it
17 has not been done systematically. We are in the
18 process of doing it.
19 DR. KATZ: So it would seem difficult to
20 put that on the top of the pedestal as an outcome
21 measure without that experience of looking at the
22 degree of change that occurs. Do you agree with
24 DR. POLYDEFKIS: I think that is fair;
1 DR. KATZ: Lunch now. We will start
2 promptly at 12:55. For people around the table,
3 head to the back of the restaurant next door and
4 everybody else enjoy your lunch. We will see you
5 at 12:55.
6 [Whereupon, at 11:55 a.m., the proceedings
7 were recessed to be resumed at 12:55 p.m.]
1 A F T E R N O O N P R O C E E D I N G S
2 [1 o'clock p.m.]
3 DR. KATZ: We will start the afternoon
4 session. This portion of the afternoon session
5 will be devoted to discussion of some of the
6 critical issues in relation to clinical trials for
7 diabetic peripheral neuropathy. In order to help
8 us focus our attention on exactly what those
9 questions are, Dr. McCormick will give us a charge
10 to committee.
11 Charge to the Committee
12 DR. McCORMICK: Welcome back. This
13 afternoon, we will bringing to the committee a
14 number of issues that have been challenging both
15 the FDA and industry as we approach the development
16 of drugs for neuropathy and neuropathic pain, as
17 you have heard this morning.
18 You have heard also this morning a bit
19 about the regulatory context in which we operate,
20 the need for a delicate balancing act and
21 thoughtful judgment as we apply new scientific
22 ideas and knowledge within our regulatory
24 We will be seeking advice from the
25 committee this afternoon on a number of questions.
1 Keep in mind that our regulatory approval and
2 policy decisions must be based on evidence.
3 Neuropathy is an area of drug development in which
4 there has been a paucity of evidence generated.
5 The elements that go into the design of clinical
6 trials and drug-development plans should be widely
7 accepted by experts in the field.
8 We will be asking your advice on outcome
9 measures, usefulness of surrogate endpoints,
10 duration of trials, effect sizes that are
11 clinically meaningful, and appropriate definition
12 of entry criteria which will help to define the
13 drug's indication.
14 As for neuropathic pain, the most commonly
15 considered question is do we know enough to
16 generalize yet. There are some clear benefits to
17 industry, as we have heard, in obtaining a broad
18 indication for neuropathy pain. If and when this
19 is something we should consider, we should
20 carefully think through what evidence would support
21 such a broad indication and be able to articulate
22 why; that is, justify it, keeping in mind that
23 policy steps that are taken with one class of drugs
24 may adversely affect another.
25 We would like to hear some discussion
1 about how much existing data you feel you have in
2 making generalizations about drug effects across
3 the vast variety of neuropathic-pain states. You
4 should also consider the risks associated with a
5 broad indication such as the risk of
6 overgeneralizing based on a narrow set of data or
7 insufficient safety exposure in the target
8 population or, in the case of some narcotics,
9 widespread availability and prescription-drug
11 These risks would ultimately have to be
12 addressed by us before and as we make an approval
13 decision. Keep in mind that if the FDA allows a
14 broad claim for a given indication, we must have
15 sufficient evidence that such a claim is really
16 applicable, the truth-in-advertising principle.
17 There is also a down side to a narrow
18 indication based on a small development program
19 particularly if the drug may have a much larger
20 target population. The greatest risk of widespread
21 off-label use is that of inadequate safety
22 evaluation during development. There are the
23 additional problems with reimbursement which has
24 plagued the neuropathic-pain community.
25 You have had a chance to read the FDA's
1 guidance for providing evidence of effectiveness
2 for human drug and biological products. You can
3 see from this document that there is some
4 flexibility in the evidence that can be accepted in
5 support of efficacy. This flexibility increases as
6 experience is gained with a class of drugs or
8 Keep these principles in mind as you enter
9 your discussions this afternoon. The afternoon
10 will be divided into discussions of neuropathy drug
11 development, disease-altering claims and the second
12 half will be discussion of neuropathic pain. A
13 debate on the issue of general versus specific
14 claims in neuropathic pain will hopefully stimulate
15 your thinking.
16 We are grateful for your willingness to
17 share your expertise with the FDA and we look
18 forward to a stimulating and very fruitful
20 I won't be going through the questions
21 since there is a long list of questions but I will
22 defer to the chair to go through them one by one as
23 we move forward through the afternoon. Thank you.
24 DR. KATZ: Thanks Dr. McCormick for
25 setting the stage for this afternoon.
1 What I would like to do first is, there
2 were a number of people in the room whose names
3 were mentioned several times by speakers as having
4 relevant experience and expertise to share with us
5 who are not sitting around the table. I would like
6 to sort of reopen in a small way the public forum
7 by asking any of those individuals in the room who
8 might be there, Mitchell Max is one and Joe Arezzo
9 is second. So, if you could just take two minutes,
10 Dr. Max.
11 DR. MAX: Mitchell Max from the National
12 Institutes of Health. I just need to say that my
13 conflicts of interest include that I either
14 collaborate with or consult for a large proportion
15 of the companies doing analgesic drug development.
16 I wanted to mention some very odd
17 phenomenon, and it is an important public-health
18 need, that the neuropathic pain condition that is
19 by far the most common has hardly been mentioned
20 today. We have been talking mostly about diabetic
21 neuropathy and a little about postherpetic
22 neuralgia, but there are ten to twenty times the
23 number of people with pain from nerve root, from
24 degenerative disease in the neck or the back.
25 It is very odd that there are essentially
1 no academic NIH-funded drug-treatment trials in
2 chronic radiculopathies and there is essentially no
3 industry development. I would urge you, since that
4 really is the bulk of the difficult neuropathic
5 pain we treat, just think of how many people in
6 your family and your close friends have complained
7 to you about sciatica or neck pain, or yourself.
8 Since this is so important, we ought to
9 think about how we can promote it. I must confess
10 that every drug company I have ever spoken to, at
11 the beginning of the meeting, I say, why don't you,
12 if you want to neuropathic pain, do a clinical
13 trial in radicular pain. And they always reject it
14 and they say no, we are going to study diabetic
15 neuropathy even though all the other companies are
16 going for that because there isn't any track record
17 yet. They are afraid that maybe it won't work.
18 There is somehow this strange hurdle.
19 Another point that perhaps you can discuss
20 later, it is not clear to me that results in
21 diabetic neuropathy and postherpetic neuralgia
22 generalize to root pain. In root pain, there is
23 generally mechanical pressure on the root or on the
24 dorsal-root ganglion and the biology of pushing on
25 the nerve cell or the root which is central to the
1 nerve-cell body. The biology must be different
2 from an injury peripherally, so maybe you need to
3 do separate trials in that.
4 So I would just urge you to think about
5 how you could encourage by a claims structure or
6 some other thing companies to get into radiculpathy
7 pain so we can treat what people have.
8 DR. KATZ: Thank you, Dr. Max.
9 Dr. Arezzo? Also, if you could start with
10 any relevant disclosures, that would be helpful.
11 DR. AREZZO: I am Joe Arezzo from Albert
12 Einstein College of Medicine. I have consulted
13 with many of the companies in terms of diabetic
14 neuropathy and a few in painful neuropathy.
15 I think one of the more intriguing
16 questions raised this morning was the issue of what
17 is the relevance of the surrogate endpoints, a
18 question that you posed, what does it mean to the
19 patient to have a change in nerve conduction,
20 particularly a relatively small change that might
21 be seen in a clinical trial or a change in
22 quantitative sensory testing, does that have impact
23 for the patient.
24 I think we have obviously limited data in
25 that respect as many of the people have already
1 mentioned. But one of the more important studies
2 in the DCCT trial. In that trial, essentially a
3 1-meter per-second per-year change in the perineal
4 nerve-conduction velocity translated to the 50
5 percent reduction in clinically evidence neuropathy
6 at the end of a five-year period of time.
7 So patients that experienced--that were
8 intensively treated and had 1 meter per-second
9 improvement per-year had 50 percent--there was a 50
10 percent difference in the clinically evident
11 neuropathy at the end of five years in terms of the
12 number of patients.
13 Another study that I think is relevant is
14 Andrew Bolton's study and Jay Sosenko's studies on
15 quantitative sensory testing. Andrew Bolton for
16 vibration demonstrated that elevation of vibration,
17 quantitative sensory-testing scores to a threshold
18 that he defined as important, 25 volts in a
19 biothesiometer, had more than a fourfold--if you
20 elevated to that score in quantitative sensory
21 testing, you had more than a fourfold increase in
22 your incidence of ulceration of the foot.
23 So that was a threshold, a point which you
24 could measure in patients before ulcerations but a
25 point which was very strongly predictive of those
1 patients that would have ulcerations.
2 Jay Sosenko did a similar study with
3 thermal thresholds demonstrating the relationship
4 between progression of quantitative scores to an
5 area of risk and the clinical development of
6 ulcerations. So these surrogate points are clearly
7 surrogate measures but I think they do have direct
8 relevance for the progression, the long-term
9 progression, to serious clinically relevant
11 Thank you.
12 DR. KATZ: Thank you very much, Dr.
13 Arezzo. Is there anybody else from the public that
14 would care to take the opportunity to share some
15 thoughts with us?
16 Okay, great. Why don't we go ahead and
17 start the discussion then.
18 Entry Criteria
19 DR. KATZ: As you can see in your agenda,
20 the first topic that we will be focussing on this
21 afternoon is the topic of entry criteria for
22 clinical trials for diabetic neuropathy. Again,
23 just to focus everybody's attention, we are not
24 talking about pain right now. We are talking about
25 disease-modifying drugs and trials of those to
1 interfere with the natural history of peripheral
3 After we discuss the entry criteria, the
4 next subject will be outcomes measurement and so it
5 will be very easy for us to slip into that. But I
6 would like to try to avoid that for now and just
7 talk about entry criteria per se so we can
8 accomplish something in that domain.
9 Now, of course, there are a number of
10 relevant questions in terms of entry criteria for
11 patients with diabetic peripheral neuropathy. Why
12 don't I take the prerogative of just starting off
13 our conversation this way with how one should make
14 the diagnosis of diabetic polyneuropathy in such a
15 clinical trial.
16 Any thoughts on that issue? Dr. Feldman,
17 you look like you are nodding your head there and a
18 thought is percolating. Would you care to start?
19 DR. FELDMAN: I think that the diagnosis
20 needs to be made on a clinical ground in terms of
21 some sort of clinical examination. We talked
22 earlier about potentially a modified NIS(LL) or
23 potentially the quantitative, semi-quantitative
24 sensory testing that David does up the leg in his
25 type of examination or ones that have previously
1 been developed by myself or Dr. Bril.
2 But the key is we need a clinical portion
3 of the examination and that needs to, then, be
4 accompanied by, I believe, a quantitative portion
5 because all electrophysiology and quantitative
6 sensory testing is just an extension of our
7 clinical exam. So I would say that a nerve
8 conduction study focused on the perineal motor
9 nerve and the sural sensory nerve, and then
10 possibly, depending somewhat what your entry
11 criteria are somewhat are dependent are what your
12 outcome measures are, of course, because, for
13 example, if you want to measure changes in cold
14 perception threshold, then you are going to need,
15 as an entry criteria--or vibration perception
16 threshold, you are going to need that to be
17 measurable upon entry.
18 I do believe that the sural and perineal
19 need to be measurable upon entry. So I think that
20 patients meeting those three criteria would be good
21 candidates for a clinical trial.
22 DR. KATZ: Just to state what probably is
23 obvious, is it obvious enough that it is widely
24 accepted that a clinical evaluation, by itself, is
25 insufficient to characterize patients on entry to
1 such trials and that the quantitative testing is
3 DR. FELDMAN: I think the DCCT is a great
4 example where, in the DCCT, a neurologist examined
5 the patient. Although there were specific things
6 you were to do at the end of the day, it was
7 whether or not the neurologist said yes or no, you
8 had neuropathy. It wasn't totally nonquantitative.
9 I think, though, because of somewhat of
10 the subjective components of the clinical exam, I
11 think a simple clinical examination probably is not
12 sufficient for entry into a clinical trial.
13 Certainly, it is very sufficient in the clinic. I
14 am sure it would be good to hear what Drs. Dyke,
15 Cornblath and Bril think about that, but I do think
16 you need to extend your clinical examination with
17 something more quantitative.
18 DR. KATZ: Dr. Dyke, do you have any
19 comments on that?
20 DR. DYKE: I agree that, for a trial, you
21 really would want objective criteria. I have
22 thought that that nerve-conduction attributes are
23 very good for that purpose and I agree that the
24 quantitative sensory could also be.
25 If I could digress a little bit more,
1 though, and broaden this a little bit, it is
2 important for us to recognize that the assumption
3 around this table has been only about diabetic
4 sensory polyneuropathy. But that is only one of a
5 series of other neuropathies.
6 So one, in thinking about entry criteria,
7 ought to say what neuropathy we are talking about.
8 There are diabetic sensory polyneuropathies. There
9 are some of them that begin during the honeymoon,
10 if you like, called insulin neuropathies. They may
11 have a totally different pathogenesis than the
12 metabolic polyneuropathy.
13 So there probably are different diabetic
14 sensory polyneuropathies. Then there are
15 multifocal neuropathies and entrapment
16 neuropathies. The median nerve at the wrist, the
17 ulnar nerve at the elbow are two common examples,
18 but the perineal is the third.
19 Then there are multifocal motor
20 neuropathies of several well-defined
21 characteristics. Osh described a brachial-plexus
22 neuropathy form. There is a well-known lumbar
23 form, thoracic form, if you like. And then there
24 is the lumbosacral form.
25 The putative mechanisms are quite
1 different. So, in my mind, one has to begin with
2 neuropathy we are talking about. Are we doing a
3 preventative or an interventative trial? What
4 pathophysiology are we going after? It could make
5 a big difference.
6 So I don't see this as something this
7 group can tackle just broadly as we are doing here
8 now but I think you would have to specify first
9 that we are probably talking about the metabolic
10 variety. We are talking about diabetic sensory
11 polyneuropathy. If that is the criteria we are
12 using, then it makes sense, I think, what you said.
13 If I could just mention two other things.
14 It also depends on the putative action of your
15 drug. There really is a need to think of trials
16 that address different issues. Clearly, there is a
17 metabolic basis for diabetic neuropathy. I think
18 everyone agrees with that, but there could well be
19 a mechanical basis for other varieties, an immune
20 basis for still other varieties, a hypoglycemic
21 anoxic basis for others and so on.
22 Then I think it also depends, in terms of
23 criteria, about the outcomes, but you want to leave
24 that for the next one. But I did want simply to
25 say we really ought to broad the idea of pain
1 because diabetic neuropathy has more than pain.
2 They have positive neuropathic sensory symptoms
3 which consist of lancinating pain, burning pain,
4 deep aching pain, itching, tenderness of their feet
5 when they walk.
6 The people who propose the study need to
7 think about those things and it would really modify
8 the criteria. So I think it would be a mistake to
9 come down with a sort of rigid set of criteria for
10 an undefined study on "diabetic neuropathy." A
11 little side pitch.
12 DR. KATZ: No; that is clearly important.
13 Let me just push you a little bit on that. It
14 sounds like you are saying that if one is trying to
15 study the metabolically based peripheral diabetic
16 polyneuropathies, stocking-and-glove neuropathy,
17 that one ought to take pains to exclude other kinds
18 of neuropathies associated with diabetes,
19 multiple-nerve entrapments, thoracic radiculopathy,
20 proximal neuropathy, et cetera, et cetera.
21 How would you suggest operationalizing
22 that attempt to exclude those other diabetic
23 neuropathies in a clinical trial?
24 DR. DYKE: It would be clear that you
25 could have an algorithm in which the neurologist
1 ends up making the final judgment. One could do it
2 having the nurse call and determining some things
3 and then going on to the neurological examination
4 as proposed by Dr. Feldman.
5 I think the bottom line is, though, that
6 there really are differences even in the diabetic
7 sensory polyneuropathy and we really ought to
8 focus. If you are going to set down criteria, you
9 ought to focus on which variety you are talking
11 DR. KATZ: Are you suggesting, then, that
12 even among the stocking-and-glove neuropathies,
13 polyneuropathies, associated with diabetes that
14 there are different physiologic subtypes there that
15 can be distinguished in some way?
16 DR. DYKE: We think so. That needs to be
17 established but the Columbia group did nerve
18 biopsies on patients with diabetic neuropathy and
19 found that there were certain ones that had
20 inflammatory infiltrates. We found the same thing.
21 We think that the patients who don't have
22 tripathy--that is, retinopathy and nephropathy, or
23 mild degrees of that, and have a lot of symptoms,
24 that immune factors might, in fact, be playing a
25 role in those.
1 So that needs to be taken into
2 consideration. So, for most of the trials that I
3 am involved with, we have always said patients
4 should have diabetes by ADA criteria. Secondly,
5 they should have stable metabolic control. Three,
6 certain categories of disease should be ruled out,
7 like other diabetic neuropathies.
8 6 to 8 percent of a diabetic cohort have
9 other kinds of neuropathy. If you don't put them
10 aside, you are mixing up the trial. Then you go on
11 from there to exclude patients with overlapping
12 neurological disease. If they have Freidreich's
13 attacks in addition to diabetic neuropathy, you are
14 not going to be able to tease it out. So other
15 neurological diseases need to be sorted out.
16 Then the question of what degree of
17 metabolic control should they have before you put
18 them into the study. The ADA criteria now hold
19 that you should try and be--you know, people in the
20 audience should know this much better than I
21 do--below 8 percent on the glycated hemoglobin or
22 maybe even on the hemoglobin A1C.
23 Now, on the other hand, you don't want to
24 exclude such patients from trials if they can't get
25 that level of control. But that is a big area of
1 concern. If they have, in addition, a little
2 uremia, the uremia, itself, can cause neuropathy so
3 that has to be a factor that has to be considered.
4 So it is a very complex issue, actually,
5 the inclusion and exclusion criteria. But I think
6 it begins with a clear focus on what you are trying
7 to improve and that makes a big difference.
8 For example, one drug might affect--you
9 might be aiming at the symptomatic group so you
10 clearly have to pick Stage 2 patients. Other
11 patients, you are trying to influence impairments
12 so you have to have a milder group that you are
14 Well, I have spoken too much.
15 DR. KATZ: You have hit on a lot of
16 critical issues that I am sure we will discussing
17 at length and I appreciate that. What I would like
18 to do is just summarize some of the key points you
19 made for the purpose of moving the discussion,
20 focussing the discussion.
21 It sounds like what you are saying is
22 that, number one, for patients to be included in a
23 clinical trial for distal sensory polyneuropathy
24 and diabetes that, number one, we ought to exclude
25 other types of diabetic neuropathy and there should
1 be some sort of algorithm or operation or proviso
2 that requires a neurologist to exclude those other
3 diabetic neuropathies.
4 Secondly, it would be important to exclude
5 a nondiabetic cause of a peripheral polyneuropathy
6 such as vitamin deficiency, alcohol, what have you.
7 Those points seem clear enough although, in my
8 experience reading results of clinical trials, it
9 is not usually done.
10 Third, we have to be careful in accepting
11 patients with uremia which may be due to diabetes
12 into the trial or at least potentially look at
13 those patients differently. Fourth, there may even
14 be subtypes within what we usually lump together as
15 diabetic sensory polyneuropathy that, although we
16 don't have any technology now to tease those
17 different subtypes out, there may be ways of
18 approaching that that we ought to keep in mind, one
19 being potentially tracking which subgroup of
20 patients has nephropathy and retinopathy since they
21 may be different than patients with neuropathy that
23 Have I captured everything you have said
24 as far as the entry criteria?
25 DR. DYKE: Maybe also add the point that
1 try and use as objective a criteria for entry as
2 you can. Usually, that means based on a normative
3 study in which it is defined as an abnormal
5 DR. KATZ: Does anybody have any comments
6 specifically about the proposals that we have just
7 had put on the table with regard to entry criteria?
8 Dr. Rendell, you have been waiting for a while.
9 DR. RENDELL: Dr. Dyke wasn't here this
10 morning when I raised just this question. The
11 question is, Peter, do you think there is a way to
12 tease out subtypes of what appear to be the same
13 disease--in other words, diabetic sensory
14 polyneuropathy--and, specifically, do you think
15 there may be certain individuals who have
16 microvascular disease as the genesis of their
17 neuropathy, others who have excessive oxidation as
18 the genesis, others who have abnormal aldose
19 reductase? Is there any way to get at a possible
20 multiple heterogeneous etiology and then be able to
21 select drugs that might treat one or the other
23 DR. DYKE: I can't answer it in any final
24 way but I think the consensus is growing among many
25 of us that, from the time of the studies at Arhus,
1 Denmark, where they showed an association between
2 retinopathy and neuropathy and nephropathy, and
3 there have been many studies since that time, that
4 there is, in general, an association.
5 If you don't, in a given patient, have
6 this association, you may not, in fact, be dealing
7 with the metabolic diabetic polyneuropathy.
8 The second trend that I think we are
9 seeing that people are recognizing that there may
10 be other mechanisms that influence the expression
11 of generalized neuropathy. One of them is,
12 obviously, immune events. I was suggesting that
13 the sort of insulin neuropathy that people talk
14 about where it actually was described from Michigan
15 in 1945 where a person who gets put on insulin
16 develops a symptomatic neuropathy and then, six
17 weeks later, improves.
18 That is common experience. They are
19 referred to by the Brits as insulin neuropathies.
20 I am not sure what that is. It could be metabolic
21 but it could also be immune. One should be careful
22 about that, I think, as a subgroup. So, most of
23 us, in our thinking about trials have tried to keep
24 that group out of it because we don't know what is
25 causing it.
1 Then, clearly, you know the compression
2 neuropathies are a real confounding variable in
3 trials. Perkins and Vera Bril and someone else
4 just wrote an article in which they were looking at
5 this question, can you tell the difference in the
6 electrophysiological features of patients who have
7 both clinical carpal-tunnel syndrome and diffuse
8 neuropathy versus polyneuropathy.
9 They said, in their equation, that they
10 were not able to show a difference. I would like
11 to suggest a few things to your study, Vera, but
12 that can be done later. But the point is well
13 made. It is hard to separate out the
14 electrophysiological features which are from carpal
15 tunnel and which are from diffuse neuropathy.
16 Then there is that whole group of the
17 radiculoplexus neuropathies which is coming in like
18 gangbusters. There is no question there is an
19 immune component. So I think, at certain levels,
20 one can do it.
21 DR. KATZ: Dr. Bril, would you like to
22 make some comments?
23 DR. BRIL: I would agree totally with
24 Peter. We have to define the type of neuropathy we
25 are planning to study in any research trial. That
1 is fundamental. I agree with Eva, you do need some
2 clinical features buttressed by objective measures,
3 electrophysiology plus or minus QST. I think those
4 are essential and they have been successful in
5 selecting populations.
6 I think what was interesting was that
7 there was no good electrophysiological measure to
8 differentiate the patients because, if you want to
9 get picky about it with electrophysiology, you can
10 almost eliminate everybody with diabetic neuropathy
11 as having carpal tunnel and then you would never
12 have a patient in your study.
13 The reason we were trying to do this study
14 was to see if we could--various algorithms have
15 been suggested to me over the years such as the
16 difference in median sensory to ulnar sensory, the
17 difference in the amplitude ratio from the median
18 to sural, from the median to ulnar, a difference
19 with a proximal conduction to the distal
21 Yet multiple different
22 electrophysiological rules to try and separate
23 carpal tunnel in someone with diabetic neuropathy
24 from the diffuse neuropathy had been suggested as
25 exclusionary rules. None of my colleagues knew
1 which was the best one. Everybody had a little
2 different rules.
3 The purpose of the study was to look at
4 patients with diabetes, look at who had clinical
5 neuropathy to find clinically in the way most
6 neurologists would do it, and then see if you could
7 separate those patients out from those with
8 neuropathy by electrophysiology, and you couldn't.
9 You just couldn't.
10 You couldn't do it in those with diabetes
11 without neuropathy. You couldn't do it with
12 neuropathy. So, if you want to exclude those
13 patients from the studies, it is not too rational.
14 Certainly, you can't measure outcomes on the basis
15 of hand symptoms, but the electrophysiological
16 studies don't do the job. Therefore, you become
17 exclusionary in a research trial, and this was the
18 only caution I had.
19 Definitely, you don't want to mix
20 lumbosacral plexopathy with a diffuse sensory-motor
21 polyneuropathy. Definitely, you don't want someone
22 who only has hand symptoms and no other evidence of
23 neuropathy at all. That is why you have the rules.
24 But I am not sure that someone with carpal
25 tunnel, for example, should just be eliminated.
1 Now, this is really fine detail for this committee
2 but that was the purpose of the paper. And yes; I
3 would like to do a prospective study and find a
4 good electrophysiological measure so that we could
5 send patients back to the neuroconduction lab.
6 At this stage, I basically give them a
7 trial of therapy. I am not very convinced of any
8 good measure.
9 DR. KATZ: Dr. Bitetti?
10 DR. BITETTI: I wanted to make a comment
11 that I think that how the drug gets labeled is
12 going to be relevant to the entry criteria in some
13 ways, too, because it seems to me that if we are
14 going to have very, very narrow entry criteria,
15 because we are now telling industry how to set up a
16 drug trial, that the more narrow we make it, are we
17 then going to only give them a label for that very,
18 very narrow section of this type of diabetic
20 I know I am jumping ahead, but depending
21 what we decide about broad versus narrow labeling,
22 I think that drug companies certainly want to think
23 about whom they are entering in their original
24 studies if that is going to determine whom they get
25 a label for.
1 DR. KATZ: Is there a regulatory
2 perspective on that issue?
3 DR. McCORMICK: There is no question that
4 the entry criteria that you set forth and use for
5 your clinical trials has an effect on what you have
6 in your label but I guess the question that I would
7 turn back to the committee would be how relevant do
8 you think, or how extrapolatable do you think, the
9 more narrowly defined population would be to the
10 general population of patients with diabetic
12 DR. KATZ: Dr. Foster?
13 DR. FOSTER: I simplistically liken this
14 in study design to a football field where, on each
15 side of the field, you have got the yard arms. How
16 wide those goal posts are going to be for the entry
17 criteria and how wide those goalposts are going to
18 be for the exit criteria for an evaluable patient I
19 think is something that the agency needs to think
20 about from the standpoint of the nature of this
21 disease, the amount of drugs, all of the issues,
22 the comorbidities that we have talked about this
23 morning, so that we wind up with studies that are
24 generalizable to a broad population of folks but
25 that subset analysis done on the other end of the
1 field after the study is over, post hoc
2 analysis--and there are designs that will
3 accomplish that.
4 So it would seem to me that what I have
5 heard this morning, not being a neurologist, is
6 that there may be a multifocal scoring system that
7 would involve both subjective and objective
8 criteria for entry and a scoring system that would
9 be agreed upon by the investigators for at least
10 pre-inclusion of those folks.
11 So, they might be Level 1, 2, 3, kind of
12 like a New York Heart Association classification
13 analogy. Then there would be a post hoc subset
14 analysis, but being able to move the goal posts on
15 both sides. I think those types of designs are
16 important to consider with a disease like this that
17 is so multifactorial.
18 DR. KATZ: Dr. McLesky?
19 DR. McLESKY: I would agree with
20 everything that has been said. In fact, obviously
21 from an industry perspective, we would like the
22 broadest claim that is reasonable. In fact, if we
23 limit the enrollment criteria or tighten it down
24 so, so finely, would the generalizability be lost,
25 number one. On the other hand, the tighter the
1 enrollment criteria, the greater the likelihood we
2 will actually be able to show a result.
3 On the other hand, the tighter the
4 criteria we have, the slower the enrollment
5 potentially would be which is also adverse. So it
6 is a delicate balance between the two extremes.
7 DR. KATZ: Dr. Dyke?
8 DR. DYKE: Yes; I agree. You know, an
9 indication that I thought makes sense is that for
10 the metabolic diabetic sensory polyneuropathy, the
11 aim is prevent or ameliorate the symptoms and
12 impairments of diabetic polyneuropathy. That is
13 broad. And that is doable, as I see it. And that
14 is measurable because you clearly have, then--and
15 by impairment, I mean, broadly, impairment,
16 neurological signs, nerve-conduction abnormalities,
17 other tests of abnormalities.
18 What we are all looking for is an
19 honest-to-god effect. If one really could prevent
20 diabetic polyneuropathy, even the first five yards,
21 hopefully, the next fifty yards would be
22 preventable also, or if we could turn the direction
23 of neuropathy from worsening to holding the same or
24 even improving, that is what we want.
25 So I think most of us are looking for
1 really hard evidence that a drug is efficacious.
2 Does it really affect the development or the
3 worsening of symptoms and impairments taken
4 broadly. My concern is, though, that for
5 regulatory purposes, they must not make the shoe
6 fit all persons or all diseases, was the point I
7 was making, not to diminish the scope of what we
8 are trying to do.
9 DR. KATZ: There is a question hanging in
10 the air and I just want to make sure that we
11 address it. Obviously, a drug that is effective
12 for hypertension is not effective for every person
13 with hypertension and a drug to lower your
14 cholesterol is not effective for every person with
15 high cholesterol. So, given the fact that no
16 matter what disease a drug is effective for, it
17 doesn't work for everyone with that disease.
18 Do people around the table feel that it
19 would be inappropriate to conduct trials in
20 patients with well-defined diabetes polyneuropathy
21 and not cranial neuropathy and radiculopathy and
22 vitamin deficiency and those sorts of things? If
23 those trials showed efficacy, do people around the
24 table feel that it would be inappropriate, then, to
25 label the drug as being efficacious for diabetes
1 polyneuropathy or is that too inappropriate a leap?
2 Specific comments about that question?
3 Dr. Farrar?
4 DR. FARRAR: Dr. Katz, what you have done
5 is to say what is the question again. I think the
6 issue, if you want a broad indication, then you
7 have to show that your drug works in the population
8 that it is intended to work in. I agree with what
9 Dr. Dyke has said about the potential differences
10 in the underlying mechanisms for the disease.
11 On the other hand, there are two ways of
12 approaching it. One way is to look for only that
13 segment of the population, test your drug only in
14 that segment of the population. It probably costs
15 a little less although enrollment will be a
17 But if you only show it in that one
18 segment, then I think there is a reason to believe
19 that you should get a label for only that one
20 segment. What would make much more sense is to do
21 something along the lines of what Dr. Foster was
22 suggesting but post hoc suggests that you think of
23 it later and do it later.
24 I think you actually plan the study with
25 the intent of looking at the global outcome in your
1 entire group and then you specifically state in
2 your goals that you are going to look at the
3 various subsets. If it turns out that your drug
4 only works in one of those subsets, then that is
5 the subset it should be used in.
6 If it turns out that it works, as you have
7 suggested, in sort of the same number of people but
8 in each of the various pieces that you want to
9 divide it into, then you get a general indication
10 because, as you say, not every drug works in every
12 I think there are just some very clear
13 ways to approach it that make sense, and the same
14 with other diseases. If you have got different
15 kinds of hypertension and your drug works in all of
16 them, then that's fine. If it doesn't, then you
17 should use the specific one.
18 The last thing I think that is important
19 is that there is a lot of concern about inefficient
20 trials if you have lots of different potential
21 etiologies and you end up with relatively small
22 differences between your groups. That is very true
23 if you insist on a mean value.
24 But there are ways to look at the data
25 which cost only a very little bit in terms of the
1 number of patients you need that actually allow you
2 to find very small differences between groups
3 without huge numbers, and it has to do with the way
4 in which the analysis is conducted.
5 But you need to decide a prior what you
6 are going to do and then you need to do it. I
7 think whichever mechanism you pick, you ought to
8 get a label appropriate for that.
9 DR. BRIL: Before we go far with the
10 splitting, I need to ask Peter and, perhaps, you,
11 how you are identifying these subsets. When I see
12 my patients, maybe sometimes they don't have much
13 retinopathy or nephropathy but they usually have a
14 bit. I am not seeing these subsets so clearly in
15 my clinical evaluation, my nerve conductions or
16 QST, the things I can measures.
17 Sometimes, yes; they have just gone on
18 insulin. They have insulin neuritis. I agree with
19 that. That is really rare. The common patients I
20 see, I can't split yet. Can you split them for us?
21 DR. DYKE: No; I didn't want to go that
22 far. I think, obviously, we shouldn't think of
23 ocular-motor neuropathy as a component of diabetes
24 sensory polyneuropathy. Yet, in many industry
25 trials, you know, from being a reading and
1 quality-assurance center, a lot of diabetologists
2 say it is diabetic polyneuropathy. They just lump
3 it together and I think we shouldn't do that.
4 And we should make clinical distinctions
5 of carpal-tunnel syndrome and ulnar neuropathy.
6 And you do that at a clinical level. I don't think
7 you and I would differ on that.
8 The difficult ones are the sensory
9 polyneuropathies from coexisting causes which does
10 happen. It has happened to me. I have had a
11 patient in a trial and, four years later, I have
12 discovered that her brother had the same sensory
13 neuropathy and so, clearly, there was at least the
14 possibility that that sensory neuropathy was
16 All I was saying is that it may turn out
17 that, even in the sensory polyneuropathy group,
18 there may be different causes and if we can pick
19 them out, we should try to do that.
20 DR. KATZ: Dr. Cornblath, you were
21 actually on deck next. Do you still have a
23 DR. CORNBLATH: I keep going back to what
24 was said very early which is we are sort of putting
25 the cart before the horse, and that is we need to
1 hypothesis-drive these entry criteria to what it is
2 we think we are doing. So we have designed a
3 study, or at least the main criteria Peter
4 suggested, were for people who had symptomatic
5 diabetic polyneuropathy in which I am presuming
6 that the outcome was to slow progression.
7 But we could design a study where nobody
8 had neuropathy if we were hoping to prevent--if we
9 had a drug that we thought, in people who had
10 diabetes but who didn't have neuropathy, would
11 prevent the development of neuropathy because this
12 data, again, has already shown that there is a
14 So what I want us to be clear--and I do
15 believe that there are lots of these little
16 subsets, depending on how far down you want to
17 drill subjects, you can drill them into large,
18 small and motor-fiber function. You can drill them
19 by quantitative sensory testing variabilities. But
20 that doesn't matter until you decide what you think
21 you can affect.
22 For the industry people, I think the
23 problem is that there is no yet effective drug.
24 One of the things that we talked about this morning
25 in Eva's summary was part of that was inability to
1 predetermine, based on the expected outcome
2 criteria, what might happen in the population.
3 So it would be important, again, if you
4 thought you were going to do NGF again, you would
5 want to enrich your population or pure your
6 population. In people who had some small-fiber
7 dysfunction, that was measurable at a site where
8 you thought you could change it. That would be
9 completely different when we looked at, for
10 example, NT3 which, unfortunately, has died. But
11 NTe is a large-fiber neurotrophic agent so you
12 would want people where you had large fibers.
13 So they are all available. They are all
14 in there but I think it needs to be
15 hypothesis-driven based on your drug rather than
16 this black box of, "We will do something with
17 diabetic neuropathy."
18 DR. KATZ: I am going to try to go in
19 order. Dr. Woolf, you were next.
20 DR. WOOLF: To me, there seems to be a
21 confusion, at least in my mind, between the issue
22 of a proof-of-concept trial where the entry
23 criteria may have to be very tight to prove that
24 the drug has an action and a second trial after
25 that, where the generalizability could be tested.
1 I think the criteria of entry for those
2 two kinds of trials may be very different.
3 DR. KATZ: Dr. Shafer, you were next.
4 DR. SHAFER: Actually, I appreciate Dr.
5 Cornblath's going before me because this is really
6 just following up on your comment, this particular
7 commentary. I wonder if our taxonomy is correct in
8 focusing so much on disease and not on mechanism of
9 drug action.
10 Presumably, there is some mechanism by
11 which the drug is acting which is why you think it
12 might be effective. If we think the drug is acting
13 in the periphery then, perhaps, the indication
14 would be a demonstration of a peripheral disease
15 which would be something like the nerve biopsy
16 where you say, okay, we can see that there is
17 degeneration of the small and myelinated nerves and
18 that will be the population which we think will
20 On the other hand, if we think the drug is
21 neuraxial in nature, then the entry criteria might
22 be demonstration of a response to a neuraxial
23 challenge as we talked about earlier. So, perhaps,
24 a way of thinking about it is not to try to
25 stratify patients by disease but rather what we
1 think is the mechanism by which the drug will work
2 and demonstration that mechanism is likely to be
3 effective in these patients.
4 DR. CORNBLATH: I think we are saying the
5 same thing.
6 DR. KATZ: Dr. Feldman.
7 DR. FELDMAN: Really, I was just going to
8 essentially say what Dr. Cornblath said that maybe,
9 at this point, rather than talking about entry
10 criteria if we talked about potential endpoints
11 with some idea if we were talking about a
12 small-fiber drug, a large-fiber drug or a drug that
13 may be efficacious in both types of disease, we
14 might be able to make some headway.
15 DR. KATZ: In lieu of making headway,
16 let's do something else. Just to deal with the
17 final issue that I think we need to deal with on
18 entry criteria and then I promise we will go on to
19 outcomes. I know it is very exciting for
21 Let's just talk, for a moment, about
22 glucose control and how that should be dealt with
23 at entry. That is a big question that comes up all
24 the time. We have heard suggestions that symptoms
25 may change in the context of increasingly tight
1 glucose control, that that can have an effect on
2 nerve physiologic monitoring.
3 What level of glucose control is required
4 has implications for the ultimate target population
5 that we are trying to generalize to so I wonder if
6 anyone has any comments about how to deal with
7 issues of glucose control upon entry into such
8 clinical trials.
9 Dr. Feldman, would you like to continue?
10 DR. FELDMAN: This is making headway. I
11 can just tell you my experience in being a
12 neurologist, not an endocrinologist, but from my
13 endocrinology colleagues who are always involved in
14 these trials, what they target for is stable
15 metabolic control, as Dr. Dyke mentioned, and
16 hemoglobin A1s in the range of 8 to 9 are
17 frequently maximum.
18 I think some trials have even accepted
19 hemoglobin A1s up to 10, but it would be unusual to
20 be greater than that. That is usually used as the
21 cutoff, hemoglobin A1, of course, in someone who
22 meets the ADA criteria for diabetes. The stable
23 metabolic control is, though, a very loose
24 definition in my experience and it is kind of the
25 endocrinologist's impression whether or not the
1 patient has been under stable metabolic control.
2 DR. KATZ: Is that the problem, that that
3 is not tightly defined enough what stable metabolic
4 control means?
5 DR. FELDMAN: When I was younger and more
6 naive, I thought that when I entered all these Type
7 2 patients into my studies, into our studies, that,
8 just because they would see us so frequently, they
9 would enter and get really better controlled. But
10 when these studies are long, which they are, a
11 year, two years, three years, sometimes there is a
12 small dip in control but usually they do have
13 stable metabolic control. It is not that entering
14 into a study--and that has really been our
15 experience at the University of Michigan and I
16 think that is a relatively global experience that
17 sometimes there is a small effect. But, usually,
18 how they were controlled is how they will go back
19 to being controlled.
20 Dr. Dyke?
21 DR. DYKE: Can we ask Dr. Ed Bayster maybe
22 to talk about this? We had a meeting recently with
23 a series of diabetologists, and he is a
24 diabetologist, where this issue was discussed at
25 some length. Ed, are you here?
1 The issue from my point of view is, for
2 John's sake, why don't we include these people with
3 very high blood-sugar levels because they cannot,
4 or they will not, get good control and they are the
5 ones that need ancillary treatment. So, Ed, why
6 are you making this fuss about metabolic control?
7 DR. KATZ: If you could just start with
8 any relevant disclosures. Those are the rules.
9 DR. BAYSTER: I appreciate that. My name
10 is Dr. Edward Bayster. I am a clinical research
11 physician with Lilly Research Laboratories in
12 Indiana as well as Clinical Associate Professor at
13 Indiana University School of Medicine.
14 The issue has come up a number of times,
15 as we have discussed trial design, on a number of
16 different levels. The issue at stake or at hand is
17 glucose control in the patient population which we
18 would like to study. The patient population is a
19 patient population with diabetic neuropathy and
20 there are a number of epidemiologic studies that
21 have been done over the years that have pointed out
22 that this particular group of patients, on average,
23 have hemoglobin A1C levels that are much higher
24 than the population, the diabetic population, in
1 The question has always been is this the
2 cause for their neuropathy or, alternatively, are
3 they unable to obtain a better control because of
4 the fact that they have high glucose levels. So
5 that is the one side. On the other side, on the
6 regulatory side and from an approach to a study and
7 an ethical side in terms of taking care of these
8 patients, we want them to have the absolute best
9 control that they can during the course of a
10 clinical trial or any kind of study.
11 With that in mind, the ideal patient would
12 have diabetic neuropathy and perfect glucose
13 control when they come into the study. However,
14 because of the fact that that perfect glucose
15 control, or that better glucose control, is
16 oftentimes impossible in that patient population,
17 what turns out or what is good enough.
18 So there are a number of strategies that
19 one can then implement in an attempt to offer that
20 patient the best glucose control during the study
21 by offering all the metabolic glucose-lowering
22 drugs that are available to do that, to lower
23 glucose and to offer the best control. Then that
24 patient, many times, is entered into the study.
25 One approach that has been population that
1 Julio Rosenstock actually published a very nice
2 abstract on about a year and a half or two years
3 ago at the American Diabetes Association suggested
4 that, quite possibly, three months of metabolic
5 control before the study for any patient with
6 glucoses that were under 12 or 13--hemoglobin A1Cs
7 under 12 or 13 percent, offered them the optimum
8 chance for the best glucose control they can and
9 that if, indeed, at the end of that period they had
10 not gotten down to the magic 7 or 8 percent, that
11 they were as good as they could be and that it
12 would be ethical, then, to continue or to study
13 that patient for diabetic neuropathy with the idea
14 that we had ethically offered them the best
15 metabolic control or the best care that we could
16 with regard to their glucose control.
17 Many times, that included insulin therapy
18 and in the implementation of insulin therapy.
19 Certainly, for long-term clinical trials for
20 polyneuropathy where adding insulin over the course
21 of a three or four-year study can confound the
22 outcomes. It offered the opportunity to actually
23 start insulin in those patients in the three-month
24 period to then take one more confounding factor out
25 at the end of the day when you come forward with
1 your results.
2 So that is one possible approach to the
3 problem. I hope that helps.
4 DR. KATZ: Thank you very much. I
5 appreciate that. Any other comments about the
6 issue of glucose control upon entry? Dr. Farrar?
7 DR. FARRAR: I think, at the end of the
8 day, the question becomes why do we worry about
9 their level of control. I think the answer to that
10 is because it has been well-demonstrated that
11 improving glucose control helps all of the
12 potential side effects of diabetes and, therefore,
13 what you need is a measure at the beginning and the
14 end of your trial that will accomplish what Dr.
15 Feldman was commenting on which is that there is a
16 stable level of whatever measure it is over the
17 course of the trial. That is what you are looking
19 I think there are a number of ways of
20 doing that but, from a regulatory perspective, what
21 would make sense to me is to ask that the measures
22 of anything that would potentially influence the
23 outcome of the trial be measured before and after
24 to be able to assess whether it had an influence
25 over what happened within the trial.
1 One last comment on it which is that, as
2 somebody whose primary interest is in studying,
3 actually, the clinical care of patients, meaning
4 not the efficacy study which clearly needs to be
5 done but looking more at the way in which patients
6 are actually treated, you can do that. The issue
7 with randomization is that you even out the two
8 groups. If you add variance to your base
9 population, you just need to expand the size of the
11 I think there are ways of handling it.
12 What you need to be able to do from a regulatory
13 perspective and from an interpretive perspective is
14 to know what has happened to your patients over the
15 course of that period. Whether the industry
16 decides to have a three-month run-in or whether
17 they decide to only use one particular group or
18 whether they decide just to wing it and see what
19 happens, if they have got the before-and-after
20 measures that are responsive enough to see the
21 difference, then, from a regulatory perspective, it
22 should be fine.
23 DR. KATZ: Dr. Bril, you were next.
24 DR. BRIL: I guess my question to
25 everybody would be since we know that, despite best
1 efforts, there are patients out there with poor
2 control, and since they are the ones with the most
3 frequent neuropathy and since, no matter how we
4 know that improving their control will reduce
5 complications, that the patient has to buy into it.
6 If they won't do it, is it truly ethical
7 to leave them out of these research trials? We are
8 ignoring them. We are saying, "You can't control
9 your sugars, you can't come into this study." They
10 are out there. They have the complications. This
11 is not too ethical to me because they have gone
12 through efforts to control their sugars.
13 They are on multiple oral hypoglycemics
14 and insulin and they are still out there with the
15 neuropathy or whatever it is. So the ethics to me
16 seem to be that we are excluding them. So I would
17 ask the agency for their comments on this. Could
18 these people, if you have made efforts, come in and
19 be randomized? This really bothers me.
20 DR. McCORMICK: There may be a number of
21 different ways that you could deal with the
22 patients who have particularly severe control. For
23 one thing, randomization should take care of some
24 of that.
25 The other thing that you might consider
1 doing is stratifying the group, looking at those
2 with poor controls separately from those--or
3 stratifying before you randomize so that you have
4 the groups that are poorly control compared the
5 well-controlled groups so that you could have a
6 trial that would include all of those.
7 DR. BRIL: So there is no set level or
8 number of A1C that they must hit?
9 DR. McCORMICK: No.
10 DR. BRIL: This is more industry-driven
11 that it is--or is it FDA-driven?
12 DR. McCORMICK: This may have been
13 FDA-driven at one point. I think this is something
14 we have really done a lot of soul searching about
15 for the same reason. I think our position has been
16 more recently that we need to make certain that
17 patients have the advantage of good--that, during
18 the trial, they have best efforts made to insure
19 good control so that they are not left to flounder
20 during a very prolonged trial, but not to
21 necessarily exclude them for poor control.
22 DR. KATZ: It sounds like many of these
23 comments, at least tangentially or at least
24 indirectly, endorse the proposal that we heard to
25 give patients the opportunity to have the best
1 control possible during some sort of baseline
2 stabilization period, then include them all and
3 either stratify them or account for them some way
4 in the analysis. I haven't heard any criticisms of
5 that approach yet.
6 Dr. Feldman?
7 DR. FELDMAN: I just wanted to comment.
8 One of the ideas I understood from our
9 endocrinology colleagues is that when patients are
10 in relatively good, loosely good, metabolic
11 control, you are not going to be adding a lot of
12 other medications. If they are under poor control,
13 even if they are in a trial, then they are going to
14 warrant other medications.
15 I can think of an example of three or four
16 patients in the Zenarestat trial that we had whose
17 control began to become relatively poor where their
18 met forman was increased. Two of them bumped their
19 creatinine. Was that from the Zenarestat which was
20 discontinued because of high creatinine or was that
21 from the met forman?
22 So my endocrinology colleagues have told
23 me that one of the reasons we are doing this is
24 because we don't really, truly understand the
25 toxicity of these drugs. As we keep adding
1 different glucose-controlling agents to the drug we
2 are studying, we may be getting interfering
4 Again, the FDA would know much more than
6 DR. McCORMICK: Again, you do have a
7 control group that you can use to sort out some of
8 those adverse events.
9 DR. KATZ: Ms. Delph, you were next.
10 MS. DELPH: I think we need to
11 differentiate between stability of control and
12 level of control because it seems to me that
13 stability of control talks about your trend, what
14 is the trend of your hemoglobin A1C or whatever
15 whereas the level of control says whether it is
16 good, bad or whatever.
17 I think that, for individuals who have
18 poor control, ethically, it would be important to
19 ensure that every effort is made to improve that
20 control but equally ethically, once every effort is
21 made, they should not be excluded simply because of
22 the level of their control.
23 If we are talking about stability as well,
24 and, personally, I think that would be important,
25 to look at the trend pre-intervention in terms of
1 glucose control but, also, I wonder whether it
2 would not be important to look at the trend of the
3 progression of the neuropathy before intervention.
4 DR. KATZ: Thank you.
5 Other comments?
6 DR. ARONSON: As I listen to the
7 discussion, it strikes me that there are two
8 perfectly laudable issues on the table. One is
9 what can we do that is most fair to test the
10 efficacy of the drug that we are wishing to test to
11 begin with and wouldn't we allow ourselves the
12 chance to do that best by only including those
13 patients that are best controlled.
14 On the other hand, it is certainly true
15 from an ethics standpoint that, by eliminating
16 those other patients, perhaps the greater majority
17 that are not likely to be tightly controlled, are
18 we serving the best good in the best way. I think,
19 again, it sort of comes back to what is our point.
20 Are we wishing to test the efficacy of
21 these drugs and design a trial in order to do that
22 and then should we go forward and see how we can
23 serve the better good the greatest. I just with to
24 have that point be made as well.
25 DR. KATZ: Dr. Cornblath.
1 DR. CORNBLATH: I think where a lot of
2 this started in trials before was that there was
3 this general association with poor control and more
4 severe neuropathy. Where the more severe people
5 were eliminated was because, on average, they had
6 more severe neuropathy and it was thought they were
7 less likely to respond to the agents.
8 So I don't think it was necessarily done
9 because there was some level of inequality or
10 discrimination but, again, it was hypothesis
11 driven. It was driven by, "We don't think this is
12 going to help them. Why should we put them in the
13 trial," A, it won't help them and B, it will hurt
14 the trial. So I think that is where the genesis of
15 this was and we shouldn't forget that.
16 So I think there is good hypothesis reason
17 for certain of the drugs to exclude these people
18 because you don't really think you are going to
19 make an effect. That, then, may have implications
20 for what you say in your "label," but there were
21 reasons to do it long ago.
22 DR. KATZ: Dr. Dyke?
23 DR. DYKE: It is clear from the
24 epidemiology data that the risk factors for
25 severity of disability sensory polyneuropathy are
1 other microvascular complications, notably indices
2 of neuropathy. The second most important one is
3 the average glycated hemoglobin control times the
4 duration of diabetes. The third one is type of
5 diabetes. So, in some ways, I have always argued
6 that it would be an advantage to take the more
7 severe neuropathies for the restrictions of
8 ancillary treatments because this is really what we
9 are talking about. We are talking ancillary in
10 addition to glycemic control. We are talking about
11 ancillary treatments.
12 So I would like it to be used for those
13 more severe patients. There is a further reason.
14 The people who have the more severe neuropathies
15 tend to worsen to a greater extent over time than
16 do the mild ones. So, from an industry point of
17 view, you stand a better chance of showing an
18 effect given that the drug works in those more
19 severe ones because the changes are more rapid over
21 So I can see the diabetologist's point of
22 view. The concern about not putting very severe
23 diabetics into the study I think comes from the
24 diabetes community. They feel it is sort of
25 unethical, in a sense, if I can speak for them, to
1 ignore their blood-sugar control and stick them
2 into a study, you know, this sort of idea.
3 So people want to cut it off at some
4 level. I have always had the inclination that they
5 should all be in there, especially those--that you
6 should have a window of time when you encourage
7 them to have good blood-sugar control. You might
8 even have an algorithm of how you do that when they
9 exceed the levels that you set.
10 But, at some point, you would allow them
11 in. But I think it is coming from the diabetic
12 community and, of course, one would have to defer
13 to them for this decision about metabolic controls.
14 So one lives with it. But if I had a preference, I
15 would include some of those more severe ones in
16 these studies.
17 DR. KATZ: Ms. Delph?
18 MS. DELPH: Thanks. My comments were just
20 DR. KATZ: I am going to summarize this
21 discussion and move on to the outcomes measurement
22 issue. What I am hearing so far about the entry
23 criteria are the following, and someone can tell me
24 if I am getting it backwards.
25 First is that there is a need to decide a
1 priori what type of neuropathy you are trying to
2 treat based on what you think your drug ought to be
3 doing and then you need to select people whom you
4 think will be responsive to that treatment.
5 Characterization of your neuropathy may
6 depend upon excluding other types of diabetic
7 neuropathy than distal-sensory polyneuropathy such
8 as thoracic radiculopathy and cranial neuropathy et
9 cetera. It also may involve the exclusion of
10 nondiabetic neuropathies that may mimic diabetic
11 sensory polyneuropathy like vitamin deficiencies,
12 et cetera.
13 The diabetes, itself, needs to be
14 diagnosed and the comment we have is that it should
15 be ADA criteria. The diabetic neuropathy, itself,
16 should be diagnosed by a composite measure which
17 includes clinical as well as electrophysiologic
18 criteria. What I seem to hear is that there is no
19 specific standard about which composite diagnostic
20 approach one should use. Someone correct me if I
21 am wrong on that since it seems like there were a
22 number of currently available approaches. So,
23 correct me if that is not right.
24 Of course the severity of the neuropathy
25 should be staged at baseline based on whatever
1 approach is chose. Lastly, glucose control upon
2 entry or upon randomization should be stable and
3 that stability is important but that tightness of
4 control is not, by itself, necessarily required
5 even though it could influence outcome and,
6 therefore, should be accounted for either in a
7 stratification or as a covariate or predictor of
9 Did I miss anything big or get anything
10 wrong? Dr. Cornblath?
11 DR. CORNBLATH: I would probably change
12 the word when you say clinical features plus
13 neurophysiology. I would probably change it to
14 ancillary studies so that potentially skin biopsy
15 could be included in that group so there would be
16 quantitative sensory testing, nerve conductions of
17 a variety of types, autonomic function, whatever
18 there is in the term ancillary studies so that none
19 of them are excluded.
20 Outcome Measures
21 DR. KATZ: Great. Outcome measures? What
22 do we know about what constitutes an outcome
23 measure that is meaningful to patients. We have
24 touched on this in a lot of different ways before.
25 Maybe someone could just boil it down into what
1 people think would be the optimal choice for
2 outcome measure in a clinical trial. Would anybody
3 like to tackle that?
4 Dr. Dyke?
5 DR. DYKE: I have thought a lot of about
6 the issue of outcome measure in diabetic neuropathy
7 especially for trials. I think there really are
8 four major groups of outcomes, perhaps five. The
9 first one is symptoms. The second one is
10 impairments. The third one is test results which
11 Dr. Cornblath just mentioned.
12 Then the fourth one would be tissue
13 alterations. The fifth one might be other outcome
14 measures of how well you are doing in work and
15 leisure and general health measures. I think it
16 depends on the trial that you are doing as to which
17 you choose or the proportion or the ratio of the
18 test which you use.
19 For example, the natural history of
20 symptoms in diabetic neuropathy is quite different
21 than the natural history of impairments. That
22 needs to be taken into account in designing a
24 We have noticed, and I mentioned it a
25 little earlier, that not infrequently a patient
1 gets rather severe positive sensory symptoms; you
2 know, prickling, asleep numbness, pain, lancinating
3 pain or constricting pain, deep aching pain and so
4 on. But, after a period of time, that goes away.
5 So if, for example, you focus on those
6 symptoms and your goal is to modify those symptoms,
7 you probably need a shorter study than you do for
8 impairment. So, what kind of duration are you
9 going to use for the clinical trial depends on what
10 you are going to emphasize.
11 So a symptomatic trial I think should be
12 relatively short because, obviously, you are going
13 to pick the people who are, if you like, in a down
14 phase. Then, by the natural history, people may
15 get better on his own or he may fluctuate. So you
16 are really better not to make that a four-year
18 There now are some symptomatic trials
19 which are positive for periods of a month, six
20 months, and so on. For impairments, the Rochester
21 diabetic study is absolutely rock solid that you
22 need a long time. You simply don't get the power
23 in a study unless you do it for about four years.
24 You can't overcome that by numbers. An
25 insight came with the OCCT. I can't think of the
1 statistician's name. Vera? Eva? The man from
2 George Washington--Kahill--the one who did the
4 DR. KATZ: It will come to you. Go on.
5 DR. DYKE: It is stopping my flow of
6 thought, though, is the problem.
7 Dr. BRIL: Peter Lachin?
8 DR. DYKE: Yes. He told me that they
9 didn't see an effect in the DCCT until four or five
10 years and then the data came in like gangbusters.
11 In the Rochester diabetic study we have shown that
12 you see a large effect at about two years but you
13 don't see it much before that.
14 So, in an impairment trial where you are
15 trying to get separation of the treatment from the
16 placebo group, you simply need time. It is a much
17 more important variable than just the number of
18 patients. So if you are using a composite score of
19 clinical impairment plus nerve conduction plus
20 sensation, I strongly recommend that you have at
21 least a four-year trial.
22 On the other hand, if you do a
23 complications trial, that is you are waiting for
24 the foot to have a plantar ulcer or to have
25 Charcot's joints, you are talking even a longer
1 time, at least data from the Sheffield group and
2 our data would support that. So it really does
3 depend what kind of a trial you are mounting and
4 the endpoint.
5 DR. KATZ: We will have a specific
6 discussion about duration of trials momentarily.
7 That was a very useful introduction. I just want
8 to make sure that we continue to--
9 DR. DYKE: Oh; you are not talking about
10 duration here. I slipped a gear. Sorry.
11 DR. KATZ: No; it will come in handy in a
13 Any other comments about outcome measures
14 that are appropriate in these clinical trials?
15 DR. CORNBLATH: I mentioned this morning
16 again, I think part of the difficulty in this issue
17 is the fact that there is only--the biggest set of
18 data comes from the Rochester group using the
19 specific measures that they have pioneered and are
20 quite good. It may be that, as I mentioned this
21 NIH report, in other populations or with other
22 measures, these time frames may, in fact, be
23 shortened substantially. We don't know that.
24 So, for example, if, rather than looking
25 at the great toe, we looked at the leading edge of
1 where pin sensation was and looked for a change of,
2 let's argue, just for example, a centimeter to go
3 one way, that may occur much quicker and you may be
4 able to detect it quicker, but we just don't know
5 that now. So there is a tremendous need for more
6 natural-history studies in which many more of these
7 outcome measures, or potential outcome measures,
8 would be applied, particularly among other
9 populations because when we admit a patient to a
10 study in East Baltimore for a Mayo-designed study,
11 it is not clear whether, in fact, the rates of
12 change apply to the East Baltimore population.
13 That is one of the several concerns I have about
14 the fixation with four years or two years.
15 DR. KATZ: Dr. Feldman?
16 DR. FELDMAN: Really, just to reiterate
17 what I said this morning, I do believe that a
18 composite score, as a primary endpoint, is a very
19 good idea. As Dr. Cornblath and Dr. Dyke said, a
20 clinical component for that composite score that is
21 quantitative, I do think, though, that it needs to
22 be heavily based towards sensory impairment and not
23 motor impairment and then couple that with two
24 quantitative sensory tests, vibration for large
25 fiber, cooling for more small-fiber modalities and
1 then, finally, couple that with some type of
2 composite nerve-conduction score, the one that Dr.
3 Dyke uses that has five nerves, the perineal, the
4 tibial and the sural with those different
5 components I discussed today, I think is very good,
6 but a composite score.
7 I think that it is important that, at the
8 end of the day, we probably concur or agree that
9 there is not one single measure that would give us
10 the home run but we really do need a composite
11 score for our primary endpoint.
12 DR. KATZ: If I am not misunderstanding
13 you, it sounds like you wouldn't necessarily choose
14 the NIS as your first choice because if its heavy
15 weighting towards proximal dysfunction and motor--
16 DR. FELDMAN: Peter, we were talking about
17 that today in the NIS(LL). The component of it
18 that I think is probably less helpful and it is
19 really based on your own work are the Questions 17
20 through 24 that look at motor strength and the
21 parts that are definitely more helpful are your
22 questions, your two reflex questions and the four
23 sensory questions.
24 So even a modified NIS(LL) or David has a
25 very nice tool that actually looks at gradient
1 changes in sensation, so something that maybe would
2 emphasize sensory more. I know your own work would
3 support that.
4 DR. DYKE: I would agree. Where you do
5 want the weakness score is when you get into more
6 severe varieties because, as you go, for example,
7 into the symptomatic 2As and Bs, you do get muscle
8 weakness and you would want to record it. So if,
9 for example, you focus on a more severe cohort,
10 don't drop off the motor weakness, I would say. A
11 mild one, an early one, I would agree.
12 Could I just speak also to the issue of
13 quantitative sensory testing. A consensus
14 statement has been prepared by a special committee
15 of AAN on which I initially was a member and then
16 dropped. It is going to be published, I
17 understand. Just to reiterate, vibration is a very
18 good measure and there are good algorithms now and
19 fast algorithms and good quantitative approaches to
20 look for the integrity of the alpha-beta sensory
22 Cool is a good measure of A-delta fibers.
23 Heat pain is a very good measure of both
24 hyperalgesia and hypoalgesia. Before patients get
25 hypoalgesia, they get hyperalgesia. Just before
1 lunch, someone talked about that and it is a good
2 marker of small-fiber disease and should be used.
3 So we now have very good approaches for this, I
4 would say.
5 DR. KATZ: It sounds like what I am
6 hearing is that, at the moment, there is no
7 validated composite-outcome measure that would meet
8 everybody's needs since we are sort of
9 deconstructing what has been done until this point.
10 Is that a misunderstanding, or is there a
11 state-of-the-art composite-outcome measure?
12 DR. DYKE: Could we restate that and just
13 say that different composite measures might be used
14 for different studies, for different outcomes.
15 That would be fine.
16 DR. KATZ: Dr. Farrar, I am reading your
17 mind and you are going to make a comment about the
18 clinical meaningfulness of the composite-outcome
20 DR. FARRAR: No.
21 DR. KATZ: You're fired.
22 DR. FARRAR: That will come later. I can
23 never resist talking about that. What I wanted to
24 point out, and Dr. Dyke has actually addressed it,
25 which is that the composite measure needs to meet
1 the needs of the trial. What I think is implicit
2 in the NIS scale is an assumption that you begin
3 with sensory abnormalities. When you get worse,
4 you get motor abnormalities and that the two are
5 somehow comparable, at least in some general
6 magnitude way.
7 So if you have both sensory and motor, you
8 are much worse than if you have just sensory. If
9 you have a little sensory, you are not as bad as if
10 you have a lot of sensory. I don't know that that
11 is true, but that clearly is an assumption of that
12 particular scale.
13 I think it is important to realize that
14 composite scales are nothing more than a bunch of
15 different questions that are added up. There are
16 different ways of adding up the scale. You can add
17 them up as simple numbers. You can multiple one
18 times another. You can weight them differently.
19 You can do what Dick Gracely did with some pain
20 scales, measure them against something else and see
21 who they work.
22 At the end of the day, the real question
23 is what Dr. Dyke said which is what is the question
24 that you are trying to answer in that trial. If
25 you are studying a full range of people, which I am
1 in favor of, you need a measure that will be
2 responsive to change in that full range.
3 If that is motor, then sensory, then
4 reflexes or whatever it is, you need to be sure
5 that that is properly included. One point; if you
6 noticed when the scale was projected there--we
7 couldn't read it, but just by viewing it--the
8 number of questions that are asked in the composite
9 scale defines the weighting.
10 If you put three questions about sensory
11 and six about motor, you mean to say that motor is
12 more important than sensory or it may be that motor
13 is not as detectable as sensory and you need six in
14 order to achieve the same amount of sensitivity as
15 for the three sensory questions.
16 There is a whole science that has
17 developed primarily around psychiatric measures
18 looking at how scales--psychiatric measures and
19 education measures. Actually, a lot of the best
20 work has been done with the SAT scores--but looking
21 at issue of how measures measure.
22 At the risk of suggesting that we don't
23 reinvent the wheel, there are some very reasonable
24 and straightforward processes that you go through
25 to achieve an understanding of what your outcome
1 scale is measuring, especially for composites.
2 Some of the things I have mentioned here
3 are fairly simple but there are also other ways of
4 making the scale do the kinds of things you want.
5 Just to mention one other method, something called
6 the Womack, which is known to a number of you, is
7 used to measure arthritis. The way that scale
8 works is it is graded so that it asks about whether
9 you can walk to the bathroom. It then asks if you
10 can walk a block. It then asks if you can walk a
11 mile. It says, can you go up and down stairs.
12 The whole purpose is that if you can only
13 walk to the bathroom, you get one point. If you
14 can walk a block and, obviously, walk to the
15 bathroom--it is usually not clear that you can't do
16 both--then you get two. If you can do that and
17 then three blocks, you get three, et cetera.
18 So that is a different way to construct
19 the scale. But I would suggest that, in terms of
20 looking at these things, ultimately the issue is
21 whether the scale measures what you want it to
22 measure and whether it gives you the right
23 weighting to the pieces that you want and that
24 depends on what you are studying and how the scales
25 are constructed.
1 DR. DYKE: Can I just agree with that?
2 You know, we never conceived as the scale being
3 sort of locked in cement and, for different
4 purposes, we use different components.
5 But I do want to make the point that
6 neuropathy is the sum-total of symptoms and
7 disparate impairments and test abnormalities and
8 outcomes. You need to be like an auto-body-shop
9 man who goes with his yellow pad to the wreck of a
10 car and writes down, "In this car, the headlight is
11 missing. The front wheel is gone. The motor needs
12 replacing," and adds it up. He ranks. Some score
13 of some kind is needed.
14 In the eyes of fifty good men and women
15 around the table, how you add that up or which
16 components you think may vary and there may be
17 better ways of doing it. But I think the
18 fundamental idea is actually quite good.
19 If I could just go back to the early days
20 of when the Social Security Administration set up
21 the criteria for how you were disabled, they didn't
22 want to just know, can you walk 50 feet and
23 additionally go to the toilet and do you need a
24 stick. They wanted to know has a scientist, a
25 doctor, also examined them and showed that they
1 were also impaired.
2 That is what we have been trying to do is
3 to give it that further evidence. I think it is
4 very good to have life scales and what can you do,
5 but Richard Hughes has a scale where, can you walk
6 seven meters with a stick. Well, it depends on how
7 big the nurse is that is helping you, et cetera.
8 I agree with them, but--well; enough said.
9 DR. KATZ: We are coming up on a break
10 momentarily. Before we use up the remaining couple
11 minutes of our time, I would just like to turn to
12 the FDA folks and see if there are any further
13 questions about polyneuropathy trials that you
14 would like to hear addressed in the last couple of
15 minutes of this session before we move on to pain
16 after the break. Anything else?
17 DR. McCORMICK: I think we have covered
18 all that we wanted to hear about.
19 DR. KATZ: Dr. Bril, you wanted to make a
21 DR. BRIL: My only additional comment--I
22 mean, the scales are just summaries of the symptoms
23 and findings. You should use ancillary tests. As
24 long as we are not locked into ancillary tests
25 because they are going to change first, I think,
1 before the symptoms and signs depending on what
2 they are.
3 The thing with quality-of-life
4 instruments, if we select patients with milder
5 neuropathy to go into these trials because they are
6 the ones who are going to respond, they may or may
7 not have a lot of impairment of quality of life.
8 So, if you are going to look for change in an
9 instrument, it would have to be impaired to begin
10 with. There may need to be more thought about
11 that, or you might have to stratify and subset
12 people so that those who have impairments in
13 quality of life can be measured for outcomes later
14 of improvement or not.
15 But not everybody is going to have a bad
16 quality of life, I think, at the beginning. So I
17 have some concerns about that.
18 DR. KATZ: Ms. Delph and then Dr. Woolf,
19 you will have the last comment. Go ahead, please.
20 DR. DELPH: I would like to urge that
21 whatever endpoint is chosen that it be something
22 that can be interpreted in a way that is meaningful
23 in clinical practice so that adequate judgments can
24 be made about whether or not it is going to be
25 beneficial and how beneficial it is going to be to
1 a particular individual.
2 I also have a question about whether or
3 not endpoints should be separated, for example,
4 between sensory and motor-type functions or between
5 the various types of nerves that are being studied.
6 DR. KATZ: Answers to that question? I
7 think the composite outcome measures that are being
8 proposed do separate out those different things and
9 can be looked at individually.
10 DR. DELPH: But it was unclear if you are
11 looking at one composite endpoint whether or not
12 you are just lumping them altogether or whether you
13 need to lump them together kind of separately
14 DR. KATZ: In other words, would the
15 components of the composite-outcome measure be
16 analyzed separately also as secondary outcome
17 measures so you can gain insight as to whether the
18 motor fibers or the sensory fibers or what have you
19 are improving differentially.
20 Any comments on that? Is it traditional?
21 Is it appropriate to separate out all the separate
22 components or the main domains of the
23 composite-outcome measure and look at them
24 separately as secondary endpoints?
25 The answer is yes.
1 DR. CORNBLATH: Dr. Dyke and I were just
2 talking. This has been done, for example, in the
3 other CIDP trials where the whole NIS was used and
4 then the weakness subset was looked at separately,
5 and you can see dramatic changes in that.
6 So it is certainly possible to do from
7 either a very large composite or even a smaller
8 composite. It is just a question, again, of asking
9 in advance what it is you want to ask.
10 MS. DELPH: And powering the trial to be
11 able to interpret those.
12 DR. KATZ: Dr. McCormick? No? Dr. Woolf,
13 last comment?
14 DR. WOOLF: Just a concern that, in the
15 creation of these composites with all these
16 weighting of these different elements, the
17 assumption may be--it hasn't been stated but it
18 often is implicit that these are linear scales
19 whereas, in fact, they may not be, that in
20 measuring them over periods of time, the
21 sensitivity may be very different at the top end of
22 the scale and at the bottom. So the significance
23 of any change needs to be understood in the light
24 that they may not be linear.
25 DR. KATZ: Yes. Validating these
1 composite outcome measures occupies professional
2 psychometricians and statisticians full-time all
3 around the globe. It is not an activity for people
4 who don't do it on a professional level and we
5 haven't certainly gotten into that discussion, and
6 I don't think we will.
7 But, clearly, the professional nature of
8 that activity needs to be kept in mind for those
9 who would take a peril of inventing their own and
10 seeing how it works.
11 Dr. McCormick, final comment?
12 DR. McCORMICK: No; actually one last
13 question, if I might. One thing that we touched on
14 a little bit this morning and some discussion about
15 pain endpoints and effect sizes I would like the
16 committee to think about in the context of
17 neuropathy trials. Let's say we do finally have a
18 drug that really demonstrates an effect, let's say,
19 in arresting the course of disease or slowing the
20 course of disease.
21 What kind of effect size would you think
22 would be reasonable to see compared to a placebo?
23 I guess we would be comparing the slopes of the two
24 arms of the study, comparing the placebo slope,
25 rate of decline, with the drug rate of decline.
1 What would be a reasonable effect size that you
2 would accept that is clinically meaningful?
3 DR. KATZ: The first question, then, is
4 which measure one would use and the second question
5 is what change in that measure would be considered
6 clinically relevant in a Phase III trial of a drug
7 to slow down the progression of peripheral diabetic
9 So what measure and what change is
10 clinically meaningful? Does anybody want to try to
11 propose an answer to that question?
12 DR. DYKE: We talked about this at the St.
13 Paul Peripheral Nerve Society four or five years
14 ago. At that time, we thought that it ought to be
15 at least two NIS lower-limb points. That is the
16 delta. Now, that sounds like a very small amount
17 but it is definable. It is the least amount of a
18 neurological abnormality that a neurologist can
19 recognize on two sides of the body.
20 But the epidemiology data actually shows
21 that it is hard to get that kind of a result in
22 trials because there is noise in all of these
23 measurements. Time is involved. These patients
24 are being treated with diabetes. But if you, in
25 fact, saw this degree of difference at the end of
1 two years in otherwise well-designed trials which
2 were truly double-blind, rigorously handled, if you
3 saw that kind of data, most of us around that table
4 at that time thought that that would be a
5 meaningful change.
6 I should tell you, there were respected
7 people including P.K. Thomas of London who thought
8 we should just have statistical significance
9 because it is sort hard to get significance in big
10 trials with rough measures and lots of people
11 involved and so on.
12 But we decided that and we needed, first
13 of all, a really well-designed trial, large enough
14 power to do the thing, double-blind and then we
15 ought to have statistical significance, an NIS
16 score of two points. The epidemiology data that we
17 produced came later and it turned out it takes
18 quite a large trial for a long period of time to
19 get that effect on the assumption that the
20 treatment arm of the trial is doing better than
22 So that was our answer at that time.
23 DR. KATZ: I am not sure that will
24 necessarily get consensus on this question right
25 now in terms of the best outcome measure and what
1 the best meaningful change is.
2 Are there other thoughts on that? Dr.
4 DR. FOSTER: It would seem to me, again
5 from the standpoint of the progressive nature of
6 this disease and your question about effect size,
7 is that effect size is not going to be a single
8 determination at X point in time after the start of
9 the study. Instead, it will probably be a series
10 of where you would look at both rate and extent of
11 the change, would you not.
12 So it would seem to me that you would be
13 designing the trial somewhat different than you
14 would, for instance, an antibiotic trial in
15 bacteremia, whether you do or you don't have
16 bacteremia. Is that not true?
17 DR. KATZ: It seems to me that whatever
18 outcome measure is chosen, it has to meet the one
19 criterion of being ultimately linked to some sort
20 of clinical benefit. From what I have heard today
21 so far, myself--in fact, I took notes on
22 this--there are three studies that I heard about
23 that correlate change in some outcome measure with
24 clinical benefit.
25 One was the increase in the vibratory
1 threshold which was correlated with foot ulcers.
2 The second one was the Sosenko study mentioned that
3 correlated changes in the thermal thresholds with
4 clinical outcome, if I heard that correctly. And
5 then there was the change in perineal
6 nerve-conduction velocity which was correlated with
7 clinically evident neuropathy. I am not sure if
8 that really meets the criterion of clinical
9 relevance or not since that sounded like a
10 physician's evaluation.
11 What would people feel about using some of
12 these quantitative measures, vibration threshold or
13 changes in thermal threshold as outcome measures
14 since there already seems to be a benchmark for
15 clinical meaningfulness.
16 DR. BRIL: We had talked about this, and
17 Peter just stepped out, I think, but if you are
18 talking about slowing progression, what we had
19 discussed at one time a few years ago was a 50
20 percent slowing of the rate of progression knowing
21 that there is a more rapid progression in those
22 with diabetic neuropathy compared to age-related
23 changes in nerve function that are usual.
24 So if you could prevent the more rapid
25 decline by at least 50 percent, there was some
1 consensus that that was going to be meaningful
2 regardless of the magnitude of that change, but
3 just that it was a 50 percent reduction.
4 You could look at VPT and see if you
5 prevent people from getting to the 25-volt level
6 that predicts foot ulceration but you would have
7 to, then, know the rate of progression of VPT over
8 the years and that is more problematic. So the
9 prevention of progression in whatever scales may be
10 the way to go, or one of the ways to go, rather
11 than an absolute magnitude of effect which becomes
12 a little problematic since you are not always sure
13 what magnitude you are going to measure with time.
14 DR. KATZ: Dr. Feldman, last word for you.
15 DR. FELDMAN: Thank you. What we have
16 done in the previous clinical trials, depending
17 upon what the primary efficacy point or points
18 were, since we know the rate of decline of these
19 points from Dr. Dyke's work and from the DCCT and
20 other epidemiological studies, what was actually
21 aimed for was only a 20 percent change, a 20
22 percent change from placebo. That is how most of
23 these studies were powered. That is how the
24 zenerestat study was powered, for example.
25 Dr. Arezzo may be able to comment is that
1 is how the Zopolrestat study was powered. I think
2 that is how the Alcar study was also powered. So a
3 20 percent change is what has been used previously.
4 Now, you are going to ask me whether that
5 is clinically meaningful. You know, I simply don't
6 know. I don't think we know if a 20 percent change
7 is clinically meaningful but the thought was a 20
8 percent change in a short duration. These studies
9 are mainly twelve months to two years.
10 DR. KATZ: A fifteen-minute break and we
11 will return promptly to start the pain session.
13 Point-Counterpoint: Extrapolation of Findings
14 from One Type of Neuropathy Pain
15 to Another Neuropathy Pain Condition
16 DR. KATZ: We have a match coming up, the
17 match of the century. There are bets being taken
18 out in the hallway if anybody is interested.
19 We will start the late-afternoon part of
20 our session on pain now. One of the major and most
21 contentious issues as we have already gotten
22 glimmers of today is whether one can extrapolate
23 from efficacy in one type of neuropathic pain to
24 other types of neuropathic pain and, if so, to what
25 extent can on extrapolate and is there such a thing
1 as a drug that works for neuropathic pain in
3 So it was felt by the conference
4 organizers that the best way to address that
5 controversy is to have two of our resident experts
6 take on the different perspectives in that
8 So, without further ado, Dr. Dworkin and
9 Dr. Rowbothom, please share your thoughts on that
11 DR. DWORKIN: The way we are going to do
12 this is I am going to talk for ten minutes. Then
13 Dr. Mike Rowbothom is going to do his prescription
14 and rebut what I have said for twelve minutes, and
15 I am going to have the right of first refusal to
16 rebut what he said in another two minutes or, if
17 all the wind is out of my sails, we will just open
18 it up to questions. By the end of Mike's talk, all
19 the wind might well be out of my sails.
20 I do want to emphasize at the outset that,
21 in terms of the positions we are presenting, the
22 position I am presenting and the position that Mike
23 is presenting, these were assigned to us on the
24 basis of a coin toss conducted by Dr. McCormick so
25 you shouldn't necessarily think that what I am
1 about to say in the next ten minutes and what Mike
2 is going to follow me and present in the next
3 minutes after me is what we believe.
4 This is a true high-school-debate kind of
5 format where we are debating what we were assigned
6 to debate. So, without further ado--
8 My presentation is a brief review of the
9 evidence that supports separate neuropathy-pain
10 indications. In thinking about how to present this
11 evidence, the rationale for separate
12 neuropathy-pain indications, it seems to me there
13 are three types of evidence supporting separate
15 One is that neuropathic-pain syndromes,
16 neuropathy-pain conditions, I think some of you
17 might refer to them as, have distinct patterns of
18 symptoms and signs. The second is that they have
19 unique combinations of underlying pathophysiologic
20 mechanisms. And the third, and arguably the most
21 important, is that there is specificity of
22 treatment response already documented in the
24 I am going to go through each of these in
2 Starting with symptoms and signs, this is
3 data that Dr. Brad Galer, who is in the audience,
4 has published in two articles, in Neurology in '97,
5 in Archives of PMNR in 1998, using the neuropathy
6 pain scale that he and Mark Jensen published.
7 What I think you can see from this slide,
8 and this is a bit of a glass half-full, glass
9 half-empty. What I have done here is the plot the
10 profile of responses in these five groups of
11 patients, postherpetic neuralgia, complex
12 regional-pain syndrome, Type 1, diabetic
13 polyneuropathy, peripheral-nerve injury and
14 Charcot-Marie tooth disease, plot the responses of
15 the patients--these are averages--across these
16 seven items or so on the neuropathy pain scale.
17 I have put an asterisk next to each of the
18 items where there are significant differences among
19 the groups. So you can see, what is it, four of
20 the adjectives, that pain qualities differ among
21 these groups--five, actually; sharp, cold,
22 sensitive, itchy and surface pain distinguish these
23 groups of patients. I think the conclusion of the
24 story, and I quote Dr. Galer, is assuming that pain
25 characteristics may reflect different underlying
1 pain pathophysiologic mechanisms, these data
2 suggest the possibility that the mechanisms that
3 produce postherpetic neuralgia pain may be
4 different than those that produce pain in other
5 neuropathy pain syndromes.
6 So there is a separation amongst these
7 syndromes in their pattern, their profile, if you
8 will, of symptoms, signs.
10 Next, we all, I think, accept that the
11 prevalence of mechanical allodynia is quite
12 different in postherpetic neuralgia and painful
13 diabetic neuropathy. For PHN, there are data. At
14 least three studies have reported the prevalence of
15 mechanical allodynia. This is almost always,
16 although not always, brush-evoked pain. It is
17 about 60 to 90 percent in PHN. I think we all
18 agree that it is quite a bit less in painful
19 diabetic neuropathy. I put down a guess of 20 to
20 30 percent. It might be lower than that.
21 A recent publication out of Israel, an
22 open-label trial of lomotrigine concluded that the
23 mechanical stimuli, paint-brush strokes, pin prick
24 and repeated pin prick, evoked only minimal pain at
25 the first visit indicating that mechanical
1 allodynia was negligible.
2 So not a lot of mechanical allodynia in
3 painful diabetic neuropathy but very prevalent in
4 PHN. So, the conclusion, with respect to symptoms
5 and signs, is that they are different among
6 neuropathic pain syndromes.
8 Moving into mechanisms, Dr. Clifford Woolf
9 who is here with us, published this illustration a
10 number of years ago in The Lancet. I think there
11 is little to disagree with in this overview diagram
12 of the underlying etiologies of neuropathic pain,
13 how those etiologies are a substrate of causes,
14 really, of different neuropathy pain mechanisms
15 that cause different kinds of symptoms, both
16 stimulus-independent pain and stimulus-evoked pain,
17 and then we have neuropathic syndromes.
18 I think one important thing that is not
19 discussed in detail in this diagram is mechanisms.
20 This is plural.
22 We know from other publications of Dr.
23 Woolf's that there are multiple neuropathic-pain
24 mechanisms. So this is a figure from another
25 recent article mechanisms of neuropathy pain and I
1 think nociceptive pain syndromes; ectopic
2 discharges, central sensitization, sympathetic
3 mechanisms. You are all familiar with this.
5 So this is my revision of Dr. Woolf's
6 figure. This I would propose is reality, that
7 there are lots of different mechanisms. Notice, I
8 used a ying-yang icon here to illustrate the fact
9 that this is a debate.
10 These hypotheses, in this figure, are, as
11 I tried to emphasize here, for illustrative
12 purposes only. I don't want to spend any time at
13 all discussing whether I am right in proposing that
14 PHN, the mechanisms of PHN, are central
15 sensitization and what else did I say, sprouting of
16 A-beta fibers into the superficial dorsal horn.
17 The point of this figure, really, is to
18 illustrate that I think we would all agree that we
19 have got neuropathy pain syndromes, PHN, DPN
20 phantom limb or breast pain, trigeminal neuralgia,
21 idiopathic small-fiber sensory neuropathy and that
22 there are a whole lot of different mechanisms that
23 sort in different combinations with respect to
24 determining the pain in those syndromes. In fact,
25 if you look clearly at this fanciful illustration,
1 there are only two syndromes here where I propose,
2 if you will, that the underlying mechanisms are
3 identical and that is diabetic polyneuropathy and
4 idiopathic small-fiber sensory neuropathy.
5 I don't know if that is true or not but,
6 for illustrative purposes, if you believe that
7 mechanism should guide treatment and you believe
8 that something like this is reality, then the only
9 two syndromes on here where you could actually make
10 the extrapolation to treatment response in one
11 implies efficacious treatment, in the other would
12 be those syndromes, diabetic peripheral neuropathy
13 and idiopathic small-fiber sensory neuropathy
14 because the mechanisms of all these others are
15 different combinations.
16 If anyone disagrees with that, I would
17 love to hear the disagreement later on. So let's
18 end up with differential patterns of treatment
21 We all know that the results of
22 placebo-controlled trials, as you can see going
23 back to the 1960s, have established--and these are
24 consistent with clinical experience, of
25 course--have established carbamazepine as
1 first-line therapy for trigeminal neuralgia. But,
2 as I say, at the bottom of the slide, no one thinks
3 carbamazepine is first-line therapy for any other
4 neuropathy pain syndrome.
5 Now, someone sent me a e-mail a week ago
6 who is not even aware of this debate urging me to
7 make the point at this meeting that trigeminal
8 neuralgia shouldn't be considered in this
9 discussion because it is just this peculiar
10 idiosyncratic neuropathy pain syndrome and nothing
11 that one would conclude about trigeminal neuralgia
12 has any relevance to the other neuropathy pain
14 So, even if we accept that argument, there
15 is other evidence of differential treatment
18 Here are two studies and authors of these
19 studies are here with us this afternoon that
20 concluded amitriptyline is not superior to placebo
21 in painful HIV peripheral neuropathy. Of course,
22 everyone in this room, I think, is aware that
23 amitriptyline, for many, many years, has been
24 considered first-line therapy in both diabetic
25 painful peripheral neuropathy and PHN based on a
1 large number, at least 13 and maybe more,
2 randomized controlled trials in those two
3 neuropathic pain syndromes.
4 But here we have, if you will, replicate
5 evidence of the lack of efficacy of amitriptyline
6 in painful HIV neuropathy. Of course, one could
7 quibble with these studies and maybe Dr. Max, who
8 is an author on both of them, will quibble with the
9 conclusion but, in fact, this study titrated
10 patients to 100 milligrams of amitriptyline and
11 this study titrated patients to 75 milligrams of
12 amitriptyline and those are reasonable doses.
14 Finally, and this is my last slide, two
15 studies of dextromethorphan both of which conclude
16 the same. One is about to be published in
17 Anesthesiology sometime in the next month or two.
18 Let me read these by way of conclusion. In the
19 first study published in Neurology in 1997, out of
20 Dr. Max's lab, the conclusion is, "In diabetic
21 neuropathy, dextromethorphan decreased pain
22 significantly relative to placebo. In PHN,
23 dextromethorphan did not reduce pain
25 In the more recent study, dextromethorphan
1 is effective in a dose-related fashion in selected
2 patients with painful diabetic neuropathy. This
3 was not true of PHN suggesting a difference in pain
4 mechanisms between the two conditions."
5 So I think very recent and quite
6 compelling evidence that treatment response on one
7 neuropathy-pain syndrome does not necessarily mean
8 that there is going to be treatment response with
9 that agent in another neuropathy-pain syndrome,
10 even one, as we all thought, as closely associated
11 with respect to treatment response as PHN and DPN.
12 There is also a published study out of Dr. Max's
13 laboratory showing lack of efficacy of
14 dextromethorphan in facial neuralgias of various
15 sorts. I just didn't have room for that on the
17 So that is the end of my talk. The
18 conclusion from these three sets of evidence that
19 on distinct patterns of symptoms and signs, unique
20 combinations, by and large, of underlying
21 pathophysiologic mechanisms and evidence of
22 differential treatment response, even when we
23 wouldn't have expected it, dextromethorphan and
24 amitriptyline in HIV sensory neuropathy I think
25 suggests that one can't make the extrapolation from
1 treatment in one or two or maybe even three
2 syndromes to treatment efficacy in neuropathy pain
3 across the board.
4 Thank you. It is Dr. Rowbothom's turn.
5 DR. ROWBOTHOM: Now for the counterpoint.
6 You would think that, with this debate, it would be
7 natural for Bob to be the lumper and me, as
8 neurologist, be the splitter and Bob, as a
9 psychologist, be the lumper.
11 After all, Bob, where you feel the pain is
12 in the brain. But that is not the way it came out
13 and, in fact, not only was I assigned the lumper, I
14 was assigned to be the lumpier of the two.
15 Although Dr. McCormick apologized for the
16 typographical error in her message, I think it
17 actually fits. No matter how much I grind away on
18 this concept that neuropathic pain could be
19 considered all the same from the treatment
20 perspective, it is still going to be a rather lumpy
21 pudding because there are differences between the
22 different syndromes.
23 So let me just go through a few things.
25 We have many different types of
1 neuropathic pain. They have different mechanisms.
2 They have different clinical presentations. They
3 have varying prevalences, varying diagnostic
4 criteria, all of which makes study of them quite
5 difficult. Most of the trials that have been
6 performed for new drugs for neuropathic pain, the
7 majority have been in diabetic neuropathy and then
8 a smaller number have been in postherpetic
10 There are some syndromes where there is
11 really even a question as to whether or not the
12 pain is truly neuropathic. I had always considered
13 CRPS Type 1 or RSD to be a neuropathy pain because
14 the mechanisms seem to relate to abnormal function
15 of the nervous system. But, even that concept, is
16 being questioned now.
17 I was at a talk that Howard Fields gave
18 last week at our pain-interest group meeting and he
19 was saying that he didn't think it was a
20 neuropathic pain really but an inflammatory
21 disorder. So even all the old concepts are being
24 The problems with spitting are--let's just
25 look at it from a couple of different perspectives.
1 First of all, is there a distinctly different
2 response to defined interventions based on
3 diagnosis. I put there, "Just prove it," and I
4 will go through a little bit of the data that Bob
6 There is a lot of variability in the
7 trials. We are so lacking in information to answer
8 some of these questions that I think it is up to
9 the scientific community and industry to try and
10 really prove whether or not different syndromes are
11 actually different from a treatment perspective or
12 if there is a very broad overlap.
13 So, for example, I.V. lidocaine. Studies
14 that we conducted many years ago and open-label
15 studies prior to that have also suggested that
16 neuropathic pain is much more likely to respond to
17 intravenous lidocaine than other types of pain
18 disorders, especially idiopathic pain or
19 musculoskeletal pain.
20 My experience had been that patients with
21 central pain were quite unlikely to respond to I.V.
22 lidocaine. Then, sure enough, about a year ago, a
23 very nice study by Nadine Natale working in France
24 came out showing that patients with central pain
25 did respond to I.V. lidocaine. So that point is
1 still up for grabs.
2 So there may be a difference between
3 neuropathic pain and nonneuropathic pain for
4 intravenous lidocaine but that, as a group, there
5 doesn't seem to be any specific neuropathic pain
6 syndrome that is particularly unlikely to respond
7 to I.V. lidocaine.
8 Of course, some neuropathic pain disorders
9 are extremely likely to respond to that. Patients
10 with trigeminal neuralgia probably have an 80 to 90
11 percent chance of having their pain greatly
12 diminished or even temporarily abolished with an
13 intravenous lidocaine infusion.
14 Second, what about tricyclics and
15 antidepressants. Tricyclics seem to be pretty
16 broad-spectrum analgesics for neuropathic pain and
17 probably the only type of neuropathic pain that
18 they haven't been well studied in is trigeminal
19 neuralgia. I would argue that, for the patient who
20 is unoperated and, therefore, has a nonmanipulated
21 trigeminal ganglion, the standard of care has been
22 to treat them with a sodium channel-blocking type
23 anticonvulsant like carbamazepine.
24 But, that said, tricyclic antidepressants
25 are very potent sodium channel blockers. There is
1 every reason to believe that they probably would
2 work in this disorder if they were to be tried.
3 There really aren't good prospective negative
4 trials showing that carbamazepine works and
5 something like amitriptyline doesn't.
6 Just to underscore the potency of the
7 tricyclics as sodium channel blockers, there was an
8 interesting small study in the anesthesia
9 literature where they actually showed that, in an
10 animal model, you could produce peripheral nerve
11 block by injecting a tricyclic antidepressant. It
12 was that potent as a channel blocker.
13 The non-tricyclic antidepressants; is
14 there reason to think that they would be unlikely
15 to work in something like trigeminal neuralgia that
16 is so sensitive to sodium channel blockers? Here,
17 there is just no information at all.
18 There are really only a couple of
19 non-tricyclic antidepressants that have much
20 evidence of efficacy and these are all the mixed
21 reuptake or the more adrenergic selective
22 antidepressants with much less evidence for
23 efficacy and, in fact, good evidence that they are
24 not effective for pain with the serotonin-selective
1 So I think, from the antidepressant
2 perspective, we either don't know or can't really
3 make a strong case that there are disease-specific
4 differences in response that are meaningful and
6 For opioids, again, trigeminal neuralgia
7 being somewhat the exception because the pain is so
8 typically phasic by the time you have got the
9 medication into your system to try and treat an
10 attack of tick, of the electrical jabs in the face
11 that are characteristic of that, the attack would
12 have ended. So that study has never really even
13 been attempted.
14 For all the other types of neuropathic
15 pain, to the extent that they have been studied,
16 there isn't a clear distinction showing that one
17 type of neuropathic pain is very responsive and all
18 the other types of neuropathic pain are
19 unresponsive. In fact, the problem is there is
20 really just too little study and almost no
21 published full-length papers on opioids for
22 neuropathic pain.
23 Anticonvulsants; again, probably the best
24 data is from the trials of gabapentin and
25 pregabalin where, for the most part, all the
1 different neuropathic-pain disorders that have been
2 studied have been found responsive to that
3 particular pair of anticonvulsant drugs and, with
4 the exception of carbamazepine for postherpetic
5 neuralgia where it failed and more recently to
6 piramate, which is a sodium channel blocker as well
7 as having other effects for diabetic neuropathy,
8 there hasn't been a lot of selectivity in that drug
9 category, either.
10 Perhaps one could make a case that the
11 topical medications, capsaicin and topical
12 lidocaine, are selective for postherpetic neuralgia
13 but that is probably, in part, at least, because
14 they have not been studied systematically for
15 disorders other than postherpetic neuralgia.
16 The next point I want to bring up is is
17 there such a thing as a pure neuropathic-pain
18 syndrome. The answer to that is yes and no. A
19 patient with acute Zoster has neuropathic pain but
20 they also have got tremendous inflammation along
21 the peripheral nerve trunk, changes associated with
22 inflammation all the way from the dorsal root and
23 the dorsal horn of the spinal cord all the way out
24 to the skin. So that is not a pure neuropathic
25 pain. Only when they end up in a chronic phase,
1 six months or more after their original Zoster
2 insult would they be considered more or less a pure
3 neuropathic pain.
4 Patients with spinal cord injury, you
5 could argue that that is pure neuropathic pain but,
6 probably, really in the circumstance where they
7 have something like an arterial-venous malformation
8 that produces a spinal-cord stroke. But the
9 majority of the patients that are going to be seen
10 in practice are patients with traumatic spinal-cord
11 injury and they may have associated spinal
12 fracture, internal injuries or other kinds of
13 tissue damage from the original injury that can
14 give them multiple reasons for their neuropathic
15 pain in addition to the spinal-cord injury.
16 Patients with multiple sclerosis may have
17 many lesions. When we tried to do a study some
18 years ago that included M.S. patients, I was
19 surprised to see how many of them presented almost
20 more like a fibromyalgia pattern. They hurt all
21 over. They had definite multiple sclerosis. They
22 met all the laboratory and imaging criteria for
23 that disorder but their pains were not focal and
24 associated with distinct abnormalities on neuralgia
25 examination that one would typically associate with
1 central pain such as would be seen with thalamic
2 stroke, for example.
3 Then, as Mitchell brought up a short time
4 ago, what about back pain or neck pain with
5 radiculopathy. The problem there is that
6 relatively few patients have a pure radiculopathy
7 without associated chronic neck pain or without
8 associated chronic low-back pain.
9 I think that the people here from industry
10 would probably agree that trying to study low-back
11 pain with or without radiculopathy is almost the
12 third rail of trying to do studies of agents
13 primarily intended for neuropathic pain. It is a
14 tough population to work with. Because of all the
15 mechanical factors involved, then it is a difficult
17 I didn't put on this slide patients with
18 CRPS or RSD which is another complicated group to
19 work with. They are difficult to recruit and they
20 often are quite unsuitable for clinical trials
21 because their pain disorders are often tied up in
22 litigation of one type or another.
23 The next point is what about the
24 diagnostic certainty. Some of these disorders, one
25 can make a quite confident diagnosis. A patient
1 with classic trigeminal neuralgia with, perhaps,
2 imaging evidence of an aberrant artery and no
3 underlying sensory deficit, that is a pretty firm
5 If you see a patient with acute Zoster and
6 they continue to have pain, that is about as easy a
7 diagnosis as you can get. But, for some of these
8 other disorders, it can be fairly difficult to
9 really establish that the pain is primarily or
10 purely neuropathic and not due, in large part, to
11 other problems.
12 Then, turning to what Bob was saying about
13 the neuropathic-pain scale, the neuropathy pain
14 scale, with the different mechanisms, and echoing a
15 point that Clifford Woolf brought up earlier today
16 that, because of the importance of CNS mechanisms,
17 there is very broad overlap among all the different
18 neuropathic-pain disorders so that there may be
19 distinct patterns between the different
20 neuropathic-pain disorders. This has been the
21 focus of a lot of the work in our laboratory in San
22 Francisco the last five or six years.
23 There still is quite a bit of overlap
24 between the different disorders and there may be
25 enough overlap that it would take very large
1 studies to try and really look at subtypes within
2 the disorder or distinct pathophysiologic
3 mechanisms to show that that had a very strong
4 impact on study outcome.
6 Turning to some of the more practical
7 clinical-trial issues, I think, from the notes that
8 I was sent by Dr. McCormick before the meeting,
9 that we certainly want to encourage--there seems to
10 be universal agreement on this point if nothing
11 else, that there should be encouragement of
12 studying a broad range of neuropathic-pain
13 disorders, that many of the disorders that I listed
14 in my earlier slide are disorders that are really
15 quite rarely studied.
16 It has been difficult to convince people
17 to study central pain prospectively. Some
18 disorders are really quite uncommon such as
19 adhesive arachnoiditis, a terrible pain problem
20 when it occurs but it would be particularly
21 difficult to do a clinical trial in a disorder that
22 is that uncommon.
23 Also, some of the disorders are difficult
24 to study from a clinical-trials perspective because
25 the population that is afflicted has many other
1 concomitant medical problems. The average age of
2 patients in postherpetic-neuralgia trials is 74.
3 When we enter patients with that disorder
4 into some of our trials, it may take two pages to
5 list all the concomitant medications they are on
6 for all their other problems. I have always
7 advocated for trying to have relatively broad and
8 straightforward inclusion criteria to try and allow
9 as many good research candidates into trials as
11 In diabetic neuropathy, we have more or
12 less given up on doing diabetic-neuropathy trials
13 because anyone who comes to a pain-research center
14 with diabetic neuropathy usually has pretty bad
15 diabetic neuropathy and enough other diabetic
16 complications or enough other sources of nerve
17 injury that they often just can't meet entry
18 criteria for a more typical industry-sponsored
20 Then, as I mentioned, particularly with
21 back pain with radiculopathy, multiple sclerosis
22 and, to some extent, postherpetic neuralgia, there
23 may be a fair amount of inhomogeneity within the
24 disorder, within the population that has the
1 Turning next to responsivity, there may be
2 differences by disorder in the overall response
3 rate. I would say that the one that is most likely
4 to have a relatively low response rate would be the
5 different kinds of central pain. Patients with
6 spinal-cord injury and post-stroke pain are
7 particularly difficult to treat.
8 The medical literature is littered with
9 the debris of failed trials where they couldn't
10 really show any change at all in pain. Diabetic
11 neuropathies had, if anything, the opposite problem
12 where the placebo response rate in some studies has
13 been so high as to make it nearly impossible to
14 show a differential effect with the active
16 In postherpetic neuralgia, perhaps the
17 disorder is just the opposite. The placebo
18 response rates in that disorder are quite low and
19 my own personal opinion on that has been that,
20 because there is such a high prevalence of
21 allodynia that a patient, even if they thought they
22 were doing better, all they have to do is touch
23 their painful area and they very quickly get an
24 index as to whether or not they are still in pain
25 or not.
1 In fact, in some of our studies,
2 allodynia, alone, is really quite a robust outcome
5 Let me just leave you with this for Bob's
6 rebuttal is that perhaps there are a few special
7 cases, postherpetic neuralgia because of the
8 presence of allodynia, and the fact that it is a
9 neuropathic disorder, it is a disease of the
10 nerves, but there is such prominent involvement in
11 the skin.
12 In trigeminal neuralgia, perhaps, because
13 it is one of the only neuropathic pain disorders
14 that is extremely responsive to surgery, to the
15 point that surgery is essentially the first-line
16 treatment once simple medication interventions have
17 failed and where there does seem to be quite a
18 fairly strong shift towards response to sodium
19 channel-blocking agents.
20 So I will leave it to you, Bob, for your
22 DR. DWORKIN: I really don't have much of
23 a rebuttal. I think we want to get this open for
24 discussion as quickly as possible. I would just
25 agree with Mike that I think, with respect to
1 patterns of symptoms and signs and of combinations
2 of underlying mechanisms, there is, obviously, a
3 kind of glass half-full, glass half-empty, issue.
4 But I think, in working on this
5 presentation, the data that I have found most
6 compelling with respect to splitting is these two
7 negative trials of amitriptyline in HIV sensory
8 neuropathy and the two negative trials of
9 dextromethorphan in PHN in the context of two
10 positive trials of dextromethorphan in diabetic
12 Those four trials, I think, are very, very
13 difficult to consider from the perspective of the
14 validity of lumping and having a broad
15 neuropathic-pain indication.
16 But that is all I have to say and I think
17 we should just throw it open for questions and then
18 general debate. Mitchell, an author of all four of
19 the trials, I am considering important in this
21 DR. MAX: In the face of our four papers
22 that you argued should be in favor of splitting, I
23 am going to argue that the FDA and the panel should
24 consider lumping in terms of a general
25 neuropathic-pain claim.
1 As I said, I think the most important--the
2 exception I will make to that is that we need to do
3 something about nerve-root pain because you can't
4 talk about neuropathic pain if you don't include
5 that. But I think we have had a conversation in
6 the past with Cynthia McCormick and Bob Rappaport
7 and I think you have said we don't know enough now
8 to know how to generalize.
9 I have got to say you are right. We
10 don't. It is clear we don't know enough. My only
11 argument for consideration of some kind of general
12 neuropathic-pain claim is that, if it is true and
13 this needs to be proven by some marketing data, if
14 the goal of a general neuropathic claim would
15 encourage industry to do more trials, that is the
16 best way to answer your questions, to learn about
18 For example, the pregabalin program of
19 studies I think has told us more about patterns of
20 pain mechanisms than all the psychophysical studies
21 we have ever done in that there are many trials
22 that show gabapentin and pregabalin relieve
23 diabetic neuropathy in postherpetic neuralgia and,
24 in one trial, doesn't relieve osteoarthritis and it
25 does not relieve the subset of people with
1 radicular pain.
2 So if a general claim would encourage more
3 trials, we would learn from them. I suspect there
4 may be a way you can do it fairly. So I would just
5 like to put that possibility on the table.
6 DR. KATZ: Just a few points of order.
7 First of all, thanks very much for doing a great
8 job. I think you both can sit down because this
9 conversation is going to go on for quite a while.
10 So thanks for pitching in.
11 If people around the table could ask me
12 before they call people up from the audience, that
13 would be very helpful since there are some rules
14 that we have to follow that I am hearing a lot
15 about. So if you could just run that through me,
16 please, although we certainly appreciate your
17 comments and expertise, Mitchell.
18 Let's go ahead then and open up the
19 conversation. The question at hand is does the
20 evidence that we have available to us today support
21 the extrapolation from success in one type of trial
22 to success in another. It may be worth a word or
23 two from the FDA folks to maybe remind the group on
24 what the requirements are for the FDA to consider
25 extrapolation in that context or to consider broad
1 labeling. Would that be all right?
2 DR. McCORMICK: I am not really sure there
3 is any policy on this or basis or requirement, that
4 is. I guess what we really need is--in order to
5 make a general claim for neuropathic pain, that
6 implies that we know that all of the various
7 components are similar or respond similarly to a
8 given drug.
9 So, while, in response to Mitchell's
10 comment which is well taken, understanding that
11 there is a need to stimulate research, I think that
12 we are looking for a stronger scientific
13 justification for making that cut, that there
14 really is a basis for being able to link all of
15 these together rather than generalizing to
16 conditions that we really aren't sure are
17 responsive to a given drug.
18 So while there is no policy, I think we
19 need to have a good scientific basis for making our
20 decisions and that is really what we are bringing
21 to the table today.
22 DR. KATZ: Let me just take a moment and
23 summarize what I heard to be your arguments and
24 then we can bring it to the floor. I think it will
25 help focus the discussion. So, Dr. Dworkin, I
1 think your points were that you feel that
2 extrapolation, or you argue that extrapolation, is
3 not appropriate because the symptoms can be quite
4 different from one neuropathic pain state to
5 another implying that the mechanisms must be
6 different and, therefore, treatment responses must
7 be different.
8 Dr. Rowbothom, I think I heard you say
9 that, well, all is speculative, you really can't
10 get a handle in most individual cases exactly what
11 the mechanisms are. The fact is that,
12 inhomogeneity, as you used the word, may be at
13 least as great within neuropathic-pain syndrome as
14 across neuropathic-pain syndrome. So that argument
15 about mechanisms being different, preventing
16 extrapolation may not hold any water.
17 I also, Dr. Dworkin, heard you summarize
18 the literature that there is, in fact, a strong
19 current within the literature supporting
20 differences in treatment responses across different
21 neuropathic pain states and the two examples you
22 gave were amitriptyline for HIV neuropathy and
23 dextromethorphan showing efficacy in diabetic
24 neuropathy and not in PHN.
25 I think the question that you raised,
1 Mike, is that, are those the exceptions or are
2 those the rules because we have these other cases
3 which are, if anything, much more well-studied
4 where we see gabapentin seems to have a relatively
5 broad spectrum of activity. Pregabalin similar for
6 neuropathic pain conditions in which it has been
7 studied. Amitriptyline, yes; it doesn't work for
8 HIV sensory neuropathy but seems to work for
9 everything else that has been looked at
11 I.V. lidocaine. You mentioned opioids and
12 there are other examples of where broad spectrum
13 seems to be the characteristic of the different
14 agents. So, are the points that you made the
15 exception or are they the rule?
16 So those seem to be the arguments as I
17 heard them. Maybe we could now open it up to see
18 what people think about those arguments for or
19 against extrapolation.
20 Mike, please?
21 DR. ASHBURN: I had a couple of remarks.
22 I will have my back to Dr. Max so then I can take
23 shots at him. Dr. Max used gabapentin as an
24 example and actually pointed out a couple of the
25 trials in general. I hope I don't misname them or
1 misdescribe some of the results of them but I
2 actually want to use them as an example for not
3 lumping. Gabapentin may be effective in
4 postherpetic neuralgia. There is some data to show
5 it might be effective in diabetic sensory
7 There is some debate with regard to its
8 effectiveness for the treatment of the radicular
9 component of low back pain. So what we think, what
10 I think, you will see when individual agents are
11 studied in different patient populations will most
12 likely be a variable response to different
13 indications since I think it is fairly clear that
14 there are lots of different mechanisms and one
15 medication is unlikely to be effective in a broad
16 spectrum of different indications.
17 In addition, gabapentin has been
18 implicated, if you will, in having other beneficial
19 effects that are taken advantage by clinicians that
20 may not necessarily lead to an indication of its
21 effectiveness in neuropathic pain such as many
22 physicians believe that it enhances the analgesic
23 effects of potent opioids when used in combination
24 with potent opioids.
25 It is also thought by the psychiatry world
1 to have anxiolytic effects which translate to
2 analgesic effects in many patients who have mixed
3 pain conditions. With that mixture, it is hard for
4 me to conceptualize how one could combine all that
5 and then get a general indication for neuropathic
6 pain particularly since the end result is trying to
7 prepare a package insert that guides a physician
8 like me who does clinical practice on how to use
9 that agent in these variable different populations.
10 How would that be written? How would I
11 look at outcomes? How would I, on an individual
12 patient faced with radicular low back pain make a
13 clinical decision with regard to the risk versus
14 the potential benefits? Gabapentin does have
15 inherent risk. Although it is a fairly safe drug,
16 there are significant dose-related side effects
17 that one has to struggle with.
18 Many of the other agents that are used for
19 neuropathic pain also share those. So I guess that
20 it the point I am making is that, at this stage, a
21 broad indication for neuropathic pain, while being
22 a long-term objective might be very difficult to
23 try to sort out because the different populations
24 are so different.
25 Ultimately, it will be difficult, I think,
1 to be able to write an insert, or write indications
2 to physicians, on how to use those medications in
3 an effective way across broad different patient
5 DR. KATZ: Other perspectives on this
6 issue? Dr. Shafer?
7 DR. SHAFER: Earlier, we talked about
8 different taxonomies and having a taxonomy
9 organized by disease or a taxonomy organized by
10 mechanism. Once again, we sort of endorsed the
11 idea of taxonomy organized by mechanism but once
12 again we are back to splitting based upon a disease
14 Since we keep coming back to this disease
15 rather than the mechanism, my real question is are
16 we just simply so ignorant about mechanisms and the
17 fact that patients will come and they will say, "I
18 have diabetes," but they won't say, "I have
19 small-fiber disease," that we should abandon
20 attempts to organize this discussion along
21 mechanisms of neuropathy and just stay with the
22 disease orientation or should we approach this, as
23 we did with the other discussion, along the lines
24 of mechanisms.
25 DR. KATZ: Dr. Woolf? Do you have
1 anything to say about mechanism-based approaches?
2 DR. WOOLF: I certainly heartily endorse
3 that. I think the problem is that we don't have a
4 full enough understanding of the mechanisms but,
5 even more than that, we don't have the tools yet to
6 identify in patients what those mechanisms are.
7 Until we do, I think we are going to have to, in
8 the real world, deal with package inserts to give
9 instructions to clinicians.
10 But I think what it does raise is the
11 issue that syndromes, and we discussed this morning
12 for diabetic neuropathy, are not homogeneous so
13 that Mike Rowbothom has shown very clearly, and I
14 am surprised he didn't actually mention this in his
15 talk, that postherpetic neuralgia is not a
16 homogenous syndrome, that not every patient has
17 tactile allodynia and he believes that you can
18 identify different subgroups with different
19 mechanisms which reasonably may respond
20 differentially to different forms of therapy.
21 So I really think, and what I find really
22 intriguing, is how much of the discussion this
23 morning comes back--to talk about generalizability
24 can only depend on the pharmacological activity of
25 the particular drug. If it is going to potentially
1 act across many mechanisms, then that may be
2 applicable. But, as we dissect out the mechanisms
3 and the molecular elements, there are certainly
4 going to be some drugs that are going to be very
5 specific in their action and that are almost
6 certainly not going to be generalizable.
7 So I think it is going to have to be done
8 on a case-by-case basis. There are some drugs,
9 like the opioids, which are not mechanism-specific.
10 They act to operate on multiple sites in the
11 neuraxis to modify sensory processing. They are
12 not affecting the mechanism of the pain. They are
13 producing an analgesia.
14 There are others such as sodium channel
15 blockers that will only work on those situations
16 where there is abnormal sodium channel expression
17 or number. So I think it is inappropriate to say
18 that, for all drugs, there may be an issue of
19 generalizability, that they need to be based on
20 what mechanisms are present, both in terms of the
21 disease state and of the drug mechanism.
22 DR. KATZ: It sounds like what you are
23 saying is that generalized activity needs to be
24 proven for each medication.
25 DR. WOOLF: Right. I think there will be
1 drugs that are generalizable and I think there will
2 be others that aren't.
3 DR. KATZ: If I could just push you a
4 little bit further on that. Do you think that, at
5 least in concept, there is a threshold that can be
6 crossed by whatever package of trials necessary to
7 get to the point where you can say, yes, this is
8 effective for neuropathic pain in general?
9 DR. WOOLF: Yes. I am not going to define
10 what that threshold is here. Maybe collectively we
11 could, but I think it needs to be science driven in
12 the same way that we now appreciate that Cox 2
13 inhibitors act by inhibiting Cox 2 and, if Cox 2 is
14 not induced, they are not going to have any action.
15 I think we now are beginning to appreciate
16 that the data is not really that strong, that the
17 alpha 2 delta subunit of the calcium channel may be
18 the target for the gabapentinoids and this is a
19 subunit that is upregulated after nerve injury. If
20 that is true across all forms of nerve injury, then
21 one can make a scientific case why gabapentin and
22 pregabalin may act in the broad spectrum.
23 But, as I said, there are other cases
24 where it is quite reasonable to suppose that an
25 alteration in vanaroid receptors may occur very
1 specifically in a subgroup of patients in which
2 case the L1 antagonist will have a much more
3 defined and smaller indication.
4 DR. KATZ: It sounds like what you are
5 saying is that, in concept, one could conceive of a
6 broad neuropathic pain-acting drug which
7 ultimately, when we get to that point, could be
8 determined scientifically through mechanism-based
9 approaches but, in the meantime, since we don't
10 have good mechanism-based approaches for people, it
11 is possible that we could achieve that goal through
12 other means which you have chosen not to define for
13 the moment.
14 DR. WOOLF: No; I am not as defeatist as
15 that. I think we are at a position now where, as
16 we design our trials, we can attempt to define
17 mechanisms as well. I think we need to use the
18 conventional methodology with all its limitations
19 but, in parallel with that, to try and get measures
20 that at least reflect the mechanisms.
21 Bob showed his spaghetti junction of
22 mechanisms. We need to try and see which of those
23 are fantasy and which are reality. Certainly, we
24 don't have all the measures to elaborate all of
1 Global pain scores, as we all use, as
2 simple, whether as a categorical scale or Brad
3 Galer's--those are so crude. We all accept that
4 they are missing those elements of the pain that
5 may be responsive to different forms of therapy.
6 So we are lumping them together and maybe losing a
7 lot of sensitivity.
8 So what I am arguing is that we need to
9 collect as much data as possible, see how the
10 different elements of the patient symptoms and
11 signs respond the different treatments and try and
12 identify that in the context of the different
13 mechanisms that may be operating.
14 DR. KATZ: Dr. Bril?
15 DR. BRIL: I was disappointed in the
16 results of tricyclics in the HIV population that we
17 found because, if there was a lumping function, I
18 could have seen it more with diffuse
19 polyneuropathies that are painful, that are similar
20 clinically, such as toxic or diabetic, because I
21 think the pain mechanisms are not necessarily
22 specific to the disease and I could have seen that.
23 But I have a little bit of difficulty with
24 just a stamp saying neuropathic pain regardless of
25 the etiology because what if you have a
1 carpal-tunnel patient. Shouldn't you be talking
2 about splints and decompression rather than trying
3 a medication right up for carpal tunnel? Maybe
4 they need surgery.
5 So, if you were going to give a
6 neuropathic pain indication, and this was in the
7 inset, people may well misuse the medications for
8 the indications you need. Something like
9 postherpetic neuralgia or trigeminal neuralgia,
10 which is treated basically with medications and,
11 perhaps, surgery with trigeminal neuralgia really
12 late, I can understand, again, lumping.
13 But just neuropathic pain of all kinds
14 doesn't make much sense to me even with what we
15 know now, and particularly the radicular question.
16 I have a real problem trying to lump
17 radiculopathies because there are so many other
18 modes of therapy for radiculopathies. So that
19 seems more problematic from a more basic point of
20 view even than the molecular level at all.
21 DR. KATZ: Dr. Aronson?
22 DR. ARONSON: I was just going to
23 reiterate so many of the comments that were made.
24 I guess, as I hear this discussion, it is almost an
25 artificial separation between lumping and
1 separating. I think there may, indeed, be, as we
2 appreciate better mechanisms of action,
3 commonalities across disease states and, if you
4 will, differential within a disease state that we
5 just simply don't appreciate.
6 The fear I have is that we will find a
7 drug works for whatever that means and however we
8 define it, but it is the right answer for all the
9 wrong reasons because we just simply don't know
10 what the reason is. So I think mechanism is so key
11 to drive this discussion rather than creating these
12 artificial silos of disease or mechanism or
14 I think we really must drive this by
15 hypothesis in the beginning, what do we expect this
16 drug to do and why do we expect to do it and test
17 it in that sort of context.
18 DR. KATZ: Dr. Dworkin?
19 DR. DWORKIN: I want to argue a moment for
20 lumping since everyone is arguing, it seems, for
21 splitting. When I think about this issue, I can
22 imagine an indication, being completely naive to
23 the way the FDA thinks, that would be something
24 like pain in peripheral-nerve injury and that would
25 be supported, for example, by replicate trials in
1 diabetic peripheral neuropathy, a single positive
2 trial in PHN, a single positive trial in HIV
3 neuropathy and a single trial in paxil neuropathy
4 that is positive, so a package of five trials
5 across four conditions, all of which were positive.
6 If a company had that kind of package, I,
7 personally, can't think of a strong argument why I
8 wouldn't be comfortable with a lumped indication of
9 pain from peripheral-nerve injury.
10 Clearly, we could come up with other
11 conditions that are not on this list that I just
12 came up with where the drug might not be
13 efficacious but this seems to me like a large
14 enough sample that I, personally, would be
15 comfortable with lumping in that delimited way,
16 pain from peripheral-nerve injury.
17 DR. KATZ: So you are saying that there
18 might be the possibility to split neuropathic pain
19 into largish subdivisions where lumping might be
20 appropriate, peripheral being the example. One
21 could also imagine central where you talk about
22 stroke or what have you.
23 DR. DWORKIN: It is easier, in fact, for
24 central because there are fewer syndromes so you
25 can kind of capture, sample the universe of central
1 neuropathic-pain syndromes easier, I think.
2 DR. KATZ: So it sounds like, while it is
3 not clear exactly what one would need to do the
4 cross that threshold of being broadly efficacious
5 for peripheral neuropathic pain, that there, at
6 least conceptually, could be such a threshold.
7 Since you are a lumper now, let me push
8 you a little bit further. Could you conceive of a
9 threshold that could be crossed with central pain,
10 the pain of spinal-cord injury, postherpetic
11 neuralgia which is probably mixed central and
12 peripheral, peripheral types of pain where one
13 could actually become a real lumper and say
14 neuropathic pain broadly.
15 DR. DWORKIN: I think, to follow the logic
16 of what I just did, if you had replicate trials in
17 one peripheral neuropathic-pain syndrome and
18 replicate positive trials in central post-stroke
19 pain, and the other peripheral syndromes I
20 mentioned and a single positive trial in
21 spinal-cord injury pain, a single positive trial in
22 MS pain, how could that not be justification for a
23 broad indication of neuropathic pain unmodified by
24 either central or peripheral.
25 If the company had really sampled the
1 domain and, within the filing, had a couple of
2 replicate trials, one peripheral, one central, I
3 can't imagine an argument why that wouldn't be a
4 broad indication.
5 DR. KATZ: Even though there might be some
6 syndromes in which that very medication might not
7 be efficacious and many patients within syndromes
8 in whom that medication might not be efficacious
9 which, as we know, is the rule.
10 DR. DWORKIN: There are always way-out
11 exceptions. I think if you have sampled the
12 universe adequately, you have to just tolerate that
13 there might be an exception that shows up five
14 years down the road in a negative trial.
15 DR. KATZ: Just to push you even a little
16 bit further on that, would you then call
17 amitriptyline a drug that is efficacious broadly
18 for neuropathic pain given that the one exception,
19 as far as we know, is HIV neuropathy?
20 DR. DWORKIN: I am troubled by the
21 replicate negative trials in HIV neuropathy.
22 DR. KATZ: But given that that is the one
23 syndrome that it seems not to be efficacious in,
24 that would seem to fit with your scheme.
25 DR. DWORKIN: In fact, when you look at
1 the other literature, what we have with
2 amitriptyline is a lot of positive trials in
3 diabetic neuropathy, a lot of positive trials in
4 PHN. There is a nice review of the literature by
5 Sindrup and Yensen that everyone should have. If
6 you look at that review, other than those trials in
7 PHN and DPN, there is a single positive trial in
8 post-mastectomy-pain syndrome and I think a single
9 positive trial in spinal-cord-injury pain.
10 So, in terms of randomized controlled
11 trials, it is not as good for amitriptyline as we
12 all kind of think every day in the clinic. It is
13 not the case that we have really sampled the
14 spectrum with amitriptyline and found a lot of
15 positive results. Amitriptyline and HIV neuropathy
16 is an exception.
17 So I don't know about amitriptyline. It
18 may not be as broadly an efficacious drug as we
19 think. But I would be surprised that, if one did
20 this program correctly with an opioid--I personally
21 would be surprised if you didn't find efficacy
22 across many of these syndromes that would support a
23 broad indication.
24 DR. KATZ: Dr. Shafer?
25 DR. SHAFER: We seem to have general
1 agreement that if we knew mechanisms, that would be
2 the right way to approach this. But we don't. We
3 are forced into a certain amount of empiricism and
4 there is this cross between virtually all the
5 mechanisms in all of the states.
6 I wonder if the other means, to take a
7 mechanistic approach, would be to actually
8 define--to use response to therapy which is how we
9 often look at mechanisms anyway, in which case a
10 strategy, sort of borrowing from arms control,
11 might be a lump-but-verify strategy where you say
12 this is broadly approved, but we will verify, by
13 therapeutic response in patients and then we will
14 put it on the sponsors to say, "If patients are
15 going to respond, they need to respond in three
16 weeks, in four weeks. If they haven't responded,
17 we are assuming that this drug is not addressing
18 the mechanism appropriately."
19 DR. KATZ: Are you suggesting, then, broad
20 labeling in anticipation of evidence of efficacy?
21 DR. SHAFER: No. What I am saying is that
22 labeling would permit trying the drug out with
23 specific instructions that, were it not to be
24 effective in four weeks, that subsequent use would
25 basically be off-label, that there would be a trial
1 period that would be part of the recommended
2 therapy with the drug and it was to be discontinued
3 if it did not reach--if it proved to be the wrong
5 DR. KATZ: Ms. Delph, you were next.
6 MS. DELPH: A couple of questions and a
7 comment. Are there a minimum number of disease
8 states that you think a drug should be tested in or
9 conditions that a drug should be tested in before
10 you can lump it and give it a broad indication?
11 Secondly, what about the use of animal models? Are
12 there specific animal models that can be used to
13 predict response even though patient populations
14 may not be tested if you can't do all of them?
15 The final comment. To the best of my
16 knowledge, HIV neuropathy, itself, is not a uniform
17 disease. You have neuropathy secondary to HIV
18 disease, itself, neuropathy secondary to toxicity
19 from drugs like didanosine, stavudine and so on.
20 So I have a question about HIV neuropathy as an
21 entity in and of itself, whether that can be just
22 looked at as one entity.
23 DR. KATZ: Dr. Rowbothom?
24 DR. ROWBOTHOM: I am glad you brought this
25 point up because I was going to address some of
1 those anyway. I agree a lot with what Clifford
2 says and that is that there probably are some drugs
3 that really should get a broad neuropathic-pain
4 labeling because they seem to work in so many
5 different syndromes.
6 The issue has gotten a little more
7 complicated in that a disorder that previously
8 seemed to be pretty drug-responsive, painful
9 diabetic neuropathy, was recently found to be
10 unresponsive to topiramate which,
11 pharmacologically, has enough similarities to other
12 drugs that one would have expected that that would
13 succeed in that disorder. So the equation has
14 gotten a little bit more complicated because of
16 Turning to HIV and also the problem of
17 central pain, if you set a criteria for a broad
18 indication that said you have to show that it is
19 effective in some list of four or five disorders,
20 then that would raise the bar, perhaps,
21 unacceptedly high because some pain disorders seem
22 to be particularly difficult to treat.
23 Central pain, spinal-cord injury and
24 post-stroke pain, very difficult pain syndromes to
25 treat, and HIV neuropathy also appears to be pretty
1 stubborn, a pretty difficult disorder to treat. So
2 we don't want to, or at least I wouldn't want to,
3 suggest to the FDA--I wouldn't want to encourage
4 the FDA to set their rules in such a way that there
5 would be a strong incentive for the pharmaceutical
6 industry to not study disorders like central pain
7 and HIV neuropathy pain which are terrible, severe
8 problems that really need more study because they
9 are unlikely to respond.
10 Certainly, no one is going to want to
11 study a disorder that never seems to get better
12 with medications; right? That is pretty obvious as
13 a bad idea economically. So what, perhaps, might
14 be a way to go would be to try and encourage good
15 studies in as many disorders as possible and, for
16 the ones that are difficult to manage in the sense
17 of being relatively unresponsive, to not
18 necessarily require those be included for pivotal
19 trials but that they be included as part of your
20 safety data so that we do collect a large database
21 on these less-well-studied disorders like
22 HIV-neuropathy pain and central-pain disorders.
23 Hopefully, with uniform enough guidelines
24 so that studies can be compared with each other so
25 that gradually a large database can be accumulated
1 and we can start to answer the question of are some
2 types of peripheral-nerve-injury pain particularly
3 hard to treat. Are some areas of injury to the
4 central nervous system--does an injury there confer
5 a particularly bad prognosis as far as responsivity
6 to treatment goes?
7 DR. KATZ: I want to make sure I
8 understand what you are saying. It sounds like you
9 are saying that you wouldn't want the FDA to
10 require efficacy in all manner of diverse syndromes
11 in order to get any indication at all but that
12 studying them in syndromes with a track record
13 should lead to those specific indications while, at
14 the same time, you would like to see encouragement
15 to study broad ranges of heterogeneous groups of
16 neuropathic-pain patients not necessarily with the
17 requirement of showing efficacy but more just to
18 see if there is a signal there efficacywise and,
19 also, to get safety data in these populations in
20 whom the drug is likely to be used maybe off-label
21 anyway, if I understood you correctly.
22 But I am still not sure what your
23 perspective is on whether it is conceivable that
24 drugs could have broad ranging efficacy and be
25 labeled as such.
1 DR. ROWBOTHOM: I think that there are
2 some drugs that should or can acquire a broad
3 neuropathic-pain label. What that would mean is
4 that their mechanism of action, where they work in
5 the nervous system, is at a critically important
6 place. Clifford mentioned the opioids, opioid
7 receptors in so many different locations in the
8 central nervous system, that really all the points
9 involved in the pain transmission and modulation,
10 there is some ability for opioids to influence the
11 signalling there.
12 So that would certainly be a potential
13 category for a broad indication. To the extent
14 that we know it, drugs like gabapentin and possibly
15 pregabalin seem to be moving in that direction.
16 The tricyclic antidepressants, partly because they
17 are such dirty drugs, they work on so many
18 different transmitter systems that they also seem
19 to be fairly broad-spectrum drugs for neuropathic
21 So there certainly should be candidate
22 compounds out there that could acquire this kind of
23 labeling. My point was to say yes to that question
24 but to also try and make sure that the criteria are
25 set up so that we continue to acquire important
1 information, especially about safety and
2 tolerability, in disorders that are less
3 well-studied rather than continuing to study
4 diabetic neuropathy and, to a much lesser extent,
5 postherpetic neuralgia over and over again.
6 DR. KATZ: It sounds like you Bob have
7 actually converged in your perspectives now. You
8 have both become partial lumpers in the process of
9 this discussion. But, if I could just push you a
10 little bit further on some of these issues which is
11 what I like to do, as you know.
12 You spoke about studies in heterogenous
13 groups of neuropathic-pain patients as being useful
14 because maybe they would identify efficacy signals
15 that otherwise we would miss because nobody is
16 going to do a trial on just patients with central
17 dyskinesthesia syndrome from spinal-cord injury or
18 whatever it is.
19 Would you see there being any role of
20 Phase III clinical trials in patients with
21 heterogeneous neuropathic-pain disorders which,
22 although those trials may be very challenging to
23 see any outcomes because of the heterogeneity of
24 the patients, but if efficacy was shown, that that
25 could be a more direct path to a broad
1 neuropathic-pain indication?
2 DR. ROWBOTHOM: If I understand you
3 correctly, what you are talking about is a large
4 trial where there were a spectrum of definable
5 neuropathic-pain disorders that would qualify a
6 potential subject for participation. So, for
7 example, they could have multiple sclerosis and
8 chronic pain related to that, focal
9 peripheral-nerve injury, all these different
10 disorders and then you would have a large study
11 looking at the overall broad spectrum of
12 neuropathic pain, or what we lump together as
13 neuropathic pain, and then, within that, substudies
14 that could, potentially, establish efficacy within
15 the component disorders.
16 I think that makes sense. That is
17 certainly a possible strategy.
18 DR. KATZ: Any other comments on that, on
19 the scientific justification for that approach?
21 DR. FARRAR: Just a quick point. If you
22 put together a group of neuropathic-pain patients
23 and did a study, at the end of the day, I think
24 what I heard Mike say is you would then have to
25 look at the subgroups individually and show that,
1 in each subgroup, you had an effect as well.
2 DR. KATZ: I didn't hear that. Did you
3 say that, Mike?
4 DR. ROWBOTHOM: I'll let him finish.
5 DR. FARRAR: Let me be specific, I guess.
6 The point is that if you put together a
7 heterogenous group and it was 50 percent diabetic
8 neuropathy, 20 percent postherpetic neuralgia, a
9 smattering of this and a smattering of that, and
10 you showed that, on average, that group got better,
11 I don't think that is evidence that it works in the
12 other groups.
13 I think, ultimately, I end up being in the
14 same camp as Clifford, and others here, in the
15 sense that, without understanding the mechanism, I
16 think it is impossible to be able to say that a
17 drug works in everything.
18 I would like to make one other comment
19 which is that Clifford also said, I think, that
20 even if we know the mechanism, predicting that
21 within an individual patient is going to be
22 somewhat difficult and I think Dr. Shafer suggested
23 a solution which is that we need to focus on not
24 only whether it works in that group but then a
25 study of why it works in particular subgroups
1 because, as we know, it doesn't work in everybody,
2 like you said.
3 DR. KATZ: Just if I could understand you
4 a little bit better. It sounds like what you are
5 saying is that, if you had a trial of heterogeneous
6 patients with neuropathic-pain, even though the
7 mean response or number of responders, whichever
8 outcome measure you like, was better in your
9 treatment group than your placebo group, you
10 wouldn't accept that as being broadly efficacious
11 for neuropathic pain because it was driven by a
12 subgroup of responders.
13 But if you had a trial of, say, something
14 like gabapentin in painful diabetic neuropathy in
15 which only 30 or 40 percent of the patients
16 responded, driving the statistically significant
17 response in your treatment group compared to your
18 placebo, would you accept that as being indicative
19 of efficacy in painful diabetic neuropathy?
20 DR. FARRAR: Absolutely. In fact, in the
21 postherpetic neuralgia study, I would argue that it
22 is not the mean value of the pain that was
23 important. If you look at the paper that was
24 published, 37 percent, approximately, depending on
25 how you define it, of the patients who got
1 gabapentin got really better, meaning moderate or
2 better, relief and only 15 in the placebo group.
3 Only a third of the patients got dramatic
4 improvement. But you give it one patient who
5 hasn't had improvement for ten years and they are
6 suddenly better, it is good evidence.
7 The second issue there, though, I think
8 somebody else had mentioned, is the issue of
9 safety. If the drug has very few side effects, or
10 serious side effects, anyway, you are much more
11 inclined to be willing to try it in a person where
12 there is a one-in-three chance of it working.
13 If the drug is like amitriptyline, I am
14 going to be much less inclined to use it. I think
15 that there are very significant side effects,
16 especially in older populations, that worry me a
17 great deal. So I think you have to make that
19 DR. KATZ: Again, just so I can fully
20 understand, what is your scientific rationale for
21 accepting success in a trial like the gabapentin
22 trial when, in fact, the success is only driven by
23 a subgroup of responders when you are not willing
24 to accept success for heterogeneous, for broad
25 neuropathic pain, when that success is also again
1 driven by a subgroup of responders.
2 DR. FARRAR: Because you can't predict
3 looking at person with postherpetic neuralgia, at
4 least not yet until Mike finishes his studies--we
5 can't predict who is going to be responding. We
6 can't divide the postherpetic-neuralgia group into
7 groups where some of them respond, where there is a
8 subsection of them that responded.
9 If we could do that, I would argue for
10 trying it only in that subgroup. But, until we can
11 do that, I think it is reasonable to try it in all
12 again because it is safe. What you are suggesting
13 is taking people that we actually think are
14 somewhat different or have some differences that we
15 can define, mixing them and then saying, because
16 30 percent of them respond that, somehow, everybody
17 in that group is the same.
18 I think there is a very distinct
19 difference. In the postherpetic-neuralgic group,
20 we cannot identify, a priori, the differences.
21 Now, Mike has started doing some research that
22 hopefully will move us toward being able to do
23 that. But, until that happens, I don't think we
24 can do it.
25 DR. KATZ: Any regulatory perspectives on
1 this issue of studying heterogeneous groups of
2 neuropathic-pain patients?
3 DR. McCORMICK: I think that we are
4 answering the question that we are all struggling
5 with with an example. I think that to embark in a
6 study that has a heterogeneous group is making the
7 assumption that we already know that lumping makes
8 sense. So I think that first we need to answer the
9 question does lumping make sense before we
10 encourage trials in heterogeneous groups.
11 DR. DWORKIN: Stating the obvious, I think
12 we have all seen data with heterogeneous groups of
13 patients where the significant efficacy is based on
14 a subgroup of not responders but a subgroup based
15 on diagnosis which suggests you shouldn't be
16 lumping, if the overall significant difference
17 comes from a subgroup of one diagnosis.
18 DR. KATZ: Does anyone else have any final
19 comments on the lumping versus splitting issue?
20 Dr. Woolf?
21 DR. WOOLF: Just to address the second
22 part of Ms. Delph's question about animal models
23 which got lost somewhere along the line. I think
24 that is a very important issue and I think animal
25 models need to be looked at as critically as we are
1 looking at the clinical development of programs.
2 There are many problems there. The
3 problems are that the animal models have been
4 designed to reliably produce symptoms that are
5 "pain related." Some of them are designed to be
6 models of disease. Very few are designed to be
7 models of mechanisms which is what we are aspiring
8 to. Again, we don't often know what mechanisms
9 operate in those models.
10 More significant is the problem that all
11 we can measure in animal models are responses to
12 stimuli. We cannot measure spontaneous pain which,
13 in diabetic neuropathy, is the biggest problem. So
14 we use outcome measures which are convenient but
15 may often be irrelevant such as heat. Hyperalgesia
16 is the commonest outcome measure in animal models
17 but is not a problem that any patient ever
18 complains of.
19 I think one of the most significant issues
20 about predictors is that we can use doses in
21 animals where humans wouldn't tolerate. It is very
22 difficult to measure side effects such as dizziness
23 or sedation. So we can get effects in animal
24 models that we would never be able to escalate a
25 dose in a patient to get the equivalent effect.
1 So I think animal models are essential.
2 They are going to, obviously, always drive the
3 drug-development program but they are never going
4 to be a surrogate for human trials, in my opinion.
5 DR. KATZ: Dr. Rowbothom?
6 DR. ROWBOTHOM: One thing that I think is,
7 perhaps, a little separate from the regulatory
8 issues and that is that, as these trials are done,
9 we want to look at the group that is responding and
10 the group that is not responding to see what we can
11 learn from a clinical-mechanisms perspective as to
12 why those patients diverged into responders versus
14 The other aspect, the regulatory aspect,
15 is that there is a model for what we are talking
16 about, large studies of mixed neuropathic pain in
17 the form of the pregabalin studies that have been
18 done where the study designs are relatively
19 similar. It is a series of studies in different
20 diagnostic groups.
21 Their approach was to study some
22 neuropathic pain and then include some disorders
23 that are thought to be nonneuropathic like
24 fibromyalgia and osteoarthritis. The value in that
25 data, of course, as John Farrar knows really well,
1 is that it is a gold mine of information about how
2 patients respond in general and pain
3 characteristics and all these other things.
4 But that is an approach that has already
5 been taken that is similar to what we were talking
6 about before where if you look at a variety of
7 chronic pain disorders including ones that are
8 known to be pretty treatable as well as ones that
9 are believed to be relatively refractory to
10 treatment, that is a valid approach as long as you
11 make sure that the number of subjects studied with
12 each diagnosis is enough that you have an
13 adequately powered look at that particular
15 DR. KATZ: Any final comments about the
16 lumping versus splitting issue before we move on?
17 Dr. McLesky?
18 DR. McLESKY: Dr. Hertz actually raised
19 this issue I thought in response to a comment real
20 early today when she was asked the question would
21 it make sense to do two pivotal trials that are
22 very similar or would it make sense to do
23 potentially two separate trials that might
24 corroborate one another in potentially somewhat
25 different populations?
1 Speaking for industry, I think we would
2 like to have feedback. Maybe you could elaborate
3 on that just a little bit more, if you would.
4 Also, I would like feedback from the panel. If we
5 were going to lump, if it did make sense to lump
6 for a particular drug, how many different kinds of
7 patient populations would it make sense to test in
8 order to be able to logically lump?
9 DR. KATZ: Boy, I don't know that we are
10 going to get that today. Does anyone want to throw
11 any proposals or comments? Sharon?
12 DR. HERTZ: The comment I made about
13 replicating studies was not replicating different
14 diagnostic populations but, within diabetic
15 neuropathies, not to mimic the exact study design
16 at the same center or group of centers but maybe to
17 take two meaningful study designs in the diabetic
18 population across centers and then have a slightly
19 different approach just to show that this wasn't
20 just one very, very large study which has a whole
21 separate discussion.
22 So it really wasn't referable to different
24 DR. KATZ: So it sounds like what you were
25 saying, Dr. Rowbothom, was that repeating very
1 similar studies in different diagnostic populations
2 is very useful for characterizing the spectrum of
3 the drug where as Dr. Hertz is saying that, to
4 really prove the point about any individual
5 indication, two trials that support each other but
6 may not be completely identical could be an optimal
9 DR. DWORKIN: What I had said earlier, and
10 I guess I would stand by it, for
11 peripheral-nerve-injury pain, I think four or five
12 different conditions for me kind of is enough of a
13 sample of the universe and, for central pain, three
14 because I just can't--beyond a certain point of
15 those numbers, there are not that many syndromes
17 DR. KATZ: Dr. Woolf?
18 DR. WOOLF: To maybe expand the lumping
19 and splitting debate, as you are about to close it,
20 we haven't split in terms of different elements of
21 the pain so that if, for example, a drug could be
22 shown to act only on spontaneous pain and leave
23 evoked pain unaffected, or the opposite, tactile
24 allodynia was sensitive, and tactile allodynia was
25 expressed across a different range of patients both
1 peripheral and central, what are the implications
2 of that in terms of either trial design or
3 potential label?
4 Patient Populations
5 DR. KATZ: I am going to use that actually
6 as a segue to move on to the next topic which is
7 related to your question which is how should we
8 characterize our patients upon entry into a
9 neuropathic-pain trial, what tests should be do,
10 what examination procedures, how can we define what
11 population we are dealing with.
12 It seems like everybody agrees that we
13 should all be working towards trying to understand
14 better what patient characteristics might confer a
15 responder status upon that patient eventually maybe
16 towards a mechanism-based approach to treating
17 these illnesses. So what do people think about how
18 we should be characterizing our patient population
19 upon entry?
20 DR. BRIL: I will start off. In the
21 specific case of diabetic neuropathy, I think we
22 need to establish the severity. I know I have been
23 involved in some trials in which the diagnosis was
24 assumed and very little independent objective
25 measure was done other than symptoms and signs, and
1 those patients--this was a study of just pain.
2 But I think the studies are strengthened
3 by the information on better responsiveness in
4 those who have, say, a sural potential present
5 meaning that staging of severity might improve the
6 trials and improve our understanding and the
8 So I think I would make a recommendation
9 or a suggestion that pain trials in diabetic
10 neuropathy be not considered simply studies of
11 reducing pain but look at the severity and relate
12 it to the outcome.
13 DR. KATZ: How would you do that,
15 DR. BRIL: Specifically, at this point, by
16 doing sural-nerve conductions and splitting into
17 sural-nerve positive or sural-nerve absent. I
18 think the QST, the vibration-perception thresholds
19 are a little less understood at this point. We
20 haven't divided them and looked at outcomes so well
21 so that I would look on severity as stage by
22 sural-nerve responsiveness.
23 DR. KATZ: Just to state the obvious,
24 duration of disease, severity of pain, all those
25 things, neurological exam, all those things. I
1 don't think we will find any disagreement about the
2 need to include those.
3 Dr. Farrar first, then Dr. Shafer.
4 DR. FARRAR: I think that we are limited
5 to a degree by what we know and that was said to a
6 great extent in our earlier discussion. I am very
7 much in favor of taking a group of patients, let's
8 say, who have postherpetic neuralgia and including
9 them all in a study in which we then measure, I
10 guess, the equivalent of a sural-nerve conduction
11 perhaps looking at allodynia and nonallodynia,
12 measuring the number limited by, obviously,
13 patients' tolerance for testing but measuring a
14 number of different features that we think might
15 actually help to differentiate subgroups within
16 that overall disease category and then looking post
17 hoc at that, not looking for the answer but looking
18 for the hypothesis for the next study.
19 By that mechanism, we can both study
20 compounds that may be useful as well as get some
21 sense about the underlying mechanisms.
22 DR. KATZ: So you are advocating
23 characterizing patients upon entry based on their
24 sensory abnormalities, basically?
25 DR. FARRAR: Yes. Certainly, that is one
1 of the components but it may also be that duration
2 of disease is important. It may also be that the
3 location of the process, whether they were treated
4 aggressively early on or not, their age, et cetera.
5 There are obviously many features and you would
6 ultimately design or look at an etiologic model and
7 a predictive model afterwards to try and generate
8 hypotheses for which groups respond and which ones
10 DR. KATZ: Dr. Rowbothom, do you have any
11 comments on the appropriate of trying to
12 characterize patients in PHN trials based on
13 sensory abnormalities or other criteria?
14 DR. ROWBOTHOM: The easiest one is to
15 characterize them in terms of the severity and
16 spacial extent of allodynia. We use a foam paint
17 brush which is inexpensive and it is quite
18 reproducible. It is something that is suitable for
19 multicenter trials because it is quite easy to
20 train somebody how to do that in a reproducible
22 Some of the more specialized techniques
23 that we have used, like capsaicin response and skin
24 biopsy is much more difficult or just much harder
25 on the patients. The capsaicin can be quite
1 painful for them so that is not something I would
2 really advocate applying large-scale across all
3 different kinds of compounds.
4 But, certainly, for that disorder,
5 allodynia should--I would very strongly advocate
6 that be followed.
7 DR. KATZ: Brush allodynia? Anything
9 DR. ROWBOTHOM: I think that you get into
10 a complexity problem when you start trying to go
11 multicenter. We were involved in one small study
12 where there were four centers and we did very, very
13 detailed quantitative sensory testing and sensory
14 mapping. Although we were able to come up with
15 pretty good agreement in the measures, it was an
16 enormous amount of work to do that, and that was
17 four university-based centers that had all
18 previously published in that area beforehand.
19 So, if you start trying to go from there
20 into the more typical multicenter study where are
21 maybe ten or fifteen or twenty centers and you
22 start going more into community-based practices,
23 then I think that level of sophistication starts
24 getting really difficult for a disorder like
25 postherpetic neuralgia.
1 DR. KATZ: Dr. Shafer?
2 DR. SHAFER: I think I am just stating the
3 obvious, but you had asked earlier about different
4 kinds of pain, spontaneous pain, for example,
5 versus evoked pain. I would say that if a company
6 doesn't know if their drug is better for
7 spontaneous pain versus evoked pain, they are not
8 ready to enter a pivotal Phase III trial where they
9 select one or the other.
10 There need to be some Phase II trials to
11 figure out what it is they think their drug does
12 before they then get around to actually designing
13 that Phase III trial and moving forward with it.
14 DR. WOOLF: I think if you look at almost
15 every published trial, you won't find that data
16 available. These are global scores. There is no
17 way of identifying whether it affects--most
18 patients are never tested to see if they have
19 stimulus-evoked pain.
20 DR. KATZ: When it has been examined, when
21 it has been looked for, how different subtypes of
22 pain respond to different medications, can anyone
23 summarize the results of that for us, trials where
24 people have tried to segregate different types of
25 neuropathic pain and see whether there is a
1 differential treatment response.
2 DR. BRIL: I remember from reading in the
3 amitriptyline studies that the stabbing pain would
4 respond to it. So would burning pain. It is
5 difficult because some of the different types of
6 pain respond to the same agents. They are not all
7 yes or no, respond or no respond, depending on
8 pain, plus the patients are not all stimulus-evoked
9 pain or spontaneous pain. They tend to have a
10 mixture of pains and that is why you don't see it
11 in the studies because they change all the time.
12 So I am not saying it not easy to drag
13 out, but what I have seen is--well, in the clinic,
14 the patients don't split into categories and, two,
15 I remember the amitriptyline story on those pains
16 and carbamazepine was better for stabbing pain, I
17 think, and not as good for burning. But I don't
18 remember all the details beyond that.
19 DR. DWORKIN: Certainly this notion that
20 persists in the literature that the tricyclics are
21 good for kind of steady burning pain and
22 anticonvulsants are good for intermittent
23 paroxysmal pain. But, in fact, if you look at the
24 studies that have assessed different kinds of pain,
25 the tricyclic studies that Mitchell has done and
1 Soren Sindrup have found responsive of intermittent
2 pain, ongoing pain and allodynia for tricyclics.
3 Peter Watson has found the same thing in
4 the OxiContin study in PHN. The gabapentin studies
5 haven't really looked at stimulus-evoked pain but
6 unpublished analyses of the McGill short form show
7 a responsiveness to gabapentin irrespective of type
8 of pain.
9 So, in fact, the data we have suggests
10 that these three types of agents, if they work for
11 one type of pain, are very likely to work for other
12 types of pain and so there isn't a symptom
14 DR. KATZ: So I think we would all agree,
15 somebody correct me if I am wrong, that assessing
16 the different subtypes of neuropathic pain is
17 important as we attempt to learn more and more
18 about this phenomenon and work towards a
19 mechanism-based approach but not to have high hopes
20 because so far it hasn't panned out.
21 Is that a fair summary?
22 DR. ROWBOTHOM: I am not sure that I agree
23 that it hasn't panned out. Part of the reason why
24 I was not strongly advocating things like skin
25 biopsy and capsaicin response is just that they are
1 not easy to do. They are not easy to get analyzed.
2 I think we need a little more data from more the
3 level of single-center or small multicenter studies
4 before you start trying to incorporate that into a
5 set of guidelines that would apply to industry as a
7 So that was really more my caution, not
8 that they hadn't worked out. It is that there just
9 really wasn't really enough known yet to really
10 push strongly on them.
11 DR. KATZ: Fair enough.
12 Dr. Farrar, you were next.
13 DR. FARRAR: I would like to just address
14 two issues to you and let you decide as to how you
15 want to approach them, but there are two other
16 issues that need to be addressed with regards to
17 pain specifically. One is whether patients who
18 have successfully been treated with another agent,
19 either similar or not similar, need to come off
20 that agent before they are tried.
21 The second is whether or not multiple
22 therapeutic options are allowed the patient. In
23 specific, this comes up with cancer patients all
24 the time is that it would be unethical to take them
25 off of their opioids to study gabapentin. What we
1 would do is to do an add-on trial, as is often
2 done with epilepsy drugs. Those are two areas that
3 we haven't covered.
4 DR. KATZ: Right. That is a great point.
5 That is actually on my list of miscellaneous things
6 to get to if we have time, and I hope that we do.
7 So let's hold that question for a second.
8 Are there any other comments about
9 characterization of patients on entry. It sounds
10 like we have advocacy for doing neurophysiologic
11 studies for diabetic-neuropathy studies and
12 presumably other polyneuropathies, at least
13 assessing allodynia in such patients,
14 characterizing symptoms based on the specific type
15 and all the other things that I think are obvious.
16 Any other points about characterizing
17 patient populations? Dr. Feldman?
18 DR. FELDMAN: Just a point that I know
19 Vera is well aware of but certainly
20 nerve-conduction studies as we have discussed
21 primarily are good for large-fiber modalities and
22 most of the pain that we are discussing today are
23 small-fiber modalities.
24 Vera, I had to step out for a moment, but
25 you are saying you want to use nerve-conduction
1 studies just to get an idea of the severity of the
2 generalized neuropathy?
3 DR. BRIL: Yes. In some previous studies,
4 the responsiveness to pain was in those who had
5 sural-nerve responses present. So it is staging
6 severity. It is just staging as you enter. I
7 mean, there is a role to look for neurotoxicity if
8 you thought you were going to get a toxic effect.
9 So, for safety, you might do it. But, basically,
10 at the beginning for staging to try and subdivide
11 the patients.
12 Primary Endpoints
13 DR. KATZ: What should be the primary
14 endpoint in neuropathic pain in clinical trials?
15 DR. BRIL: Reduction of pain.
16 DR. KATZ: Thank you. Anybody disagree?
17 DR. RENDELL: That, of course, seems to be
18 reasonable but one of the questions I have always
19 had is why do we only do such short-term pain
20 studies? They are always twelve weeks. The answer
21 to that is that is what the agency wants. But, is
22 that reasonable? The reason I am asking whether
23 that is reasonable is because in the
24 diabetic-neuropathy area we are now using pain
25 studies as a surrogate for studies of actual
1 diabetic neuropathy realizing we can't find any
2 drugs that will ever treat or meet the criteria for
3 approval of diabetic-neuropathy drugs. Why are we
4 only going twelve weeks? Why don't we go a year?
5 Why don't we go two years?
6 DR. KATZ: You are referring to
7 placebo-controlled trials that last that long,
9 DR. RENDELL: The studies we are now doing
10 are very short-term. They require that patients go
11 off all their other pain drugs but the problem is
12 companies are simply substituting pain studies for
13 diabetic-neuropathy studies. They are doing that
14 intentionally with the hope of getting approval.
15 DR. KATZ: Anyone have any thoughts on
17 DR. BRIL: Can I ask--the reason that I
18 see that patients have to come off their other
19 drugs and the reason I have always thought that was
20 a good idea for pain and painful neuropathy was
21 that there seems to be that refractory core of
22 patients who have painful neuropathy.
23 If you start recruiting these patients
24 into studies, you may be biasing yourself to a
25 failed study whereas if you have patients who are
1 not on multiple drugs, you stand a better chance of
2 showing efficacy. That is what I think. But maybe
3 that is just a wrong opinion and maybe an add-on to
4 somebody who is on two or three other drugs, and
5 add-on study, would still have the potential of
6 showing an effect.
7 DR. RENDELL: Do we want patients with
8 such severe pain, at least in those studies that
9 are surrogates for diabetic-neuropathy studies?
10 DR. KATZ: I am interested in that point
11 about studies on pain being used as surrogates for
12 disease on occasion--I was not aware of that.
13 DR. RENDELL: What is happening the
14 companies are admitting that they cannot get a drug
15 approved for diabetic neuropathy. What they are
16 doing is they are using pain as a surrogate at this
18 DR. KATZ: Anyone have any knowledge about
20 DR. RENDELL: Vera certainly does?
21 DR. BRIL: No. I don't think I agree that
22 that is what happening. I do see that there are
23 medications being developed strictly for pain or
24 that are out there already, gabapentin being one,
25 and that is being studied more now for control of
1 painful symptoms and different agents.
2 I know of novel antidepressants and novel
3 anticonvulsants that are being studied strictly for
4 the control of painful symptoms. I do know of some
5 agents that are being tested to reverse or
6 interfere with disease progression, to halt or slow
7 down or reverse disease progression and that these
8 agents are being studied from multiple points the
9 way we discussed earlier so that they are being
10 assessed with respect to their effect on
11 neurological deficits on examination and on
12 ancillary measures such as nerve conductions and
13 quantitative sensory thresholds.
14 Some are being studied by the Peter Dyke
15 scale composite score. They are also being looked
16 at with respect to their effect on symptoms which
17 is what I think we all want. I mean, it would be
18 wonderful to have a specific agent that reduced
19 neuropathic symptoms and improved nerve function
20 and reduced the sensory loss on exam.
21 If you had an agent that did all of that,
22 it would be a tremendous advance in the field
23 because we have nothing that does that.
24 DR. RENDELL: But is twelve weeks enough?
25 DR. BRIL: But these studies that I am
1 talking about are not just twelve weeks. The
2 twelve-week studies are basically the ones that are
3 designed just to show an analgesic effect for the
4 painful symptoms the same as they are in
5 postherpetic neuralgia or whatever other pain thing
6 you want.
7 I think that the issue is a little bit the
8 placebo-control group. It is difficult to go
9 beyond twelve weeks. But perhaps you need longer
10 studies to see if the pain really is sustained. I
11 mean, that is not a bad idea.
12 DR. RENDELL: That is one of the
13 questions. Isn't one of the endpoints how long the
14 pain is relieved and what happens after the pain is
16 DR. KATZ: Dr. McCormick?
17 DR. McCORMICK: I hear two questions. One
18 is the question of why are trials that are designed
19 to look at symptomatic relief of pain only three
20 months long? Is that the agency's standard and why
21 is that?
22 I think that we have considered three
23 months for most conditions an adequate length of
24 time to determine that a drug is either working or
25 not working for symptomatic relief of pain. Now,
1 that may not be correct and I would like to hear
2 further discussion on that point as to why longer
3 trials might be needed.
4 As to the other point of the twelve-week
5 trials in pain being used as a surrogate for
6 disease progression, that has not been our
7 experience. Clearly, trials that are intended to
8 look at the progression of disease are far longer
9 than that and all sponsors that have come to us to
10 date have come to us with that realization and with
11 that expectation that they are in trials for the
12 long haul, that these are going to be very long
14 So I don't think there is a single sponsor
15 yet that has come to us. Now maybe these trials
16 that you are thinking of are still in the
17 conceptual phase but, for the most part, sponsors
18 that have come to us have not had the perception
19 that a three-month trial would suffice for an
20 alteration-of-disease claim.
21 DR. KATZ: Ms. Delph, you were next.
22 MS. DELPH: I would like to add in the
23 discussion of safety when we are talking about
24 duration of trials because I would like to hear
25 what people think would be an adequate duration for
1 pivotal trials especially looking at safety and,
2 secondly, postmarketing studies. I don't know if
3 you are going into the postmarketing period but
4 certainly, in HIV where we have a lot of fast-track
5 approval of drugs, one of the big, big, big
6 problems we have had is postmarketing safety
8 DR. RENDELL: In what respect?
9 MS. DELPH: Getting companies to do them.
10 DR. KATZ: Let's focus on the duration of
11 trial issue and then we can talk about safety
12 monitoring as well. Does anybody else feel that
13 three months is not an adequate length for a trial
14 and you can be specific about what can be
15 accomplished by longer trials.
16 Dr. Farrar?
17 DR. FARRAR: I think it is important to
18 keep in mind that the two different lengths of
19 trial are going to answer different questions, both
20 of which are valid. I think it is up to the agency
21 to decide what it requires in order to do that.
22 The twelve-week trial, the three month
23 trial, is does it work for any length of time that
24 is reasonable and three months is certainly a
25 reasonable period to consider. A year trial is
1 does it then continue to work, and that is going to
2 be confounded by issues related to development of
3 tolerance, changes in the disease process and in a
4 host of other things.
5 A very reasonable question but I think a
6 different one. One of the issues, though, that is
7 very clear is that, in a symptomatic trial for
8 pain, it is unethical to allow somebody to continue
9 in substantial pain for a long period of time.
10 What that means is that if you are talking about a
11 trial for a year, you can't possibly expect a
12 patient to stay in the trial if they are not
13 getting an effect.
14 What that means also is that the way you
15 would have to analyze that data would be, then, to
16 look at success or failure, sort of a dichotomous
17 outcome. But it would be unethical to take
18 everybody off their medicines and have them go for
19 a year.
20 There may be ways to structure it
21 differently and I would be open for--
22 DR. KATZ: Dr. Rowbothom?
23 DR. ROWBOTHOM: There are two things that
24 have been brought up in the last couple of minutes.
25 One is the issue of whether or not patients can be
1 on other therapies that might alter their pain
2 while they are in clinical trial. The second one
3 is duration of treatment in a blinded clinical
5 From the perspective of the persons that I
6 see with these different chronic-pain disorders, it
7 is very hard for them to see the possible benefit
8 for them as individuals to go into a
9 placebo-controlled trial of a drug that is, let's
10 say, in Phase II when they have a 50:50 chance of
11 being randomized to placebo and then they have no
12 access to the compound open-label afterwards. So
13 there is really nothing in it for them.
14 So it is difficult enough just to convince
15 patients that, in the interest of medical science
16 or their own agenda, to try something when there is
17 nothing really for them at the end of the trial.
18 Eight weeks has been long enough for many drugs to
19 separate quite clearly from placebo and twelve
20 weeks, certain, if you can't show efficacy over
21 twelve weeks, then I think that the drug doesn't
22 work for pain.
23 For disease modification, of course a year
24 makes much more sense. But I think from our
25 discussions this morning it was pretty clear that
1 if you focus just on peripheral nerve anatomy or
2 physiology that that is very complicated and is
3 only partially related to the complaint of pain.
4 Conversely, going the other direction,
5 touching on what Dr. Shafer brought up this
6 morning, is that if you follow a pure
7 quality-of-life outcome measure, there are so many
8 components in that--pain is just one of them--that
9 it also makes it difficult to show that your drug
10 is really working for pain and that is why the
11 patients are generally coming into the clinic is
12 they have pain and they want that to be relieved.
13 So I think twelve weeks is fine. I think
14 for a patient, especially somebody with
15 postherpetic neuralgia, to say, "I want you to be
16 in a placebo-controlled study so you will get
17 placebo for the next twelve months," they would
18 say, "I am 78-years old. Twelve months is a long
19 time for me. Thank you very much, but forget it."
20 I just don't think I could really advocate that.
21 Now, if they were allowed to be on their
22 other medications and the purpose of the study was
23 to see if thermal-sensory function in their area of
24 shingles pain improved, if their allodynia was
25 going to get better, if we were going to do serial
1 skin biopsies to look if the nerve fibers
2 normalized, those kinds of measures, and they were
3 allowed to stay on other treatments and we were
4 following purely a disease-modification type of
5 paradigm, then I don't think that would be such a
7 But, from a pure analgesia perspective,
8 twelve weeks is a pretty long time for a subject.
9 DR. KATZ: Dr. Shafer and then Dr.
11 DR. SHAFER: Is there a role in these
12 chronic-pain studies for the way we would approach
13 an acute-pain study which would be essentially like
14 an opioid sparing. You wouldn't take a patient
15 post-op--so you don't get any pain relief but you
16 would them on PCA morphine and you would look at
17 sparing. Are these patients on opioids or on
18 another drug which they can essentially
19 self-titrate and you can use that to assess the
20 efficacy of the new measure.
21 DR. KATZ: Thoughts on that?
22 DR. ROWBOTHOM: That has been a problem in
23 trials, looking for opioid-sparing effect. If you
24 look at the two initial gabapentin trials that were
25 published, the subjects were allowed to us other
1 medications. They were allowed to be on an opioid,
2 at least in the postherpetic neuralgia study. In
3 many of the clinical trials that we have been
4 involved in, subjects are allowed to continue using
5 an opioid as long as it is something they
6 previously were on and they are on relatively
7 stable doses.
8 It is always possible that they are going
9 to reduce their dose during the treatment trial and
10 so you would be showing an opioid-sparing effect.
11 It is a tough outcome measure to really assess
12 because you would then be looking for a fairly
13 restricted group; okay, I want postherpetic
14 neuralgia, they have got to have four out of ten
15 pain or worse and they have to be on opioids and
16 then have one of your measures be opioid sparing.
17 I think that is probably cutting it too
18 fine to be practical. I was referring really more
19 to the ethical aspects where if you require
20 patients to go off all their medications in order
21 to be in a trial and then it is a very long trial
22 with a placebo control, that is really difficult
23 for subjects.
24 What you tend to get in those trials
25 because we have done a couple of them, and this is
1 purely my own person experience. This is, of
2 course, completely anecdotal; we get pretty strange
3 subjects for those studies. You get people that
4 either no one really believed that they had pain or
5 everything completely and totally failed and so,
6 therefore, they are just on nothing.
7 That is a bit of an unusual group. I am a
8 little more comfortable with the--and, again, I am
9 speaking from my experience more on postherpetic
10 neuralgia because that is such a kind of average
11 slice of the 55- to 80-year-old age range that they
12 are getting a little bit of response to some things
13 but it is not really enough at the doses that they
14 can tolerate. So, therefore, they are interested
15 and are able to participate in the clinical trial.
16 DR. DWORKIN: I would like to second what
17 Mike said. I personally believe that three months
18 is enough and, in fact, for a placebo-controlled
19 study of pain--in fact, I think I could argue that,
20 so if we are going to do three months, then that
21 should certainly include any titration at the
22 beginning within the three months.
23 I think I could argue that eight weeks
24 would be enough to show durability. I don't know
25 what we would get for the extra month because I am
1 hard-pressed to think of drugs where you lose
2 efficacy from week 8 to 12. The original Nurontin
3 trials published in JAMA were eight-week trials and
4 I don't think any of us thinks that if those trials
5 had gone out to twelve weeks that we would have
6 lost the efficacy of Nurontin versus placebo.
7 So I think that twelve weeks is more than
8 enough and I think I might even be comfortable with
9 eight weeks.
10 DR. KATZ: I am going to refocus the
11 discussion now back to the outcome measures because
12 it is very important that we address some questions
13 in that domain. I think that somebody said, and I
14 don't think that anybody disagreed, that pain needs
15 to be the primary outcome measure. I don't think
16 we need to quibble about whether it is a VAS or a
17 numerical rating scale or a categorical scale or
19 What about secondary outcome measures in
20 neuropathic-pain trials. What would be relevant?
21 I think that we all said that we should
22 characterize the subtypes of pain as well to see if
23 there is any sort of differential effect on one
24 symptom versus another.
25 We spoke about measuring allodynia as an
1 entry criteria for characterizing our patients and
2 I think we would all accept that as a relevant
3 outcome measure as well. So, correct me if I am
4 wrong. Any other secondary outcome measures that
5 would be particularly important in neuropathic-pain
7 Dr. Dworkin and then Dr. Farrar.
8 DR. DWORKIN: Some kind of measure or
9 measures of psychological distress, psychological
10 psychosocial morbidity, and then function, quality
11 of life, is the patient out going to the movies and
12 shopping more than they were before the trial
13 began? Those would be the other two classes.
14 DR. FARRAR: Very specifically, those
15 factors need to be measured at the beginning of the
16 trial to serve as evidence that your two groups,
17 the placebo and the treatment group, are, in fact,
18 the same in terms of the level of depression and
19 the level of function, and so on.
20 They are also vital as outcome measures
21 not necessarily because they should be the primary
22 outcome but because if I saw the pain getting
23 dramatically better but people didn't do any more
24 and they stayed as depressed or got worse, you
25 would really begin to wonder whether it was just a
1 chance finding.
2 What we are looking for, really, is to see
3 all of them headed in the right direction. If that
4 is the case, then you feel much more comfortable
5 with them. So I am strongly in favor of measuring
6 at least those two and there is a lot of reason to
7 think that you ought to be looking at coping
8 mechanisms and what patients' expectations are with
9 the trial at the base because both of those clearly
10 influence the potential outcome of the trial.
11 DR. KATZ: Speak a little bit more about
12 the expectations issue, what you are talking about
14 DR. FARRAR: This is an area that is
15 relatively new in terms of some of the ways that it
16 has been looked at. But it is very clear that
17 patients' expectation for the effect of the drug
18 influences their placebo response. If patients
19 believe that that drug that they are going to be
20 tried on has a very significant possibility of
21 helping them, then, whether or not they get the
22 real drug or not, they are going to have a better
23 expectation for it.
24 The opposite is also true. If you try and
25 enroll somebody in a trial of an nonsteroidal
1 antiinflammatory and they have tried it five times
2 before and it has never worked but you are paying
3 them $300 so they are going to do it, it doesn't
4 matter whether the drug works or not. It is not
5 going to work for them.
6 So I think it is important that you
7 measure it up front in terms of understanding,
8 perhaps, why the trial either succeeded or failed
9 and then the expectation is not an issue that you
10 would measure again as an outcome but it is very
11 clear that, at baseline, it could have an influence
12 over how your study ends up.
13 DR. KATZ: I had the opportunity to spend
14 some time with Patrick Wall at Mass General
15 Hospital. Just before he died, he visited Boston.
16 He summarized for me everything that he had learned
17 about the placebo effect in his years of
18 researching it, in just a few words, which is that
19 if you want to know who is going to have a placebo
20 effect, just ask them what they expect is going to
21 happen at the end of the trial and he can tell you
22 right up front who is going to have a placebo
23 effect and who is not. It amazed him that that
24 wasn't done routinely in the clinical trials.
25 It remains to be seen about that.
1 DR. BRIL: This is not exactly an endpoint
2 but the other thing that should be collected is
3 safety data for all of these drugs so you have this
4 balance between efficacy and safety and I am seeing
5 that more with open-label extensions that are going
6 a year or so, that people are collecting more
7 safety data to balance against the side effects
8 than against the efficacy than had been done
10 DR. KATZ: Dr. Shafer?
11 DR. SHAFER: I think this just
12 reemphasizes perhaps ground we went over but if we
13 are talking about strictly pain as the endpoint of
14 the trial, then, yes, we should measure pain. But
15 if we are back to thinking to about things that are
16 modifying disease, then functional studies like
17 nerve-conduction studies maybe we be the
18 appropriate endpoints rather than just pain,
20 DR. DWORKIN: Somehow, we have left out
21 John's favorite measure which is also my favorite
22 measure and that is some patient rating of their
23 global impression of improvement. I think that is
24 essential and awfully easy to get and may actually
25 be some kind of integration in the patient's mind
1 of pain relief and quality-of-life improvement and
2 satisfaction and side effects and psychological
4 So that is essential.
5 DR. KATZ: One question comes up from time
6 to time about the inclusion of quality of life as
7 an outcome measure. Does anybody feel that quality
8 of life should be a required coprimary outcome
9 measure meaning that, let's say, for example, pain
10 was reduced but quality of life was not changed,
11 that that would constitute a failed trial? Anybody
12 in the room endorse that perspective.
13 DR. FARRAR: I don't endorse the
14 perspective but I do want to make the point that
15 different quality-of-life scales have different
16 responsiveness. If you use a scale that is not
17 going to respond, it won't respond. So if you were
18 going to require that, you would need to be very
19 careful about using the right kind of
20 quality-of-life scale designed for that specific
22 DR. KATZ: It sounds like everybody agrees
23 that pain is pain and we don't need to second guess
24 it overly and quality of life is important as a
25 secondary outcome measure but not as the sole
1 required primary.
2 DR. BRIL: How much does pain have to
4 DR. KATZ: John, tell us. How much does
5 pain have to improve?
6 DR. FARRAR: It depends on the question
7 you are trying to answer, but if we take the point
8 of view of the patient, I think, ultimately, the
9 question is if you had a choice of taking this
10 medicine or not, would you continue to take it.
11 For a chronic-pain study, I think ultimately that
12 is the question.
13 For an acute-pain study, I think the
14 answer is a little easier because we know that, in
15 looking at whether drugs work or not--i.e., do
16 patients feel that they need to take an additional
17 dose of medication for that episode. So you give
18 somebody a study medication and thirty minutes
19 later, it should have worked.
20 At thirty minutes, you say, "Is this good
21 enough, or do you want something else?" Then they
22 can answer. If it is not good enough, then you are
23 quite convinced that that is likely to be the case.
24 What we have learned from that is that a change of
25 about 33 percent on a pain-intensity scale seems to
1 correlate very nicely with that outcome.
2 There is some data to suggest that
3 although only in a couple of studies and it needs
4 to be replicated.
5 DR. RENDELL: Just to try to rephrase the
6 issue that I am trying to get at with the length of
7 time of pain studies, if we are going to do pain
8 studies, we ought to limit them to pain. But if
9 you are going to try to add measures of
10 functionality, it doesn't make any sense to do that
11 in a twelve-week trial.
12 There is a current trial scheduled that
13 involves two sets of nerve conductions on two
14 separate days at beginning and end of trial. That
15 just doesn't make sense. I don't care what
16 nerve-conduction specialists say, you can't see a
17 chance in that short a period of time.
18 DR. KATZ: Break for an official
19 announcement. We have officially gone below our
20 quorum if we were to need to take a vote on
21 anything. But we are perfectly fine to continue
22 our general discussion. Sorry for the
24 Any further comments on the issue of
25 outcome measures and neuropathic-pain clinical
1 trials? Have we missed anything important?
2 DR. FARRAR: One way of perhaps getting at
3 this issue of length of time, I think one of the
4 primary questions in the study that you would want
5 to look at for longer than three months is whether
6 or not the drug continues to provide benefit
7 because, as Mike and I think Bob, also, clearly
8 said, if it doesn't work by eight weeks, it is time
9 to stop.
10 For prevention trials, preventing
11 progression of disease, that is a different issue,
12 very different. In someone with pain, if you
13 haven't created some benefit for them by eight
14 weeks, then it is not going to work at all. What
15 you may want to look at and, in fact, some
16 companies now tout this in some of their
17 discussions, which is to say, in the follow-on
18 trial, 30 percent of the patients stayed on the
19 drug for a year as evidence that it continued to
20 work for that patient. I think, in some ways, that
21 is a valid way of looking at it.
22 DR. KATZ: Is there a wy of making that
23 work stand from a clinical-trial point to obtain
24 some statistical evidence that the drug is working
25 by influencing the disease process rather than as a
1 pure analgesic?
2 DR. FARRAR: I don't know that I can
3 answer that question. I think it is sort of mixing
4 apples and oranges. If we are trying to treat the
5 symptoms, then what we are measuring is the
6 symptoms. If you think the drug actually has an
7 effect on the disease process, then you need to
8 structure your trial completely differently. I
9 think that point is valid, but if you are looking
10 simply at pain, then I think the issues are pretty
11 much straightforward.
12 DR. KATZ: Dr. Dal Pan?
13 DR. DAL PAN: Dr. Farrar mentioned earlier
14 that he didn't like mean values. I was wondering
15 if the group could just discuss a bit a responder
16 type analysis where patients are treated as
17 successfully treated or not successfully treated
18 and the analysis essentially a comparison of
19 proportions between two groups versus making
20 inferences based on mean values of pain scores or
21 changes in pain scores or something like that.
22 DR. ROWBOTHOM: That is being done quite a
23 bit now with this number-needed-to-treat analysis
24 where you look at the proportion who meet some
25 criterion in the active group, subtract it from the
1 placebo response rate and you come up with a
2 number, and the smaller the number, the better.
3 So a drug with an NNP of between 3 and 5
4 is considered a really good drug because that would
5 mean that you would need to treat between 3 and 5
6 patients before you got one that had this level of
7 response. So that is being done quite a bit.
8 I did want to mention something about what
9 Dr. Rendell brought up a couple of times about the
10 monitoring. I think, perhaps, there is some
11 confusion or, if not confusion, lack of clarity or
12 trying to do two things at once, and that is if you
13 are doing a lot of complicated electrophysiologic
14 testing, nerve conductions and things like that
15 that would require an experienced person to do, and
16 a lot of equipment, and you are doing them so close
17 together, then what you are really doing is some
18 kind of intensive safety monitoring rather than
19 disease modification.
20 It seems that, at least what I am coming
21 away with from the discussion today is that disease
22 modification and pain are really different things
23 and so trials should be designed to look at those
24 issues separately and not necessarily try and do
25 both at once by either doing very, very short
1 disease-modification studies because you can
2 monitor pain over twelve weeks, because that is not
3 enough time to look at disease modification, or
4 require that studies of pain be extended to very,
5 very long periods of time because that is how much
6 time you need to look at disease modification.
7 It is probably good that they be kept
8 somewhat distinct.
9 DR. KATZ: Actually, Dr. Woolf, you were
10 on deck first.
11 DR. WOOLF: I terms of outcome, we haven't
12 discussed active comparator as an element. We are
13 talking about detecting efficacy but one issue is
14 efficacy relative to what, just to placebo or to
15 something that has been shown in the literature to
17 DR. KATZ: So there are a number of issues
18 hanging in the air right now that haven't been
19 addressed. So I am going to try to force us to
20 address them one at a time.
21 Let's go with Dr. Dal Pan's question
22 first. The advantages and disadvantages of using a
23 mean change in a pain score as, say, the primary
24 outcome measure for a trial versus a dichotomous
25 response index of some kind, you are a responder or
1 you are not a responder and you compare the
2 proportion of responders in each group.
3 Let's just deal with that. Actually,
4 John, you have written on that so maybe you would
5 like to summarize the advantages and disadvantages
6 of each approach.
7 DR. FARRAR: I am happy to do so although
8 I think Bob was actually first. The primary issue
9 revolves around deciding whether a medication for
10 symptom management works or not. I think it is
11 important to differentiate that from one that
12 influences the course of a disease because I think
13 there clearly is a difference in considering those
14 two entities.
15 DR. KATZ: We will focus on pain for now.
16 DR. FARRAR: Yes; I understand. With
17 regards to pain specifically, the issue is that a
18 mean value or any central-tendency value--it can be
19 mean, median or mode--does not provide a unique
20 solution to the idea of how many people actually
21 get better.
22 The primary reason that mean values are
23 used, at least for historical reasons, is because
24 there is some misconception that a mean value or
25 using a continuous analysis provides you more power
1 so you don't need as large a study.
2 There are also some issues related to how
3 you actually then determine the effect of the
4 study. One of the biggest criticisms that I hear
5 is, well, if you decide that you want to do--if you
6 do a responder analysis, you have to decide what a
7 response is. That makes people uncomfortable. So
8 a number of people have said to me at various
9 points, well, if you just look at the mean value,
10 you don't have to decide what is important. It is
11 just statistically significant or not.
12 My argument is that it doesn't matter
13 whether you get a mean or you do a proportional
14 analysis, you have to, at some point, decide what
15 is clinically important and you may as well do that
16 up front.
17 The second issue with regards to
18 proportional analysis or looking at a responder
19 analysis in pain specifically is that all of our
20 measures measure a subjective response of the
21 patient. Since every patient responds differently
22 and uses the scales differently, the appropriate
23 approach, it seems to me, is to look at the
24 clinically important difference within the patient,
25 decide what is important for that patient, whether
1 it is 33 percent or being able to walk or whatever
2 measure you would like to use, and then looking at
3 the number of people who actually respond within
4 the two groups.
5 There is a third issue which I think I
6 have not yet been able to find a trial that
7 actually clearly demonstrates this but at least
8 theoretically it is possible to have a mean value
9 that is identical in two trials and have the
10 proportional analysis be distinctly different.
11 There is the possibility that if you have one group
12 that responds and one group that doesn't that we
13 don't know a priori, that you could actually get
14 the wrong answer using a mean value.
15 DR. KATZ: Dr. Dworkin.
16 DR. DWORKIN: We have a paper under review
17 now that is a PHN trial. After John's paper
18 appeared, we did the analysis that is kind of
19 suggested in John's paper which is we looked at the
20 proportion of responders who respond with a 33
21 percent reduction in the active arm--
22 DR. PERLMUTTER: I don't think there can
23 be a fully general answer to that question. The
24 answer is it depends. There are certainly
25 situations in which you will lose quite a lot of
1 power by dichtomizing a variability and there are
2 others in which you won't. My sense of what Dr.
3 Farrar is getting at is there are methods that
4 actually have most of the advantages of both, that
5 you can do methods with good power which,
6 nevertheless, can be interpreted in this elegant
7 way in terms of responses.
8 DR. KATZ: Could you expand on that a
9 little bit in terms of the methods that you are
10 referring to?
11 DR. PERLMUTTER: For example, the Wilcoxin
12 Mann Whitney Rank Sum Test can be viewed as based
13 on the ensemble of all possible dichotomies. So
14 someone just said a few minutes ago that one of the
15 problems with the responder analysis is you have to
16 decide up front what a responder is.
17 Well, suppose you don't decide up front
18 what a responder is but you consider all possible
19 definitions of what a responder is. You can
20 actually do a statistical analysis based on all of
21 those tests simultaneously with the appropriate
22 corrections for the fact that you are doing all of
23 them and sort of picking the best one.
24 The rank sum test and what I think is a
25 little better than even than normal scores test can
1 be viewed in this way. Other methods along those
2 lines I think can get you most of the advantages of
3 both of the responder analysis and the purely
4 parametric analysis.
5 DR. KATZ: Thank you.
6 DR. DWORKIN: John, I am not sure I
7 understand the power issue because if we agree that
8 this is an elegant endpoint that really captures
9 what we are interested in then, if we have lost
10 some power, so what? We just have to have some
11 more patients in the trial.
12 DR. KATZ: Are you paying for them?
13 DR. DWORKIN: If that is the endpoint, so
14 we pay for it. Power doesn't do it for me if we
15 agree that that is an elegant valid endpoint.
16 DR. FARRAR: My guess is that the people
17 on that side of the room probably care more about
18 it than you do. But I would like to expand and
19 suggest something that maybe actually we could talk
20 about later which is that using an ordinal analysis
21 gains you almost all of the components of other
22 forms of regression with very small loss of power
23 and ultimately gives you an analysis of whether at
24 every possible cutoff--and I think that is what you
25 were getting at--at every possible cutoff, one is
1 better than the other.
2 What is really nice about it, and we have
3 a paper that is currently being put together on
4 this is that you can actually draw a graph and show
5 that--so that if Bob likes 33 percent because I
6 told him it was the right thing and Dr. McKway in
7 the U.K. prefers 50 percent, you can look at the
8 graph and see the differences between the two
9 groups at all levels. There the power issue is
10 tiny, so I think it would even make these folks
12 DR. KATZ: Certainly others have suggested
13 the approach of using the parametric analysis as
14 the primary outcome measure and then using a
15 responder analysis which may be more intuitively
16 understandable as a secondary outcome measure so
17 you have potentially the best of both worlds that
19 Any other thoughts about outcome measures?
20 Ms. Delph, did you have a comment?
21 MS. DELPH: As far as outcome measures are
22 concerned, I wondered about the value of adherence
23 to medication. The other thing, in terms of
24 measuring outcome, does the baseline severity of
25 disease, whether measured by intractability, level
1 of pain or whatever, does that matter in terms of
2 patient population when you are assessing outcome?
3 DR. KATZ: I think it has become routine
4 to use baseline pain as, in some studies, actually
5 frankly stratifying based on baseline pain and in
6 others at least using it as a covariate analysis at
7 the end because that does seem to be associated
8 with treatment response in many studies.
9 Does anyone have anything to add to that?
10 The other issue was adherence. Certainly,
11 I would guess in most industry-sponsored trials,
12 adherence is monitored but based on things like
13 pill counts, I think is the standard and who knows
14 if the patient took them or flushed them down the
15 toilet in the waiting room.
16 Go ahead.
17 DR. FARRAR: A quick comment. A friend of
18 mine down the hall studies HIV in patients and
19 where adherence to the use of drugs that make
20 patients feel really lousy is a big issue. The
21 advantage we have in pain management is that if you
22 have got a drug that makes people feel better,
23 there is no problem with adherence.
24 MS. DELPH: Which is why I am asking
25 whether it would be valuable as an outcome measure.
1 DR. FARRAR: I don't think so. I think
2 whether they take them or not is going to be
3 dependent on many, many different factors and what
4 you are really looking for is efficacy since you
5 know that patients who get better are going to take
6 the drug. I don't think it is an issue of trying
7 to figure out whether the ones who don't get better
8 don't take it. I think it is really an issue of
9 whether they feel better or not and how you measure
11 DR. KATZ: I actually wonder
12 whether--certainly there is experience in other
13 areas of clinical trials where adherence is a huge
14 issue and trials have failed because people have
15 side effects and don't take their medications.
16 There have been a variety of approaches that have
17 been used and I am sure the folks from the FDA know
18 a million times more about this than I do but, for
19 example, putting inert markers in tablets and
20 measuring urine tests to make sure people are
21 taking their medication, having bottles that it
22 records it digitally when you open the bottle.
23 Of course, you can open it and flush it
24 down the toilet again, but it still gets you one
25 step closer to understanding a true adherence,
1 diaries for medication consumption, that sort of
2 thing, because a small number of nonadherent
3 patients in the treatment arm can completely
4 distort the end results of the trial. But I don't
5 know how commonly--I don't think these things are
6 commonly done. I don't know what the regulatory
7 perception is about how big a problem it really is
8 in actual practice.
9 DR. McCORMICK: I think they are fairly
10 commonly done in trials but I am not sure how much
11 we really use that information. I think the point
12 was a good one that the fact that patients who
13 don't take their medications don't respond doesn't
14 really help us in the end.
15 Dr. Hertz just pointed out that more
16 frequently than not, if patients are not tolerating
17 the drug, they drop out of the trials.
18 DR. ROWBOTHOM: I just wanted to make one
19 comment picking up on what you are saying. It is
20 not really about adherence, per se, but it is an
21 important one and that is if you are following and
22 intent-to-treat type study and data analysis, then
23 the patients who are either not adherent or who
24 drop out of the study, they are still counted by
25 the outcome.
1 Some of the trials that have been
2 mentioned during the day today are problem trials
3 because they didn't follow an intent-to-treat
4 analysis. They only looked at the subjects who
5 completed the entire study.
6 So if you take a study and you break down
7 the data and you look at the patients who completed
8 everything and leave out the data from the subjects
9 who didn't complete, then you will get very
10 different results and it will usually overestimate
11 the treatment benefit.
12 DR. KATZ: Dr. Perlmutter?
13 DR. PERLMUTTER: I agree with that. I
14 just want to say there are some good ways, I think,
15 of taking adherence to treatment into account in
16 the analysis of trials without violating the
17 intent-to-treat principle, but I agree with you
18 completely that the way to do that is not just to
19 leave out the nonadherent patients.
20 DR. DWORKIN: Your point, Ms. Delph, made
21 me think of something we haven't discussed which is
22 that industry seems to have consensed on an average
23 pain rating of 4 or greater for entry into these
24 trials in a kind of baseline week of ratings.
25 So now I am going to say something that is
1 going to make me even more unpopular on that side
2 of the room. When we place ads in newspapers or
3 get referrals from primaries, there are a whole lot
4 of patients who come in and fail entry criteria
5 because their baseline weak of pain ratings is a 3.
6 So their average over seven days is a 3 and they
7 are not close to the 4 required in these trials.
8 And those patients are real disappointed that they
9 can't participate in the research.
10 To my mind, that begs the question of
11 whether we are setting the bar too high. Now, I
12 realize there is going to be a loss of power if we
13 set the bar at 3 or even 2 or 2.5, but there are a
14 lot of people out there with chronic pain that we
15 might say is in the mild to moderate range who
16 desire treatment enough to be interested in
17 enrolling in a placebo-controlled trial and we are
18 excluding them from all ongoing studies that I am
19 aware of.
20 DR. KATZ: Thoughts on that issue? You
21 are correct in that there is literature suggesting
22 that there would be a loss, a floor effect and a
23 loss of power from dropping below there.
24 Dr. Farrar?
25 DR. FARRAR: I had a patient that I
1 treated who had the worst pain I had seen in years
2 and his pain was never worse than a 3. The reason
3 is because on the worst end of the scale, he
4 imagined his father in a concentration camp and
5 that was enough to move him all the way down the
6 scale. So there clearly are people who use 3 or
7 have 3 as a measure who have intensely bad pain.
8 The issue, I think, primarily is that you
9 have to have enough of the scale to move in order
10 to be able to accurately measure the amount. One
11 might be able to put patients in who are at a 3.
12 If my data is correct and 33 percent is a
13 reasonable drop, then going from 3 to 2 would be 33
15 The trouble is you get to 2 and it is 50
16 percent or nothing. So you end up losing the
17 ability to be able to differentiate that. We could
18 argue about 3 and 4, but I think the issue is not
19 whether there aren't patients that would be good to
20 have in the trials but, rather, a measurement issue
21 and that makes it sort of the reason we have to
22 stick with that.
24 DR. KATZ: We are now officially in the
25 wrapup phase of our session. So I would like to
1 turn to the FDA folks and ask them if they would
2 like to focus the discussion in any particular
4 DR. McCORMICK: Actually, I have one
5 question that is an extension of the debate that we
6 heard earlier and the discussion that surrounded
7 that debate. You have all received a copy of the
8 guidance for industry on the burden of evidence, of
9 establishing evidence in clinical trials.
10 The writers of the guidance envisioned
11 situations in which a single clinical trial might
12 be used to--when an indication had already been
13 established, to extent that indication of there was
14 sufficient pathophysiologic similarity across
15 disease states to warrant that.
16 I guess my question for the committee
17 is--and I feel that we haven't quite come to
18 closure on the lumping and splitting. I feel that
19 is still up in the ethernet somewhere--that my
20 further question is do you think that this group of
21 diseases or disorders that manifests themselves
22 with pain are sufficiently similar such that we
23 might be able to, let's say if we have had an
24 indication for postherpetic neuralgia and then we
25 have another single clinical trial in another
1 neuropathic pain state that that might be
2 sufficient to get a claim for that other disorder,
3 not necessarily a broad general claim for
4 neuropathic pain but an additional condition.
5 DR. KATZ: Thoughts on that? Do people
6 feel that if you have, for example, two adequate
7 and well-powered trials for a painful diabetic
8 neuropathy showing a very believable successful
9 result and now you have got another trial that
10 comes along, single trial, postherpetic neuralgia,
11 very believable, should that be sufficient to hold
12 in abeyance this replicate-trial rule and would we
13 believe that that drug is probably efficacious in
14 postherpetic neuralgia based on a single trial?
15 DR. McCORMICK: I guess the follow up to
16 that is what evidence would you need to be able to
17 say yes to that.
18 DR. KATZ: From the specific trials that
19 have been done.
20 DR. McCORMICK: In that specific trial.
21 DR. BRIL: I would accept a single trial
22 as an add-on to another--I would lump that far. So
23 if you had two replicate trials in one indication
24 and then a very robust trial as well as in another
25 indication but a single one, I think there are
1 enough similarities in neuropathic pain to enable
2 that to happen.
3 We were going to lump all neuropathic
4 pain. So this is similar enough to me and the same
5 level of evidence you had in one of the two trials
6 that were for the original indication, if you had
7 that in another indication, I think that would be
8 good, so a 33 percent pain reduction or a responder
9 analysis or whatever particular measure was being
10 used in these trials, I am not sure you would need
11 to replicate that.
12 DR. KATZ: You would want to see, though,
13 that there was a satisfactory clinically meaningful
15 DR. BRIL: Oh, yes.
16 DR. KATZ: As an example of robustness.
17 DR. BRIL: It would have to be a robust
18 study. If it were weak or marginal or uncertain or
19 there was criticism of the study for some reason,
20 the patient population was skewed somehow or it was
21 all in one center--there are things that would
22 limit it but if it was a multicenter, well-run,
23 well-powered study with well-defined patients and
24 the results were very clear and unequivocal and
25 replicated what had happened in the other two
1 trials, I that would be acceptable.
2 DR. DWORKIN: I am uncomfortable with
3 saying yes to your question, Dr. McCormick, and so
4 I will answer it with a question. I don't know
5 what the precedent is for SSRIs. If I have two
6 positive trials, say, for generalized anxiety
7 disorder with my favorite SSRI and I think do a
8 positive trial in social anxiety disorder, is the
9 precedent that that gets me the second indication?
10 I think if the precedent is yes in the
11 context of anxiety disorders, then I would be more
12 comfortable in going from two positive PHN to an
13 additional indication for DPN if it is positive.
14 But if the precedent in anxiety disorders is no,
15 then I wouldn't be comfortable in our domain.
16 DR. McCORMICK: I can't comment on the
17 precedent for anxiety disorders but I can comment
18 on epilepsy trials where an indication has been
19 granted in many trials in many of the drugs that we
20 have for complex partial seizures and then a single
21 trial in Lennox Gasteau was granted based on--an
22 indication was granted for Lennox Gasteau based on
23 a single trial.
24 DR. DWORKIN: But isn't it the case that
25 the percentage of failed trials in things like
1 depression and anxiety is much higher than in
3 DR. McCORMICK: Yes.
4 DR. DWORKIN: And probably the better
5 analogy for precedent would be psychiatric
6 disorders than epilepsy for neuropathic pain. I
7 guess that is the way I was thinking because my
8 understanding is that the packages for SSRIs in
9 depression had as many negative trials as positive
11 DR. KATZ: But just to focus on the issue
12 at hand, Dr. Dworkin, if you had two trials sitting
13 in front of you that were adequate and
14 well-controlled for painful diabetic neuropathy
15 that you had no questions about and then another
16 one came along in postherpetic neuralgia, enough of
17 a sample size, results seemed robust, what would
18 you believe? Would you believe that that drug was
19 likely efficacious in postherpetic neuralgia or
21 DR. DWORKIN: I am uncomfortable. I would
22 want to know if other trials had been done in that
23 indication and what the results were.
24 DR. KATZ: They haven't.
25 DR. DWORKIN: They haven't. I don't know
1 what other people think. I would be uncomfortable.
2 DR. KATZ: Would that be a yes or a no?
3 DR. BRIL: He is a splitter.
4 DR. DWORKIN: It is a no. I'm a splitter.
5 DR. KATZ: Dr. Woolf?
6 DR. WOOLF: I actually did do my homework
7 and I read it and it seemed to positively exclude
8 symptom control, the guidelines. It specifically
9 said for life-threatening or serious--is that true?
10 Are these the criteria for a single additional
11 trial? I do remember it saying symptom control was
12 not envisioned as being--
13 DR. McCORMICK: I think, in this
14 particular section, studies in closely related
15 disease, it really was not referring to terminal
16 illnesses or serious life-threatening diseases, but
17 in general.
18 DR. FARRAR: Bob's question and Nat's
19 pressure to answer does raise a question that fits
20 in with what you have asked which is getting a
21 positive trial, a single positive trial, given the
22 nature of p-values, does suggest that at least in 1
23 out of 20 products you might get a single trial
24 that is positive by chance.
25 One of the questions, then, would be if
1 there were a bunch of negative trials and then a
2 couple of positive trials whether that becomes
3 adequate for an indication. I honestly don't know
4 how that fits with your criteria in terms of the
5 initial indication and then subsequent indications.
6 DR. McCORMICK: I think we would weigh the
8 DR. KATZ: Any other thoughts on this
9 single-trial issue? Dr. McLesky?
10 DR. McLESKY: I was just going to say, in
11 response to Clifford's comment, that I was pleased
12 that the guidance was delivered to us to read and,
13 in fact, to help focus us. From my reading of it,
14 Dr. McCormick--you are the expert in this, but from
15 my reading of it, it seemed to imply that there is
16 judgment left with the agency to determine, in that
17 particular drug class and in that particular
18 patient population and disease groupings, if it
19 does make sense to have just one single trial for a
20 new indication or a new patient subunit tested.
21 That is really the question I think that
22 the FDA would like to hear answered here. Are you
23 comfortable in this group of disease states? Are
24 they similar enough, if there is good evidence with
25 a specific drug, to have that drug then, if there
1 is corroboration in another kind of a similar
2 disease state, are you comfortable having that
3 indication spread over?
4 DR. KATZ: It seems to me that
5 false-positive clinical trials in neuropathic pain
6 are unusual if they exist at all. To have a
7 clinical trial show that a drug works for
8 neuropathic pain but then find that, through some
9 subsequent process, clinical practice, surveillance
10 studies, you know, what have you, that it actually
11 doesn't work.
12 Can anyone think of an example of that?
13 Maybe dextromethorphan is the one example I can
14 think of which I swear it doesn't work at all in
15 clinical practice but there are trials. But if you
16 look at the details of those trials, they would not
17 meet what one calls robust criteria of any sort, I
18 don't think.
19 Mexiletine? No, again, I think that is a
20 debatable point. I have a number of patients on
21 long-term mexiletine treatment. You agree with
22 that? So I don't know. I think that my own
23 understanding of the literature and what I have
24 seen, and I would welcome other people's
25 perspective on this, that I am not aware of a true
1 false-positive trial where a single trial appears
2 robust but then the medication winds up actually
3 not being efficacious in clinical practice. I am
4 not talking about the things that eventually come
5 off because of safety reasons.
6 Does anybody disagree with that? Dr.
8 DR. ROWBOTHOM: I was going to say yes to
9 your earlier question that Bob was having such
10 difficulty with saying yes or no. But I will say
11 yes to that one. Obviously, it is going to get
12 more complicated if, let's say, a very similar drug
13 was studied in that disorder and proved inactive or
14 if you were in a situation where there are now
15 multiple studies, some positive, some negative, and
16 you were trying to get a second indication.
17 That, of course, goes to the agency to
18 sort out but if you have--the premise, as you
19 stated it, I have no problem with. I think the
20 only comment I would make in response to or in
21 follow up to my yes is that I think you still want
22 to try and encourage study, if not to establish
23 efficacy, to at least look at safety and
24 tolerability in some of the less well-understood or
25 seemingly less-responsive disorders so that we can
1 get out of the current cycle we are in where there
2 are really only a few neuropathic-pain disorders
3 that are being studied and a very large collection
4 of neuropathic-pain disorders that are going
6 DR. KATZ: Clearly, your point about
7 safety is worth--a trial that would satisfy us with
8 a demonstration of efficacy would not necessarily
9 satisfy us with a demonstration of safety in that
10 particular population.
11 Dr. Dworkin?
12 DR. DWORKIN: I guess an example that
13 occurs to me and that is carbamazepine where there
14 is an indication for trigeminal neuralgia and there
15 are four or five trials that are inconsistent among
16 themselves in diabetic neuropathy. By this
17 criterion, given that there is an indication for
18 trigeminal neuralgia, the existence of one or two
19 positive trials in diabetic neuropathy should give
20 carbamazepine an indication for diabetic
22 But my sense, and you guys know much more
23 than I do, is no one thinks that carbamazepine is
24 an especially efficacious drug in diabetic
25 neuropathy or that we don't know what the answer is
1 to that question.
2 DR. KATZ: I think it is not used that
3 much because there are agents that are more
4 well-tolerated that don't require monitoring of
5 blood tests. I don't have, myself, any reason in
6 my own experience to think it is not efficacious.
7 Do you disagree?
8 DR. ROWBOTHOM: I would agree with what
9 you are saying but also I don't think that any of
10 those studies of carbamazepine for diabetic
11 neuropathy would meet at least my conception of the
12 hypothetical situation you were putting forward.
13 Those were not large robust well-controlled
14 studies. They were mostly older studies, smaller,
15 and they don't really meet the current criteria for
16 how good multicenter properly controlled clinical
17 trials are conducted.
18 DR. KATZ: It sounds like what you are
19 saying is that it does get back to the judgment
20 call and that there are circumstances where the
21 robustness of the program of the whole and the lack
22 of any other negative mitigating factors could give
23 the agency reason to approve that second indication
24 from just a single positive trial.
25 Yet there are other circumstances where a
1 trial that might not be so strong or that might be
2 contradicted by other evidence would allow them to
3 make a judgment against that second indication. Is
4 that more or less what you are saying? Does
5 anybody disagree with that perspective that there
6 are circumstances where a second indication could
7 be given based on a single positive trial in the
8 right circumstances and that it shouldn't be
9 absolutely ruled out? Do you agree with that,
10 John, Clifford?
11 DR. FARRAR: Yes; I do. I think the issue
12 is how you define robust and, just to be absolutely
13 clear, it has nothing to do with the statistical
14 significance. So, provided that there is adequate
15 evidence that it really creates a clinically
16 important improvement in the patient population, I
17 have no problems with it.
18 DR. KATZ: Clifford?
19 DR. WOOLF: Because, as we recognize,
20 there will be a 1 in 20 chance of a false positive,
21 I feel just a little bit uncomfortable. I would
22 feel much more comfortable if there had been two
23 replicate studies of diabetic neuropathy and
24 postherpetic neuralgia and then a third one for
25 radicular. Then I would be very comfortable.
1 DR. KATZ: Everybody wants to be
2 comfortable. Ms. Delph, you had a comment?
3 MS. DELPH: I have a question for the FDA.
4 Does the indication have to be an all-or-none, yes,
5 it is indicated or no, you don't give the
6 indication or can you, in the labeling, give the
7 clinical-trial information that is available and
8 give some conditional indication that it may be or
9 under some circumstances or that kind of wording.
10 DR. McCORMICK: I didn't catch the first
11 part of your question which I think defined what
12 the results of the trial were.
13 MS. DELPH: No; sorry. I think we are
14 assuming that you have two good trials that give a
15 particular indication and then a third one that is
16 scientific sound that gives a possible second
17 indication. What I am asking is, for that second
18 indication, does it have to be all or none? In
19 other words, do you either give the indication or
20 not give it or, in the labeling, can you indicate
21 that this is the scientific information available
22 to us and, therefore, it may or may not be
23 indicated in certain individuals with this
25 DR. McCORMICK: First of all, by a
1 possible indication or possible positive trial, do
2 you mean that the results are equivocal in the
3 trial or the results are positive and we are the
4 point of deciding whether or not to grant the
6 MS. DELPH: Yes; that is what I am saying.
7 You have the one trial and the results are
8 unequivocally positive in that trial.
9 DR. McCORMICK: The reason for asking this
10 question now or beforehand is because we really
11 need to know what our criteria are for granting an
12 indication. We really don't give provisional
13 indications in the labeling. We either have to
14 make a determination at the time of approval that
15 the drug will be indicated for that condition or
16 not. We can't really say, "You decide." That is
17 really not an option. So we really have to make
18 that determination, do we have the grounds, based
19 on the evidence that we have before us, that this
20 drug will be indicated for that condition. That is
21 why we are deliberating about it now.
22 DR. KATZ: In the few minutes we have
23 left, I wonder if we could address the issue that
24 Dr. Farrar mentioned earlier which is the whole
25 issue of adjunctive therapy because that comes up a
1 great deal and there are a lot of important
3 So, for example, you want to do a trial on
4 Drug X for neuropathic pain but we know now that,
5 let's say, for postherpetic neuralgia, we know now
6 that gabapentin is effective for postherpetic
7 neuralgia. We know that amitriptyline is and other
8 tricyclics. Can we justify withholding those
9 medications from people? Can we carry out a
10 clinical trial that is likely to show efficacy when
11 the patient already has other analgesic medications
12 on board?
13 If we did decide to do that for ethical
14 reasons, would we then be granted a label for
15 adjunctive therapy and not monotherapy which I
16 think many sponsors look at as a potential
17 albatross. What do people feel about those issues?
18 Everyone is numb by now.
19 DR. BRIL: I would like to see some
20 adjunctive studies because I would think that some
21 of the medications could be synergistic and you may
22 have more relief than you would have with either
23 alone. However, for the reasons I said before, I
24 think it is more difficult. You may pick
25 refractory patients who are going to fail to
1 respond. If patients are already on a drug or two
2 and they still have a lot of pain, usually I am
3 very pessimistic about their outcome.
4 So, although I would like to see
5 adjunctive studies, I think that monotherapy is
6 probably the initial thing for a new drug for
7 diabetic-neuropathy pain anyway and then, perhaps,
8 there could be some requirement or suggestion that
9 adjunctive studies--or that there be an adjunctive
10 arm, there be a placebo arm, a single drug arm and
11 an adjunctive arm, something of that nature.
12 DR. KATZ: So even though that wasn't
13 necessary for an indication, an adjunctive--I mean,
14 that is a separate thing. But, still, I think we
15 all would like to see data on potential synergism.
16 Do people feel that it is ethical to have patients
17 off of neuropathic pain medications completely for
18 twelve weeks now that we know that several of them
19 are efficacious? Dr. Farrar?
20 DR. FARRAR: I think Mike actually alluded
21 to this earlier which is that if a patient with
22 diabetic neuropathy is 100 percent better on a
23 drug, they are not going to volunteer for your
24 clinical trial. I think what you are going to get
25 is patients who got 30 percent relief, some percent
1 of relief, but not adequate relief from, say,
2 tricyclic antidepressants.
3 In a setting where the drug you are going
4 to be testing can be demonstrated to be safe in the
5 combination, and that is obviously the issue, I
6 would argue for doing the study allowing patients
7 to come into the study on whatever they are on,
8 stay on whatever they are on for the period of the
9 twelve weeks to see whether what you are using
10 makes them better or not.
11 The argument is that, I think, if you
12 have--or the argument would be that if you have
13 patients who are completely cured by one particular
14 drug, they may remove from the population people
15 who are more responsive. I think that is true.
16 But I don't think that gets away from the ethical
17 issue of if something is helping a patient a little
18 bit, it is hard, ethically, to take them off.
19 You don't lose anything, I think, by
20 trying to treat them with a second drug as long as
21 there is not an interaction.
22 DR. KATZ: Granted that there may be an
23 ethical advantage of allowing patients to remain on
24 their baseline medications even though their
25 provided only partial relief, you would advocate
1 adding on the study drug or placebo to what they
2 are on. What type of indication would such a drug
3 get? Would it be indicated as adjunctive therapy
5 DR. McCORMICK: If you were to lave
6 patients on their existing medications and then do
7 a placebo-controlled trial with the new agent, then
8 they would get an indication for adjunctive
10 DR. KATZ: Even if they were on a
11 hodge-podge of different medications.
12 DR. McCORMICK: Even if they were; right.
13 DR. KATZ: So how do we deal with that
15 DR. DWORKIN: Even if those medications
16 have no indications for this condition?
17 DR. McCORMICK: That's a tough one.
18 DR. DWORKIN: They are on tricyclics which
19 don't have an indication for neuropathic pain.
20 DR. McCORMICK: In reality, I think that
21 they are being used to treat the pain. I think
22 what we would probably do is describe that in the
24 DR. FARRAR: To try and be concise about
25 it, there is a tremendous argument in the
1 literature about whether it is even ethical to do
2 studies with placebo in pain-related clinical
3 trials. I would have a great deal of difficulty
4 getting a study through my IRB that said I had to
5 take a patient off something that they were already
6 on that was working at least partially for them.
7 What I think may be the mechanism is what
8 is used in epilepsy trials which is that the
9 initial study is an adjunct study and then, at the
10 end of the study, if patients get dramatically
11 better, you can say, let's take you off of the
12 tricyclic and then show that they continue to have
13 benefit, showing that monotherapy ultimately
14 provides them with the benefit.
15 Now, I don't know how to structure that
16 trial specifically with regard to regulatory issues
17 but that would certainly convince me.
18 DR. McCORMICK: Actually, the way those
19 trials are usually done, or usually what happens in
20 those scenarios is that the product has
21 demonstrated efficacy and then subsequent trials
22 were done as monotherapy. You don't have the
23 withdrawal effect or issues of crossover.
24 DR. KATZ: Other comments about the
25 adjunctive therapy issue?
1 MS. DELPH: I find it very difficult to
2 agree with taking patients off medication that is
3 working for them without proven efficacy of the
4 investigational agent. I like the suggestion of
5 the design that you are giving. I know,
6 statistically, when you start involving other
7 agents, it is a nightmare. But I also wonder about
8 things like crossover trial designs, whether those
9 would be useful.
10 DR. KATZ: It is clearly a complicated
11 issue and crossover trials have their own baggage
12 that makes them frequently difficult to interpret.
13 It is interesting to note in the context of taking
14 patients off their medications that many of the
15 trials that I have seen in osteoarthritis and
16 similar indications where people are taken off
17 their baseline medications, there is a flare. They
18 are enrolled in the trial and they get put on
19 either an active treatment or placebo, the patients
20 in the placebo arm typically have at least as good
21 pain relief as on their previous acting drug if not
22 actually better. So I am not sure that, in real
23 life, there is actually any consequence of taking
24 patients off medications that they think are
25 working for them.
1 I don't know if it is the fact that they
2 are in the loving hands of a clinical-trial unit or
3 if the drug was actually not that efficacious for
4 them, but I am not sure if it is more of an
5 imaginary issue or a real issue.
6 DR. BRIL: I think, in practical
7 experience, if you have a patient who is doing well
8 on drugs, you don't take them off them because the
9 response rate is so uncertain. If you have someone
10 who is responding to therapy, you don't really put
11 them into these drugs. It is the people who don't
12 respond, who are on drugs, they are not any better.
13 They still have a lot of pain and those are the
14 patients who will come off their drugs because they
15 are not helping them anyway.
16 Usually, yes; these are short-term trials.
17 This is why they are short, I guess, and you
18 usually have rescue medications and then you have
19 dropouts. So there are ways to handle the ethical
20 issue of having a placebo arm in the trial, or ways
21 of considering it, that I think are fairly ethical.
22 But it is not standard practice to have
23 someone who is well-controlled and take them off
24 their pain medications and put them in a pain
1 DR. KATZ: It is hard to be prescriptive
2 about that, I think. I think that when I was
3 seeing patients actively, I would spend as much of
4 my time taking people off medications that they
5 thought were working but, in retrospective, weren't
6 after they came off than I did putting people on
7 medications. So I think it is often very difficult
8 to tell and patients are often wrong about whether
9 medications they are on are actually helping them
10 or not.
11 I think it is hard to be prescriptive.
12 Any final comments about any important
13 issues related to neuropathic-pain clinical trials?
14 Any final questions from the FDA side of the table?
15 DR. McCORMICK: I would like to thank the
16 committee for a wonderful discussion today. It has
17 been a great honor to have such distinguished
18 guests here with us sharing your thoughts. Thank
19 you very much.
20 DR. KATZ: Let me thank everybody as well
21 for a wonderful discussion and we will see you next
23 [Whereupon, at 5:30 p.m., the meeting was
25 - - -