1
DEPARTMENT OF HEALTH AND HUMAN SERVICES
FOOD AND DRUG ADMINISTRATION
CENTER FOR DRUG EVALUATION AND RESEARCH
ARTHRITIS ADVISORY COMMITTEE
Monday, July 29, 2002
8:00 a.m.
Holiday Inn Bethesda
Versailles I and II
8120 Wisconsin Avenue
Bethesda, Maryland
2
PARTICIPANTS
Gary S. Firestein, M.D., Chairman
Kathleen Reedy, R.D.H., M.S., Executive Secretary
MEMBERS
Jennifer Anderson, Ph.D.
Kenneth D. Brandt, M.D
Leigh F. Callahan, Ph.D.
John J. Cush, M.D.
Ildy M. Katona, M.D., CAPT, MC, USN
Susan M. Manzi, M.D.
Wendy W. McBrair, R.N., M.S., C.H.E.S.
Yvonne S. Sherrer, M.D.
GUESTS
ARTHRITIS ADVISORY COMMITTEE
Steven B. Abramson, M.D.
Raymond A. Dionne, D.D.S., Ph.D.
Janet D. Elashoff, Ph.D.
Clifford J. Woolf, M.D.
ANESTHETIC AND LIFE SUPPORT ADVISORY COMMITTEE
Michael Ashburn, M.D., M.P.H.
Nathaniel P. Katz, M.D.
Mitchell B. Max, M.D.
NONPRESCRIPTION DRUGS ADVISORY COMMITTEE
Frank F. Davidoff, M.D.
Alastair Wood, M.D.
INDUSTRY REPRESENTATIVE
Charles H. McLeskey, M.D.
GUESTS
David Borenstein, M.D.
John T. Farrar, M.D. MSCE
Vibeke Strand, M.D.
3
C O N T E N T S
PAGE
Call to Order and Introductions:
Gary S. Firestein, M.D. 5
Meeting Statement:
Kathleen Reedy 8
Welcome:
Lee S. Simon, M.D. 10
Introduction:
James Witter, M.D., Ph.D. 13
1992 Guidance:
Christina Fang, M.D. 24
Sharon Hertz, M.D. 34
Basic Science:
Clifford J. Woolf, M.D., Ph.D. 40
Discussion Points #1, 2 62
Claim Structure:
Lee S. Simon, M.D. 95
Discussion Points #3, 4 115
Back Pain, Chronic Issues:
David Borenstein, M.D. 142
Discussion Point #5 168
Open Public Hearing
Najib Babul, Pharm.D. 180
Kenneth M. Verburg, Ph.D. 190
Eugene Laska, Ph.D. 200
Mason Diamond, D.D.S. 203
Abraham Sunshine, M.D. 211
Daniel Carr, M.D. 216
Ann Berger, M.D. 227
Thomas J. Schnitzer, Ph.D. 234
Z. Shainhouse, M.D. [Letter] 245
Michael R. Hufford, Ph.D. 247
Introduction:
James Witter, M.D., Ph.D. 257
Tolerance and Toxicity:
Nathaniel P. Katz, M.D. 263
4
C O N T E N T S(CONTINUED)
PAGE
Statistical Issues for Measurements: 298
Laura Lu, Ph.D.
Open Discussion of Points: #1, 2, 3, 4 and 5 306
5
1 P R O C E E D I N G S
2 Call to Order and Introductions
3 DR. FIRESTEIN: Welcome to everybody to
4 this meeting of the Arthritis Advisory Committee
5 along with a number of esteemed guests.
6 My name is Gary Firestein. I am the chair
7 of the committee. Before we get started with the
8 actual agenda, because there are so many new people
9 here today, it might be valuable to go around and
10 have everybody around the table introduce
11 themselves briefly.
12 As I said, I am Gary Firestein. I am from
13 UC/SD and I am a rheumatologist.
14 Why don't we go around to my left.
15 DR. SHERRER: I am Yvonne Sherrer. I am a
16 rheumatologist. I am from Fort Lauderdale.
17 DR. CUSH: Jack Cush. I am a
18 rheumatologist from Presbyterian Hospital of
19 Dallas.
20 DR. CALLAHAN: Leigh Callahan. I am an
21 epidemiologist from the University of North
22 Carolina in Chapel Hill.
23 DR. WOOD: I am Alastair Wood. I am a
24 clinical pharmacologist from Vanderbilt.
25 DR. DAVIDOFF: I am Frank Davidoff. I am
6
1 an internist and a recovering journal editor.
2 MS. McBRAIR: Wendy McBrair. I am a nurse
3 and health educator from Virtua Health in New
4 Jersey.
5 DR. WOOLF: Clifford Woolf. I am a
6 biologist from Massachusetts General Hospital and
7 Harvard Medical School.
8 DR. DIONNE: Ray Dionne, clinical
9 pharmacologist, National Institute of Dental and
10 Craniofacial Research.
11 DR. MAX: Mitchell Max, neurologist,
12 National Institute of Dental and Craniofacial
13 Research.
14 DR. WITTER: Jim Witter from the FDA.
15 DR. SIMON: I am Lee Simon, Division
16 Director of 550, FDA.
17 DR. McLESKEY: Charley McLeskey, an
18 anesthesiologist, serving as the industry
19 representative here from Abbott Labs.
20 DR. STRAND: Vibeke Strand. I am a
21 rheumatologist, teach at Stanford, and work as a
22 consultant.
23 DR. BORENSTEIN: David Borenstein,
24 rheumatologist, Clinical Professor at George
25 Washington University.
7
1 DR. FARRAR: John Farrar. I am a
2 neurologist interested in pain management at the
3 University of Pennsylvania.
4 DR. ELASHOFF: Janet Elashoff,
5 biostatistics, Cedars-Sinai and UCLA.
6 DR. ASHBURN: Michael Ashburn,
7 anesthesiologist, from the University of Utah.
8 DR. ANDERSON: Jennifer Anderson,
9 statistician, from Boston University Medical
10 Center.
11 DR. KATZ: Nathaniel Katz. I am a
12 neurologist at Harvard Medical School.
13 DR. MANZI: Susan Manzi. I am a
14 rheumatologist from the University of Pittsburgh.
15 DR. ABRAMSON: Steve Abramson,
16 rheumatologist, NYU and Hospital for Joint
17 Diseases.
18 DR. KATONA: Ildy Katona, pediatric
19 rheumatologist, from the Uniformed Services
20 University.
21 DR. BRANDT: Ken Brandt. I am a
22 rheumatologist from Indiana University.
23 MS. REEDY: Kathleen Reedy, Food and Drug
24 Administration.
25 DR. FIRESTEIN: As I mentioned, we do have
8
1 a very full schedule and we have a large number of
2 people in this committee today, so it will be
3 impossible for everybody to take the podium for
4 prolonged presentations, and I would just ask the
5 members of the committee to try to keep comments to
6 the point, so that everybody can have an
7 opportunity.
8 We will begin the meeting with a meeting
9 statement read by Kathleen Reedy.
10 Meeting Statement
11 MS. REEDY: This is the meeting statement
12 for the Arthritis Advisory Committee meeting on
13 July 29th and 30th, 2002.
14 The following announcement addresses the
15 issue of conflict of interest with regard to this
16 meeting and is made a part of the record to
17 preclude even the appearance of such at this
18 meeting.
19 The Food and Drug Administration has
20 approved general matters waivers for the following
21 special government employees which permits them to
22 participate in today's discussions: Gary
23 Firestein, Kenneth Brandt, Ildy Katona, Yvonne
24 Sherrer, Susan Manzi, Jennifer Anderson, John Cush,
25 Alastair Wood, Nathaniel Katz, Michael Ashburn,
9
1 Janet Elashoff, Mitchell Max, Raymond Dionne,
2 Steven Abramson.
3 A copy of the waiver statements may be
4 obtained by submitting a written request to the
5 Agency's Freedom of Information Office, Room 12A-30
6 of the Parklawn Building.
7 In addition, Leigh Callahan, Frank
8 Davidoff, Wendy McBrair do not have any current
9 financial interests in pharmaceutical companies,
10 therefore, they do not require a waiver to
11 participate to today's discussions.
12 We would like to note for the record that
13 Ms. McBrair's employer's interests in two drug
14 companies are exempt under 2640.203(g).
15 The topics of today's meeting are issues
16 of broad applicability unlike issues before a
17 committee in which a particular product is
18 discussed, issues of broad applicability involve
19 man industrial sponsors and academic institutions.
20 The committee participants have been screened for
21 their financial interests as they may apply to the
22 general topics at hand. Because general topics
23 impact so many institutions, it is not prudent to
24 recite all potential conflicts of interest as they
25 apply to each member, consultant, and guest.
10
1 FDA acknowledges that there may be
2 potential conflicts of interest, but because of the
3 general nature of the discussion before the
4 committee, these potential conflicts are mitigated.
5 We would like to note that Dr. Charles
6 McLeskey is participating in today's meeting as a
7 non-voting industry representative. As such, he
8 has not been screened for conflicts of interest.
9 In the event that the discussions involve
10 any other products or firms not already on the
11 agenda for which FDA participants have a financial
12 interest, the participants involvement and their
13 exclusion will be noted for the record.
14 With respect to all other participants, we
15 ask in the interest of fairness that they address
16 any current or previous financial involvement with
17 any firm whose product they may wish to comment
18 upon.
19 DR. FIRESTEIN: Thank you very much.
20 Now we will move on to the Welcome from
21 Dr. Simon.
22 Welcome
23 DR. SIMON: Thank you, Gary, and I would
24 like to welcome the committee. We are grateful
25 that you are willing to come, take time out of your
11
1 practice and busy days to join us here in this
2 rather hot and humid land, but nonetheless, the
3 fact that you have been able to take time out
4 Monday and Tuesday, we are quite grateful about.
5 We recognize that much of what you do here
6 is done involuntarily and we recognize that that is
7 a burden and the government appreciates your
8 commitment.
9 Having been recently on the other side of
10 this microphone and having sat around the table
11 with you as a committee member previously, I can
12 appreciate really what it takes to do this, so
13 thank you.
14 I want to make clear that this meeting is
15 the first of many meetings in an iterative way for
16 us in 550, and hopefully other divisions in the
17 future, to participate with you all in discussing
18 issues of pain, which we find a very critical time
19 in the development of new therapies for pain.
20 We have advanced the science of
21 understanding mechanisms and we believe that part
22 of our role at the FDA is to foster new therapeutic
23 development by discussing all different kinds of
24 ways to look at pain indications and how one would
25 approve such drugs. We believe that these kinds of
12
1 discussions will allow us to further understand
2 better how to create and construct guidances for
3 industry.
4 Much of what we will discuss today will
5 not, and has not, been generally discussed within
6 the entire FDA. I would like to make clear that
7 much of the discussion will inform us in 550, the
8 Analgesics Anti-inflammatory and Ophthalmologic
9 Product Division, about issues that we have been
10 grappling with and have been advising industry
11 about, about the products that we are responsible
12 for.
13 However, much of what we will discuss
14 today will be brought back for further discussions
15 with other divisions, such as 170, Anesthetics and
16 Critical Care, that is particularly interested in
17 this topic since they are responsible for drugs
18 like opioids.
19 So, we feel very strongly that today's
20 discussion, although not directly product oriented,
21 will help us and inform us significantly about
22 where we are going in the future in guidance
23 development.
24 So, again, thank you very much, and I will
25 turn the meeting back to Gary.
13
1 DR. FIRESTEIN: Thank you, Lee.
2 We will move ahead now with an
3 introduction to the topic from Dr. Witter.
4 Introduction
5 James Witter, MD., Ph.D.
6 DR. WITTER: Good morning. Thank you,
7 Gary, Dr. Simon.
8 I am the clinical team leader in 500 and,
9 as such, I would like to thank the members of the
10 team that have spent a lot of time and energy
11 getting ready for today. I particularly would like
12 to acknowledge the help of Barb Gould and I think
13 you will appreciate some of her work here shortly.
14 [Slide.
15 In case you missed it, we are here to talk
16 about pain, and pain is one of those words that
17 even, standing alone, evokes an emotion out of I
18 think everybody. Maybe, in fact, some of you have
19 some of this right now. It is generally not a good
20 emotion, though.
21 [Slide.
22 Pain is really quite fascinating because
23 it is, in one way, the ultimate symptom and
24 therefore, the target for drug development, which
25 is part of the interest today, but it crosses some
14
1 magical line and can become a disease, which we
2 talked about at a meeting just down the street a
3 couple of months ago. So, it kind of goes through
4 what we might think of as I guess a phenotypic
5 change.
6 [Slide.
7 The purpose of the meeting really, then,
8 is what we are going to try and do is simplify
9 things down to concepts and really examine two main
10 aspects of pain and its relief. One of those is
11 how have analgesics been studied and labeled to
12 date, and how should analgesics be studied and
13 labeled in the future.
14 The ultimate goal, then, is to inform
15 analgesic labels in a meaningful way for both
16 patients and clinicians. So, a lot of the focus is
17 going to be a discussion of labels.
18 [Slide.
19 Let's starts off with some definitions
20 then. Can we say, then, that since acute pain is
21 generally considered a self-limiting condition,
22 that that should inform us on how the drug should
23 be studied, labeled, and used? Use is what we are
24 particularly concerned about because we know that
25 off-label use has resulted in serious adverse
15
1 events and death with certain analgesic drugs in
2 the past.
3 [Slide.
4 Can we, on the other hand, say that
5 chronic pain is defined as daily or intermittent
6 pain that occurs either on or off medication and
7 lasts more than 3 months for patients who do not
8 have cancer, but lasts more than 6 weeks for those
9 who have cancer, if what we are trying to do is
10 recruit patients into trials, we don't want to keep
11 them out, particularly those that have cancer.
12 [Slide.
13 So, I am going to ask you to answer a
14 series of questions in your head. Please don't
15 raise your hand unless they apply.
16 But I would like to know: Who has never
17 experienced pain? Who thinks that pain doesn't
18 hurt? Who thinks that pain doesn't interfere with
19 your activities? And who thinks pain doesn't
20 impact your life?
21 I see no hands.
22 [Slide.
23 So, do we then have an agreement,
24 unspoken, that an analgesic should: relieve pain,
25 should improve function, should improve quality of
16
1 life, and do so in a safe manner?
2 And this is important. The other side of
3 the equation from working is safety, and we are
4 going to talk about that today and tomorrow also,
5 but it is always going to be I think in the back of
6 our minds.
7 [Slide.
8 As I mentioned earlier, we are really
9 going to be focusing on labeling, so if you look in
10 the draft OA or in the RA guidance document, you
11 will see something that states the following:
12 "Although label claims have legal and regulatory
13 uses, their central purpose is to inform
14 prescribers and patients about the documented
15 benefits"--and I have inserted in here (and
16 risk)--"of a product."
17 [Slide.
18 Now, this isn't the first time that we
19 have talked about labels and analgesics. We did so
20 about four years ago. We took only one day, and I
21 think by the end of tomorrow, you will realize that
22 that was not sufficient. We broke it up into a
23 morning and afternoon session, and I think I see
24 some people that were here then.
25 The morning session, we really discussed
17
1 the onset of pain relief, what we call fondly,
2 internally, the "fast-faster" wars, and we also
3 studied design of Rx prescription/OTC analgesics.
4 In the afternoon, we devoted it to pain
5 claim structures for both acute and chronic pain.
6 [Slide.
7 We asked some questions, as we usually do,
8 of the Advisory Committee. We asked: Should pain
9 claims be categorized as: for acute versus chronic
10 versus unrestricted or I guess general pain claims?
11 Should they be by categories, for example,
12 neuropathic pain, or should they be by
13 subcategories, for example, diabetic neuropathy?
14 [Slide.
15 We also then asked: Of these studies, how
16 many should there be, how long should they be, what
17 kind of pain "models" should we be using to inform
18 such labels, and what is this concept of
19 "clinically meaningful" benefit and how should it
20 be determined in both the setting of acute and
21 chronic pain?
22 [Slide.
23 But we are here to talk about the future,
24 so what we are going to be discussing throughout
25 these two days are some ideas about how to move
18
1 forward and how to make pain claims for the future,
2 and what we might able to do, for example, is break
3 them up into two basic categories, a clinical and a
4 mechanistic.
5 Clinical is first because, as I mentioned
6 before, pain is the ultimate symptom, so we need to
7 make sure that we address that. Tomorrow, in
8 particular, we are going to be discussing the acute
9 pain setting and, in particular, what we have
10 called the ABC's of doing studies to look at
11 analgesics in the acute situation.
12 Later today, we will be talking more about
13 chronic and what those studies should be, in
14 particular, then labels that should have a specific
15 chronic claim, such as osteoarthritis, which we
16 routine give out in the division, or should we
17 talking about more general claims, replicates of
18 three models, which Dr. Simon will be going into in
19 just a bit, but I think one thing that Dr. Simon is
20 going to stress is that we are trying to set up
21 many ways, particularly for chronic pain, many ways
22 to get approved.
23 Then, I think we are going to be
24 discussing some mechanistic approaches or what we
25 might call some bridging studies, and I will talk
19
1 about that in a bit.
2 [Slide.
3 So, let's just stop for a moment and think
4 about a mechanistic claim. We don't have such a
5 thing, but we wonder what it might look like if we
6 did have one, would it look like, for example,
7 something that would say prevents neuroplasticity,
8 does that make sense to people, or reducing
9 prostaglandin levels, or reducing substance P in
10 CSF, are those the kinds of things that we would
11 mean by a "mechanistic claim."
12 [Slide.
13 So, mechanisms, I have come up with
14 something here called "Mechanisms of Total Pain
15 Relief," and this is a hypothetical model--and
16 please blame me for anything that is wrong
17 here--but let's just say that we can categorize
18 things in terms simply of we will call them Factor
19 X, which are NSAIDs, and like-related compounds,
20 Cox-2's, for example, and let's take a Factor Z,
21 which are opioids and related compounds, tramadol,
22 for example, and then Factors Y, which are future
23 drugs either in development or still in somebody's
24 mind somewhere.
25 Let's say that these then contribute to
20
1 this called chronic pain.
2 [Slide.
3 If we do some mathematics on this, can we
4 say that--let's form some hypotheses here. Can we
5 say Hypothesis 1, for example, that if you take any
6 X or any NSAID, and you add that to any Z or any
7 opioid, you will get 100 percent pain relief, is
8 that the correct hypothesis?
9 Or is it, Hypothesis 2, that we take any
10 combination of X and any combination of Z, we have
11 to add in something else, something else that is
12 missing, the Y factor, to really get 100 percent
13 pain relief?
14 [Slide.
15 Now, once we have answered or tried to
16 answer that, then maybe we then have developed a
17 plan for everybody. Plan 1, for example, going
18 back to Hypothesis 1, would be, well, we really
19 have all we need out there. All we need to do is
20 improve the safety of these existing compounds.
21 Or do we say Hypothesis 2 is true, and
22 sure, of course, we want to optimize use of
23 existing drugs, but what we really need to do is
24 develop and improve new drugs.
25 If that doesn't work, we have an
21
1 alternative plan and we are ready to go here, we
2 have the extra strength pain relief--and thank you,
3 Barb.
4 [Slide.
5 So, I think it is important, the ideas
6 that we discuss today, they sift down, they
7 eventually become drugs. They get into research,
8 both pre- or non-clinical and clinical. If they
9 are lucky, they come to us. If they are lucky
10 again, they get labeled and they get out there for
11 use.
12 [Slide.
13 We are very much a part of this process,
14 and we have become more so thanks to the help of
15 Dr. Meyer Katzburg, who I would like to acknowledge
16 for all his work in setting up what we now have as
17 we are live on the air. The Division has a web page
18 accessible through--go to the CDER web site. You
19 will see there is an announcement of this web page.
20 We are excited about it, it is still growing, and
21 we would love your comments. I can assure you what
22 you send to us, we will all read it, so make it
23 good.
24 [Slide.
25 A couple of months ago I had the pleasure
22
1 and pain experience to work with Dr. Dionne, who is
2 sitting here today, on the NIH-FDA Analgesic Drug
3 Development Workshop.
4 [Slide.
5 We had some objectives for that workshop.
6 We wanted to define pain in terms of the unmet
7 needs for pain management and where to go for unmet
8 needs in terms of pain research, and we discussed
9 how to harness the emerging technologies and
10 improve the development and ultimate FDA approval
11 of new therapies.
12 [Slide.
13 Of course, we had some outcomes and
14 suggestions from this. There was a concern that
15 this separation of pain into acute and chronic may
16 miss addressing the nervous system "plasticity"
17 that many feel goes on.
18 It was acknowledged that there is no
19 consensus for a pain metric, but that one, in fact,
20 needs to be developed to allow for comparisons and
21 poolings of results across the analgesic trials.
22 There was a lot of discussion as to
23 whether new analgesics need to be evaluated as
24 supplementary medications on existing ones because
25 that represents more accurately the pattern of
23
1 clinical use.
2 [Slide.
3 We talked about the need for new therapies
4 to treat pain mechanisms and we talked about how to
5 translate these scientific advances into improved
6 pain relief when it comes down it, it is going to
7 really take a cooperative effort between academics
8 and industry and the regulatory agencies, such as
9 us.
10 Then, we talked about the FDA guidance of
11 1992 and how it needs revision. Let me just talk
12 about that. Dr. Fang will be discussing it in much
13 more detail.
14 [Slide.
15 Let me just mention to you, so that we are
16 on the same page, that the document really
17 discusses analgesic approaches in the 1980's, and
18 if you read it, it assumes that revision would be
19 necessary with time, so I think we all are in
20 agreement that we have arrived.
21 Maybe one of the most distressing features
22 is that it encourages "me too" types of drugs
23 rather than encouraging the "me first" types of
24 drugs that I think we all agree we need in the
25 future.
24
1 So, without further delay, I would like to
2 introduce Dr. Christina Fang from the FDA.
3 I have omitted here, my mistake, I am
4 sorry, Dr. Sharon Hertz, also from FDA, will be
5 discussing the '92 guidance document and some of
6 the positives and negatives from that.
7 We will have Dr. Clifford Woolf from the
8 Mass. General talk to us about the issue of
9 plasticity, our own Lee Simon, who will be
10 discussing the pain claim structure, and Dr.
11 Borenstein will talk to us about what might be one
12 of those new indications in particular lower back
13 pain.
14 Thank you.
15 DR. FIRESTEIN: Thank you very much.
16 As you noted, we are going to move ahead.
17 If the FDA is going to revise the 1992 guidance, it
18 might be useful to first review what they are.
19 So, Dr. Christina Fang and Dr. Sharon
20 Hertz will do that now.
21 1992 Guidelines
22 Christine Fang, M.D.
23 DR. FANG: Good morning. My name is
24 Christina Fang. I am a medical reviewer for the
25 Division of Anti-inflammatory Analgesics and
25
1 Ophthalmic Drug Products.
2 [Slide.
3 I am going to talk about 1992 analgesic
4 guidance document and the current issues.
5 [Slide.
6 The 1992 Guideline for the Clinical
7 Evaluation of Analgesic Drugs had provided the
8 guidance to analgesic drug development and the
9 research in last 10 years. It was originally
10 developed with the focus on NSAIDs and opioid type
11 drugs.
12 With the emerging new molecular entities
13 and with our growing knowledge about analgesics and
14 analgesia, we see the need to resolve many major
15 issues.
16 [Slide.
17 The major areas for improvement in 1992
18 guidance document will be presented at the
19 subsequent slides. Each will be followed with a
20 brief discussion on major issues.
21 [Slide.
22 The 1992 Guidance document recommended the
23 analgesic indications to be for the management of
24 pain. It is stated that evidence of pain. It is
25 stated that evidence of pain of several different
26
1 etiologies will justify general purpose analgesic
2 labeling, also the inclusion of specific labeling
3 indications for preoperative medication, for
4 support of anesthesia, for obstetrical analgesics,
5 or the dysmenorrhea requires specific studies.
6 [Slide.
7 How general and how specific the
8 indications should be has always been in debate.
9 The indication recommended should be based on the
10 number of acute and chronic pain model studies.
11 All the analgesics should be studied
12 sufficiently to include representative
13 subpopulations of major types of pain. The purpose
14 is to provide guidance to practitioners and to
15 minimize unsafe and ineffective off-label use.
16 In terms of specific indications, there
17 are some limitations. For example, we are not able
18 to study all of the indications because of the lack
19 of model sensitivity. If a drug only works for
20 very specific indications, it should be
21 demonstrated that the drug has unique
22 pharmacodynamic activities directed only at the
23 specific indication.
24 [Slide.
25 Acute and chronic indications. This topic
27
1 has always been in debate, as well. We see the
2 need to study the short-term and long-term
3 efficacy, but how much should we have regulatory
4 requirement in terms of models, in terms of
5 replications, we see the same model and the
6 different models, and in terms of length of study.
7 How short-term or the multiple-dose study
8 will help us to study the initial dosing regimen to
9 see if loading dose is necessary and to determine
10 optimal dosing interval.
11 [Slide.
12 In the discussion of chronic studies, the
13 1992 Guidance stated that the focus of the
14 multiple-dose studies of more than 2 to 3 days in
15 duration is to provide documentation of clinical
16 acceptability and the safety of the test drug
17 rather than providing pivotal proof of efficacy.
18 [Slide.
19 Today, we no longer think of studies of 2
20 to 3 days in duration as chronic studies. We need
21 to determine the length for long-term efficacy
22 study. If adequately designed and well controlled,
23 the long-term studies should be able to provide
24 pivotal proof of efficacy.
25 It is especially valuable for drugs with
28
1 delayed onset. The reason we ask for long-term
2 studies is because we see the problem with
3 off-label use for chronic pain. Also, these
4 long-term studies will provide useful information
5 for product labeling, about long-term benefit-risk
6 ratio and the durability effect.
7 [Slide.
8 In terms of pain models, the 1992 Guidance
9 stated that the selection of pain model depends on
10 the strength of analgesia, route of administration,
11 model sensitivities, active controls, and mechanism
12 of action.
13 Also, the initial Phase II studies should
14 explore a wide enough range of pain models.
15 [Slide.
16 We see the need for more acute and chronic
17 pain models because we only have limited models for
18 study of acute pain and most of which were
19 developed for the development of NSAID type drug
20 and also we have limited models for chronic pain,
21 and most of those to be studied were
22 musculoskeletal in origin.
23 We also see the need for models to study
24 the worst type of pain because of the dosing
25 regimen that could be different for this kind of
29
1 setting, and maybe there is a need for concomitant
2 and rescue analgesics.
3 [Slide.
4 In terms of dosing, the 1992 Guidance
5 stated that Phase II studies "should explore the
6 entire dose-response curve of the test drug and
7 should be the basis for selecting the dose used in
8 later Phase II and Phase III studies."
9 Phase III studies are "intended to assess
10 the effectiveness of the recommended dosage
11 schedule under conditions of use."
12 [Slide.
13 We see the need for studying both dose
14 levels and dosing intervals at acute and chronic
15 settings. The dosage obtained from acute setting
16 may not apply to chronic use, and the dosing
17 recommendations should be based on optimal
18 benefit-risk ratio rather than dosing many for
19 convenience.
20 We should also differentiate fixed dosing
21 in clinical trials for establishing efficacy from
22 the variable dosing used in clinical practice.
23 [Slide.
24 In terms of efficacy parameters, the 1992
25 Guidance stated that, "The development program for
30
1 an analgesic should collect data to describe
2 adequately onset of effect, peak effect, and
3 duration of effect. There many ways to collect
4 data on these measures of efficacy."
5 Then, there is a long list of measured and
6 derived parameters in the 1992 Guidance document.
7 [Slide.
8 The choice of efficacy parameters should
9 be based on minimizing bias, demonstrating time
10 course of effect, and providing useful information
11 for dosing recommendations.
12 Pain curves, onset, and the duration
13 should all be studied using valid and reliable
14 tools, and should be studied for both acute and
15 chronic settings.
16 [Slide.
17 For chronic pain evaluations should
18 determine how much the pain-related functional
19 status and the patients global satisfaction should
20 be used for primary or supportive evidence.
21 [Slide.
22 In terms of study controls, the 1992
23 Guidance recommends the placebo and active control
24 for single-dose study, the active control or
25 placebo control with rescue for short-term,
31
1 multiple-dose study, and active control for
2 long-term or multiple-dose study.
3 [Slide.
4 We see the need for adequate controls in
5 both acute and chronic analgesic studies. The
6 placebo controls should always be considered
7 whenever applicable because of the high placebo
8 response in analgesic trials.
9 The superiority design versus equivalence
10 design should be planned accordingly. There are
11 some special considerations for chronic studies in
12 terms of differential dropout rates and in terms of
13 how to keep blinding intact if there are different
14 safety profiles between the drugs to be compared.
15 [Slide.
16 In terms of effect and sample size, the
17 1992 Guidance stated that the calculation of sample
18 size "depends on the variance, the magnitude of
19 difference to be detected, and the desired power."
20 Special consideration should be given to
21 the "validity and the implications of the clinical
22 significance of the differences or similarities to
23 be detected."
24 [Slide.
25 How do we determine clinically meaningful
32
1 effect size has been a debate. There is no
2 consensus on how to define up-to-date. There are
3 did approaches. Whichever approaches are used, a
4 wide database should be applied. The sample size
5 determination is closely related to the
6 determination of clinically meaningful effect size.
7 [Slide.
8 In terms of safety, the 1992 Guidance
9 stated that for peripherally acting or NSAID oral
10 analgesics, the study should regular dosing for a
11 least 6 months. For centrally acting oral
12 analgesics, there should be regular dosing for at
13 least 1 month, continuing for at least 3 months if
14 feasible. For oral combination analgesics, the
15 studies should have regular dosing for at least 1
16 month.
17 [Slide.
18 We see the need to study the safety in
19 terms of the relationship between extent of
20 exposure and adverse events. The extent of
21 exposure includes the level of exposure and the
22 length of exposure.
23 We see the need to study the maximum
24 recommended dosing proposed. The ICH guidelines
25 for chronic pain only provides the minimum
33
1 requirement for minimal number of subjects and the
2 length of exposure.
3 There may be a need to study the
4 representative study population. There may be a
5 need to study the special population with high
6 risks. The large safety trial may be needed if
7 there are serious safety concerns.
8 [Slide.
9 In terms of opioid sparing, we need to
10 determine the clinical relevance of opioid sparing.
11 We need to see the extent of dose sparing that is
12 clinically meaningful.
13 We need to decide if opioid sparing could
14 be discussed in terms of concurrent analgesics or
15 in terms of adjuvant analgesics. For opioid
16 sparing study design to be treated as a concurrent
17 analgesic, there should be consideration of
18 standardization of opioid use and also the data
19 analysis that combines pain data and the rescue
20 medication data, and we need to determine how to
21 evaluate efficacy and safety for this kind of use.
22 [Slide.
23 You can see we have many issues to be
24 resolved. We need a strong need to updating 1992
25 Guidance document. We see the need for proposals
34
1 for future analgesic research. There is also the
2 need for consensus among researchers, drug
3 sponsors, and the regulatory agency.
4 Here, I am just introducing the general
5 concepts and the details will be discussed by my
6 colleagues in the subsequent presentations.
7 Thank you very much.
8 DR. FIRESTEIN: Thank you very much.
9 Now we will go to the second half of this
10 presentation by Sharon Hertz.
11 Sharon Hertz, M.D.
12 DR. HERTZ: Thank you.
13 [Slide.
14 First of all, I would like to thank Dr.
15 Simon and his division for inviting us to
16 participate in this Advisory Committee. I am from
17 the Division of Anesthetics, Critical Care, and
18 Addiction Drug Products. As many of you may know,
19 we also work with a lot of the analgesic products.
20 I am going to present some highlights from
21 our internal discussions on analgesics development,
22 and there will be some overlap with Dr. Fang's
23 presentation. I think what may came out is that
24 there is tremendous overlap in the Division's
25 concerns and in a lot of our approaches to this
35
1 process.
2 [Slide.
3 The 1992 Guidance has been in use for over
4 a decade and we know that subsequent advances in
5 pain research and in pain management really are
6 calling for new approaches to analgesics
7 development.
8 The 1992 Guidance places what we feel is
9 an undue emphasis on models rather than on really
10 looking at particular clinical settings of intended
11 use and target populations, and this has led to
12 some ambiguous labeling and perhaps an inadequate
13 exploration of drugs in the context of the actual
14 clinical use.
15 [Slide.
16 We think that the guidance lacks an
17 adequate emphasis on Phase II dose finding and we
18 have seen many development programs that have come
19 through with very abbreviated Phase II programs.
20 [Slide.
21 There is not an adequate addressing of
22 duration of clinical trials, particularly for drugs
23 intended for chronic administration, and study
24 designs that are recommended in the guidance are no
25 longer considered practical and have been shown to
36
1 lead to somewhat ambiguous results.
2 [Slide.
3 Selection of adequate control groups, as
4 described in the current ICH guidelines, has
5 replaced some of the older thinking represented in
6 the older guidance.
7 [Slide.
8 While the 1992 Guidance makes a
9 distinction between pain due to inflammatory and
10 noninflammatory conditions, it fails to recognize
11 the greater variability in pain etiologies and how
12 this may impact on the response to different
13 analgesics.
14 [Slide.
15 Here are some of the basic development
16 points that we tend to focus on and request when we
17 discuss program development with sponsors.
18 Obviously, for Phase I, we like to see an adequate
19 characterization of the PK profile, but not just
20 for single dose, but also multiple dose studies.
21 We like to see preliminary safety and
22 tolerability over a very broad range of doses
23 potentially anticipating what will be used later
24 on.
25 [Slide.
37
1 During Phase II, we like to see the
2 product explored in potential target populations.
3 Pain conditions identified as responsive in
4 preclinical trials or experience with drugs of a
5 similar class may help define populations to begin
6 exploring during Phase II.
7 [Slide.
8 Analgesics are rarely used only as a
9 single dose agent, so single dose studies shouldn't
10 be proposed for support of marketing applications.
11 Rather, these should be used more to explore early
12 on, analgesic properties.
13 [Slide.
14 We like to see a wide exploration of
15 dosing during Phase II to help inform what would be
16 appropriate arms in Phase III trials.
17 [Slide.
18 Phase II provides a lot of very important
19 opportunity to explore outcome measures and
20 determine what approach is most likely to
21 demonstrate the best way to demonstrate efficacy of
22 this particular product.
23 [Slide.
24 Is there a subgroup that responds well,
25 suggesting a responder analysis is a better primary
38
1 analysis? If so, what are the characteristics of
2 that group? Or do most patients exhibit a moderate
3 but important improvement suggesting an analysis of
4 mean scores as most informative?
5 [Slide.
6 Are there products that are already
7 approved that are better than the studied product,
8 so that even though the study drug beats placebo,
9 it doesn't necessarily lend itself to the target
10 population in that study, that there may, in fact,
11 be another, better indication for the product?
12 [Slide.
13 During Phase III, we ask the sponsor to
14 consider ways to prospectively define a clinically
15 meaningful response for the primary pain variables,
16 preferably using validated measures. As Christina
17 mentioned, this is a very difficult thing to do,
18 because we don't necessarily know yet what
19 clinically meaningful represents.
20 We really prefer the use of validated
21 measures particularly for the primary outcomes.
22 [Slide.
23 For a product likely to be used
24 chronically, we request studies of adequate
25 duration. Typically, we request 12 weeks on final
39
1 titrated dose. This affords an opportunity to
2 assess durability and it is a concept, the 12-week
3 concept is also used for other products in other
4 areas of the Agency.
5 [Slide.
6 Also, for our particular drug groups,
7 particularly the opioids, these 12-week studies can
8 offer an opportunity to provide information
9 concerning tolerance if designed accordingly.
10 [Slide.
11 Efficacy needs to be replicated, not
12 necessarily in an exactly duplicated design, but in
13 a similar population, and these studies are going
14 to provide the basis for informing the label and
15 how the product is to be used.
16 We look forward to getting together with
17 the hosting division to discuss the outcome of this
18 Advisory Committee and to work together on further
19 guidance development and approach to analgesic
20 development.
21 Thank you.
22 DR. FIRESTEIN: Thank you, Dr. Hertz.
23 The next item on the agenda is a
24 discussion of some of the basic science behind pain
25 and analgesia by Dr. Clifford Woolf.
40
1 Basic Science
2 Clifford J. Woolf, M.D., Ph.D.
3 DR WOOLF: Thank you very much for this
4 opportunity to share a basic science perspective on
5 this very important issue.
6 [Slide.
7 What I would like to try and discuss today
8 is how the advances that have occurred in the last
9 10 years, since the 1992 Guidelines, some of the
10 advances that have been made and the implications
11 for them in looking at analgesia and analgesics,
12 and this issue of labeling.
13 Some of the particular issues I would like
14 to address is whether there is a basis for the
15 differentiation of pain in terms of its chronicity,
16 intensity, and how our understanding of the
17 mechanisms that are responsible for pain can drive
18 and may actually be included in any discussion
19 about indication.
20 [Slide.
21 To begin with, to look at pain chronicity,
22 I think it is important, when we look at the
23 difference between acute and chronic pain, to try
24 and identify whether chronic pain may be the
25 results of the persistence of a mechanism or may be
41
1 the result of the recruitment of a novel mechanism
2 that is not present in those patients that have
3 acute pain, because these clearly are quite
4 different.
5 [Slide.
6 So, doing a kind of an analysis of those,
7 we can readily appreciate that acute pain
8 characteristically is transient, it may be
9 recurrent, but it is always reversible. That is a
10 key element implicit in our definition of acute
11 pain, whereas, chronic pain, I think we can
12 usefully divide into two very broad categories.
13 There are those patients who have
14 long-lasting pain which is reversible, so that if
15 the driving mechanism responsible for that pain is
16 removed, that pain will tend to disappear, whereas,
17 there are other patients where the pain is truly
18 persistent and we can even say irreversible.
19 I think these are very distinct
20 subcategories and we need to recognize and solve
21 that.
22 [Slide.
23 In terms of looking at pain intensity,
24 again, the issue is whether there is a continuum of
25 pain mechanisms that can generate pain of different
42
1 intensity divided between mild, moderate, and
2 severe, or whether each of these levels of
3 intensity of pain reflect discrete mechanisms that
4 operate, that are recruited at different levels of
5 disease or as new etiological factors come into
6 play.
7 Another important aspect we need to take
8 into account is when we look at the intensity of
9 pain that is experienced by an individual, whether
10 that reflects an increase in some stimulus, some
11 external driving force, some disease factor, or,
12 indeed, may be an alteration in the responsiveness
13 of the nervous system.
14 Certainly, there is now increasing belief
15 amongst basic scientists that the responsiveness of
16 the nervous system can alter quite profoundly, and
17 an increase in intensity may not necessarily
18 reflect an increase in stimulus.
19 [Slide.
20 The simple underlying approach to pain
21 until quite recently was that multiple etiological
22 factors operating by means of inflammation, tissue
23 damage, nerve lesions, or a number of other ways,
24 could act on a highly specialized sensory apparatus
25 in the nervous system to drive the symptoms and
43
1 signs that we now collectively call pain, and that
2 there was, if you like, this convergence of
3 etiological factors acting on the nervous system to
4 initiate a set of changes which generated the
5 response that we interpret as pain and that we
6 could then subdivide the pain depending on the
7 etiological factors, the duration, the associated
8 changes into different pain syndromes.
9 What I would like to argue today is that
10 we need to move away from this very simple model,
11 and I would like to show you why it is neither
12 correct nor helpful in defining the approach the
13 analgesics.
14 [Slide.
15 One of the main reasons is that it has
16 become increasingly clear that we are dealing with
17 multiple distinct pain mechanisms. This is an
18 incomplete list. Almost certainly this list is
19 going to change as our understanding of pain
20 improves, but it is clear that there is a distinct
21 mechanism that is responsible for nociception by
22 which I mean the sensory mechanism that is
23 responsible for pain in response to a transient
24 non-damaging, noxious stimulus.
25 There are distinct mechanisms that operate
44
1 to alter the sensitivity of the high-threshold
2 nociceptive primary afferents that are responsible
3 for nociception, and these changes at the
4 peripheral terminals of these nociceptors are what
5 we call peripheral sensitization and are a major
6 driver of inflammatory pain.
7 In addition, it is increasingly apparent
8 that changes in the processing of sensory
9 information within the central nervous system, that
10 collectively we can call central sensitization,
11 play a major role in the shaping of the pain
12 experience and may in some individuals and in some
13 situations be a major factor responsible for the
14 pain.
15 After nerve damage, we now appreciate
16 there is the development of ectopic excitability,
17 sensory inflow with a sensory stimulus. There are
18 also increasing indications that lack of inhibition
19 and structural alterations in the nervous system
20 may play a major role particularly in chronic pain
21 associated with nerve damage.
22 Today, I am going to stick my discussion
23 to the first three mechanisms and try and
24 illustrate how understanding of them has
25 implications for determining the efficacy of
45
1 different groups of analgesics.
2 [Slide.
3 In addition to multiple pain mechanisms,
4 we need to recognize that pain is not a monolithic
5 single entity. There are different pain symptoms
6 that may complicate a way to reflect these
7 different mechanisms, and that if we use global
8 pain scores, we may be missing some of the
9 different mechanisms that operate in different
10 conditions, so it is important for us to appreciate
11 that there is spontaneous pain, pain that
12 apparently arises without any peripherals or
13 without any stimulus, and evoked pain, pain that
14 occurs in response to some input.
15 Spontaneous pain itself may be divided
16 between that that appears to derive from the skin,
17 from the superficial structures of the body, and
18 that which is deep. Indeed, there are differences
19 between the pain that is continuous and that which
20 is intermittent, and clinically, we certainly
21 recognize that these are not the same.
22 Evoked pain, again there is enormous
23 difference between pain that is evoked by thermal
24 and mechanical stimuli, and it is important to
25 differentiate pain that occurs in response to a
46
1 stimulus that normally would not be painful, what
2 we call allodynia, and an exaggeration of the
3 response to a noxious stimulus, that which we call
4 hyperalgesia.
5 What I would like to argue is that each of
6 these different categories reflects different
7 activities in the nervous system and it is
8 essential in performing clinical trials to try and
9 capture as much of this information because it
10 reflects some of the processing that generates the
11 pain experience.
12 [Slide.
13 To illustrate the points that I have made,
14 I am going to look at the COX-2 selective or
15 specific inhibitors and try and identify from our
16 increased knowledge of the mechanisms that operate
17 to produce pain, how there may be elements of pain
18 that are sensitive to these classes of drugs and
19 others that are not, and for that reason, why the
20 discussion of whether it is appropriate to discuss
21 global analgesics or even analgesics that are
22 appropriate for all acute pain or all chronic pain
23 needs to take into consideration some of these
24 factors.
25 [Slide.
47
1 So, to begin with, to come back to
2 nociception, as I said before, this is the term
3 that we use to describe the capacity of the nervous
4 system to respond to particular intense stimuli,
5 noxious stimuli, those stimuli which have the
6 capacity to damage the body.
7 These stimuli are detected by highly
8 specialized primary sensory neurons, the nociceptor
9 neurons, which respond only to intense, and not to
10 innocuous stimuli, and they feed into particular
11 neurons within the central nervous system that
12 transfers this information to that part of the
13 cortex that eventually results in the sensation or
14 the perception of pain.
15 This, if you like, is the "ouch" pain, the
16 pain we feel in response to a pinprick or touching
17 something that is too hot or too cold, and clearly,
18 it has a major role as a protective mechanism, an
19 early warning device, and that is something we need
20 to appreciate because abolition of no nociception,
21 while appropriate in some conditions, such as
22 during surgical intervention, is not appropriate in
23 the chronic setting.
24 [Slide.
25 How does nociception generate? Well, if
48
1 we think back to 1992, we had almost no information
2 on how noxious stimuli act on the nervous system to
3 generate nociception, and in the last 10 years, the
4 progress has been extraordinary. Only in the last
5 few months has the receptor, the CRM1 receptor been
6 cloned that converts cold stimuli into cold pain.
7 Heat pain is detected by a number of
8 different receptors. About five years ago, the
9 vanilloid receptor 1 was identified as being a heat
10 transfuser, and only in the last month has another
11 member of the vanilloid family, the TRPV3, the TRP
12 channel V3 been identified.
13 So, we now know the individual ion channel
14 receptors that respond to these noxious stimuli and
15 produce generated potentials. There are also
16 receptors that respond to chemicals released at the
17 time of tissue damage, such as bradykinin, the B1
18 and B2 receptors, and we are at the point of
19 understanding how intense mechanical stimuli are
20 transfused into electrical activity.
21 Now, the point of going through all of
22 these is that you will see there are no
23 prostaglandin receptors, there is no COX-2 here, so
24 that the process by means of which an intense
25 thermal chemical or mechanical stimulus produces
49
1 nociception is COX-2 insensitive. No amount of
2 COX-2 inhibitors given at anytime will affect the
3 way we respond to pinprick or heat stimulus, so
4 that COX-2 is not appropriate for that indication.
5 [Slide.
6 If we look at the transfer of information
7 from the primary sensory neuron to central
8 neurons--and this is an attempt to cartoon the
9 central terminal of nociceptors and their synaptic
10 interaction with neurons in the spinal cord--we
11 have identified the key transmitters that act to
12 transfer this information.
13 There are both excitatory amino acids,
14 such as glutamate and neuropeptides, such as
15 substance P, and they act on a number of receptors
16 on the postsynaptic neuron, both inotropic
17 receptors and metabotropic receptors, and these can
18 be modulated in different ways by a number of
19 receptors which play a role in inhibitory
20 mechanisms.
21 The GABAergic, particularly the GABA-A
22 receptors, which control presynaptic release of
23 transmitters and a number of other receptors,
24 particularly the opiate receptors, which are
25 expressed both pre- and post-synaptically, and can
50
1 reduce synaptic transmission.
2 So, opiate receptors and opioids, opiate
3 receptor activation and opioids can certainly
4 modify this transmission process and can reduce
5 nociception, but again, you will see that there is
6 no COX-2 or prostaglandins involved in this, and
7 once again, nociception, both peripherally and
8 centrally, is not COX-2 sensitive.
9 [Slide.
10 That is essentially the conclusion made
11 here.
12 [Slide.
13 If we talk about COX-2 as being an
14 analgesic, we need to take onboard that it is not a
15 global analgesic, it does not reduce all pain in
16 all circumstances, and it certainly will not reduce
17 nociception, which is actually a desirable
18 consequence of all chronic usage as I have
19 indicated.
20 [Slide.
21 We now move on to peripheral
22 sensitization. This is the setting now where we
23 have inflammation in the periphery. The peripheral
24 terminal of nociceptors are exposed to inflammatory
25 mediators, and this changes the peripheral terminal
51
1 in the way that this terminal can now be activated
2 by stimuli that have a lower intensity, so that
3 both stimuli that would normally not produce pain,
4 and noxious stimuli produce a greater response, and
5 this creates the situation where we have what is
6 called primary hyperalgesia, which is abnormal pain
7 sensitivity in the site of tissue damage, and one
8 of the particular roles that peripheral
9 sensitization has been shown to operate in is
10 primary heat allodynia, the reduction in the heat
11 threshold for producing pain.
12 Normally, we require stimulus of about 42
13 degrees for the conversion of a hot to a painful
14 stimulus, but in the presence of inflammation, this
15 can fall quite substantially.
16 What are the mechanisms involved in
17 generating peripheral sensitization? Well, they
18 are multiple, but the one that I want to highlight
19 today is that as a result of the inflammatory
20 response and the release of cytokines, particularly
21 IL-1 beta and TNF-alpha, there is the induction of
22 changes in cells surrounding the inflamed area of a
23 number of enzymes and growth factors and
24 chemokines, but the one here that I want to
25 emphasize is COX-2, but if COX-2 and phospholipase
52
1 are induced at the site of peripheral inflammation,
2 that results after action by specific tissue
3 isomerases and the production of prostanoids, such
4 as prostaglandin E2, which can then act on EP
5 receptors, prostaglandin receptors that are
6 expressed on the peripheral terminal of the primary
7 nociceptor.
8 Prostaglandin, when it acts on the
9 peripheral terminal, does not directly produce an
10 activation of the peripheral terminal, it does not
11 itself produce pain. What it does do is alter the
12 excitability of the peripheral terminal, and we now
13 know how that occurs. It is via activation of
14 kinases that are present in the peripheral terminal
15 that phosphorylate either transducive proteins,
16 such as the vanilloid VR1 heat transducer, reducing
17 its threshold of activation or it phosphorylates
18 ion channels that are present in the peripheral
19 terminal making the peripheral terminal
20 hyperexcitable, so that less of a stimulus or less
21 transducer action is required to activate the
22 peripheral terminal.
23 I indicate there is a northern blot on the
24 side showing that in normal skin, there is
25 undetectable COX-2 levels, but within several hours
53
1 of peripheral inflammation, there is an enormous
2 induction of this enzyme, and the point being that
3 this particular pain is COX-2 sensitive. You
4 cannot have COX-2 action if there is no target
5 COX-2 expressed, but after peripheral information,
6 it begins to be expressed, so this particular
7 mechanism is COX-2 sensitive.
8 There are, in addition to prostanoids,
9 other mechanisms that can drive peripheral
10 sensitization, which means that COX-2 inhibitors
11 may not completely eliminate this process.
12 Bradykinin, amines may also produce these changes,
13 this activation of kinases, which can phosphorylate
14 some of these proteins.
15 Conceivably, drugs may be developed that
16 can block these kinases and even their targets,
17 such as the vanilloid receptor or the ion channels,
18 and may actually totally abolish the changes that
19 are produced by peripheral inflammation.
20 [Slide.
21 I now want to move on to changes that can
22 occur within the central nervous system, changes in
23 the excitability of neurons which alter its
24 responsiveness, and the situation here is that we
25 now recognize that noxious stimuli produced by
54
1 irritants, tissue damage, inflammation, anything
2 that can activate nociceptors can result in a use
3 or activity-dependent plasticity within the central
4 nervous system, altering the excitability of these
5 central neurons, and this results in a situation
6 whereby these neurons respond to normal inputs in
7 an exaggerated or abnormal way.
8 This generates two broad changes that we
9 can recognize in pain. One is secondary
10 hyperalgesia, which is a change in sensitivity to
11 pain outside of an area of tissue damage or
12 inflammation.
13 Peripheral sensitization contributes to
14 the pain sensitivity at the site of tissue damage,
15 but central sensitization, this abnormal
16 responsiveness of central neurons, contributes to
17 the change in sensitivity that spreads into normal
18 non-damaged or non-inflamed tissue outside the area
19 of tissue damage.
20 One particular mechanism that we now
21 recognize as being driven by central sensitization
22 is tactile or brush-evoked allodynia. This is the
23 pain that can occur by the activation of normal
24 low-threshold mechanoreceptors that would be
25 activated by lightly touching or brushing the skin.
55
1 After the induction of central sensitization, such
2 stimuli can begin to produce pain, and this is a
3 reflection of this mechanism.
4 [Slide.
5 The reason why central sensitization
6 produces changes in pain is it turns out that the
7 pain projection neurons within the nervous system
8 do not exclusively receive input from nociceptors,
9 the high-threshold sensory fibers.
10 They receive, in addition, an input with
11 weak synaptic input from low-threshold
12 mechanoreceptors. This synaptic is normally too
13 weak to drive the cells, so that activity generated
14 by light touch, movement of a joint will not
15 normally generate an output in the pain projection
16 neurons, but if the excitability of the central
17 neurons is increased, then, this normal input in
18 normal, low-threshold mechanoreceptors can begin to
19 drive these abnormally excitable central pain
20 projection neurons and result in the recruitment of
21 pain in response to this normal input.
22 This is the mechanism for brush-evoked
23 mechanical allodynia.
24 [Slide.
25 What actually produces the increase in
56
1 excitability of the central neurons and the
2 specific details are not important for the purposes
3 of this discussion, but just to say that it turns
4 out there are two phases to the production of
5 central sensitization.
6 There is an acute phase that occurs within
7 seconds of the activity of nociceptors. If you
8 activate nociceptors intensely, and this can be
9 done by an irritant stimulus or heating the skin or
10 tissue damage, that will result in the release of
11 glutamate and beyond it, if there is enough
12 glutamate released as a result of repetitive
13 activity in nociceptors, that will induce
14 activation of intracellular kinases, cyclic
15 AMP-dependent protein kinase A, and
16 calcium-sensitive protein kinase C, which will
17 phosphorylate the receptors and ion channels on the
18 postsynaptic membrane, altering their
19 responsiveness.
20 So, there is an activity-dependent change
21 in the excitability of the postsynaptic membrane
22 due to the synaptic release. Again, you can see
23 that while there are multiple players invoked in
24 here, COX-2 is not a feature. So, this component of
25 central sensitization, the acute component that is
57
1 switched on almost immediately by intense
2 nociceptor activity is not COX-2 sensitive.
3 [Slide.
4 However, it turns out that peripheral
5 inflammation, in addition to inducing COX-2 in the
6 site of tissue damage, as I have indicated, also
7 induces COX-2 within the central nervous system, in
8 the spinal cord, and this occurs after several
9 hours.
10 The question is does this have any role in
11 central sensitization.
12 [Slide.
13 Well, there are two things to first
14 recognize, is that the central induction of COX-2
15 occurs only in response to peripheral inflammation,
16 and not in response to peripheral nerve damage, so
17 again, we need to differentiate when we are looking
18 at this mechanism the way it operates after tissue
19 damage and inflammation is quite distinct from what
20 happens after peripheral nerve injury.
21 It turns out that the late phase of
22 central sensitization, that phase that occurs hours
23 and days after tissue damage does involve COX-2,
24 because COX-2 begins to be induced in neurons
25 within the central nervous system, produces
58
1 prostaglandins which have multiple actions,
2 increasing transmitter release, increasing the
3 excitability of postsynaptic receptors, as well as
4 blocking some inhibitory actions.
5 The net result is that the increase in
6 excitability of central neurons acutely is not
7 COX-2 sensitive, but that which occurs some hours
8 after tissue damage begins to have a component that
9 is COX-2 sensitive.
10 [Slide.
11 So, the conclusions I would like to make
12 from this is that there are COX-2 sensitive
13 peripheral and central components of inflammatory
14 pain, but not necessarily of the pain associated
15 with peripheral nerve injury, that COX-2
16 inhibitors, as an example, can only act when their
17 target is expressed. It needs to be induced. This
18 takes a finite amount of time.
19 The cytokines IL-1 needs to produce, it
20 needs to act on cells, which then switch on
21 transcription factors, such NF kappa B, which then
22 switch on the COX-2 gene, the messenger RNA has to
23 be made, translated into protein, and this needs to
24 be transported to the appropriate place in the
25 cell.
59
1 This takes several hours, so that after
2 peripheral inflammation, you only get a COX-2
3 sensitive component when the COX-2 is expressed and
4 there.
5 There are also non-prostanoid contributors
6 to inflammatory pain, and this may explain why
7 COX-2 selective or sensitive inhibitors cannot
8 produce a complete relief of pain. Other
9 mechanisms continue to operate. So, that may
10 contribute to the ceiling effect of these class of
11 drugs.
12 I have already mentioned that peripheral
13 nerve injury may not be present.
14 [Slide.
15 So, I think we need to consider then what
16 are the models that are appropriate for looking at
17 the relationship between etiology and the symptom
18 that we call pain.
19 Well, one possibility may be that
20 different etiologies may act on the nervous system
21 to produce different distinct mechanisms that may
22 produce particular symptoms. If you need to treat
23 the particular kind of pain associated with a
24 particular etiology, you can target the individual
25 mechanism.
60
1 Unfortunately, the reality as far as we
2 can judge is more like this, that a single
3 etiological factor can operate on the nervous
4 system to operate multiple mechanisms. Peripheral
5 sensitization and central sensitization are not
6 independent, both can be switched on by peripheral
7 inflammation.
8 Peripheral nerve injury can produce both
9 ectopic excitability and central sensitization, and
10 part of the challenge that we have is to try and
11 identify the links between different etiological
12 factors and the mechanisms they operate, as well as
13 how the different mechanisms can change, produce
14 the symptoms that the patient complains of.
15 [Slide.
16 What I would like to try and suggest is
17 that we need to differentiate between analgesic
18 drugs, drugs where the implication is a global
19 relief of pain, and drugs where there is a
20 reduction of the abnormal sensitivity of the
21 nervous system, and that this is a useful
22 distinction.
23 I hope I have indicated to you that both
24 the temporal and intensity characteristics of pain
25 do not, by themselves, reflect mechanisms, that
61
1 they are different mechanisms that can operate to
2 produce both acute and chronic pain, and that for
3 this reason they may not, by themselves, be useful
4 predictors of analgesic action.
5 I would like to argue that as we begin to
6 understand more about pain mechanisms and the very
7 particular mechanisms that individual drugs have,
8 that it is this combination that is going to
9 provide the most useful input for determining
10 indication and efficacy.
11 [Slide.
12 In order to make progress, we need to move
13 away from using exclusively global pain scores as
14 our outcome measures. We need outcome measures
15 that are sensitive or specific to particular
16 mechanisms, and that is a big challenge.
17 We need clinical trials that can validate
18 mechanistic hypotheses and that are designed
19 specifically to address the issue of which drugs
20 acting on which mechanisms can alter the symptoms
21 in particular groups of patients.
22 We need to consider labeling claims and
23 the like to the action of drugs, with the
24 interaction of the drugs with specific pain
25 mechanisms, as well as the more traditional
62
1 approach, which has been empirical trials looking
2 for efficacy.
3 My final conclusion is that I think--and
4 this overlaps to some extent with the comments made
5 by Jim Witter--are there going to be global
6 analgesics. I think this is unlikely.
7 Pain has too many different mechanisms
8 operating that it is very unlikely that a single
9 drug is going to affect all of them and that the
10 challenge we have is to try and optimize the way to
11 detect which particular mechanisms an individual
12 drug is operating to see the utility of blocking
13 that mechanism for particular groups of trials and
14 let that drive the labeling of the drugs.
15 Thank you.
16 DR. FIRESTEIN: Thank you very much for an
17 excellent discussion.
18 Discussion Points #1 and 2
19 DR. FIRESTEIN: At this point, we can move
20 into some of the discussion issues that were raised
21 by Dr. Simon and the Agency. I believe that we
22 were going to discuss Points No. 1 and 2. I will
23 just read the first one and then open it to the
24 group for comment.
25 1. A revised analgesic guidance may
63
1 include indications intended to inform labels for
2 the management of acute versus chronic pain, rather
3 than a general pain claim. Please comment on the
4 clinical relevance of this distinction in terms of
5 efficacy and safety.
6 if there is anybody who would like to get
7 the ball rolling here? I suppose that then becomes
8 the Chair's prerogative to comment and then have
9 everybody disagree with me.
10 I think that the discussion that we have
11 already had, defining the distinct mechanisms of
12 pain, raised some of the issues about separate
13 labels for acute versus chronic pain as opposed to
14 a general pain claim versus a specific claim that
15 is mechanism based.
16 I think in particular, one of the things
17 that was discussed earlier was the question of
18 whether chronic pain in some cases merely
19 represents persistence of acute pain mechanisms,
20 and how can one distinguish that in a chronic pain
21 labeling is going to be quite difficult.
22 I don't know, Dr. Woolf, you might want to
23 comment on that particular aspect.
24 DR. WOOLF: The point I was trying to make
25 using the COX-2 inhibitors would be, to get down to
64
1 specifics, that although they may have an
2 indication for chronic pain based on a number of
3 replicate trials showing efficacy in chronic pain,
4 the evidence indicating that there is no COX-2
5 induction of peripheral nerve injury, which may
6 certainly produce chronic pain, would indicate that
7 most patients with neuropathic pain may not respond
8 to COX-2 inhibitors, so that an indication of
9 chronic pain by itself is incomplete and may lead
10 to inappropriate use of analgesics, which may not
11 have efficacy in certain particular groups of
12 patients.
13 So, the issue then is does chronic pain,
14 by itself, have a meaning. I think we have just
15 got to be a little cautious of that.
16 DR. FIRESTEIN: I guess on the other hand,
17 it might at least bring us a little closer to
18 reality as opposed to a more global pain
19 indication, in other words, although there are
20 clearly limitations between acute versus chronic
21 pain, that is less problematic than trying to have
22 a global pain indication that would cover all
23 aspects of all pain indications.
24 DR. MAX: Gary, you have already in your
25 question, you already indicated that this
65
1 distinction is mechanistically insufficient,
2 because you said chronic back pain can have acute
3 inflammation on top of it. I think it is clear
4 from Clifford's talk that this does not do very
5 much for us with mechanisms.
6 However, just from a practical clinical
7 setting point of view, I think it is clear that
8 when we talk about acute pain, we are talking about
9 a specific clinical orientation of the patient.
10 They have sudden bad pain and they are willing to
11 do anything they can for a few days to handle it,
12 and a little bit of impairment of work might be
13 okay.
14 On the other hand, in chronic pain, we
15 really need evidence from day-in, day-out living,
16 not just the single dose trial, that the patient
17 has got to be able to live with the analgesic
18 regimen and the way of evaluating it is going to be
19 much different.
20 So, I think the main argument for this
21 division being important is the practical
22 considerations, the clinical setting, are so much
23 different that they really imply completely
24 different clinical trial designs.
25 I mean once we take each, then, we can
66
1 bring in some of the mechanistic considerations
2 that will be hard.
3 DR. FIRESTEIN: Dr. Brandt, did you have a
4 comment?
5 DR. BRANDT: Yes. I think, Dr. Woolf,
6 that was really a beautifully lucid and useful
7 dissection of mechanisms. To bring it to
8 osteoarthritis pain, I would like to ask whether it
9 suggests a research approach.
10 Nonsteroidals for patients with
11 osteoarthritis improve pain on average, on visual
12 analog scales, 20, 25 percent. Some patients get
13 terrific relief, some patients get worse, but on
14 average, 20, 25 percent.
15 If you add acetaminophen to a
16 nonsteroidal, you get a further increment, but
17 there still is a significant amount of residual
18 pain. Based on what you said, presumably, there is
19 another mechanism that is driving it, how does one
20 get at that, how does one study that to know what
21 sort of drug might be useful or might be reasonably
22 tested to get at that residual pain.
23 DR. WOOLF: Chronic osteoarthritis is a
24 very interesting disease from a basic science point
25 of view. The problem we have is that there are
67
1 very poor preclinical models that it is very
2 difficult to test in the preclinical setting what
3 the mechanisms are.
4 The fact that there is a response, even
5 though modest, to standard NSAIDs when in most
6 patients there is not ongoing inflammation, raises
7 the issue of where is the COX-2 that presumably
8 they are acting on, so I think the first research
9 question is, is this a disease of the periphery in
10 terms of COX-2 mechanisms or is the COX-2 induced
11 in the central nervous system.
12 The fact that there is an additional
13 contribution of acetaminophen would imply that that
14 is likely to be the case.
15 The ceiling effect of NSAIDs is as you
16 indicate, and the fact now with the
17 second-generation COX-2's, where the doses can be
18 pushed to a level where all conceivable COX-2 is
19 likely to be inhibited certainly indicates that
20 there is a residual mechanism that is not COX-2
21 sensitive.
22 What it is, is obviously the big
23 challenge, and I could speculate, but I think this
24 is where new drugs with new targets are coming onto
25 the market. Some of them may be useful by
68
1 themselves, but I think in clinical practice, we
2 know already that polypharmacy is a standard way in
3 which patients are treated.
4 So, it is very likely that these new
5 drugs, acting on different independent targets,
6 will have a role, sometimes by themselves, but
7 often in combination with existing therapy.
8 DR. FARRAR: Understanding that even in
9 the realm of arthritis, it is very often difficult
10 to identify in any given patient the primary cause
11 for their discomfort, I wanted to ask Dr. Woolf
12 whether, if we were able to identify a subset of
13 arthritic patients who had, in fact, a very similar
14 peripheral mechanism, whether that nice
15 pathophysiology slide you showed with all the
16 various mechanisms, whether all of those mechanisms
17 would apply in every patient or whether, in fact,
18 there would be within even a mechanistic approach,
19 differences in the way that a particular patient
20 responds to both the pain and the underlying
21 treatment based on the fact that some may have a
22 predominance of one kind of receptor over another
23 or a predominance of one response over another.
24 DR. WOOLF: I think it is even more
25 complicated than that. I think it is not only the
69
1 problem that individual patients within a
2 particular group or clinical entity, a particular
3 form of arthritis may have different mechanisms,
4 but an individual patient over the evolution of
5 their disease will almost certainly have different
6 levels of contributions of the different
7 mechanisms.
8 The challenge is how to identify them, and
9 the fact, the comment that was made that some
10 patients may respond extremely well to NSAIDs than
11 others, I think that gives part of the clue. I
12 think one of the tools that we are going to have to
13 use are drugs to try and identify mechanisms.
14 Those patients who respond very well to
15 COX-2 inhibitors, by definition, we are defining at
16 least one component of their pain is COX-2
17 sensitive, whereas, those patients that don't,
18 assuming the drug, the notions of bioavailability
19 or PK, we can conclude that in those individual
20 patients, there is not a COX-2 component.
21 So, I think we are going to have to use a
22 combination of trying to link up symptoms with
23 mechanisms, which is difficult, but not impossible,
24 as well as the responsiveness of the patient to
25 very specific forms of therapy.
70
1 DR. SHERRER: A question as it relates to
2 chronic pain, because it was mentioned earlier by
3 Dr. Witter, and it is certainly true clinically,
4 that there are two types of chronic pain. There is
5 the chronic persistent pain, and there is the
6 chronic acute intermittent pain or intermittent
7 pain at least.
8 Do those patients represent people with
9 repetitive acute pain mechanisms even though it is
10 one disease, such as the osteoarthritis patient who
11 flares every few weeks or with a weather change or
12 with activity, or, in fact, is that a different
13 mechanism of chronic pain?
14 DR. WOOLF: I gave an example just to try
15 and differentiate in the most global sense, but
16 there will again be patients, such as those with
17 trigeminal neuralgia, who will also have
18 intermittent pain where the mechanism will be
19 completely different from an OA patient with flare,
20 so I hope I didn't give the impression that that
21 represents two distinct mechanisms.
22 There may be again multiple mechanisms
23 that operate between those two classes, but I think
24 we are all aware of patients who have OA of the
25 hip, when the hip is replaced, can do extremely
71
1 well with minimal recurrence of pain, where there
2 are patients with peripheral neuropathic pain where
3 the neuroma is removed, and they have a transient
4 response and the pain comes back, so the point
5 being that in some cases, removing the etiology,
6 the cause, the hip, can actually remove the pain,
7 whereas, in other patients, it appears as if the
8 mechanisms have now been hard wired, if you like,
9 and are resistant to, are no longer driven by the
10 initial disease process.
11 DR. FIRESTEIN: Let's come back to one of
12 the issues raised here, and that is whether or not
13 there is utility to differentiating between acute
14 versus chronic pain as compared with a general pain
15 claim and, in particular, issues that relate not
16 only to efficacy but safety.
17 One example of that would be for the
18 selective COX-2 inhibitors where one dose might be
19 approved for the treatment of acute pain and has
20 had either a dosage creep that has then at least in
21 the clinic led to use of some of these higher doses
22 for chronic treatment, and some of the safety
23 issues may not have been addressed in the clinical
24 trials because of that.
25 Does anybody have a comment? Yes.
72
1 DR. ELASHOFF: What I wanted to ask is in
2 the first day or so of pain, if you are labeling
3 things for acute or for chronic, does one know in
4 the beginning whether you ought to be using the
5 ones labeled for acute, because you don't know
6 whether it might turn out to be chronic or not, or
7 might you have the knowledge to say you ought to be
8 starting in with chronic, so would one always start
9 with acute things and then switch, or does one
10 potentially have the knowledge at the beginning
11 that you might start out with chronic things.
12 So, it seems to me that the issue of the
13 labeling has to also say, well, practically
14 speaking, how would you know in any given situation
15 which ones you are going to be using.
16 DR. FARRAR: I think we need to very
17 carefully differentiate between how we use the
18 medicine and what we are treating. The question
19 you are asking really relates to whether the
20 medicine is used over a long period of time or
21 whether it is used over a short period of time.
22 I think the issue is not answerable from
23 an acute or chronic perspective. If you take
24 migraine headache, there are medicines that are
25 used to prevent it, that are used regularly over a
73
1 long period of time, and then there are medicines
2 that are used to treat it, which may be used over a
3 very short period of time.
4 I think we need to differentiate between
5 whether it is used over a long or short period,
6 which can be done in a label, to say this drug can
7 only be used for, it has only been shown to be safe
8 for six weeks versus saying whether you are
9 treating acute or chronic pain. I think those two
10 are very different.
11 DR. CUSH: But aren't you just saying the
12 same thing? I mean it is acute, a short period,
13 and chronic if it's long term. We know that based
14 on what the etiology of the pain is, the problem,
15 whether it's postsurgical or dysmenorrhea or
16 migraine, what our goals are as far as short term
17 or long term.
18 But the terms of acute therapy and chronic
19 therapy are useful. They dictate how we use these
20 drugs. They dictate our expectations for these
21 drugs. To go with a more general pain claim is too
22 vague and not applicable to many patients that we
23 use.
24 DR. FARRAR: But don't confuse acute
25 treatment and chronic treatment with acute pain and
74
1 chronic pain. As was said here, you don't know
2 when you start necessarily whether it is going to
3 be a 2-day treatment or a 10-day treatment.
4 DR. CUSH: I think most physicians do know
5 when they start out with managing pain what the
6 goals are for pain management. Now, it is not to
7 say that patients who start out with a migraine
8 don't have a migraine that might be extending out
9 beyond a few days, and acute therapies may not
10 work, but I think that there are goals when you
11 make a diagnosis and see a patient as far as
12 whether it is going to be short-term therapy or
13 long-term therapy.
14 DR. WOOD: I also found the last talk very
15 interesting, but it seems to me the question that
16 we need to debate is where the science is with this
17 and whether the science is mature enough to
18 actually make decisions on this.
19 I mean I would characterize this as being
20 a bit like, say, leukemia. Leukemia is
21 characterized by an increased white count, and
22 clearly the management of leukemia depends on
23 knowing a lot more than just that the number of
24 white blood cells is increased.
25 You need to know the etiology, you need to
75
1 know the subset of patients, the subset this
2 patient belongs to in order to define an
3 appropriate therapy.
4 So, my question I think is the following -
5 is the science mature enough or likely to become
6 mature enough in the foreseeable future to divide
7 patients into subsets based on the kind of
8 divisions that Dr. Woolf described, and are we or
9 will we be at a stage in the near future when we
10 could make treatment decisions based on such
11 subdivisions, or alternatively, is this solely at a
12 stage where this should guide or direct drug
13 development, and are you proposing this, not as a
14 treatment decision paradigm, but one that would
15 allow us to identify potential new targets for drug
16 development, which--and this is important for this
17 discussion--which we would then need to define in
18 some way, a way in which we would approve the drug,
19 because it is improbable that the approval will be
20 based on some surrogate for the subsets you are
21 talking about.
22 Does that make sense?
23 DR. WOOLF: Yes, I think so. The
24 situation we are at currently has been based on the
25 experience with both NSAIDs and opiates, and we now
76
1 have a sense of which patients are likely to
2 respond, the kinds of outcome measures that are
3 sensitive to that.
4 My concern is that the basic science is
5 now revealing new targets which industry are
6 developing new molecules, and the current models
7 that the 1992 Guidelines reflect are not
8 appropriate for that, that if we use these models,
9 there may be heterogeneity of mechanisms in the
10 patient groups that we study that will dilute the
11 outcome measures to a point where it may look as if
12 there is no efficacy globally, whereas, in fact, in
13 the subgroups that do have the particular
14 mechanisms, you would get very high efficacy, and
15 that was a point that was raised by Dr. Fang
16 earlier, that the responder rate may reflect the
17 different incidences of mechanisms.
18 We are at a transition point where it is
19 difficult to predict exactly how useful clinically
20 the identification of mechanisms is likely to be,
21 but I think equally, there is now enough evidence
22 from the COX-2's where we are defining exactly how
23 they produce the effects and efficacy to recognize
24 that we can divide patients into COX-2 sensitive
25 and COX-2 insensitive groups.
77
1 With that knowledge, we can identify some
2 of the best ways to identify efficacy, as well as
3 clinical utility.
4 DR. WOOD: But presumably, the COX-2
5 insensitive group includes all of the above, I mean
6 everything that is not prostanoids mediated, so the
7 heterogeneity in that group is probably at least as
8 large as the heterogeneity in the total group. It
9 is just lacking the prostanoids insensitive group.
10 So, how would you guide either therapeutic
11 decisions on the basis of that, or alternatively,
12 and more importantly I guess for this group, how
13 would you guide the definition of patients to
14 include in the trial that would demonstrate such
15 efficacy, that is not just an exclusion?
16 DR. WOOD: Well, in terms of COX-2's, for
17 example, that if the COX-2's have a label for acute
18 pain, I think that would be too generous in the
19 sense that procedural pain, pain associated with
20 minor acute procedures that would generate
21 nociceptor pain, would not be sensitive to COX-2's,
22 and therefore, that would be an inappropriate
23 usage.
24 Equally, there is minimal clinical data
25 available, but if there were, I think it is likely
78
1 that postherpetic neuralgia and diabetic neuropathy
2 are going to turn out not to be COX-2 sensitive, so
3 that a chronic pain indication, a global chronic
4 pain indication for COX-2's again would be
5 inappropriate.
6 There would be some patients where that
7 would not be likely to produce efficacy. The
8 problem is there is still heterogeneity in the
9 other groups, I accept that, and that is what makes
10 it very difficult.
11 DR. FIRESTEIN: Dr. Ashburn, any comment?
12 DR. ASHBURN: One thing I wanted to point
13 out is that we have been talking about several
14 different definitions of acute versus chronic.
15 Dr. Hertz talked about that the 1992
16 advisory on analgesic drug approval discussed the
17 concept of acute pain as being pain that existed
18 very early on, had a fairly sudden onset and a
19 short duration of action, and chronic pain was pain
20 that had persisted for six weeks in a cancer
21 patient, although I have cancer patients who would
22 say that if it persists for two day, it is chronic,
23 and chronic pain, for people who are not dying of
24 cancer, has to last six months before it meets the
25 definition.
79
1 Dr. Woolf gave what I believe is a more
2 appropriate definition regarding the expected
3 impact on the body and the expected reversibility
4 of the pain.
5 On the other hand, some of the other
6 speakers have really alluded to something which may
7 be more important with regard to drug review, and
8 that is, the duration of therapy, which is much
9 more different, if the expected therapy is of short
10 duration rather for long-term, chronic delivery.
11 I want to just point out that one issue
12 has to do with regard to safety and durability of
13 effect, which I think are very important factors
14 that need to be investigated when a medication is
15 being looked at for outcome. The other one has to
16 do with defining different disease states with
17 which to do studies. That had to with appropriate
18 labeling with regard to dosing interval.
19 DR. FIRESTEIN: That actually begins to
20 bring us towards the second question. We have a
21 couple of other comments that people wanted to
22 make, and then we will move on. But I think most
23 people here seem to be in agreement that a general
24 pain claim is rather vague and it is going to be
25 difficult to approach from a mechanistic or even a
80
1 clinical perspective.
2 I think one of the things that we might
3 want to consider, after hearing the elegant
4 discussion on pain mechanisms, is in addition to
5 acute and chronic, whether or not there might be a
6 place for a third category, such as acute
7 persistent, where patients that have acute
8 mechanisms of pain, that are persistent and
9 reversible, but need to take the medication for a
10 prolonged period of time, might have even different
11 criteria than other chronic indications.
12 Dr. Cush was next, then, we will get a
13 couple of other comments, and then we will move on.
14 DR. CUSH: My comment is to Dr. Woolf. I
15 think that many of us would like to see pain
16 defined mechanistically in an effort to better
17 control pain, maybe use complementary regimens to
18 get more total control, if that were possible, a
19 disease, such as osteoarthritis, but at this point,
20 would you not say that we can maybe define
21 mechanistically how certain drugs may work, and
22 that might well go into some of the preclinical
23 work that would go into maybe how a drug is defined
24 as far as its mechanism of action, but we do not
25 yet have the tools to define mechanistically how
81
1 these drugs work in clinical trial meaning that we
2 don't have the tools for different diseases to say
3 that this going to be a peripheral sensitizing drug
4 or central, and whatnot.
5 DR. WOOLF: If we conduct clinical trials
6 the way they have been at the moment, then, the
7 answer is yes, because global pain scores are not
8 going to identify mechanisms.
9 The big issue there is if we can gather
10 more information, for example, I indicated the
11 peripheral sensitization had a particular property,
12 which is abnormal heat sensitivity in the site of
13 inflammation, whereas, central sensitization was
14 associated with tactile allodynia.
15 Now, if that inflammation were collected
16 as part of secondary outcome measures, maybe we
17 could get an indication whether new forms of
18 therapy acted on those particular mechanisms in
19 addition to whatever global effect they had on pain
20 scores.
21 So, I think we need to move from seeing
22 pain as this monolithic entity with a single
23 expression, which is what the patient feels, to try
24 and collect more data, in the same way that if we
25 look at heart failure, we would make a number of
82
1 measurements - peripheral edema, hypertension,
2 cardiac output, and treat those specifically.
3 I think we need to do the same with pain.
4 The trouble is we are not exactly sure of the
5 durability of these different components and their
6 reflection to mechanisms, but I would argue global
7 pain scores, by themselves, are too insensitive to
8 pick up these individual mechanisms, and therefore,
9 some drugs with some utility may be lost.
10 DR. FIRESTEIN: Two other quick comments.
11 Dr. Davidoff, did you have a comment to make, and
12 then Dr. Abramson, and then we will move to the
13 second issue.
14 DR. DAVIDOFF: Yes, I would also like to
15 add my appreciation for the discussion, which I
16 think was very lucid. But in thinking about that
17 and some of the other comments, it occurs to me
18 that there might be another spectrum in which to
19 make useful distinctions, perhaps even in terms of
20 labeling.
21 That is, there appear to be certain
22 clinical situations which are analogous to some of
23 the, as you put it, preclinical models where the
24 mechanism is relatively pure, and the models are
25 chosen to be able to study a particular type of
83
1 pain.
2 There are others, mostly clinical
3 situations, where it seems pretty obvious that the
4 mechanisms are mixed, and the difficulty is trying
5 to sort them out on some clinical basis whether it
6 is from subtle clinical cues, maybe the development
7 of testing that would allow you to identify the
8 mechanism, or the therapeutic trial.
9 The power of a therapeutic trial, as
10 Alastair has suggested, may actually reemerge as
11 something very powerful, just the way the treatment
12 of hypertension has evolved, so that it is not
13 clear.
14 There are certain relatively pure forms of
15 hypertension, like a pheo or primary aldosteronism,
16 where the treatment is highly specific and narrowly
17 defined, whereas, with most hypertension, it is
18 much more difficult, and, in fact, patients are put
19 on one drug and then a second drug, and a third
20 drug, and nowadays, frequently four drugs, and the
21 therapeutic response is really the way the
22 diagnosis is made, if you were smart enough to know
23 what each of those drugs was doing.
24 So, I wonder if it might be useful to add
25 sort of a dimension of purity versus--how should I
84
1 say--pure versus mixed mechanisms as being one way
2 to consider approaching the labeling.
3 DR. FIRESTEIN: Dr. Abramson.
4 DR. ABRAMSON: I think I had a related
5 comment because it seems that the issue is less
6 whether we should have an acute versus a chronic
7 label, which I think we should because of the
8 different clinical syndromes, but the issue is the
9 heterogeneity of what we are going to be calling
10 indications for clinical pain, and having to
11 grapple with, it that too broad a concept.
12 I mean you are describing different pain
13 mechanisms, and whether we will have a broad-based
14 label is something I think is going to be difficult
15 to grapple with.
16 I am a little concerned in that context,
17 therefore, that to try and dichotomize mechanisms
18 may be premature, in other words, many of these
19 syndromes have to be mixed, as was just said, and
20 some of the science is early and some of the
21 observations don't take into account perhaps the
22 kinetic changes over time.
23 So, I guess the question again for Dr.
24 Woolf is how advanced are the preclinical models in
25 terms of the expression of the different molecules
85
1 in the central and peripheral system and how might
2 we think about, when we do clinical trials in
3 chronic pain, differentiating these different
4 mechanisms based on tissue expression of some of
5 these molecules.
6 DR. WOOLF: I think your point is well
7 made. We are certainly at a point where I think it
8 is appropriate to discuss it and to try and build
9 in our view of the way in which pain is generated
10 to take into account mechanisms, but this is early.
11 This is a point where the kinetics I agree
12 are poorly defined particularly in patients.
13 Unfortunately, many of the changes, the expression
14 of different molecules occur within the nervous
15 system, so access in patients to tissue to actually
16 determine them is extremely difficult.
17 The reliability of animal models for
18 clinical diseases is a separate issue, which is
19 obviously complicated, but I think we just need to
20 try and include this as part of our operating
21 definition of what pain is, and not just ignore the
22 mechanism, particularly since we are at a point
23 where we are about to get new forms of analgesics
24 that have actions that are different NSAIDs and
25 opiates, and as a consequence, may need different
86
1 outcome measures reflecting the action of a
2 particular mechanism.
3 So, we are not there yet, but I think we
4 are a point where, as new trials have been
5 designed, we may need new approaches to them.
6 DR. FIRESTEIN: Actually, we have been
7 migrating slowly towards Discussion Point 2, which
8 specifically asks about mechanistic approaches
9 versus clinical approaches, and maybe we can steer
10 for the final five or 10 minutes of the session,
11 the conversation towards the utility of those two
12 approaches, whether scientifically we are at the
13 point where we should be focusing strictly on
14 mechanistic targets or whether or not the gold
15 standard will be the patient's clinical syndrome.
16 DR. MAX: Let me follow up on Dr. Wood's
17 question on where are we with the science of
18 clinical analgesia. I think it is pretty primitive
19 compared to the animal models because pain is a new
20 enough field, with so few clinical investigators,
21 mostly doing single center trials, that we haven't
22 had the size of the clinical trials combined with
23 the rigor to answer these questions.
24 I think we agree that we are mammals, and
25 if Clifford can demonstrate all these different
87
1 mechanisms in rats, we can in people, and there are
2 a number of examples in the laboratory with humans
3 where we can do, say, a selective nerve block and
4 knock out one kind of pain.
5 We expect that if we looked hard enough
6 with the right tools and the large cohorts in many
7 industry trials, we might find some interesting
8 correlations to learn how to use the drugs better.
9 That is why better tools, if we could
10 develop the equivalent of the arthritis trial
11 groups' scales, we might find things, and I think
12 Clifford's group is working on this, but we are
13 quite primitive, like we have just done a trial
14 with Hopkins looking at a crossover trial of
15 placebo tricyclics and opioids in postherpetic
16 neuralgia in 70 patients, and we find that one
17 group responds to opioids, and an independent group
18 responds to tricyclics, but to really prove that,
19 you would need to replicate, you would need to give
20 the patient back the same drug.
21 We haven't separated that from the
22 possibility of random variation. So, I think the
23 problem for this committee is to provide enough of
24 an incentive for industry trials to try to look for
25 mechanistically based advantages.
88
1 I don't think we can count on that coming
2 out, but I think if we look a little harder, they
3 are going to emerge.
4 DR. FIRESTEIN: Ken.
5 DR. BRANDT: I don't think that Question 2
6 is necessarily an either/or proposition. Coming to
7 responsibilities of safety and efficacy and looking
8 at drugs, if we come back to a way guidelines for
9 management of OA both by the ACR and by ULAR,
10 basically recommends starting with acetaminophen,
11 and if that doesn't work, moving on basically on
12 NSAIDs, and so on.
13 It occurs to me in thinking about Dr.
14 Woolf's comments, we don't know how patients who
15 fail acetaminophen respond to an NSAID. We assume
16 that they are NSAID responsive and they will do
17 better. We don't know that, and it might be useful
18 in terms of this dissection, admittedly at a very
19 crude level and admittedly with the caveat we don't
20 have a clue how acetaminophen works, to get that
21 sort of information in and see whether
22 acetaminophen failures, how frequently they respond
23 to NSAIDs and to agents that perform differently
24 than COX-2 inhibition.
25 I think there is a place to start in this,
89
1 taking a disease that is understood to some extent.
2 DR. FIRESTEIN: But is it more useful to
3 have a musculoskeletal approach or a mechanistic
4 approach for these drugs, for instance, do we need
5 to have separate rheumatoid arthritis and
6 osteoarthritis indications?
7 In spite of what has been said, there
8 actually is a fairly prominent inflammatory
9 component, for instance, do we want inflammatory
10 pain versus non-inflammatory pain, for instance, in
11 musculoskeletal diseases.
12 DR. BRANDT: Well, I think the issue is
13 that there are a number of origins of pain beyond
14 inflammation. There is not any disagreement that
15 OA has an inflammatory component, but, for example,
16 I think that bone pain may be significant in
17 osteoarthritis because of the alterations in bone
18 hemodynamics.
19 That might evoke interest in a whole
20 different class of drugs that would be relevant to
21 OA pain, vaso-active types of medications, that it
22 provides an opportunity by considering the
23 pathophysiology of the disease, and I think you
24 would agree there are differences between RA and OA
25 in a broad sense, not just with regard to pain or
90
1 inflammation.
2 That might provide opportunities to
3 explore different approaches to developing disease,
4 perhaps specific analgesics.
5 DR. KATONA: My question is for Dr. Woolf.
6 Do you have any idea on the developmental aspects
7 of the different pain mechanisms? Just working
8 along with children and adults, it is very obvious
9 that in any inflammatory disease children, who have
10 somewhat less pain, it is easier to be controlled,
11 as well as acute situations don't get chronic as
12 often as adults.
13 I am just wondering if you have ever
14 looked at or whether you have any data on it.
15 DR. WOOLF: There certainly is a major
16 interest in the developmental aspects of pain, and
17 this is an area that I, myself, do not work on, but
18 it appears as if the very early interventions in
19 neonates may have consequences, long-term
20 consequences that are quite different from a
21 similar intervention in children and adults. That
22 is one aspect that needs to be looked at, and then
23 the separate aspect of the responsiveness of
24 children themselves.
25 That raises the whole issue of what are
91
1 the mechanisms that operate or are responsible for
2 the conversion of acute pain to chronic pain. We
3 have heard discussion earlier of when you are
4 giving an analgesic acutely, you may not know
5 whether the patient is going to require that for a
6 long time.
7 Our knowledge of why some patients go on
8 to develop chronic pain, and others do not, is
9 quite poor, and the difference between children and
10 adults in that is certainly an important issue.
11 DR. FARRAR: I think the discussion point
12 asks the question of whether a mechanistic approach
13 or a clinical approach has a rationale, and I think
14 that what we are hearing from Dr. Woolf and Dr.
15 Brandt, and others, is that both of them are
16 clearly applicable to the appropriate use of any
17 medication.
18 It seems to me, though, that the point
19 before the FDA is that we are not yet at the point
20 to be able to mechanistically identify each and
21 every patient that comes to see us. We are also,
22 frankly, not even able to clinically identify at
23 the beginning, the underlying clinical reason for a
24 patient's disease process the first time they come
25 to see us.
92
1 Understanding that the nature of the
2 science of medicine is still very nascent, it is
3 still very much at the beginning, that it is
4 appropriate to consider the way in which a drug is
5 labeled, to consider the way in which patients
6 present and the way in which physicians will then
7 treat them.
8 I am a neurologist. I would love to know
9 what the underlying mechanism is of half the
10 patients that I see who come to me for pain. In
11 fact, I can't do that, even in patients with the
12 same disease process, we cannot identify,
13 necessarily identify their response.
14 In thinking about how a drug company
15 therefore must perform tests to look and see
16 whether the drug is working, I think it needs to
17 focus on the way in which patients present, so that
18 if we can develop a mechanism, Dr. Max was
19 suggesting, a mechanism to be able to actually
20 identify certain subgroups, then, it makes sense to
21 perform trials in those particular subgroups.
22 Until that science catches up, we are left
23 with treating patients with osteoarthritis.
24 Treating patients with osteoarthritis means testing
25 in osteoarthritis and understanding that the
93
1 underlying mechanisms may be very different in that
2 same patient.
3 Where that leads to is again the issue of
4 differentiating between the long-term use of a
5 medication and treating a long-term process,
6 because the two are very different, and I think we
7 need to stick with the way in which medicines are
8 likely to be used for the time being.
9 DR. FIRESTEIN: You have made some very
10 cogent points. I think that while the science has
11 progressed considerably with regard to mechanisms,
12 in the end right now we are faced with patients
13 that come into the clinic that may have multiple
14 mechanisms for a particular clinical syndrome that
15 we are going to be treating.
16 It is likely that at least for now, we
17 need to focus on the clinical presentation for many
18 patients.
19 Lee, I know that there is lots of people
20 that had additional comments, but we need to move
21 on. Are there any additional points that we need
22 to address for this section?
23 DR. SIMON: Not right now except Dr.
24 Goldkind has one more bit of information to add and
25 a question to ask.
94
1 DR. GOLDKIND: Some of this has been
2 addressed by Dr. Firestein. We need to remember
3 that ultimately, the common pathway for approving
4 an analgesic relates to the experience of pain, and
5 so it may be worth discussing whether an indication
6 that is mechanistic in development, but ultimately
7 relates to a metric that is somewhat global, might
8 not be the hybrid, you know, is allodynia
9 associated with a condition, that could be a
10 mechanistically driven indication, but it would
11 still have to ultimately be reflected in the
12 patient's experience.
13 I think we need to remember that the
14 patient ultimately needs to be impacted in a
15 meaningful way, and if it drives development to
16 allow more detail and description in the label or
17 some creativity in an indication, if there is an
18 important benefit to be accrued.
19 DR. FIRESTEIN: There is probably general
20 agreement with that.
21 I think we will end this session here. We
22 will take a 10-minute break, so that we can get
23 back on track. We will see you in a few minutes.
24 [Break.]
25 DR. FIRESTEIN: The next speaker is going
95
1 to be Dr. Lee Simon, the Division Director, and he
2 is going to talk to us about chronic pain and the
3 claim structure.
4 Claim Structure
5 Lee S. Simon, M.D.
6 DR. SIMON: Thank you, Dr. Firestein. I
7 would like to thank again the members of the
8 committee. I would like to take a moment and thank
9 the Divisions of OTC and 170 Anesthetics and
10 Critical Care, for lending us members of their
11 committee to join with the Arthritis Advisory
12 Committee given the fact that pain is such a broad
13 and extraordinary large indication, it affects so
14 many different syndromes and diseases, and much of
15 what you can see our discussion relates to, do you
16 do models or do you do diseases, and ultimately
17 end, as Dr. Witter had suggested, how we do that
18 depends on what we are trying to inform patients,
19 are we trying to inform patients about the
20 syndromes and diseases they suffer from and what
21 kinds of drugs then interfere with them, or are we
22 trying to think about ways that will do also
23 driving new drug development.
24 I think much of these next several
25 discussions that will be presented to you will have
96
1 a lot to do with that.
2 I would also like to just take a second to
3 acknowledge my entire division that has spent weeks
4 in putting these talks together. They have really
5 done a spectacular job, and I would like to
6 acknowledge the fact that this has been one of Dr.
7 Jim Witter's pet projects over the years, even
8 prior to my arrival, and is the culmination of a
9 lot of work for Jim, and I think he has done a
10 terrific job.
11 I would like to thank all of the guest
12 speakers, some of which you have not yet heard, but
13 given Dr. Woolf's superlative presentation, you can
14 imagine the level of conversations we will have and
15 presentations we will have.
16 In the context of chronic pain, let me
17 remind you I am talking now about things that our
18 division in 550, Analgesics, Anti-inflammatory and
19 Ophthalmologic Drug Products, have grappled with
20 and some of the advice that we have been providing
21 some of you sponsors in the audience so far as it
22 relates to the identity of chronic pain.
23 I think that it has been a really
24 informative discussion to think about chronic pain,
25 not just in the context of its chronicity, but also
97
1 in the context of how one uses a drug and how one
2 then thinks about the safety, thus, how one would
3 design a clinical trial to inform you about chronic
4 pain.
5 [Slide.
6 So, pain is always a subjective
7 experience. Some people are quite stoic. My wife
8 never seems to need any kind of anesthesia to get
9 her teeth worked on, whereas, I have to put to
10 sleep to get my teeth cleaned.
11 So, I think that the subjective experience
12 really defines a lot of what we are trying to
13 target here, and that is very important although
14 Dr. Woolf has mentioned that the patient global
15 response is not necessarily going to tell us much
16 about mechanisms, but don't forget the subjective
17 experience, it is important to know what the
18 patient feels about the therapeutic response and
19 whether they are adequately treated.
20 Everyone learns the meaning of pain
21 through experiences usually related to following
22 off your bike or falling around when you are trying
23 to be a toddler and trying to reach that breakable
24 thing on the chair or table above you.
25 As an unpleasant sensation, it becomes an
98
1 emotional experience over time, and it is clearly
2 not only a physical stress, but an emotional
3 stress, as well.
4 [Slide.
5 I have had a really interesting
6 opportunity. I was given the Merck Manual from
7 1899 as a gift when I participated as an author in
8 the Merck Manual of 1999, so it allowed me to look
9 back on pain and the therapy of pain in 1899 versus
10 what we think about in 1999, and what the changes
11 have been.
12 So, in one hundred years, as you heard
13 from Dr. Woolf's talk, there has been clear
14 progress in the field of understanding of pain,
15 defining painful disease states and syndromes,
16 along with delineating appropriate therapy.
17 [Slide.
18 That is shown by this comparison between
19 the original 1899 and now, 1999. So, this, in
20 fact, is the original page from the index of
21 indications from the 1899 Merck Manual,
22 demonstrating pain and the definitions of pain.
23 You will notice that hepatalgia is a very
24 important syndrome of pain in 1899, as was
25 odontalgia, otalgia, ovarian neuralgia, very
99
1 specific definitions as you can see, clearly
2 delineating the way we do today about different
3 kinds of pain.
4 Furthermore, this is the entire list of
5 available pain medications in 1899 that were
6 suggested. Yellow are some of the things that have
7 fallen out of favor, such as iodine or potassium
8 cyanide, something that would not be readily
9 available today for us to use.
10 On the other hand, the white actually
11 demonstrate the drugs that were available in 1899,
12 belladonna, chloral hydrate, codeine, morphine,
13 menthol, some of which may be similar to the kinds
14 of things we use today, like Arthritis-Eze, which
15 is always advertised on the TV about the use of
16 menthol, phenacetin, the parent product for
17 acetaminophen, and sulpyrine was what they referred
18 to as aspirin in those days. I actually didn't
19 know that.
20 [Slide.
21 So, looking now in 1999, this is just one
22 of the pages of the index on pain. As you can see,
23 we have clearly moved forward about categorizing
24 pain in various different ways, both by some of the
25 things you have heard about from Dr. Woolf, as well
100
1 as descriptors, such as after tooth extraction or
2 bladder pain, abdominal pain, psychogenic pain,
3 carpal tunnel syndrome, and this then actually goes
4 on for three pages.
5 [Slide.
6 What also interested us, the separate
7 Analgesics Index, which, in fact, goes on for
8 multiple pages, describes pain relief in terms of
9 acute postoperative pain, or in cancer pain
10 syndromes, or non-opiate drugs for pain,
11 nonsteroidals, opiate drugs, so, in fact, it is
12 really quite interesting how we have come along,
13 where we have been, and where we are today.
14 [Slide.
15 So, we have actually furthered our
16 description of pain, but even 100 years ago, we
17 fundamentally are using today the same fundamental
18 drugs that they were using then - opioids, morphine
19 and codeine, for example, nonsteroidals, as
20 evidenced by salaparendi [ph], "effective aspirin,"
21 it was called in those days, forms of sedatives
22 like chloral hydrate.
23 Well, we don't usually use chloral hydrate
24 today for pain relief, but we certainly use other
25 kinds of things that help people tolerate pain. We
101
1 don't really know how they work, for example, the
2 tricyclic antidepressants, and then muscle
3 relaxants.
4 So, I am not sure that we have come a long
5 way in the analgesic development area. One of the
6 reasons for that has to do with the issue of
7 various descriptors of pain.
8 [Slide.
9 This is an archaic way of actually
10 bringing this about, and I thought that we would
11 start here with this. Dr. Cush actually jokingly
12 referred to this kind of archaic description prior
13 to beginning this session.
14 Somatic pain, visceral pain, and
15 neuropathic pain, not that neuropathic is archaic,
16 but this issue of somatic and visceral are, so
17 somatic pain - caused by the activation of pain
18 receptors in either the cutaneous body surface or
19 deeper tissues, such as musculoskeletal tissues,
20 whereas, visceral pain, pain that is caused by
21 activation of pain receptors, gee, a really similar
22 kind of description, not exactly the way Dr. Woolf
23 would have necessarily described the various
24 different effector agents of somatic or visceral
25 pain.
102
1 So, pain receptors from infiltration,
2 compression, extension or stretching of the
3 thoracic, abdominal, or pelvic viscera, such as
4 chest, stomach, and pelvic areas.
5 What has actually survived this archaic
6 descriptors is the neuropathic pain - caused by
7 injury to the nervous system either as a result of
8 a tumor compressing nerves or the spinal cord, or
9 cancer actually infiltrating the nerves or spinal
10 cord, but unfortunately, this now definition
11 removes or leaves out the issue of inflammation to
12 the nerve root as part of the causal relationship
13 of neuropathic pain.
14 [Slide.
15 Then, we move to something we have already
16 talked about, not just the sense of where it is in
17 the body, but, in fact, the descriptors of how
18 severe it is, so mild, moderate to severe. They
19 are very useful as descriptions. Patients
20 understand severe pain versus mild pain, but to any
21 one patient, that might be very different, so for
22 me, I think walking into the dentist office is
23 severe pain without even having them do anything.
24 So, it does not provide any rigor.
25 Perhaps these should be used to modify the
103
1 definitions of acute and chronic pain indications,
2 which perhaps might allow patients to understand
3 more about how to use, but what measure do you
4 apply for mild, moderate, severe, and ultimately,
5 that measure, either defined by the sponsor or by
6 the agency in evaluating that measure, ultimately,
7 it is the bias of the agency, investigators, and
8 sponsors to suggest which is really which, which is
9 mild, which is severe, which then brings us up to
10 acute versus chronic pain.
11 [Slide.
12 I would like to remind you when we think
13 about this, I think the discussion that was ensuing
14 right before we took the break was really a
15 critical one. It is both a temporal sequence, as
16 well as the idea that the mechanisms are separate.
17 It shouldn't necessarily mean that we are defining
18 them absolutely. This is an area that is
19 iterative, it is still in development.
20 We don't have a clue about all the
21 aspects, as you have already heard, and, in fact, I
22 expect that in 10 years from now, we will know a
23 lot more than we do today.
24 So, acute pain - short-lasting, so
25 temporal component, manifesting in objective ways,
104
1 perhaps that is mechanistic. It can be easily
2 described and observed.
3 It may be clinically associated with
4 diaphoresis and tachycardia, so there are clinical
5 events that take place associated with the
6 transient events, the transient stimulus that leads
7 to the acute pain.
8 Maybe only lasting several days,
9 increasing intensity over time, which might lead to
10 this issue of that bridge between acute and
11 chronic, the subacute pain. It can occur
12 intermittently, episodic or intermittent pain. Dr.
13 Sherrer referred to an OA flare superimposed on top
14 of a more chronic event.
15 Usually related to a discrete event for
16 onset, such as postoperative, post-trauma,
17 fracture.
18 And then there is chronic pain -
19 long-term, typically defined if it lasts for
20 greater than three months, in the context of cancer
21 pain, perhaps less based on survival issues. More
22 subjective and not as easily clinically
23 characterized as acute pain, and has a more
24 psychological overlay.
25 I don't mean to suggest that we are
105
1 incapable of understanding and identifying chronic
2 pain, but tachycardia and diaphoresis is not
3 necessarily associated with the onset of chronic
4 pain. This kind of pain usually affects a person's
5 life, changing personality, and their ability to
6 function, as well as their overall lifestyle.
7 [Slide.
8 That brings us to a discussion that Dr.
9 Firestein led just before - what about the general
10 descriptor of pain, why can't we just label these
11 things for pain and let the marketplace decide, why
12 can't we just say it works in this kind of pain,
13 and you could try it in something else, and if it
14 doesn't work, you try something else.
15 That might be helpful and useful, but it
16 is not particularly informative to patients,
17 particularly with what we know today. The general
18 pain definition has been broadly used in the past,
19 however, acute and chronic indications use
20 different models, may be mechanistically different,
21 and have different safety issues.
22 Furthermore, the psychological component
23 clearly separates acute pain from chronic pain, and
24 that may have very important implications for
25 therapeutic intervention, patient response, and
106
1 patient safety claims.
2 [Slide.
3 Unfortunately, one of the major proponents
4 of this kind of meeting was not able to make it
5 today, and I wanted to allow Dr. Lipman to seem
6 like he is actually in the audience by bringing up
7 some of the things that he has referred to in the
8 past, one of which is this particular statement
9 from a paper in Cancer Nursing, which is that
10 chronic pain has a psycho-social component that
11 must be dealt with before depression becomes a part
12 of the clinical picture. Chronic pain should be
13 recognized as a multi-factorial disease state. So
14 it is a state that is responding to something, but
15 nonetheless, may be an independent disease state
16 requiring intervention at many levels.
17 [Slide.
18 This diagram actually reflects these many
19 levels and demonstrates the interaction that over
20 time basically, whatever the pathologic process is,
21 associated with the interaction with physical
22 factors, leads to anxiety, depression, and
23 psychological factors overlying each of these
24 events, so that in the right circumstance and in
25 the right patient, there could be issues of
107
1 isolation and loneliness, totally informing the
2 patient leading to increasing anxiety and
3 depression, the issues of hostility, why me, why is
4 this happening to me, why can't I deal with this,
5 and then the issues of social factors, which lead
6 to the increasing loneliness and anger associated
7 with this increasing isolation, thus suggesting a
8 time period that we are liable for being able to
9 intervene, to be able to allow this cascade of
10 events perhaps not to progress.
11 [Slide.
12 So, in thinking about trial design from
13 the regulatory point of view, we have to think
14 about again how Dr. Witter suggested, what are the
15 issues regarding how to inform patients about their
16 use of these particular therapeutic interventions.
17 So, look for trial designs that will allow
18 us to see the result of how to translate the use to
19 the patient, so as Dr. Hertz suggested before, we
20 are becoming much more interested in disease states
21 to be studied than models to be studied.
22 At the time, we didn't have a lot of
23 understanding of the diseases. It seemed
24 reasonable to try to look at models, but is
25 alveolar bone pain in dental extraction the same as
108
1 bunionectomy, is dysmenorrhea, which actually has a
2 clear mechanism of understanding of why there is
3 cramping and abdominal discomfort, is that actually
4 extrapolatable in a general way to other forms of
5 pain.
6 So, some of the models that we were
7 looking at are disease states that we have been
8 looking at, have been osteoarthritis, chronic low
9 back pain, which has been a big debate, some of
10 which we will be informed in a little bit by Dr.
11 Borenstein, fibromyalgia, an area of great and
12 intense investigation, which has some very
13 interesting aspects to the psychological overlay of
14 how people deal with their pain, and perhaps
15 genetics, about who selects out the individual
16 response to an inciting event, and then who goes on
17 to develop a chronic pain syndrome without further
18 inciting episodes.
19 Neuropathic pain, and there are many of
20 those, I just selected out two - diabetic
21 neuropathy and amyotrophy, cancer pain and the old
22 issues associated with that, that are quite unique.
23 Temporomandibular joint pain, peripheral vascular
24 disease perhaps, and then not only the disease
25 states or models, but what about mechanistic
109
1 approaches.
2 [Slide.
3 I am going to present three different
4 possibilities for your consideration. I almost
5 feel like Rod Serling in creating the Twilight
6 Zone. These are all just for your consideration.
7 We would like to throw out the possibility that we
8 want to engender drug development.
9 We think this might be a good way to go,
10 but now that I am on the light side rather than the
11 other side, perhaps I don't have the right
12 perspective that other people have about what is
13 necessary, so we have to think about this together
14 as whether or not these are the right ways to do
15 things.
16 So, possible indications of one disease or
17 model, one could even add in mechanism perhaps, an
18 example, signs and symptoms of osteoarthritis. Not
19 everybody knows that OA is osteoarthritis but us
20 rheumatologists do. So, an example, signs and
21 symptoms of OA, two replicate randomized and
22 controlled trials, three co-primary outcomes in
23 which each must win, so it would be pain, function,
24 and a patient-determined global. And why would we
25 want that latter one is again it is important for
110
1 us to know how the patient feels, not unimportant
2 in labeling and allowing other patients to know
3 what that means. There yet may be other measures
4 that become important as we will talk about in a
5 second.
6 There needs to be superiority to placebo
7 or perhaps superiority to an active comparator.
8 There could even be discussions, although it is not
9 on this slide, about non-inferiority to an active
10 comparator, but, in fact, that would have to be
11 defined based on some issues as shown in the
12 appended paper from Ellenberg and Temple about
13 placebo responses and things like that.
14 [Slide.
15 There is also the possibility of thinking
16 about a whole organ system indication, such as
17 musculoskeletal disease, and then one might think
18 about, for example, improvement in the pain of
19 musculoskeletal disease.
20 Three models of diseases, though, might be
21 required to achieve this, all within the rubric of
22 musculoskeletal disease, so low back pain perhaps
23 in association with studies in osteoarthritis, and
24 then perhaps also in fibromyalgia, all of which
25 affect the musculoskeletal system, we believe, and
111
1 perhaps inform us somewhat about the use in a
2 general way in musculoskeletal disease.
3 You will need two replicate randomized,
4 controlled trials for each model or disease state.
5 There need to be three co-primary outcomes, each of
6 which have to be won on, of pain, function, and
7 patient-determined global, and it could be
8 superiority to placebo or superiority to active
9 comparator, or maybe in the right circumstance
10 non-inferiority that we could discuss.
11 The important aspect of this would be that
12 the label would reflect, not just the idea of
13 musculoskeletal disease, but reflect the approval
14 of all the disease or models that had been studied,
15 so therefore, you would get the approval for
16 musculoskeletal disease in osteoarthritis and
17 fibromyalgia and chronic low back pain, which would
18 be actually in the label, as well as in the
19 Clinical Studies Section, to inform people about
20 the responses.
21 Furthermore, we would be even interested
22 in discussing the issue of, well, gee, in
23 fibromyalgia, maybe wind-up, the concept of wind-up
24 pain is really critical, and perhaps, in fact, if
25 you could interfere with that, in drugs that are
112
1 quite unique, that have nothing to do with what we
2 have thought about pain before, such as an NMDA
3 inhibitor, perhaps that might be the right way to
4 go and achieve that for fibromyalgia.
5 [Slide.
6 Then, the big discussion point that a lot
7 of people have heard before and we have informed
8 people about is the idea of a general chronic pain
9 indication. Now, this seems to be quite a high
10 bar, however, just think about how high a bar it
11 reflects, meaning it could be suggesting that drugs
12 could be used in any form of chronic pain.
13 Now, this leads us to a discussion of
14 lumping and splitting, and some of the discussion
15 we have had to date would suggest that it is going
16 to be impossible as we learn more mechanisms to
17 actually get a drug that would be appropriate for
18 chronic pain totally, and that may well be true.
19 Thus, I would take you through this
20 argument, suggesting that replicate trials in each
21 model should be in disparate diseases, so you would
22 have to study one aspect of musculoskeletal
23 disease, one aspect of cancer pain, and perhaps one
24 aspect of neuropathic pain, and that product,
25 whatever that product might be, would have to win
113
1 in all three areas.
2 However, this is not to limit the possible
3 areas. It may be that you could figure out
4 something else besides neuropathic pain to study
5 and thus get the same rubric - must measure pain,
6 patient global, and some functional outcomes are
7 the co-primaries, and again win, must be superior
8 to placebo in all three and superior to the active
9 comparator, and again, I point out that the label
10 reflect two issues.
11 One would be the approval for the broad
12 category, limited specifically by safety
13 considerations, and the label will also, based on
14 the data accumulated to achieve this, would
15 demonstrate that the therapy is approved for the
16 indication of chronic pain, but also the three
17 diseases or models or mechanisms that had been
18 studied, so therefore, it is kind of four things.
19 You get all three areas, perhaps other
20 areas that you were also studied in, so if you did
21 musculoskeletal disease into two different areas of
22 osteoarthritis and chronic low back pain, they also
23 would be referenced in the label and in the
24 Clinical Trial section as thought appropriate for
25 patients information and clinician information.
114
1 [Slide.
2 Yet, there is still yet another approach,
3 which we certainly want to encourage, although we
4 are not entirely sure how to go about doing it, I
5 don't know if you are, is the mechanistic approach.
6 We don't yet know how to do it, we don't really
7 know the models, but possible examples, as Dr.
8 Witter alluded to, perhaps alteration of wind-up by
9 inhibition of NUDA receptors in fibromyalgia,
10 alteration of brain plasticity or neuroplasticity,
11 alteration of early markers that might predict
12 specific and verified clinical outcomes, thus
13 giving a broad opportunity to really drive the
14 science and improve drug development.
15 [Slide.
16 All of this has to be remembered in the
17 context that we, at the Agency, have to label
18 things in the context of benefit to risk. So, as
19 this cartoon suggests, as this unfortunate person
20 sitting at this particular cafe selecting out which
21 food to choose, and seeing the risks and benefits
22 that are listed up on each one, it would not be
23 dissimilar from a physician, patient, or clinician
24 choosing particular drugs to choose based on their
25 benefits to risk, as listed within documentation
115
1 that had been accumulated in trial development.
2 Thank you very much.
3 DR. FIRESTEIN: Thank you, Lee.
4 Discussion Points #3 and 4
5 DR. FIRESTEIN: At this point, we have
6 been asked to discuss Points 3 and 4 here. Yes?
7 DR. MAX: I would like to comment to Lee.
8 As I have said to you before, I really like one
9 thing you said, and I am really profoundly worried
10 and I really hate another thing you said.
11 What I really like is that your primary
12 goal is to advance the science by encouraging many
13 clinical trials in many diseases, and I have
14 written a review article in Anesthesiology last
15 July with Clifford, where we conclude that the best
16 way to learn about mechanisms in human is from
17 clinical trials in many diseases, and your approach
18 does that.
19 The one thing--and I think it is a detail
20 that I am very concerned with--is your stipulation
21 that each trial needs to demonstrate, at the same
22 time, a win for not only pain, pain scores over
23 placebo, but in addition, a global outcome, global
24 patient preference, and quality of life.
25 I would argue that if you look at large
116
1 databases of opioid trials and malignant and
2 nonmalignant pain, as my colleagues in the
3 Anesthetic Division have, and in my experience
4 looked at chronic neuropathic pain and chronic back
5 pain in other trials, it is unusual that one shows
6 all three at once, and maybe we are behind you in
7 OA, and I am afraid if you tell industry that you
8 need to have a win in all three for each positive
9 trial, that it's a why study pain, let's give that
10 up, it's an impossible thing to meet.
11 I would propose the alternative, that you
12 show pain is reduced more than a placebo by
13 statistically significant outcomes, and at least
14 you show evidence that you are not intoxicating the
15 patient, there is no deterioration in the global or
16 in the patient preference, and perhaps as an
17 additional tier, you can get additional claim to
18 give the incentive to develop better quality of
19 life. That's my counterproposal.
20 DR. SIMON: I would just like to point out
21 that, and I am delighted that I have stimulated
22 this kind of discussion, that the quality of life
23 measures are not necessarily the same thing as
24 function, and what we are relating to are
25 functional measures, not necessarily requiring the
117
1 bar of achieving an improvement in quality of life,
2 although that is very important to us and certainly
3 would be a secondary outcome that we would be
4 looking for.
5 It is unfortunate that a lot of the
6 definitions of health-related quality of life
7 measures have been assumed to be measures of
8 function. It is not necessarily clear that all
9 are measures of function, and I am not yet sure
10 that we have all the measures that we need to
11 achieve this particular proposal.
12 It may well be that measures of function
13 yet need to be developed in cancer, for example,
14 that will allow us, to inform us in the relative
15 short term of study, that patients with cancer
16 whose pain is improved would benefit from function,
17 as well.
18 This is a suggestion of not just the
19 development of new drugs, but new outcome measures
20 that is critical, and I think Dr. Strand will be
21 discussing some of the issues about the tiered
22 nature of how to look at that question.
23 DR. FIRESTEIN: Dr. Strand.
24 DR. STRAND: I just wanted to comment back
25 to you, Mitch, that, in fact, we know from
118
1 certainly musculoskeletal diseases, OA and RA, that
2 when you improve pain, and even if that is the most
3 that you seem to improve in terms of the disease,
4 such as the COX-2's in, say, rheumatoid arthritis,
5 you are still getting responder analyses, you are
6 still showing improvement in physical function, and
7 improvement in health-related quality of life.
8 So, in fact, these domains are affected
9 very significantly by pain and they are improved by
10 pain, so I think that perhaps the bar is not as
11 high as you might think.
12 Obviously, we have to look at it in terms
13 of what disease states or what mechanisms of pain
14 we are trying to treat, but it goes to show that
15 with the multiple ways pain affects people in their
16 day-to-day lives, if we are improving that, we
17 should see it in these other aspects.
18 DR. FIRESTEIN: I guess the other issue is
19 whether pain and these other outcome variables,
20 especially quality of life, are independent. I
21 think we have had a lot of these discussions with
22 regard to rheumatoid arthritis and osteoarthritis
23 where quality of life is a dependent variable on
24 pain, as well as other aspects of joint
25 destruction.
119
1 So, it is not clear to me that you gain a
2 lot from a measure of quality of life if you don't
3 get a win because of statistical vagaries or an
4 inaccurate instrument for measuring that when the
5 patient is subjectively better based on other
6 criteria for pain.
7 Yes, and then Dr. Katz.
8 DR. ELASHOFF: Yes, the whole issue of
9 exactly what the correlation is between these
10 measurements across patients or across studies is
11 an empirical one. I suspect that they are never
12 completely independent, but that the correlation in
13 some cases might be low and in other cases it might
14 be high.
15 I think one needs to think conceptually of
16 what one might expect in any given situation and
17 why you might expect them to be less correlated or
18 more correlated, but this is an empirical question
19 on which a lot of light could be thrown by proper
20 analysis of older studies.
21 Typically, there isn't enough in-depth
22 analysis of exactly what the relationships are
23 among various outcome measurements, and I would
24 like to encourage that not only new studies be
25 asked to really look in detail at the relationships
120
1 between these outcome variables, but that older
2 studies could be re-analyzed to address that
3 question.
4 DR. KATZ: I would like to caution against
5 a "one size fits all" strategy with regard to what
6 domains one might require to say that a trial is
7 successful or not successful, and I would also like
8 to caution against an overly enthusiastic
9 generalization from the rheumatic diseases to other
10 types of pain in that regard.
11 For example, it is clear that if somebody
12 is on their death bed with cancer pain, you know,
13 one's obligation is to relieve pain and its
14 associated suffering, and the opioids are a
15 miraculous and time-proven strategy for that.
16 To then require that that patient get out
17 of bed and walk down the block, or do some other,
18 you know, or improve functionally in some way would
19 be a big mistake and would prevent us from really
20 achieving our primary goals in that situation.
21 Certainly, one could design a functional
22 measure heavily weighted towards pain that might
23 show function, but that is, you know, just a
24 remeasurement trick that doesn't really accomplish
25 anything I don't think.
121
1 Similarly, in the patient, a 75-year-old
2 with postherpetic neuralgia, with a 4 out of 10
3 pain, they might be pretty much doing what they
4 need to do every day anyway, and that doesn't meant
5 that relieving their pain is not an accomplishment
6 even though it would be very tricky to design a
7 functional or quality of life measure that would
8 show dramatic improvement.
9 Lastly, you have got some really bad power
10 calculation issues in terms of powering a trial to
11 improve an SF-36 or something like that. It really
12 sets a very high financial and feasibility
13 threshold when, in many cases, relieving pain is
14 really the primary goal.
15 Although in osteoarthritis, I can
16 certainly accept that function is an intrinsic part
17 of what we are trying to improve there, and in that
18 context, it may make more sense, so I think we need
19 to think carefully about each individual situation.
20 DR. FIRESTEIN: Dr. Callahan and then Dr.
21 Cush.
22 DR. CALLAHAN: First, I would like to
23 agree probably in musculoskeletal diseases, they
24 are very different, but I do agree with Dr. Strand
25 in terms of pain and function are highly
122
1 correlated.
2 My question was for Lee. When you say
3 pain based on our discussions this morning, are you
4 talking about a global pain or talking about
5 various types of pain to get a global pain, as well
6 as specific pain that would get at more of what was
7 presented by Dr. Woolf?
8 DR. SIMON: Dr. Firestein, can I answer
9 that?
10 DR. FIRESTEIN: Of course.
11 DR. SIMON: Thank you.
12 DR. FIRESTEIN: The Chair appreciates your
13 request.
14 DR. SIMON: I learn from previous
15 experience.
16 I think that your question really relates
17 to the lack of development of the area. If this
18 was five years hence, and Dr. Woolf's scenario was
19 translated to a specific new receptor inhibitor, we
20 would likely be thinking exactly in the terms that
21 you have just said.
22 Our problem is, is that we are not yet
23 there. I could envision three different receptor
24 inhibitors demonstrating improvement and perhaps
25 even getting a moniker chronic pain indication
123
1 depending on whether or not they are broad enough
2 to warrant that, again going back to the lumping
3 and splitting concept.
4 Yes, I believe in the splitting concept
5 because I think that, and I think much of our
6 division does, many in our division do, because I
7 think the reasons for that are very logical and
8 disease-specific and mechanistic understood.
9 For example, in acute pain, I can't
10 imagine that a drug that necessarily works in
11 dysmenorrhea will necessarily work in bunionectomy,
12 and just because it works in dysmenorrhea and is a
13 good model to study for that particular event, and
14 it tells you something about one day of use,
15 doesn't mean it is translatable to other forms of
16 pain, but I think we are limited.
17 We don't have all of that information yet.
18 I would like to believe that what I have proposed
19 or what we have proposed may actually lead us in
20 the way to develop more, not less.
21 DR. CUSH: My comments are directed at Lee
22 and Jim, that I think given the comments of Dr. Max
23 and Dr. Katz, I think that to consider a pain
24 indication is reasonable and then to define that,
25 that the indication here is pain, but there is also
124
1 improvement, not only in pain, but in quality of
2 life or function or in a patient global, that could
3 be in the indication as determined by the research
4 that is done, might be very useful to users and to
5 patients and whatnot.
6 To get to your suggestions regarding
7 indications, I like the idea of disease-specific,
8 organ-specific, and then global indications, I
9 think that that sets sort of sequentially more
10 difficult tasks, but greater implications to the
11 populace, and I think that the design you laid out
12 would be very useful.
13 DR. FIRESTEIN: Dr. Abramson and then Dr.
14 Ashburn.
15 DR. ABRAMSON: Lee, I would just like
16 address the splitters versus lumpers question and
17 make a case for splitting.
18 Even in the realm, the domain of
19 musculoskeletal disease, because fibromyalgia, OA,
20 and low back pain are obviously going at different
21 mechanisms perhaps, and I think we are at a moment
22 now where we can hypothesis test some of the
23 mechanistic concepts, and we can do it using
24 clinical studies.
25 I think if we look at fibromyalgia
125
1 differently, if we lump them, we may lose the
2 opportunity to looking at different
3 mechanistic-based pain pathways. So, I would argue
4 for splitting largely as a way to do clinical
5 trials to test these different potential mechanisms
6 neatly and cleanly.
7 DR. ASHBURN: I found your presentation to
8 be quite interesting and I think that many of your
9 aspects were starting to be well thought out, but I
10 have the same sort of love-hate relationship that
11 Dr. Max presented before, because one of the things
12 that you alluded to even when you were talking
13 about your experience in the dentist and your
14 wife's experience in dentists, is that pain is many
15 things.
16 Pain is not purely nociception, which many
17 physicians think of it, but rather, pain is a
18 global area, and it is best treated using a
19 bio-psycho-social model of care including
20 interdisciplinary care of which medical management
21 is only one part of the care.
22 When one is talking about taking care of
23 patients with complex disease, even I think of
24 headache as complex, maybe my neurology colleagues
25 don't think of it, but those patients are fairly
126
1 complex. Medical management is only one part.
2 The NIH Consensus Conference was done
3 almost a decade ago now, presented that
4 self-management techniques were equally efficacious
5 to the medical interventions that we frequently
6 focus on.
7 So, one of the issues is that setting
8 study and outcome measurements in those patients is
9 a good start, but is fairly difficult to do. There
10 are disease-specific measures of health that Dr.
11 Carr may talk about that are under development with
12 regard to the care of individuals who have complex
13 pain problems, but they are in their infancy.
14 They frequently look at function, they
15 look at physical function, as well as mental
16 function, and they usually have several different
17 scores enveloped into one area, and then the
18 question would be, drilling down, is improvement in
19 one functional score adequate, is improvement in
20 many adequate, does it matter.
21 Those are the sort of issues that make me
22 nervous, and the concern that I have is, is that
23 while it is an excellent idea to integrate
24 measurement of outcomes amongst a wide variety of
25 fields as a requirement to looking at new
127
1 medications, requiring that positive benefit be
2 shown may be a barrier to care and may actually
3 decrease interest in the development of new
4 medications for the treatment of these patients.
5 DR. FARRAR: I have to say that I really
6 enjoy coming to these meetings because I get to sit
7 in a room with a group of real experts and hear
8 them disagree vehemently about things that we are
9 all talking about, and yet with the same common
10 goal, which is to strive to make patients' lives
11 better, which is ultimately what medicine is about.
12 I think, in part, I won't comment on what
13 I loved and hated about Dr. Simon's presentation,
14 but one of the things that he said that certainly
15 is applicable to this, is that things are going to
16 change and that we are not targeted today or we are
17 not charged today with coming up with the final and
18 ultimate answer, that we are charged with coming up
19 with what makes the most sense for right now.
20 It made me think about the fact that we
21 really have to be honest with ourselves. If we had
22 a drug that was absolutely spectacular in the
23 treatment of pain, in the way that penicillin was
24 with pneumococcal pneumonia, you wouldn't need a
25 randomized trial and you could use any measure you
128
1 care to use, and you would come up with a positive
2 result.
3 What comes to mind in pain management is
4 hip replacement in an old patient who has a broken
5 hip that is amenable to that treatment. I mean any
6 way you look at that, the patient is better. The
7 patient's pain is better, they can walk again, they
8 can get out of bed. Any measure you care to use
9 would work.
10 The unfortunate part is that in
11 medications, we are not yet at that step. It seems
12 to me, therefore, that what we are charged with
13 really is providing enough information to the
14 people who are going to be using these medications
15 to allow them to make reasonable choices about how
16 they treat their patients.
17 I agree that, you know, the clinician on
18 the front line is faced with a whole bunch of
19 different choices, and if we can figure out the
20 mechanism and figure out a test that will give them
21 the mechanism, then, by all means, a mechanistic
22 approach makes sense.
23 If can figure out whether we know this
24 patient is going to develop an allergic reaction
25 and this one is not, then, we should choose
129
1 obviously only the group that has the allergic
2 reaction.
3 It occurs to me that we are not there yet,
4 and that really, in many ways, what the label needs
5 to reflect--and I keep coming back to the label
6 because ultimately, that is what gets out to the
7 public and then obviously clinical trials on top of
8 that, but what the label needs to reflect is what
9 is it that we know about this drug, do we know that
10 it is safe given in three doses, do we know that it
11 is safe given in 1,000 or in 500 milligrams, do we
12 know that it is safe in terms of kids or adults or
13 pregnant and not.
14 In terms of efficacy, do we know that it
15 works when given in a single dose--that is
16 important--do we know that it works when it is
17 given over a long-term period of time.
18 With that kind of information in hand, I
19 think it is possible to practice medicine, and that
20 is really what we are targeted at doing today.
21 Clearly, one size does not fit all, and every drug
22 is going to have a different set of underlying
23 things that we need to know about it.
24 That makes the job very, very complicated,
25 which is clearly indicated by the amount of
130
1 disagreement that we have, but I think we need to
2 focus on that.
3 DR. FIRESTEIN: Thank you, although I
4 don't think the sham surgery for hip replacement
5 protocol has been completed yet.
6 DR. STRAND: I just wanted to say that
7 neither should we be trying to shove responder
8 analyses based on other diseases into the pain
9 field, and the fact that RA and OA have actually
10 been addressed very differently from that point of
11 view, but that we should really be thinking about
12 these things as domains, domains of physical
13 function or function period domains of
14 health-related quality of life, and not pick the
15 instrument.
16 We have lots of disease-specific
17 instruments for various kinds of diseases, we have
18 ones for cancer pain, et cetera, so that we don't
19 have to shove the idea into a situation where it is
20 not clinically appropriate.
21 DR. McLESKEY: Well, Lee, you certainly
22 stimulated the discussion. As the industry
23 representative, I would probably be negligent in my
24 duty here if I didn't have at least some response
25 at this stage.
131
1 I would like to echo Dr. Farrar's comment
2 of a minute ago that obviously our entire goal for
3 being here, your agency, and the various roles of
4 the folks in this room is to advance the practice
5 of medicine, to advance the options available to
6 treat patients.
7 I hope we keep that foremost in our minds
8 as we discuss all of these various issues, what
9 will optimize that result, what will optimize the
10 advance of the practice of medicine and how can we
11 safely achieve that goal with advances in the
12 medications available to our patient public.
13 The pushback that I have heard you receive
14 already or your comments receive already from a
15 couple of the members of the committee on this side
16 of the table specifically, I think probably is
17 representative of the novel concept that you have
18 approached, the innovative concept that you have
19 approached, and expected kind of a result from
20 that, understanding our current knowledge base of
21 disease models, and so forth, and how to measure
22 accurately the effectiveness, and so forth, of
23 various medications.
24 The concept that you mentioned especially
25 for a general claim of three disease states and
132
1 having to hit on all three of the aspects of pain,
2 function, and global, to me seems like a pretty
3 high bar, and I wonder if the industry colleagues
4 of mine in the room would not feel similarly, and
5 yet, on the other hand, we don't want to act like
6 antagonists and pull back and push back and oppose
7 advances as the advances in the understanding of
8 the mechanisms of pain have been discussed earlier
9 today.
10 So, I would just suggest that we don't
11 want to make the hurdle so high that, in fact, it
12 will stifle innovation and move exactly in the
13 direction we don't want to go. We want to
14 stimulate innovation and advance and move forward.
15 So, again, I hope I am not coming across
16 as somebody who is antagonistic to advance, I am
17 not, but I think to accurately represent industry,
18 we would like in the future to work closely with
19 the regulatory authorities and with the
20 academicians, and so forth, to come up with some
21 kind of a compromise approach that is reasonable,
22 that provides a hurdle that we think we can get
23 over and accomplish the eventual mission of pushing
24 medicine forward.
25 DR. FIRESTEIN: On the other hand, maybe
133
1 the bar for a global pain indication needs to be
2 high because a drug that really is or a therapeutic
3 that really is appropriate for all pain
4 indications, as a global pain indication would
5 suggest, is not really practical at least with the
6 current state of knowledge.
7 There are so many mechanisms of pain, it
8 is actually unlikely that we would find something
9 that is effective for wind-up pain and fibromyalgia
10 and osteoarthritis and cancer pain, and the
11 question is whether or not, under those
12 circumstances, the graded approach that has been
13 suggested, in particular a disease-oriented
14 approach followed by an organ-oriented approach,
15 followed by a global pain indication is reasonable
16 because the final Holy Grail of global pain is, in
17 practical terms, not really approachable based on
18 the science that we have heard today and has been
19 written about over the past several years.
20 DR. McLESKEY: Perhaps so, but on the
21 other hand, the comments that I have heard from Dr.
22 Farrar and others indicate that maybe we are not
23 quite there yet, and are we trying to run a little
24 bit too soon before we have perfected the issue of
25 walking.
134
1 But, nevertheless, as you have said, that
2 in order to achieve a global claim, which would
3 obviously be attractive to industry, and I would
4 argue would be attractive to clinicians to some
5 degree, as well, to offer them flexibility, and so
6 forth, if we are to hit on three separate
7 indications or diseases and to perform those
8 indications in replicate, and on each of those hit
9 on the three issues of pain, function, and global,
10 that implies to me that the sponsor would have to
11 perform a substantial number of pivotal trials in
12 order to achieve that mission, which again makes
13 the hurdle extremely high.
14 DR. BRANDT: Just a question for
15 clarification based on what you just said, Gary.
16 You referred to global pain. My understanding of a
17 patient global, for example, is a little different
18 from that, and one of the problems is there are
19 many, many, many globals, it depends on how you ask
20 the question.
21 For example, taking all things into
22 account, how is your arthritis or how is your
23 disease doing, which takes into account side
24 effects, it takes into account other joints than
25 the index joint and so on.
135
1 Perhaps Lee could clarify what he meant by
2 his global.
3 DR. FIRESTEIN: Well, my understanding is
4 that global means all pain, all indications.
5 DR. SIMON: Actually, let's be very clear.
6 A patients global response is very different than a
7 global indication, and so we would ask for patients
8 to tell us how they feel, as Dr. Brandt has
9 suggested, but Dr. Firestein, I think--I don't mean
10 to put words in your mouth although I am delighted
11 about what you said--was referring to the concept
12 that this high bar would likely stimulate further
13 development because, in fact, it would allow us to
14 look at a therapeutic that would be active in very
15 different disease states, thus, a global chronic
16 pain indication. A very different use of the
17 "global."
18 DR. WOOLF: I think this issue has
19 implications for the preclinical development of
20 analgesics which we haven't really spoken about,
21 but the information that can be derived in terms of
22 global action across a matrix of pain models is
23 essential.
24 I think that as the development plan for
25 any given analgesic is entered into, we need to
136
1 have as good an evidence as possible of the action
2 of the particular drug, its specific action in
3 terms of which targets it is interacting and its
4 relative efficacy in a broad range of different
5 models, models that are maybe more sophisticated
6 than some of the ones that are being currently
7 used.
8 DR. MAX: Let me put forth what I hear is
9 the consensus around the table and see if it really
10 is. I think we may be suggesting to you that there
11 is no objection to having a general pain claim that
12 requires two studies in each of three different
13 disease categories.
14 We could learn a lot from all the
15 different studies that will come in, and I just
16 hear some objection to making the lowest level
17 general pain claim have each of the six trials get
18 all three endpoints, and the counterproposal is
19 that general pain can be six trials, 3 times 2,
20 each getting pain, is reduced significantly, but to
21 get statistically significant global patients and
22 function would be incentivized by a higher level
23 reward, just like the rheumatoid arthritis claims
24 do that, and I think I agree with Vibeke and others
25 that it is important to have an incentive to
137
1 develop better measures because there are real
2 issue, should we be spending for COX-2's, should we
3 be giving opioids chronically.
4 Function makes a difference in these
5 questions, and we need to know more about it, but I
6 think we are suggesting to you that there be an
7 additional carrot for this.
8 Does that capture what you are saying,
9 Charles?
10 DR. McLESKEY: I am not sure, Mitchell, I
11 am not sure that there is universal unanimous
12 agreement that there would be three separate
13 disease states studied in order to achieve a
14 general claim.
15 I am one guy representing, obviously,
16 trying to represent industry, but I work for one
17 company, and I would suggest that before such a
18 generalization or a statement like that of general
19 acceptance were achieved, that there be some kind
20 of working group formed where there would be
21 representatives from several of the major players
22 in this area to make sure we have consensus of that
23 kind of an approach.
24 DR. FIRESTEIN: But it is important to
25 remember that whether the number is three, you
138
1 know, three disease areas, or four, or two, or five
2 or six or more, that the global pain indication
3 should, by necessity, be a very high standard,
4 because it needs to cross all mechanisms.
5 The question is whether it serves the
6 clinicians well to have a global pain indication
7 for a drug that does not work well in neuropathic
8 pain, for instance, if you have done one or two
9 other diseases or organ systems.
10 I think the bar is, by necessity, going to
11 be high for global pain because that is in essence
12 all pain under all conditions. It seems to me
13 based on what I have heard today that there are
14 lesser labeling criteria that still are very broad
15 and still would be probably more reachable than we
16 are today with current technology.
17 So, asking for all pain under all
18 conditions when it hasn't been demonstrated is
19 perhaps asking for something that is not really
20 appropriate at this point.
21 DR. McLESKEY: I appreciate your comments.
22 My only retort to that is that we need to balance
23 incentives in order to advance the field versus the
24 hurdles that are placed in order to achieve those
25 goals, and that kind of a consensus development at
139
1 this stage, I would suggest needs input from some
2 others who perhaps are not at this table.
3 DR. WOOD: Just to respond to this, I
4 think it is important. I don't think we have
5 consensus, at least certainly not from me, that the
6 global pain indication would be required for
7 approval.
8 So, I would visualize that a drug would
9 come to the Agency and get approved perhaps with a
10 more restricted label and could progress
11 incrementally up that scale as experience, and so
12 on, increased.
13 It would seem improbable to me that a
14 company would go for a global pain indication as
15 its first step. That would be an awfully high-risk
16 strategy and one that would seem to me
17 counterintuitive anyway.
18 So, I would be less concerned I think that
19 you are, Charles, at the dangers of that, because
20 you would only be going for a global pain
21 indication once you had received approval for
22 probably multiple other indications and had
23 reasonable level of experience.
24 So, I think we are sort of arguing about
25 something that is not likely to be even an early
140
1 step in drug development. Maybe I am wrong.
2 DR. FIRESTEIN: Why don't you go ahead and
3 respond, and then Dr. Ashburn, and then we will
4 probably move on.
5 DR. McLESKEY: Well, that is certainly a
6 presumption that you have made, and there is
7 actually a history, a recent history that global
8 claims have been achieved, maybe with hurdles not
9 quite so high, and again, obviously, the broader
10 the claim can be, the greater the incentive there
11 is for innovation from the sponsors.
12 All I am saying is that if we make the
13 hurdle quite high or, as you say, if we have to
14 incrementally approach it, the costs go up with
15 that approach, and the resistance to innovation
16 then may rise, which obviously, we don't want to
17 see happen, as well. We want to encourage
18 innovation.
19 DR. ASHBURN: I think the point that you
20 make is something that one needs to bring out,
21 flesh out a little bit more, and that is, is that
22 if you make a global claim too difficult, then
23 companies I think tend to go for a very narrow
24 focus or very narrow indication to get a product on
25 market with the expectation that that product for
141
1 pain will be used in a wider range of patients, so
2 as an off-label use, and that has a double-edged
3 sword, that if you make the bar too high, people go
4 for a narrow indication. Then, the medication will
5 be released, and then it will be used in patients
6 in whom it has not been studied.
7 Not only is that a problem with regard to
8 lack of good outcome data to guide clinical
9 judgment, but also has a problem with regard to
10 safety. That is one of the issues, trying to
11 strike a balance, so that you encourage people who
12 are developing products to widely study them the
13 medication, but not make the barrier so high that
14 they go for a narrow indication and actually
15 increase the risk of harm to patients once a drug
16 is released.
17 I also want to just re-flesh on the
18 outcome measurement, is that I think it is a
19 wonderful idea to include outcome measurement as a
20 part of the clinical trials for these products.
21 The concern that I have is that it is sending the
22 voice that positive benefit in all those different
23 fields are a requirement.
24 So, I think that tracking outcome
25 measurement can be a vital important required part,
142
1 but I visualize that data being used to guide the
2 development of the label rather than being a
3 primary indicator for approvability.
4 DR. FIRESTEIN: Well, now that we have
5 resolved this problem, I don't know that we
6 answered the questions that you raised in No. 3
7 here by providing you with a list of appropriate
8 models, but we did discuss No. 4 in some detail.
9 Again, just to reiterate, the notion is
10 that there are still very broad claims that would
11 still be available without a global pain
12 indication, is that correct?
13 DR. SIMON: Correct.
14 DR. FIRESTEIN: At this point, we will
15 move on to a discussion of back pain by Dr.
16 Borenstein.
17 Back Pain - Chronic Issues
18 David Borenstein, M.D.
19 DR. BORENSTEIN: I wanted to thank the
20 Advisory Committee and Lee Simon for asking me to
21 speak today. He said I should make it practical,
22 and I try to be a practical person, so hopefully,
23 what I will speak to you today about in regards to
24 back pain will, in fact, be practical.
25 It was one of the things I did want to
143
1 raise my hand and speak, but having the option of
2 actually being able to speak and having the
3 microphone allowed me to use it at this time.
4 [Slide.
5 I just wanted to give you a little
6 background about myself for those who may not know
7 me. I am from the George Washington University
8 Medical Center, not the other one across town. So,
9 if you want to find me, that is where you will find
10 me. I have been involved with low back pain in its
11 various forms, both on clinical trials and from the
12 standpoint of taking care of patients, I guess now
13 about 24 years, so I think I have some experience
14 at least in regards to low back pain.
15 [Slide.
16 When the Advisory Committee and Lee asked
17 me to speak, there were some issues that they
18 wanted me to discuss, so I thought I would sort of
19 put them out and say what they were in one form,
20 but what they also truly meant, and that was to
21 find the forms of chronic low back pain and its
22 prevalence.
23 What does that really mean? Is it
24 frequent enough and important enough to study? If
25 we have it, it's a problem that everyone talks
144
1 about, but is it really big enough a problem for
2 which it is worthwhile to actually look at?
3 Will patient selection including etiology
4 and severity influence the performance of drugs in
5 development? That means is it possible to identify
6 and separate the individuals who have back pain.
7 This may be all moot if we can't really
8 separate them out, they are just going to be one
9 group of people, then, we may just need to discuss
10 back pain, but there may be subgroups that we
11 really want to identify.
12 Which are the appropriate outcome
13 measures? That is, can improvements in back pain
14 be related to therapy, in other words, can it be
15 determined? If we have back pain patients and we
16 treat them, can we actually tell whether we do
17 anything for them?
18 [Slide.
19 4. Will a general indication be useful
20 for different labeling claims? I know Lee beat
21 this up already, somatic versus neuropathic versus
22 chronic headache. So, if you have someone who has
23 pain, it's in the low back, will it, in fact,
24 translate to them as far as their headache is
25 concerned, will there be some applicability?
145
1 Finally, with chronic low back pain, will
2 it serve as a measure for efficacy for a general
3 chronic pain indication, or should it remain
4 independent for a specific disease, exactly what we
5 have been discussing this morning?
6 I don't know if I have all the answers for
7 it, but I figured I would be discussing them and at
8 least I will give you my point of view.
9 [Slide.
10 So, what is chronic low back pain, what
11 does that mean, and what is its prevalence? How
12 often does it occur?
13 [Slide.
14 Well, in a lot of different studies, low
15 back pain is described as the pain that occurs in
16 the area with boundaries between the lowest and the
17 crease of the buttocks. So, when we talk about low
18 back pain, we are really not talking about leg
19 pain, we are not talking about sciatica, although
20 that is part of what we see with low back pain, so
21 depending upon how you define it, one can have a
22 wide variety of people.
23 If you just define chronic low back pain,
24 this would be the anatomic area that you might want
25 to study. That doesn't mean you wouldn't
146
1 necessarily study individuals with sciatica, but
2 that might be a special group.
3 [Slide.
4 What is chronic low back pain? It has a
5 duration. Duration may be as defined previously up
6 to three months, that is, up three months there is
7 this opportunity of having a repair, being in this
8 acute nociceptive stage, so that the body may heal
9 itself and then go back to its baseline state.
10 However, after possibly three months,
11 maybe sooner, this neuroplasticity has occurred and
12 thereby you are in a state where the nervous system
13 has had a response to this injury and you are now
14 in a chronic pain state.
15 Others have described chronic pain as pain
16 that persists longer than the expected period of
17 time for healing, so some people have described
18 chronic pain occurring within two days or two
19 weeks, not even waiting two months to be in a more
20 chronic stage because it is no longer in this acute
21 healing phase.
22 So, once again, these are at least two
23 different definitions that one might want to
24 describe in regard to chronic low back pain.
25 [Slide.
147
1 What is its epidemiology? Is it
2 worthwhile to study? Is it frequent enough, will
3 you find people who would want to be in clinical
4 trials because of this problem?
5 Well, 20 percent of the U.S. population
6 develops back pain yearly, so 1 out of 5 is a
7 potential candidate for a clinical trial. That
8 doesn't mean all of them have chronic low back
9 pain, but certainly 1 out of 5 do develop it.
10 Back pain is the second most common cause
11 of disability in the United States, and it is the
12 most common among men, accounting for 16.5 percent
13 total disabilities in individuals greater than 18
14 years of age in 1999. So, I propose that it is an
15 important problem. Not only does it cause pain,
16 but it also causes disability, and it's expensive.
17 If you look at Workers' Compensation
18 claims, which is far from all the individuals with
19 low back pain, from 1986 to 1996, during this
20 one-year period of time, 8.8 percent of the claims
21 were for back pain, but was up to almost 85 percent
22 of the costs.
23 So, having better therapies for low back
24 pain is important. Not only it a frequent problem,
25 but it also is potentially disabling and
148
1 significantly expensive.
2 [Slide.
3 So, there are at least reasons for which
4 having better therapies would result in betterment
5 to the individuals with it and society in general.
6 Is there a fiction as regards to how low
7 back pain does over time? In other words, the
8 usual story has been that most patients get better
9 within a two-month period, so we don't have very
10 many going on to a chronic phase.
11 [Slide.
12 Well, this study was done and reported in
13 the Annals of Rheumatic Disease back in 1998. This
14 was done in the Netherlands where they had
15 individuals who were a bit younger, those
16 individuals who we might want to think about being
17 involved with low back pain. They had about 450
18 individuals where they sent out postal
19 questionnaires over a 12-month period and followed
20 them over time to see what happened.
21 Most people, in fact, got better. The
22 median was about 7 weeks. However, still, at 3
23 months, 1 out of 3 still had back pain. You say,
24 well, they still might get better. If you want to
25 know whether these individuals will still be there
149
1 one year later, this study would suggest 1 out of
2 10.
3 So, with the individuals who have low back
4 pain, 1 out of 10 in general will still be having
5 it one year later. So, we do, in fact, have
6 individuals who are available to be studied. You
7 have, if you think of at least 2 percent, let's
8 say, of the U.S. population each year going into
9 the chronic back pain category.
10 [Slide.
11 Now, it's very funny to me, when people
12 ask me what is back pain, having written a 700-page
13 book on it, it is very difficult for me to answer,
14 and so when I hear people saying chronic back pain,
15 I just go twirling around saying which one do you
16 mean.
17 In my book that I wrote on this, we had 60
18 different reasons for developing the symptom, the
19 symptom of low back pain. Now, it takes a little
20 time to figure which disease is causing that, and
21 we will talk about whether we are good at that or
22 not, but this is one of the ways one might look at
23 the various categories with low back pain, whether
24 it's mechanical, rheumatologic, infectious,
25 psychiatric, so there are a wide variety of
150
1 disorders which can be associated with this
2 problem.
3 [Slide.
4 Now, let's simplify it a little bit. What
5 turns out to be the case is that the systemic, the
6 rheumatologic, the endocrinologic, the psychiatric
7 types of illness associated with low back pain are,
8 in fact, relatively few.
9 This is probably being generous on the low
10 side. Probably mechanical pain may be more like 90
11 or 95 percent of all the individuals looking at a
12 large enough population of individuals. So,
13 mechanical low back pain can be defined as one of
14 these various problems.
15 It can be associated with disorders
16 dealing with the muscle, ligaments, or tendons
17 which have been injured. It can be discogenic, it
18 can be the intervertebral disk which has been
19 affected, and that is a whole separate topic of
20 whether that causes pain or not, but can be
21 associated with a herniated disc which may also
22 result in a radiculopathy or sciatica.
23 There is also apophyseal joint disease,
24 and I am sure that Dr. Brandt would agree that
25 osteoarthritis affects the lumbar spine, so there
151
1 is some osteoarthritis there, as well. There is
2 spinal stenosis, spondylolysis, and
3 spondylolisthesis, which is an instability of the
4 spine, and then scoliosis can, in fact, be
5 associated with chronic low back pain.
6 These can all occur acutely. Some are
7 more associated with a more chronic situation. So,
8 you can have some that are acute and some, then,
9 that will go on to the chronic phase.
10 [Slide.
11 I would certainly like to hear what Dr.
12 Woolf has to say about my sources of pain as
13 regards to the lumbar spine. My suggestion is it
14 is once again complicated as to which structures
15 are being affected.
16 Superficial somatic, I love when comes in
17 as far as the back is concerned. I can pick up
18 herpes zoster and cellulitis pretty easily, and
19 that is easy to do. It gets more complicated the
20 deeper in the body you go, and that is why this is
21 so complicated because we are very good at
22 osteoarthritis of the fingers, but it becomes much
23 more difficult when it is osteoarthritis of the
24 zygo-apophyseal joints, because we can't get our
25 fingers around them. It becomes much more
152
1 difficult to diagnose.
2 So, deep somatic structures, such as the
3 muscles, the joints, the bursa, and fascia, also
4 have a characteristic kind of pain, which I would
5 propose is different than superficial somatic pain
6 in its character, in its clinical symptoms.
7 The same for radicular pain associated
8 with nerve root difficulties compared to visceral
9 referred pain mediated through sympathetic
10 afferents versus neurogenic pain, which may be more
11 of this diabetic neuropathy or amyotrophy,
12 psychogenic pain, which exists totally in the
13 cerebral cortex.
14 So, when you deal with low back pain,
15 depending upon which structures may be affected,
16 and which nerves may be affected, you can get a
17 different character of pain. I truly believe that
18 I can tell the difference between somatic and
19 radicular.
20 [Slide.
21 It was also suggested that we have
22 difficulty in deciding what pain intensity is, and
23 I was always quite interested in knowing what
24 minimal, mild, moderate, and severe was. Dr.
25 Simon's definition was he gets it as soon as he
153
1 even gets close to the dentist's office.
2 Well, this is the way I decide about it
3 because we don't have any specific machine that
4 measures it specifically. I do it on the basis of
5 function, and this is what I do in the office every
6 day.
7 Minimal is mentioned in passing and its
8 normal function. The person came in because they
9 had knee pain, but when you go through your total
10 review of systems, they mention that their back
11 bothered them once in a while.
12 Mild is a component of symptoms with mild
13 dysfunction. They are concerned that they are not
14 running as far as they used to because their back
15 bothers them. That is starting to concern them.
16 It doesn't bother them with the rest of their
17 activities, it is their recreational activities.
18 Moderate, it is getting in the way of what
19 they do with their work, it is becoming an impact
20 upon how they do their daily lives, and severe, the
21 point that Dr. Simon didn't tell you, is that he
22 brings his wife with him when he goes to the
23 dentist because he will need someone to help him
24 put on his clothes after he gets done.
25 That is the equivalent when I have someone
154
1 with severe back pain who comes to me, they come
2 with someone else, because they can't function to
3 put on their clothes to get in the office or to get
4 out of it.
5 So, there are ways of differentiating
6 among the various types of discomfort these
7 individuals experience.
8 [Slide.
9 The diagnosis of back pain is nonspecific
10 in 80 percent of patients. This is a dictum which
11 is repeated again and again and again, and it is
12 based upon some studies which been in the
13 literature for quite a long period of time, really
14 before there was an MRI or CT scan.
15 It is easier to just repeat it as to go
16 out and really find out if it's true or not, so it
17 is repeated and said most of the time you really
18 can't tell what is going on with these individuals.
19 That might be a problem if you were going to base a
20 whole indication on an entity which you really
21 couldn't diagnose, and I could understand why that
22 might be a problem.
23 [Slide.
24 Is that truly what happens? There was
25 just a very interesting paper, set of papers, which
155
1 appeared in the Archives of Internal Medicine just
2 in the last month. It came perfectly on time in
3 regard to this meeting.
4 Basically, it was a pro and con situation.
5 What the authors were saying is specific diagnosis
6 is possible or specific diagnosis is impossible.
7 On one side there was this physician, Dr.
8 Abraham, who raised the point that, in fact,
9 specific diagnoses are possible, that there are
10 clinical symptoms and signs associated with
11 differentiation of muscle, joint, and ligamentous
12 structures, that it is possible to, in fact,
13 differentiate mechanical versus systemic disorders,
14 that you can categorize these clinical symptoms,
15 that can be done, and that subtyping these
16 individuals does have the possibility of improving
17 therapy, that is, if you can separate the specific
18 mechanisms either or pain generators or the nerves
19 that are mediating it, might it be possible to get
20 a better therapy because you could identify them.
21 [Slide.
22 On the other side was Rich Deyo, and he
23 has been long known for being of the school that
24 you really can't make diagnoses. His point,
25 however, his not hidden agenda, quite clear agenda,
156
1 he is concerned about individuals utilizing health
2 services to make diagnoses, which don't really make
3 a difference, so people doing MRI's and x-rays, and
4 all this.
5 His point is specific diagnosis is
6 impossible. You can find anatomic abnormalities
7 in asymptomatic individuals. This will result in
8 overutilization of imaging techniques. There is
9 inconsistency with physical findings. In general,
10 if we look at it, nonspecific therapy works,
11 nonsteroidals can work in a wide variety of things,
12 so if they do, why bother to try and find the
13 specific pain generator, they work in general.
14 My point is probably a mixture of both. I
15 think both have points to be made for their side.
16 I think it is possible to separate these
17 individuals a bit better, and I think even Dr. Deyo
18 in his response said yes, it probably is
19 recognizing that his concern was about utilization,
20 and not the fact that you couldn't diagnose some of
21 these more specific problems.
22 So, I do think it is possible, but until
23 we categorize and study a bit more specifically, we
24 may not be able to come up with better therapies,
25 and that is part of what this group needs to
157
1 decide, is that worthwhile, and that is what the
2 committee will have to sort of deal with.
3 [Slide.
4 Also, is it possible to differentiate
5 among these various types of problems, can you do
6 the difference between somatic, neuropathic, and
7 radicular pains? Yes, they can be differentiated,
8 and specific pain generators are difficult to
9 identify, but localization is not essential for
10 effective therapy.
11 So, my point would be this, that it is
12 possible to categorize some of these individuals
13 with chronic low back pain, you can put them in
14 broad categories, and then you can study them to
15 see, in fact, they are responsive to different
16 types of therapies.
17 I think it is important to try and
18 separate somatic versus radicular, but that doesn't
19 mean they should be mutually exclusive, and some
20 therapies may, in fact, work in both areas.
21 [Slide.
22 Now, as my third point, are there pain
23 outcome measures or low back pain measures which
24 have been shown to be effective in picking up
25 differences? Now, Dr. Strand is I am sure going to
158
1 do an excellent job talking about this tomorrow,
2 and I am not stealing any of her thunder at all,
3 because I am just going to go into this for a
4 minute or two, because I don't want to tread too
5 far afield.
6 But I do believe, at least as part of our
7 discussion, do we have these outcome measures, do
8 we have back-specific function measures, do we have
9 pain measures, and do we have patient global
10 satisfaction measures that make a difference?
11 [Slide.
12 Well, back-specific function measures do
13 exist, and these have been tested for a long period
14 of time. They are the Roland Morris Disability
15 Questionnaire and the Oswestry Disability Index.
16 [Slide.
17 For those who may not be aware of them, I
18 am just going to take one minute to just describe
19 them to show you that they do, in fact, exist, they
20 do function assessments as a means of telling how
21 back pain patients are functioning and how they are
22 doing.
23 There are 24 items from the Sickness
24 Impact Profile. The functions that they pulled out
25 affect back pain that day. The scores are added,
159
1 and this has been a validated and reproducible
2 instrument for a number of years since it first
3 came out in 1983, and has been associated with
4 picking up differences and improvements in patients
5 with low back pain on a function basis.
6 [Slide.
7 Then, we have individuals who have been
8 measured with the Oswestry Disability Index, and
9 this is also a pain and function assessment. There
10 are 10 sections on various functions with 6 levels
11 of assessment in each.
12 They measure physical and social functions
13 that day. They can once again be added up to 100,
14 and have been validated and are reproducible
15 instruments, as well. So, from the standpoint of
16 function, we certainly have capabilities.
17 [Slide.
18 In regards to pain assessments, I will
19 leave it once again to others to describe whether
20 these are the appropriate ones or whether there are
21 others that are better in describing specific
22 different types of pain.
23 One may have a general type of pain
24 assessment tool, and if you have a specific
25 character of pain, a neuropathic pain, or another
160
1 type of pain, one might use that specific tool, as
2 well, in that specific circumstance.
3 [Slide.
4 Then, in regards to global satisfaction, I
5 would ask this group to strongly believe that a
6 question to the patient asking how are you doing
7 and are you doing better is a worthwhile outcome,
8 and should always be, period, case closed.
9 It doesn't take too long to ask, it takes
10 very little time to circle, but that is what I ask
11 every day, and you can do it with smiley faces, you
12 can do whichever which way you want, but that is
13 what the patient cares around, do I feel better all
14 over, and what was said in regards to toxicities
15 and frequency of dosing and everything else all
16 gets wrapped up into the way the patient feels.
17 So, I think whether they are satisfied
18 with their therapy, very much, a little, mixed
19 reviews, or I really hate it, really does get to a
20 significant outcome as far as these studies are
21 concerned, and I think it is a very simple question
22 to ask, but a very important piece of information
23 to know.
24 [Slide.
25 Then, of course, optional measures are
161
1 also possible depending upon whether you think
2 there is depression associated with these
3 individuals with chronic pain. There is the
4 general health status circumstance with SF-36 and
5 various depression scales, I just picked out one.
6 This could be optional if you think
7 depression is playing a significant role in regards
8 to these chronic back pain patients.
9 [Slide.
10 So, I do believe there are instruments
11 that exist that measure the effect of drug
12 interventions on chronic pain for function, pain,
13 global satisfaction, and for general health status.
14 [Slide.
15 Now, what was mentioned also is quite
16 clear, that is, chronic pain therapy is
17 multimodality. Depending upon how long it has been
18 present, one may use one drug, two drugs, three
19 drugs, four drugs. One may use a variety of other
20 physical modalities, physical therapy, exercises, a
21 wide range of things in order to take care of back
22 pain.
23 I am not sure how one wants to deal with
24 that in saying they need to be additive or have a
25 baseline state and then take one aspect away and
162
1 seeing if substitution makes a difference, either
2 making the patient go back to their baseline state
3 or, in fact, improve upon their baseline state.
4 So, these are some of the therapies that
5 are available as far as back pain is concerned.
6 [Slide.
7 These are the therapies, the drug
8 therapies associated with low back pain. I want
9 you to know that I looked in the PDR to see if one
10 had an indication for chronic low back pain. None.
11 So every day that I work in the office, I have no
12 indication for any of the drugs that I am using.
13 I feel comfort with that, but uneasy. I
14 have to tell my patient if they are smart enough or
15 willing enough to ask me is this indicated for
16 this, the answer is not specifically, but I think
17 you have a problem that will respond to this.
18 So, here is a wide range. This isn't my
19 list, this is culled from a number of different
20 papers and studies looking at what has been
21 effective as far as chronic low back pain occurred.
22 This has been nonsteroidals, muscle relaxants,
23 analgesics, antidepressants, anticonvulsants,
24 alpha-2 adrenergic agonists, and a miscellaneous
25 group including the NUDA receptor antagonists.
163
1 [Slide.
2 I am not going to go through all of these.
3 Certainly many of you know them already. There are
4 the nonsteroidals. This was recently reviewed in
5 Spine in 2000, suggesting that these medications,
6 in fact, do have benefits as far as chronic low
7 back pain is concerned.
8 The ones that are short-lived, have short
9 half-lifes, they can be used for the acute
10 exacerbations that Dr. Sherrer was talking about,
11 that if you have someone who has a baseline state,
12 but has an acute exacerbation, one can use a short
13 half-life nonsteroidal, long, sustained effects for
14 long half-life medications, and certainly from the
15 standpoint of COX-2 inhibitors, decreased toxicity
16 because the people will be on drugs for extended
17 periods of time is certainly an important
18 indication and concern, that it may be good for a
19 week or two, but when you are talking about one or
20 two years, it is still going to be safe.
21 I am not suggesting that one needs to
22 study it that long a period of time, but there are
23 patients who are on these drugs for extended
24 periods of time, so toxicity is something I am
25 concerned about when I start these patients, but I
164
1 don't really know how long they are going to end up
2 on them, but if they work, I keep using them.
3 [Slide.
4 Then, there are, of course, the muscle
5 relaxants as they have been described previously,
6 and these are important adjuncts to therapy. If
7 you wanted to see the effect of any one of these
8 for longer than six months, I couldn't show you a
9 study that really did that on any regular basis.
10 [Slide.
11 Non-narcotic and narcotic medicines are
12 all used in patients who have chronic low back pain
13 depending upon their status.
14 [Slide.
15 I am almost out of time, but I wanted to
16 be practical. We have been very much talking about
17 mechanisms and all. I deal with patients just like
18 many of you, and I thought what I would do to end
19 up my discussion today is live my life.
20 You have a few patients with chronic low
21 back pain. This is what they are getting. This is
22 a 52-year-old person who had a work-related
23 myofascial injury in the lumbar spine. It is mild
24 to moderate, she is still able to function. We
25 changed her nonsteroidal to a diclofenac product.
165
1 She remained on her muscle relaxant when she has an
2 acute exacerbation, so she can stay at work. She
3 knows that she can dose with an extra short form of
4 the medicine, and she knows that she is supposed to
5 be on her exercise program in order to maintain her
6 function.
7 [Slide.
8 There is a 67-year-old person who has
9 facet joint disease, has basically osteoarthritis
10 as part of their chronic low back pain. This
11 individual is treated with a COX-2 inhibitor and a
12 muscle relaxant, and has been on this regimen for
13 an extended period of time.
14 This, I would say was the mild to moderate
15 chronic somatic type of pain.
16 [Slide.
17 Then, I have another individual who has
18 had a laminectomy, some of these are post-surgical
19 individuals, who happens to have a fractured screw
20 in his back, but he doesn't really want to get it
21 taken out.
22 So, this individual, over time, and I have
23 been taking care of him over 10 years, has gone
24 through a variety of therapies now where he is now
25 currently on a COX-2 inhibitor nortriptyline, a
166
1 fentanyl patch, and a short-acting narcotic when he
2 has his acute exacerbations.
3 [Slide.
4 Then, finally, for the individual who has
5 moderate to severe neuropathic pain, who is still
6 in this chronic back pain situation since he has a
7 component of pain, he has had a traumatic
8 neuropathy to the sciatic nerve.
9 He is on a long-acting nonsteroidal,
10 gabapentin, oxycodone, long acting, and short-term
11 narcotic for when he has an exacerbation.
12 That is what chronic low back pain therapy
13 can be depending upon who you are seeing and what
14 kind of status they are in. I do believe it is
15 possible to separate these individuals out. Many
16 of these individuals have been on a variety of
17 therapies for an extended period of time.
18 [Slide.
19 So, I would like to conclude with this and
20 hopefully have answered some of these questions,
21 but probably have raised more. I do think that
22 chronic low back pain is a model for chronic pain.
23 I think it is an important problem.
24 I think there are enough people in the
25 society for which it is worthy of being
167
1 investigated. There are outcome tools available I
2 think at this time that can at least give us a
3 handle as to how to measure it, but certainly
4 others, as they are developed, would be useful.
5 Somatic pain is identifiable, that is,
6 pain related to musculoskeletal disorders, and for
7 terms if you don't like somatic, but prefer
8 musculoskeletal system, would be where I would put
9 that, are identifiable and can be seen and studied.
10 The degree of pain and effect of study
11 design I think is also possibly differentiated.
12 For those who have mild to moderate pain, it might
13 be possible to do a single drug versus placebo with
14 an active comparator, however, when you have these
15 individuals who have more severe pain where there
16 may be more mechanisms involved, there, you may
17 have individuals who may be on a stable multidrug,
18 multimodality therapy, but there, take the drug
19 away, have them flare, and then replace it with the
20 study agent and thereby be able to determine
21 whether they did better or worse from their
22 baseline state.
23 That is where I will conclude. Thank you
24 very much for your attention.
25 DR. FIRESTEIN: Thank you very much. We
168
1 have about 10 minutes to discuss Point No. 5, which
2 is to comment on the value of chronic low back pain
3 as a separate labeled indication versus part of a
4 broader claim.
5 Discussion Point #5
6 DR. MAX: A question for Dr. Borenstein.
7 One big distinction that seems to come out of your
8 talk is the distinction between people who have low
9 back pain every day for a year or two years and
10 those who are having clear-cut, new injury, where
11 perhaps the disc is getting another little tear,
12 and all the studies, like the postcard study you
13 show, had people with new relapses.
14 Do you think it would be appropriate in
15 clinical trials to make some sort of distinction
16 between these people who probably have some acute
17 inflammatory pain on top of it, which might respond
18 to different drugs and how would you do it?
19 DR. BORENSTEIN: Well, I do think it is
20 possible to separate these individuals out. Some
21 people have a chronic ongoing back pain, which it
22 may vary a little bit, but is essentially there for
23 extended periods of time. We are talking months
24 and months and months.
25 There are other individuals who have
169
1 exacerbations of their pain, they wax and wane.
2 Those individuals do have a different kind of
3 story. Some of those may think it is the weather
4 that bothers them or certain activities that will
5 have an effect upon their pain.
6 So, I do think it is possible through the
7 appropriate questions at the start of such a study
8 to differentiate from these individuals who has a
9 chronic stable type of pain versus those who are
10 having acute exacerbations, which may have more an
11 inflammatory component.
12 DR. MAX: Has the methodology been
13 developed yet in any of the published clinical
14 trials of back pain to distinguish these two
15 classes?
16 DR. BORENSTEIN: Well, as I tried to show,
17 there is great debate about whether one can define
18 or describe low back pain, and this has just been
19 written about last month. I think if people do take
20 care of back pain patients, you can separate these
21 individuals out.
22 There are a certain criteria where one
23 might say their level of pain has remained at a
24 certain level for a period of time. So, I do
25 believe that it is possible to separate them out,
170
1 but has it been studied specifically as to which
2 group this may be, whether it is osteoarthritis
3 with more a flare component? No, that hasn't been
4 done.
5 DR. FIRESTEIN: Dr. Sherrer.
6 DR. SHERRER: I think Dr. Borenstein has
7 shown that low back pain has all the general
8 problems that chronic pain has in general, and I
9 don't think it is going to offer us anything
10 specific.
11 You pointed out that you have
12 osteoarthritis affecting the low back, you have
13 inflammatory joint disease affecting the low back,
14 you have soft tissue pain affecting the low back,
15 and I think we see that clinically.
16 Then, you have the chronic persistent
17 pain, the chronic intermittent pain. It is the
18 same thing we see with chronic pain elsewhere. So,
19 I don't see that separating low back pain out per
20 se is going to be beneficial unless we are going to
21 be able to separate out inflammatory low back pain
22 or osteoarthritic low back pain.
23 DR. BORENSTEIN: My suggestion would be
24 that we could. If you have a sed rate greater than
25 20, you have an inflammatory process which
171
1 separates out most, I do think it is possible to
2 separate out those individuals who have
3 inflammatory back pain.
4 I think we can separate out the
5 spondyloarthropathies. Those people have
6 infections, and all those. I would not suggest
7 that it is so difficult to do. I think it is
8 clearly possible to identify those individuals who
9 have mechanical pain.
10 Now, if you want to separate out those who
11 have it solely on muscle discomfort versus joint
12 discomfort, it may become a bit more difficult, but
13 from the standpoint of an inflammatory versus
14 non-inflammatory standpoint, I think that is not a
15 difficult process to undergo.
16 DR. GOLDKIND: I would like to ask Dr.
17 Borenstein what evidentiary base are you familiar
18 with that speaks to the polypharmacy, not
19 surprising at all, but striking how patients with
20 chronic low back pain, and that is probably going
21 to be true in other chronic pain situations, are on
22 polypharmacy, but is it fully anecdotal or do you
23 see studies that incorporate that aspect.
24 DR. BORENSTEIN: Most of them are
25 anecdotal. I mean it becomes most of the way drug
172
1 studies are done for the most part except in
2 rheumatoid arthritis, and they haven't necessarily
3 been transposed into chronic low back pain, is that
4 you have a stable therapy, which can be on a wide
5 variety of drugs, and then you take one drug away.
6 This, I do not believe has been
7 specifically done in chronic back pain patients who
8 are on more than one drug. That is the problem that
9 we face. If this was thought to be a good
10 process, then, in fact, that could be done, but
11 that is the way some of these patients with chronic
12 back pain need to be treated.
13 Some, in fact, can be treated with one or
14 two drugs. Others with more severe pain are
15 treated with multiple drugs.
16 DR. GOLDKIND: Do you think there would be
17 any value in studying specifically combinations,
18 how we put drugs together, or does the current
19 clinical trial design where there is background
20 that includes the remainder suffice for clinical
21 practice?
22 DR. BORENSTEIN: Well, getting back to
23 what Dr. Woolf was talking about before, this may
24 be one of the ways of trying, in fact, to identify
25 those individuals. Just hypothetically, you have a
173
1 group of people who are on a nonsteroidal, muscle
2 relaxant, a tricyclic. You have three drugs.
3 You come along and find out that you
4 intervened with one, you take one of those out and
5 intervene and find a subgroup of people who have a
6 specific response, this might be interesting in
7 identifying those individuals who have a response
8 to that specific group, because it is going to be
9 very hard to find these people who have chronic
10 back pain, who are going to be on placebo versus
11 the active comparator, and nothing else.
12 So, I think this may be one of the ways of
13 getting those drug trials done and also identifying
14 those individuals who may be doing subgroup
15 analysis to see how they may have responded above
16 and beyond what the mean might have been otherwise
17 to get at some of these mechanism problems.
18 DR. FIRESTEIN: Dr. Cush.
19 DR. CUSH: I think from Dr. Borenstein's
20 comments we should be very concerned that despite
21 the prevalence of the condition, the number of
22 agents which have targeted back pain for an
23 indication are very few, and that is surprising and
24 disappointing.
25 I think that the FDA should make an effort
174
1 to try to make this an indication if, on one hand,
2 to spur and excite research in this area as an
3 indication, but obviously, this was always out
4 there and people could have gone after it, and
5 companies may have stayed clear of low back pain as
6 an indication for a variety of reasons, maybe the
7 difficulty in studying patients, the outcome
8 measures, and whatnot, but this is an inherent
9 problem in there and maybe the FDA can look forward
10 to try to develop ways of pushing people in this
11 direction as far as research and clinical trials.
12 One way might be for that global
13 indication that we argued about in the last
14 session, maybe one of the defining diseases under
15 that heading might be low back pain.
16 MS. McBRAIR: As the consumer rep, I just
17 wanted to thank Dr. Borenstein for supporting the
18 idea of studying patient function, their patient
19 global assessment, and possibly quality of life.
20 People can have a lot of pain medication
21 and pain control, and not be able to function very
22 well, as oftentimes noticed by employers and
23 families and others, and I think we need to take a
24 clear look at what we are doing to people when we
25 offer them all these medications.
175
1 DR. FIRESTEIN: Dr. Brandt.
2 DR. BRANDT: Polypharmacy is not unique to
3 low back pain. It may be a general phenomenon in
4 patients who have chronic pain. In osteoarthritis,
5 those people who are given a prescription NSAID, a
6 very significant proportion are taking also an
7 over-the-counter NSAID and acetaminophen and
8 glucosamine. So, it is not unique to low back.
9 DR. BORENSTEIN: If I could just comment
10 on that. Once again, although ideally from a
11 scientific basis, it is nice to think of nice
12 straight lines as the only way things happen, but
13 dealing with human beings, they always find ways of
14 making the lines curve.
15 I have never seen a straight one yet, and
16 there is always a little bit of everything, and the
17 trouble that we have is trying to identify those
18 people and how they verge away from this line, this
19 straight line, where the curves come in.
20 So, that is why I was saying polypharmacy,
21 yes, there may be this peripheral sensitization and
22 other things playing a role, as well as
23 nociception. Some people may have some of both,
24 and that even though it may be what we would
25 expect, where a COX-2 or a nonsteroidal may have no
176
1 effect, in certain circumstances, they do seem to,
2 and so we are always surprised, we are always happy
3 it happens, but I can't really always explain it.
4 So, though knowing the basic science is
5 clearly essential, and the better we get at it, the
6 better we will be able to have therapies. At this
7 point, I still think that we still have to use a
8 little bit more leeway in the way we actually use
9 these drugs to try and maximize the effect in our
10 patients.
11 That is once again the basic goal is to
12 make the patients better. The science will catch
13 up with the human beings as they tell us how they
14 are doing.
15 DR. FIRESTEIN: One of the problems with
16 looking at low back pain as a single entity is that
17 it becomes difficult to manage them with an
18 individual agent for diseases, that has multiple
19 etiologies, just as we don't have a single
20 indication for heart disease, for instance, but we
21 wouldn't necessarily even desire a single
22 indication for low back pain, which is in part
23 caused by osteoarthritis or other mechanical
24 derangements, and the like, or neuropathic
25 diseases.
177
1 I wonder if we would be doing the patients
2 a service or a disservice by lumping all those
3 patients together rather than trying to be more
4 specific in targeting our approaches.
5 For instance, you already mentioned that
6 90 percent of the patients have mechanical issues,
7 and that might be one way of at least getting one's
8 arms around the indication rather than just trying
9 to include all back pain.
10 DR. BORENSTEIN: My response with that
11 would be exactly that. I think you can separate
12 out these individuals who have musculoskeletal
13 versus the systemic illnesses, and make a
14 difference for those individuals.
15 It becomes difficult to say that it is
16 only joint, and not muscle, because you can get
17 referred pain, as well, so if you are able to deal
18 with that process and make a difference, even
19 though it may be muscle first and joint second, or
20 joint first or and muscle second, you can still
21 make an impact in this musculoskeletal arena.
22 DR. FIRESTEIN: Dr. Katz, and then we will
23 finish up.
24 DR. KATZ: I would like to come down on
25 the side of an entity of chronic low back pain.
178
1 Again, as Dr. Borenstein did, we are talking about
2 non-neuropathic low back pain, we are talking about
3 eliminating systemic diseases, but I would be in
4 favor of that being an indication unto itself and
5 also that being a disease model that could be used
6 to work towards a musculoskeletal claim.
7 All of these diseases can be split
8 infinitely into different subgroups that may
9 respond more or less well. We just heard earlier
10 that hypertension is actually a number of different
11 diseases that respond differently, but nobody is
12 bothered by the idea of having a drug for
13 hypertension, diabetic neuropathy, it is the same.
14 Postherpetic neuralgia, it doesn't bother
15 us to approve a drug that works at best in a third
16 or 40 percent of patients, knowing that our
17 approval only applies to a subgroup, but knowing
18 equally well that because we don't know how to
19 segregate out that subgroup, we need to provide a
20 physician a reason to use the medication.
21 We also know that much more harm has been
22 done by under-recognition and undertreatment of
23 chronic pain than by overtreatment, so if we had to
24 come down on which side we would want to occur, I
25 would prefer to err on the side of seducing
179
1 physicians into treating their patients.
2 The fact that the chronic low back pain,
3 even musculoskeletal pain is somewhat
4 heterogeneous, I think is a strength in the sense
5 that if you can show in a good trial that your
6 medication works for this admittedly heterogeneous
7 group of disorders, then, all the more it should be
8 applicable to a broader musculoskeletal pain
9 diagnosis where its heterogeneity is actually a
10 strength, and not a weakness.
11 DR. FIRESTEIN: I would agree with that as
12 long as we are primarily discussing mechanical back
13 pain.
14 DR. KATZ: Yes, as Dr. Borenstein defined
15 it.
16 DR. FIRESTEIN: That brings us to the end
17 of this morning's session and we will reassemble at
18 1 o'clock. Thank you.
19 [Whereupon, at 12:02 p.m., the proceedings
20 were recessed, to be resumed at 1:00 p.m.]
180
1 A F T E R N O O N P R O C E E D I N G S
2 [1:05 p.m.]
3 DR. FIRESTEIN: The next segment is the
4 open public hearing, which involved a number of
5 individuals who will make short presentations from
6 5 to 10 minutes.
7 For those individuals that will have
8 PowerPoint slides, I would ask you to make your
9 presentation from up here if you already have them
10 loaded onto the computer, and if you haven't
11 already done that, then, you will not be making
12 slide presentations unless you are very fast.
13 Again, the various individuals have
14 already been apprised of the time limitations and
15 while we don't have a gong up here to cut them off,
16 I would ask them, please, to try to adhere to them
17 as closely as possible.
18 The first is Dr. Najib Babul, Chief
19 Scientific Officer of TheraQuest Biosciences.
20 Open Public Hearing
21 DR. BABUL: Good afternoon. I want to
22 thank the Advisory Committee Chair and the Division
23 Director for allowing me to speak at this meeting.
24 I am particularly pleased to speak at this meeting
25 because I think the briefing document raises a
181
1 number of important issues both to regulators and
2 to drug development scientists.
3 [Slide.
4 Let me just introduce myself briefly. My
5 name is Najib Babul. I am Chief Scientific Officer
6 for TheraQuest Biosciences, a Philadelphia,
7 Pennsylvania based company, consulting in the area
8 of analgesia rheumatology drug development.
9 [Slide.
10 This is my conflict of interest statement,
11 pharmaceutical sponsors that I work with, have
12 submissions or pending submissions before Division
13 550 or Division 170, and the views that I express
14 are solely those of TheraQuest Biosciences.
15 [Slide.
16 Much has been said earlier today about the
17 regulatory framework for approval of drugs and what
18 is lacking in the existing guidelines. Certainly I
19 can tell you as somebody who has been using the
20 1992 Analgesic Guidelines that before we throw the
21 baby away with the bathwater, these guidelines are
22 fairly robust and certainly help those of us who
23 are in the trenches and developing drugs for acute
24 pain, these guidelines have been exceedingly useful
25 and continue to be useful.
182
1 That is not to say that we don't presently
2 have challenges in drug development. In addition
3 to these FDA guidelines, there is the CPMP
4 document, draft document, which also provides
5 additional guidance to those of us who are doing
6 international clinical trials in analgesia.
7 Of course, we have rather well put
8 together OA guidance document from the FDA and from
9 the CPMP, which can provide a bit of a foundation
10 for going forward if a decision is made to put
11 together additional guidance documents.
12 [Slide.
13 Now, having said that, there are certainly
14 gaps in the regulatory framework for development of
15 analgesics. This is a short list of some of the
16 gaps as I see them, and the include multi-dose
17 evaluation in acute pain, evaluation of drugs with
18 slow onset of effect in acute pain, and there are
19 clearly some drugs, including drugs that have a
20 depot effect, that provide sustained analgesia in
21 the perioperative period, for instance, that would
22 fit into that category.
23 Drugs for neuropathic pain, which was the
24 subject of a separate Advisory Committee meeting in
25 May, drugs for cancer pain, which perhaps fit in
183
1 the mandate of this Advisory Committee, and, of
2 course, the possibility of putting together
3 guidance documents for low back pain, for
4 fibromyalgia, and for myofascial pain, then,
5 broadly speaking, looking at chronic pain as an
6 indication.
7 [Slide.
8 This is a brief list of some of the models
9 of chronic pain. One can categorize chronic pain
10 in a number of different ways - mechanistically, by
11 diagnosis, etiology, et cetera, and this is an
12 attempt at categorizing some of the models,
13 myofascial pain, low back pain, osteoarthritis,
14 fibromyalgia, some have argued and actually
15 demonstrated successfully that mixed model
16 populations with chronic non-cancer pain can be
17 successfully evaluated as a heterogeneous
18 population, and then, of course, neuropathic
19 chronic pain and cancer pain.
20 [Slide.
21 Now, there are some compelling reasons why
22 we have lagged behind in chronic pain in contrast
23 to acute pain in developing guidelines and in
24 developing drugs and getting a label claim. I
25 should note that there is a bit of a divergence in
184
1 terms of the labeling history for opioids,
2 particularly sustained release opioids in the
3 Division of Anesthetic Critical Care, Addiction
4 Drug Products, where there has been a de facto
5 chronic pain indication without stating chronic
6 pain, and in the Anti-inflammatory, Ophthalmic,
7 Analgesics Group where, in fact, that indication I
8 don't believe has broadly existed although clearly
9 there are some drugs historically that have been
10 approved for the treatment of moderate to severe
11 pain implying acute and chronic.
12 Now, some of the challenges that drug
13 developers like myself find in developing drugs for
14 chronic pain is that the etiology is rather
15 diverse. Dr. Borenstein earlier talked about I
16 think 60 or 70 potential etiologies for low back
17 pain alone, so certainly even with a heterogeneous,
18 a seemingly identical "diagnosis," broadly
19 speaking, or presenting a complaint like low back
20 pain, you can have a rather heterogeneous
21 population.
22 Having said that, perhaps much more is
23 made of that than we ought to. A substantial
24 number of patients, as Dr. Borenstein noted, have
25 mechanical low back pain, and while there is some
185
1 disagreement, many have argued that as many as 90
2 percent of patients with low back pain have
3 idiopathic low back pain, and there is now
4 pharmacologic evidence from work that has been done
5 demonstrating that that group, whether it is
6 homogenous or heterogeneous, in fact, is a
7 worthwhile group to evaluate analgesics in, and we
8 have certainly been able to separate active from
9 placebo.
10 In addition, these patients have
11 considerable amount of psychological overlay which
12 varies a great deal from patients who may have some
13 myofascial pain post-motor vehicle accidents to
14 patients with osteoarthritis who may have
15 considerably less access to diagnosis.
16 We also have a situation that is
17 confounded by disability payments and litigation
18 and secondary gain issues which make it very
19 difficult for us to look at issues of function, for
20 instance, in this population.
21 I think, finally, there are unrealistic
22 outcome expectations. There are a number of
23 stakeholders in this debate, not just drug
24 developers and regulators. In fact, insurance
25 companies and other third parties sometimes view a
186
1 successful outcome not as relief of pain per see,
2 but a return to work situation, which of course
3 means that their exposure to liability, financial
4 liability is significantly reduced.
5 [Slide.
6 Recently, Division 550 has suggested that
7 replicate evidence in three chronic pain states or
8 chronic pain models are necessary for a chronic
9 pain indication. While I appreciate that the brief
10 document suggests to the committee that this is
11 subject to consideration and some debate, and that
12 it is not cast in stone, I think, Dr. Simon, you
13 referred to this as an iterative process, there are
14 some potential implications that I think we need to
15 consider.
16 [Slide.
17 I think the first issue that concerns me
18 is that there may be an absence of established
19 models to provide evidence in three chronic pain
20 states. While one can throw fibromyalgia into this
21 chronic pain basket, some would argue that, in
22 fact, it is a rather distinct entity and that it
23 may not respond to many of the agents that other
24 drugs perhaps respond to in chronic pain.
25 I think certainly the suggestion contained
187
1 in the briefing document and indeed at the NIH-FDA
2 Workshop, that replicate evidence for a specific
3 sub-indication would be a basis for approval is
4 reasonable. I think that very few would disagree
5 that at least in a 505(b)(1)/ new chemical entity
6 approval strategy, that if you are going to go for
7 a specific sub-indication, that perhaps some degree
8 of replication is necessary.
9 However, I would suggest to the division
10 and to the committee that replication across three
11 models, models, which we have yet to fully
12 establish and validate, might be too onerous a
13 requirement to put on the pharmaceutical sponsors,
14 and that perhaps, and this is a suggestion for
15 potential discussion by the committee and by the
16 division, that perhaps replication in two models of
17 chronic pain or perhaps robust and internally
18 consistent evidence in single trials in three
19 models might be sufficient to provide a broad
20 indication of chronic pain with the proviso that
21 the Clinical Pharmacology Section of the package
22 insert would speak to the specific evidence that is
23 available on that drug.
24 One concern that a number of us interested
25 in chronic pain have is that if the burden is too
188
1 high for a broad indication, we may end up people
2 being expeditious, and there was some reference to
3 this earlier, people just taking the quick and
4 dirty route out, just getting a specific narrow
5 sub-indication with the potential for substantial
6 off-label use and orphaning of other indications
7 for evaluation purposes.
8 [Slide.
9 There are a number of additional issues
10 which I would like to just very briefly address.
11 In the briefing document, there is reference to the
12 use of co-primary endpoints. Indeed, pain function
13 and patient global are important endpoints. There
14 is little debate on this issue, and I believe Dr.
15 Strand at the NIH-FDA Workshop led a breakout
16 session on this particular issue, and there was
17 general consensus that these are important
18 endpoints.
19 There is indeed some precedents at least
20 at the division in terms of for OA, in terms of
21 having a win on three co-primaries, however, it
22 does increase the statistical burden required for
23 approval, and I think that for function, function
24 the way it is viewed through self-reports, we need
25 to be careful that we define function carefully
189
1 because function, the way it is viewed, say, in OA,
2 using WOMAC as an instrument, is very different
3 than the way function is viewed by pain physicians.
4 So, before talking about function as a
5 self-report, perhaps that may be achievable,
6 although I am not certain about that, in all
7 chronic pain states. Certainly, we don't ask that
8 in depression, we don't ask that in migraine in
9 terms of return to work or restoration of function
10 per se, and that it may be too unrealistic a
11 pharmacologic expectation to put on what is really
12 a complex disorder, and I would ask the Division
13 and the Advisory Board to consider this.
14 DR. FIRESTEIN: Dr. Babul, could you wrap
15 up, please.
16 DR. BABUL: I am pleased to note the
17 Division was prepared to consider placebo versus
18 active control, that have some assay sensitivity.
19 I would urge the Division to consider some
20 clinometric flexibility, so that we don't have
21 ossification of trial design methodology. Finally,
22 I would suggest that we need some degree of
23 harmonization to the extent we can between Division
24 170 and Division 550 as we go forward in terms of
25 approaches that are acceptable for opioids and for
190
1 non-opioid analgesics.
2 Thank you.
3 DR. FIRESTEIN: Thank you.
4 The next speaker is Dr. Kenneth Verburg,
5 Vice President, Clinical Research, Pharmacia.
6 DR. VERBURG: Good afternoon. My name is
7 Ken Verburg. I am here representing Pharmacia
8 Corporation today. We appreciate the opportunity
9 to contribute to the meeting.
10 [Slide.
11 I would like to limit my comments and my
12 brief presentation today to just some general
13 observations about the development of new
14 guidelines for analgesics or drugs intended for the
15 treatment of pain, and then focus on some
16 observations specifically directed towards chronic
17 pain and acute pain.
18 [Slide.
19 As we heard this morning, it makes at
20 least some sense based on the information we have
21 at hand to set up and use a mechanistic basis as a
22 framework at least for the indications or the way
23 that we think the indications should be laid out.
24 This would lead to using this particular
25 mind-set, an easy separation if you will, of
191
1 nociceptive and neuropathic pain, but a subdivision
2 of nociceptic pain into somatic and visceral pain
3 is not quite so clear with both being acute and
4 chronic, and substantial overlap between the two
5 conditions.
6 Also, we could use a mechanistic basis, as
7 we have heard this morning, about the chronicity of
8 pain, separating out acute and chronic pain into
9 separate indications, and not necessarily having to
10 have or having to demonstrate acute pain a priori
11 before getting an indication for chronic pain.
12 We could also use mechanisms to gauge pain
13 severity. Particularly here, I think, we are most
14 interested in the differences across models and how
15 that might translate to effective regimens, and
16 finally, in terms of just general overall
17 considerations, a notion or a realization that
18 there are different classes of analgesics that may
19 be effective as either monotherapy or multimodal
20 therapy under certain conditions in the particular
21 sites of action.
22 [Slide.
23 We would also encourage that the
24 development programs expedite therapies to meet the
25 clear unmet medical need in this particular area,
192
1 that efficient programs are set up that provide the
2 information that is needed for registration, cut
3 down on the white space with the gray area that us,
4 as sponsors, sometimes confront, and also that we
5 consider conditions of clinical practice, that
6 being preoperative administration or preemptive
7 administration and/or postoperative administration,
8 multimodal analgesic regimens for certain
9 conditions, and also differences in the treatment
10 of acute and chronic pain.
11 [Slide.
12 Some comments now about chronic pain
13 specifically as the previous presenter outlined.
14 These are the models that we have the most
15 experience in, but limited in terms of their
16 duration in 12 weeks or longer, primarily to the
17 arthritides or osteoarthritis and rheumatoid
18 arthritis. You find very rare cases or even
19 nonexistent, that the other conditions have been
20 studied beyond one or two weeks.
21 [Slide.
22 In terms of the approach to the
23 determination of efficacy, we have used
24 successfully the three-domain approach in both
25 osteo and rheumatoid arthritis, which would include
193
1 pain intensity, a global assessment and functional
2 or disability assessments, and are pushing forward
3 into chronic low back pain using specific
4 instruments for those conditions, but again using
5 the three-domain approach.
6 As was mentioned this morning, however,
7 this approach may not be applicable to all
8 conditions. Our experience in cancer pain, albeit
9 limited, we have experienced difficulty in showing
10 functional improvement in combination with improved
11 global or pain severity scores.
12 As I mentioned before, on the previous
13 slide, there is a limited number of models, at
14 least in our hands, that would appear to be
15 suitable for the study of three months, and even
16 those that may be approachable for this duration of
17 time, you are always left with the dilemma of what
18 to do with patients on extended placebo treatment,
19 how to handle that both in the clinical trial, as
20 well as the statistical imputation that results.
21 Also, we would like to propose that
22 serious consideration be given to models of chronic
23 intermittent pain, what particular endpoints might
24 be useful, the duration of treatment, and/or the
25 numbers of cycles that would be needed to be
194
1 treated.
2 Finally, we need to clearly outline as Dr.
3 Witter mentioned this morning, the safety
4 requirements for chronic pain in much more detail
5 than the current guidelines now described.
6 [Slide.
7 Due to the heterogeneous nature of chronic
8 pain conditions, as we have heard this morning, we
9 have also proposed a tiered approach slightly
10 different than Dr. Simon had outlined this morning,
11 but we would also agree that a separate indication
12 for each condition with replicate studies would
13 seem to be of benefit.
14 An indication for chronic musculoskeletal
15 pain, we would propose could be achieved with a
16 single study in three chronic musculoskeletal
17 conditions, in a sense, a replication would be
18 achieved, as well as spreading out the
19 observations, if you will, across a number of
20 musculoskeletal conditions.
21 Finally, we would propose that a way
22 forward for a general chronic pain indication would
23 be a single study in two chronic musculoskeletal
24 models and/or cancer pain, and a single study in
25 two neuropathic models, again tracing back t the
195
1 differences in the mechanisms at least as we best
2 have them identified now. This would seem to fit
3 very well with that particular model and the
4 limitations.
5 I just want to make a couple simple points
6 here. One model does not achieve all necessary
7 objectives, in particular, demonstrating acute
8 onset of analgesia is difficult in some of these
9 models due to the high placebo response and the
10 self-limiting nature of the pain often confounds
11 demonstration of effective regimens, particularly
12 on the days 2 and beyond.
13 It is also very variable as to what an
14 effective regimen might be depending on the model
15 that is selected, so we would advocate that studies
16 with multiple doses over a number of days be
17 conducted in both musculoskeletal conditions, as
18 well as post-surgical conditions.
19 Finally, one additional comment
20 particularly related to primary dysmenorrhea. This
21 is sort of an orphan here. It is a stand-alone
22 indication, however, data from this particular
23 model has been used in many cases to support an
24 overall acute pain claim, particularly with respect
25 to onset of action.
196
1 [Slide.
2 While the current guidelines provide
3 adequate criteria in our view to evaluate
4 single-dose analgesic efficacy, it is traditionally
5 understood that replicate studies in dental pain
6 and post-surgical pain are required.
7 There are well-defined efficacy measures
8 assessing onset, extent, and the duration of
9 analgesia. Again, it is generally understood that
10 the time to onset of analgesia should be
11 demonstrated to be less than one hour in replicate
12 trials, and that the time to rescue medication from
13 single-dose studies is used to support the dose
14 regimen on day one and subsequent days.
15 [Slide.
16 The criteria to demonstrate multiple-dose
17 efficacy, i.e., an effective regimen, are less well
18 defined by the current guidelines, however, and
19 that is where significant work I think needs to be
20 focused.
21 Also, while we are doing this, study
22 design and study conduct considerations are also
23 important to bring into the mix, and that includes,
24 as I mentioned before, the self-limiting nature of
25 the pain in some models and also that the severity
197
1 of the initial pain in other models may not be
2 controlled by monotherapy alone.
3 [Slide.
4 Just to give you a little example of the
5 self-limiting nature of pain, this is data from a
6 thinly disguised COX-2 specific inhibitor trial in
7 laparoscopic cholecystectomy looking at the percent
8 of patients with moderate to severe pain plotted on
9 the Y axis versus the days post-surgery on the X
10 axis.
11 Here, you can see significant treatment
12 effects on days 2 and 3, but overall by day 4, and
13 particularly by day 5 and beyond, you can see that
14 the numbers of patients experiencing moderate to
15 severe pain even in the placebo group is quite
16 small and does not allow adequate assay sensitivity
17 to see a drug effect.
18 [Slide.
19 Finally, just a comment about multimodal
20 analgesia, obviously, the premise here is to obtain
21 additional clinical benefit by controlling pain
22 with agents from two to more classes. Ideally,
23 these would be operating through different
24 mechanisms or at least different sites, and the
25 efficacy measures versus monotherapy would be
198
1 reduced medication requirements, improved analgesia
2 over monotherapy, a reduction in adverse effects,
3 and improved patients global assessments.
4 [Slide.
5 Again, just to give you a little taste of
6 what that looks like, this is from a total knee
7 arthroplasty study looking at morphine alone here
8 in the white line down at the bottom, and then two
9 doses of a COX-2 specific inhibitor in the blue
10 line at the full therapeutic dose, and in the
11 yellow line, at half-maximal therapeutic dose.
12 You see that there are significant
13 improved analgesia scores in terms of reduction in
14 pain intensity with both doses versus morphine
15 alone.
16 [Slide.
17 One acute pain model does not fill all
18 criteria for determination of a single dose and
19 multiple dose efficacy, and we would propose that
20 new guidelines specify in more detail which models
21 are best to define onset, peak effect, and
22 duration, specify compartmental approaches perhaps
23 for pain studies, for example, single dose,
24 multiple dose studies on day one and subsequent
25 days, and then propose models best for monotherapy
199
1 versus combination therapy.
2 [Slide.
3 Finally, specify what acute pain models
4 are needed to obtain a broad acute pain indication
5 by severity and/or etiology. We have spoken about
6 that in the context of chronic pain today, and I
7 suspect some of the same conversation will surface
8 tomorrow.
9 Specify how many models and whether
10 replication is needed in each. If models are of
11 similar etiology, we would propose that really only
12 one model should need replication in that
13 particular instance.
14 Finally, as was my comment with chronic
15 pain, we need to more carefully define the safety
16 requirements for acute pain with a new agent,
17 either when studied alone and/or in combination
18 with pursuit of a chronic pain indication.
19 Thank you.
20 DR. FIRESTEIN: Thank you. Could you
21 clarify just one point, and that is, for the
22 chronic pain indication, the alternative proposal,
23 two studies and three models is one study and four
24 models?
25 DR. VERBURG: Yes, I was proposing one
200
1 study across four different models.
2 DR. FIRESTEIN: So, there would be fewer
3 studies for each individual indication, but an
4 increase in the number of total indications
5 examined for the chronic pain?
6 DR. VERBURG: Yes.
7 DR. FIRESTEIN: Thank you.
8 The next speaker is Eugene Laska, Director
9 of Statistical Sciences Division, Nathan Kline
10 Institute for Psychiatric Research, sponsored by
11 Merck Research Laboratories.
12 DR. LASKA: The business of doing clinical
13 trials in the context of randomized, double-blinded
14 clinical trials is to develop inferences that are
15 causal, to be able to claim that the reason we see
16 drug differences are because the different
17 treatments that were in the trial were causal.
18 [Slide.
19 As a consequence, any of the decisions
20 made in terms of what must be demonstrated to get a
21 claim has to be done within that context. We have
22 not spent a lot of time this morning, some of the
23 speakers before me have, on the details of clinical
24 trial design and methodology, and I think that both
25 the beauty and the devil are in those details.
201
1 How to do clinical trials in the chronic
2 and acute framework are clearly needing additional
3 input, improvements in design, styles, and methods,
4 and methods for inference. I will be very brief now
5 because I have some time to talk about the acute
6 setting, so now I just want to say one brief word
7 about doing research in the chronic framework.
8 [Slide.
9 Right now there are precious few, if any,
10 I am not aware of any clinical trials that have
11 really answered the question about what to do about
12 the fact that placebo patients in a chronic
13 framework drop out very rapidly, and statisticians
14 have developed both crude and very sophisticated
15 methods for imputing data, the crudest being the
16 last observation carried forward and variance
17 thereof, and the more sophisticated using methods
18 of multiple imputation developed by some quite
19 credible and rather brilliant statisticians.
20 In my view, none of those satisfies the
21 criteria needed to draw valid causal inference
22 because there is some form of informative censoring
23 going on in these trials, in particular, placebo
24 patients are dropping out because they are not
25 getting adequate relief, and adverse effects are
202
1 coming into play, so the censoring mechanism may
2 very well be informative.
3 A design has been used in other areas of
4 medicine, appears to me to be potentially very
5 relevant in this arena, and that is the so-called
6 withdrawal trial. The withdrawal trial is an
7 enrichment trial in which patients stay on the
8 trial for the 12 weeks, as Lee proposed, for
9 example, and dropouts are taken note of and there
10 is some kind of inference on the dropout rates
11 done, but the only patients who are relevant are
12 those who have stayed on and had satisfactory
13 response from the test treatment by the 12th week.
14 Those people, I believe should have a
15 criteria, for example, the one I described, at
16 least some X percent of the patients who started
17 the trial have to be around for the 12th week for
18 the drug to be considered a chronic medication.
19 At the end of that week, patients are
20 randomized into one of two groups. Half remain on
21 the trial that they started with, on the treatment
22 that they started with, they remain on the drug,
23 the other half go off the treatment they started
24 with, and go on to a placebo, and proof that the
25 drug works is contained in demonstration of placebo
203
1 treatment superiority during the subsequent period
2 of time. Depending on the drug, it might be a week
3 or two weeks thereafter.
4 This particular approach does away with
5 the need for imputing the values of dropout
6 patients to the end of the trial, and when patients
7 are dropping out in the first and second and third
8 week, the imputation really looks quite silly.
9 This is a proposal that I think needs some
10 time and attention, and hopefully will allow us to
11 draw better inference about the treatments we wish
12 to investigate.
13 Thank you.
14 DR. FIRESTEIN: Thank you.
15 The next speaker is Mason Diamond,
16 pharmaceutical consultant.
17 DR. DIAMOND: Thank you. My name is Dr.
18 Mason Diamond. I am independent consultant,
19 pharmaceutical consultant from the Boston area. I
20 am also Vice President at Engenium [ph] Research,
21 which is a contract research organization based on
22 North Carolina.
23 I am speaking today on my own behalf and I
24 paid my own expenses to attend this meeting. At
25 this moment, I have no financial arrangement nor
204
1 financial interest in any company or CRO currently
2 involved in the development of analgesics.
3 Before I begin, I wish to thank the FDA
4 and the Arthritis Advisory Committee for giving me
5 the opportunity to address this group.
6 Furthermore, I would like to commend CDER, Division
7 550, and specifically Dr. Simon and Dr. Witter for
8 taking this much needed initiative. To my
9 knowledge, no other regulatory authority has done
10 this.
11 My purpose in speaking today is to
12 highlight some concerns regarding the needs of the
13 elderly population. I strongly believe that these
14 concerns should be addressed in analgesic drug
15 development.
16 There are over 34 million Americans over
17 the age of 65 that are affected by pain. Research
18 has shown that at least 62 percent have taken
19 prescription medication for more than six months to
20 treat their pain.
21 More disturbing are the estimates that as
22 much as 80 percent of nursing home residents suffer
23 from painful conditions that go untreated.
24 Arthritis has been identified as the
25 single most common cause for chronic pain in the
205
1 elderly, however, it is not uncommon to see more
2 than one indication requiring analgesic therapy.
3 In addition, most elderly persons have multiple
4 medical problems that require multiple medications.
5 Many drugs used to treat these concomitant
6 conditions have not been sufficiently evaluated for
7 co-administration with each other, let alone with
8 many analgesics. As a result, the comprehensive
9 guidelines necessary to deal with the complex
10 safety issues in this population are not available.
11 It is the fear of possible serious and
12 life-threatening side effects that is often the
13 barrier to adequate pain treatment in older adults.
14 The situation is further complicated by progressive
15 cognitive and emotional difficulties encountered in
16 this population.
17 This makes medical evaluation and
18 management even more challenging. The net result
19 is that while in many cases the pain management
20 with drugs and other treatments are possible, each
21 year millions of older people are forced to endure
22 unbelieved suffering.
23 The elderly represent the largest number
24 of pain sufferers and purchasers of analgesic
25 products, yet, they remain in the greatest need of
206
1 innovative therapies.
2 In an effort to address this need, I would
3 like to offer some points to consider as we move
4 forward in our discussions of analgesic pain models
5 and clinical study designs.
6 First, inclusion/exclusion criteria. In
7 order to minimize response variability in our
8 clinical studies, it is common for us to enroll as
9 homogeneous a population as possible. While
10 scientifically sound, this approach tends to
11 exclude those individuals who may be most
12 representative of the target population.
13 For example, in arthritis trials, the
14 actual effectiveness and safety profile common to a
15 more frail elderly population may not be reflected
16 in the Phase III study results. My recommendation
17 would be to ensure a more representative patient
18 cohort in our pivotal clinical trials or conduct
19 separate studies specifically in this population.
20 Second, the pharmacokinetics and
21 pharmacodynamics of drug interactions significantly
22 complicates pain management in older adults. The
23 resulting side effects from polypharmacy, coupled
24 with the underlying medical conditions, can be
25 daunting to deal with.
207
1 It is not uncommon for the elderly to be
2 on five or six medications at a time and often
3 more. Although these issues have been discussed in
4 the FDA and ICH guideline documents, and drug
5 companies do go to great lengths to evaluate drug
6 interactions, these studies need to include more
7 older adults who are being treated for multiple
8 medical conditions since they represent the
9 ultimate beneficiaries of these new therapies.
10 Third, the duration of evaluation. The
11 most common pain problem in the elderly are chronic
12 and patients often take analgesic medications for
13 long periods of time, if not for the rest of their
14 lives.
15 Many adverse events become evident only
16 after long term use. Evaluations of 12 weeks or
17 even 12 months may not be sufficient to capture the
18 long-term risks and benefits of a particular drug.
19 I am sure that everyone here agrees that we are all
20 committed to bringing safe and effective
21 medications to the public as rapidly as possible,
22 however, we must also ensure that our research
23 provides the necessary information to enable
24 practitioners to better manage their patients
25 especially those on complex treatment regimens.
208
1 This could be accomplished by blinded
2 studies of longer duration or by employing longer
3 open-label follow-up extension studies, which would
4 provide this much needed information while not
5 impeding the drug development process.
6 Finally, outcomes evaluation, I think on
7 everybody's mind. In a search for better methods
8 to evaluate pain, we are focusing on objective
9 measures to incorporate into our study designs,
10 mechanism-based assessments, determination of
11 biomarkers for underlying diseases, and levels of
12 pain modulating biomolecules are some of the
13 options under discussion.
14 I feel that all these options should be
15 actively pursued, however, these approaches will
16 take some time to validate. Also, in many cases,
17 the objective evidence for underlying disease may
18 not correlate with the symptoms, and symptoms may
19 wax and wane spontaneously.
20 One solution is the utilization of
21 multidimensional pain outcomes. This includes pain
22 assessment, functional assessment, psychological
23 outcomes, and quality of life measures.
24 New assessment tools designed for both
25 cognitively impaired and unimpaired elderly adults,
209
1 such as the geriatric pain measure developed at
2 UCLA, are in the process of being validated. In
3 addition, there are very many well-established and
4 highly validated tools dealing with each of these
5 areas that are currently available, however, since
6 pain affects so many aspects of people's lives, no
7 one measure can adequately capture the overall
8 effect of any therapy.
9 For example, in an arthritis trial, it is
10 possible to show no change in pain level, but a
11 significant impact on the patient's ability to
12 function. This is due to an individual's ability
13 to adapt their level of activity to the level of
14 pain tolerance.
15 So, if a patient takes an analgesic that
16 enables them to climb stairs, walk a greater
17 distance, take care of themselves, or play with
18 their grandchildren, but continues to report pain,
19 I would still consider this a clinically
20 significant outcome.
21 In addition, the impact of pain on an
22 individual's psychological state and overall
23 quality of life is no less relevant than pain level
24 or functional status. Therefore, until we have one
25 system that measures all of these parameters, we
210
1 should evaluate efficacy based on more than one
2 outcome.
3 It is clear that the treatment of pain in
4 older adults is an enormous undertaking. No less
5 so is conducting clinical trials in the elderly
6 population. We must remember that the information
7 captured during drug development provides guidance
8 for practitioners in addition to satisfying
9 regulatory requirements.
10 Therefore, I believe that by addressing
11 the needs of the elderly during the drug
12 development process, we will enable the medical
13 community to more effectively treat the millions of
14 elderly patients through a need and bring them the
15 benefits of these new drugs.
16 Thank you.
17 DR. FIRESTEIN: Thank you very much.
18 The next speaker is Daniel Carr from Tufts
19 University.
20 [Pause.]
21 DR. FIRESTEIN: While we are waiting to
22 sort out our technical difficulties, why don't we
23 move ahead to the next person that is not using
24 slides.
25 Dr. Abraham Sunshine from Analgesic
211
1 Development.
2 DR. SUNSHINE: Thank you. I am Abraham
3 Sunshine, Professor Clinical Medicine at NYU School
4 of Medicine. I am President of Analgesic
5 Development. I appear here on my own, and I have
6 not received any compensation from pharmaceutical
7 companies to appear.
8 I was asking myself why did I want to
9 speak, and I think I can contribute in giving some
10 historical perspective on the analgesic guidelines.
11 The 1993 Guidelines, which we well
12 described by Dr. Fang and her associates, really
13 began in the eighties, and it took 10 years to get
14 a document that went through all the hurdles,
15 first, to get a consensus and then to get it
16 through the FDA.
17 So, that document is over 20 years old. I
18 want to acknowledge the work of Lee Simon and his
19 associates for initiating this conference, and also
20 the work of Ray Dionne who ran the consensus
21 meeting at the NIH.
22 The 1992 Guidelines really were driven by
23 investigators and industry who just didn't know
24 what to do to get an analgesic approved, and the
25 ground rules were changing with each drug that was
212
1 approved, so to move forward, it was thought that a
2 consensus would be helpful.
3 Now, the guidelines served us well. The
4 drugs that were being developed at that time were
5 acute analgesics. There were no drugs for chronic
6 pain, and the last thing a pharmaceutical company
7 would be interested in is developing a treatment
8 for neuropathic pain.
9 So, there was no discussion, as Dr.
10 Firestein pointed out, about how to conduct chronic
11 trials because there were very few chronic trials
12 or drugs being considered, and opioids for chronic
13 nonmalignant pain was a no-no. People didn't use
14 opioids for chronic nonmalignant pain.
15 I think advances have been made now, as we
16 saw fentanyl being used, patch being used in low
17 back pain, but we also know about the OxyContin
18 story, that anybody that had a backache was put on
19 dope and got into trouble.
20 The guidelines did permit us to develop
21 many of the NSAIDs both for Rx and also to define
22 an OTC dose. The technology was developed, so that
23 one could pick up the effects of 12.5 milligrams of
24 ketoprofen, and even 100 milligrams of ibuprofen,
25 and dose-response work was done using these
213
1 guidelines.
2 The guidelines also helped avoid
3 pseudospecificity, and I think this is an important
4 point because we are at a road where I think as I
5 hear rumblings, that we are going to
6 pseudospecificity. For example, dysmenorrhea was
7 understood to be a drug, recycled oxygenase was
8 involved, but in order to get a claim for treatment
9 of dysmenorrhea, one had to show that the compound
10 work as a general pain medication, and then, in
11 addition, in dysmenorrhea.
12 I was on the web site that Lee talked
13 about, and it really is a good web site and I see
14 that Google has helped you get this web site
15 working, and yesterday morning I came across CDER's
16 policy on OTC analgesics 1994, signed by Dr.
17 Woodcock, who clearly points out that to get a
18 claim for menstrual cramps, one needed two positive
19 clinical trials in appropriate pain models, and in
20 addition, positive clinical trial in an OTC
21 dysmenorrhea model.
22 I don't think these guidelines are being
23 followed at the moment, and now we are getting
24 pseudospecificity where drugs which really have a
25 broad implication in terms of pain management, are
214
1 brought labeled for dysmenorrhea, and not for
2 general pain.
3 The other that was important to emphasize
4 in the eighties and nineties is that small sample
5 sizes of 30 to 50 patients per treatment in a
6 single center generated important data, and data
7 where you got dose response to the NSAIDs.
8 Ketoprofen, from a dose of 12.5 milligrams up to
9 100 milligrams was clearly defined.
10 Today, and I don't know the reason, one
11 needs hundreds of patients per treatment arm and
12 then there is a lot of deliberation is the drug
13 better than placebo.
14 One of the problems, I don't know that it
15 was discussed so far, is combination therapy. Very
16 few combination drugs have been approved. I mean
17 there are combinations of ibuprofen with opioids,
18 and there is a combination of tramadol with
19 acetaminophen, so polypharmacy didn't get ahead.
20 One of the reasons, it was extremely
21 difficult to show the contribution of each of the
22 ingredients. Although we know that codeine works,
23 and we know ibuprofen works, put them both
24 together, and the results were not convincing, so
25 there is no ibuprofen-codeine product even though
215
1 it was attempted many times.
2 I think as you move forward with the
3 guidelines, it is clear that polypharmacy is here
4 to stay. The other thing, polypharmacy was
5 discovered by patients, not by CDER, not by the
6 industry, but if you look back, there was Empirin
7 compound, acetaminophen, and aspirin--Dr. Brandt
8 talked about that--and caffeine. Then, there was
9 Empirin with codeine, and these were drugs that
10 just over time were found to be helpful, but when
11 pure science came to play, combination therapy was
12 a no-no, and you had to prove the contribution of
13 each of the compound.
14 When Burroughs-Wellcome took caffeine out
15 of Empirin compound, the sales of Empirin compound
16 plummeted, much like the stock market is doing
17 today, and that compound is off the market. I
18 think that caffeine has a role as an analgesic
19 adjuvant.
20 DR. FIRESTEIN: Dr. Sunshine, could you
21 please wrap up? Thank you.
22 DR. SUNSHINE: Okay. I think as we go
23 ahead that we have to develop tools to explore all
24 the contributions of the neuroscientists that Dr.
25 Woolf discussed today, so that we can utilize the
216
1 information to develop better drugs. Time does not
2 permit me to go into that aspect, but in five
3 minutes I couldn't answer the question, so I think
4 it is going to take maybe not 10 years, but a
5 couple of years.
6 Thank you.
7 DR. FIRESTEIN: Thank you very much.
8 I believe now our information technology
9 problem has been solved, and we can now go back to
10 Dr. Carr's presentation.
11 DR. CARR: I thank the committee very much
12 for having invited me down here. In particular, I
13 think Lee and Jim Witter, and as did the prior
14 speaker, I thank Ray Dionne for having organized a
15 preconference and also Ms. Reedy for getting me
16 down here.
17 As I was listening to the erudite and
18 complex discussion earlier today, I wonder what
19 might there be that hadn't yet been said. So, I
20 titled the title of this 10-minute presentation
21 "What might still be said, that hadn't yet come
22 across," and I am speaking from a rather
23 distinctive point of view of a clinician, but I
24 would like to call attention to a great resource
25 that I think has yet not been tapped, and should be
217
1 tapped, which is that the evidentiary body upon
2 which clinicians seek to make recommendations for
3 therapy and to treat their patients, insofar as
4 analgesics are concerned, in large part, derives
5 from approval trials.
6 So, I would say that there is an
7 opportunity to render this very robust
8 data-generating process much more useful to
9 clinicians and therefore, their patients.
10 [Slide.
11 Now, to try to lighten the postprandial
12 stupor, I thought I would begin by posing four
13 simple questions. The first is--and these are
14 reasonable questions--who won the last presidential
15 election? Did X Corporation make money or lose
16 money? As Dr. Sunshine mentioned, we are all
17 interested in that.
18 What kind of pain does my patient have,
19 and what is the most effective treatment for my
20 patient's pain? In the interest of time, I am not
21 going to cover the first two questions, but I will
22 say that in try to cover or provide mustering of
23 evidence to answer the third and fourth questions,
24 I have had the privilege to be involved with some
25 wonderful individuals over the years, with Ada
218
1 Jaycox for the old AHCPR acute and cancer pain
2 guidelines, and more recently with Joseph Lowe and
3 Leo Gudis and others for work with AHRQ.
4 So we have actually made an earnest effort
5 to try to muster the evidence. This report, which
6 can be cited or traced through the AHRQ web site,
7 on cancer pain, involved screening over 18,000
8 titles. A couple of weeks ago, there was an NIH
9 State of the Science Conference held here in
10 Bethesda, as well, just down the block, and for
11 that we screened an incremental 6,000 titles
12 relating to cancer pain.
13 So, we have made an effort to try to
14 muster the evidence.
15 [Slide.
16 At the same time, and I am sorry if I
17 repeat what you have heard before, but I am just
18 putting things that I think clinicians might tend
19 to focus on, is that recent insights, much of them
20 accomplished by individuals in this very room, to
21 my mind have blurred the boundary between acute and
22 chronic pain.
23 Pain is itself a widely distributed
24 process, and I am not sure we have mentioned the
25 brain yet, but the brain and imaging of the brain
219
1 are both very important factors to consider in
2 understanding pain.
3 I think we have heard, although perhaps
4 not in these words, that chronic pain is itself a
5 disease, and a theme that has popped up again and
6 again amongst different speakers is that the field
7 itself has arrived at what you might term
8 combination analgesic chemotherapy, much as one
9 uses combination chemotherapy for other conditions.
10 In fact, the onset of the disease of
11 chronic pain is potentially very rapid. If one
12 looks at epidemiologic data from the 1999 IASP book
13 on Epidemiology of Pain, edited by Crombie or the
14 2000 Review in Anesthesiology by Perkins and
15 Kehlet, it is quite clear that many patients who
16 undergo operations of any kind will develop
17 persistent pain.
18 I think this is an under-recognized
19 epidemiologic factor, but it is very, very
20 important, and I am actually surprised that this
21 market opportunity hasn't been seized upon. There
22 is also much insight into the long- and short-term
23 benefits of aggressive therapy, although in the
24 preemptive analgesia area, it is clear that a
25 single drug is unlikely to make an impact.
220
1 We have also had evolving understanding of
2 drug pharmacokinetics and pharmacodynamics in
3 particular appreciating the diversity of
4 individuals according to gender or ethnicity or
5 even as far as interpretive aspects go, culture.
6 There has been tremendous insight into
7 understanding the mechanisms of opioid tolerance,
8 and we are just beginning to see the emergence of
9 insight into disease-specific mechanisms, such as
10 in cancer.
11 For example, I refer to work by Debar and
12 colleagues on identification of endothelin-1 as a
13 cancer-specific mediator. Nonetheless, as one has
14 tried to consolidate all these published trials,
15 and by the way, I think the efforts to
16 consolidation are themselves an advance through
17 Cochrane or evidence-based practice centers, the
18 fact remains that the vast majority of most pain
19 treatment is empiric and generic.
20 In other words, one starts with
21 acetaminophen, perhaps switches to a nonsteroidal,
22 perhaps has a so-called weak opioid, or perhaps
23 changes the weak with a strong opioid, which is the
24 same algorithm you might follow for a badly
25 sprained ankle, as cancer pain.
221
1 [Slide.
2 One of the big problems in trying to
3 organize the evidence is that the evidence itself
4 is quite flawed, and I think the FDA can help
5 future generations. Randomized, controlled trials
6 are a tiny fraction of the pain literature. It is
7 quite shocking, but when we did the acute pain
8 guideline in '92, we pulled 13,000 titles, of which
9 675 were randomized, controlled trials.
10 Last year, when we did the cancer pain,
11 roughly 20,000 titles screened, as you saw, about
12 180 were randomized, controlled trials, and for the
13 interim State of the Science NIH Consensus
14 Conference, we got another 6,000 titles. We
15 boosted that figure from 180 to 216.
16 What are all these other trials? The vast
17 majority are observational or describe a technique.
18 Because of the nature of the literature, so many
19 different types of diagnoses, patients, and outcome
20 measures, it is impossible to do a quantitative
21 meta-analysis for most of the clinically important
22 questions.
23 In fact, for the State of the Science
24 Conference two weeks ago, of the 218 retrieved pain
25 trials in cancer pain, there were 125 different
222
1 pain-related instruments that were employed.
2 Now, granted, some of the differences were
3 in a 3-point scale versus a 4-point, versus a 10-
4 or 11-point scale, but the fact of the matter is
5 there could really be a great service done to
6 insist upon some standardization for pooling of
7 this colossal, but difficult-to-combine body of
8 knowledge.
9 The generalizability of the trials, as you
10 have heard before, is limited by inclusion and
11 exclusion criteria. The clinician is treating an
12 individual who has comorbidity, who may be elderly,
13 who is taking other drugs, and these are not
14 represented in the data upon which the evidence is
15 based.
16 A very important factor is the relatively
17 small amount of focus placed upon side effects.
18 Side effects, including adverse events, but even
19 predictable side effects are what keep many
20 patients from achieving good pain relief, such as
21 with opioids, and it would be wonderful if there
22 were a non-punitive shift in the process, so that
23 side effects could be monitored prospectively and
24 with greater precision than in the past without
25 penalizing the sponsor of the trial.
223
1 One has a sense from the literature that
2 previously, there was a process set up which
3 encouraged actually underpowered trials, that is,
4 few patients per trial. If one looks at the actual
5 retrieved trials for cancer pain treatment, for
6 example, these are on the orders of dozens of
7 patients per trial, but if you look at cancer
8 treatment, such as primary chemotherapy, through
9 collaborative groups, these number hundreds or
10 thousands.
11 In fact, if one were to calculate the
12 number of patients, let's say, with cancer pain
13 versus the number of patients enrolled in trials,
14 these are a tiny, tiny fraction of those with the
15 condition.
16 [Slide.
17 Well, what about that question, is this
18 treatment helping, well, to translate efficacy data
19 into effectiveness is the mission of a clinician,
20 and thus far I have called attention to some gaps
21 in the literature and what FDA can do to help.
22 I would say that to patients and their
23 families, the primordial outcome is low pain
24 intensity. On the other hand, particularly with
25 the treatment of chronic non-cancer pain, quality
224
1 of life often trumps the pain intensity on a visual
2 analog scale. Very often the approach to treatment
3 of chronic non-cancer pain is to encourage patients
4 to do more even if their visual analog scale does
5 not go down, and as you have heard, very commonly
6 in the clinical setting, patients self-titrate to a
7 visual analog scale, which may be moderate pain,
8 but they are able to do more.
9 We need standardized consensus
10 instruments. Right now there is an effort underway
11 that I am privileged to be involved with. It's a
12 tripartite collaboration of the Joint Commission
13 AMA and NCQA to try to develop performance measures
14 to evaluate the implementation on site of JCAHO
15 guidelines, but this is a bit of a struggle.
16 We will get the job done, but is not
17 helped by the proliferation of instruments.
18 Obviously, you have heard a lot of erudite comment
19 about the need for generic versus
20 condition-specific instruments.
21 One caveat is that coarse instruments, and
22 the SF-36 is a coarse instrument, may overlook
23 benefit, which is actually done to patients. I
24 guess it's a disclaimer, I have been involved in
25 the development of the Treatment Outcomes of Pain
225
1 Survey from Tufts or TOPS scale, that is
2 essentially an augmented condition-specific SF-36
3 validated for patients with chronic pain.
4 Of course, we are aware that we can't just
5 administer endless questionnaires because of the
6 burdens on patients and clinicians. I have already
7 mentioned that side effects seem to be approached
8 very differently in the literature, in a much more
9 cavalier haphazard way than are the desired
10 outcomes, but they are often the thing that stops
11 the patient from getting better. They just can't
12 increase the dose.
13 So, are there things one do towards an
14 answer?
15 [Slide.
16 Well, I personally believe that to frame
17 compartments about acute pain or chronic, to say
18 when does acute become chronic, it is a little bit
19 of a misleading question because it equates a time
20 course with a mechanism, but we all know there are
21 many instances of prolonged acute pain, such as
22 labor pain or arthritis, a sunburn or if someone
23 comes in with an obstructed viscus, which are
24 cured, and they never become chronic pain, or even
25 repetitive pain like muscle bruises or soreness in
226
1 athletes, for instance.
2 Therefore, one must infer that nociception
3 itself rarely induces chronic pain except perhaps
4 when there are psychosocial factors. These are the
5 small accidents that evolve into disabilities.
6 On the other hand, the progression of
7 acute to chronic pain is well documented
8 clinically, and as I have mentioned, is a big
9 problem in epidemiologic terms.
10 DR. FIRESTEIN: Dr. Carr, would you wrap
11 up. Thanks.
12 DR. CARR: The last slide, I think, but I
13 will wrap this up in a minute.
14 [Slide.
15 I would submit to you that we have to look
16 at the evidence and apply logic and distinguish
17 between intense nociception, which most of us imply
18 by the phase acute pain, versus the rapid onset of
19 peripheral and central nervous system
20 reorganization, that Professor Woolf spoke to you
21 about.
22 There seems to be a clue that if you have
23 concurrent nerve injury and intense nociception or
24 inflammation, that increases the risk, so in an
25 ideal world, if we all did our jobs, there would be
227
1 prospective identification, planning for patients
2 at risk, individualized anti-nociceptive and
3 behavioral interventions, effective treatments
4 chosen according to evidence, and combined, these
5 would be titrated, we would monitor standardized
6 outcomes to validate and calibrate our practice.
7 In so doing, we would accomplish our
8 mandated continuous quality improvement, we would
9 meet JCAHO standards and identify best practices.
10 Then, we would follow up people and we would assess
11 long-term cost and benefits.
12 Thank you very much for your attention.
13 DR. FIRESTEIN: Thank you.
14 The next speaker is Dr. Ann Berger, Chief,
15 Pain and Palliative Care at the NIH.
16 DR. BERGER: Thank you. I want to also
17 thank Radion and James Witter. In looking at what
18 I could offer here, it is similar to Dan in that I
19 can offer the clinical perspective of pain and
20 palliative care.
21 Prior to coming here, I had run both the
22 Pain and Palliative Care Service at Yale and at
23 Cooper Hospital, which is part of the University of
24 Medicine and Dentistry of New Jersey, so I have had
25 a lot of experience with palliative care patients,
228
1 as well as chronic benign pain patients.
2 In looking at the total pain picture, I
3 brought a handout and I am sorry I didn't make a
4 slide, I didn't know we could show slides, the
5 total pain picture is really made up of the
6 physical pain, which at least clinically, from my
7 experience, is usually not just neuropathic pain,
8 it's not just visceral pain, it's not must somatic
9 pain, it is usually a combination pain.
10 So, it is going to be pretty difficult to
11 say you are going to do a study just on neuropathic
12 pain because unless you are talking about something
13 like brachial plexopathy or diabetic neuropathy,
14 because many of the pains are mixed pains.
15 We see this all the time with patients,
16 but then besides the total pain picture of being
17 all those physical different mechanisms, we have a
18 whole element of suffering, and I think that is
19 where we really miss the boat in medicine.
20 The suffering components is not only
21 depression, it is not only the psychological
22 states, but it is social issues, it's loss issues.
23 When somebody came up and spoke about pain in the
24 elderly, that's a huge problem and partly it's a
25 huge problem because the loss issues are so huge.
229
1 These are people who have lost their pets,
2 their furniture, their families, their friends, and
3 that is something we never take into account.
4 Suffering also involves spiritual concerns, and for
5 anybody in pain, whether they are religious or not,
6 it is always a spiritual issue because anyone who
7 is sick or anyone is in pain, it's why is this
8 happening to me, purpose-meaning type issues, as
9 well as social family functioning, physical
10 disability, and for palliative care syndromes, it
11 is fear of death.
12 Now, the only difference in my mind
13 clinically, when I look at a patient, is, is this a
14 palliative care patient or is this a chronic benign
15 pain patient, and the way I define that is
16 palliative care are patients that can ultimately
17 die from their disease, so they have a
18 life-threatening disease, something like cancer,
19 something like HIV disease. Clearly, there are
20 lists of those, you know, because many diseases we
21 don't cure, so COPD, CHF, you know, many diseases.
22 Chronic benign pain are patients like with
23 low back pain, fibromyalgia, endometriosis, chronic
24 pancreatitis, and these people are not going to die
25 from their disease, but the treatments really need
230
1 to be very similar to the cancer pain population.
2 My background and how I got into this, I
3 was initially an oncologist and I consider myself a
4 reformed oncologist, and actually started the
5 Palliative Care Service at Yale, and at the time
6 started ending up seeing a lot of chronic benign
7 patients.
8 How did that happen? It happened that an
9 oncologist was doing that because the principles
10 were the same principles. So, you know, it is not
11 unusual to get lower back pain, reflex sympathetic
12 dystrophy, fibromyalgia, and I was a little
13 concerned with looking at the guidelines to say,
14 well, you are going to just divide it up into
15 little departments of all these different pains,
16 when it is really a much broader issue, and these
17 chronic pain patients are very similar in many,
18 many ways.
19 What has struck me so many times, you
20 know, initially, when I got into more of the
21 chronic benign pain part, but just all the time, is
22 that the suffering issues of these patients are at
23 least as much, if not more, than the palliative
24 care, cancer pain, HIV population, overwhelming.
25 So, I say that this is a component that we
231
1 have missed in medicine, we have missed the boat
2 because we always think that there is a medication
3 for that, and there is no medication for suffering.
4 I would like to share an example of a
5 patient that I took care of for a while in New
6 Jersey, a man who had back pain after being
7 disabled on his job as truckdriver, and he ended up
8 going for all kinds of epidural injections, facet
9 blocks, and continued to have pain, then had
10 surgery, and continued to have pain.
11 I mean we all know the story, we have all
12 seen it many times, and he actually became more
13 depressed, was seeing psychiatry, was put on four
14 or five different antidepressant type medication
15 anti-anxiety medicines, was in a stupor, but was
16 still having pain, and ultimately ended up going to
17 a neurosurgeon to have a dorsal com stimulator
18 placed, which failed. Not a big surprise that this
19 failed.
20 At this point, they said all right, send
21 him to Ann, she seems to know how to fix these
22 people. He came to my office crying, crying,
23 crying with his wife, and so we started--the
24 assessment I do is the same like I would on a
25 palliative care patient. I am like what is going
232
1 on here, what is going on.
2 He was a truckdriver, had lost his job,
3 again, all these losses, had lost his job, lost his
4 finances. This was his whole self-esteem to be a
5 truckdriver. Six months later his daughter
6 actually died of a brain aneurysm and left him with
7 a six-month old baby. Two years after that, his
8 father died of Alzheimer's, and a year after that,
9 his sister died of bone cancer.
10 This is not an unusual story. This is a
11 story that comes into my office every day, whether
12 the patient has low back pain or RSD or
13 fibromyalgia, the stories are usually very similar.
14 The losses are very similar.
15 In terms of the suffering component, the
16 only thing that helps that is all the
17 nonpharmacologic things, counseling. There is no
18 Prozac, there is no Zoloft, there is no medicine.
19 It is counseling, it's art therapy, it's music
20 therapy, it's pet therapy, it's acupuncture, it's
21 Reiki, it's spiritual, it's all these other
22 components.
23 In terms of, in my mind, when I look
24 clinically at a palliative care patient versus
25 chronic benign pain, really, the most important
233
1 difference in terms of how I treat them medically,
2 with the medications, is clearly, if they are
3 palliative care, quality of life has to come first,
4 and you are absolutely correct, function may not
5 increase.
6 You know, sometimes just being awake and
7 breathing is increased function. Whereas, in
8 chronic benign pain, yes, we expect function to
9 increase, and that is the big difference. I don't
10 care what numbers the patients are using. This
11 guy I was talking about before was on heavy doses
12 of oxycontin, up to actually 2,400 milligrams, and
13 still remains at that dose.
14 It didn't matter because he started
15 working, he was functioning after this, and that is
16 the important thing, are you functual again if you
17 have chronic benign pain.
18 The things that I think we don't have
19 enough data on, we clearly don't have enough data
20 on cancer drugs, on neuropathic pain, and also on
21 things like post-treatment pain syndromes. It is
22 very interesting that we don't look at
23 post-treatment pain syndromes.
24 Again, in the elderly, people who have
25 multiple, multiple operations, it is not unusual
234
1 that they are going to have pain after their
2 operations, and this is not something that we think
3 about. It is not only postmastectomy pain,
4 postnephrectomy pain, but it is anytime a surgeon
5 lifts the knife, you could ultimately end up with
6 chronic pain, so a lot of people with abdominal
7 surgery, it is from endometriosis, from
8 pancreatitis, from whatever.
9 DR. FIRESTEIN: Thank you very much.
10 The next speaker is Dr. Thomas Schnitzer
11 from Northwestern.
12 DR. SCHNITZER: I appreciate the
13 opportunity to be here to speak today. I am here,
14 although I do interact with the pharmaceutical
15 industry significantly, I am really here
16 representing myself as a rheumatologist, a
17 Professor of Medicine, and Assistant Dean for
18 Clinical Research at Northwestern University,
19 Feinberg School of Medicine.
20 [Slide.
21 I actually wanted to talk about three
22 specific things. I had three topics that I thought
23 I would want to discuss, but, first, I would really
24 like to commend the FDA, both of the divisions that
25 are here, and Dr. Witter and Dr. Simon for their
235
1 ability to bring together this discussion, which I
2 think is clearly, after the discussions we have
3 heard today, much need.
4 There were three topics I really wanted to
5 talk about, but given the fact that I had limited
6 time, which manages to focus you intensely, decided
7 to really cut down to really just speaking about
8 two of these, the nosology of chronic pain, which I
9 think we have heard a lot about, I will not speak
10 to further.
11 But I would like to talk about the
12 methodology of the efficacy trials, particularly in
13 musculoskeletal pain, really in an attempt to
14 demonstrate I think some of the limitations and
15 some of the opportunities and that exist in terms
16 of methodology.
17 As I am talking to my clinical
18 pharmacology colleagues, I think what is clear, as
19 they say, is that a really good investigator can
20 design a trial that will give the results that he
21 or she wants. So, study design is actually
22 critical, and what I would like to do is focus on
23 the traditional study design we have used to
24 demonstrate some of the limitations of this design,
25 and then to talk about opportunities.
236
1 [Slide.
2 In the area that certainly I have had 15
3 or 20 years experience, a flare design, whether it
4 is osteoarthritis, rheumatoid arthritis, or other
5 types of musculoskeletal disease, is typically what
6 is done.
7 This is what we use for these conditions
8 to be able to demonstrate efficacy. What we
9 haven't really I think given enough thought about
10 is the issue of defining an analgesia-dependent
11 population that we are studying, that we are
12 dealing with high levels of pain, so at the time of
13 randomization, when we actually start to treat
14 patients, their mean pain score is often greater
15 than 70 mm on a 100-mm visual analog scale, so this
16 is not minor league, minor pain, this is I think
17 high intensity pain.
18 I would submit that we are really not
19 looking at a chronic pain model, but we are looking
20 at a subacute pain model, and I was glad to see Dr.
21 Simon in his definition of acute pain actually
22 include subacute pain, which I actually think the
23 models we use would fit very well.
24 Finally, I think we are selecting for
25 drugs that work in acute pain rather than looking
237
1 for drugs that work in a chronic pain mode.
2 [Slide.
3 To be able to perhaps explain that better,
4 I will just take a slide here, which really
5 represents no specific trial, but is similar to
6 what we see in many of these OA trials, looking at
7 pain on walking.
8 The first point represents the patient
9 population that we are screening, so when they come
10 in on their medication. What I would want to
11 indicate is the fact that these patients, in many
12 of these trials, are required to be on full doses,
13 prescription doses of analgesic medication, so they
14 need to be on this medication.
15 To qualify to be in the trial, they need
16 to have an increase in their pain. So, they are
17 analgesia-dependent patients.
18 Now, this population is hardly
19 representative. As an active investigator and as
20 an investigator who believes in collecting metrics
21 at our research center, I can tell you that when we
22 advertise for patients with knee pain, that for
23 every 20 telephone calls we get, we may have one
24 patient enter a trial.
25 So, that is 5 percent of those people who
238
1 were willing to pick up the telephone, call us, and
2 say they have a problem and they would like to be
3 in a trial. Of the patients who actually come in
4 and we can talk to, and we put in the trials, about
5 20 percent qualify in this type of trial.
6 So, the idea that this is giving us a
7 representative sample of patients with
8 osteoarthritis or rheumatoid arthritis is clearly
9 not the case. This is a subset, this is not a
10 general population.
11 The second point to be made is clearly
12 these patients have to flare, so they have now a
13 chronic pain background, but we are requiring that
14 they have the onset of acute pain over the course
15 of usually five half-lives of a drug. Their pain
16 gets up in the range of 70 to 80 mm on a 100-mm
17 visual analog scale, and I will submit this is not
18 looking at chronic pain, this is looking at a flare
19 of acute pain that has been induced by the study
20 design.
21 This is hardly what we, as clinicians,
22 typically see. We don't start patients in our
23 clinic on another drug after they have stopped
24 their previous drug for three or four days. So,
25 this is an artificial situation.
239
1 As I said, I would submit that we are
2 looking at a subacute pain model, not a chronic
3 pain model. When you think about it, what type of
4 drug are we going to select? We need a drug which
5 is going to work quickly. Patients are going to
6 drop out if this drug doesn't work fast. This is
7 going to sound very much like the acute pain
8 argument.
9 So, we need a drug that works quickly, and
10 we need a drug, in addition, not only working
11 quickly, but a drug that is effective for high
12 levels of pain, not mild or moderate levels of
13 pain, but high levels of pain.
14 So, we are selecting for drugs that have
15 already proven that they work in the acute pain
16 setting. We have just gone through a dental pain
17 model for acute pain, which looks at issues not
18 dissimilar to this, and actually has pain levels
19 that are very similar to what we are seeing here.
20 So, I would submit that we are probably
21 not using the right model even though it has been
22 clearly validated and does develop, we will approve
23 drugs, but probably for acute for subacute uses.
24 [Slide.
25 Now, is there another way? Well, it is
240
1 hard to believe, but I actually did not speak to
2 Dr. Laska before this meeting, but I would like to
3 talk about withdrawal trials, as well, and
4 actually, having such an accomplished statistician
5 present this information before I am means that I
6 don't have to deal with the statistical aspect of
7 this at all, which I don't feel qualified to do.
8 But I think there are significant
9 advantages to looking at a withdrawal design. Now,
10 this is not unusual, it has been used in pediatric
11 studies repeatedly for ethical reasons. It is
12 actually included in the RA guidance document, so
13 this is not something which does not have a
14 history.
15 The advantages, in addition to the
16 statistical strengths that Dr. Laska submitted, is
17 that all subjects receive active medication, so
18 this is a real advantage. Everybody gets treatment.
19 For many patients, if you get them for trials, this
20 is important.
21 There is no necessity for disease flare
22 although you can put one in if you want, but there
23 is absolutely no necessity to have a disease flare,
24 so you can actually look at baseline pain levels on
25 treatment, and there is no artificial definition of
241
1 responders.
2 What I mean by that is we are going to
3 have a long discussion, I am sure, both today and
4 tomorrow, about how many millimeters if a
5 clinically meaningful response.
6 Well, in this model, the patient decides
7 that. I mean we don't have to have physicians
8 sitting back trying to make the decision about how
9 much is appropriate. What you really have is the
10 patient says I have had enough, I want out of the
11 trial. That will be different for each patient,
12 but it doesn't matter, because you will actually
13 have a response.
14 [Slide.
15 So, this is what a trial might look like,
16 and there is run-in phase here, which I shouldn't
17 leave out the importance of, because this run-in
18 phase on active medication, so patients are first
19 on active medication for a number of days, allows
20 you to learn a lot about the use of that drug in an
21 open-label fashion. I think that is also an
22 important aspect.
23 Patients are then randomized at some
24 point. The other point about this is they can be
25 randomized at anytime, so the investigator nor the
242
1 patient has to know when that occurs. Then, you
2 see patients dropping out for lack of efficacy or
3 whatever you want to use as your objective
4 endpoint, and a differential dropout rate between
5 patients on active therapy, which would be
6 indicated here, and on placebo or another less
7 active therapy on the bottom line.
8 The intent is really not to say the
9 withdrawal trials are the way to go. It is just to
10 say that I think we need to consider a number of
11 other approaches in terms of methodology, and this
12 may be one of them.
13 [Slide.
14 The last thing I want to talk about is
15 long-term safety. It is really something that has
16 not been talked about today, but I think is
17 absolutely critical.
18 We know from discussions here at the
19 Agency and I think eloquent discussions, that the
20 datasets at the time of NDA are really inadequate
21 to be able to detect uncommon events. We know that
22 some sort of postmarketing surveillance program is
23 required if we want to be able to determine these
24 uncommon events. So, I would say it is required or
25 let's say needed rather than making it a
243
1 requirement.
2 These studies need to be well defined,
3 they need to be carefully planned, and I think,
4 most importantly, they need to be done in a timely
5 manner, so these programs are going to be of any
6 value if we have them shortly after a drug is
7 approved, and long after it is history.
8 I think the way we go about this is to
9 provide appropriate incentives to pharma to do
10 these studies. What I mean by that is I think we
11 should take a page out of the book that exists, we
12 ought to look at what has been done in the
13 pediatric world, and saying that we should give
14 incentives to industry, and say if you do an
15 appropriate postmarketing surveillance study, that
16 you have the potential--and this will be something
17 clearly the Agency cannot do alone, but will take
18 Congress--the potential to have perhaps six months
19 of additional patent protection if these long-term
20 surveillance programs are put into place.
21 I think it is a shame that this country,
22 that spends so much money on health care, can't
23 spend money in determining safety of these drugs we
24 use. The point about this is that if we have a
25 drug that is used, these uncommon events, even with
244
1 the datasets that are as large as we see for
2 NSAIDs, 10- 12,000 patients, we can't rule out an
3 uncommon event that occurs 1 in 4,000 patients,
4 let's say, we will take rule of 3.
5 If we are treating millions of patients
6 with these drugs, which we will, very successful
7 drugs, we have the potential for having thousands
8 of people have an adverse event that may be
9 life-threatening, that could not be detected in the
10 NDA dataset.
11 So, I think we need to develop these
12 surveillance programs, and I think the only way to
13 do it is really to provide the incentives
14 appropriately.
15 [Slide.
16 So, in summary, I would like to say I
17 think we need to stimulate new approaches, and I am
18 glad to see this conference is really focusing on
19 that, different and appropriate methodologies, and
20 I think we need more in the way of safety and
21 outcomes data.
22 I really believe that the way to do that
23 is really through an effective partnership among
24 government, industry, academia, and the public, who
25 are all demanding this.
245
1 Thank you very much.
2 DR. FIRESTEIN: Thank you.
3 The final presentation will be by Dr.
4 Michael Hufford, Vice President, Scientific
5 Affairs, The Science of Patient Experience.
6 While he is getting set up, I would just
7 let the panel know that there is, in addition, a
8 letter from Dr. Shainhouse that will be entered
9 into the record, but will not be read today.
10 Letter from Z. Shainhouse, M.D.,
11 Dimethaid Health Care, Ltd.
12 "As Dimethaid Health Care, Inc. has an
13 interest in topical NSAIDs for symptom relief of
14 rheumatic diseases, we would appreciate the panel
15 taking into consideration the application of any
16 proposed trial models and designs to a topical
17 NSAID.
18 "In trial design for topicals in OA
19 symptom relief, one can use as a model the usual
20 designs for oral NSAIDs. The efficacy variables of
21 pain and physical function, which are used to
22 assess the study joint, are readily studied with
23 topicals. The role of the Patient Global
24 Assessment is less clear.
25 "Questions on Patient Global Assessment
246
1 are often used to inquire about the non-signal
2 joints which are treated simultaneously by oral
3 NSAIDs that provide full, systematic distribution
4 of a therapeutic concentration of drug.
5 "The site-specific nature of topical
6 treatment is unlikely to deliver fully-therapeutic
7 systemic drug levels to provide 'global' benefit to
8 other, non-study joints. Even if one restricts
9 enrollment through trial design, non-study joints
10 may flare during the trial. A Patient Global
11 Assessment for a topical cannot mean the same thing
12 as for an oral.
13 "There are other aspects unique to the
14 study of topicals. Approvability trials, for
15 reasons of practicality and design standards,
16 always study the hip or knee. Topicals are not
17 appropriate for treatment of hips. There is very
18 little literature for oral NSAIDs, let alone
19 topicals, in the treatment of other joints. Do we
20 have sufficient studies on the natural history and
21 spontaneous remission of symptoms in other joints
22 to determine the appropriate duration of study?
23 For that matter, is the now-standard 3-month trial
24 design for OA of the knee or hip based on any such
25 evidence on the natural history of the disease?
247
1 "Clinical experience suggests that where
2 disease is less than bone on bone, symptoms do,
3 indeed, tend to resolve with time - which is
4 perhaps the basis for the usual recommendations to
5 stop oral NSAIDs when symptoms resolve. Is this not
6 further proven by the failure of so many patients
7 to 'flare' during the screening, washout-out stage
8 for drug studies?
9 "The literature describes a significant
10 placebo effect for topicals, thereby complicating
11 study of the onset of pain relief.
12 "In Europe, topical NSAIDs are usually
13 approved and prescribed for the treatment of soft
14 tissue injuries. We are aware of no guidelines for
15 trial design for such studies. Duration would of
16 necessity be shorter because of the self-limited
17 nature of the disorder.
18 "We will appreciate comments from the
19 panel members on the applicability of any
20 guidelines they may propose to the field of topical
21 NSAIDs."
22 "Sincerely, Z. Shainhouse, M.D."
23 [End of letter]
24 DR. HUFFORD: You can see I have tried to
25 rise to the challenge to do a very quick swapout.
248
1 [Slide.
2 Let me begin by saying the company that I
3 work for, In Vivo Data, provides electronic diaries
4 to sponsors in clinical trials, and as such, a
5 number of compounds either are or will be under
6 review by the Agency.
7 [Slide.
8 What I would like to speak to you about is
9 something I have been working on myself for 10
10 years, and my colleagues, for an additional five,
11 using diaries to help patients succeed in providing
12 real-time, real-world data in clinical trials.
13 Of course, diaries are used widely in
14 arthritis trials to capture patients' experiences
15 in a variety of real world settings, and has been
16 mentioned throughout the day today, as well as at
17 the NIH-FDA Conference on Analgesic Drug
18 Development a while back, the collection of pain
19 data in particular, either using the VAS or Rick
20 Graceley's modified VAS scale, is one common
21 implementation, as well as collecting data on
22 functional attributes, stiffness, physical
23 functioning, and nighttime awakenings, and there is
24 good psychometric reasons for this.
25 A number of studies have shown that diary
249
1 data can be more sensitive to medication effects
2 than recall-based reports at the site. One key
3 concern, though, about paper diaries, in addition
4 to the generally poor data quality in terms of
5 legibility, is really noncompliance, because when
6 you use paper diaries, compliance with timely
7 completion if left completely up to the patient to
8 enter the time and date, and you go by that record.
9 Of course, that is very vulnerable to
10 hoarding and falsification, as I am sure many
11 people in this room, including myself when I was a
12 professor, can testify, it is not uncommon to catch
13 patients filling out a week's worth of diary cards
14 immediately before a site visit. Indeed, this
15 happens so often that John Urquhart [ph] has termed
16 it "parking lots compliance."
17 Noncompliance importantly, not only
18 violates the protocol, but it undoes the expected
19 advantage of the diary method because the reason
20 that you implement diaries is to avoid the
21 systematic inaccuracy and bias inherent in recall.
22 It is not pain patient's fault, but simply the way
23 they encode and retrieve information.
24 So, one of the best known biases is
25 patients in a great deal of pain will
250
1 systematically exaggerate their mean pain over the
2 course of the week. Again, it is not fault, but
3 you can't extract yourself from current pain to
4 provide an accurate estimate or recall-based pain,
5 so diaries are used as a way to avoid their recall
6 biases.
7 [Slide.
8 I would like briefly to present a study
9 that my colleagues and I recently published in the
10 March 18th issue of the British Medical Journal.
11 Dr. Arthur Stone, who is the Vice Chair of
12 Psychiatry at SUNY-Stonybrook, what we did is we
13 had two objectives. We wanted to quantify
14 subjects' compliance with paper diaries in a way
15 that was objective really for the first time, and
16 to compare that paper diary compliance to an
17 electronic diary benchmark, something that a number
18 of us, including myself, have been working on in an
19 academic context for over a decade.
20 The endpoints was reported compliance,
21 what patients said they did in terms of telling us
22 about their real-world pain, actual compliance,
23 which we will get to in just a moment, as well as
24 hoarding, that parking lot compliance that I
25 mentioned.
251
1 This was a randomized, parallel, two-arm
2 study with 80 heterogeneous chronic pain patients
3 being assigned to one of two groups, either a paper
4 diary or an electronic diary. What patients didn't
5 realize--and this is actually a sample one--is the
6 paper diary was covertly instrumented, such that
7 photo cells, that we built into the binder, would
8 detect the change in light and write the time and
9 date stamp to an onboard wafer-thin computer chip
10 that we had built into the binder.
11 This was unique insofar as for the first
12 time, you could have an objective documentation.
13 So, the patient said it's Monday at 10:00 a.m. and
14 I am telling you about my pain, well, you could
15 look at the objective electronic record and say,
16 well, is it possible, was the diary even open on
17 Monday for them to complete that report.
18 Again, half of the patients were then
19 assigned to a compliance-enhanced electronic diary
20 with a variety of features that helped them be more
21 compliant with the protocol.
22 It was a three-week pain study with
23 patients completing three reports of their pain,
24 both in the morning, afternoon, and evening, and we
25 asked them to do them at specific times of the day.
252
1 What we found is when you simply look at
2 the paper diary cards, it looks like they were 90
3 percent compliant, that is, 90 percent of the time
4 you had paper diary cards at the date and time that
5 you asked the patient to give the report, so you
6 would be thrilled.
7 Of course, we, for the first time, had an
8 objective records team and could look at actual
9 compliance.
10 [Slide.
11 To our surprise, we thought it would be
12 bad, we didn't think it would be this bad, we had
13 11 percent compliance. So, 79 percent of the time,
14 the patients were not completing the diary card as
15 they told us that they were.
16 [Slide.
17 When we compared that to the patients
18 randomly assigned to use the electronic diary,
19 because one could argue that it was an artifact,
20 chronic pain patients can't possibly be expected to
21 fill out diaries, although we asked them to all the
22 time, what we found is with the variety of
23 compliance enhancing features, we were able to get
24 very high rates of compliance documented over the
25 course of the study, time and date stamp verified
253
1 as required by the protocol.
2 [Slide.
3 So, we looked at the completion of those
4 paper diary cards in batches, trying to understand
5 what happened to those other 79 percent of diary
6 cards. It turns out 1 out of every 3 days, the
7 diary was never even opened. On those days,
8 reported compliance was 96 percent. So, it on the
9 very days that patients forget to do anything with
10 the diary that they are most likely to go back and
11 back-fill a day's or at times even a week's worth
12 of diary cards, so we found a great deal of
13 back-filling really more disturbing to all of us,
14 including myself. Having written the statistical
15 analytic plan, I can tell you that we did not even
16 originally take this into account.
17 We also found forward-filling, that is,
18 there were instances where the patient, say, on a
19 Wednesday evening, would open the diary for about
20 30 minutes. This was a very short pain assessment,
21 only took about 2 minutes to complete. If you open
22 it for 30 minutes and then closed, closed all day
23 Thursday, closed all day Friday, they come in for a
24 site visit on Saturday, and lo and behold, they had
25 Thursday's and Friday's diary cards, so there was
254
1 clear evidence of forward-filling, as well.
2 [Slide.
3 To give you a sense of whether or not the
4 high rates of compliance achieved in the electronic
5 diary group were a fluke, this is a sample of my
6 colleagues and I's peer-reviewed publications, not
7 all of them, but stretching back nearly a decade
8 now.
9 This was the paper compliance at 11
10 percent, the electronic diary compliance at a
11 verified 94 percent compliance, and this is just a
12 sample of some of the work we have done across
13 therapeutic categories showing that patients can
14 succeed in providing real-time, real-world data,
15 but they do need help to do it.
16 [Slide.
17 So, in sum, diary data are critically
18 important to a variety of trials including
19 arthritis trials to avoid retrospective bias that
20 Ike and Rademeyer and Com, and Bradburn, in his
21 famous 1987 Science paper, have outlined so
22 cogently.
23 Paper diaries, though, are vulnerable. In
24 fact, we were able to show objectively both poor
25 and faked compliance using paper diaries. On the
255
1 other hand, electronic diaries with science-based
2 compliance principles can be used to provide
3 documented high, real-time compliance rates. They
4 can also enable more sophisticated diary designs.
5 I don't have time to get into this, but there is an
6 entire field of study called ecological momentary
7 assessment who aim is to densely sample patients'
8 waking experience including dynamic sampling to
9 capture things like time of onset, time to relief
10 in trials.
11 Then, lastly, of course, the validity and
12 integrity in diary data is essential obviously to
13 the evaluation of medication. So, reprints of the
14 British Medical Journal study, I believe have been
15 distributed.
16 Thank you very much for your time.
17 DR. KATZ: May I ask a question, Dr.
18 Firestein?
19 DR. FIRESTEIN: Sure.
20 DR. KATZ: Let me just first congratulate
21 you on a wonderful little study.
22 DR. HUFFORD: Thank you very much.
23 DR. KATZ: I think it is a good example of
24 how methodological issues can be subjected to
25 rational analysis and empirical investigation. We
256
1 so often talk about these important methodological
2 issues, and it is so unusual that we see somebody
3 that actually tries to test a hypothesis in
4 practice.
5 It also matches perfectly with our
6 experience including our published experience in
7 comparing paper and electronic diaries.
8 My question is, were the pain ratings
9 different?
10 DR. HUFFORD: That is one thing we are
11 actually currently pursuing. That has actually
12 taken a tremendous amount of time ironically, to
13 clean and lock the paper diary data. So, that is
14 something that we are working on currently, to look
15 at the psychometric differences.
16 One of the challenges is with the
17 forward-filling in particular, and how to deal with
18 that, but that is something that we are following
19 up on right now.
20 DR. KATZ: Right. We are still cleaning a
21 database that was locked in 1996 from an electronic
22 diary study, it's no small task.
23 DR. FIRESTEIN: Thank you very much for a
24 very provocative discussion.
25 At this point, we are going to take
257
1 another break. At five minutes to 3:00, we are
2 going to start.
3 [Break.]
4 DR. FIRESTEIN: We are going to begin this
5 session with an introduction from Jim Witter.
6 Introduction
7 James Witter, M.D., Ph.D.
8 DR. WITTER: Good afternoon.
9 [Slide.
10 What we thought this afternoon, what we
11 will try and do, and it's going to be an imperfect
12 division, was to make sure that we don't lose the
13 focus on safety, but there is going to be a little
14 bit of a schizophrenia in the sense that we will be
15 talking about some efficacy also this afternoon,
16 and then we will open it up for more general
17 discussion.
18 [Slide.
19 If we were to, for example, take, as I
20 have done here, a line, and on one side of it,
21 write "pain," and the other side "pleasure, we
22 could probably spend these two days just talking
23 about the meanings behind that.
24 What we are interested in really are these
25 concepts of safety, tolerance, and tolerability,
258
1 and as you look, for example, at NSAIDs and opioids
2 as general medicines, they would fall somewhere on
3 this particular line.
4 [Slide.
5 The real question then would be what is
6 the perfect drug and it should be totally safe, but
7 how safe is safe and who should be deciding that,
8 and it should be totally effective, and as we all
9 know, there is no such drug, be it analgesic or
10 otherwise.
11 [Slide.
12 What we thought we should do is take some
13 time to discuss safety and really what we do as an
14 assessment of drug safety, during the development,
15 during the IND phases, before NDA approval--and
16 realize we don't want to confuse on some of these
17 acronyms, but I think we want to use these, so that
18 everybody gets familiar with them if you are
19 not--and then what happens at approval and then
20 after that. We don't want to lose focus on any of
21 these.
22 So, before the NDA is approved, we have
23 preclinical, or I guess we should be referring to
24 this now as non-clinical studies to help guide us,
25 to get some idea of what the profile of the
259
1 compound looks like.
2 Then, we have, as well, various phases,
3 Phases I through III, which enroll larger and
4 larger numbers of patients, and by the time these
5 are completed, if everything has gone well, this
6 information is submitted to us, we look it over, we
7 review it and make an assessment as to whether it's
8 efficacious, really trying to judge effectiveness,
9 and then whether it is also safe enough.
10 If that is approved, then, we have a
11 compound that has a label, and yet that is not the
12 end of the drug's life cycle. There are things
13 that happen post-approval and as Dr. Schnitzer
14 noted before--and maybe we had talked about this
15 beforehand, but we didn't--there really is an
16 incomplete safety assessment when a compound is
17 released, no matter how hard we try, it is just not
18 possible.
19 [Slide.
20 So, we need to be looking at adverse
21 events. As I described, we look at adverse events
22 both before and after approval, and these are from
23 the patients and they are also from the
24 investigators.
25 Now, there has been a discussion, and
260
1 maybe we should have that continue today, that the
2 patient global is also something that should really
3 be intended to catch that something is not quite
4 right experience with an analgesic. Maybe that is
5 what this is best geared for in these particular
6 trials.
7 [Slide.
8 But I think it is safe to say that drug
9 safety is really synonymous with drug information.
10 The more information we have, the better.
11 [Slide.
12 Now, once something is approved, there are
13 various tools--and this important because again we
14 don't catch everything pre-approval--we have this
15 AERS database, adverse events reporting system,
16 which is a passive surveillance system, which has
17 various problems in and of itself, Weber effects,
18 when something is on people's minds, they report
19 it, when it is not, they forget it, but we have
20 other mechanisms, as well.
21 We have abilities to look for drug
22 utilization in certain databases. We can look at
23 external databases for other issues, whatever may
24 be of interest to us. We can look at background
25 incident rates of various adverse events, for
261
1 example, and then we can actually also undergo
2 active surveillance real-time and prospective types
3 of programs, and they have all been employed to
4 some extent.
5 [Slide.
6 So, what these are termed really is risk
7 management tools, and some these then,
8 postmarketing, there are some routine things that
9 we do. For example, we can change the product
10 labeling, we can add adverse events, we can add
11 contraindications, precautions and warnings, and,
12 in fact, the dreaded black box warning.
13 We can make recommendations on monitoring,
14 in fact, we can make this directive - you shouldn't
15 give this until that, for example, follow a lab
16 result, and we can also change indications to make
17 them second line.
18 [Slide.
19 Other things that we can do, which are
20 less commonly done, are to provide patients with
21 information, medication guides as an example here.
22 We can provide clinicians with Dear Doctor letters.
23 We can make public announcements through other
24 forums, such as today.
25 [Slide.
262
1 We can also have patient registries either
2 on a voluntary or a mandatory basis, and there was
3 some discussion about that earlier, too. Then, we
4 can also, and I think this is the thing that
5 everybody tries to avoid, is the product can be
6 withdrawn.
7 [Slide.
8 What are some of the lessons we have
9 learned postmarketing? With regards to labeling
10 changes, there is a feeling that in many ways,
11 these are largely ineffective for widely used drugs
12 because they send out just too complex messages,
13 and that there have, in fact, been failures due to
14 persistent adverse events or studies--some of those
15 active surveillance that I had mentioned
16 before--studies showing that contraindications have
17 been ignored, have led to market withdrawal.
18 Tomorrow, we will be hearing discussion about Durak
19 as an example.
20 [Slide.
21 Patient registries are useful for
22 estimating the denominator, so to speak, in
23 long-term safety. They don't manage risk per se,
24 but certainly overseas I think it is safe to say
25 that they are heavily utilized for gathering safety
263
1 information.
2 So, without further delay, I would like to
3 introduce then Dr. Katz, who will be discussing
4 some of the issues of safety and tolerance with
5 opioids, and then Dr. Lu later will follow with
6 some discussion on some efficacy issues.
7 Tolerance and Toxicity
8 Nathaniel P. Katz, M.D.
9 DR. KATZ: Good afternoon. Let me begin
10 by thanking the Division, Dr. Simon, Dr. Firestein,
11 Dr. Witter, and everybody else for giving me the
12 chance to come and share some thoughts with you
13 about side effects of opioids, also to Drs.
14 McCormack and Rappaport from the other division who
15 have given me an opportunity to gain some
16 experience in the regulatory world on that side.
17 I will be talking about side effects of
18 opioids and what I think are the potential down
19 sides of opioid therapy that are of concern to
20 patients and to physicians, and that need to be
21 understood in order to inform our risk-benefit
22 assessment.
23 I will also be trying to address what we
24 know to date about those potential side effects
25 from the clinical trials that are available.
264
1 [Slide.
2 Let me just begin by saying that when you
3 give a talk just on the down sides of a medication
4 or a class of medications, it may come across as
5 being very unbalanced and that you don't get a
6 chance to emphasize the up side, so let me just get
7 my balance statement out of the way upfront.
8 It has been universally acknowledged now I
9 think, at least in Western medical professional
10 societies, that opioids have an essential, an
11 unreplaceable role at this point in time in the
12 treatment of both acute and chronic pain, and that,
13 in general, they are safe medications.
14 Now, having said that, let me try to
15 expand a bit on the potential down sides of that
16 class of medications.
17 [Slide.
18 Here is what people want to know about -
19 do people get addicted, tolerance, well, I guess
20 that is not really a toxicity, is it, but it is a
21 phenomenon that may result in loss of efficacy over
22 time, potentially side effects, and so it is
23 important to talk about.
24 People are concerned about
25 neuropsychological effects of these medications,
265
1 can people drive, do they lose their ability to
2 function, has their psychomotor reaction time
3 changed, all those sorts of things, can they write
4 their will, can they engage in business, et cetera.
5 Then, there is the plain old garden
6 variety symptoms - nausea, vomiting, constipation,
7 dizziness, sweating, itching, et cetera, et cetera.
8 There are a bunch more. You can pick up any
9 package insert and see what they are.
10 These are the things that are of concern
11 to people, maybe others, and let's see what we know
12 about them in terms of opioid therapy, and I will
13 be focusing mainly on chronic pain.
14 [Slide.
15 Just first to get a couple of definitions
16 out of the way. I am sure that folks in this room
17 know these things, but just to make sure that we
18 are using the same language because language has
19 been a terrible problem in the study of these
20 phenomena.
21 Addiction, which is also known as
22 dependence, psychological dependence, abuse, all
23 related terms, it implies that patients on opioids
24 lose their control over their use of the drug.
25 This is the loss of control model, sort of the
266
1 modern model of what addiction is, compulsive drug
2 use, continued used despite harm.
3 These are things that it is sort of like
4 art or pornography. Everyone knows what it is when
5 they see it, but when you actually try to define
6 it, it is very difficult to come to any consensus.
7 But what we are talking about here is loss of
8 control over the medication.
9 Physical dependence just means that when
10 you stop the drug, you have a withdrawal syndrome,
11 or you suddenly reduce your dose, or you get an
12 antagonist or something like that, and this is
13 something that is expected of people on opioid
14 therapy.
15 It is not an adverse effect per se, it is
16 not connected with addiction in any particular way,
17 and it is just when the terminology was changed
18 from addiction to dependence, it created this
19 confusion between addiction and physical
20 dependence.
21 So, get that out of your mind right now, I
22 will not talk any further today about physical
23 dependence because it is not, as far as I can see,
24 a toxicity we need to worry about if we counsel our
25 patients appropriately.
267
1 Tolerance means less bang for your buck
2 over time in a word, less effective medication
3 after prolonged use, or if you want to look at it
4 the other way, you need to increase your dose in
5 order to maintain the same effect. So, these are
6 the phenomenon that I am going to be talking about.
7 What I would like to add just
8 parenthetically in a moment is that there may be
9 other negative behavioral syndromes of opioid
10 therapy that we don't have good words for, that the
11 syndromologists have not really defined yet.
12 For example, there is something that we
13 all have seen that Steve Passaic is calling "the
14 chemical coper syndrome," where we have all I think
15 seen these patients, where you have a patient on
16 high-dose opioid therapy, they are telling you that
17 they need it and that it is helping them. Their
18 pain score is still a 9 out of 10.
19 If you ask them, well, you know, how is it
20 helping you if it is a 9 out of 10, and they will
21 say it would be a 20 out of 10 without my pain
22 medication. They can't get off of it, they may
23 have subtle side effects.
24 They would give you a positive global
25 satisfaction rating, by the way, to you fans of
268
1 global satisfaction ratings, although their pain
2 relief is not there. These are the patients who
3 may do well after opioid detoxification. Their
4 pain scores may be no different, if not better, and
5 they may feel more alert, et cetera. There is a
6 literature on this.
7 Again, this is not a syndrome that has
8 been well defined, but it is something that we all
9 see, and we can keep it in the back of our minds.
10 I won't talk about it any further.
11 [Slide.
12 So, what do we know about these things?
13 First of all, there is nothing new under the sun.
14 In my worst moments sometimes I think I am the
15 first person to think about these things.
16 Diagoras of Melos, Third Century B.C., a
17 Greek physician, "It is better to suffer pain than
18 to become dependent upon opium." Again, they are
19 talking about the use of opiates for chronic
20 nonmalignant pain. This is what was being
21 discussed in the medical literature of the third
22 century B.C. 2,400 years ago.
23 Again, Erasistratus, if you ever want to
24 look him up, his name is spelled a number of
25 different ways, a Greek physician who actually was
269
1 one of the heads of the Alexandrian School of
2 Medicine in ancient Egypt. Mainly, he got his name
3 through anatomical studies, but he also said opium
4 should be completely avoided, period, and he was
5 referring there to the risk of dependence.
6 At the same time, there were other
7 physicians who were promoting the use of opioids as
8 a cure-all for all sorts of illnesses, again, just
9 showing you this does not give a balanced
10 historical approach, but it does suggest that
11 people have been concerned about these things for a
12 long time.
13 Of course, in the modern era, with the
14 advent of the randomized, controlled trial that has
15 been available to us for more than 50 years now,
16 doubtless we have high quality evidence concerning
17 the incidence of these side effects, and you will
18 soon see the quality of the evidence that we have.
19 [Slide.
20 Now, we do know that opioids are abused,
21 that is no secret to anybody. This is DAWN data
22 and shows the prescription analgesics. This is ER
23 Mentions [ph], for what that is worth, it is gives
24 you some sort of a signal, and it is really of the
25 same order of magnitude as cocaine, a bit less than
270
1 alcohol, far greater than marijuana, et cetera.
2 So, are these patients abusing them, are
3 they addicts who are non-patients? Again, we don't
4 know. We suspect that they are mostly
5 non-patients, but again you will see the quality of
6 the information that we have, clearly, it is an
7 issue.
8 [Slide.
9 In the 70's and 80's, during the era, as
10 was pointed out earlier by Dr. Sunshine, where
11 treating pain with opioids was basically a no-no, a
12 few radical and provocative studies were published.
13 There was one by Medina and Diamond that
14 looked at drug dependency and people treated
15 primarily with intermittent opioids for chronic
16 headaches, pointing out that of their 2,000
17 some-odd patients, few, if any, became addicted.
18 Porter and Jick, this is probably the most
19 famous study which has been quoted millions of
20 times, addiction rare in patients treated with
21 narcotics. This study, published in 1980, again,
22 11,000 some-odd patients treated for acute pain in
23 Boston area hospitals over a period of time, and
24 only something like 4 out of this 11,000 were later
25 on felt to have become addicted to their opioids.
271
1 Then, Perry and Heidrich, another one,
2 similar study, management of pain during burn
3 debridement, use of opioids in many thousands of
4 patients, only rarely was addiction noted.
5 These studies created a new vocabulary for
6 the discussion of addiction with opioid therapy.
7 Now, for the first time in a long time, or at least
8 we thought, we could actually discuss the
9 possibility that maybe opioids are okay for the
10 treatment of pain.
11 Then, at the same time, you had the cancer
12 pain literature that was coming out demonstrating
13 the safety and efficacy of opioids in treating
14 cancer pain. There were a number of retrospective
15 survey studies in non-cancer pain, suggesting that
16 addiction was rare.
17 From this, there created a climate, at
18 least among pain specialists, that you wouldn't get
19 your patients addicted if you gave them opioids for
20 pain, although none of these studies actually
21 addressed the issue at hand.
22 These three studies, the most famous one,
23 the Porter and Jick one, is actually a
24 one-paragraph Letter to the Editor in the New
25 England Journal of Medicine. None of these studies
272
1 actually defined addiction in any way. None of
2 them actually implemented any particular plan for
3 how they were going to detect addiction.
4 They were all retrospective based on the
5 judgment of the physician, and none of them were
6 related to the use of opioids for the treatment of
7 chronic pain. So, again, whether or not opioids
8 are addictive in the management of chronic pain,
9 maybe they aren't, maybe they are, maybe there is a
10 number, but we certainly don't know anything about
11 it from these particular studies.
12 [Slide.
13 It is fair to summarize this at this point
14 and say that no published study of opioids for
15 chronic pain has prospectively evaluated the
16 incidence of addiction by any definition. That is
17 the state of the literature at this point in time.
18 [Slide.
19 There are some methodological issues
20 buried in how one would assess this if one wanted
21 to anyway. There are lot of very thorny
22 methodological issues. The first issue is which
23 population.
24 The studies that I showed you earlier, in
25 general, dealt with a patient population with no
273
1 history of addiction, no psychiatric comorbidity as
2 are most of the randomized, controlled trials that
3 are done today.
4 So, we became interested in what happened
5 if you gave opioid therapy long term for patients
6 with a history of substance abuse, which is
7 probably not an insignificant proportion of the
8 patients that we see in pain management centers.
9 If fact, those prevalence numbers vary between
10 around 3 and 20 percent.
11 This is a retrospective study of all of
12 our patients that we could find who had a history
13 of substance abuse documented in their chart.
14 There were only 20 patients. The bottom line is
15 about half of them did fine and half of them
16 self-destructed. We tried to outline some risk
17 factors for who would be in the good outcome group
18 and who would be in the bad outcome group.
19 The only point I am trying to make here is
20 not that there is a great study either, but that
21 the choice of population determines the results
22 that you see.
23 [Slide.
24 Another very thorny issue is what
25 instrument would you use to measure the rate of
274
1 addiction in patients on opioids for chronic pain.
2 I think the most widely subscribed-to assessment
3 tool for opioid addiction, in the first place, is
4 the DSM-IV or various measurements, the DIS, et
5 cetera, that are based on the DSM-IV, and these are
6 the criteria. You need to have 3 of the following
7 9 symptoms. This is all based on self-report and a
8 doctor-patient interaction, and the self-report is
9 an issue that we will talk about momentarily.
10 But the bottom line is that this doesn't
11 really make sense in people on opioids for chronic
12 pain, and without spending a lot of time going
13 through the details, diminished effect with same
14 dose, does that mean you are addicted? I don't
15 think so.
16 Dose escalation or prolonged use is a sign
17 of addiction. Does that mean you are addicted? In
18 our population, I don't think so. Desire to cut
19 down, excessive time spend obtaining, using, or
20 recovering from use of the substance, well, you can
21 ask most of your patients on chronic pain whether
22 they ever had to spend excessive time obtaining
23 their medication, they have, et cetera, et cetera.
24 So, this it the most well-established
25 criteria, and they are really not relevant to the
275
1 patients that we are looking at, and there actually
2 is no instrument right now that has been validated
3 for detecting addiction in this population although
4 I am happy to say that there is some work being
5 done on that.
6 [Slide.
7 The measures that have been used in the
8 addiction world are based primarily on self-report.
9 Certainly, all the prevalence information that I
10 gave you based on these few quasi-studies are all
11 based on either self-report or impressions of the
12 physician, again based on patients behaviors and
13 patient reporting.
14 What do we know about self-report measures
15 in patients on opioids for chronic pain? There
16 have been four studies, to my knowledge, that look
17 at that. One is the study by Brian Ready, which
18 showed that patients with chronic pain don't report
19 accurately their use of the medications that have
20 been prescribed to them. This was based on
21 inpatient charting by nurses of what the patients
22 were actually given.
23 Another study by David Fishbain comparing
24 self-reported drug use to urine toxicology screens
25 and other measures showing that validity is not
276
1 reliable.
2 We did a study comparing behavioral
3 monitoring of patients to urine toxicology again.
4 I will show you that in a second. There was
5 another study that basically did what we did in a
6 way and confirmed our findings.
7 Again, in our study, I won't spend a lot
8 of time on this, but just very, very briefly. In
9 122 patients from two centers, we instituted urine
10 toxicology monitoring on all patients over a
11 three-year period of time that were on opioids.
12 The bottom line is that 29 percent of our
13 patients had a positive urine toxicology screen.
14 These are patients who had an opioid contract in
15 effect. It said we are not supposed to be doing
16 other things. Twenty-nine percent had a positive
17 urine toxicology screen meaning either illicit
18 substances, cocaine, marihuana, et cetera, or
19 things in their urine that they were not supposed
20 to have.
21 We have them on methadone, they have got
22 hydromorphone. We have them on codeine, they have
23 fentenyl, et cetera. About one-third positive, and
24 if you looked at the monitoring behavioral issues
25 suggestive of inappropriate medication use, about
277
1 22 percent of our patients had inappropriate
2 behaviors of one kind or another, 43 percent either
3 had a positive urine toxicology screen or a
4 suggested behavior.
5 The interesting thing to me is that there
6 is this dogma prevalent in the pain management
7 community that an astute physician, if you monitor
8 your patients carefully and you are attuned to
9 their behaviors, you know what is going on with
10 your patients, you don't need anything fancy, and
11 you can unmask the diverters and drug sellers and
12 criminals and drug addicts simply by your own
13 astute presence and by monitoring self-report.
14 This data suggests that if you only
15 monitored patient behaviors, you miss about half
16 the patients who have a positive urine toxicology
17 screen. I think it is this sort of data, which is
18 also confirmed by this other study I won't tell you
19 about in detail, that confirms, I think in my mind
20 anyway, that self-report measures alone, if you are
21 trying to monitor for noncompliance anyway, are
22 inadequate.
23 I should issue a very quick caveat just so
24 that I don't give the wrong impression. We were
25 not measuring addiction in this study. I don't
278
1 have any idea of the extent to which these signs
2 correlate with addiction. As far as I know, none
3 of these patients were addicted, but certainly if
4 somebody on opioids has cocaine in their urine or
5 they have opioids that they are getting from
6 another source, that is something that I think I
7 want to know about.
8 [Slide.
9 Another potential source of external
10 information outside of patient self-report that has
11 not really been talked about as a patient
12 monitoring tool on a formal basis, is the whole
13 idea of using prescription monitoring program data.
14 Many of you know that right now I think it
15 is 19 states in the United States have prescription
16 monitoring programs that track some or all of the
17 scheduled medications that these patients are on.
18 In Massachusetts, we have a prescription monitoring
19 program that tracks only Schedule II data, and not
20 any other scheduled medications.
21 So, the idea of using this as a way of
22 getting verification of patient self-report of
23 compliance has really not been pursued, and there
24 is a lot of interesting data buried in these
25 prescription monitoring programs that could be
279
1 used.
2 For example, we found--we are just
3 starting to validate this database--in
4 Massachusetts, in the year 2000, there were over a
5 million Schedule II opioid prescriptions that were
6 given. There is only 6 million people in the State
7 of Massachusetts, which is interesting, and it
8 looks like there were about half a million unique
9 individuals in Massachusetts that got a
10 prescription for opioids.
11 Now, this database happens to exclude the
12 VA, which is probably not a small issue, and there
13 are a few other exclusions, as well. So, about 9
14 or 10 percent of the Massachusetts population got
15 Schedule II opioids. If you include the other
16 schedules, that probably would double, triple, or
17 quadruple this number.
18 Before I started looking at this, there is
19 really no notion of the epidemiology of opioid
20 therapy, and we do have information on this
21 database on what proportion of people have five or
22 more prescribers, what proportion of people use
23 five or more pharmacies, what proportion of people
24 run out of their day's supply early every month.
25 We can get this data, and we are hoping to
280
1 actually report these numbers as our work goes on.
2 I think one could consider even using this in a
3 clinical trial or postmarketing or risk management
4 program to look at noncompliance.
5 I am going to leave the issue of addiction
6 there with the unfortunate conclusion that we don't
7 know a lot about the incidence of addiction in
8 patients given opioids for chronic pain.
9 [Slide.
10 Tolerance is another issue and also it
11 seems so easy when you first look at it, and then
12 it gets very complicated when you try to figure out
13 exactly what you mean by tolerance and how you are
14 going to measure it.
15 This is just a concept slide to give you a
16 sense for how one might think about tolerance and
17 begin to approach the idea of how to measure it.
18 Look at these green lines here for a minute. These
19 are little graphs looking at--and this is all
20 invented out of my mind, this is not clinical trial
21 data, this is all conceptual--this is the dose
22 required to produce analgesia over time.
23 In an ideal world, a medication that did
24 not produce tolerance would have a flat line. Here
25 is a different way it might go. You might have a
281
1 bit of a dose escalation at the beginning and then
2 you might be stable over time, in fact, there is a
3 school of thought that suggests that this is what
4 happens to most people on chronic opioid therapy,
5 or it might escalate over time, or it might
6 escalate faster over time.
7 So, this is fine. Looking at dose
8 escalation is a perfectly good place to start I
9 think if you allowed patients to free titrate to
10 the dose that gives them adequate analgesia.
11 The complexities start to emerge, though,
12 and one of the complexities is side effects.
13 Because the usefulness of the drug, or if you want
14 to call it the therapeutic index of the drug,
15 really depends upon having a dosage range for an
16 individual patient where they can get adequate
17 analgesia without intolerable side effects, that is
18 what we are talking about.
19 If that difference between the dose they
20 need for analgesia and side effects remains in a
21 useful range, that is more useful sign of a
22 medication that is not associated with problematic
23 tolerance. Of course, if both of them escalate
24 equally, then, that is fine, too.
25 Tolerance might even be a good thing. For
282
1 example, we know from clinical experience that
2 people often become tolerant to nausea and
3 dizziness and neuropsychological side effects, and
4 other bad things, so you may find that, in fact,
5 tolerance can work in your favor. Your therapeutic
6 index may broaden over time.
7 On the other hand, it is conceivable that
8 your does that you need for analgesia increases,
9 but you don't become as tolerant to the side
10 effects, in which case you crash and burn on your
11 drug. They maybe is someone who drops out of your
12 clinical trial.
13 Unless these things are assessed, unless
14 you are assessing adequacy of pain relief, unless
15 you are assessing overall tolerability of your
16 drug, which is never done to my knowledge, and you
17 are modeling how those go over time, then, you
18 can't really say anything about tolerance or you
19 can't make a sophisticated statement about
20 tolerance, to my view.
21 [Slide.
22 So, what do we know from clinical trials?
23 This, sorry to say, I know nobody can read this,
24 but it is just there to give you a visual
25 impression, anyway, these are all the randomized,
283
1 controlled trials that have been published using
2 non-opioid comparators, placebo or a non-opioid,
3 for chronic, non-cancer pain where we are watching
4 the patients for at least one month. I think that
5 is a reasonable benchmark if you are having a
6 discussion about tolerance.
7 These are all the ones in the published
8 literature. For those of you with good eyes, if I
9 have forgotten one or two, then, you can come up
10 and yell at me after we talk, but this will give
11 you a good visual.
12 I put the asterisks next to the trials
13 where you can learn something about tolerance from
14 the trial, usually because there is a prolonged,
15 so-called open label extension period where
16 patients are watched open label on their drug for
17 some period of time.
18 I will just briefly highlight what it is
19 that we know. Again, here is one trial where pain
20 relief was stable at 19 weeks, don't have dose
21 information, and again, in all these trials, a
22 blurb doesn't really do justice, and you can learn
23 a lot more from getting to the trials themselves.
24 There are people in the room who have been involved
25 with these trials who could probably educate us
284
1 further about them, but just to give a visual.
2 Here, this is the trial that we did. We
3 found that actually in our patients, only 36 dose
4 and pain relief were stable after an initial period
5 of escalation. This is the Watson and Babul, Najib
6 Babul addressed this earlier today, their very nice
7 study of oxycontin for postherpetic neuralgia.
8 Again, in their open label extension,
9 there was a small subgroup of patients--Najib, you
10 will have to remind me--I think it was about 11 or
11 so out of the 50 patients were still there at the
12 end of follow-up, still enjoying analgesia, and you
13 can go on down the line.
14 The bottom line is that as you follow
15 patients out, here is an example, about 18 months,
16 only 15 of 106 patients still in the trial, still
17 getting good analgesia, still at a stable dose.
18 I think what these sorts of studies tell
19 us is that although none of these studies have
20 actually, to my knowledge, said we define tolerance
21 in this way, this is how we are going to measure
22 it, this is our result. That has never been done,
23 to my knowledge. Somebody can challenge me if they
24 think I am wrong about that, but all we can get is
25 an indistinct window about what happens long term.
285
1 It looks like only a minority of patients
2 are still on drug over time. Now, should we expect
3 that everyone should be on drug a year later?
4 Obviously not. If you look at trials of NSAIDs for
5 osteoarthritis, you are also not going to have
6 everybody on trial at the end of a year because
7 that's not how it works.
8 People get better people get worse and
9 drop out, people move to Florida, people die of a
10 heart attack, all sorts of things happen to people,
11 but it still suggested to me that--it doesn't
12 really reassure me that tolerance is not a problem
13 in clinical practice--and it suggests to me that we
14 need a methodology for evaluating this
15 prospectively with some rigor.
16 Interestingly, this study, which I put in
17 italics, is a study of tramadol. I excluded
18 tramadol except for this one study for patients
19 with painful diabetic neuropathy, 117 patients.
20 Tramadol is a drug that is an opioid and a
21 non-opioid in the same drug, and clinically
22 speaking, we don't think tramadol is associated
23 with tolerance or at least not much.
24 Interestingly, only 4 out of 117 patients
25 at six months dropped out due to lack of efficacy,
286
1 which is interesting because that is dramatically
2 different than what we see in the trials of the
3 pure new agonist, and it makes me wonder whether
4 the fact that only a small number of patients are
5 in these new agonist trials is indeed indicative of
6 tolerance developing because we didn't see that to
7 the same extent in the tramadol study.
8 [Slide.
9 Now, this is all speculation, nuance. I
10 think really the only robust conclusion is that we
11 need to start measuring tolerance. Again, just to
12 give you a quick visual of that, what we often see
13 in the way these studies are reported--and again
14 this is whitewash data of not any particular drug,
15 is that as the months wear on, the patients' dose
16 or their pain score, if you want to look at pain
17 scores, remains stable, but the trick is that only
18 a small fraction of the patients are present here
19 that started here, and we no doubt have informative
20 censoring, and can't say too much about long-term
21 efficacy from this type of report.
22 [Slide.
23 In my view, it is fair to say that the
24 phenomenon of tolerance to opioids in the treatment
25 of chronic pain has not been systematically
287
1 investigated in the published medical literature.
2 [Slide.
3 Neuropsychological function, I outlined
4 the concerns earlier. I am not going to really
5 speak about that because again, there is actually
6 no published prospective controlled trial on
7 opioids for non-cancer pain that has evaluated
8 neuropsychological function.
9 There is a published uncontrolled trial
10 where patients on a hodgepodge of opioids were put
11 on controlled release opioids. That is Jennifer
12 Hathorne Waites [ph] trial that actually suggested
13 in that setting, neuropsychological function
14 improved.
15 There is a study that, Mitchell, you
16 alluded to earlier that you did with Raja and those
17 folks that is still unpublished, that I have heard
18 rumors about, that I have heard rumors is going to
19 reassure us all about neuropsychological function
20 measured in a prospective way.
21 I, myself, have been involved in yet
22 another unpublished trial that I hope will come to
23 light soon, that also will find reassuring, so I
24 think that this is going to probably work out okay,
25 but at this point in time, this remains the fact of
288
1 the matter.
2 [Slide.
3 One final note on another sort of occult
4 toxicity that has been getting a little more press
5 lately, but hasn't really been addressed formally,
6 is the whole issue of opioids in endocrine
7 function. I think this is actually a very big
8 deal.
9 It is known that in animals, every animal
10 endocrinologist knows this. When I go up an animal
11 endocrinologist and I say, you know, I am a little
12 concerned about opioids and testosterone, they say,
13 da, what are you talking about, we have known about
14 that for 100 years already, about opioids and
15 testosterone.
16 It is known that opioids lower
17 testosterone and actually have other endocrine
18 effects, as well, in animals. There is one study on
19 heroin addicts showing low testosterone levels, one
20 study on methadone maintenance patients showing low
21 testosterone levels, and two studies now of
22 patients on intrathecal opioids showing profoundly
23 lower testosterone levels in men who develop a
24 central or hypogonadotrophic hypogonadism on
25 intrathecal opioids.
289
1 In the intrathecal studies, those were the
2 only ones that tried to address symptoms, and it
3 does turn out that loss of libido and impotence are
4 associated with low testosterone seen in those
5 trials.
6 In one of the two trials, it was actually
7 a pre-post study where they measured endocrine
8 function before going on intrathecal opioids and
9 then after, showing the declines, so very
10 interesting information. We have known about that
11 anecdotally for a while. In women, we see
12 amenorrhea and infertility, and other things.
13 What are the symptoms of low testosterone?
14 Fatigue, loss of muscle mass, you don't want to get
15 up and go, mood disturbances, osteoporosis and
16 compression fractures, so a potential public health
17 hooked to this.
18 So, has anyone seeing patients with
19 chronic pain ever seen any of these symptoms in
20 anybody? I think that these symptoms are basically
21 universal. So, you would think that somebody would
22 have asked the question of what proportion of
23 patients on opioid therapy for chronic pain have
24 low testosterone levels. You would think that that
25 question would have been asked.
290
1 [Slide.
2 This is preliminary data from our group,
3 our data, trying to address this question. Again,
4 I am always a little bit nervous about presenting
5 unpublished and non-peer-reviewed data, but I think
6 this is big enough to at least flag your interest
7 in this area.
8 All of my patients on opioid therapy for
9 nonmalignant pain had to undergo an endocrine
10 battery of blood tests at least once a year, and
11 this has been going on for about four years now.
12 There were complete enough data available on 25
13 males. I haven't tried to understand the female
14 data because it is just too confusing.
15 We found that free testosterone, which I
16 think is the more sensitive of the two, was below
17 the reference range in 63 percent of our patients
18 age 25 to 49. This is how the normal testosterone
19 levels come packaged at least at our institution,
20 25 to 49, and 50 to 75.
21 Free testosterone levels were below the
22 reference range in 88 percent of patients age 50 to
23 75, the older group, and our mean LH and FSH
24 levels, compared to normal controls, were below
25 normal, suggesting that the majority of our
291
1 patients had central hypogonadism, were on opioids
2 for chronic pain.
3 We looked at mean levels compared to
4 healthy controls, et cetera, and also found that
5 they were low.
6 Again, I think this is very provocative
7 and needs to be followed up further by a properly
8 controlled trial, and suggests to me anyway that
9 endocrine dysfunction may actually be the major
10 organ toxicity of opioid therapy.
11 [Slide.
12 Let's not forget about the little
13 symptoms, the garden variety symptoms I spoke about
14 earlier - nausea, vomiting, blah-blah-blah. In
15 clinical trials, we all know how these side effects
16 are captured. They are captured by the passive
17 capture methods. The patient has to raise their
18 hand and speak up and say I am dizzy or I am
19 nauseous.
20 Then, the study coordinator has to write
21 it down. Then, it has to be coded by somebody and
22 put in the database. We know from a variety of
23 sources of information that passive side effects
24 captured like that are inadequate in the sense they
25 don't nearly tell you what you would find if you
292
1 asked patients how they are feeling.
2 We know that dropouts due to symptomatic
3 side effects are substantial in both acute and
4 chronic pain trials of opioids, and the chronic
5 pain trials that I see, that range from 10 to even
6 50 percent, so it has got to be that these inform
7 the risk-benefit analysis of opioids for chronic
8 pain.
9 We also know that if you look at--I am not
10 going to take the time to present data--but if you
11 do symptom distress assessments prospectively by
12 giving patients a checklist on how they are
13 according to a variety of symptoms, and how severe
14 they are, you can find out a lot more, and you can
15 actually get data that predicts dropouts more
16 accurately than just passive side effects captured,
17 and there are some very nice studies by Richard
18 Anderson and Marsha Testa and other people showing
19 that these are very sensitive measures of how
20 patients are doing.
21 You would think that somebody would have
22 asked the question about how patients with opioids
23 do if you give them a prospective symptom checklist
24 to inventory. We did that in at least a
25 preliminary way in our study that came out in 1998
293
1 of patients and back pain.
2 We gave them a checklist like this, it had
3 20 items. It had them rate none, mild, moderate to
4 severe, and got a lot of interesting information,
5 which I won't take the time to give you, but one of
6 the interesting things was that we were able to
7 discriminate side effects intensity scores between
8 a high dose and a low dose opioid regimen and also
9 from a nonsteroidal anti-inflammatory drug regimen.
10 So, this checklist analysis did
11 discriminate between regimens. We also found
12 interestingly--I don't really know how to
13 understand this--people on low-dose opioids had
14 fewer side effects, but were more bothered by them,
15 people on high-dose opioids were less bothered by
16 their side effects, strangely.
17 So, it seemed like maybe opioids
18 influences how much you are bothered by whatever it
19 is that ails you. Maybe you understand that better
20 than I do. Anyway, do this, that is what I am
21 trying to say.
22 [Slide.
23 I will end with just a quick comment on
24 the use of opioid sparing as an outcome measure
25 since that was mentioned as a question in the
294
1 background materials, so everybody knows what this
2 means. You have a drug X compared to placebo or
3 some comparator, and you look at how much opioid
4 the patients in both groups use in outcome measure,what does
5 that mean, is that good, is that bad.
6 First of all, just conceptually, if a
7 patient in one treatment group has decreased opioid
8 requirements, there is a few things that could be
9 due to. The first, which is the one that we are
10 all interested in, is that your study drug is an
11 analgesic. That is good, and the obvious examples
12 there are NSAIDs compared to placebo in
13 postoperative pain, where patient controlled
14 analgesia or other things are very nice
15 discriminative analgesic effect.
16 The other possibility is that your drug is
17 not an analgesic by itself, but together with
18 opioids, enhances opioid analgesia, and some people
19 think that are some NMDA receptor antagonists that
20 might do that. It is hard to discriminate between
21 an analgesic and an opioid enhancer in that sort of
22 model.
23 The other possibility I will just mention,
24 although you maybe you won't like hearing it, is
25 that the study drug, all it does is enhance opioid
295
1 side effects, so that patients can't use as much,
2 and that certainly is a conceptual possibility
3 although one should be able to tease that out by
4 looking at pain scores and by looking at side
5 effects, if you look at side effects in an
6 appropriate way, which is often not done.
7 So, you have to be able to provide
8 supportive data to classify what is going on in
9 terms of these possibilities, should you have
10 opioid sparing.
11 [Slide.
12 Lastly, is opioid sparing meaningful in
13 your clinical trial. I am remind of the
14 expression, "A difference is only a difference if
15 it makes a difference," and so if you do reduce
16 your opioid dose, does that mean anything.
17 Well, I think it does mean something if
18 the scientific question is whether the drug has
19 analgesic activity in the model that you chose, so
20 for a proof of concept trial, for example, if you
21 are just trying to show does your drug have
22 analgesic effects or not, given the caveats I
23 mentioned earlier, you know, I think that answers
24 your question, but if you are trying to show does
25 the treatment help the patient, which I think
296
1 ultimately is what we need to have an evidentiary
2 body of information about, the answer is no, by
3 itself, if you are on 10 milligrams of morphine or
4 20 milligrams of morphine, that doesn't mean you
5 are better or not better.
6 You need to show I think, in my opinion,
7 if you are interested in whether the patient is
8 benefiting, some benefit, which could be decreased
9 pain, it could be decreased side effects, which
10 again you are not going to get unless you address
11 in an aggressive way.
12 By decreased pain, we have to be a little
13 bit careful there. The example that comes to mind
14 for me is that we know that in the postoperative
15 setting, opioids work pretty well for rest pain,
16 but not as well for movement-associated pain,
17 whereas, NSAIDs tend to work well for
18 movement-associated pain, maybe even better than
19 opioids in some circumstances.
20 In the postoperative world,
21 movement-associated pain is where the rubber meets
22 the road, because patients get up and rehab
23 themselves and ship themselves out of the hospital
24 these days.
25 So, one could conceive of showing benefit
297
1 of NSAIDs by focusing specifically on
2 movement-associated pain compared to an opioid-only
3 regimen as opposed to just global pain. As people
4 were saying earlier, just looking at global pain,
5 you may miss the boat on something important.
6 So, I think that opioid sparing, by
7 itself, needs to be looked at very carefully, and
8 you have to really address the scientific question
9 of the study by looking at clinical benefit.
10 [Slide.
11 In conclusion, opioid toxicity, just to
12 recapitulate, opioids are generally safe
13 medications. We don't have 17,000 patients a year
14 dying of GI bleeding in the United States from
15 opioids.
16 So, looking at the big picture, opioids
17 are generally safe medications. I think it is fair
18 to say that the treatment response does appear to
19 be durable in a subgroup. How large is that
20 subgroup, I don't know, and again, tolerance has
21 really not been systematically looked at in any
22 published studies.
23 In my view, symptom distress scales or
24 toxicity scales, especially trying to look at why
25 people drop out, so that you don't have informative
298
1 censoring going on, must be used to assess the
2 overall treatment effect.
3 Addiction, the major concern in chronic
4 treatment I think has not been investigated, in my
5 view, using any legitimate methods, and
6 endocrinopathies may, in fact, wind up if this
7 preliminary data pans out to be actually the major
8 organ toxicity of opioids as we go forward.
9 Thank you for your attention.
10 DR. FIRESTEIN: Thank you very much, and
11 we will have an opportunity to discuss some of this
12 in a few minutes during our open discussion after
13 the next talk, which is Statistical Issues for
14 Measurements by Dr. Lu.
15 Statistical Issues for Measurements
16 Laura Lu, Ph.D.
17 DR. LU: Good afternoon. I am going to
18 discuss issues in time-specific measurements and
19 time-weighted average for pain in chronic and acute
20 analgesia trials.
21 This discussion is to set a stage for
22 tomorrow's further discussion of endpoints.
23 [Slide.
24 First, I am going to introduce
25 time-specific measurements and time-weighted
299
1 average. Then, I will discuss issues in chronic
2 analgesia trials for those measurements in terms of
3 interpretation of drug benefit and data imputation
4 methods, and the parallel issues in acute analgesia
5 trials. At the end, I will provide summary.
6 [Slide.
7 I will use an individual patient's pain
8 curve to illustrate those measurements I will talk
9 about. Suppose a patient's pain was evaluated at
10 time 2, 4, 8, and 12, and these vertical segments
11 represent change from baseline in pain scores at
12 each specific time 2, 4, 8, and 12. So, these are
13 what I call time-specific measurements.
14 I will refer to the area under this pain
15 curve as AUC later.
16 [Slide.
17 I denote those time-specific measurements
18 for change from baseline in pain as d1, d2, d3, and
19 d4, and the time intervals between each
20 neighborhood measurements as t1, t2, t3, and t4.
21 [Slide.
22 The time-weighted average can be defined
23 as AUC divided by the patient's treatment period.
24 In another form, it can be also described as a
25 weighted average of time-specific measurements, and
300
1 the weights are decided by the neighborhood
2 intervals of disorder and the treatment period.
3 That is why we call this normalized AUC
4 measurements as time-weighted average, and one-time
5 weighted average is used as an endpoint we quite
6 often refer to it as AUC approach.
7 [Slide.
8 Now, the issues in chronic analgesia
9 trials. First, the interpretation of drug benefit
10 by those measurements.
11 [Slide.
12 End-of-the-trial measurement is a
13 time-specific measurement. It is commonly used in
14 chronic analgesia trials. It measures drug effect
15 at the end of the trial. Time-weighted average is
16 another endpoint being used. It measures average
17 effect through the trial.
18 The two measurements actually describe
19 different aspects of drug effect, and no matter
20 which measurement is used at the endpoint, the
21 consistency of drug benefit over time is always an
22 important review issue.
23 [Slide.
24 As shown in this graph, when two
25 treatments switch advantage over time, then, there
301
1 is inconsistent drug effect. In this situation,
2 none of the single measurements can really describe
3 drug benefit for one over the other.
4 [Slide.
5 When there are missing values, for the end
6 of the trial measurements, the last observation
7 carried forward is a commonly used imputation
8 method. It imputes measurement at withdrawal time
9 to later period.
10 For time-weighted average, one would say
11 that there is there is no imputation as long as
12 there is at least one post-baseline measurement,
13 but actually, it is not true.
14 When the patient dropped out earlier, the
15 average treatment effect before withdrawal time
16 will be used to represent the average effect in
17 overall treatment period. So, this is a form of
18 imputation.
19 [Slide.
20 Both of the imputation methods imply
21 assumptions that later evaluations of drug efficacy
22 is similar to that of earlier evaluation. This is
23 a very artificial assumption, and cannot be
24 verified by data we have seen.
25 Also, the results generally favor drug
302
1 with imputation than without imputation due to
2 different dropout mechanisms in treatment groups,
3 for example, different dropout rates and dropout
4 reasons.
5 [Slide.
6 We have seen those problems with
7 imputation methods. Can we make any improvements
8 in terms of trial design and data analysis? First,
9 I think we should continue efficacy evaluation even
10 after a patient drops out even the patient is on
11 rescue medication, and these measurements can
12 provide additional treatment information, so a true
13 ITT analysis can be performed.
14 Also, if a clinically sensible responder
15 analysis can be performed like a definition can be
16 found, now, we can perform responder analysis in
17 terms of time to respond, percentage of responder,
18 and duration of response.
19 A responder analysis may better
20 characterize drug effect and avoid artificial
21 imputation methods by taking into account of
22 dropout status.
23 [Slide.
24 Parallel issues in acute analgesia trials.
25 [Slide.
303
1 In single-dose acute analgesia trial, we
2 focus on onset, duration, and pain curves. For
3 multiple-dose acute trial, we focus more on
4 duration of effect.
5 [Slide.
6 In single-dose trials, time-specific pain
7 measurements provide more information about onset
8 and duration, but time-weighted average
9 measurements, such as some of pain intensity
10 difference or some of pain relief and intensity
11 difference do not.
12 So, in single-dose trials, we prefer more
13 of the time-specific pain measurements over
14 time-weighted average. In multiple-dose trials,
15 time-specific measurements and time-weighted
16 average face similar issues as those in chronic
17 analgesia trials, so I will only focus on the
18 imputation methods for time-specific pain
19 measurements in single trials.
20 [Slide.
21 The three commonly used methods we have
22 seen for data imputation are these three -
23 last-observation-carried- forward approach,
24 baseline-observation-carried-forward, and
25 worst-observation-carried-forward methods.
304
1 The last two methods are generally more
2 conservative than the
3 last-observation-carried-forward approach, but all
4 these three approaches are very unrealistic by
5 carrying forward earlier pain intensity scores into
6 later period. This is against the self-limiting
7 nature of acute pain.
8 [Slide.
9 I will use this example to show the
10 artificial effect of those imputation methods.
11 This is not a real example, but it represents the
12 common scenario we have seen in trials.
13 Suppose patients' pain was evaluated for
14 24 hours after dental surgery, and these two curves
15 represent the mean pain intensity a long time for
16 placebo and the treatment group. These are
17 observed curves without any data imputation.
18 Because of the short duration of dental pain, at
19 the end of 24 hours, no matter how many patients
20 left in the trial, the patients' pain will be very
21 mild, so the mean scores approach zero.
22 [Slide.
23 Now, if we use early pain intensity scores
24 to impute later period, these two red curves
25 represent the imputed curves for pain intensity,
305
1 and then we got the impression that at the end of
2 the day, the patients are still in pain and also
3 the drug is still effective over placebo, this
4 artificial effect is caused by different dropout
5 mechanism. Mainly it is because more placebo
6 patients drop out in the early stage, and also most
7 of those patients drop out due to lack of efficacy.
8 [Slide.
9 In summary, for chronic analgesia trials,
10 end-of-the-trial measurement and time-weighted
11 average represent different aspects of drug effect,
12 and consistency of drug benefit through the trial
13 is always an important issue for review.
14 In acute analgesia trials, time-specific
15 measurements are more informative than
16 time-weighted average in single-dose trials.
17 [Slide.
18 We should continue to measure efficacy
19 even after patients withdraw, even after patient is
20 on rescue medication, and these measurements can
21 provide additional treatment information for drug
22 effect.
23 Also, if we can come up with clinically
24 sensible responder definition, we can carry out a
25 responder analysis, which may better characterize
306
1 drug effect and avoid artificial imputation by
2 taking into account the dropout status.
3 Thank you.
4 DR. FIRESTEIN: Thank you very much.
5 Open Discussion of Points #1, 2, 3, 4 and 5
6 DR. FIRESTEIN: Now, we come to the time
7 at the end of the say where there is a spirited
8 discussion, and we can resolve all of the issues
9 that have been raised, so that the FDA can go ahead
10 and make its formal recommendations.
11 Before we move ahead, I just wanted to try
12 to briefly summarize some of the points that have
13 been brought up and then open them up for
14 discussion.
15 One of the issues was the notion of
16 whether or not separate acute versus chronic pain
17 indications has utility not only for drug
18 development, but also for our patients compared
19 with simply a single indication for pain, and also
20 whether or not this should be more mechanism versus
21 clinical indication oriented.
22 With regard to the chronic pain
23 indication, a proposal was put on the table that
24 this could potentially be achieved with a very high
25 bar where three separate indications would be
307
1 looked at, each with two studies and each involving
2 three separate domains.
3 Notably, there were a couple of
4 alternatives that were proposed during the open
5 discussion or the public forum, one involving two
6 separate indications and then another involving
7 four separate indications, but with only one study
8 for each one.
9 Then, we talked about low back pain,
10 whether or not that would be one of these potential
11 clinical indications for chronic pain, and, in
12 particular, whether or not all low back pain could
13 be lumped together or whether or not there is some
14 rationale for taking the vast majority, which is
15 mechanical low back pain, and then using that as a
16 separate location.
17 Finally, we have talked a bit about safety
18 and the issues regarding dose and indication creep,
19 as well as off-label use. That was raised a number
20 of times.
21 So, those are I think the major issues
22 that are before us right now.
23 DR. MAX: I would like to return to the
24 issue of mechanism-based diagnosis and ask my FDA
25 colleagues about some possible incentives for this.
308
1 If we go back to Dr. Woolf's talk, he
2 mentioned several dozen molecules involved in pain
3 processing, and actually, we could probably get
4 very close to some mechanisms in patients right
5 now, because imagine, let's say we have the results
6 of a large chronic pain trial, say, in back pain
7 with some novel drug that works on one of those new
8 mechanisms, and overall, there is just
9 nonsignificant trends towards efficacy.
10 However, it is already known that probably
11 half a dozen of the molecules Clifford was talking
12 about this morning have common human polymorphisms
13 with two forms of the molecule, either one made in
14 higher volume expressed with a molecule expressed
15 more or with higher functioning levels of the
16 molecule and with some very common people with less
17 expression or less functional forms of the
18 molecule.
19 So, what if the company could for a few
20 cents an assay take all the pain molecules and
21 characterize the patients as high functional or low
22 functional for that, so what if they do that for a
23 number of different molecules and found that if
24 they just take the subset, say, with a hyperactive
25 NMDA NR2B molecule function polymorphism, in those,
309
1 the drug really was effective.
2 So, now they have found by dredging a
3 prospective mechanistic-based subset, so they come
4 to you and say, okay, could we now go and do one
5 more study and get approval for this, what might
6 you say to a company like this?
7 DR. GOLDKIND: We might say a number of
8 things. I think that the assay that would
9 differentiate a responder or potential responder
10 from a non-responder has to be something clinically
11 available, so that a doctor can use that in
12 guidance, so it has to be referable to the
13 population. It wouldn't really help a doctor or
14 patient if they didn't have that.
15 In terms of the evidentiary base, is an
16 exploratory analysis adequately supportive of a
17 prespecified primary outcome for a second trial,
18 that has been used before. There is not a global
19 answer to that question, but that is what you are
20 describing is an analysis where a subpopulation is
21 looked at and where you are exploring for an effect
22 on subpopulation, and you identify one, and then
23 you confirm that in a second study.
24 That, I would say is really dealt with on
25 a case-by-case.
310
1 DR. MAX: With regard to that, I think,
2 you know, the tests themselves now cost like about
3 25 cents a genotype, so the company might even
4 provide that. To say just that you need one new
5 trial for it, that sounds pretty encouraging,
6 because if I just came up and dredged a database
7 with a new hypothesis, I think your earlier
8 guideline, Lee, would suggest you are starting from
9 scratch and you should have two trials for
10 replicate evidence for a new indication. So, if
11 you said that, that would be very encouraging.
12 DR. SIMON: Well, let's be clear. I
13 always like being clear. What we did propose was
14 that mechanistic models that had clinical relevance
15 would be acceptable without further definition of
16 the number of trials that would be necessary. We
17 don't know yet how to go about this. One could even
18 envision that the argument could be that such a
19 design would lead to a definition in only
20 subpopulations, and it would not be extrapolatable
21 to the general population.
22 The down side would be that. The up side
23 would be, well, so what. You have identified a
24 patient population that would respond, you have a
25 clinically measurable test that is clinically
311
1 applicable and accessible to the treating
2 clinician, so therefore, you can identify the
3 patient that could potentially respond, and that
4 should be something that should be rewarded.
5 We would believe that that should be
6 rewarded. There is nothing in our presentation that
7 precluded a unique way of going about this. All we
8 suggested was in a traditional trial design, that
9 the three-model, two-replicate, three co-primaries
10 would be important.
11 But if a mechanism could be defined, could
12 be reproducible, and could be clinically applicable
13 and available, then, I think all bets are off.
14 DR. FIRESTEIN: I think the key point is
15 that it must be clinically applicable.
16 DR. DAVIDOFF: I was going to say that I
17 have a feeling that the statisticians in the room
18 are having acute epigastric pain hearing that by
19 dredging a single database, you can, in fact, have
20 the basis for approval.
21 I would think that that should be handled
22 with extreme caution and that there should be
23 required at least one replication of a planned
24 trial.
25 DR. WOOD: I would like to return to the
312
1 opiate sparing issue. I was very concerned that
2 there has been absolutely no discussion of the
3 underlying assumption in these studies, and the
4 underlying assumption in these studies is that
5 there is no alteration in the pharmacokinetics of
6 the opiate induced by the co-administered drug.
7 That may seem somewhat obscure, but when
8 you recognize that erythromycin would be an
9 extraordinary effective opiate sparing drug if
10 administered with fentenyl or that inducing
11 codeine's metabolism to morphine would be extremely
12 effective by some drug with no primary analgesic
13 effect, or more subtle changes, like we can turn
14 Imodium, the anti-diarrheal drug, into a very
15 potent analgesic and a very potent opiate by simply
16 inhibiting the transporter responsible for normally
17 keeping it out of the brain.
18 The ability to have unrecognized effects
19 that have nothing to do with analgesia, I think are
20 substantial. In addition, some of the metabolites
21 that are produced from these drugs produce
22 toxicity, and if they accumulate or are induced,
23 they are likely to produce side effects that may or
24 may not be recognized as being due to the
25 metabolites.
313
1 So, it seems to me that there is an
2 absolute necessity in an opiate sparing trial that
3 we have a standard that dictates that the drug does
4 not produce some pharmacokinetic interaction. That
5 is tough actually. It is relatively easy to define
6 the obvious ones like the drug concentration in
7 plasma doesn't increase.
8 It is much harder to do that in, for
9 example, supposing Imodium was on the market--well,
10 it is on the market over the counter--we can turn
11 Imodium into an extraordinarily potent sensory
12 acting opiate by simply administering drugs that
13 inhibit the transporters.
14 That is not something you would spot from
15 an obvious plasma concentration time profile. So,
16 I think there is a great danger in an overly
17 simplistic analysis of opiate sparing as an
18 endpoint, and there needs to be independent data
19 that demonstrates that the drug has analgesic
20 effect on its own.
21 DR. FIRESTEIN: Maybe Dr. Katz can address
22 that concern with regard to the pharmacokinetics
23 and opiate tolerability, and then Dr. Farrar, if
24 you had anything to add, that is.
25 DR. KATZ: I agree.
314
1 DR. FIRESTEIN: Thank you.
2 DR. FARRAR: I think the point about the
3 use of opioid sparing as a potential measure is an
4 important jumping-off point to consider what was
5 brought up in the last two discussions, the last
6 one in particular, which is that what is it we are
7 trying to do here.
8 I would argue, as I think Dr. Katz did
9 very nicely, that opioid sparing might be a nice
10 way to at least think that maybe the drug has some
11 effect, but ultimately, what we are interested in
12 is making the patient better.
13 At the end of the day, whether you are
14 using a specific protein that you assay to identify
15 a group in which people get better, which I think
16 is a great idea and hopefully will pan out, but at
17 the end of the day, we really need to decide what
18 it is when a patient gets better.
19 I would ask Mitchell, in terms of the
20 situation that he is talking about, would you want
21 a particular group to respond a lot or a little,
22 does it matter whether you have got a BRAC gene, so
23 that you have got a 90 percent chance of developing
24 breast cancer or a 90 percent of responding to a
25 drug, or does it matter whether you have got a 51
315
1 percent chance of responding to the drug, because I
2 think no matter how we slice this and no matter how
3 we look at it, at the end of the day, we are left
4 with the issue of does it make the patient better
5 or not.
6 You can use any statistical technique you
7 like or you can use any analytic technique you
8 like, you can use any assay technique you like, but
9 we can't escape that issue.
10 In terms of the discussion today, we have
11 talked about a lot of different mechanisms, and I
12 wonder what these people's thoughts are on that.
13 DR. FIRESTEIN: Janet and then Dr. Katz.
14 DR. ELASHOFF: In terms of the data
15 dredging to find a subgroup that you then test in
16 that subgroup, and that that might be a very good
17 way to find subgroups in which it does, in fact,
18 work, from a statistical point of view, the
19 likelihood of the second trial coming out should be
20 pretty small because you are mainly picking up
21 false positives with that kind of multiplicity of
22 testing, so that it might be that the first 5, 10,
23 15 times somebody tries that, it doesn't pan out in
24 the second trial.
25 DR. KATZ: I just wanted to add one more
316
1 point about the opiate sparing trials, because I
2 don't want us to leave the discussion with having
3 trivialized the opioid sparing. I mean there are a
4 number of clinical scenarios in which you have to
5 give the patients concomitant opioid therapy with
6 whatever your analgesic of interest is.
7 For example, in the postoperative
8 thoracotomy or postoperative pain setting, it would
9 be unimaginable to not allow the patients to have
10 access to opioids, and the setting of cancer pain
11 would be another example.
12 So, you often have to co-administer your
13 study drug with an opioid analgesic, and then
14 opioid sparing is a natural thing to look at. So,
15 having said that, there are reasons to look at
16 opioid sparing, but the bottom line is that you
17 still need to decide whether or not your patients
18 are better on your study drug.
19 DR. WOOD: A patient would not be better
20 on a study drug just because you inhibited fentenyl
21 or fentenyl's metabolism. I mean that is exposing
22 them to the same dosage exposure as they would have
23 got from a higher opiate dose, and we need to make
24 that distinction.
25 DR. FIRESTEIN: And the patient wouldn't
317
1 necessarily be better, they would just use less
2 opiates.
3 DR. KATZ: That is exactly my point and
4 that if it was just a pharmacokinetic interaction,
5 presumably, the patients would be the same. Your
6 outcome measures would fail to show in that case
7 that your patient was better off despite the opioid
8 reduction, and it should be considered a failed
9 trial. That is what I am trying to say.
10 DR. FIRESTEIN: Lee.
11 DR. SIMON: In fact, that is the
12 conundrum. We are confronted in proposals to look
13 at the question of opioid sparing as a primary
14 outcome, and the reason we ask the question for
15 this debate was we don't know what to do with that,
16 (a) we don't know what is minimally clinically
17 important decrease - is a 3 mg decrease, a 30 mg
18 decrease clinically important unless you tell us
19 what the measures are that tell us that it is
20 important, meaning is the patient more aware, are
21 they able to walk faster, is the recovery
22 postoperatively improved, is there less pneumonia,
23 if, in fact, pneumonia is an issue.
24 These are the issues that have to be
25 clinically relevant to make a measure, such as a
318
1 change in opioid use, important, and that
2 discussion is no different than the one that was
3 raised by Mitchell just before.
4 The measurement of a receptor change or
5 whatever is really not different than the
6 measurement in the change in how much morphine that
7 one might use unless there is a change in the
8 clinical relevance and an improvement to the
9 patient care.
10 I just want to make it clear to Dr.
11 Davidoff that we would not be looking at only one
12 unique database for such an event. One would have
13 to define clinical relevance by multiple databases.
14 Thank you.
15 DR. FIRESTEIN: Dr. Davidoff and then Dr.
16 Brandt.
17 DR. DAVIDOFF: I was really just going to
18 say essentially the same point about opiate
19 sparing, that it might not be necessary to find
20 better overall pain relief, but fewer side effects
21 associated with it.
22 After all, some of the major distinction
23 between antidepressants is not that there is
24 overall better therapeutic efficacy between SSRIs
25 and tricyclics, but that there are fewer side
319
1 effects.
2 DR. BRANDT: I think this whole discussion
3 on opioid sparing is very interesting, but I would
4 suggest that in the context of the meeting, it is
5 perhaps a little too narrow, we could raise the
6 same issues with regard to NSAID sparing or chronic
7 NSAID use.
8 DR. SIMON: So, in that case, Dr. Brandt,
9 would you propose that a primary outcome for a new,
10 perhaps analgesic that would not have opioid
11 effects and would not have the traditional effects
12 one associates with the traditional nonsteroidal
13 anti-inflammatory drugs, could use as an outcome
14 measure for primary approval, the decrease in
15 requirement for the rather ineffective nonsteroidal
16 anti-inflammatory drugs?
17 DR. BRANDT: When you consider the side
18 effects associated with NSAIDs, the answer is yes.
19 DR. WOOD: But only provided you have
20 demonstrated it is not just due to a simple
21 interaction.
22 DR. BRANDT: Surely.
23 DR. FARRAR: At the end of the day, it
24 makes no difference if you reduce the opioid or the
25 NSAID. What makes the difference is whether the
320
1 patient is better, and if they are better, as Dr.
2 Davidoff was suggesting, because the side effects
3 are better, that is better. It is not that they
4 are using less of one drug or another drug.
5 It really doesn't matter. I mean I agree
6 with you, and I am not arguing the issue about
7 opioid sparing, I think opioid sparing is
8 suggestive at best, and you clearly need to
9 differentiate between the amount of opioid that
10 they are actually taking orally and the amount
11 absorbed and the amount that is reaching the active
12 sites and the amount that is causing the effect,
13 and there are lots of drugs in which you get the
14 buildup of toxic byproducts, as well.
15 But at the end of the day, what you really
16 need to know is whether that patient postsurgically
17 had a better experience with the combination of
18 drugs that you gave than if you didn't.
19 How you define better depends on the
20 circumstances that you are looking at, but I think
21 there are clearly lots of indicators that we can
22 use to look to see what we should be measuring and
23 how we should be measuring. But at the end of the
24 day, the question is, is the patient better, would
25 I want to give that patient that drug the next time
321
1 around because they said, you know, I had three
2 surgeries so far, this was the best experience I
3 had so far.
4 That was very true with epidural
5 anesthesias. I mean there is absolutely no
6 question that people post-op with thoracotomies did
7 better because they were able to breathe better, et
8 cetera, et cetera. How much opioid you gave them
9 didn't make a difference.
10 DR. FIRESTEIN: Dr. Sherrer and then Dr.
11 Anderson, Dr. Strand.
12 DR. SHERRER: I think that at the end of
13 the day, it is, is the patient better. I think
14 that is very important, but I also think we need to
15 consider some of the social issues with the chronic
16 use of opiates, that impact on whether the patients
17 are actually better.
18 We have many patients who are afraid to
19 take opiates because of the issue of addiction, and
20 there are many physician who are afraid to
21 prescribe opiates because of the issue of
22 addiction, and the bottom line of that is it
23 impacts on whether the patients are better, because
24 if they are not going to take the drugs or if the
25 drugs are there and the physicians are afraid to
322
1 use them, then, it is not going to make the patient
2 better even if theoretically they could.
3 So, I think we do need to look at this
4 issue of addiction and tolerance, and what is the
5 relationship more, and what I am hearing is that we
6 can't really define that well enough to do that, or
7 at least we don't have measures of predicting or
8 defining addiction.
9 I think that is very important. One of
10 the major issues with the use of opiates and
11 chronic pain is whether, despite those six studies
12 that you showed us that suggest there is not
13 addiction, there is still fear on behalf of
14 physicians and patients that there is addiction and
15 that tolerance itself may lead to addiction.
16 DR. ANDERSON: My concern is about what
17 you were saying just now, about the patient, at the
18 end of the day, the patient being better, and that
19 if this was solely in terms of having fewer side
20 effects, that was okay.
21 I didn't like that, I guess because, you
22 know, side effects don't happen, you know, happen
23 sporadically or should happen sporadically, but
24 efficacy is something that one would hope would
25 happen in a large proportion of patients.
323
1 Historically, the FDA has kept efficacy
2 and safety, I mean they are linked, but they are
3 not considered the same thing, and it bothers me
4 that a drug combination could be considered could,
5 not because it was efficacious, but just because it
6 had fewer side effects. I may be misunderstanding
7 what you are were saying.
8 DR. FIRESTEIN: In some cases, the side
9 effects are mechanism based, and that is a
10 situation where it would be optimal to lower the
11 dose. So, for instance, with opiates, constipation
12 or nausea or vomiting, those are clearly based on
13 the pharmacology of the molecule, and so if one can
14 get past those by using a lower dose, and using
15 another adjunctive therapy, then, there would be
16 some benefit to the patient.
17 Dr. Strand.
18 DR. STRAND: I would just like to say this
19 reminds me of some steroid sparing discussions that
20 some of us have had in the past, and it seems to me
21 that it is all find and good if we can decrease the
22 dose of opioids or the dose of steroids, but if, in
23 fact, there isn't some benefit that is measurable
24 in addition, in terms of patient-reported outcomes
25 of efficacy and/or tolerability, then, I don't know
324
1 that we have demonstrated very much of anything.
2 The other point that I would like to make
3 is that I think data dredging is not the way we are
4 going to get approvals or try to look at different
5 ways of approving products, say, in chronic pain,
6 or possibly even subacute pain or whatever we are
7 calling it, but there is room to develop these
8 analyses from the Phase II data, particularly since
9 there is much more emphasis on doing better Phase
10 II trials, dose finding and dose interval finding
11 or schedule.
12 From that point of view, one could, in
13 fact, develop evidence-based, responder type of
14 outcomes, or one could combine certain outcomes for
15 a certain type of response in the Phase III trials.
16 That has been done before.
17 DR. FIRESTEIN: Dr. Cush and then Dr.
18 Farrar.
19 DR. CUSH: My summary of what I heard
20 today that I would hope that the Agency would take
21 away is I think that we are probably still wedded
22 to some of the methods of the past, and that would
23 be acute and chronic indications and some of the
24 primary outcome variables that have been used for
25 those indications, but that we hear that the
325
1 science has come along and we would like to see
2 mechanistic issues being raised, may be secondary
3 outcomes measures where applicable, and that would
4 be ideal as we move forward and designing better
5 trials that mean something.
6 Secondly, I think that making low back
7 pain a priority and either incentivizing that or
8 requiring that in some way would be nice, and
9 lastly, the words of Dr. Carr reminded me of
10 something that Ted Pinkus said at a meeting that I
11 think Lee and I were at, which is that as
12 clinicians and biometricians we have done a good
13 job in defining outcomes and coming up with
14 acceptable measures, but we have missed the boat
15 because we are still not at a point where clinical
16 trials are approximating what goes on in the
17 office, so clinicians and patients won't understand
18 an ACR-20 or a WOMAC, and whatnot, and at the
19 Agency, I think it could go more towards that
20 direction, I think it would also further not only
21 clinical trials, but patient care, as well.
22 DR. FARRAR: To take off from what was
23 just said by Dr. Cush and perhaps try and persuade
24 Dr. Anderson that there may be some aspects of this
25 that don't apply to everyone.
326
1 I agree with you. I think, Dr. Cush, that
2 making the trials understandable to the clinical
3 circumstance is of paramount importance, so that
4 when I, as a clinician, sit down with my patient, I
5 know what to do, and I don't just know that
6 patients got better on the WOMAC by an average of
7 4. I don't know what that means now, and I know
8 what the WOMAC is, even use it.
9 I think, though, the issue that I wanted
10 to bring up more specifically is that what Dr. Max
11 was suggesting was not, I think, that data dredging
12 should be used as the sole purpose or the sole way
13 in which a drug should be approved, but that it be
14 used as a hypothesis-generating event, and I think
15 that makes sense.
16 Then, he was trying to see whether one
17 trial after that would be enough in terms of
18 stimulating that kind of research, and I agree that
19 there is issues there on whether it is one or two
20 can be debated.
21 What he was getting at, though, was that
22 with a 50 by 50 slab of gel, you might be able to
23 tell what the makeup of that patient is with
24 regards to their response. This gets at what Dr.
25 Anderson I think was saying was that, in fact, the
327
1 drug that we use has to be good for lots of people,
2 and we are getting to the stage now where we are
3 developing drugs, especially in neuropathic pain,
4 perhaps not so much in arthritis, where individuals
5 who respond to a single drug are a minority of the
6 patients that we are treating.
7 You can look at that two ways. One is we
8 just don't know how to predict who is going to
9 respond, and that is very true. If we could
10 predict who was going to respond, then, 100 percent
11 of those patients would respond, but the clinical
12 fact is that people see arthritis, they don't see
13 the variance of the arthritis that we might able to
14 see here.
15 People see pain. They don't see the
16 variance and the subtleties of it that an expert
17 might see, and they treat them with the medications
18 that we have.
19 There are some very good examples in
20 postherpetic neuralgia and diabetic neuropathy
21 where drugs that are clearly effective worked in
22 about a third of the patients treated. About a
23 third of the patients got a moderate or better
24 improvement. That is 1 out of 3 and if I am
25 treating in the office, and only 1 out of 3 people
328
1 get better, I am might decide that is not the right
2 drug.
3 On the other hand, I might look at it and
4 say 1 out of 3 in something where nothing else has
5 worked, that is really good. The same applies in
6 arthritis in that there are clearly differences
7 between the NSAIDs, and they are not as dramatic
8 perhaps as the differences in the anticonvulsants,
9 but there are differences, and it may be that one
10 group responds better to one kind of NSAID and a
11 different one to a different.
12 So, the idea that we have to somehow have
13 a drug that works in 50 percent or 70 percent of
14 our patients in clinical trials is not I think the
15 issue. I think the issue is being able to identify
16 the people in whom it does work, and it really
17 works, not just a little, but it makes them really
18 better.
19 DR. FIRESTEIN: Dr. Dionne and then Dr.
20 Abramson.
21 DR. DIONNE: I have heard the phrase
22 "end-of-the-day" mentioned a few times. I am
23 struck by the fact that this is the end of the
24 first day that was supposed to be devoted to
25 chronic pain, and I have heard a minimum consensus
329
1 of opinion on some of the issues that were raised
2 for the Agency, and I would be afraid that they
3 might go back up Rockville Pike and disappear into
4 the back room, so to speak, and come back in four
5 years or 10 years, as Al Sunshine said it took last
6 time, with a document that reads like the Ten
7 Commandments.
8 I am wondering, is there room for
9 discussion of the processes that might allow us to
10 resolve some of these issues based on some sort of
11 a scientific process rather than an opinion-based
12 process.
13 For example, the 125 pain measurement
14 scales that Dan Carr mentioned are ones that it
15 would be hard to imagine we could sort through and
16 just by opinion say these are the two or three that
17 should work, yet, we are still using Category and
18 VAS, which are as old as the drug classes we use to
19 test them on, ignoring all the new technology,
20 which might include the electronic diary we heard.
21 Other outcome measures, how would we go
22 about getting at which ones are desirable, let
23 alone grappling with the issues like analgesic
24 combinations, what would be the criteria for those.
25 That was an issue that raged all through the 80s.
330
1 I am not sure whether it got resolved or people
2 just stopped trying to get combos of NSAIDs and
3 opiates put together.
4 Is there room for some discussion of the
5 process that the Agency might use to arrive at from
6 where they are now to where they would be when a
7 document appears?
8 DR. FIRESTEIN: Is there room, Dr. Simon?
9 DR. SIMON: There is always room at the
10 table. I think that this meeting and two meetings
11 that have been held by the Advisory Committee of
12 170, talking once about neuropathic pain and issues
13 about opioids reflects the fact that we are very
14 interested in dialoging with the community, the
15 patient community, about these particular areas.
16 We are talking on a regular basis, and
17 will be talking on a much more regular basis, with
18 the individuals in the FDA who are interested in
19 pain and issues regarding pain, particularly the
20 other Division 170, and coming up with a consensus
21 as much as we can as it relates to the various
22 different products that we are assigned
23 responsibility for, and those products that we can
24 possibly imagine will be developed in the future,
25 to then lead us towards a document.
331
1 Furthermore, there are discussions that
2 are ongoing with the NIH about establishing a
3 meeting to discuss outcome measures, both acute and
4 chronic, addressing issues regarding function
5 versus health-related quality of life that need to
6 be addressed before we can put pen to paper to try
7 to design and craft a document that will fulfill
8 all the needs that we have been talking about just
9 so long today, not the less tomorrow.
10 So, that is the process. The process has
11 got a was to go. We have got more internal debate
12 to do, more external debate to do, more to learn,
13 and to address Ray's issue of going to the evidence
14 and the science using the science as we interact
15 with the group at the NIH, in understanding more
16 about outcome measures as we did at the last March
17 meeting. So, that is the process.
18 DR. FIRESTEIN: Steve.
19 DR. ABRAMSON: I guess part of the process
20 I would like to express is that we have this
21 dilemma of wanting, at the end of the day, to do
22 the best globally for the patients, and yet we are
23 confronted by very specific syndromes that differ,
24 and we have an iterative process to get a global
25 overarching kind of indication, but, in fact, that
332
1 iterative process is going to take a lot of very
2 focused specific kinds of analyses of different
3 pain syndromes, developing clinical criteria of
4 those syndromes, the way we have done in other
5 diseases, in OA, and outcome measures, as Mitchell
6 was getting at, even prospectively looking at
7 certain biomarkers in those areas.
8 So, I think it is a time of great
9 opportunity to look at different pain syndromes, to
10 use this new development of analgesics as a way to
11 use the clinical trial tool to answer questions
12 that are mechanistic.
13 Part of the dilemma, the conflict is that
14 one does not want to get a global approach too
15 early without this iterative process having been
16 gone through to really understand these different
17 diseases, which, in fact, are quite distinct one
18 from another, even in the musculoskeletal, so that
19 is just the process comment.
20 Going to back to Dr. Anderson's, and Dr.
21 Katz mentioned this, and it is a very focused
22 question, back to the opioid use as a surrogate
23 endpoint. There is a difference I think between
24 what is good for the patient at the end of the day
25 versus the regulatory agencies need to determine
333
1 whether a drug is efficacious.
2 Absent the metabolic effects of opioid,
3 metabolism, for example, and drug-drug interaction,
4 the question still is, is opiate use a legitimate
5 endpoint, primary, secondary, by which you can
6 judge the efficacy of a new drug.
7 That doesn't mean whether the patient is
8 better to be on one or two, and I think you alluded
9 to this, but I am not sure sillet [ph] isn't a
10 valid measure. I don't know about the area, but it
11 is worth discussing, which is not the patient's
12 contentedness with their combination of drugs, but
13 whether it's a tool, an instrument to judge the
14 validity of a new drug being presented to the FDA.
15 I am just curious what people think. I
16 don't know if I want to open that up to
17 discussion, it is just kind of a comment.
18 DR. FIRESTEIN: That is another major area
19 of discussion in and of itself. One of the reasons
20 that the Division gathered this meeting was to
21 address certain specific questions, and as we are
22 getting towards the end of the day, although it is
23 only 1:30 in San Diego right now, so I am just
24 waking up, I think.
25 From what I heard said, I don't know
334
1 whether or not it can at least try to offer some
2 more concrete guidance or at least advice to the
3 Division with regard to some of the key questions,
4 and one is whether or not there is, in fact,
5 utility to having acute and chronic pain as opposed
6 to just pain as a potential indication.
7 It seems to me that that is not an
8 unreasonable approach, and I was wondering if there
9 is any additional discussion that would help sort
10 that out or if people are relatively comfortable
11 with that.
12 DR. ASHBURN: I would say yes with the
13 caveat that the definition goes away from time
14 lines with regard to duration of the pain, and kind
15 of goes towards the acute versus chronic pain
16 definitions that Dr. Woolf presented to us earlier
17 this morning with regard to pain that is expected
18 to be of short duration with some expectation that
19 it goes away over time.
20 Again, that goes towards a concern that
21 chronic pain states sometimes can be rapid onset
22 and can deserve study and therapy early rather than
23 late in their time line, and should not wait three
24 or six months prior to being allowed to include
25 patients for investigation, and the example, that
335
1 is, patients with postherpetic neuralgia or with
2 cancer pain.
3 DR. FIRESTEIN: I think that is an
4 excellent point, and again raises the question of
5 an acute persistence pain and acute chronic--I
6 don't know.
7 DR. ASHBURN: One terminology that comes
8 to my mind when we talk to medical students about
9 this concept is short-term pain versus long-term
10 pain, and the perception of getting away from the
11 terms acute and chronic, which mean different
12 things to different people, but rather, the
13 expectation of whether this pain is of short
14 duration, of limited area, whether or not the
15 expectation is, unless one intervenes on the
16 patient's behalf, that the pain will persist over
17 long periods of time.
18 DR. FIRESTEIN: Dr. Borenstein.
19 DR. BORENSTEIN: Well, one of the points I
20 wanted to make is what happens in the clinical
21 trial situation and what comes into the clinic. I
22 think all the basic scientists would agree if you
23 can attack pain early, you would like to keep it
24 from becoming chronic, so intervening as early as
25 possible in the process to keep that from happening
336
1 may have a mechanistic way of trying to keep
2 chronic pain from appearing, but if the patient
3 appears to you already with a process which seems
4 to be chronic pain, then, I think what you may find
5 to be effective there may be somewhat similar to
6 what you would use in the very acute circumstance,
7 but you may need more interventions at that point
8 to really make a difference in that individual.
9 So, what you would do if you had someone
10 who was your patient over time, you would treat
11 them differently than you might if you find them
12 later on in the process when you have them as your
13 patient.
14 DR. FIRESTEIN: Brief comment from Dr.
15 Farrar and then Dr. Katz.
16 DR. FARRAR: There are I think two
17 important components of this, and very briefly, one
18 is just to remind us that acute and chronic are
19 time frames and that the acute pain and chronic
20 pain does not necessarily imply acute treatment and
21 chronic treatment, and I think that those two
22 things are very different in terms of thinking
23 about the safety of a drug and the overall use.
24 The second issue I think has been brought
25 out before, but would suggest that what we are
337
1 really talking about is reversible pain versus
2 non-reversible, and there are certainly syndromes
3 which occur and can, as I was learning at lunch
4 today, snake bites last an awfully long time. If
5 you don't know what I am talking about, you will
6 find out at dinner, I guess.
7 But the point is that there are pains that
8 occur for a very long time, but are reversible and
9 are treated aggressively, and there are acute types
10 of pains best brought up I think by Clifford
11 earlier, which is that, you know, trigeminal
12 neuralgia is an acute pain that is very, very
13 different than postsurgical pain.
14 I think that it is very important to
15 differentiate, but we have to be careful about the
16 way in which we do that.
17 DR. KATZ: I was going to make a similar
18 point, I think, which is that when we think about
19 treatment of acute pain, the way it actually works
20 out very frequently in real life is that patients
21 are actually treated for months often for their
22 so-called acute pain, which we normally might think
23 of as just a few days. Thoracotomy, you know, 50
24 percent of patients six months after a thoracotomy
25 have moderate to severe pain, spinal fusion
338
1 surgery, the patients are often on analgesics for
2 six months or a year, knee replacement, et cetera,
3 et cetera.
4 So, I think it is also worthwhile keeping
5 in mind that how is the medication likely going to
6 be used in practice, and the trials that are done
7 to support that use ought to have some relationship
8 to the actual way that they are used.
9 DR. FIRESTEIN: A couple of more brief
10 comments over here and then we will go to the next
11 point.
12 DR. WOOLF: It seems to me, coming back to
13 the issue of what encouragement we can give to the
14 Agency in terms of development plans, we have heard
15 from Dr. Farrar that 30 percent of these patients
16 may respond to a certain treatment, and he has no
17 way of predicting at the moment who those patients
18 may be.
19 My plea would be that in any discussion
20 with the industry in terms of any development plan,
21 as we are in this transition mode from a rather
22 empirical approach to the management of pain to one
23 where mechanisms can be identified, is to try and
24 get as much information as possible.
25 While, on the one hand, of course, we all
339
1 agree we want the patient to feel better and we
2 need some global measure of that, but the point I
3 was trying to make this morning was that there are
4 many aspects of a pain that are simply ignored in
5 trials and that may be very useful in terms of
6 seeing whether patients do respond in different
7 ways to different forms of therapy.
8 So, I think part of the process has to be
9 not to prejudge and to try and gather as much
10 information as possible from the patient as to what
11 their pain is composed of and how different aspects
12 of the pain respond to different therapies.
13 DR. DIONNE: I think Clifford just said
14 what I was going to say, but let me just try to
15 restate it. If the Agency is interested in
16 mechanisms, and if we think the way to the future
17 is having a better understanding of the mechanistic
18 process by which a new drug works rather than just
19 extrapolating from animal models which may or may
20 not be relevant, would there be a possibility of
21 developing some sort of incentive into the claim
22 structure or the approval process that would give
23 greater favorability to coming up with a rational
24 study of the mechanism underlying an acute drug
25 versus a chronic drug, so you might discover, in
340
1 fact, as has been stated all day, that some drugs
2 may be actually acting on a chronic pain mechanism
3 that may be starting at the first day or two, and
4 this may have long-term benefit for preventing the
5 pain a preemptive fashion rather than having to
6 wait two or three months and then try a treatment
7 that is ineffective because that mechanism is no
8 longer active.
9 So, have some mechanistic approach built
10 into the acute versus chronic studies that allows a
11 little bit of information to be gathered, and the
12 best way to harness the resources that the industry
13 could bring to that, of course, would be to have
14 some sort of incentive in the approval process for
15 that.
16 DR. CALLAHAN: I was just going to say I
17 think you made a compelling argument this morning
18 about the mechanisms, but if we don't have the
19 instruments to measure the components, is it fair
20 to ask the industry to look at those components
21 until those measurements are available, or should
22 we go with the global and ask them to look at that
23 sort of in a secondary fashion as they evaluate the
24 new drugs that are coming on.
25 DR. FIRESTEIN: This really brings us to
341
1 the second question, and that is, whether one
2 focuses on mechanism-based indications of clinical
3 indications, and by and large, over the course of
4 the day, most of the emphasis is that while
5 mechanism-based indications are of tremendous
6 interest, the science isn't there yet in order to
7 use that as the touchstone for specific drug
8 approvals, and that we still are relying primarily
9 on clinical situations and clinical indications
10 even at this point.
11 I was wondering if again there was any
12 comment or disagreement for that. Did you want to
13 comment on that?
14 DR. KATZ: I agree that right now it is
15 premature to begin a drug development program for
16 pain due to excitable nociceptors or central
17 sensitization or something like that, but one has
18 to be careful not to just by default allow any
19 clinical classification system.
20 Some of them make a lot more sense than
21 others. For example, the idea of having a
22 medication for cancer pain makes no sense to me
23 whatsoever, because some people with cancer pain
24 have a brachial plexopathy from tumor invasion,
25 some people have bone metastasis, some people have
342
1 visceral obstruction with almost no connection
2 whatsoever.
3 So, to me, that would not make a lot of
4 sense specifically because the mechanisms are so
5 different among those different types of pain, so
6 again, you can't forget the mechanism either,
7 whereas, musculoskeletal pain, it seems to me that
8 medications that work for one kind of
9 musculoskeletal pain tend to work for another -
10 osteoarthritis, rheumatoid arthritis,
11 non-neuropathic low back pain, et cetera, I would
12 suspect because the mechanisms are similar in those
13 disorders.
14 I don't think that you can just allow any
15 clinical classification system, but you can pick
16 and choose from ones that make more sense.
17 DR. MAX: I just want to mention what I
18 heard Lee Simon and Larry Goldkind saying a few
19 minutes ago seemed very new, that they said that if
20 they get some novel evidence about how can you
21 reliably predict response with a new mechanistic
22 test, they have the authority to approve it after
23 the post-hoc searching with one new prospective
24 trial, and since it so new and they don't have to
25 maintain a level playing field when there aren't
343
1 many other things, every situation is unique, it
2 sounds like it is a real green light for industry
3 to try to be imaginative and scientifically
4 creative. That is the first time I have heard
5 that.
6 DR. FIRESTEIN: I would not overinterpret
7 those comments.
8 [Laughter.]
9 DR. FIRESTEIN: It is clear that exciting
10 new discoveries, novel targets that have clear
11 proven efficacy in clinical situations can move
12 very quickly into the clinic, into approval.
13 An example of that would be some of the
14 TNF inhibitors for rheumatoid arthritis. Under
15 those circumstances, I suspect that what you
16 envision would be possible although I wouldn't dare
17 speak on behalf of the Agency--well, I think I just
18 did.
19 This actually brings us to the other sort
20 of difficult problem, and that is the notion of if
21 there is going to be a chronic type of indication,
22 what is the benchmark for that. I don't know what
23 the right answer is. We have a couple of different
24 possibilities. I didn't know if anybody on the
25 committee had specific recommendations.
344
1 From my own perspective, I was intrigued
2 by the proposal of the four different indications
3 with single studies, because if you are using
4 chronic pain as the actual indication, then, you
5 are not going for the separate indication of OA
6 versus something else. You are using chronic pain
7 as the indication, and the second confirmatory
8 study would be in a different indication.
9 So, there is actually a rationale and
10 maybe a middle road whereby you actually require
11 fewer studies, but more indications.
12 I would want to know if anybody had a
13 comment there.
14 DR. WOOD: As written here, it seems to me
15 to be counterintuitive. It seems to me that to put
16 a bar up that says you have to demonstrate, for
17 example, response in low back pain and diabetic
18 neuropathy and cancer pain, seemed to me to be
19 counter to everything we have discussed in terms of
20 mechanisms.
21 It would seem to me that demonstrating
22 that a drug is effective in multiple indications
23 demonstrates just that, that the drug is effective
24 for multiple indications, and at that point,
25 physicians can and do make decisions every day
345
1 about extending the drug's use into other
2 indications for which it has not been tested.
3 But making the leap in terms of a labeling
4 for indications for which it has not been tested
5 seems to me something that has never been done in
6 any other setting. I don't see even why you need
7 to do it. If you have studied the drug in four
8 indications, that is normally what you label it
9 for.
10 Just to follow that up, the ACE inhibitors
11 were all approved for the treatment of heart
12 failure with subtle differences in the indications
13 for which they were approved, reflecting the
14 studies that were actually done.
15 That hasn't obviously affected their use
16 in these indications.
17 DR. FIRESTEIN: Dr. Woolf and then Dr.
18 Goldkind.
19 DR. WOOD: I find myself feeling a bit
20 uncomfortable with this notion that there is going
21 to be a global chronic pain analgesic. I think it
22 goes against everything we know and everything that
23 we are beginning to understand, and I think that is
24 exactly what your comment relates to.
25 So, which four different indications would
346
1 you need in order to make sure that it was global?
2 How many neuropathic pain and how many
3 musculoskeletal pain, what is the balance that one
4 would feel comfortable with, that would encompass
5 all forms of chronic pain that crossed all
6 mechanisms?
7 I don't think we have a consensus on that.
8 I think that if one were careful in selecting four
9 indications that were predominantly
10 musculoskeletal, that would leave you with a
11 situation where you may have a drug with an
12 indication for chronic pain that would still not
13 work in many patients who have postherpetic
14 neuralgia or diabetic neuropathy or radicular pain.
15 DR. FIRESTEIN: Well, the FDA would have
16 to think very carefully about how one would choose
17 those particular indications, it seems to me.
18 You were going to make a comment.
19 DR. GOLDKIND: In response to Dr. Wood's
20 comment, our current reality is that we are
21 approving drugs as analgesics, and there is an
22 assumed generalizability, and that is part of why
23 we wanted to discuss this, but we do see drugs that
24 have dental pain and maybe one particular post-op
25 setting that form the pivotal basis for approval.
347
1 They are marketed as analgesics and even
2 if we describe the particular pain settings or
3 model, depending on semantic difference, in the
4 Clinical Trial Sections, so people know where the
5 evidentiary base came from, it still is an
6 analgesic indication.
7 This lumping and splitting, we play out
8 all the time, and we do want to optimize that.
9 DR. FIRESTEIN: But that is precisely why
10 the bar is so high potentially for a true global
11 chronic pain indication.
12 DR. MAX: I am very sympathetic towards
13 setting such a high bar for general chronic pain
14 claim. That is the part of Lee's proposal that I
15 love, and I think it is because of this syllogism.
16 When I talk to company marketing people,
17 they say we would really like a chronic pain claim,
18 or even if it is neuropathic pain, a generalized
19 neuropathic pain claim, because we can send our
20 marketing people and our detail men and sell more
21 drug, and have higher profits, and I think the
22 logic is that incentive would lead to many more
23 trials and from multiple trials, multiple trials in
24 many different disease conditions are the best way
25 to advance the science, and I think that is a great
348
1 way to go about things, so we will be able to
2 generalize even better later on.
3 The missing piece of data, however, is I
4 have asked whenever I have had those conversations,
5 I have asked the marketing person, industry, is
6 there any evidence how much a general claim is
7 worth, why it makes a difference, do you need it
8 for the managed care organization or the pharmacy
9 to pay for it, et cetera, and I haven't encountered
10 any rigorous data or modeling, so let me ask
11 anybody from the committee or agency, would we be
12 better served if there were some economic model or
13 data, if that is partly underneath the reason for
14 going for this high bar.
15 DR. McLESKEY: I won't respond in any
16 detail, but I would say there is a general
17 understanding that the bar to the claim largely,
18 the likelihood is the larger the market will be.
19 DR. ABRAMSON: I just want to pick up on
20 Dr. Woolf's comment that a general global approval,
21 if we lower the bar for individual approvals is
22 going counterintuitive to the notion that we are
23 funding their differences among the different pain
24 syndromes, and I think the concept of general
25 chronic pain, I think we have to be very careful
349
1 about given this morning's discussion.
2 I would argue that even if a broad
3 indication was met because you had three
4 indications in these separate areas, that we
5 shouldn't lower the bar in any of those individual
6 indications by the numbers of studies that you
7 would need to show that your drug worked in
8 neuropathic pain, fibromyalgia, low back pain,
9 whatever it is.
10 So, my concern about having one study in
11 four different areas is that you are diluting the
12 individual iterative process and that everything
13 should be able to stand alone as an indication in
14 that area, and if you hit three or four, you have a
15 global marketing advantage, but you haven't diluted
16 the process for any area.
17 I think the word "counterintuitive"
18 becomes very critical that we separate all these
19 different pains as much as we can as we better
20 understand them.
21 DR. FIRESTEIN: Dr. Katz.
22 DR. KATZ: I wonder if it might be useful
23 to use the opioids as a model to do a thought
24 experiment with the idea of a chronic pain
25 indication. Barring the issues that I mentioned
350
1 earlier about addiction and tolerance, and all
2 that, we know that there are clinical trials
3 supporting efficacy of opioids in neuropathic pain,
4 musculoskeletal pain, there are a bunch of
5 different studies, headache, short-term studies
6 even if we want to go that far, cancer pain
7 certainly, does anybody feel that opioids would not
8 meet anyone's threshold to be a general analgesic
9 for chronic pain?
10 Now, granted, they don't work for every
11 kind of pain. Probably they are not effective for
12 central pain, I would guess, but does a medication
13 have to be effective for every single kind of pain
14 in order to be considered generally to have broad
15 applicability, just as a medication for
16 hypertension might not work for every single
17 patient or subtype of hypertension, but still might
18 have broad applicability within hypertension?
19 It seems to me that the opioids are a
20 broad spectrum analgesic. Why, therefore, is it
21 not possible that another medication could be a
22 broad spectrum analgesic?
23 DR. WOOD: Let me just respond. I think
24 one thing that we were talking about over here is
25 there seems to be the impression that 30 percent is
351
1 a bad response rate. That is about the average you
2 get in every trial of almost any indication. It is
3 40 percent for anti-hypertensives, it is lower for
4 antidepressants.
5 I mean 40 percent is about the rate of
6 response you get to a single drug in the pivotal
7 trials which are submitted to the Agency, less than
8 that for some. So, 30 percent ain't so bad, and if
9 it's 33 percent, so expecting that we will see
10 substantially more than that seems to me to be
11 counter to what we have seen with almost every
12 other drug class we have approved.
13 DR. FIRESTEIN: Dr. Borenstein will get
14 the last comment from the committee today.
15 DR. BORENSTEIN: One of the points I
16 wanted to be sure about from the clinical situation
17 is when the drug is approved for a general pain
18 indication or is used in one area, it does get used
19 in another to see if it works.
20 That ends up being what happens in the
21 clinical situation. I think what it is for the
22 Agency is to decide whether three out of four at a
23 certain level, and pretty good on another is close
24 enough, or is it really great on two and okay on
25 two others, is that good enough.
352
1 What you see in patients is whether they
2 respond or not. In individuals, it is really yes
3 or no, do they get an effect and can they tolerate
4 it. So, I think the question for the group is what
5 is adequate to allow a drug to have this indication
6 to allow it to be used in the general public for a
7 variety of pain syndromes that will allow patients
8 to get better and at the same time, use it
9 reasonably safe.
10 That is what I think the group has to
11 decide, whether that is three or four, certainly
12 the Agency has a better idea of what that truly
13 means. In the clinical situation, I see a patient
14 where if I have a drug where I think it might be
15 helpful, I am going to try it. At some time, I am
16 going to be smart enough to figure out the
17 mechanism by why it works, but sometimes you just
18 have to try.
19 DR. FIRESTEIN: Before I adjourn, I did
20 want to see if there is any of the officers from
21 170 that had any additional comments. No? Okay.
22 Thank you very much, everybody. It has
23 been an exciting day and we have more in store for
24 tomorrow. Thank you.
25 [Whereupon the proceedings were recessed
353
1 at 4:45 p.m., to reconvene on Tuesday, July 30,
2 2002, at 8:00 a.m.]
3 - - -