1 DEPARTMENT OF HEALTH AND HUMAN SERVICES FOOD AND DRUG ADMINISTRATION INFECTIOUS DISEASES SOCIETY OF AMERICA INTERNATIONAL SOCIETY OF ANTI-INFECTIVE PHARMACOLOGY FOOD AND DRUG ADMINISTRATION ANTIMICROBIAL DRUG DEVELOPMENT WORKSHOP Thursday, April 15, 2004 9:05 a.m. Advisors and Consultants Staff Conference Room 5630 Fishers Lane Rockville, Maryland 2 PARTICIPANTS John E. Edwards, Jr., M.D., Moderator INDUSTRY: Lisa Benincosa, Ph.D. Mike N. Dudley, Pharm.D. Barry Eisenstein, M.D. Dennis M. Grasela, Pharm.D., M.D. Timothy J. Henkel, M.D., Ph.D. John H. Rex, M.D. Frank Tally, M.D. Gregory A. Winchell, Ph.D. ACADEMIA: Ralph Corey, M.D. William A. Craig, M.D. Harmut Derendorf, Ph.D. George L. Drusano, M.D. Thomas Fleming, Ph.D. Robert J. Guidos, J.D. Sheldon L. Kaplan, M.D. W. Mike Scheld, ,M.D. Jerome J. Schentag, Pharm.D. George Talbot, M.D. Paul M. Tulkens, M.D. FDA: Renata Albrecht, M.D. Edward Cox, M.D., MPH Mark J. Goldberger, M.D., MPH John Powers, M.D. David Ross, M.D., MPH Janice Soreth, M.D. MISCELLANEOUS: John S. Bradley, M.D. Dennis M. Dixon, M.D. J. Todd Weber, M.D. 3 C O N T E N T S Opening Remarks, John E. Edwards, M.D. 4 I. Progress Report Since Last Workshop IDSA, W. Mike Scheld, M.D. 9 FDA, Edward Cox., M.D., MPH 21 Discussion 36 II. Continuing Discussion on Incentives for Drug Development IDSA, George Talbot, M.D. 41 FDA, Mark Goldberger, M.D., MPH 62 Discussion 85 III. Microbiologic Surrogate Endpoints in Clinical Trials IDSA, Sheldon Kaplan, M.D., Baylor College of Medicine 137 INDUSTRY, Barry Eisenstein, M.D., Cubist Pharmaceuticals 148 FDA, John Powers, M.D. 163 Discussion 187 IV. DISCUSSIONS ABOUT BACTEREMIA AS AN INDICATION/ ISSUES WITH CLINICAL TRIALS IN ENDOCARDITIS INDUSTRY, Timothy J. Henkel, M.D., Ph.D. 249 IDSA, Ralph Corey, M.D. 261 FDA, David Ross, M.D., MPH 277 Discussion 288 4 1 P R O C E E D I N G S 2 Opening Remarks 3 DR. EDWARDS: Welcome, everyone. My name 4 is Jack Edwards and I am from the Harbor UCLA 5 Medical Center School of Medicine at UCLA, and also 6 am a member of the Antimicrobial Availability Task 7 Force of the IDSA, and I will be coordinating the 8 meeting today. 9 I would first like to say that we are very 10 grateful to the staff at the FDA and the IDSA for 11 several months of a great deal of work that has 12 gone into the preparation of this meeting, which is 13 really the second in what we hope is a continuum of 14 meetings to address this issue that we, from the 15 IDSA, have become quite acutely aware of, and that 16 is the long-term, if you will, paradoxical 17 diminishment in the availability of anti-infectives 18 over the last several years, particularly in the 19 area of antibacterials, which is somewhat 20 paradoxical as it is occurring at a time when our 21 needs are just increasing dramatically due to 22 resistance problems and we have the specter of 5 1 bioterrorism in the background as well. 2 The IDSA has been studying the issue 3 acutely over the last two years and this meetings 4 is a result of efforts to try to bring the issue 5 into a sharper focus and for all of us to think of 6 solutions that will correct this trend. The 7 meeting, we are hoping, is going to be informal. 8 This is not an FDA advisory board meeting, and I am 9 going to be encouraging informality and interchange 10 as much as possible during the whole meeting. Some 11 of you may remember at the last meeting we had a 12 little trouble breaking the ice at the beginning 13 and I started to tell my favorite biostatistician 14 joke, but then I realized that there were several 15 biostatisticians in the room who might not think it 16 was funny so we passed on that. But I do want to 17 try to keep this informal, collegial, interactive 18 and we will be trying to stimulate conversation as 19 we go along the whole way. 20 Just a couple of brief announcements, we 21 have changed the order a bit from what you saw 22 posted on the web site. There are revised 6 1 schedules on the table there and I hope you all 2 have them. So, the order will follow that 3 dispersed schedule. I will make some other 4 announcements later regarding lunch, and so forth, 5 as we go along. 6 Without delaying, I want to formulate 7 thanks to some individuals and I am going to do 8 that throughout the meeting here. I would like for 9 us to begin by going around the table and each 10 person at the table introducing themselves with 11 their affiliation. Then we will get right into our 12 first discussion area. Mike, would you begin? 13 DR. SCHELD: Sure. I am Michael Scheld. 14 I am from the University of Virginia and I am here 15 representing the IDSA. I am the immediate past 16 president of the Society. DR. TALBOT: Good 17 morning. My name is George Talbot. I am with 18 Talbot Advisors and I am here with IDSA today. 19 DR. DERENDORF: I am Harmut Derendorf, 20 from the University of Florida, and I am president 21 elect of ISAP. 22 DR. ROSS: David Ross, and I am with the 7 1 Division of Anti-infective Drug Products, FDA. 2 DR. COX: Ed Cox, Deputy Director for ODE 3 IV, CDER, FDA. 4 DR. GOLDBERGER: I am Mark Goldberger, the 5 Director of ODE IV, CDER, FDA. 6 DR. POWERS: John Powers, Lead Medical 7 Officer for Antimicrobial Drug Development and 8 Resistance Initiatives in ODE IV. 9 DR. ALBRECHT: Renata Albrecht, Director, 10 Division of Special Pathogen and Immunologic Drug 11 Products, FDA. 12 DR. WEBER: Todd Weber, I am the Director 13 of the Office of Antimicrobial Resistance, National 14 Center for Infectious Diseases at CDC. 15 DR. DIXON: I am Dennis Dixon. I am from 16 the National Institutes of Health, National 17 Institute of Allergy and Infectious Diseases, and I 18 am Chief of the Bacteriology and Mycology Branch. 19 DR. FLEMING: Thomas Fleming, Chair of the 20 Department of Biostatistics, University of 21 Washington, and I am disappointed I didn't hear 22 your opening joke last time! 8 1 [Laughter] 2 DR. TULKENS: I am Paul Tulkens, from 3 Brussels, Belgium. I am the past president of ISAP 4 and I am running a pharmacology group at the 5 Catholic University of Louvain, Brussels. 6 DR. WINCHELL: I am Greg Winchell, from 7 the Clinical Drug Metabolism Department at Merck. 8 DR. HENKEL: I am Tim Henkel. I am the 9 Chief Medical Office of Vicuron Pharmaceuticals. 10 DR. BENINCOSA: I am Lisa Benincosa, from 11 Pfizer Global Research and Development. 12 DR. REX: I am John Rex. I am the Medical 13 Director for Infections, AstraZeneca 14 Pharmaceuticals. 15 DR. COREY: I am Ralph Corey. I am from 16 Duke University, here on behalf of IDSA. 17 DR. TALLY: I am Frank Tally. I am Chief 18 Scientific Officer at Cubist. 19 DR. DRUSANO: George Drusano, Co-director 20 of Ordway Research Institution and a third past 21 president of ISF. 22 DR. EISENSTEIN: Barry Eisenstein, Head of 9 1 R&D at Cubist. 2 DR. CRAIG: Bill Craig, from the 3 University of Wisconsin, also a past president of 4 ISF. 5 DR. BRADLEY: John Bradley, from 6 Children's Hospital, San Diego. I am representing 7 the IDSA. 8 DR. EDWARDS: Thank you very much. At the 9 end of the last meeting we really developed a 10 rather intricate set of plans for progress to be 11 discussed at this meeting. I am going to now ask 12 Mike Scheld to begin with his comments regarding 13 the progress report from the IDSA since the last 14 meeting. Mike, thanks. 15 I. Progress Report Since Last Workshop 16 IDSA 17 DR. SCHELD: Thanks very much, Leo and 18 thank you Jack. 19 [Slide] 20 I thought I would start out with a slide 21 that I took of the rotunda at the University of 22 Virginia. As you have heard, I am Michael Scheld. 10 1 I am from the University of Virginia. If you can 2 dim the lights in here, it looks a lot better. If 3 you don't know, Mr. Thomas Jefferson founded the 4 University of Virginia and he considered it one of 5 his greatest achievements. In fact, if you visit 6 Monticello, his home and his grave site you won't 7 even find the fact that he was the third president 8 of the United States on his head stone. It 9 mentions the University of Virginia and a few other 10 things, like the Declaration of Independence. 11 To put his life in perspective, I will 12 just tell you one little anecdote. When John 13 Kennedy was president he had a dinner here in 14 Washington, at the White House, for all the Nobel 15 prize winners that were then alive. This was back 16 in 1963. At the end of the dinner he made the 17 comment--there were about 150 Nobel laureates in 18 attendance and he made the comment that this room 19 has not seen such an assemblage of intellect since 20 Thomas Jefferson dined here alone. That sums it up 21 pretty well. 22 [Laughter] 11 1 [Slide] 2 As you have heard, we formed an 3 Antimicrobial Availability Task Force within the 4 Infectious Disease Society of American, and what I 5 am planning on doing today is to just remind 6 everybody where we were in November, 2002 in our 7 workshop. 8 A perception that we had at the time which 9 was basically that as resistance was rising 10 research and development of antibacterials was 11 declining; some evidence for that perception in the 12 interim, and then I will tell you a little bit 13 about the work of the task force. 14 This is really a two-part presentation. I 15 am going to go over the membership of the task 16 force and what I think their charge should be and 17 how they should go about their business, because I 18 put this task force together during my presidency 19 at IDSA. Then George Talbot will take up the 20 second half where we will tell you a little bit 21 more about what we have learned along the way, as 22 well as some potential solutions. 12 1 [Slide] 2 So, back in November of 2002 I think the 3 themes or the workshop were basically a delta 4 issue, and we are not going to go there today, 5 fortunately; that antibacterial resistance was 6 increasing at a time when antibacterial research 7 and development was declining. We talked a good 8 deal about PK/PD and surrogate endpoints, and you 9 will hear more about that during the course of this 10 workshop. Then we discussed three disease states, 11 acute exacerbation of chronic bronchitis, 12 meningitis and hospital-acquired pneumonia. At the 13 end I will make a comment about the AECB efforts of 14 the IDSA as well. 15 [Slide] 16 So, this was our perception, and the years 17 at the bottom are completely arbitrary. Obviously, 18 antibacterial research and development was 19 declining before 1998, as well antibacterial 20 resistance was increasing. 21 [Slide] 22 Some evidence for this, and this is just 13 1 one example from the CDC where you can see the 2 increase in MRSA as well as vancomycin resistant 3 enterococci, resistant pneumococci and 4 fluoroquinolone resistant Pseudomonas. In my own 5 hospital about 40 percent of our bloodstream 6 isolates of Staph. aureus are now MRSA. About 15 7 percent of enterococcal bloodstream isolates are 8 VRE. We have about 35 percent overall penicillin 9 resistance to pneumococcus in central Virginia. 10 Our fluoroquinolone resistant Pseudomonas is now in 11 excess of 30 percent. 12 [Slide] 13 There are many threats to antibacterials. 14 I think one is bacterial resistance. Obviously, 15 drug shortages in recent years--we have had 16 problems with a supply of penicillin, gentamicin, 17 meropenem and others. The pipeline is rather dry 18 and there is a void in public policy which is 19 something that the IDSA has tried to address. 20 [Slide] 21 This shows antimicrobial research and 22 development from a paper by Spellberg et al. which 14 1 includes Dr. Edwards as one of the co-authors, 2 which I am happy to say is in press and maybe it is 3 published today. 4 DR. EDWARDS: I think it is today. 5 DR. SCHELD: Today--tax day; happy tax 6 day, everybody. You can see your taxes at work 7 here! It is published today in Clinical Infectious 8 Diseases, if you want to access it. This shows you 9 the total number of new antibacterial agents during 10 five-year intervals. Back in 1983-'87, you can see 11 that that number was 16 and it steadily went down 12 until 1998-2002 when it was only 7. Last year 13 there were only 2 new antibacterials approved. 14 [Slide] 15 This is just since 1996 where you see the 16 number that have been actually approved and 2002 17 was a pretty bad year for antibacterial approvals. 18 We had zero in that year. In 2003 there were two 19 again. 20 [Slide] 21 So, in 2002, out of 89 new drugs there 22 were no new antibacterial drugs that were approved. 15 1 If you look through the annual reports of the 15 2 major pharmaceutical companies, you find nearly 400 3 agents in development but only 5 of them are new 4 antibacterials. 5 [Slide] 6 Since 1998 there have been a new drugs 7 approved, but I would just point out here that only 8 a few of them are truly novel in that they have a 9 new target, like linezolid as well as daptomycin, 10 in 2003, and we have clarithromycin which was 11 approved. I guess it was about April 1 of this 12 year. So, there are not very many, and many of 13 them also hit another target. 14 [Slide] 15 So, back in December of 2003 the NDA 16 pipeline, or so-called "pink sheet," lists a few 17 drugs in Phase 3 antibacterial development. Only 18 tigecycline is really a broad spectrum agent that 19 has activity against many gram-positive and 20 resistant gram-negative pathogens. I just point 21 this out because in the same publication there are 22 18 novel oncology agents listed. There are 16 1 obviously a lot more you could add to this list, 2 like dalbavancin and others but this is just one 3 example in the published literature of what is in 4 Phase 3. 5 [Slide] 6 I am going to skip this in the interest of 7 time because George is going to show it in a few 8 minutes. 9 [Slide] 10 So, we put together an Antimicrobial 11 Availability Task Force and this is their charge, 12 which is to develop novel public policy to ensure a 13 sustainable supply of safe and effective 14 antimicrobial drugs to protect public health. The 15 initial focus was on the antibacterial development 16 because that was the area where we felt the crisis 17 was most acute. 18 [Slide] 19 The task force members are listed here. 20 It is probably pretty hard to see them, but I would 21 like to acknowledge and thank each of them 22 individually. John Bartlett was kind enough to 17 1 chair the task force. He couldn't be with us 2 today. He is from Hopkins. John Bradley is here. 3 Jack Edwards is here. David Gilbert deserves our 4 thanks because he has really done a lot of work on 5 the task force and, in addition, he worked very 6 hard at putting this workshop together. I serve. 7 Dave Shlaes is here, as is George Talbot, Frank 8 Tally and Dave Ross and John Powers have been 9 really a tremendous help in framing our thoughts 10 for how to move forward. I also should acknowledge 11 Bob Guidos who is sitting here as head of public 12 policy as IDSA. Our staff have been 13 extraordinarily helpful in putting together our 14 work plan. 15 [Slide] 16 This is the work plan as I envisioned it 17 when we formed our task force. It was basically to 18 understand the problem; publish our research and 19 the findings of our surveys. The first of these 20 publications is published today. Discuss with the 21 stakeholders--we made many field trips and I will 22 give you a couple of examples. The "white paper" 18 1 is in production, which will be used to discuss 2 this issue with other stakeholders, including 3 policy makers, congressional leaders, etc. We hope 4 to have that out within another few weeks. Then, 5 develop some solutions and you will hear about some 6 of our suggestions from Dr. Talbot in a few 7 minutes. 8 [Slide] 9 Back in October of 2001, CDC, NIH, FDA and 10 others put together a public health service action 11 plan to combat antimicrobial resistance, and there 12 were three major elements: to stimulate the 13 development of priority products to combat 14 antimicrobial resistance; streamline the regulatory 15 process and to identify incentives for development. 16 I think it is fair to say that all three of these 17 elements are included in our "white paper" and have 18 formed our thinking on many of these issues over 19 the last 18 months. 20 [Slide] 21 We call our "white paper" "Bad Bugs, No 22 Drugs." We have sought input from the major 19 1 stakeholders in this issue, including our 2 membership which now is over 7,500 physicians, 3 researchers and health care providers. To put that 4 in some perspective, that membership in the early 5 1990s was about one-third of that, about 2,500. We 6 have had many discussions with the FDA, CDC, NIAID, 7 Health and Human Services and senior pharmaceutical 8 executives, as well as venture capital companies 9 and members of Congress and legislators. 10 [Slide] 11 So, we have met with a number of the 12 groups listed on the right. We met with Tony Falci 13 and many others at NIAID. We have had discussions 14 with HHS and CDC, Judy Gerben and Jim Hughes. We 15 met with Commissioner McClellan as a group from 16 IDSA. We have had many meetings with Congress. We 17 have met with representatives of a number of 18 pharmaceutical companies that are listed there: 19 Abbot, Bristol-Myers Squibb, GaxoSmithKline, 20 Novartis, Pfizer and Vicuron, and others including 21 venture capital. We will outline what we learned 22 from some of those field trips in a few minutes 20 1 with Dr. Talbot's presentation. 2 [Slide] 3 Lastly, when we met before I think most of 4 us would agree that the literature on the treatment 5 of acute exacerbation of chronic bronchitis and the 6 role of antimicrobial therapy, at best, is 7 unsettled. So, we have basically put together 8 another joint task force between the IDSA and the 9 American Thoracic Society. There are 12 10 individuals involved in this. The chair from the 11 IDSA side of it is Tim Murphy, from State 12 University of New York in Buffalo. And, they have 13 started their work which is basically to develop a 14 protocol and budget. I had envisioned this to be 15 placebo, an old antibiotic and a new antibiotic 16 type of clinical trial. Implement a network 17 because such a network for bacteriology and 18 bacterial infections really needs to be done 19 nationwide; and then submit this to the NIAID for 20 funding. 21 Where it stands is they have started their 22 work and I am hoping that by our June board of 21 1 directors meeting for the IDSA we will have a draft 2 protocol to start to discuss at that level and 3 submit it later this summer. 4 So, Jack, that is kind of what we have 5 been up to in the IDSA and I will leave it up to 6 Dr. Talbot to give the second half of this 7 presentation, but thank you very much. 8 DR. EDWARDS: Thank you very much, Mike. 9 I am now going to ask Ed Cox, from FDA, to comment 10 from the FDA perspective. 11 FDA 12 DR. COX: Good morning and welcome, 13 everyone. 14 [Slide] 15 What I will be doing today is providing an 16 update since our last meeting in November of 2002 17 of some of the activities that folks at the FDA 18 have been involved with related to antimicrobial 19 resistance and antimicrobial drug development. 20 I will cover a variety of topics, 21 including some of the activities we have been 22 involved with at scientific meetings, both advisory 22 1 committee meetings and workshops; the work that is 2 ongoing with regards to guidance development; the 3 work that John Powers has been working tremendously 4 hard at with the folks at Focus Technologies and 5 our database through our contract with them. Then 6 I will also discuss some of the input that we got 7 from an advisory committee with regards to the 8 criteria for resistant pathogens of public health 9 importance, and then just briefly mention the newly 10 announced Critical Path initiative. Then I will 11 also just touch on some of the work that we have 12 been involved in at FDA with regards to preserving 13 the utility of our existing antimicrobial agents. 14 [Slide] 15 First, just to mention the March 4 FDA 16 anti-infective advisory committee. As part of that 17 meeting, there were presentations on the patterns 18 of antimicrobial resistance in Streptococcus 19 pneumonia and how this patterns, based on the data 20 analysis from the Focus database, related to 21 scientifically-based resistant pathogen claims for 22 Streptococcus pneumoniae. 23 1 Again, John Powers work showed the high 2 rate of cross-resistance between penicillin 3 resistant strains of Streptococcus pneumoniae and 4 many of the other commonly used antimicrobial 5 agents for respiratory tract infections. The 6 result of this was the concept of the multi-drug 7 resistant Streptococcus pneumoniae labeling claim. 8 [Slide] 9 Just to show some of the data from Focus 10 Technologies that were used in supporting the 11 concept of the multi-drug resistant Streptococcus 12 pneumoniae claim, this sort of schematic of the 13 scatter plot, that I will be showing you in just a 14 minute, shows drug Y on the Y axis with the 15 increasing MICs from the origin outward. Then on 16 the X axis, drug X, again increasing MICs from the 17 origin outward. So, those strains that are highly 18 susceptible to both drug X and drug Y will 19 congregate in the lower left hand corner. Those 20 strains that are resistant to both strains will 21 congregate in the upper right-hand corner. 22 [Slide] 24 1 If we look at a drug like cefuroxime, 2 which is shown here on the Y axis again with 3 increasing MICs and penicillin on the X axis with 4 increasing MICs from the origin outward, we see the 5 correlation of resistance between these two drugs 6 with the isolates congregating on the diagonal. 7 [Slide] 8 Just to contrast, this is a quinolone 9 antimicrobial agent on the Y axis and penicillin on 10 the X axis, and here we don't see the correlation 11 that we had seen with cefuroxime and penicillin 12 during the time period at which these isolates were 13 collected. 14 So, these data were very helpful to us as 15 one of the advisory committee discussions about 16 multi-drug resistant Streptococcus pneumoniae, and 17 were important in the subsequent MDRSP claims that 18 have since been awarded in labeling. 19 [Slide] 20 Also, on March 5 at the FDA Anti-Infective 21 Advisory Committee we had the opportunity to get 22 advice from the committee with regards to 25 1 developing criteria for resistant pathogens of 2 public health importance. The idea behind these 3 criteria is to identify those resistant pathogens 4 of public health importance that would warrant 5 claims in product labeling. The discussions at the 6 advisory committee helped refine the criteria for 7 resistant pathogens of public health importance. 8 In addition, there was discussion about the pros 9 and cons of a list that was brought up at the 10 advisory committee meeting. 11 [Slide] 12 The criteria that were discussed and 13 somewhat refined based upon the advice of the 14 advisory committee are that for a resistant 15 pathogen of public health importance in a 16 particular indication, that there be limited 17 available therapies due to multi-drug resistance of 18 the organism if the organism caused serious or 19 severe disease; that the drug to which the organism 20 is resistant, that is the drug within the resistant 21 pathogen claim, is commonly used in the disease 22 under study; that the organism is of sufficient 26 1 prevalence in the population with the disease under 2 study, or that the drug is used to control the 3 spread of disease within a population, for example 4 an anti-tuberculosis agent in the treatment of TB 5 would prevent the spread of TB in a population. 6 Then, also that there be clinical correlation of in 7 vitro resistance with core clinical outcomes. 8 [Slide] 9 Just to provide a little more comment on 10 the criteria, the idea is not that the pathogen 11 would need to fulfill al the criteria on the list, 12 but that these would be the criteria that would be 13 evaluated in looking at the resistant pathogen to 14 determine if it rose to the level of being a 15 resistant pathogen of public health importance 16 within a particular indication. 17 When we look at resistant pathogen claims, 18 typically what we are looking for is evidence of 19 activity in the treatment of the particular 20 indication, including successful treatment of 21 susceptible strains of the pathogen. Then also, in 22 addition, clinical activity in the treatment of the 27 1 resistant pathogen of interest in the particular 2 individual of interest. Part of the rationale here 3 is that there may be differences in the patient 4 characteristics of those patients who have 5 resistant organisms compared to those who have 6 susceptible organisms. 7 With regards to the ultimate application 8 that would result, priority review be based upon 9 the results of the clinical trials. It is also 10 important to note that there may be different 11 approaches here to bringing the drug application in 12 and that a drug application that comes in that 13 doesn't have all the data in hand for the resistant 14 pathogen claim can still be judged on the merits of 15 its safety and efficacy for the treatment of the 16 indication, and may then subsequently come in at a 17 later point in time for a resistant pathogen claim 18 if sufficient data were accrued. 19 [Slide] 20 There are a number of previously granted 21 resistance claims. I will just mention a couple of 22 them here, methasone resistant Staph. aureus with 28 1 claims for VRE, claims for beta-lactamase producing 2 H. flu, penicillin resistant Strep. pneumo. and 3 then more recently multi-drug resistant Strep. 4 pneumo. in resistance claims that have been 5 previously awarded in product labeling. 6 [Slide] 7 Also during the March 5 meeting, there was 8 an interesting discussion about relating clinical 9 data from one disease to another. The idea really 10 and the concept behind this is that for a drug 11 application that comes in and that covers a number 12 of different indications within the respiratory 13 tract, if that package of data were well anchored 14 in two adequate and well-controlled studies in 15 community-acquired pneumonia and there were also 16 studies in AECB and acute bacterial sinusitis, it 17 may not be necessary, you know, for a program that 18 is well anchored in two adequate and 19 well-controlled CAP studies to have two additional 20 studies in each of the additional respiratory tract 21 infection indications that are coming in as part of 22 the package. Part of this would depend upon the 29 1 single studies that would be conducted in the other 2 indications would be done in a well characterized 3 population and have good microbiologic 4 characterization of the patients under study. 5 Some of the concepts that were brought up 6 at the committee with regards to such an approach 7 would be that it would be important to have similar 8 microbial etiologies across the indications; 9 similar tissue distribution of the antimicrobial 10 agents in the target organs; and also that other 11 factors that should be considered would be the 12 severity of the disease. For instance, typically 13 any sort of supporting data we extrapolate from the 14 more severe disease to the less severe disease, not 15 the opposite way. Also, that there be 16 consideration of the host factors. It would be 17 more difficult or the directionality here would be 18 one from taking the data from--it wouldn't be 19 appropriate to take data from an immune competent 20 population and use that to support an indication in 21 an immune suppressed population because there may 22 be host factors there that would need to be 30 1 considered. 2 [Slide] 3 In September of 2003 there was a BAMSG/FDA 4 workshop that involved folks from academia, the 5 industry and also the FDA in discussions of 6 clinical trial design for empiric antifungal 7 therapy in febrile neutropenic patients. There 8 were a lot of interesting discussions about the use 9 of fever both as an inclusion criterion and then 10 also with regards to endpoints in studies of 11 patients for antifungal therapy for febrile 12 neutropenia. There were also discussions on 13 clinical trial design for investigating combination 14 antifungal therapies. For looking at a regimen of 15 A plus B, how do we know or how do we demonstrate 16 that there is an advantage of A plus B beyond the A 17 and B agents from which the regimen is composed? 18 [Slide] 19 Then, in October, 2003, additional 20 discussions of the anti-infective drug products 21 advisory committee on clinical trial design for 22 diabetic foot infections, defining the condition 31 1 itself; getting the inclusion and exclusion 2 criteria that would be used in a clinical trial. 3 Also issues about getting microbiologic diagnosis 4 in patients in patients with diabetic foot 5 infections were discussed, and also how to measure 6 the outcomes in these trials. 7 Clinical trial design in acute bacterial 8 sinusitis--there was also a lot of discussion that 9 was helpful to us in our ongoing efforts to update 10 the acute bacterial sinusitis guidance document. 11 The discussion centered around ways that we may 12 enrich the population in the acute bacterial 13 sinusitis studies such that a larger proportion of 14 the patients have bacterial disease. Really, you 15 know, the ultimate role here is if we can enhance 16 the population for those patients that have acute 17 bacterial sinusitis we may be able to demonstrate a 18 larger effect size in such a population and, 19 thereby, the trials my be able to be done more 20 efficiently. 21 There was also discussion about the role 22 of microbiologic diagnosis. Currently, sinus tap 32 1 is the procedure that is used in the microbiologic 2 diagnosis of acute bacterial sinusitis. There was 3 also some discussion about some newer technology 4 that might be used to diagnose acute bacterial 5 sinusitis and make a microbiologic diagnosis of 6 acute bacterial sinusitis. 7 Another issue that was discussed by the 8 committee was in those patients with acute 9 bacterial sinusitis, who don't have urgent acute 10 bacterial sinusitis, the possibility of dong 11 superiority trial designs which would be something 12 like an active plus symptomatic therapy versus 13 other symptomatic therapy. Again, such an approach 14 may allow for more efficient study of acute 15 bacterial sinusitis to be conducted. Another 16 possible option would be a dose-response approach. 17 [Slide] 18 We are updating and developing new 19 guidance documents in a variety of areas, including 20 acute bacterial sinusitis, acute bacterial 21 exacerbation of chronic bronchitis, acute otitis 22 media, acute bacterial meningitis and drug 33 1 development for resistant pathogens. A lot of the 2 advice that we have gotten at the advisory 3 committees over the last year and a half has really 4 been very helpful to us in updating our guidance 5 documents, and we hope to have several of these out 6 really quite soon. 7 [Slide] 8 One other thing I just wanted to mention 9 is the Critical Path initiative. The Critical Path 10 "white paper" was issued in March of 2004, and it 11 recognizes the advances in basic biomedical 12 sciences and the high cost of bringing a medical 13 product to market. The idea behind the Critical 14 Path initiative is to make the process of bringing 15 a product to market a more efficient process. It 16 calls for advances in the applied sciences in 17 medical product development, in essence, to develop 18 a better product development tool kit. The idea is 19 if we can get better tools to assess safety and to 20 demonstrate efficacy earlier on, we may be able to 21 more efficiently guide drug development. This will 22 be an effort that will require the joint efforts of 34 1 academia, industry and FDA in order to successfully 2 bring it forth. 3 [Slide] 4 Then just changing gears a little bit and 5 talking about some of our efforts to preserve the 6 utility of our existing antimicrobial agents, the 7 final rule labeling requirements for systemic 8 antimicrobial drug products intended for human 9 issue was issued in of 2003 and became effective in 10 February of 2004. This rule amends our labeling 11 regulations for systemic antimicrobial drug 12 products to require that they include a statement 13 in labeling about appropriate and prudent use of 14 antimicrobial agents. It also encourages 15 physicians to counsel their patients about 16 appropriate antimicrobial use. It is estimated to 17 impact approximately 669 systemic antimicrobial 18 drug products. 19 [Slide] 20 Another component of our efforts to 21 preserve the utility of existent antimicrobial 22 agents is the "Get Smart" program that is 35 1 co-sponsored with the CDC. It is an education and 2 outreach program that is directed towards both 3 consumers and then also health professionals. The 4 goal here, again, is to inform both the public and 5 health professionals about the importance of using 6 antimicrobial agents appropriate and prudently in 7 order to preserve their useful life span. 8 [Slide] 9 In summary, since the November 2002 10 meeting we have had a number of scientific meetings 11 on labeling issues and clinical trial design for a 12 number of indications. There has subsequently 13 been, since the meeting in 2003, awarding of 14 multi-drug resistance Streptococcus pneumoniae 15 claims. We have gotten important input on the 16 criteria for resistant pathogens of public health 17 importance. We have had discussions about 18 streamlining drug development through relating 19 clinical data from one disease indication to 20 another and how we might approach that in a 21 conceptual fashion. We have gotten a lot of 22 important input about clinical trial designs for a 36 1 variety of different therapeutic indications. 2 These meetings and the science that has come out of 3 them have been very important to us in updating the 4 guidances, and our work is ongoing there and we 5 hope to have a number of the guidances documents 6 out soon. Then, just briefly, the Critical Path 7 initiative, which will have an impact on drug 8 development in the near future; then our ongoing 9 efforts to preserve the utility of our existing 10 antimicrobial agents. So, there has been a lot 11 done since the last meeting and there is still more 12 to do. With that, I will close. Thank you. 13 Discussion 14 DR. EDWARDS: Thank you very much, Ed. 15 That was very nice. Now we have a few moments for 16 discussion and I would like to invite any comments 17 anyone would like to make. I might just start by 18 reflecting on the fact that I think it is obvious 19 that a wholehearted effort has been made by 20 everyone involved as a result of the meeting last 21 time. Virtually every point both of you have 22 addressed was brought up at that meeting and plans 37 1 were made to go forward to address them, and it is 2 very gratifying to see the responsiveness. 3 Are there any questions from the audience? 4 I am going to ask one and I realize is going to be 5 somewhat difficult to answer, but regarding the 6 guidances, you mentioned soon. Is it possible to 7 give us a little more thought about what that means 8 exactly? 9 DR. COX: We have made good progress and I 10 think there are three or four that we hope to have 11 out prior to quarter four of 2004. We are working 12 hard and we hope to have them out soon. You know, 13 one other comment too, from these meetings we do 14 get a lot of important scientific information and 15 we can informally or actually in the setting of 16 meetings guide companies and use this information. 17 So, the updating of the guidance documents is 18 something that we are working hard on and hope to 19 have out there soon. 20 DR. EDWARDS: Any other questions? Let me 21 ask one other, Ed. In your description of the 22 finding of resistance, it wasn't quite clear, is 38 1 that a combined effort with CDC or is that within 2 the FDA advisory committee exclusively? 3 DR. COX: Yes, there has been input that 4 was discussed at the FDA advisory committee and 5 folks from the CDC were present, along with our 6 guest representatives and also the advisory 7 committee. So, in essence, I would say that is an 8 effort in developing those criteria where we have 9 had input from both our advisory committee and also 10 from folks at the CDC to help guide the further 11 development of those criteria. 12 DR. EDWARDS: Yes, in the back? 13 PARTICIPANT: [Not at microphone; 14 inaudible]. 15 DR. EDWARDS: To get it on the record we 16 will have to have it repeated into the microphone. 17 DR. COX: So, the question was about the 18 use of a development program which would be 19 anchored in, say for instance, two studies in 20 community-acquired pneumonia and then for 21 additional studies in the respiratory tract, such 22 as acute bacterial sinusitis, acute exacerbation of 39 1 chronic bronchitis were it may not be necessary to 2 do two adequate and well-controlled studies; it may 3 be possible to do one well characterized adequate 4 an well-controlled study in combination with the 5 two studies anchored in community-acquired 6 pneumonia, and the question asked would it be 7 advisable to meet with the FDA, if one were 8 considering such an approach. I think the answer 9 there is definitely yes, to get folks in and talk 10 to us about their proposed program because this is 11 an issue where I think it is important to get an 12 understanding of the protocols that will be used 13 for those other studies, and certainly having a 14 well characterized study will be very important in 15 the setting of trying to use support across the 16 indications. Also, I think it will be important to 17 have a discussion with anyone who is proposing to 18 take such an approach in order to understand which 19 indications they were planning to rely upon for 20 some degree of cross support. So, I do think it 21 would be advisable to come in and talk with us and 22 we would certainly welcome the opportunity to do 40 1 so. 2 DR. EDWARDS: Thank you. George? 3 DR. DRUSANO: This is just a quick 4 question for Dr. Scheld. Michael, one of the 5 things that you had presented was the IDSA 6 initiatives to provide incentives for sponsors to 7 stay in or reenter antimicrobial drug development. 8 Many of those proposed kinds of incentives I think 9 are probably beyond the scope of regulatory 10 agencies and actually fall within the purview of 11 Congress in terms of actually having to change 12 laws. Is there any way that the outcomes of these 13 kinds of meetings are being sent to appropriate 14 congressional leaders for this kind of 15 consideration? 16 DR. EDWARDS: George, if I may, that was a 17 perfect introduction to George Talbot's address and 18 he is going to be addressing the specific answer to 19 that question. So, George, I will ask you to go 20 ahead now, if you will, please. 21 II. Continuing Discussion on Incentives for 22 Drug Development - IDSA 41 1 DR. TALBOT: Well, good morning, everybody 2 and thank you for the opportunity to continue the 3 discussion on incentives. 4 [Slide] 5 I am doing so on behalf of the 6 Antimicrobial Availability Task Force of IDSA. It 7 has been my privilege to participate in that group. 8 I have to say, in looking at this first slide, that 9 it makes me realize that it is always different to 10 follow Mike Scheld as a speaker, not only because 11 of his articulate expression of difficult concepts 12 but also because of his first slide, which is that 13 beautiful slide of the rotunda and I have no 14 similar slide. As I was sitting here I was 15 thinking, you know, I work out of the home and it 16 probably wouldn't be appropriate to show a picture 17 of the rotunda in my home, but I will have to come 18 up with something, Mike, to match you next year. 19 [Slide] 20 My presentation objectives for today are, 21 first of all, to update workshop attendees on 22 AATF's efforts to clarify factors responsible for 42 1 the decrease in antimicrobial R&D, which we have 2 been discussing over the years. 3 And, I would like to discuss the full 4 range of possible solutions and not just the 5 financial incentives. The title is incentives but 6 we have been looking at a range of possible 7 solutions and it is not fair really to call all of 8 them incentives. 9 [Slide] 10 As an overview, I can summarize what AATF 11 has learned during the past year. Some of these 12 statements may seem to be truisms but I think it is 13 important to state them up front anyway. 14 We do believe that there is a problem, 15 that there is a decrease in antimicrobial R&D. We 16 had some discussions about whether that was in fact 17 correct or not at the beginning of our meetings, 18 but it seems clear to us that there is a problem. 19 It is clearly a problem that is complex and has a 20 multifactorial etiology. It is also susceptible to 21 oversimplification, the "if only," if only pharma 22 would put more resources into this; if only the 43 1 regulatory process would be quicker; if only 2 Congress would do this or do that. It is just not 3 that simple. There is no easy, single solution 4 that we can see at this point. 5 The good news is that there are potential 6 approaches apparent. One thing that is absolutely 7 clear is that the progress will require a long-term 8 commitment by all of us and its success will be 9 dependent upon the active collaboration of 10 essential partners. 11 [Slide] 12 The problem, as Mike stated, is clear. It 13 is summarized in this Institute of Medicine report 14 for the year 2003. I have highlighted in italics, 15 my italics at the bottom, the statement that as of 16 the time of this report only four large 17 pharmaceutical companies with antimicrobial 18 research programs remained in existence in 2002. 19 One of the questions we had was, well, is that 20 really true or not and we attempted to answer it. 21 [Slide] 22 Regardless of the result or the output of 44 1 what is going on is this slide that Mike also 2 showed you from the Spellberg paper, showing the 3 decrease in total approved antimicrobials in the 4 U.S. We do have two in 2003 and one now in 2004 5 but the trend is certainly down. Interestingly 6 enough, in looking at the FDA's Critical Path 7 paper, there seems to have been a general trend in 8 approvals of new entities over this time frame so 9 it is not just restricted outcome the 10 antibacterials, although all of us, of course, 11 feels this most acutely. 12 [Slide] 13 So, have learned a lot, and primarily from 14 interactions with our stakeholders which include 15 pharmaceutical companies, venture capital 16 interests, FDA, CDC, NIAID and HHS. We have also 17 reached out to the scientific and lay press to 18 discuss the issues with them and, as Mike 19 mentioned, we have had interactions with Congress. 20 [Slide] 21 What are some of the issues for pharma? 22 Well, one thing I want to start with here is 45 1 something that came across to us loud and clear, 2 that many individuals and groups within the 3 pharmaceutical industry are deeply concerned about, 4 and also committed to, the future of antibacterial 5 R&D. There is no question that there are many 6 totally devoted individuals and even some groups 7 for this area and that has to be acknowledged. 8 But it is true also to say that "big" 9 pharma is becoming disengaged. There are some 10 notable exceptions but what we are finding is that 11 the greatest concern is the dearth of resources 12 being applied at the discovery level. So, although 13 there are things later in the pipeline, what we are 14 seeing is a diminution in the resources applied at 15 the discovery level. The reasons for this are 16 clear to those in the room. They have been 17 elucidated elsewhere, but it simply is reduced to 18 the fact that "big" pharma sees a better return 19 from the treatment of chronic diseases, whereas, in 20 contrast, antibacterial therapies are costly to 21 develop, on a part maybe with many other 22 therapeutic agents. But since they are short 46 1 course and used for acute illnesses the potential 2 economic return is less. 3 There is another specific issue here, 4 which is this third one, that many of these 5 antibacterial therapies are not embraced by the 6 marketplace. We are all familiar with the issues 7 of cost which apply across therapeutic areas, but 8 there is the issue of resistance, legitimate 9 scientific concern about promoting the emergence of 10 resistance and that does play a role in how rapidly 11 the uptake of these agents occurs. 12 The other interesting thing we have hard 13 from senior pharma people, which surprised us as ID 14 physicians, is that this is often viewed as a 15 satisfied market. There are antibacterials. We 16 have heard this in three different areas that may 17 reflect the fact that they see it as one market as 18 opposed to segments, but when they look at the 19 market overall they see a satisfied market in 20 comparison to some other potential areas. Finally, 21 these are rarely blockbusters. 22 [Slide] 47 1 What else did we did we hear? Well, for 2 "big" pharma any further uncertainty, regulatory 3 uncertainty for example is a disincentive. Within 4 "big" pharma there is competition for resources 5 among different projects so whenever an 6 antibacterial project is discussed at a resource 7 allocation meeting and it is competing against 8 other therapeutic area products, if there is 9 regulatory uncertainty or marketplace uncertainty 10 it is a disincentive, at least at the margin. 11 Further intellectual property protections 12 are of interest to some but question upside. Tax 13 credits were actually very interesting to one 14 person in senior level pharma, and that is 15 something I will come back to. 16 One key message that came across though 17 from those who are both in and those who are out is 18 that because of the enormous hurdles for 19 establishing and maintaining a discovery 20 infrastructure, it is essential to try to keep 21 those companies that are in this field in because 22 once they go, reestablishing that infrastructure is 48 1 extremely different and time consuming. 2 [Slide] 3 What about "small" pharma? "Small" pharma 4 is more engaged I think in our interactions with 5 them for several reasons. First of all, the 6 financial return is better matched to their size. 7 The market opportunity for them is more clear and 8 for them regulatory uncertainty is probably a 9 lesser concern. We heard this from a few people. 10 If you are focused on antibacterial therapies, 11 well, there is no competitions with other 12 therapeutic areas and if you have a question about 13 how to approach a certain product you can go and 14 talk to Mark, Ed, John and David for example. 15 The focus for these companies is 16 dichotomous. Some are really developing 17 in-licensed compounds only, late stage. Others do 18 have robust discovery efforts, as far as we can 19 tell from looking at the publicly available 20 information on web sites. But the question in the 21 end is will this be enough? Even if there are six 22 or eight of these companies involved in discovery 49 1 research, will they be able to advance these 2 products through to the market? Especially the 3 community market with large trials may be required. 4 So, one of the things that we are seeing is that a 5 lot of these products are focused on the hospital 6 market where the hurdles are less. 7 [Slide] 8 One of the things smaller companies need, 9 of course, is access to venture capital. It is 10 essential. In a totally unscientific sampling 11 which involved speaking to a few VCs, we saw again 12 a dichotomous opinion on this. Some VCs see the 13 dearth of discovery efforts as an opportunity for 14 long-term investment, but others consider this 15 whole area very high risk, specifically because of 16 the restrictions on use of marketed products. What 17 we are seeing happening at the moment is that when 18 companies are being evaluated often it is the 19 late-stage products in the portfolio that are 20 driving the decision whether to finance the company 21 or not. There is a lot of expertise in the room. 22 People may have had different experiences. These 50 1 are generalizations but I think at least in some 2 cases that is true. 3 [Slide] 4 Turning to FDA, we had several 5 conclusions. First of all, it is clear to us, 6 absolutely clear that FDA understands the problem. 7 FDA wishes to partner in finding solutions and 8 regulatory uncertainty they understand, when 9 present, further clouds the development process. 10 This is highlighted in the Critical Path report. 11 With this as a foundation for moving forward 12 though, the FDA has certain unavoidable 13 constraints, and appropriately so. They must 14 maintain scientific rigor in their evaluation. 15 They have to give due consideration to adequate 16 data on safety and efficacy. Also, they have 17 limited flexibility per statutory constraints. For 18 example, they cannot waive user fees for specific 19 high priority products. That requires, as George 20 was asking, legislative intervention. 21 [Slide] 22 We have colleagues here from CDC and 51 1 NIAID. We realize that there are substantial 2 relevant efforts that have been proposed by these 3 groups. I have listed two of them. There have 4 been progress reports for the first. The outline 5 of the NIH "Roadmap" is impressive in terms of the 6 potential help it might give to this process. So, 7 we are enthusiastic about that and hope that these 8 programs can be implemented. 9 We certainly think that more funding could 10 be used for critical efforts, and more could be 11 done to foster inter-agency collaboration, training 12 and outreach specifically regarding antibacterial 13 drug development. Now, I say that last point with 14 the recognition that if you look at the "Roadmap" 15 there really is due acknowledgement of these issues 16 and we are simply adding our voice to try to 17 encourage work in this area. 18 [Slide] 19 What about the scientific and lay press? 20 Well, pipeline concerns are of interest to both the 21 scientific community and the public. That has been 22 clear to us and we are seeing articles directed to 52 1 a wide readership across both areas. 2 One point we have concluded is that our 3 communications to the media should highlight not 4 only measures to decrease resistance or decrease 5 the emergence of resistance, but also to ensure the 6 pipeline. 7 [Slide] 8 I have a sampling on three slides of just 9 some articles that have come out that show that 10 this issue has captured the attention of the lay 11 press, as shown on this slide-- 12 [Slide] 13 --general scientific readership with 14 articles in Nature and Science and The Lancet. 15 [Slide] 16 Then ID specialty journals, with this last 17 one being the one that Jack just mentioned has come 18 out today. 19 [Slide] 20 Congress is a key part of the solution 21 process and there are some vocal and effectiveness 22 supporters for addressing the issues. We have 53 1 found that; it is clear. But it is also clear that 2 the focus of policy-makers is elsewhere at the 3 moment, quite understandably, for bioterrorism and 4 for everything else that faces the country at this 5 point. 6 Bioshield I, which has passed the House 7 and has been given some money by Congress, offers 8 some hope of solutions but it is pretty narrowly 9 focused and, for that and other reasons, has 10 substantial constraints. It seems to us that more 11 attention and action are needed at this level. 12 [Slide] 13 So, what could IDSA do to define 14 solutions? Well, first of all we want to raise 15 awareness of the problem in multiple venues, always 16 speaking outcome the needs of patients. We would 17 like to brainstorm with you, with all our partners 18 in this, on possible solutions, partly because many 19 of these potential solutions are not in our area of 20 expertise. We do hope, however, that we can act as 21 a catalyst when appropriate to help things move 22 forward. 54 1 [Slide] 2 Part of rasing awareness is making a lot 3 of visits. You have heard about the field trips. 4 I have listed here some of the other things that we 5 have done that Mike alluded to also. I would 6 highlight also the "white paper" which will 7 summarize our interpretation and some detailed 8 recommendations, and that is due to be released in 9 May. 10 [Slide] 11 Turning now to potential solutions, I will 12 discuss these in a very broad way and, hopefully, 13 this will foster some discussion if you have all 14 had enough caffeine this morning--I see mostly open 15 eyes out there. One key thing is partnering of 16 stakeholders. This cannot be, in our view, a 17 finger-pointing exercise. We all have different 18 responsibilities, different goals, different 19 agendas but I think there is agreement and the AATF 20 things there is agreement that this is an issue 21 which requires everyone's attention in the interest 22 of the public health, and we hope that we can 55 1 continue to play and important and constructive 2 role in partnering efforts. 3 [Slide] 4 Another potential solution that must be 5 mentioned is whether there could be changes in the 6 marketplace that would alter the economic equation 7 by enhancing greater receptivity to new 8 antibacterials. We heard this theme a number of 9 times. We have to say, however, that we feel that 10 potential change in this area are constrained due 11 to cost, concern regarding promotion of resistance, 12 and a desire to hold antibacterials of last resort 13 in reserve. 14 Actually, I was speaking to a VC the other 15 day and he was very politic. He said, you know, 16 one of the things we see is that because infectious 17 disease physicians are among the most educated and 18 most sophisticated of all physicians, there is a 19 tendency for them to hold these drugs in reserve. 20 It was a phone conversation so I couldn't see if 21 his tongue was in his cheek but I think there is 22 truth to that but, you know, from their perspective 56 1 it is a real disadvantage. 2 Our conclusion at the moment is that in 3 this area change is unlikely unless scientific data 4 are developed to justify different usage patterns. 5 We are not suggesting that; we are just saying that 6 that is what would be required to drive any change, 7 in our opinion. 8 [Slide] 9 Regulatory adjustments are also important. 10 You have heard a lot this morning already from Ed 11 about what is going on. The coming updated 12 guidelines are I think going to be a major step 13 forward. We all believe that. One thing we would 14 encourage is that there be periodic and timely 15 review and revision so that companies would know 16 what the cycle might be for that and perhaps have 17 greater assurance that changes in clinical medicine 18 would be reflected in the guidelines. That is 19 something for FDA to consider. 20 We also believe that it would be useful to 21 encourage novel clinical trial designs to gather 22 information on drug efficacy against resistant 57 1 pathogens. The final bullet is something we will 2 be discussing later today but we do believe that is 3 important. 4 [Slide] 5 NIAID is responsible for implementing the 6 "Roadmap" for translational research. That is an 7 incredibly excellent document in my opinion, having 8 not waded through all of it but through the 9 executive summary, to be honest. 10 There are some specific things we would 11 suggest for consideration that could foster 12 antimicrobial R&D--more collaborative planning with 13 industry and academia, a point made in the 14 "Roadmap." More training, fellowship curriculum, a 15 point made in the "Roadmap." Perhaps using the 16 NCI/FDA model to create an NIAID/FDA program to 17 help streamline development. Funding research into 18 rapid diagnostics in the document is something we 19 have talked about for a long time that would really 20 help in appropriate antimicrobial use and in 21 clinical trial design. Then, Mike mentioned 22 funding placebo-controlled trials. 58 1 [Slide] 2 To turn briefly to legislative, there are 3 ongoing activities that have an impact. The GAO 4 study of this problem that we are talking about 5 today has yet to be launched. Bioshield is in a 6 bit of abeyance at the moment. But the Best 7 Pharmaceuticals for Children Act is a good example 8 of how Congress can come together to create some 9 incentives for industry to pursue particular 10 studies. 11 In the future we have what has been called 12 Bioshield II, appropriately or not, S.666, 13 Lieberman and Hatch, and Dr. Guidos has been 14 working very closely with Chuck Ludlam in that 15 office to understand the implications of that Bill. 16 But we believe overall that unique problems will 17 require unique solutions. 18 [Slide] 19 Here are a few of the things that we are 20 suggesting for legislative consideration. These 21 will be discussed in more detail in the "white 22 paper." But we believe that for priority 59 1 antibacterials, those that meet unmet medical 2 needs, there are several things that might help 3 spur R&D. Use of incentives that have been shown 4 elsewhere to successfully spur R&D, such as R&D tax 5 credits which give up-front now dollars as opposed 6 to dollars at the end of a products life when, in 7 fact, you don't know if the product is even going 8 to be there and making more money. As one 9 pharmaceutical person put it to us, you know, 10 getting us another year at the end of the life of a 11 drug when a drug doesn't make any money to begin 12 with doesn't help us. Up-front dollars might. 13 Supplemental IP protections ought to be 14 considered. There is the wild card patent 15 exclusivity or extension that we can discuss more 16 if there are questions. Also, mechanisms to 17 facilitate the interest and success of smaller 18 companies. I have listed one here. 19 [Slide] 20 We have also suggested that it might be 21 useful to consider a commission on antibacterial 22 resistance with a broad representation from 60 1 stakeholders, those in this room. The charges 2 would be to identify priority pathogens and decide 3 which antibiotics should receive the benefits of 4 legislative initiatives and incentives. 5 [Slide] 6 Another important legislative interaction 7 is to increase funding for essential programs. 8 Those of you who conduct these programs know what 9 you need. We would like to be there to support you 10 in terms of requests that you can verify for 11 appropriate increases in funding. 12 [Slide] 13 Finally, again for fair balance I have to 14 identify the number of groups that have 15 accountabilities and responsibilities. We would 16 note that the corporate world can point with pride 17 to many pro bono initiatives for human health. I 18 have listed a few; there are many others. We don't 19 view this as an actual crisis at the moment but if, 20 for example, this impending crisis explodes with 21 vanco. resistance in community staph., in addition 22 to all these other things that need to be done, the 61 1 public will need the help of the pharmaceutical 2 industry to address this problem. 3 [Slide] 4 These considerations will all be discussed 5 in detail in the "white paper," to be released in 6 May. 7 [Slide] 8 By conclusion, I would say there is a 9 problem. It is multifactorial with no single, easy 10 solution. We strongly believe that essential 11 partners are engaged. That is very, very 12 encouraging to us. We also believe that potential 13 solutions are apparent. 14 As for IDSA's role, we stand ready to make 15 a long-term, constructive commitment to help 16 address this brewing public health crisis. 17 [Slide] 18 So, the question is "Bad Bugs, no Drugs." 19 However, can we help? And we want your input. 20 [Slide] 21 Finally, some final notes--I have listed 22 potential conflicts and I would like to add 62 1 acknowledgments to the AATF members, to the factual 2 input we have received from John and David who are 3 ad hoc members of our task force; IDSA staff and 4 all the people with whom we have spoken. Thank you 5 very much. 6 DR. EDWARDS: Thank you very much, George. 7 That was a very nice summary of a very complex set 8 of issues, all of which we could discuss in 9 extensive detail and, hopefully, will off an on 10 during the course of the meeting. 11 I would like now to ask Mark Goldberger, 12 from the FDA, to comment on the incentives issue. 13 Mark? 14 FDA 15 DR. GOLDBERGER: It is a pleasure to be 16 here. I think George has covered a lot of the 17 issues already. Hopefully, I can made a few 18 additional remarks without being overly redundant 19 to allow adequate time for discussion. 20 [Slide] 21 As everybody has said, the issue is that 22 antibiotic resistance is increasing. Whether it is 63 1 truly the crisis now, we need to keep in mind, as 2 has already been emphasized, that development of 3 new drugs, of course, does not occur overnight. I 4 was interviewed--I forget by what 5 publication--recently and we were talking about 6 this issue and that is the message I tried to 7 emphasize. The great majority of patients can 8 still be treated with the available antibiotic 9 therapy, although there are already pockets in 10 ICUs, etc., where real problems are occurring. But 11 we can't look just to the present. We have to look 12 at the trends in resistance and take into account 13 what we think might happen three, five, seven years 14 from now and recognize that the trends suggest that 15 the problem will get worse and that we need a lead 16 time to get new products out there. That is one of 17 the things always to keep in mind. You can't be 18 thinking just about where you are today; you have 19 to be thinking about where you are likely to be 20 tomorrow. 21 Also, as has been noted and really gets to 22 some of the core tensions in trying to move 64 1 forward, we have to keep in mind that, on one hand, 2 we are looking for new products but, on the other 3 hand, realistically we would like to preserve the 4 usefulness of both those new products and existing 5 products as long as possible. If you think about 6 it, you realize there is a certain tension between 7 doing those two things. 8 [Slide] 9 A lot of the issues have already been 10 covered in terms of thinking about development. 11 One way, maybe very simplistically, to think about 12 them are regulatory/clinical trial issues; 13 scientific/medical issues and then economic issues. 14 George has already talked a lot about all these 15 three categories and I will just make a few 16 additional comments. 17 [Slide] 18 I think there has been a lot of discussion 19 about the need for formal guidances and I think we 20 recognize we have, in some respects, been somewhat 21 slow in getting some updated guidances out there. 22 Sometimes one of the problems with having a lot of 65 1 meetings and interactions is that you are 2 constantly getting new information which, of 3 course, make you think maybe we should modify the 4 guidance but I think it is time to move forward 5 with those that are really of the highest priority. 6 There is some thought a little higher up 7 within our Center, for instance, that maybe one of 8 the issues why there are so many guidances that 9 haven't been completed--and I should say that 10 although we certainly have more than our share on 11 antimicrobial drugs, there are plenty of other 12 guidances that haven't been completed either. Is 13 that they are, in fact, a little too long, 14 sometimes a little too detailed and perhaps there 15 is a way of simplifying them which would allow the 16 time for preparation and getting them out to be 17 shortened. So, that is something that we obviously 18 need to work on. 19 We certainly have been using advisory 20 committee input and we find that to be quite useful 21 as, of course, are meetings like this which have 22 the advantage of being a little more flexible in 66 1 terms of the type of participants and I think that 2 that is extremely helpful. 3 There are also a lot of regulatory tools 4 that already exist. The reason I emphasize this 5 is, as has already been mentioned, you have to look 6 at the things that you can do potentially now and 7 the things that you might like to do but would, for 8 instance, require new legislation. I think 9 everyone understands that the legislative process, 10 for the most part, does not occur overnight, and 11 getting things passed with the many levels of 12 competing priorities is not an easy thing. 13 Therefore, if there are some things you can do now 14 that would be beneficial, you want to really try to 15 take advantage of those. 16 We have regulatory tools that I won't get 17 into now but I talked about a couple of years ago. 18 Our Subparts E and H, fast track designation, all 19 of which talk about how one can expedite the 20 development of drugs for serious illness. They 21 involve increased communication. They involve 22 looking at clinical trial programs. They involve 67 1 uses of surrogate markers, etc., and I will talk a 2 little bit more about some of these issues in a 3 couple of minutes. Some of them will be the 4 subject of additional discussion over the course of 5 this meeting. It is important to keep in mind 6 though that one of the advantages is that these are 7 tools that exist now. Finally, there are certain 8 types of exclusivity that already exist which may 9 be useful, and I will talk a little bit about that 10 in a couple of minutes. 11 I will say for complex issues and for 12 innovative products, at the end of the day I think 13 that still the most important tool that exists, 14 which we certainly use and which someone asked a 15 question about a couple of minutes ago, is actual 16 communication with the company in question over the 17 specifics of their product. Even when guidances 18 exist, remember, guidances are tailored broadly to 19 a specific disease entity or maybe large numbers of 20 types of drugs. They rarely would provide the 21 level of detail that an individual firm, especially 22 with an innovative product, needs to decide how to 68 1 move forward. At the end of the day, that is still 2 the most useful tool that exists. Certainly, in 3 recent times I have had, with a variety of 4 products, that kind of intense interaction and that 5 probably increases, more than anything else, the 6 chance of bringing things to a successful 7 conclusion. I want to emphasize, as I said a few 8 moments ago, that that is still something that we 9 are very interested in doing for products that have 10 the potential to add value. 11 [Slide] 12 We are going to talk throughout the course 13 of this meeting about some clinical trial issues, 14 and one of the goals is to try, as much as is 15 possible, to reduce the size of the clinical trial 16 program. That involves ultimately addressing a 17 tradeoff between our ability to assess, for 18 instance, effectiveness and the resources required 19 to perform a trial. That is, the more data you 20 have about the drug, in general the more you might 21 understand about it for efficacy and certainly from 22 safety, at least insofar as one talks about, for 69 1 instance, rare events. 2 That is not necessarily an insurmountable 3 problem and something you have heard us talk about 4 in the past is the idea of substituting quality for 5 quantity in at least some clinical studies. That 6 is, smaller studies performed on well characterized 7 patients might yield more useful information than 8 very large, open-label studies which may enroll 9 hundreds or thousands of patients but may not 10 provide much in the way of really useful data. I 11 think that is an issue we, hopefully, can move 12 forward on with some of these concepts, as well as 13 what I like to call strengthening the link to 14 clinical inference. It is sort of following up on 15 what Ed talked about a few minutes ago. That is, 16 how studies and data fit together as a package as 17 to how much mileage we can get from getting 18 indications to support one another. 19 Now, what Ed proposed and discussed may 20 not have sounded that radical but, in fact, what we 21 are thinking about doing is going beyond the old 22 model where hospital-acquired pneumonia, for 70 1 instance, might support community-acquired 2 pneumonia. But moving outside the respiratory 3 tract, you aren't going to get much support for a 4 respiratory indication. So, the idea that for 5 serious illnesses looking at several serious 6 individuals across the body, so long as you have 7 adequate data, for instance, on tissue penetration, 8 etc., and some decent microbiology, might form a 9 package that would allow you to have a smaller 10 number of overall studies than would otherwise be 11 the case. It is something we have talked about 12 once with our advisory committee. We think it is a 13 fruitful area for moving forward. We have had 14 discussions with at least one firm about using this 15 type of approach, which we also think may serve as 16 a tool to also look at studies of resistance 17 indications as well. 18 [Slide] 19 What are some of the consequences of the 20 preceding? I mean, serious illness should be able 21 to lead to expedited evaluation. Expedited 22 development should be able to lead to reduced 71 1 costs. One of the issues that you have to always 2 worry about is does expedited development equal 3 less certainty re benefit and risk. George sort of 4 commented on this as well. 5 You know, those are always issues that, as 6 a regulatory body, we have to deal with but in 7 general everyone has to deal in this area with 8 uncertainty. The FDA deals with uncertainty. The 9 clinicians who will be using the drugs will have to 10 deal with uncertainty. Industry has to deal with 11 uncertainty. People don't always do well with 12 uncertainty, and without wanting to reopen the 13 delta issue at all, that is a good example with 14 uncertainty. That is, some staff within FDA were 15 concerned about widening the delta, increasing the 16 level of certainty as to whether the drug really 17 works. Companies then perceived that this idea of 18 narrowing the delta would produce greater 19 uncertainty as to whether they could get their 20 products approved. 21 So, no one really ultimately came out 22 ahead with that effort of looking about 72 1 uncertainty, but we have to keep in mind when we 2 talk about drugs for infectious diseases that we 3 know, first of all, some of these diseases are 4 serious so you would like to have some idea of how 5 well the drug works. If you have a new drug for 6 serious illness, you are willing to tolerate a 7 certain amount of uncertainty about how well it 8 really works, particularly if the alternatives are 9 few and far in between and you will tolerate some 10 degree of uncertainty about safety if, in fact, the 11 drug seems to be producing some decent activity. 12 Certainly, HIV and oncology are areas where that 13 has been the case. 14 As you think about drugs more broadly for 15 less serious illnesses, naturally the willingness 16 to tolerate uncertainty becomes a little less. But 17 this is an issue that we always have to keep at 18 least in the back of our minds as we move forward, 19 that as we seek to expedite a program we do have to 20 deal with some of the ramifications of expedited 21 development. 22 [Slide] 73 1 Well, what are some of the scientific 2 issues that exist? Certainly there has been a lot 3 of interest in the use of surrogate markers, and 4 certainly infectious diseases is one of the big 5 areas where surrogates have been used successfully. 6 Certainly everyone is familiar with how useful they 7 have turned out to be for HIV drugs. We have 8 certainly got enough data to know that surrogate 9 markers can be quite useful in getting a handle on 10 a drug for tuberculosis. We have some good data 11 now from studies that were done a few years ago 12 that allow us to look at two-month sputum 13 conversion rates, and more particularly, at early 14 relapses to predict how well a drug will look down 15 the road and allow approval actions to be taken 16 probably years earlier. 17 You can get into somewhat messy situations 18 with surrogates. An example that I was very 19 involved in a number of years ago was the use of 20 clarithromycin for the treatment of M. avium 21 bacteremia. I think we were all comfortable that 22 reduction in bacteremia was a good thing and 74 1 subsequent long-term trials that were conducted, 2 using a second drug to prolong bacteremia, I think 3 did show some real benefit to patients. On the 4 other hand, we also noted that even though higher 5 doses of clarithromycin led to somewhat greater 6 suppression of bacteremia, survival was actually 7 worse. 8 Outside the infectious disease area, we 9 know from some of the studies in cardiac disease 10 that some of the surrogates, such as suppressing 11 ventricular premature depolarizations, don't always 12 lead to the kind of result that you are looking 13 for. So, surrogate markers are extremely 14 promising. They do require though some attention 15 to detail. You need to have the right trials, the 16 right data to feel comfortable. 17 One other thing that may not be a big 18 issue with short-term studies of antibacterial 19 drugs but can be a bigger issue with longer-term 20 study of disease, and certainly in infectious 21 disease it is, let's say you measure a surrogate 22 early on in the course of the disease and it looks 75 1 like it is responding very well. There is an 2 assumption there that what you do in terms of 3 managing the patient subsequent to measuring that 4 surrogate is the appropriate thing. Remember, the 5 surrogate measures what happened before it. If, 6 for instance to use tuberculosis, you measure what 7 happens to the sputum at two months and it looks 8 very good, but then part of your new experimental 9 regimen is a radical change for the last several 10 months of treatment in your regimen and that is not 11 such a good follow-up regimen, your overall result 12 may not be what you expect. You always need to 13 keep that in mind. Nonetheless, we think that this 14 is a very fruitful area for moving forward with. 15 What are other things? We are going to 16 talk a lot about the use of preclinical and early 17 clinical trial data, PK/PD data in combination with 18 clinical trials to dose select, to more efficiently 19 move forward in development and, hopefully, to end 20 up with smaller clinical trials. I think that is 21 again a very fruitful area. 22 We always have to keep in mind that a lot 76 1 of the PK/PD we often get in terms of data is, of 2 course, from the blood since the blood is most 3 accessible but, in fact, sometimes the infections 4 are in the tissue and we have to be comfortable 5 about how we make those extrapolations. 6 An area that I think may turn out to be, 7 for new, innovative drugs, fruitful is the role of 8 infections due to susceptible organisms in the 9 study of drugs for resistance claims. What I mean 10 by that is if you have a brand-new molecular entity 11 that, say, for enterococci has the same activity 12 against vancomycin-susceptible, 13 vancomycin-resistant enterococci, what role can the 14 susceptible organisms in diseases cause by the 15 susceptible organisms play in the overall 16 evaluations since, frankly, things like that would 17 simplify the overall approach and have a greater 18 number of patients. I think that is an area we 19 need to do some thinking about as well. 20 An area I am not sure what to say about 21 and I kind of put this up so that, hopefully, we 22 will hear from industry is issues about discovery, 77 1 which have already been covered some by George. 2 You know, I hate to phrase it like this but suppose 3 we were able to wave a magic wand and a lot of the 4 problems with economic return and other things that 5 are of concern to industry disappeared magically, 6 and they felt that this was a fruitful area to put 7 resources in, where are we with the level of 8 science that would allow us, in fact, to think that 9 the discovery programs would start to yield 10 fruitful entities that could be scaled up and be 11 effective in treating people? The question is how 12 comfortable are we with our science at this point, 13 and I leave that to folks from industry who are in 14 a much better position to address it. 15 [Slide] 16 This is a graphic following up on what Ed 17 brought up about the Critical Path. It is just to 18 point out one of the goals that FDA would like to 19 do, basically assisting in looking at problems in 20 safety, efficacy and even manufacture that may be 21 of concern to industry and, hopefully, over time a 22 partnering with industry in areas that we can be of 78 1 assistance in for looking at some of the road 2 blocks, some of the problems. As is known, we do 3 have access to more data than any one company has. 4 There are always questions about how much of that 5 data can be made public but certainly certain 6 analyses can be done that may be of help to 7 industry. 8 An area, for instance, that is relevant to 9 what we are talking about here today is how much 10 data we could get or have additional companies 11 supply from older applications to allow us to do 12 even more work than what we have already started to 13 do in the validation of surrogates in certain 14 infections. That is just one example. 15 The Critical Path initiative is something 16 that is just under way. The hope is that over time 17 it will yield useful information that will assist, 18 for instance, in the safety area. One of the big 19 problems is that you get through your development 20 program and you are in your Phase 3 studies or 21 sometimes, actually, even after approval when 22 unexpected safety problems come up and that can 79 1 have a very negative impact either on drug approval 2 or what happens to the drug in the postmarketing 3 period. 4 One of the questions always is are there 5 better tools that might exist that would allow one 6 to reduce the likelihood of things like that 7 happening, short of having much, much larger 8 trials. So, that is one example of the kind that 9 might out of the Critical Path initiative and, 10 hopefully, in moving forward that would be of help 11 in this particular discipline as well. 12 [Slide] 13 As far as some of the economic incentives, 14 there is orphan exclusivity that exists, seven 15 years of marketing exclusivity. I believe 16 Waxman-Hatch exclusivity, which provides 17 substantial add-on to drugs to make up for the 18 development time that was lost or the patent life 19 that was lost during development, is now available 20 for new entities that were not the subject I think 21 of applications before late 1997. So, that is 22 something for brand-new entities that I think may 80 1 be of help. I think many of you are familiar with 2 the pediatric exclusivity as well. 3 One of the reasons that, of course, and 4 this is not any great shock to industry, as to why 5 economic incentives are considered desirable is 6 that it has become very expensive to develop new 7 drugs. Now, there are a lot of numbers floating 8 around and they seem to have gone up rather 9 dramatically over the last few years but, suffice 10 it to say, whether it is 1.1 billion or 1.7 billion 11 or even a little less than a billion, it costs a 12 lot of money to get a product through and you, 13 therefore, needs to have a substantial return and 14 that is one of the concerns that you are hearing 15 from industry as to perhaps why anti-infective 16 drugs are not that desirable--it costs this much 17 money; what are your chances of getting it back? 18 Remember, without getting into the 19 concepts of net present value, you are paying the 20 dollars up front. The money you are getting back 21 is down the road so those dollars are worthless. 22 That is also an issue as well in terms of thinking 81 1 about the streams of income, income and outgoing. 2 [Slide] 3 What are the other economic incentives? 4 George touched on this, you know, expanding 5 eligibility for orphan designation; the Bioshield 6 like purchase arrangements; issues of wild card 7 exclusivity, that is, you develop a new 8 antibacterial and you would like to add some months 9 on to any product you like, or other exclusivities 10 or enhanced patent protection. 11 The issue is that all of these require 12 legislation. I don't think I have to remind people 13 that the budget situation is tight and things that 14 are going to add costs and, for instance, wild card 15 exclusivity adds can add a lot of healthcare costs. 16 I am not sure how attractive these things are at 17 the level of Congress, but one of the nice things 18 about having the IDSA involved is that the IDSA, as 19 a scientific medical organization, is the kind of 20 organization that can bring ideas forth, whereas, 21 you know, within the government you are not 22 supposed to be lobbying Congress. But the IDSA is 82 1 free to talk about these issues and see what kind 2 of feedback they get back from them. 3 [Slide] 4 What are the downsides? Again, George 5 covered this pretty well. Fundamentally, most of 6 the that is short course and, in spite of the 7 issues about resistance, the fact is that most of 8 the therapy works pretty well. The need is 9 greatest for resistant and related infections but, 10 of course, the market is most attractive for 11 infections in the primary care setting. So, we 12 have that sort of tension as well. What companies 13 would like are drugs that large numbers of people 14 would take. On the other hand, that has its own 15 set of problems in terms of perhaps some excess use 16 and what that would do to the usefulness of the 17 drug. The need is greatest for resistant and 18 related infections. In some respects, certain IV 19 drugs that might be more hospital based would be 20 highly desirable but not very attractive, at least 21 to larger companies. 22 To think about this, it is useful to 83 1 remember that it is not all of anti-infective drug 2 development that has been adversely affected. We 3 only need to think about what is going on with HIV 4 drug development where the therapy is not short 5 course. It is not as highly effective and there is 6 enormous activity going on. So, I think it is not 7 so much that it is anti-infective therapy per se; 8 it is the particular characteristics of the type of 9 anti-infective therapy that we are talking about. 10 [Slide] 11 An unresolved issue that we bring up every 12 time that we have no good solution, but it goes 13 back to the uncertainty issue of which we spoke a 14 few minutes ago, there is a basic tension between 15 encouraging antimicrobial development and 16 preserving the usefulness of current and new drugs. 17 In essence, what you are saying is we want you to 18 spend X amount of money to develop a new drug, 19 knowing that people will be going out there and 20 telling the practitioners not to use it too much. 21 You know, if you strip it of everything else, it 22 gets down to that. It is hard to know how to work 84 1 through that problem to get a level of comfort 2 sufficient to make the necessary investments. 3 Absent sufficient balance between these activities, 4 on one hand, adequate investment may not occur or 5 the benefits of such investment may be short-lived. 6 [Slide] 7 We obviously can't develop a drug, and 8 that is not just FDA but IDSA as well so, 9 obviously, industry needs to play a great role. I 10 think it is worth mentioning again that new types 11 of exclusivity and patent protection would require 12 new legislation. We do know that if we can 13 expedite development we can lower costs and I think 14 that is one of the goals that we are all talking 15 about. It may not look like a direct cost but it 16 has the same value, shortening overall development 17 will cost less. We always worry internally that 18 promotional claims derive from statements in 19 labeling. Statements in laboratory come from how 20 much data you have about the product and we would 21 probably have less here. Truthfully, if we could 22 work through all the other problems I don't really 85 1 view that as something that is insurmountable. All 2 right, thanks a lot. 3 DR. EDWARDS: Thank you very much. It was 4 a very beautiful discussion. Comments? We have 5 now a period of time for discussion. Dr. Rex? 6 Discussion 7 DR. REX: Those were great talks. I want 8 to pick up a theme that has been left sort of to 9 the side of both of them but one that is very 10 important in that it is part of my daily life. 11 What has gone on so far with the antimicrobial task 12 force is absolutely laudable in that it has started 13 the conversation; it is asking very hard questions 14 about real issues, things like delta and regulatory 15 uncertainty. 16 But I was particularly pleased to see the 17 discussion of the question of infrastructure, the 18 need to build the infrastructure and the need to 19 maintain the infrastructure. I can make some 20 contrasts between "small" and "large" pharma here. 21 I want to make them mainly to say that they are not 22 really the key driver. "Small" pharma excels in 86 1 certain areas; perhaps does a little better with an 2 established molecule or an early possible molecule. 3 "Big" pharma has the bigger tools; may have the 4 bigger library; can do the high throughput 5 screening. In my own particular case at AZ we have 6 about 200 discovery scientists that are working on 7 antibacterials and antimicrobials. 8 But what you really have to do, you have 9 to roll the dice enough times. You have to be 10 smart and pick your targets but you also have to be 11 lucky and that takes a lot of dice rolls, which 12 takes a bunch of money over a bunch of time. 13 Ultimately, we are all working in a cost 14 constrained environment, even small and large. 15 Somebody has to go to the bank. My banks is the 16 rest of the company. The bank for a small firm is 17 the venture capital. 18 So, what does it take to protect this 19 infrastructure? You have this very valuable group 20 of people who are working away and you want them to 21 keep going. Well, the point that I want to make, 22 and this is sort of an unpopular theme these days 87 1 but I can state it simply by saying without profit 2 on drug number 1 there will be no drug number 2 3 because there will be no company to carry forward 4 with drug number 2. 5 This then speaks to the larger issue of 6 what is going on with the way that pharmaceutical 7 industry is being pressed with issues about 8 reimbursement--payor strategies, cross-port 9 importation. These issues are bigger than what is 10 going on in this room today. They are bigger than 11 the NIAID. Some of them are at the NIH level; some 12 of them are the multinational level and you are not 13 going to fix them today. 14 So, I do not want to dissuade this group, 15 the IDSA. What you are doing is incredibly 16 valuable. Only you can make this contribution to 17 influencing the conversation, but the bit that I 18 would like to add as one more comment on your 19 slides is that there is this other thing that has 20 to go on at the same time. It is sort of like the 21 comment about acting locally but thinking globally. 22 You have to act at your sphere of influence. By 88 1 improving the quality of the guidelines you and the 2 FDA can help us reduce our uncertainty internally. 3 And, that is very valuable when you are working 4 internally to get one program funded versus another 5 program. 6 But there is this other bit that really 7 does need to be considered in the same breath. I 8 do not have an answer for it but if you don't start 9 asking questions about it, it is kind of like the 10 ten-year lag on drug development, you won't have an 11 answer when you need one. So, that is my comment 12 about the additional piece that is here that needs 13 to be addressed from our perspective. 14 DR. DRUSANO: Thanks. I would like to 15 actually address this comment to both Dr. Scheld 16 and Talbot. In both presentations there is a 17 dramatic slide looking at the number of approvals 18 of anti-infectives over time. As we saw, there was 19 an inexorable down slope to that number. So, the 20 real issue becomes two things. Number one, how 21 much of a crisis are we in, in terms of new drugs 22 coming along? Number two, and a little bit more of 89 1 a subtle point, is how long is that trough going to 2 last? Because, as John pointed out, you do need a 3 certain infrastructure. When companies get out you 4 don't jump back in and then go straight to full 5 production overnight. 6 One of the things that I think would be 7 helpful to make the case, particularly to Congress, 8 is to examine two things. Like we do with 9 controlled clinical trials, it is nice to see not 10 just anti-infectives in the number of approvals per 11 year, but to have some control therapeutic 12 areas--what is happening in pulmonary? What is 13 happening in oncologics? How many approvals are 14 there per year in those to make a point to see if, 15 indeed, this is as bad as we think it is. 16 As to the duration of the trial, how many 17 anti-infective INDs are being filed because that is 18 the other side of the coin? You don't know how low 19 that trough is going to be and how long it is going 20 to be down there unless you know what the activity 21 currently is in companies that are filing new 22 anti-infective INDs. I think we know the answer to 90 1 that but it would be nice to have that data to make 2 the point to regulators, to Congress. It will make 3 I think the situation much clearer. 4 DR. EDWARDS: Thanks, George. 5 DR. POWERS: George, ask and you shall 6 receive. We will now show you the slides that 7 actually answer your question. This is twice in a 8 row and I am not a shill, I promise. We didn't pay 9 George off for this one! 10 This keeps getting asked and, George, I 11 was thinking when you showed your slide of how many 12 people from the press have called, Jason Brosky, 13 from our press office, calls every day and asks 14 this question. So, we tried to put some slides 15 together and one of the reasons I want to show this 16 is to get this information out into the public 17 domain. 18 [Slide] 19 Mike showed 1996 onward, but I thought it 20 would be very instructive to show back to 1980 what 21 has actually happened with these drugs, and you can 22 see that it is an up and down process, as you would 91 1 expect. We go from some highs and we got some lows 2 that actually even preceded. 3 The other thing that is important to 4 notice though is that in the years in which there 5 are no drug approvals, look what happens 6 afterwards. So, we see this blip. So, some of 7 this is related just to the vagaries of the 8 calendar where there was a drug that was in 9 development that just happened to spill over into 10 the next year. 11 [Slide] 12 If you do a trend analysis of this, it is 13 true that the overall trend is decreasing. It goes 14 from an average of 2.7 drugs out here, in the 15 1980s--I did it in a ten-year span. In the 1990s 16 it is 2.5 and there is really not enough data yet 17 in the 2000s to actually say what is going on, but 18 the average there was 1.25 from 2000 to 2004. 19 Ed Cox and I were discussing yesterday 20 when you talk about safety analysis and you say, 21 well, this drug had a 1.5 percent adverse event 22 rate and the other one was 0.2, but what you really 92 1 need to look at is the details of what is in there. 2 So, let's look at the details of what are these 3 drugs because I think the idea is too what are we 4 asking for. 5 [Slide] 6 George, here is the question you were also 7 asking about all drug approvals, what is happening 8 overall. As you can see, there is this up and 9 down. We had a blip in 1996 and then it is coming 10 back down overall. The dark line here is for small 11 molecule drugs, which is what we deal with in the 12 anti-infective divisions. The bottom line here is 13 for biological products. So, that is actually a 14 little bit on the up-slope whereas what we are 15 seeing is the small molecules come down. So, that 16 is the answer to your question of what is going on 17 overall in terms of drug development. This slope 18 has been going down since 1996 for everything so 19 that mirrors the trend of what is going on. Then 20 let's look at the mean clinical and approval phase 21 links. The yellow one here is how long it takes 22 your drug in the clinical time and the blue part is 93 1 how long it takes to get approved. This was done 2 by the Tufts Center for the Study of Drug 3 Development so this was independent of the FDA that 4 somebody looked at this. 5 When you look at this over time you can 6 see that the clinical phase has lengthened, yet the 7 time to approval phase where the FDA looks at the 8 drugs has actually shrunk over time. So, on the 9 whole it is a little bit longer than it was, but 10 part of this is due to the increase in the clinical 11 phase program. 12 [Slide] 13 Then this asks the other question of how 14 long does it take to get your antimicrobial 15 approved. This looks at all of the drug 16 classes--anesthesia drugs, cardiovascular, 17 anti-infectives, CNS and anti-neoplastics. Here 18 are the anti-infective drugs. They are at the 19 bottom of this rung almost the entire time from 20 1982 to 2001. So, when you look at this the mean 21 clinical and approval phase lengths are shorter, 22 which would argue that it is actually less 94 1 expensive in terms of time for anti-infective 2 drugs. 3 [Slide] 4 If you go to the next slide, it is the 5 median. So, this way, looking at the median we are 6 saying, well, are outliers driving this system? Is 7 this shorter just because HIV approvals take a 8 shorter time? The answer to that doesn't look like 9 that is true either. If you look, anti-infectives 10 are again at the bottom of this. The median 11 clinical and approval phase lengths for 12 anti-infectives are also shorter than the other 13 drug classes as well. 14 [Slide] 15 So, if you look at this, back in the 1980s 16 to 2001 the mean was 18.6 months and the median was 17 14. Over here, in 1998 to 2001 it is now 8.8 in 18 six months mean and median for anti-infectives. 19 Some of that is really short for HIV, as you can 20 see, 4.6 and 4.1. So, some of this is driven by 21 the HIV model. 22 [Slide] 95 1 But here is the real important part to 2 look at, what was getting approved back in the 3 1980s and the 1990s. This gets to what I think 4 Mark was talking about, about quality versus 5 quantity. What do we want to see? If you look at 6 the left side of this graph, it is pretty much 7 yellow so what was in there was predominantly 8 beta-lactam drugs and we looked at every one of 9 these. You have some real big sellers here like 10 bacantacillin, Ceftin, cefuroxime and cefpirome. 11 No insult intended to people who made those drugs. 12 But when you look at these numbers, they are driven 13 by a lot of drugs that really didn't get a lot of 14 clinical usage. 15 If you look to the right here, what we see 16 in the 1990s is the blue that starts to pop here 17 with all the qinolones in the 1990s. But then when 18 you look over here there is much more variety of 19 colors because the kinds of drugs we are trying to 20 see is a greater variety than what we were seeing 21 in the past. 22 [Slide] 96 1 This gets to the issue of new drug 2 classes. Mike mentioned that there have been two 3 new drug classes approved since 2000. It occurred 4 to us, well, when were the last new drugs approved? 5 So, without just looking at the overall picture you 6 may get a skewed view. 7 If you look at this, the first antibiotic 8 approved are the sulfonamides in the 1930s, and I 9 shouldn't say approved, just clinical usage because 10 the efficacy standards for the FDA were not 11 introduced until 1962 with the Kefauver-Harris 12 amendments, and you can see the majority of drug 13 classes were introduced prior to that point in 14 time. 15 Now, what we used to define drug class 16 here is what the IDSA used in their "white paper," 17 and that means novel binding site for the 18 antimicrobial. That is why we lumped macrolides, 19 lincosamides and streptogamins together and 20 telithromycin got lumped in with the macrolides as 21 well. 22 So, what you see here is that most of the 97 1 drug discovery occurred in the 1930s, '40s and '50s 2 and the last truly novel class was trimethoprim in 3 1968. So, there hasn't been any new drug discovery 4 for 40 years and those two new drug classes from 5 2000, if you look at it from that point of view, is 6 actually a pretty good thing. 7 That is enough; I will stop there but I 8 just wanted to show sort of the idea and get back 9 to the question of what are we asking for. I think 10 when IDSA folks met with Commissioner McClellan one 11 of the points he made was that we need to be very 12 focused about what we are asking for. So, the idea 13 is do we want a lot more, you know, oral 14 cephalosporin drugs to add to the mix to make the 15 numbers go up, or are we looking for more quality 16 drugs that are going to relate to antimicrobial 17 resistance? 18 DR. EDWARDS: Would either of you two like 19 to comment on those comments? If you don't, I 20 will. 21 DR. SCHELD: Well, I will say what I think 22 we are asking for--I am not particularly interested 98 1 either professionally, personally or as a 2 representative of IDSA in a number of new oral 3 cephalosporins that have overlapping spectra with 4 what is already on the market. What I think we, as 5 physicians, are most concerned about, as George 6 rightly put it, to put the patient first, is the 7 escalation in gram-negative resistance in 8 Acinetobacter of untreatable infections in our 9 soldiers coming back from Iraq with Acinetobacter 10 pneumonia and things like that. How to go about 11 putting that case out there with the legislative 12 solutions is not going to be easy. We also 13 recognize, just as everybody said here this 14 morning, if we had a new drug which would be active 15 against Acinetobacter that is only susceptible to 16 Clistin tomorrow, the ID people in every hospital 17 around this country are going to try and reserve it 18 in some capacity. So, you have this Catch-22 that 19 we have talked about before. 20 As George said, the solutions are not 21 simple. It is going to be multifactorial but I 22 think we, as an organization, should put that type 99 1 of a global out there for a legislative solution. 2 That is where I am coming from. 3 DR. EDWARDS: We have had about 50 years 4 of experience with anti-infectives now as a 5 species. I think we are still in the beginning 6 stages of learning about them. If we imagine that 7 in the last few years--well, in some ways, John, 8 you have shown some data that shows a little bit of 9 a consistency in terms of the types of agents, new 10 agents that have been coming out. We are saying 11 there is a critical diminishment in the larger 12 pharmaceutical company research and development 13 programs at the present time. I think we have as 14 good data as we can possibly get for that. Then, 15 we also have the curve that we keep referring to of 16 the approval of new entities. 17 How that compares to other classes of 18 drugs I am not sure is all that relevant, to tell 19 you the truth, because we are faced with an 20 existent problem now. We have learned that 21 antibiotics, to quote without naming specifically 22 some of the people we have talked to, define their 100 1 own life span, and they are unique drugs in 2 comparison to other drugs for two reasons. One is 3 that they do define their own life span. I mean, 4 that is not totally unique but it is a very 5 prominent characteristic of the antibiotics. 6 Secondly, most of us can't think of very many other 7 classes of drugs where the thought leaders 8 immediately start discouraging their use once they 9 are developed. That is certainly not the case with 10 oncology drugs. A new oncology drug is just 11 immediately embraced by the community. So, we have 12 a unique problem with this class of drugs from 13 those two perspectives. 14 Yes, George, we feel we are in a crisis. 15 I just wanted to make two small points about that 16 because data is being collected, and all, and I 17 don't want to emphasize that, but those of us who 18 are seeing patients on a regular basis are spending 19 the majority of our time trying to figure out 20 strategies for how to deal with them within the 21 confines of the resistance issues. Mark has made a 22 very important point, which is that for most of the 101 1 organisms we are dealing with we do have available 2 drugs that work. When we start talking about VRE 3 and MRSA, then we have a real strong exception but, 4 even though that fact exists, one has to look at 5 what physician behavior is. 6 I would like to use this example, in a 7 large metropolitan hospital where many patients are 8 seen per day with a localized cellulitis, prior to 9 the last few years those patients were started on a 10 cephalosporin and were dismissed from the hospital 11 emergency room, and the cost of providing care for 12 that individual was relatively small. Now the 13 concern is that they may have something like an 14 MRSA, and it is so high that the specimen is 15 cultured; arrangements are made for a follow-up 16 visit; the patient is started on a regimen of 17 antibiotics that is more expensive and more toxic 18 than a cephalosporin would be, such as rifampin and 19 trimethoprim sulfa. 20 So, if you were really to do the statistic 21 and say how many of those people did have MRSA 22 simple cellulitis requiring a specific antibiotic, 102 1 the answer would probably be very small but, 2 nevertheless, a tremendous expenditure is made 3 towards taking care of that patient because of the 4 spectra of the problem. So, that is a whole 5 separate issue that is related to the fact that 6 while there is a certain armamentarium, there is a 7 shift away to a different strategy in order to 8 protect the individual patient. I hope I have made 9 that point clear. 10 One question that George asked that I also 11 was wondering about is do we have access to IND 12 data? I know that is confidential data, but do we 13 have access to numbers? DR. POWERS: I can tell 14 you that as a part of the Critical Path initiative 15 there was a review done of INDs. I think this is 16 in that document and this is not just for 17 anti-infectives, which I don't think we would be 18 able to release for proprietary reasons, but the 19 overall number of INDs in all the drug classes 20 being submitted to the agency has gone down. I 21 don't remember the exact numbers. But that was one 22 of the reasons why that Critical Path was 103 1 undertaken. 2 DR. EDWARDS: George? 3 DR. TALBOT: Just a couple of comments, 4 that graph in the Critical Path paper was really 5 quite dramatic I think. I don't know the numbers 6 behind it, but for all classes the number of 7 submissions has gone down which has two points. 8 One is it is a great thing that the FDA is looking 9 at this problem across all classes but, as Jack 10 said, we still have a unique problem with 11 antibacterials. 12 I think the other question that you had, 13 George, is whether it is a crisis or not. 14 DR. DRUSANO: I believe it is a crisis, 15 George, no question. I just wanted to clarify a 16 little bit. 17 DR. TALBOT: Well, Jack expressed that 18 concern also. I think I would point out that one 19 person's crisis is another person's problem. The 20 word crisis is very easy to use. Public health 21 crisis is a very easy phrase to use. I would just 22 urge that we be precise in our use of that because 104 1 trumpeting public health crisis from now in 2 perpetuity may eventually work against us. I am 3 not sure that there is a consensus on that, as to 4 whether it is a current crisis. It is certainly a 5 crisis for the person who has some of these 6 acinetobacters, but whether it is a public health 7 crisis or an impending public health crisis is 8 something I think we really ought to reflect on in 9 terms of our communication about it. That is my 10 personal perspective but I don't know that we serve 11 ourselves well by falling into rhetoric. 12 DR. EDWARDS: I just can't help but make 13 one comment about that, George. You know, I was 14 discussing this whole issue with my wife not too 15 long ago and she provided me with a concept that I 16 really like, and that was that I think what we are 17 working on here is to try to stop a potential 18 catastrophic situation before it occurs. There is 19 a lot of discussion about all those sorts of 20 strategies going on currently as they relate to 21 some congressional hearings. 22 DR. TALBOT: Yes, and I agree with that 105 1 completely. As I said in my presentation, if 2 vanco. gets out broadly into staph. and vanco. 3 resistance gets out broadly in the community, I 4 mean that is going to be a catastrophe. But, 5 again, just to urge that we be precise and 6 consistent in our terminology so that, as 7 scientists, we clearly state the problem and state 8 its potential repercussions, one of which could be 9 exactly that. And, to George's point, it is 10 exactly the reason why we need to keep the 11 companies that are in, in since we know from our 12 visits that once you are out, to get back in is 13 going to take six, seven, eight years, we have been 14 told. So, we have to keep them in if they are in. 15 DR. DRUSANO: Just to make a point, there 16 are two drivers for this whole problem. Number one 17 is resistance but the second driver is economics. 18 Now, we have talked a little bit about the 19 economics from the point of view of getting some 20 congressional changes so that reimbursements for 21 companies can incentivize them to develop 22 anti-infectives, and that is wonderful. 106 1 There is a whole other part of the 2 economics mix that has not been talked about, and 3 Dr. Powers touched on it a little bit, which is to 4 say that when you look, development times are the 5 shortest for anti-infectives for any therapeutic 6 area. That is number one. So, there has to be, I 7 think, an economics analysis by somebody 8 independent for the companies. 9 The other issue is what is the Phase 3 10 failure rate? If you look at one of Dr. 11 Goldberger's slides--I don't care whether you use 12 1.3 or 1.7 billion dollars, that is a lot of money. 13 What is the overall Phase 3 failure rate by 14 therapeutic area? I don't know the numbers but I 15 would be willing to bet you, because of the say 16 that we collect preclinical data on 17 anti-infectives, that the Phase 3 failure rate is 18 much smaller. So, you have to amortize the cost of 19 a Phase 3 program failure across all of the income 20 that is coming in. So, I think when you look at 21 that, when you look at development times, when you 22 look at margin in terms of profit, that these are 107 1 all things that are economically based that need to 2 be analyzed so that the industry itself sees that 3 this is an economically viable area. 4 DR. EDWARDS: I am going to take the 5 prerogative to break at the moment unless it is a 6 very short comment. 7 DR. HENKEL: It is fairly short. There is 8 one more link between economics and resistance. A 9 couple of times this morning, including in Mark's 10 talk, there has been an assumption that use of a 11 new product is necessarily linked to resistance to 12 that product, and I am not sure that is necessarily 13 true. I would like to challenge that in that we 14 should look at it in a little more sophisticated 15 fashion in terms of the activity of a new drug, the 16 concentrations at the site of infection, how well 17 it eradicates the pathogen, and we may actually 18 discover that some new drugs more effectively 19 prevent development of resistance if they are used 20 in an appropriate way as opposed to an older agent 21 which may be marginal in some of those respects. 22 DR. EDWARDS: It is an excellent point. 108 1 Clearly, enough has been spoken already this 2 morning to stimulate a tremendous amount of 3 discussion, which is the purpose and we hope that 4 it will go on, and on, and on. We are going to 5 take a 15-minute break. We are already behind 6 schedule but if we could come back just a little 7 bit after 11:15. 8 I want to thank the speakers from the 9 beginning of this session for these absolutely 10 beautiful presentations. Thank you very much 11 again. 12 [Brief recess] 13 DR. EDWARDS: I am going to take the 14 prerogative to reopen the discussion before we move 15 on to the next part and see how it goes here. If 16 the quality of the discussion becomes meritorious 17 enough, we may just take the meeting a little bit 18 further; we will play it by ear. I really felt 19 that we had to terminate that discussion quite 20 prematurely, and by the intensity of the discussion 21 that has gone on in the break, it is really 22 amazingly gratifying to see everyone talking so 109 1 much. I think we are going to go ahead and reopen 2 the discussion for this first part now for a few 3 moments. Would anyone like to introduce a new 4 concept and continue on at this point? John? 5 DR. BRADLEY: The discussion took several 6 turns and I wanted to address some of John Powers' 7 questions on trying to focus where we are going 8 with our proposals with respect to what the FDA can 9 do and what industry can do, and perhaps an example 10 will help this. 11 In 1998, at the Anti-Infective Advisory 12 Committee, information on community-acquired, 13 multiresistant pneumococcus was presented, and for 14 pediatric meningitis we were very concerned with 15 alarmingly increasing rates of resistance to 16 penicillin and cephalosporins, reports of 17 vancomycin resistance, and we pleaded for the need 18 for testing for quinolones in pediatrics. There 19 were also presentations on otitis media that wasn't 20 responding to any currently available oral agents. 21 And, the green light was given. Industry stepped 22 up to the plate and many of these studies actually 110 1 have gone on, taken place, and results, some of 2 which will be presented to the committee next 3 month, are fruitful indicators that this discussion 4 produced a product. 5 But we didn't know that resistant 6 pneumococcus, at least that causing meningitis, 7 would turn out not to be a problem because of the 8 introduction of the pneumococcal conjugate vaccine. 9 Ceftriaxone resistance rates are dropping and the 10 vancomycin resistance that was seen in Memphis 11 turns out to not have been applicable to many other 12 centers. We certainly never saw it in San Diego. 13 So, the imperative to study drugs in meningitis, 14 which would seem clear in 1998, here, six years 15 later isn't, and for all of the investment that 16 industry puts into these protocols, and it is 17 substantial, they are left with a drug which has no 18 economic benefit to them. That goes back to Dr. 19 Rex' comments that if you don't get successful drug 20 1 you won't go to drug 2. 21 So, in part the uncertainty that we are 22 all facing is how bad will the resistance be and 111 1 when will it happen? I think everyone agrees it is 2 going to happen but how can we help legislate some 3 sort of incentive that will keep industry looking 4 for new, novel agents that will address the issue 5 of resistance and allow industry to have an 6 economic incentive so at the end of the day they 7 can go to their shareholders and say, you know, 8 this was a reasonable business decision we made and 9 there is benefit to patient populations. 10 I think to try to focus our discussions on 11 how we can make it worthwhile to all the 12 stakeholders and take it out of just, you know, you 13 make a drug, you sell it, you make a profit 14 scenario is a unique aspect of this particular 15 workshop. The equation for community-acquired 16 infections for things like otitis media, at least 17 in pediatrics, and the equation for 18 hospital-acquired infections like multi-resistant 19 Pseudomonas and Acinetobacter are two different 20 things because, certainly, for hospital-acquired 21 infections they will never be the huge population 22 so they will never be the economic incentive. By 112 1 working together, hopefully, we can come up with a 2 solution that works for everybody. 3 DR. POWERS: Jack, can I respond to that? 4 DR. EDWARDS: Yes. 5 DR. POWERS: John, I think you are raising 6 an important point there about the level of 7 uncertainty and where that comes in. One is 8 regulatory uncertainty related to how to do the 9 trials, etc. But what you just pointed out was 10 scientific uncertainty, something none of us can do 11 anything about. So, if you were developing a new 12 drug for cancer and somebody came up with a cure 13 for all cancers tomorrow, you are kind of out of 14 luck as the company who was developing that drug 15 for cancer. And, that is not a regulatory issue. 16 What I think we hear a lot of is, well, 17 the agency promised us this four years ago. You 18 know, the science changes as we go on, therefore, 19 the risk/benefit for those things actually changes 20 as well. If you kind of try to project into the 21 future though, this is the idea of focus and where 22 we are seeing the least amount of drug development 113 1 for multi-drug resistant gram-negative rods and 2 hospitalized infections but, as you say, that is 3 where the market is the smallest. So, how do we 4 come up with an incentive there that does not seem 5 to fit the supply and demand model? 6 DR. EDWARDS: Would anyone from industry 7 like to comment on that? What would be the 8 incentive? 9 DR. REX: I will make one comment but I 10 really think that some of the other people around 11 the table should weigh in as well. I want to go 12 back to the theme that I raised a minute ago. You 13 know, George asked before the break does industry 14 understand about the fact that anti-infectives are 15 lower risk to develop, and I promise you we do. We 16 understand that once you prove that the drug is 17 safe in Phase 1, doesn't cause some funny bad side 18 effect, the likelihood of an anti-infective going 19 on to become a marketable compound is much better 20 than it is for almost any other class. I project 21 that message every time I talk about why we should 22 be doing anti-infectives, along with the fact that 114 1 the cost of anti-infective--the spend profile is 2 actually very nice. So, all those things are 3 clearly understood. 4 However, we all live in a resource 5 constrained environment, every one of us, in your 6 personal budget, in the budget of any group that 7 you work with, the laboratory that you run, and you 8 can only spend so many dollars on things that don't 9 generate return. You also have to spend dollars on 10 things that do generate return. John, you talked 11 about the fact that, you know, a cure for cancer is 12 found and now all the cancer drugs are no longer of 13 value. Companies also recognize that, and that is 14 part of your mixture of risks so you know that you 15 are spreading out in terms of variety of things so 16 if any one thing goes down, you are not about that 17 one thing; you are about the whole pattern of 18 stuff. 19 It is ultimately that overall support for 20 reimbursement for drugs as a whole is actually the 21 critical driver. Without some degree of comfort 22 with that everybody is going to retreat to their 115 1 next position of comfort, which is the big things 2 that make the big money because you know that that 3 is at least stable for the next couple of years. 4 So, it is all of that that really makes a 5 difference. 6 What could IDSA do? You guys are working 7 on anti-infectives and that is great but can you 8 broaden that? Are there other academic groups? 9 Does AAP have a broader initiative that has to do 10 with new drugs for X, Y, Z, whatever these other 11 things are? Is there a way to broaden the scope of 12 what is going on here? 13 DR. EDWARDS: Do you want to comment, 14 Mike? 15 DR. SCHELD: Well, John, we certainly can 16 and we have actually discussed this with a number 17 of other societies, but we have also discussed it 18 with, say, organizations like AARP which represents 19 35 million Americans. I wouldn't say that that has 20 gone very far but it is on our radar screen. 21 DR. EDWARDS: Excellent point. Yes? 22 DR. EISENSTEIN: I would just like to 116 1 weigh in a little bit on the problem that everybody 2 recognizes, which is drug resistance and, 3 therefore, physician behavior which is appropriate 4 in terms of trying to maintain the best newest 5 products activity, which is essentially perverse 6 from the standpoint of the marketplace. So, if the 7 marketplace drives or should drive use and this 8 works for most situations. In places where the 9 compound should be limited for public health 10 reasons new ways of thinking I think need to occur. 11 The best example that comes to my mind offhand is 12 that the U.S. Congress has mandated, 13 controversially I would say, for subsidies to be 14 given to certain industrial groups, and the farm 15 population comes to mind, where the farmers are 16 actually paid not to plant crops. If there were 17 such an incentive for drug development where a 18 product could be actually subsidized could be 19 developed and, in a sense, to be put on the shelf, 20 there might be very interesting means by which 21 companies could be willing to take the path towards 22 innovation if they saw that as an opportunity. 117 1 Now, that is fraught with major political 2 issues, certainly not to be solved here but, in a 3 way, it is almost the best use of government, which 4 is to protect extraordinarily difficult to preserve 5 desires of society that aren't otherwise solved by 6 the marketplace. 7 DR. EDWARDS: George, let me ask you to 8 comment. 9 DR. TALBOT: That is a very interesting 10 idea. It follows on the theme that Tim mentioned 11 which is the marketplace might. Certainly, the 12 receptivity of the marketplace to new products is a 13 major issue. I think though that my conclusion 14 there was correct. What is going to be required to 15 change that will be scientific data, solid, very 16 robust scientific data, for example, showing that 17 cycling works or something along those lines. I 18 guess the question is what sort of scientific data 19 would be necessary to change that perception and 20 who would generate it. 21 DR. EDWARDS: Yes? 22 DR. TULKENS: Maybe I can make a comment. 118 1 We have a number of examples in countries where 2 consumption is low. One example is Holland where 3 resistance is also extremely low. Now, companies 4 don't make much money on that. So, the question is 5 really to discuss can we identify places where 6 compounds are needed but will be restricted? 7 Therefore, what needs to be changing maybe is the 8 pricing. We need to subsidize or to give the 9 companies a high price for compounds we need and 10 which we know will not be used. I don't know 11 exactly what the economics are in Holland, but I am 12 not sure that companies get a lot of money out of 13 that country but the point is interesting that 14 sales are about one-third of what they are in other 15 countries, and the resistance rates are almost to 16 zero, very, very low. So, that is maybe one way to 17 answer the question. In all situations where 18 restriction has taken place has always led to a 19 slowdown of the resistance and sometimes no 20 resistance at all for many, many years. So, we 21 really have to design the sort of economic model 22 where compounds are developed, left on the shelf 119 1 and all used very sparingly. 2 DR. EDWARDS: George? 3 DR. DRUSANO: Over the break a number of 4 colleagues came up to me and beat me about the head 5 and shoulders and indicated that, yes, indeed, 6 pharmaceutical companies really did understand in 7 extreme detail the economic utility of antibiotics 8 or lack thereof. 9 I think perhaps I was a bit misunderstood 10 in terms of what I said. It is all the economics. 11 We are proposing here oftentimes that we go to a 12 legislative solution so that we ultimately would be 13 approaching Congress to change the law. Now, there 14 are going to be political issues about supporting 15 such a change in law from legislators. If 16 something like that is going to happen, and this is 17 where my comments were meant to go, there has to be 18 a real clear and transparent economic analysis of 19 what the companies currently do get back from 20 anti-infectives, accounting for all of these 21 issues, so that if you are going to put in place an 22 incentive program nobody can say that the companies 120 1 are being over-incentivized. That has to be 2 transparent as to what you are doing in order to 3 get the companies to get back in. That is where I 4 was coming from. 5 DR. EDWARDS: Other comments? That is an 6 excellent point, complicated. Yes, please? 7 DR. BAX: Richard Bax. We have heard a 8 lot for the last 15, 20 years about antimicrobial 9 resistance. We have heard a lot of people talking 10 with great concern about a lot of things which have 11 been discussed many, many times. Could I just ask 12 the panel and the group what really needs to happen 13 now in order to address this issue of where are we 14 going to get new antimicrobials from. We have had 15 lots of proposals but what simple things--and I 16 know the answers are complex, but what simple 17 things really need to happen in order to at least 18 start addressing this problem? 19 DR. EDWARDS: George is cringing over 20 there! Perhaps the question is where would the 21 major focus be. I think George has very 22 beautifully shown that solution of this problem at 121 1 this point and the perspective looks like it is a 2 multifactorial issue. There isn't a single clear 3 course of action that looks like it would solve the 4 situation on the immediate front. George, I am 5 going to ask you--if you don't want to, I will 6 express my opinion but, you know, what should we 7 all as a body put in the highest level of priority, 8 realizing that a multifactorial approach is 9 obviously ultimately going to be the situation. 10 This is really a difficult question because I think 11 it depends on exactly what perspective you are 12 coming from, and I could comment on that. Mike, go 13 ahead. 14 DR. SCHELD: I will take a crack at it, 15 Richard. I think, first of all, when we bring out 16 this "white paper" next month it will crystalize 17 some of our thoughts as an organization. We have 18 kind of listed in there many of the solutions that 19 Dr. Talbot so beautifully presented in somewhat of 20 a priority order. 21 But what needs to happen is some of the 22 things that are already on the table. The NIAID 122 1 needs to implement their "Roadmap." The FDA needs 2 to get their guidances out there. We need to get 3 this issue up on Capitol Hill for hearings to be 4 held on antimicrobial resistance and the drying up 5 of the pipeline for antibacterials this year. So, 6 that would be what I would suggest as the next 7 step. We are not going to have a wild card 8 exclusivity legislation written and passed in 2004. 9 It may never happen. But what we need to do is get 10 the topic out there and have people talking about 11 it, not just in the media but really at the level 12 of congressional staffers. So, that would be my 13 response. 14 DR. EDWARDS: George? 15 DR. TALBOT: Mike, thanks for taking the 16 heat off me. I was going to be more simplistic 17 about it and say I don't think there is one 18 solution. You didn't actually say that in so many 19 words but your answer actually was the same. That 20 is why partnering in this is needed. 21 DR. EDWARDS: I think that is a very 22 excellent response, George. I might make one 123 1 comment that would be indicative of a direction we 2 are heading in, that is, we just had quite a 3 discussion a few days ago about actually starting 4 to put words on paper regarding proposed 5 legislation. I know that makes Bob Guidos pretty 6 nervous but getting to the words on paper appears 7 to the IDSA as an extremely high priority among all 8 the other things which need to be done. 9 DR. BAX: Maybe it is just the rapid 10 effect of development of niche antibiotics with new 11 modes of action which are actually coming to the 12 marketplace, and nothing actually succeeds like 13 success. 14 DR. EDWARDS: Yes? 15 DR. CRAIG: I wanted to ask Ed Cox, is the 16 guideline for development of drugs for resistant 17 organisms one of the ones that is going to be 18 coming out relatively soon? If not, is there some 19 way that there can be some collaboration between 20 IDSA and the FDA to try and speed that up? I am 21 not sure that taking guidelines to the advisory 22 committee is necessarily the best group. I think 124 1 the number of experts on that committee is somewhat 2 diluted by the legal requirements for membership on 3 that committee. Sure, you can add a few experts to 4 it but I think there are probably other venues that 5 can help speed up guidelines and get more input 6 from many other people that are critically involved 7 in clinical trials. 8 DR. COX: With regards to the guidance 9 document on drug development for resistant 10 pathogens, scientifically among the ones listed up 11 there, it is one of the more complex. Certainly, 12 there are some unresolved scientific issues that we 13 are going to need through. So, that one is 14 certainly one that we are very interested in 15 getting out but it may be one of the ones that does 16 take a little bit more time because of the 17 scientific issues that we have to work through 18 there. 19 As part of the process, we do oftentimes 20 take to the advisory committee the guidance 21 documents in draft form and we do get a lot of 22 helpful advice from our advisory committee. Then 125 1 we also post the guidance document publicly for 2 comment. So, you know, I hear your comment and we 3 are working diligently to try and get the guidance 4 documents out there and updated, and that will be 5 one that we are working on but, you know, given the 6 scientific issues there it may take a little bit 7 longer. 8 DR. CRAIG: I think industry wants to know 9 what their costs are going to be, and the guidances 10 are going to determine some of those costs so I 11 think it is an important part of the equation that 12 we need to get answers for. 13 DR. COX: Yes, and I do think with regards 14 to companies that are venturing into development 15 programs, certainly we do welcome the opportunity 16 to have companies come in and we can certainly give 17 advice and guidance with regards to drug 18 development, and specifically drug development for 19 resistant pathogens, in the interim. 20 DR. EDWARDS: Janice and then John. 21 DR. SORETH: Dr. Craig, you remember when 22 you were a chair of advisory committee and we put 126 1 forth drafts of a dozen and a half different 2 guidances. I am reminded, as Dr. Talbot and others 3 have said, there are no easy solutions but there 4 wasn't exactly a swift upturn in the development of 5 antibacterial compounds when those reissued 6 guidances came out, and as important as it is for 7 us to get new drafts out and brand-new entities 8 that we haven't put pen to paper on, such as the 9 resistance guidance, it won't serve as any magic 10 bullet. That said, we will have a number of these 11 out by the end of the year, as Dr. Cox has said, 12 but I think it has to come in conjunction with a 13 number of other things going forward to have a 14 meaningful impact and a mature hope that we will 15 have new antibiotics and new classes on the market. 16 DR. CRAIG: My comments came from the fact 17 that I was the chair and I was not convinced, when 18 we did go over them, that that was the best venue 19 for discussing guidelines. 20 DR. POWERS: I think that is part of the 21 point but that is the way we have to operate. We 22 first discussed the criteria for pathogens of 127 1 public health importance in this venue in November 2 of 2002, then took that to the advisory committee 3 in March of 2003. But what I think people need to 4 understand is that as a regulatory agency we cannot 5 partner with just one group when it comes to trying 6 to develop this kind of things. We can have open 7 public meetings like this one to try to gather the 8 information. We put that out in the Federal 9 Register where everybody can comment. 10 The other thing I sort of wanted to say is 11 that when we looked at the development of those 12 previous guidances--and I have to thank the two 13 people to my right who did a huge amount of work on 14 this, Dr. Albright and Dr. Soreth--they got called 15 "draft" and when I think of these, they are almost 16 always going to be draft. As George pointed out, 17 we will have to update these at some point but we 18 will never make the political mistake, ever again, 19 of calling them "draft" because I think there was a 20 lot of uncertainty that went along with that as if 21 they were going to change tomorrow. 22 But when we look back at when the FDA did 128 1 have a contract with IDSA to do this, it took IDSA 2 over four years to put those guidances together. 3 That is not a knock on IDSA; it is to sort of show 4 the amount of work that has to go into actually 5 developing these. Mike called me up and asked me 6 about some slides we showed about sinusitis, back 7 in October of 2003, and said, hey, who did all 8 review of those placebo-controlled trials for you, 9 and we said we did. We reviewed 17 10 placebo-controlled trials that went back from 1962, 11 and let me tell you how interesting it is to try 12 and find what the primary endpoint is in a trial 13 from 1962-- 14 [Laughter] 15 --so there is a lot of work that goes into 16 it to try to actually do this and I think of it 17 like a research project. But if somebody came to 18 me and said, you know, go find a cure for cancer in 19 a year and a half, I would say I need a little bit 20 more time to be able to do that. So, we have done 21 a lot of research on trying to pull these together 22 so that we have the best information that can go 129 1 into them. 2 DR. EDWARDS: Frank, did you have a 3 comment? 4 DR. TALLY: Yes, I have a comment. When 5 you asked about industry and what the problem is, I 6 think we need to come back to industry. If you 7 have a compound, as John said, that survives Phase 8 1 there are multiple development pathways and, yes, 9 you want guidelines and interactions with FDA so 10 you can get there as quick as you can in the least 11 expensive way. But I think one of the crises that 12 you have identified today is really the discovery 13 of novel new compounds against novel new targets, 14 which is a process that is going on right now. I 15 think the crisis that we are seeing is the amount 16 of dollars in "big" pharma. The research has been 17 going up dramatically but the amount of dollars to 18 discover those new molecules and antibacterials is 19 going down and therein lies the crux of the 20 problem, and it is because of all the economic 21 stuff we have talked about. 22 It is different for "big" pharma and 130 1 "small" pharma because "small" pharma can see a 100 2 or 200 million dollar product is good for them. 3 For "big" pharma that doesn't even hit the screen. 4 So, I think the identification of what is going to 5 get us novel new compounds is the crux of the 6 problem that we have here. 7 I must say, at the advisory committee 8 meeting last March there was a bit of nihilism, 9 that we have discovered all the antibiotics we are 10 going to discover and we are in this hopeless abyss 11 at this point. I really don't think that is true 12 but I think there are new ways, but it is going to 13 take longer to get them because of the definite 14 decrease in the amount of dollars addressed into 15 this area. 16 DR. EDWARDS: Yes, George? 17 DR. TALBOT: I wanted to comment on the 18 timing of the guidances and I have honestly nothing 19 but respect for how long it takes and how few of 20 you there are, even though John seems to have 21 cloned himself, as far as I can tell, in terms of 22 everything he is doing. That is one of the reasons 131 1 why IDSA certainly supports increased funding for 2 all these efforts. 3 But I would like to make a comment or an 4 observation from my experience and ask industry 5 representatives to tell me whether I am off base or 6 not. The thing about a 6- or 12-month delay in 7 guidances is that industry operates on an annual 8 budget process, and every time the budget goes up 9 before top management there is a full discussion of 10 the risks, regulatory risks, scientific risks, and 11 so forth, to pursuing a particular project as 12 opposed to the risks that exist for other projects 13 in other therapeutic areas. So, I guess I would 14 just be concerned that, in the absence of a 15 guidance and knowing, in fact, that maybe it is 16 going to come out in 3 months or 6 months or 12 17 months, without that certainty that that could 18 delay commitments at the level of industry and even 19 potentially result in cutbacks or 20 what-have-you--again, I understand your position 21 but I am not sure that it is correct to say that, 22 you know, a difference of 3 months, 6 months, or 132 1 whatever, really may not make too much difference. 2 I think it could in some situations and I would 3 like to hear what industry would have to say about 4 that. 5 DR. WILLIAMS: Tim? 6 DR. HENKEL: Well, I think you and I have 7 had a couple of conversations about guidances 8 before, and while I find them very useful I have 9 operated under the impression that I have heard 10 John mention before. They are starting points for 11 discussion with the agency. They are not anything 12 written in stone. So, where there are voids, where 13 there is nothing out there is where there is the 14 greatest uncertainty. We are going to talk about 15 resistant pathogens; we are going to talk about 16 bacteremia later today. I think those are some of 17 the more problematic areas where some kind of 18 guidance would be helpful so we know if it is 19 feasible to go forward. But that is a later stage 20 of things. As Frank said a moment ago, you don't 21 worry too much about what the guidance says if you 22 don't have a discovery program and you don't have 133 1 compounds coming into development. So, it has to 2 happen at multiple levels. 3 DR. EDWARDS: From the contact we have had 4 with industry, the word "guidance" and the issuance 5 of the guidances consistently has been expressed to 6 us for whatever exactly is the reason, and the 7 reason might be multifactorial, at least it keeps 8 emerging as a very prominent part of their 9 response. Renata? 10 DR. ALBRECHT: Just to agree with the 11 comments that have just been made. Again, clearly, 12 we think guidances are important and we are working 13 hard, as many of my colleagues have said, on 14 finalizing those. But just to echo a comment that 15 Ed and others have made, even when there isn't a 16 written guidance there is a lot of communication, 17 interaction and meetings with industry. Without 18 mentioning specific drugs or companies, certainly 19 our division--and Janice can comment on this--has 20 met and had discussions with multiple companies 21 about developing certain procedures for resistant 22 organisms and what it would involve. I think that 134 1 progress has been able to continue even as we 2 continue to work on the resistant pathogens 3 guidance document. So, the guidance is a starting 4 point, as has been said, but it is not the end-all 5 and be-all, and we are still working with companies 6 to make sure that we don't delay any developmental 7 plans that they come and bring to us. 8 DR. EDWARDS: Good. Thank you for your 9 comments. David, let me go ahead and call on you 10 and then I am going to have to shift into the rest 11 of our agenda, I am afraid. 12 DR. SHLAES: I didn't want to prolong 13 things but on the issue of the importance of 14 guidance, I think the audience in this room is the 15 wrong audience. I mean, we all know that we can 16 come to the FDA and discuss issues around specific 17 compounds. We have all had that experience. We 18 appreciate it; we know how important it is. It is 19 the management of the companies that need the 20 guidance. Because when we tell them it is okay, we 21 can go to the FDA; don't worry about it, they don't 22 believe us-- 135 1 [Laughter] 2 I am serious. Actually, I was just 3 talking to Dr. Goldberger about this earlier, but 4 the importance of the guidance is not for us; it is 5 for the management of companies and that is why 6 when the IDSA went around to all the companies and 7 talked to CEOs, every other word out of their mouth 8 was "guidance." So, that is the issue. 9 DR. EDWARDS: Well, that is a very good 10 point. Frank? 11 DR. TALLY: It is not only the management 12 of the "big" pharma but without a guidance document 13 when you go to the venture capital world or the 14 public market to raise money for small companies to 15 develop drugs, they point to the lack of guidance 16 too, that it is in the area that gives uncertainty. 17 Once you bring in uncertainty to Wall Street they 18 won't invest. So, there is a point for a guidance 19 so maybe it should be made much more general so 20 there is something there because in the development 21 of a couple of drugs for resistant organisms it was 22 only the interaction, as we have just talked about, 136 1 with the agency that led us to put the plans 2 together to be able to accomplish approval of those 3 drugs. So, there is a general background and I 4 think in the advisory committee meetings over the 5 last two years that guidance has come out and I 6 have partaken in those, and we have addressed a lot 7 of the issues and I think a lot of them have been 8 resolved, and it would be nice to get them out 9 there, in the general principle, to take that 10 uncertainty away from the investment pocket and 11 from "big" pharma, big management. But all of the 12 people doing the research know kind of what to do 13 and interact with the agency. 14 DR. EDWARDS: Those are great points that 15 have just been made that we really haven't 16 discussed much in the past, and it is something 17 that we might really want to develop in the future 18 discussions. So, thank you very much. DR. POWERS: 19 I think one of the things we want to say about that 20 though is that for you guys, when you come in--for 21 us, you know, at the level when the drug is 22 actually getting developed--for a CEO it seems like 137 1 all we need to put on paper is bring us the drug; 2 we will develop it. But for the nitty-gritty brass 3 tacks of doing those trials, those guidances need 4 to be fleshed out a little more in terms of the 5 level of detail, and I think that is what we have 6 been tried to work on, getting them to that point 7 so that we don't have to rewrite them in a year. 8 DR. EDWARDS: At this point we are going 9 to move on to the third part of the program, which 10 is the beginning of the discussion of surrogate 11 endpoints. I will call on Sheldon Kaplan, from the 12 IDSA, to begin the discussion. 13 III. Microbiologic Surrogate Endpoints 14 in Clinical Trials - IDSA 15 DR. SHELDON: Thank you. 16 [Slide] 17 Let me say I have a little upper 18 respiratory infection myself and I m actually not 19 taking any antibiotics for that. But I was asked 20 to address, on behalf of the IDSA, microbiologic 21 surrogate endpoints in clinical trials. 22 [Slide] 138 1 That has been the subject of a number of 2 workshops and FDA discussions, and I actually took 3 this slide right from the FDA web site, discussing 4 distinctions between biomarkers, a clinical 5 endpoint and a surrogate endpoint. A surrogate 6 endpoint is a biomarker intended to substitute for 7 a clinical endpoint and it has to predict clinical 8 benefit, harm or lack of benefit or harm. 9 [Slide] 10 As a result of using a surrogate endpoint, 11 the FDA might reasonably likely have a new drug 12 approach if these endpoints predict clinical 13 benefit on the basis of an effect on a clinical 14 endpoint other than survival or irreversible 15 morbidity. 16 [Slide] 17 So, in thinking about this, of course 18 there are already a number of infections for which 19 microbiologic surrogate endpoints are already 20 useful and are used for clinical trials, for 21 example, group A strep. pharyngitis; uncomplicated 22 lower urinary tract infection; Shigella 139 1 gastroenteritis, and there are a number of others 2 and we will hear about some other clinical entities 3 for which microbiologic surrogate endpoints are 4 useful. 5 [Slide] 6 For group A streptococcus, symptoms are 7 going to resolve regardless of therapy. The time 8 to resolution could be compared but you will never 9 be able to really compare two agents with respect 10 to suppurative or noon-suppurative complications 11 that occur way too infrequently to use as clinical 12 endpoints. 13 [Slide] 14 There are other infections for which 15 surrogate endpoints probably are not useful or are 16 unproven. For skin and skin structure infections, 17 if you are not dealing with an abscess, if the 18 treatment is clinically successful there is nothing 19 to reculture. 20 For pneumonia there is no organism 21 isolated so it would be difficult to determine 22 that, at least in pediatric studies. Perhaps 140 1 sputum cultures in adult pneumonia trials might be 2 a surrogate endpoint. 3 In acute hematogenous osteomyelitis or 4 septic arthritis you are not going to resample the 5 bone at different points into treatment to prove 6 that the bone is sterile. The same would be true 7 for intra-abdominal infections and for viral 8 meningitis and encephalitis it might be difficult 9 to use a surrogate endpoint. 10 [Slide] 11 Why is this true? Again, the site might 12 be difficult to resample. In other cases the lack 13 of eradication of an organism may not actually 14 equal clinical failure. If someone has a 15 ventilator-associated pneumonia and the tracheal 16 aspirate still contains the organism, that does not 17 equal clinical failure at all. In other cases you 18 may eradicate the organism but that may not equal 19 substantial clinical benefit, such as in upper 20 respiratory infection due to enterovirus. 21 [Slide] 22 There may be other infections for which 141 1 surrogate endpoints are useful. This requires 2 further study, in my opinion, for bacterial 3 meningitis, and I know there has been a lot of 4 discussion on acute otitis media and sinusitis. 5 Actually, I was asked to talk about acute otitis 6 media initially and I decided I would really rather 7 not discuss that issue. It has been one of 8 tremendous controversy and one that has been 9 discussed at this type of venue many, many times. 10 VP shunt infections I think might be an infection 11 for which a microbiologic endpoint makes a lot of 12 sense, to me, for a study. Coagulase-negative 13 staphylococcus line-associated bacteremias seems 14 like a surrogate endpoint that is microbiologic 15 makes sense there, and perhaps pertussis. 16 [Slide] 17 But I want to focus on bacterial 18 meningitis again. This has been the subject of 19 major discussions. This was discussed at a 20 workshop in November, 2002. Here are some of the 21 questions that were brought up at that particular 22 workshop: Are there data to show that a 142 1 microbiologic outcome is as good a marker as 2 clinical outcome, or would it be a good surrogate 3 for clinical outcome? Or, would you perhaps miss 4 potential differences of effects on inflammatory 5 responses? How do you best get preclinical and 6 early phase clinical trial data to use in 7 meningitis trials to address some of these issues? 8 That is a very interesting question. 9 [Slide] 10 What is the best timing of repeat lumbar 11 punctures to determine the eradication of an 12 organism in meningitis? How many organisms might 13 even be required to say you have delayed 14 sterilization, and what about quantitative 15 cultures? These are really all difficult questions 16 to address. There are very few studies that we can 17 turn to, to look at these issues. 18 [Slide] 19 The IDSA guidelines, developed in 1992, 20 have clinical endpoints of cure; survival with mild 21 neurologic sequelae; survival with severe 22 neurologic sequelae which I think are somewhat 143 1 dependent on the observer--are these clinicians, 2 pediatricians? Are they neurologists doing these 3 examinations? And, of course, death. 4 One of the other problems with some of 5 these endpoints are that clearly some of the 6 neurologic sequelae improve with time so are you 7 going to be looking at the patient at the end of 8 therapy, two or three months later or a year after 9 they have been involved in the study? 10 We can't really look at mortality between 11 two agents in the U.S. because our mortality rates, 12 thank goodness, are very, very low. 13 So, I have selected to look at audiology 14 results as one potential surrogate marker. This is 15 an objective measure. It is quantitative. 16 Although, as with other neurologic sequelae, many 17 of these hearing losses may improve with time. 18 [Slide] 19 We took a couple of studies, comparing 20 head-to-head antibiotics. We have this study that 21 was published in The New England Journal of 22 Medicine, ceftriaxone versus cefuroxime, a 144 1 randomized trial, very small number really of 2 patients, but we have a repeat CSF culture at 18-36 3 hours. There were no significant differences in 4 the clinical characteristics of the patients at the 5 time of enrollment. 6 [Slide] 7 If we focus on the repeat LP culture 8 results and hearing loss, we see that a positive 9 CSF culture in the ceftriaxone group occurred in 10 1/52 versus 6/52 in the cefuroxime group. This 11 actually was not a significant difference although 12 it is clearly a trend. All of the positive 13 cultures were Hemophilus influenzae type b. There 14 was a 4 percent incidence of hearing loss in the 15 ceftriaxone group versus 17 percent in the 16 cefuroxime group, which was almost clinically 17 significant. So, we actually have perhaps a 18 dichotomy where we don't see a significant 19 difference in the positive CSF cultures at the 20 second tap but almost a significant difference in 21 hearing loss. 22 [Slide] 145 1 If you look further at this, 2/27 patients 2 with H. follow-up type b meningitis had hearing 3 loss, 7 percent in the ceftriaxone group versus 17 4 percent in the cefuroxime group. However, I would 5 mention that all the patients with positive 6 cultures had Hemophilus influenzae type b. Two of 7 six patients who had hearing loss after cefuroxime 8 therapy had delayed sterilization of CSF. But four 9 of the patients who had hearing loss did not have 10 delayed sterilization. 11 [Slide] 12 For pneumococcus we also see a slight 13 difference perhaps, 0/8 in the ceftriaxone group 14 and 2/6 in the cefuroxime group. None of the 15 patients with pneumococcal meningitis with hearing 16 loss had a delay in their CSF sterilization. 17 [Slide] 18 So, if we do the little calculation that 19 has been performed for acute otitis media comparing 20 bacteriologic success and clinical success, we see 21 that the sensitivity of a sterile repeat CSF is 91 22 percent but its specificity for a repeat CSF being 146 1 positive and showing clinical failure is low, 2 around 30 percent. 3 [Slide] 4 If we also look at studies that were done 5 in Dallas over several years, these are four 6 prospective trials conducted consecutively, with 7 three of them also looking at dexamethasone so it 8 is really a hodge-podge of clinical trials, none of 9 these trials compared agents directly. But overall 10 there were 174 children who received ceftriaxone 11 and 159 who received cefuroxime and there were no 12 significant differences between the groups at the 13 initiation of therapy. 14 [Slide] 15 If we look at the CSF culture at 16 follow-up, for ceftriaxone it was stated in the 17 paper that they were uniformly sterile; 9 percent 18 of the cefuroxime follow-up cultures were positive, 19 highly significantly different. Whereas, for 20 hearing loss there was no significant difference. 21 So, actually we see something quite different than 22 what was shown in the earlier paper by Schaad. 147 1 [Slide] 2 If we look at some more recent studies, 3 meropenem versus cefotaxime for bacterial 4 meningitis, and look at the sequelae--these are the 5 IDSA sequelae groups, cure, mild sequelae and 6 severe sequelae, and you can see that the two 7 agents were really quite comparable. There were 8 virtually no patients who had positive CSF cultures 9 at the follow-up tap; two in the meropenem group, 10 one in the cefotaxime group. They all happened to 11 be Hemophilus influenzae. So, obviously, many of 12 these patients with severe sequelae had negative 13 repeat CSF cultures. 14 [Slide] 15 The same thing is true for a more recent 16 study comparing trovafloxacin to ceftriaxone. 17 There were 8 children who had positive repeat 18 cultures. I believe 5 of these were pneumococcus 19 but, again, the vast majority of patients who would 20 have had severe sequelae must have had a negative 21 CSF culture. 22 [Slide] 148 1 So, I think, at least as far as bacterial 2 meningitis is concerned, it is really not clear to 3 me how well a repeat CSF culture at 24-36 hours or 4 18-36 hours after initiation of therapy predicts 5 hearing impairment or the overall outcome with the 6 fact that the majority of patients with severe 7 sequelae have a sterile CSF culture at follow-up. 8 It is not clear at all if the findings for 9 Hemophilus influenzae type b meningitis are then 10 applicable to pneumococcal or meningococcal 11 meningitis, and these are the infections, of 12 course, that we are seeing more commonly now in the 13 United States, although pneumococcal meningitis is 14 really on the decline. 15 So, those are my comments about 16 microbiologic endpoints as surrogate markers, 17 focusing really on bacterial meningitis. Thank 18 you. 19 DR. EDWARDS: Thank you very much. Yes, 20 Barry Eisenstein, from Cubist. 21 Industry 22 DR. EISENSTEIN: Thank you. 149 1 [Slide] 2 I am going to be representing the thinking 3 of a number of people within Cubist, particularly 4 those of us involved in clinical development and 5 drug evaluation. What I think the morning 6 discussion was reasonably focused on a fairly high 7 strategic level. What I am going to be talking 8 about is much more tactical, and I like the comment 9 that was made earlier today about thinking globally 10 but acting locally. Perhaps it is the combination 11 of small steps that help us move along the path 12 towards easier registration and then uptake that 13 can help to incentivize industry. 14 What I would like to do is to briefly 15 review some aspects of the Subpart H accelerated 16 approval process and talk about the pros and cons 17 and where that may be helpful to industry; where it 18 may not be as helpful, as least in some of the 19 indications that we are thinking about; and then 20 lead to discussion of a particular example where 21 microbiologic surrogates could be of some value in 22 helping us at least think about ease of 150 1 registration. 2 [Slide] 3 The bottom line with surrogate endpoints, 4 of course, is that we want to consider them as a 5 substitute for a clinical endpoint and presumably 6 they should be easier to obtain; they ought to be 7 quicker to obtain. Combination should either make 8 trials quicker or less expensive or smaller. To do 9 that though, we have to be able to demonstrate I 10 believe two things, one of them, you can measure 11 the surrogate precisely and I think this goes to 12 the comments made earlier by the agency that 13 well-designed trials is really what matters here in 14 terms of being able to shorten them and make them 15 smaller. Obviously, there has to be prediction of 16 clinical outcome. 17 [Slide] 18 So, if you look at Subpart H, some of the 19 key features for accelerated approval are that, 20 number one, there has to be reasonable likelihood 21 to predict clinical benefit. You have to be able 22 to verify and describe clinical benefit, a key 151 1 component. In this case, there has to be a 2 clinical follow-up after the approval based on the 3 surrogate, a very important point that I will get 4 back to in a moment. I think appropriately, the 5 conditions for which this should be used are 6 limited to those that are considered to be either 7 serious or life-threatening. 8 [Slide] 9 So, if we look at some aspects of Subpart 10 H, the incentive to industry, I think their 11 particular value is in situations where there are 12 large patient populations and truly delayed 13 clinical endpoints. A good example is 14 hypertension, cholesterol and HIV, which has been 15 mentioned before. The surrogates in any of these 16 cases allow earlier patient evaluation, submission, 17 approval, but no reduction in overall development 18 cost because of the subsequent requirement for 19 clinical confirmation, on the other hand and in 20 particular, if the population to be treated is 21 large enough the cost can be offset by the earlier 22 approval. 152 1 On the other hand, if one is dealing with 2 smaller populations, and this is particularly the 3 case for some of the "niche" antimicrobials that 4 many of us considering given the drug resistance 5 aspect that we are all trying to focus on--with 6 smaller populations you are obviously dealing with 7 more difficult accruals, higher development cost 8 because of that. The cost of demonstrating 9 clinical benefit still remains. 10 So, the cost of the earlier approval may 11 not be offset and, in fact, you may have to 12 over-enroll in the beginning in a trial that has to 13 be followed for years to ensure that one has the 14 appropriate follow-up. I will give an example of 15 that. Well, I will show my hand. The example I am 16 going to be talking about is osteomyelitis. If one 17 considers the long-term need for follow-up in 18 osteomyelitis, it is quite obvious that one has to 19 start off in this mind set at least with a very 20 large number of individuals to get the right 21 follow-up. 22 [Slide] 153 1 So, what about the other approaches? If I 2 say the traditional approach to registration, can 3 one find a way that surrogates can be used? Well, 4 there are some examples. They have been mentioned 5 already. In infectious diseases, urinary tract 6 infection, gonorrhea, pharyngitis where organisms 7 can be shown to be eliminated. But full validation 8 is also often impossible or not feasible for 9 certain situations where one might think a 10 surrogate could be of great use. So, one might be 11 in the VRE case superimposed on an ultimately fatal 12 disease like cancer, where you would like to be 13 able to demonstrate the value of eliminating the 14 VRE because one knows there is high attributable 15 mortality, but in the background of the enormous 16 amount of noise with an ultimately fatal disease it 17 might be difficult and, therefore, one might need 18 to have extraordinarily large populations. 19 Another example, and the one I am going to 20 talk about at more length, is osteomyelitis where 21 there is need for a very long follow-up. Obviously 22 more limited validation could provide enabling 154 1 incentive by reduction error in endpoint 2 measurements. That increases the power of the 3 study, leading to smaller study populations. These 4 study populations, being smaller, are more 5 feasible. IN some cases there can be a shorter 6 time to evaluation, in the case I mentioned about 7 osteomyelitis, and there could be major impact in 8 serious indications affecting smaller number of 9 patients. 10 [Slide] 11 Examples of Subpart H and examples of the 12 other approach, I could say--accelerated approval 13 has been demonstrated with chronic viral 14 infections, particularly HIV. Whether this can be 15 applied to hepatitis C I think is yet to be 16 demonstrated. Tuberculosis has been mentioned. 17 Other antibacterial examples include Synercid for 18 VRE; Biaxin for MAC, but the big caveat there is 19 that mortality was actually higher at the higher 20 dose even though elimination of the microorganisms 21 was shown to be better. So, there are major 22 caveats. Then, ciprofloxaciin for the elimination 155 1 of anthrax where, in this case, an animal model is 2 used to substitute for the obvious inability to do 3 the clinical study. 4 Surrogate endpoints with limited 5 validation for traditional approvals could include 6 bacteremia. I believe we are going to hear much 7 more about that later. I am looking forward to 8 that discussion. Osteo. and prosthetic joint 9 infection, I will have more to say on that in a 10 moment. Then, let me just make a plug again, we 11 have heard this before, for having sensitive 12 isolates being used as surrogates for resistant 13 isolates, VSE, VRE, MSSA, MRSA. An example here is 14 that it might be difficult to be able to get enough 15 samples of, say, VRE such that a label ends up 16 saying enterococcus faecalis and then, in 17 parentheses one sees, vancomycin sensitive or 18 susceptibility organisms only, giving, as I have 19 heard from the field, some misinterpretation that 20 your drug may, in fact, not be adequate at all, 21 useful at all for VRE when, in fact, from all of 22 the scientific evidence, shortly of a fully-fledged 156 1 out clinical study, you have very strong evidence 2 that it can be very effective and very useful. So, 3 just a little plug on that with some very recent 4 experience in that field. 5 [Slide] 6 What does it take to make a good surrogate 7 marker? I think Tom Fleming had a very nice paper 8 about eight years ago. Interestingly though, it 9 did not talk about antimicrobials so that one could 10 apply the same teaching. I think here one is 11 dealing with a discussion of causation versus 12 correlation. The best surrogates obviously are 13 both necessary and sufficient in the causation of 14 disease. For microbiologic surrogates, at the very 15 least, one must satisfy Koch's postulates and then 16 go beyond that, in a sense, to demonstrate that not 17 only is the organism necessary and sufficient for 18 disease, but elimination of the organism is 19 necessary and sufficient for elimination of the 20 disease. 21 One of the problems, I would say, with 22 meningitis is that because of the significant 157 1 inflammatory cascades that get evoked with the 2 destruction of microorganisms one could, in fact, 3 get some peculiarity and de-linking of some of the 4 very strict causation-correlation coupling there. 5 Obviously, culture sample acquisition is 6 critical. You need to ensure sterile access to 7 otherwise sterile samples. That avoids false 8 positives. You need to avoid sampling error, on 9 the other hand, and you need to avoid false 10 negatives. 11 [Slide] 12 Let's go on to the example I would like to 13 give, which is osteomyelitis. One could think of 14 several different conditions for osteo., 15 hematogenous, contiguous and prosthetic joint, 16 which one initially might not think would be a way 17 to develop a drug but there might be a way in here 18 that could be interesting to discuss. 19 Let's talk about hematogenous. The 20 initial culture is clean. False positives are 21 unlikely if one gets biopsy and culture results. 22 False negatives though are possible, particularly 158 1 on repeat evaluation, and one also has the 2 difficult problem, as has been mentioned before, of 3 being able to actually go in and get individuals to 4 get their culture results after therapy. Then 5 years later there is the potential for late 6 recurrence. This is obviously a difficult topic to 7 study. 8 For contiguous osteomyelitis, as was 9 discussed in part at the AC meeting on diabetic 10 foot, culture diagnosis is fraught with error, with 11 false-positive and false-negative sampling errors 12 and obviously, again, the potential for late 13 recurrence. 14 Prosthetic joint infection with adjacent 15 osteomyelitis has an initial culture that is clean. 16 It is an open surgery; you can get a biopsy. The 17 post-treatment culture is clean and standard of 18 care, particularly in the two-state procedure that 19 I will describe. There is infrequent late 20 recurrence if, and only if--and here is the key 21 linkage--if, and only if, the culture results are 22 negative. To me, this is the definition of a good 159 1 surrogate. Moreover, antibiotics are needed in 2 addition to surgery--evidence I will show 3 briefly--indicating that there is no or very weak 4 placebo effect so that there truly is a value of 5 the antimicrobial therapy. 6 [Slide] 7 Let's talk about the example of Staph. 8 aureus prosthetic joint infection. This is a 9 serious disease I think by any definition. 10 Clinical evaluation of cure requires prolonged 11 follow-up. It is difficult and costly to do the 12 study for a large number of patients if one is 13 going to power this adequately for long-term 14 follow-up. 15 [Slide] 16 The two-stage procedure that I mentioned 17 just briefly to describe it is that the infected 18 joint is removed, debrided. There is typically 19 adjacent osteomyelitis. In fact, individuals in 20 the study must have, by the description that I am 21 giving, adjacent osteomyelitis to get in the study. 22 There are optional local antibiotics that I would 160 1 call standard of care, similar to the debridement. 2 Importantly, systemic antibiotics are given for 4-6 3 weeks, after which time individuals go off 4 antibiotics for a period of anywhere from 203 5 months, at which time the surgeon goes back into 6 what should now be a clean field as is able, under 7 visual inspection, to look very carefully at the 8 disease state and do the appropriate biopsies. I 9 would argue one would see high specificity and 10 sensitivity. The results of culture reimplantation 11 currently is used as the indicator of therapeutic 12 success so that the surrogate, in fact, does become 13 the means by which the surgeon makes his or her 14 definition of cure. 15 [Slide] 16 Why is it that we can say that systemic 17 antibiotics have an effect? Why wouldn't surgery 18 alone be of value? I think the argument could be 19 made in two ways. One of them is that if you just 20 do the one-stage reimplantation where you do all 21 the surgical things but you don't allow the 22 antibiotics for a period of time and then a test of 161 1 cure there is a significant failure, double-digit 2 failure rate as opposed to a much lower failure 3 rate with the two-stage procedure, arguing that 4 systemic antibiotics do make difference. 5 I think an even more compelling case can 6 be seen with the other publication in Clinical 7 Orthopedics, where in the two-stage procedure 8 individuals had organisms that were either defined 9 as sensitive or resistant and the failure rate was 10 significantly higher in the resistant organisms 11 than in the sensitive organisms, I think strongly 12 arguing again for the importance of the systemic 13 antibiotics. 14 [Slide] 15 On the other side, the negative culture I 16 think is highly predictive by the evidence that I 17 show here, and I won't go through it in detail but 18 only to say that the negative predictive value, 19 including the possibility of reimplantation of a 20 new organism at the time of surgery, at worst, is a 21 negative predictive value of 97 percent at 5 years, 22 92-97 at 9 years, so extraordinarily long 162 1 follow-up. 2 [Slide] 3 In conclusion, I would say that 4 microbiologic surrogates for accelerated approval, 5 particularly with the Subpart H category, is 6 extremely helpful for large patient populations but 7 microbiologic surrogates with limited validation 8 could be used for traditional approval for the 9 smaller populations. That will enable companies to 10 try to develop drugs for these smaller "niche" 11 indications. 12 I would argue that a promising example 13 could be prosthetic joint infection. It is a 14 serious disease. It is difficult to enroll many 15 patients. Clinical test of cure is not a feasible 16 endpoint because it takes 5-9 years or more. There 17 is a clean, predictive microbiologic surrogate 18 endpoint that closely tracks with the ultimate 19 clinical endpoint, and there is a negligible 20 placebo effect. Thanks. 21 DR. EDWARDS: Thank you very much. I 22 think we will go right to Dr. Powers for the third 163 1 presentation. 2 FDA 3 DR. POWERS: Thanks, Jack. This is the 4 fun part, to get off the policy stuff, no offense 5 but we will do some science now. 6 [Slide] 7 So, what I wanted to talk about now is 8 sort of our perspective on what we have looked at 9 in terms of surrogate endpoints in clinical trials 10 in infectious diseases. As Dr. Goldberger said, we 11 are in a unique position in that we get to see this 12 information over a broad range of drug applications 13 and studies. I wanted to go through some of the 14 information we have actually seen. 15 You have already heard the definition but 16 I just want to reiterate what Dr. Kaplan gave you. 17 I am going to go over some of the strengths and 18 limitations and the utility of surrogate endpoints 19 in the overall drug development process; talk about 20 some of our regulatory considerations with the use 21 of surrogate endpoints; and then talk about some 22 issues and what we need to do to validate a 164 1 surrogate endpoint. 2 [Slide] 3 This doesn't show up very well but I just 4 wanted to bring up the issue again of what is 5 really important to patients. Patients don't come 6 in and say, "I've got this E. coli I want to get 7 rid of." Patients come in because they have 8 symptoms they want to get rid of, therefore, it is 9 important for us to actually look at how the 10 patients are doing. 11 This is a cartoon from the New Yorker that 12 has a gastroenterologist scope and this person is 13 saying, "look, this is great; your endoscopy is 14 normal." You can't read this, but the patient says 15 has all these persisting complaints that are still 16 continuing. This is the situation we want to 17 avoid, getting a surrogate that is telling us you 18 ought to feel great when, in fact, the patient 19 feels crummy. 20 [Slide] 21 Just to go over these definitions again, a 22 clinical endpoint is a direct measure of how a 165 1 patient feels, functions or survives. That means 2 we are looking at things like mortality or 3 resolution of the symptoms of disease. The real 4 easy answer to what is a surrogate is that it is 5 anything other that measures how a patient feels, 6 functions or survives. So, that would be things 7 like laboratory measurements or a physical sign 8 used as a substituted for a clinical endpoint. We 9 use these all the time in clinical medicine. 10 Getting an X-ray on a person is really a surrogate 11 of how their pneumonia is doing. We are very 12 comfortable in their use in clinical practice, but 13 the surrogate endpoint by itself does not confer 14 direct clinical benefit to the patient. We all 15 know that we see patients, especially elderly 16 patients, in whom their pneumonia is clinically 17 resolved yet it takes their X-ray 6-8 weeks to get 18 better. 19 [Slide] 20 Some examples of surrogates that we use 21 all the time are measurements of the organism 22 itself, like culture or viral load; antigen 166 1 testing; X-rays; CAT scans; endoscopies; and even 2 histologic changes. 3 One of the other things we do is we get 4 into this issue of a surrogate of a surrogate. 5 When you start feeling comfortable with the initial 6 surrogate, you then start to drift even further 7 away from the patient, you look at changes in 8 cultures associated with histology. That is a 9 surrogate of a surrogate. Or, presence of an 10 organism on a catheter tip. I always tell medical 11 students and residents that I teach that catheters 12 don't get infected, people do, yet we refer to 13 catheter infections all the time. Finally, there 14 is the issue of elimination of organisms in a 15 population without the disease. Well, gee, if I 16 make the bug go away in uncomplicated UTI, can't I 17 just show that making the bug away in a person with 18 asymptomatic bacteria is the same thing? But that 19 ignores the host organism response and there also 20 may be some differences in the organisms that 21 affect those various populations. 22 [Slide] 167 1 As Dr. Kaplan brought up, surrogate 2 endpoints are a subset of biomarkers, and there is 3 an entire NIH conference on this with results that 4 were published in Clinical Pharmacology and 5 Therapeutics, in 2001. But biomarkers are very 6 useful for purposes other than measuring endpoints 7 in clinical trials and I think this is a real point 8 of confusion when we start talking especially about 9 culture data. 10 One of the things we have talked about in 11 the advisory committees is that we do want 12 microbiologic information to diagnose the disease 13 in question. So, using a biomarker as a diagnostic 14 tool increases the specificity of the diagnosis. 15 In October of 2003 we talked at great length about 16 this in acute bacterial sinusitis. When we look 17 through our new drug applications if you just took 18 all-comers your rate of actually getting bacterial 19 disease was somewhere around 35 percent. So, that 20 meant that in your trial you were enrolling 70 21 percent of the people that didn't have bacterial 22 disease. 168 1 Also, a biomarker can be a risk factor for 2 a disease. For instance, looking at the patient's 3 white blood cell count in a patient with a 4 hematologic malignancy gives you a clue that that 5 person is now at risk for a bacterial or a fungal 6 infection when they are neutropenic. Neutropenia 7 is not the disease however. Also, resolution of 8 neutropenia decreases the risk. That doesn't mean 9 the disease has gone away. 10 But I think the trickiest part here is 11 looking at biomarkers as an indicator of disease 12 prognosis and contrasting that to a biomarker as a 13 surrogate endpoint for the outcome of the disease 14 in question. I think we get these mixed up all the 15 time. For instance, HIV viral load and CD4 counts 16 in HIV, and the reference I put here is an NIH 17 consensus conference that was published in JAMA in 18 1993. So, 11 years ago these folks got together 19 and showed that HIV viral load is both a good 20 indicator of disease prognosis in HIV and a good 21 measure of treatment outcome. CD4 count is a good 22 measure of prognosis and is not a good measure of 169 1 treatment outcome because a number of people will 2 die even though their CD4 count rises on treatment, 3 and I think that is a really important distinction 4 to make. 5 [Slide] 6 Here is one way to look at this. When you 7 are looking at a biomarker as a prognostic 8 indicator what you are doing is you are taking all 9 the people that have a microbiologic failure and 10 all the people that have a microbiologic success, 11 and what you show is that the people who have 12 microbiologic failure are more likely to be a 13 clinical failure than the people who have 14 microbiologic success. What you just came up with 15 was a relative risk. That may or may not tell you 16 anything about the results of treatment. The 17 reason I point out this example is that, if you 18 look here, there are more clinical successes in 19 both of these arms than there are clinical 20 failures. 21 [Slide] 22 So, that gets us to the idea of what we 170 1 really want to do when we are evaluating a 2 biomarker as a surrogate endpoint. What we now 3 want to do is look at microbiologic outcomes and 4 outcomes means successes and failures, not just 5 failures. I think that is part of the issue, as 6 Dr. Kaplan referred to in this issue about otitis 7 media which is that that data has concentrated on 8 looking at failures and not at the kids who were 9 also successes. So, what we want to do here on the 10 other side is look at the clinical outcomes, both 11 successes and failures, and match up the concordant 12 patients who are micro. successes and clinical 13 successes or micro. failures and clinical failures 14 to the people who are also discordant in terms of 15 micro. successes, who still fail therapy, or kids 16 in whom the bug is still present but who get better 17 anyway. 18 [Slide] 19 Another way to visually represent this is 20 to look at this overlap. So, as Dr. Eisenstein 21 pointed out, you will never expect this to be 100 22 percent; 30 percent of people with severe 171 1 community-acquired pneumonia die no matter what 2 drug you give them. So, we are not expecting this 3 overlap to be 100 percent. But what do you do when 4 the micro. and clinical outcomes don't overlap by a 5 great amount, as Dr. Kaplan pointed out in 6 meningitis, where the microorganism is eradicated 7 and the person still does badly, and the otitis 8 media is the flip side where the organism is not 9 eradicated and the kids get better anyway. 10 [Slide] 11 So, what are the strengths of surrogate 12 endpoints? Well, they are reproducible. They are 13 standardized, at least in some cases where we have 14 standardized ways of culturing and detecting 15 susceptibilities. They are objective. They can be 16 dichotomous, they can give us a yes/no answer which 17 is easier to analyze; and they can lower the sample 18 size when the success rate is higher with the 19 microbiologic endpoint and, importantly, in chronic 20 diseases, as Dr. Eisenstein pointed out, we can 21 measure it earlier. 22 [Slide] 172 1 Let me just show you some information with 2 sample size. If we have a 90 percent success rate 3 here with a microbiologic endpoint, and let's just 4 take a 10 percent non-inferiority margin, you need 5 142 patients per arm. If the clinical endpoint 6 there were 70 percent instead of the micro. outcome 7 of 90 percent, you now need 330 patients per arm 8 with the same power and the same non-inferiority 9 margin. So, you can shrink the sample size 10 substantially if you can use a validated surrogate 11 endpoint in a trial. So, that is how a 12 microbiologic surrogate can help us in an 13 individual trial and, obviously, if you can use 14 those in a number of diseases you can shrink the 15 overall size of your drug development program. 16 [Slide] 17 Surrogate endpoints, even if they are not 18 validated, can still be helpful early in the 19 clinical development program. They can give us a 20 mechanism of understanding of how the drug actually 21 causes disease. They can give us some knowledge 22 about the clinical pharmacology of the drug or 173 1 guide us in dose selection, which we will talk 2 about at great length tomorrow. They can also give 3 us some proof of principle of efficacy in Phase 2 4 trials, and they can be a basis for selecting 5 compounds to go forward into Phase 3 testing, which 6 is a big part of the FDA's Critical Path about 7 trying to eliminate drugs that are not likely to 8 work. 9 Also, they can be helpful in later stages 10 of drug development by bringing treatment benefits 11 to patients earlier than with trials with clinical 12 outcomes when the surrogate endpoint precedes the 13 clinical endpoint by a significant amount of time, 14 such as in HIV infection, to contrast that from 15 acute otitis media where the clinical and the 16 surrogate endpoint will be measured at exactly the 17 same time. 18 [Slide] 19 But there is a payoff to this too, and 20 that is, that smaller trial may not provide you 21 enough data to analyze the safety of the drug. So, 22 the absence of an adverse event in a safety 174 1 database of 300 patients only rules out risk of 2 1/100 or one percent. How does that stack up if 3 you see one case of liver failure in those 330 4 patients? We know that the background rate of 5 liver failure in that population is one in a 6 million. So, when you see an issue like this that 7 pops up in a clinical trial's database, it really 8 is incumbent on you to study it in more detail 9 before you let the drug out onto the market. But 10 that is also dependent upon the risk-benefit of the 11 drug in question. If it is another drug for 12 sinusitis, you are probably a lot less tolerant of 13 a severe adverse event like liver failure. 14 Also, the benefits in terms of the 15 surrogate endpoint may overestimate the benefit on 16 the true endpoint. For instance, in prophylaxis 17 trials you may show a 90 percent rate of 18 decolonization which translates only into one 19 percent reduction in prevention of infection. That 20 still may be very useful because if that remains 21 constant we can still use that as a measure of drug 22 efficacy. In a treatment trial the rate of culture 175 1 negativity may overestimate the clinical cure rate. 2 I think that is something you can certainly handle, 3 but it wouldn't be something that you want 4 clinicians to say, "well, because it's got a 90 5 percent microbiologic cure rate, that means 90 6 percent of my patients are going to get better." 7 [Slide] 8 But the real issue here for us is that the 9 surrogate endpoint may not predict clinical benefit 10 at all, and that is the really trick issue. Why 11 may it not predict clinical benefit? First of all, 12 the surrogate endpoint must be on the causal path 13 of disease and there is not a lot of question 14 related to this in infectious diseases. Clearly, 15 the microorganism is on the causal path that 16 actually gets you sick. But the real question here 17 is does the intervention which results in a 18 microbiologic effect, is it the only thing that 19 affects the clinical outcomes? Dr. Kaplan already 20 alluded that in meningitis it may not be in that 21 the majority of people who don't do well have 22 negative cultures, and there may be some other 176 1 reasons for that. So, the surrogate must capture 2 the full benefit of treatment effect and assumes no 3 other pathway of the mechanism of drug effect. 4 The important point here is that the 5 mechanism of action of the drug may be different 6 from the actual clinical effect of the drug. We 7 have had several meetings where we have been told 8 by consultants who come in to talk to us that the 9 only thing antimicrobials do is eradicate the bug. 10 [Slide] 11 We don't think that is true, and the data 12 that actually speaks to that talks about this, 13 there may be unmeasured benefits or unmeasured 14 harms that actually occur with the administration 15 of an antimicrobial, and the surrogate needs to 16 take that into account as well. This presumes that 17 our knowledge about how the drug works is actually 18 complete and in many cases it is not. 19 [Slide] 20 So, what are some of the unmeasured 21 benefits that may occur? Well, there may be 22 effects of the drug other than eradication. We 177 1 know that there are some inhibitory effects of 2 antimicrobials on organisms. We know that 3 bacteriocidal therapy is not necessary in many 4 infections. It certainly is in endocarditis and 5 meningitis, but no one has ever proven you need 6 bacteriocidal therapy in acute otitis media. And, 7 there may be direct effects of the antimicrobials 8 on the host immune system, and it is fascinating to 9 look this up when you see some of these articles 10 that actually talk about the effects of 11 antimicrobials on phagocytosis and the effects on 12 the immune system in general. So, there are direct 13 antimicrobial host interactions that may impact on 14 the outcome as well. 15 There may be unmeasured harms. As Dr. 16 Kaplan pointed out, lysis of organisms in 17 meningitis may actually have an inflammatory effect 18 that causes worse clinical outcomes. In 19 prophylaxis trials you may be able to get rid of 20 one organism, which then may be replaced by another 21 organism, which then goes on to cause disease as 22 well. Or, you may have other sources of infection, 178 1 other than that affected by a prophylactic drug. 2 For instance, we might sterilize somebody's urine 3 and then they go on to get an infection from their 4 skin, or somewhere else. 5 [Slide] 6 The other issues are how do we actually 7 measure the surrogate endpoint? Do we know 8 everything we think we know about these surrogates? 9 Dr. Eisenstein already brought up the issues about 10 false-positive and false-negative cultures that may 11 impact on that. 12 [Slide] 13 Finally, there is the issue about are 14 there problems with measuring the clinical 15 endpoint? I think especially in relationship to 16 otitis media, investigators have points out this 17 idea of the "Pollyanna" phenomenon for years. 18 However, the result of that has been, well, that 19 means we ought to just chuck the clinical endpoint. 20 Well, the thing we need to realize is that this is 21 the gold standard, the clinical endpoint. So, what 22 that means is that we need to find a better way of 179 1 measuring the clinical endpoint. When your gold 2 standard has a problem it doesn't mean you can 3 chuck it and just use the surrogate. You need to 4 find a better way of looking. 5 [Slide] 6 Here is one of the things we presented at 7 the October, 2003 advisory committee about 8 sinusitis. Although this relates to the 9 oseltamivir trial in influenza, you can see that 10 most people are going to be better in a short 11 period of time from their influenza. If you 12 measured the primary endpoint out here you would 13 never be able to show a difference between drugs or 14 between drug and placebo because you have gone out 15 beyond the time of the natural history of the 16 disease. 17 The other way I looked at this, I jokingly 18 said to one of our medical officers if we evaluated 19 mortality and we had 85-year follow-up, most of the 20 people in both arms of the trial are going to be 21 dead. That doesn't have anything to do with the 22 drug's effect. So, what we want to do is to look 180 1 at a clinical time point that is relevant to the 2 natural history of the disease. So, looking at 3 some of these self-resolving illnesses, we want to 4 go back either to a fixed time point that we know 5 is relevant to the disease or use a time to 6 resolution of symptoms, and that is what influenza 7 and traveler diarrhea trials did, they used time to 8 resolution of symptomatic disease. 9 [Slide] 10 Well, how do you go about looking at some 11 of this? Well, Dr. Kaplan showed you this idea of 12 sensitivity and specificity. If you do a 2 X 2 13 table that compares success with that surrogate 14 marker to clinical success, what you get here is 4 15 cells. These cells were the success of the 16 surrogate and success of the clinical outcome or 17 failure with the surrogate and failure with the 18 clinical outcome are where we have concordant 19 cells. On the other hand, we have these discordant 20 cells, here in red, where surrogate failure still 21 results in clinical success or there is clinical 22 failure even though there is success with the 181 1 surrogate. 2 What we really want to do then, if we look 3 up and down here, on the left, that is sensitivity. 4 If we look up and down here, on the right, that is 5 specificity. But what we really want to do is show 6 the relationship between sensitivity and 7 specificity, which we refer to as concordance or 8 correlation. 9 One way of doing this is actually do a 10 kappa coefficient of correlation, which is another 11 way for a dichotomous variable to actually look at 12 this concordance rate. What makes a good kappa? 13 Well, that is a judgment call obviously, but at 14 least one article in the literature that reviews 15 all this suggests that a kappa of less than 0.4 is 16 marginal or no agreement; a kappa of between 0.4 17 and 0.75 is good agreement; and a kappa of greater 18 than 0.75 is excellent agreement. 19 [Slide] 20 What I did, I took these kappas and I 21 turned them into lines so the slope of the line 22 would be equal to what that kappa is. A perfect 182 1 correlation slope would be a slope of 1. So, if 2 you had 80 percent success with the surrogate, you 3 would get 80 percent success in the clinical 4 outcome as well. We know that is not going to 5 happen, especially in severe diseases. There are 6 going to be people who die, no matter what drug 7 they get. So, we are going to see something less 8 than 1 in terms of the correlations of these 9 surrogate outcomes and the clinical outcomes. 10 [Slide] 11 What we actually did is to go back and we 12 reviewed some of this information for some of the 13 diseases where we could pool the information on how 14 well did those surrogate outcomes correlate with 15 clinical outcomes. We reviewed 12 clinical trials 16 on acute otitis media that were done with 17 tympanocentesis and came up with a combination 18 kappa of 0.37 for those 12 trials. 19 We also took that Lebel trial that Dr. 20 Kaplan showed and did the same kind of correlation 21 and showed that for acute bacterial meningitis the 22 kappa was 0.14, almost down to the level of a coin 183 1 flip. 2 Again, these are for different reasons in 3 these different diseases. In acute otitis media 4 the discordant part is people in whom the culture 5 is still positive but they are getting better 6 anyway. In acute bacterial meningitis it is the 7 exact opposite where the culture is negative but 8 the people are still having sequelae anyway. 9 [Slide] 10 Let me just show you another example. 11 Here are the 12 trials where we looked at the acute 12 otitis media outcomes. It is very interesting that 13 they all kind of group down here, depending upon 14 which drugs you look at, but there is a good bit of 15 variability from 0.14 to 0.42. But look what 16 happens in the one trial in which the surrogate 17 outcome and the clinical outcome were measured at 18 the same time. In the rest of these trials what 19 you saw is that the surrogate was measured on day 20 4-6 on therapy and, yet, the clinical outcome was 21 measured way out here, somewhere beyond 10 days, 22 which we know is beyond the natural history of the 184 1 resolution of otitis media. But this is only one 2 trial where this was done. Is this because this 3 drug worked better? Is it because the surrogate 4 was better for that drug? Or, is it because the 5 timing of the surrogate was done in a more relevant 6 way? We can't tell because it is the only trial 7 that was done that way so we still have unanswered 8 questions here. 9 [Slide] 10 What happens though if that correlation is 11 different for drug A versus drug B? As I showed 12 you here, it is very different for all these 13 different drugs in acute otitis media. So, if you 14 have a difference between drug A and drug B, what 15 happens is that when you look at the clinical 16 outcomes or the surrogate outcomes here, it appears 17 that drug B is actually superior to drug A. But if 18 those lines don't line up, what should happen over 19 here is that on the true clinical outcome drug A is 20 actually superior to drug B. So, what happened is 21 you got told the exactly wrong thing by using that 22 surrogate outcome. 185 1 So, you can see that these are some of the 2 issues that we have to deal with when we are 3 talking about validating surrogate endpoints and 4 the potential for them to tell us the wrong thing, 5 and why it is important to validate the surrogate 6 ahead of time. 7 [Slide] 8 In traditional approvals that are based on 9 surrogate endpoints we can only do this, 10 regulatory-wise, where the endpoint is already 11 validated to predict a clinical endpoint. We can 12 do this under Subpart H in accelerated approval but 13 that only applies to serious and life-threatening 14 diseases. Therefore, acute exacerbations of 15 chronic bronchitis, acute bacterial sinusitis and 16 acute otitis media would not really qualify under 17 Subpart H. It also requires confirmatory 18 post-approval trials based on a clinical endpoint. 19 [Slide] 20 But here is part of the issue, when you do 21 a trial like HIV, what usually happens is you check 22 the viral load at 24 weeks and then you follow 186 1 those same patients, and they continue on in the 2 trial and you get longer-term follow-up. What do 3 you do in a trial of acute bacterial disease where 4 the clinical and the surrogate endpoint are 5 measured and these people are cured? There is 6 nobody to follow. So, this is the situation like 7 in clearance of VRE bacteremia with Synercid. How 8 does one go on and validate that surrogate endpoint 9 which is a requirement under Subpart H? It then 10 requires you to initiate a second trial. 11 So, what we are getting told is it was 12 really tough for us to do this first trial. Then 13 we have to ask the question how are you going to 14 get the second trial to confirm that surrogate 15 endpoint when it was so difficult to do it the 16 first time, and will we run into logistical 17 problems because once the drug is on the market 18 people won't be willing to randomize patients 19 anymore? Accelerated approval requires a 20 confirmatory trial, and we would like to have some 21 discussion today about the logistics of doing that. 22 Finally, validation of the surrogate 187 1 requires a meta-analysis correlating all of these 2 studies that we can possibly look at, which we have 3 tried to do at least for some of these diseases, 4 and even if it correlates it still needs to capture 5 the full treatment effect. 6 [Slide] 7 This is one of my favorite quotes. I know 8 John Rex has seen this because I showed this a 9 couple of months ago, but I think it really puts 10 this in place: It is far better to have an 11 approximate answer to the right question which is 12 often vague, than an exact answer to the wrong 13 question which can always be made precise. So, I 14 will stop there and turn it back to you, Jack. 15 Discussion 16 DR. EDWARDS: Thank you very much, John. 17 Those were three very nice presentations. Thank 18 you very much. We are behind schedule. What I 19 think I would like to do is begin the discussion at 20 this point, and then we are going to break for 21 lunch in a few minutes and we will finish the 22 discussion after lunch. So, let us get started, if 188 1 we could. George? 2 DR. DRUSANO: I know that Dr. Powers was 3 at the University of Maryland for a while in the 4 infectious disease section, and he probably 5 remembers Alice Kaplan in the trauma unit. One of 6 my earliest memories was taking care of a trauma 7 patient and in the bad old days they did instant 8 autopsy programs. So, what we found was this 9 patient, with gram-negative pneumonia who got 10 treated with very aggressive antimicrobial therapy. 11 The patient died--it was pseudomonas--because 12 exotoxin A had rotted out his lungs. He went to 13 autopsy; had absolutely sterile lungs. The drug 14 did what it was supposed to do. 15 I recognize some of the issues that I 16 think have been very clearly and nicely presented 17 by Dr. Powers. I guess my view is there is only so 18 much that you can ask out of an anti-infective, 19 which is to kill the bug. Now, it is true 20 macrolides may have some immunomodulatory 21 activities but if you are explaining variants, how 22 much of the variants' outcome is explained by the 189 1 antimicrobial effect versus the immunomodulatory 2 effect? When we talk about the issues of using a 3 clinical endpoint, oftentimes we are missing 4 information. For instance, I think it was of great 5 interest that the neuraminidase inhibitor trial was 6 shown. That was Fred Hayden's study and in that 7 they showed very nicely that the clinical 8 endpoints, how you felt, was driven by basically by 9 how much IL6 you had in your plasma. 10 All of this just goes to one point, which 11 is that there are certain issues with clinical 12 endpoints, one of which is where in the clinical 13 course of your disease do you enter the treatment 14 trial and initiate therapy? Consequently, you may 15 have already had too much exotoxin A and there is 16 nothing other than killing the bug that the drug is 17 going to do. So, that is one thing. 18 Another thing is that you may have a 19 particular virulent strain of bug of an influenza 20 virus that you react to quite well and get a lot of 21 IL6. The best neuraminidase inhibitor in the world 22 is not going to make you feel better. 190 1 So, while there are clear-cut problems 2 with just uncritically accepting a microbiologic 3 endpoint, it is also I think important to look at 4 the issues and the sources of variants in clinical 5 endpoints. We make things gold standard by 6 acclamation. We do not make things gold standard 7 because we know from on high. I think one of the 8 first things that needs to be done in this area is 9 to look at the sources of variability and to make a 10 reasoned judgment about when we measure and what is 11 important to us. 12 DR. EDWARDS: Do you want to take that on 13 or do you want to have lunch first? 14 DR. POWERS: Let me give a parting phrase 15 before we head off to eat. That person that died 16 of pseudomonas pneumonia can't answer the question 17 of did the drug do what he thought it was going to 18 do for him. I think this gets back to a really 19 important point. Let me use an example from 20 October of 2001. You can give all the drugs you 21 want to somebody with inhalational anthrax and you 22 can make the Bacillus anthracis go away and those 191 1 people die. We cannot say those drugs are 2 effective in the treatment of pulmonary 3 inhalational anthrax. 4 So, I think we have to disabuse ourselves 5 of the notion that it is okay to have a drug just 6 to eradicate the bug when everybody dies anyway. I 7 remember when Mike Ryan was teaching me, back at 8 the University of Virginia, he had this one quote I 9 just love, and he said that is like rearranging the 10 deck chairs on the Titanic. You make all this 11 stuff look good--we only do this in the ICU, right? 12 Nobody dies without a sodium that is 140 on the 13 button, but the question is what does that do for 14 patients? I would offer that by acclamation the 15 clinical endpoint is the gold standard, and it is 16 by acclamation of the patients and that is really 17 whom we are trying to help. 18 DR. DRUSANO: I can't let that one pass. 19 Yes, everything that we do is for the patient but 20 that is the wrong issue. What we are trying to 21 answer, I think and, you guys are the regulatory 22 folks so I don't want to get too strong on this-- 192 1 [Laughter] 2 --but the idea is why are we approving an 3 agent? An agent has to be safe and effective, and 4 there is no issue that there will be times--and I 5 think your mention of inhalational anthrax in much 6 the same way that I mentioned the pseudomonas 7 patient on purpose--there is no drug that is ever 8 going to salvage those patients. Okay? If I were 9 sick, I would want the drug to salvage me but that 10 is not the right question I would put forth to you 11 for approval of the drug. The only thing you can 12 ask of the drug is not to salvage the patient but 13 to allow the patient to be salvaged if they have 14 entered treatment at a point in their clinical 15 course when they can be salvaged. That is the 16 issue, at least in my view. As a clinician you 17 always know that there is going to be the guy for 18 whom you do everything right, you sterilize him, 19 but for toxin production, for elaboration of TNF, 20 for whatever reason, it is too late. That is not 21 an indictment of the drug; it is an indictment of 22 the pathogenic process. 193 1 DR. POWERS: Before we break for lunch, 2 and I am glad there is so much discussion, in a 3 randomized trial the hope is that those things 4 would even themselves out between the arms of the 5 trial. What we are doing is we are comparing the 6 two drugs and how they function, and if those 7 people die from all those other reasons in both 8 arms equally, that is what we are looking for. 9 Does randomization work perfectly? No. But does 10 that mean you chuck out the baby with the bath 11 water and stop randomizing? That is probably not a 12 good idea either. So, I think the idea is that we 13 know those things occur but it is in a randomized 14 trial that we are trying to even out all those 15 things except for differences between the drugs. 16 DR. DRUSANO: But the issue is not 17 randomization. The issue is endpoints and what can 18 you expect of the drug. It has nothing to do with 19 randomization. You are absolutely right about 20 that. The question becomes what can you ask of the 21 drug, and except in the rare instances, like 22 macrolides where you have some immunomodulatory 194 1 effect, the majority of the variants of outcome is 2 explained by the impact of the drug administration 3 on the number of bacteria that are there. 4 Sometimes, you are absolutely right, it can even go 5 the wrong way. If you get a burst of endotoxin 6 because you have lysed too many organisms, you may 7 get yourself into trouble. But those are the 8 exceptions rather than the rule, and I think it is 9 worth really examining what question it is that we 10 want to answer for the drug. 11 DR. EDWARDS: At this point we are going 12 to break. We should be back at two o'clock. 13 [Whereupon, at proceedings were recessed 14 for lunch, to resume at 2:00 p.m.] 195 1 A F T E R O O N P R O C E E D I N G S 2 DR. EDWARDS: Dr. Drusano had mentioned 3 some points that Dr. Powers was responding to and 4 we had to end in the middle of that discussion. I 5 had to cut Tom Fleming off at the time we stopped 6 and I apologize Tom for that, and I promised him 7 that I would let him begin this part of the 8 discussion with some comments he wanted to make 9 that were directly related to the interchange 10 between John and Dr. Drusano. So, please, Tom. 11 DR. FLEMING: Thanks, John. I think the 12 presentations this morning were really excellent in 13 laying out a lot of the complexities that we face 14 when we are looking at use of potential surrogate 15 endpoints for assessing clinical endpoints and the 16 effects on clinical endpoints. There is a lot that 17 needs to be said and I would like to say but I 18 would like to just begin by following up on 19 something George was saying before the break. 20 In my words, I think it was something to 21 the effect that a drugs should be evaluated on the 22 basis of whether they achieve the anticipated 196 1 biological effect on the anticipated mechanism. 2 Just to look at a few examples in chronic 3 granulomatous disease, gamma interferon was 4 anticipated to affect bacterial killing superoxide 5 production. It didn't, yet it did provide the 6 clinical benefit of reducing the risk of recurrent 7 serious infections. 8 We heard reference to the setting of post 9 MI arrhythmias which are a known risk factor for 10 sudden deaths, and anti-arrhythmics provide the 11 intended beneficial effect on suppressing 12 arrhythmias and were used by half a million 13 Americans a year until a trial showed that they 14 tripled the death rate. As we had reference this 15 morning to AIDS patients with MIA bacteremia, 16 increasing clarithromycin doses led to the intended 17 five-fold reduction in bacterial load and, yet, 18 also led to a five-fold increase in mortality. 19 So, if I look at these measures and say 20 what were the intended effects, I think we should 21 use the anti-arrhythmic drugs and we should use 22 high doses of clarithromycin and we shouldn't use 197 1 gamma interferon. You would get the opposite 2 conclusions if I were the patient saying what is it 3 that I really care about, and do these agents 4 provide me with what I really care about, in which 5 case I would be used gamma interferon for CGD. I 6 would be using lower doses of clarithromycin and I 7 wouldn't be using ecanide and flecanide to suppress 8 my arrhythmias. 9 So, my sense about this is I think you are 10 absolutely right in Phase 1 and 2 trials. I think 11 the intention of those studies should be to 12 establish plausibility of efficacy in a manner that 13 can be done in a timely, doable trial. And, one of 14 the best ways of doing that is to establish whether 15 or not we achieve the intended biological effect on 16 these markers. Absolutely. 17 But once you get to Phase 3, isn't it 18 important from the patient perspective to 19 ultimately know whether these agents provide the 20 actual tangible clinical benefit? So, in part of 21 the answer is does the drug achieve the targeted 22 biological effect? That is part of it. But the 198 1 other part of it is, is that effect integral to 2 achieving clinical benefit? In a severe sepsis 3 patient with multiple organ failure I might achieve 4 one element of what would be beneficial, but in the 5 myriad of complexities of what this patient is 6 facing that might not translate to any improvement 7 in 28-day survival. 8 The other critical thing is does the drug 9 affect other and often unanticipated or at least 10 unexpected mechanisms that influence clinical 11 outcomes that could be either positive or negative? 12 If positive, I could be underestimating effects. 13 If negative, by releasing too much endotoxin, as 14 you were saying, I could be achieving unintended 15 negative effects. So ultimately I think there is a 16 great deal of appropriateness to stating if we are 17 developing agents we need to try, as best possible, 18 to understand mechanisms of action of the disease 19 process and how it influences the outcome, and what 20 are the types of interventions that could impact 21 those markers and, mediated through that, plausibly 22 achieve clinical benefit. Phase 2 studies are an 199 1 ideal setting to see whether that is the case. 2 But ultimately, if I am going to market 3 this product for clinical care in a non-research 4 setting, I want to know whether it benefits the 5 patients. So, I want to do a trial that 6 establishes direct evidence about clinical benefit, 7 and in that trial obtain additional evidence about 8 whether or not I achieved the intended biological 9 effects. 10 DR. DRUSANO: We agree about 99 percent of 11 the way, Tom. Let's explore the other one percent. 12 The examples that you gave are excellent and clear. 13 The difference is two words, Koch's postulates. 14 The difference between what we have shown before 15 and what we are looking at now is we don't have a 16 perfect understanding of what goes on with 17 anti-infectives and organisms and pathogenesis and 18 the disease process, but we have a deeper 19 understanding of what was going on and the linkages 20 to outcome in this particular circumstance relative 21 to arrhythmias, relative to CGD disease and many of 22 the other examples that you gave. So, I think 200 1 there is a difference in kind, not a difference in 2 degree here. I think anti-infectives really are 3 different in kind. So, that is number one. 4 Number two, nobody, and certainly not me, 5 is saying throw away the clinical endpoint. The 6 issue is not whether or not you throw it away, it 7 is what is the place where you get the cleanest 8 inferences and, for a variety of reasons, many of 9 the times when we do clinical studies people fail 10 with sterile lungs and there is no way in God's 11 green earth is going to salvage that patient. 12 Would you want to be making the mistake of saying 13 that a drug doesn't work for what it was intended 14 to do, which is kill microorganisms, when what it 15 is failing at is its inability to salvage a patient 16 who has gone beyond its ability to do that? 17 So, I think that I would agree 18 whole-heartedly, you do need to look at clinical 19 endpoints because we don't know perfectly well 20 everything. There certainly are unanticipated 21 effects. There certainly are unanticipated 22 toxicities. I think the clarithromycin story is 201 1 certainly one of those. So, nobody is saying throw 2 it away. The issue is how to make the best 3 inferences on whether a drug is doing what you want 4 it to do. There will always be times when we are 5 wrong because that is the human condition. But I 6 can only speak for myself and if I was going to be 7 making the choices for my disease or my family's 8 disease, I would want to know what it does for the 9 bacteria. 10 DR. FLEMING: And I would add that I would 11 ultimately like to know what it does in terms of 12 the clinical conditions that are the ultimate 13 driving factor to my taking the intervention. 14 If I hear what I hear you saying 15 correctly, George, I would agree with it. That is, 16 not all settings are the same and in some settings 17 the level of our understanding about the biological 18 processes through which the disease process 19 influences clinical endpoints and the nature of our 20 interventions and the plausibility that it would 21 have unintended effects, these are critical and the 22 better we understand those issues, the more likely 202 1 it is that the effects on the markers will reliably 2 predict effects on the clinical endpoints. 3 I would only caution that in every 4 clinical setting in which I work with 5 investigators, because of their particular 6 knowledge about that clinical setting, I get a 7 sense from them that they think their setting is 8 different. In their setting, unlike all the other 9 experiences that we have had, we will do better in 10 this particular setting. I would just caution that 11 even if we have a pretty good idea that we have to 12 achieve decolonization in order to prevent serious 13 bacteremia and clinical infection, and we think it 14 is decolonization of the gut, maybe it is, in fact, 15 a broader effect that is necessary. Maybe I am 16 mismeasuring it. Maybe it is a more durable 17 effect. There are many things that make this 18 complicated. Even in a setting where I feel it is 19 perfectly clear, such as mother to child 20 transmission of HIV, what am I trying to do? I am 21 trying to prevent the child from becoming infected 22 with HIV. It is that simple. That is only a 203 1 surrogate for subsequent AIDS-defining events and 2 death but it is that simple. It has to be right, 3 unless what I am doing in a developing country 4 setting is offering formula feeding instead of 5 breast feeding to reduce breast feeding 6 transmission and, thereby introducing impurities 7 from the water supply and introducing infection 8 through that manner. So, even in a simple case 9 like that I can be creating a net effect that isn't 10 what I expect. 11 DR. DRUSANO: Tom, I would agree with 12 that, and I think the issue is you never get it all 13 the way right. You never explain 100 percent of 14 the variance and nobody is saying throw away 15 clinical endpoints. It is how you draw your major 16 inferences on whether the drug is working or not. 17 DR. EDWARDS: David? 18 DR. SHLAES: Thanks. If I could just try 19 and go back to the specific example that was raised 20 by Barry because I think there are a lot of 21 problems there because basically it is current 22 medical practice; there is good data to suggest 204 1 that there is a correlation between sterilization 2 and success of reimplantation of a new prosthetic 3 joint. What I would like to do is go beyond that 4 and take what you said and what Dr. Goldberger said 5 and put them together and ask if one successfully 6 carried out a small trial using bacteriological 7 endpoints with an ultimate follow-up for clinical 8 success in prosthetic joints, would one also then 9 get approval for treatment of osteomyelitis and 10 septic arthritis? Because that would be a way to 11 do a reasonably short trial in a serious disease 12 with the complication of a prosthesis present, but 13 then you might be able to extrapolate to some of 14 these other conditions. I wonder how the FDA would 15 feel about something like that. 16 DR. EDWARDS: Dr. Powers? 17 DR. POWERS: I think this gets to our 18 first question up there, David. That is, in what 19 situations and in what kind of diseases would 20 surrogate endpoints be most useful? That gets back 21 to the Subpart H discussion of serious and 22 life-threatening diseases and biological 205 1 plausibility. But it gets also to that last issue 2 on there and, Barry, you can comment on this, if 3 you are going to accept that surrogate endpoint 4 under Subpart H, much like HIV trials, you have to 5 continue to follow those people and see what 6 happens to them clinically. You can't just wave 7 goodbye to them and say, well, we have accepted 8 this. That may be different for a chronic 9 osteomyelitis model where you could continue to 10 follow those people as opposed to other diseases 11 where a person is cured and that is it. That is 12 why we wanted to get to these discussions. 13 DR. SHLAES: In what Barry presented he 14 pointed out that that is exactly what he would do 15 so you would get the bacteriological result at a 16 certain point, and then there was the follow-up 17 which essentially was the reimplantation X months 18 later, which would be the final result. Or, you 19 could even wait several months beyond the 20 reimplantation for your endpoint. But I think what 21 he proposed was exactly what you are talking about. 22 So, my question is, (a) do you buy it based on the 206 1 data and, (b) if you buy it, would you extrapolate 2 then to some of the other indications which are 3 studies which nobody does anymore which people 4 would like to do. People would like to have 5 indications for treatment of osteomyelitis and 6 septic arthritis and this might be a reasonable way 7 of getting there. That is all I am asking. 8 DR. POWERS: Well, let me put it this way, 9 as Jack had said earlier, we are not setting policy 10 at this meeting; we are just trying to discuss 11 this. So, can't say do we buy it or not at this 12 particular meeting. We are just trying to get 13 these points out on the table to discuss them. 14 The other issue is, and I think maybe we 15 will ask Dr. Fleming to comment on this--you know, 16 I have a bunch of other slides that I didn't 17 include because of time, but just because a 18 surrogate works in one disease and one population 19 does not mean you can extrapolate it to another 20 population. I think that again requires a separate 21 validation. I think we sort of touched on that in 22 the asymptomatic bacteria versus the uncomplicated 207 1 UTI situation. So, I think the question of 2 extrapolating to other people gets to be a sticky 3 one. 4 DR. EDWARDS: Yes, Bill? 5 DR. CRAIG: John, in your analysis where 6 you had the very slow slopes in acute otitis media, 7 how did age impact that? Were those mostly studies 8 done where the average age was older, where 9 placebos worked just as well as the drug and that 10 is why we had the very low values? The one that 11 did start to show some correlation was done in 12 younger kids where, I guess, most of the studies 13 would suggest antibiotics may have a role? 14 DR. POWERS: You have a good point, Bill. 15 That is, is that poor correlation something that 16 has to do with the drug, something that has to do 17 with it being a poor marker, or does it have to do 18 with the natural history of the disease? You saw 19 all the data we have. That is it; that is all I 20 can tell you. Those trials run the range of ages 21 of patients, etc., and one of the things that we 22 want to see, obviously, that I talked about at the 208 1 end is a meta-analysis of effects across all those 2 different populations. You are asking the exact 3 question about can we extrapolate from one 4 population to another and the answer is we don't 5 know because we don't have the data to be able to 6 cut it that fine and to be able to look. Would 7 there be some better correlations in kids under the 8 age of two? Would there be better correlations if 9 we measured the clinical and the microbiologic 10 endpoint at the same time? Well, I showed you one 11 hint that that may be the case but we don't know. 12 So, I think this also gets back to the 13 issue of the FDA's Critical Path. That is, what 14 that Critical Path says is that we need to develop 15 better tools, and this is all about tomorrow, 16 developing tools, and we can't do that. We need 17 your help in order to develop those tools to try to 18 streamline the drug development process. So, we 19 have done about all we can in looking at what is 20 already out there. Is there a better way of doing 21 it? We hope you guys will build a better mouse 22 trap and come in and show it to us. 209 1 DR. EDWARDS: Other questions? George? 2 DR. TALBOT: Obviously, one of the issues 3 with this whole discussion is the fact that there 4 are some extremely substantive and relevant 5 concerns about using surrogate endpoints broadly, 6 as Tom has pointed out and John, related to, for 7 example, host organism response in a certain 8 setting that can't be predicted. I guess, as a 9 broad framework for this and related to your 10 questions, I would urge that progress in this area 11 not be aborted because there can't be agreement on 12 a universal philosophy for using surrogate markers. 13 What I would suggest is that if we could even agree 14 on one setting, such as the prosthetic joint one 15 where it seems reasonable to expect that 16 eradication of the bug would be beneficial and 17 intended, and where it also seems unlikely that 18 there would be substantive host-organism 19 interaction related to inflammatory cascades, and 20 so forth, and where there are also some data--let's 21 go for that and at least try and get movement on 22 that. 210 1 With that in mind, I would say this seems 2 quite logical. There is a precedent actually for 3 this. I believe early or part way through the 4 Synercid program a very similar study design was 5 proposed in terms of prosthetic joint infection 6 using the culture result at the reimplantation or 7 post-antibiotic phase as a clinical surrogate 8 endpoint that would be confirmed ultimately by a 9 late clinical follow-up. So, the idea has been 10 around and I think it really deserves a lot of 11 attention. So, I would propose that that be a 12 specific indication that we try to move forward. 13 The residual question is where would that 14 fit in the context of an entire drug development 15 program? Because there is not only the issue of 16 host organism response specific to an indication, 17 there also is a question of do you have an adequate 18 safety database anyway for all the other things 19 that can go wrong with a drug? So, in thinking 20 about that part of the framework, I would say that 21 you could use, for example, this surrogate endpoint 22 for this indication provided that the size of your 211 1 safety database in other indications is adequate, 2 greater than X thousand patients, 100 patients, or 3 what-have-you. If you meet that criteria and if 4 you demonstrated efficacy in other indications, and 5 if you know you have tissue concentrations, then 6 you could go for this as a reasonable indication 7 for a surrogate marker. It would have to meet all 8 those criteria. 9 DR. GOLDBERGER: I want to agree with 10 George. You wouldn't be thinking in any case of 11 the design that was talked about for the joint 12 prosthesis as a study that would certainly be there 13 just by itself and that would be the entire 14 submission. I think when you think of it, as 15 George just said, in the context of perhaps a 16 couple of other indications where you can see that 17 the drug was successful in treating serious 18 illness, then I think the extrapolation which, 19 truthfully, in the case of the prosthetic joint is 20 that really that big a leap in any case but, even 21 if it were somewhat of a leap, when it is 22 buttressed by this additional data in serious 212 1 illness, I think that is an extrapolation that is 2 reasonable to make. I think it is always important 3 as we start thinking about these issues to think of 4 the package, the several studies that would come in 5 together and how they could naturally reinforce one 6 another and give people confidence that if you use 7 something unusual as an endpoint in one study, you 8 have other data to make you more comfortable with 9 it. 10 DR. EDWARDS: George? 11 DR. DRUSANO: A lot of what I am hearing 12 harkens back to a little bit earlier this morning 13 when people were talking about how much uncertainty 14 can you bear. You know, everything that we have 15 heard today from Dr. Fleming, from the FDA, Dr. 16 Powers--everybody is really centered around how 17 much certainty can we bear; how much uncertainty 18 can we bear. Really, I think it is possible in 19 certain defined circumstances where the severity is 20 high, where it is relatively rare, that you might 21 want to bear a little bit more uncertainty to be 22 able to study this in a reasonable manner and use a 213 1 surrogate endpoint--not every time. If you can 2 study it better and you can get lots of them and 3 get microbiologic and clinical, that is great. 4 But I think we have to focus on where this 5 thing fits in. Tom, I think quite appropriately, 6 said, hey, if you are talking about Phase 1/2, I am 7 your man. It is not too much of a leap from that 8 to say, hey, are we talking about this very severe 9 area where things are relatively rare? It might be 10 worthwhile then to consider what we are talking 11 about in that context. 12 DR. POWERS: I think that is getting back 13 to our questions again. Rather than agreeing is 14 this appropriate for prosthetic joint infection, 15 what we wanted to do was come up with some 16 characteristics of where this would most likely 17 fit. George, you have already enumerated one which 18 is already in Subpart H, and that is serious and 19 life-threatening diseases where perhaps there are 20 not as many therapies out there. So, we already 21 know that under the Subpart H regulations. The 22 question is what other characteristics of a disease 214 1 would people want to see? 2 So, the question I certainly have when I 3 think about this is, you know, it seems to fit more 4 with chronic diseases like HIV where you can keep 5 following the people over time. The question we 6 have run up against is what happens when the people 7 are either cured or dead and you need to go do that 8 second confirmatory trial. If you told us how 9 tough it was to do the first one, how are you going 10 to do the second one to be able to do that 11 confirmation? So, there seems to be a difference 12 between the chronic illnesses where this would seem 13 to apply more easily at least versus the acute 14 illnesses, and do people see any differences 15 between those? 16 DR. DRUSANO: Absolutely. 17 DR. POWERS: Let me ask another question. 18 Bill, you brought up a really good question, and 19 that is are we doing this the best way we can? So, 20 let's take otitis media as an example. Do we want 21 to see another double tympanocentesis trial with a 22 tap at day 4-6 and a clinical endpoint getting 215 1 measured 28 days out? If not, how are we going to 2 get the information that we need to do this in a 3 better way? 4 DR. CRAIG: Well, as I said before I think 5 most of the data, at least that I have seen, 6 suggests that children under two are the ones where 7 antibiotics have their role. So, clearly, I think 8 if you wanted to try and show the clinical success 9 and possibly differences from so-called approved 10 drugs that are out there, that would be the 11 population to look at. If you keep adding in older 12 kids for which placebos work just as well you 13 dilute out the effect that you are going to see and 14 you end up with equivalence when really it might 15 have been a superior drug. 16 Now, you probably can, possibly by using 17 double taps and then still looking at clinical, be 18 able to reduce your numbers because you are down 19 where it can be complete eradication to where you 20 don't get any eradication. On the other hand, you 21 know the placebos will bring it on up to 60 percent 22 response, 70 percent response even when the 216 1 organism is not eliminated. As a result, you are 2 looking at differences that are much smaller and 3 much harder to see differences in. 4 So, I think you need to get to the 5 patients that really need the antibiotics. I think 6 that is why we have a problem with AECB, that is, 7 clearly identifying who those patients are. We 8 think purulence in the sputum is very important but 9 that still doesn't necessarily identify those from 10 the data that is out there right now. So, I think 11 that is a much harder disease to look at. 12 Sinusitis--I don't know how right now to 13 be able to identify the people for which 14 antibiotics are really going to work. Hopefully, 15 some of the tap studies that they are doing with 16 catheters, in that situation, will give us more 17 data on the efficacy of antibiotics and let us know 18 how much spontaneous response occurs because that 19 is the other part that you need in the equation to 20 be able to look at these things rationally to know, 21 if someone is given a placebo, how much of the 22 organism is going to be eliminated, and those 217 1 things. We have that kind of data for otitis. 2 There is not a lot of reason for me to think that 3 it is going to be markedly different in sinusitis 4 but I think we need the data to be sure. 5 DR. POWERS: It sounds like what you are 6 saying is that we need two pieces of information. 7 We need to know how these people would do with 8 placebo from both the bacteriologic and the 9 clinical point of view so we can do that 10 correlation in the people who don't get drugs, and 11 that would seem to apply to otitis, AECB and to 12 sinusitis as well, in all three situations. 13 DR. CRAIG: Although I think I would have 14 difficulty doing that based on current data of 15 doing placebo in kids under the age of two. I 16 really do feel that that is a group where 17 antibiotics are needed and we would probably have 18 difficulty getting that by our ethical boards. 19 DR. POWERS: Well, at least it is the 20 population that really has the disease. 21 DR. CRAIG: Yes. 22 DR. POWERS: I guess that is part of the 218 1 other issue. When we looked through this 2 information, it seems to be is it that those kids 3 do worse or do they actually just have the illness 4 as opposed to the kids above the age of two who may 5 be the kids who don't really have it. 6 DR. EDWARDS: George? 7 DR. DRUSANO: Dr. Powers, I would just 8 like to kind of reflect a question back at you. I 9 mean, part of what we are talking about is the 10 difference between the clinical and microbiological 11 outcome, and a lot of this we just don't understand 12 well enough. For instance, the oseltamivir study 13 was a key one to have shown, in a sense, that they 14 did measure IL6 area under the curve. I guess I 15 would ask the agency suppose I was able to do a 16 study in kids less than two years of age in otitis 17 media, and we were looking at a beta-lactam or 18 quinolone and then a macrolide, and we were able to 19 look every day at a clinical score and then we were 20 able to not do the same kids but randomized kids to 21 get tapped at day 4, day 5, day 6 and day 7 so we 22 could look at time to sterilization, and if we 219 1 could explain all of the clinical stuff on the 2 basis of damping down IL6, or something like that, 3 how would the agency interpret that relative to 4 something like, you know, we killed all the bugs 5 and over here they got better earlier because not 6 all the bugs were gone but they got rid of the IL6? 7 How would those things be balanced in an 8 evaluation? 9 DR. POWERS: I think, George, what you are 10 saying is validating IL6 as an endpoint, if I am 11 interpreting what you are saying correctly, and 12 that is what we are asking people to do. The 13 question in short-term, self-resolving acute 14 illness is what would have happened if those people 15 got nothing? I think that is a lot tougher 16 question than saying how would we interpret this in 17 a serious and life-threatening disease where we 18 would not expect people to get better 19 spontaneously? So, in short-term disease there is 20 an added complexity to that that goes in there. 21 But I am glad you brought up oseltamivir. 22 The primary endpoint in those trials was not IL6. 220 1 The primary endpoint was time to the patient 2 getting better. But what the company did in that 3 case was they actually do a patient-reported 4 outcome tool that they validated ahead of time 5 where patients filled out information twice a day. 6 Without that you never would have been able to show 7 that or do that correlation with IL6. I think that 8 is one of the pieces of the puzzle that is missing 9 here. It isn't easy sometimes to measure these 10 clinical endpoints in a valid way. If you don't 11 have a valid outcome measure tool for measuring 12 clinical outcomes the analogy I use is you took a 13 piece of paper and you put lines on it and you 14 called it a ruler, even though it may not be 15 measuring anything that you can be clear on what 16 you got at the end of the day. People have done 17 this before in terms of measuring patient-reported 18 outcomes. In fact, I was just looking through this 19 in Rosenfeld and Kline's book on evidence-based 20 otitis media. I was surprised to find an entire 21 chapter on patient-reported outcomes in acute 22 otitis media. 221 1 So, what is clear to us is this idea, 2 again going back to the Critical Path, that we need 3 to develop better tools here on how to measure 4 clinical endpoints. Because it is tough doesn't 5 mean you can just throw away the clinical endpoint. 6 We still need to develop a good way to actually do 7 this. 8 DR. EDWARDS: John? 9 DR. BRADLEY: In your discussion prior to 10 lunch you were making correlations between 11 microbiologic outcomes and clinical outcomes. To 12 go back to otitis because it is such a messy 13 infection to look at, one of the reasons that we 14 started doing double tap studies or single tap 15 studies is that the clinical diagnosis of otitis is 16 extremely difficult and you won't get a lot of 17 agreement between clinicians as to what is bona 18 fide otitis or not. Viral mediated disease gets 19 you something that looks very close to bacterial 20 otitis. So, getting the actual microbiologic 21 information has been critical. Likewise, in 22 following up to see if you got a microbiologic 222 1 effect of the antibiotic, the second tap has turned 2 out to be very helpful. 3 Clearly, there are kids who will stay 4 clinically symptomatic despite sterilization, 5 probably because of the viral disease that set them 6 up for their otitis in the first place. We all 7 recognize that. And, there are a few children who 8 will get better, as you commented, despite having 9 persisting organisms in the middle ear. 10 For each illness there is the spontaneous 11 cure rate and I think that if you develop 12 guidelines which include a kappa value that is 13 appropriate for that particular infection entity 14 you can take the information regarding the 15 discontinuity of a micro. and clinical endpoint and 16 build it into that particular illness based on data 17 that has been collected. So, the kappa for 18 meningitis would be different than the kappa for 19 otitis. 20 DR. POWERS: Right, John. You are 21 bringing up exactly what we are saying. To be able 22 to use those--you are again talking about what 223 1 George was saying, validating the endpoint first. 2 DR. BRADLEY: Right. 3 DR. POWERS: And there could be some 4 issues in validating that endpoint. We already 5 talked about it. Is it a problem with how we 6 measure the surrogate? Is it a problem with how we 7 measure the clinical outcome? One of the things I 8 didn't get to talk about is what was the clinical 9 outcome that we compared it to in otitis. There 10 may be some problems with how that clinical outcome 11 is measured. 12 But there are two points you brought up 13 that I wanted to get at. First, the major 14 discordant area in acute otitis media is not kids 15 with prolonged viral illness who still remain ill. 16 There are very few of them. The major discordant 17 area is the huge number of kids that still have the 18 bug that got better anyway. I think it is 19 important to understand that because when we talk 20 about measuring clinical outcomes in otitis, that 21 is what people say to us--well, but if they are 22 still sick it is because of the viral illness. 224 1 That is not the cell where the problem is. The 2 cell in that 2 X 2 where the problem is everybody 3 is better regardless of whether the bug is gone or 4 not. 5 The second issue--you brought up two 6 issues about it. One was making a microbiologic 7 diagnosis at entry versus using it as an endpoint, 8 and those are very different. As I was saying, a 9 biomarker for diagnosis is different than using it 10 for an endpoint. The July 2002 advisory committee 11 on otitis voted unanimously that we ought to be 12 using a baseline tympanocentesis to get these kids 13 into the trial. What remains at issue now is how 14 useful is microbiologic information to us as the 15 sole measure of efficacy? We are not saying it is 16 not important as a part of the mix, but as the sole 17 measure of efficacy in these trials. 18 DR. CRAIG: John, all the data that you 19 showed was not double tap. 20 DR. POWERS: Yes, it was. There are 12 21 double tap studies in the literature. All of those 22 that I showed you were double tap trials comparing 225 1 the bacteriologic outcomes at day 4-6 with clinical 2 outcomes, usually those clinical outcomes being out 3 sometime beyond day 10, except for the one trial I 4 showed you where it was day 4-6. 5 DR. DRUSANO: John, in the same vein, in 6 the discordant cell--and I know nothing about those 7 trials--was there any quantitation? I mean, there 8 is a big difference between saying the bug is there 9 or the bug is not there as opposed to saying 10 6 10 versus 4/ml or something. 11 DR. POWERS: That is a great point. There 12 is no quantitation. That is why I was asking you 13 guys the question of is there a better way to do 14 these? What we are seeing is these correlations 15 don't look so great the way we are doing it. So, 16 there are a couple of pieces of the puzzle that fit 17 here. Can we measure the surrogate in a better way 18 or a better time? Can we measure the clinical 19 outcome in a better way or a better time to be able 20 to put these together? 21 What I am trying to articulate to you guys 22 is we can't do that. We can only review what we 226 1 are handed. So, to be able to validate these 2 surrogates we need somebody out there, either on 3 the industry side or the academic side, to think 4 about this; realize that maybe we can improve upon 5 the way we are doing it and actually try to look at 6 those differences. 7 DR. DRUSANO: The other question is who 8 should be validating endpoints? I mean, is there a 9 place here--maybe not the agency as a regulatory 10 body but maybe NIH as a governmental agency to see 11 here is a standard way. You know, we are going to 12 invest not in drug A but we are going to invest in 13 learning how to have an endpoint for thus-and-such 14 a disease amongst anti-infectives. It seems to me 15 if you are having sponsors do this all the time, it 16 is learning and throwing it away because then the 17 next sponsor maybe doesn't do it. But if we have 18 something that is more general where there is an 19 outside body that does not have an interest, or as 20 much of an interest, it may be something that can 21 be done once well to everybody's satisfaction, and 22 then picked up so that the actual level of the 227 1 submissions gets better. 2 DR. EDWARDS: George, I need a little 3 clarification on that. You are asking the question 4 who should validate the endpoints. Are you asking 5 that question in reference to clinical endpoints 6 only or both? 7 DR. DRUSANO: Both. 8 DR. EDWARDS: All right. Dennis? 9 DR. DIXON: Certainly the NIH is 10 interested in that. We are approaching that in a 11 number of ways, through both investigator-initiated 12 approaches and through our contract mechanisms. 13 One of the things we try to bring into our clinical 14 trials is incorporation of diagnostic and endpoint 15 assessments. We even have free-standing contracts, 16 such as invasive aspergillosis where the focus is 17 on developing the diagnostic. We did try and put 18 this in context, as a couple of people have said 19 today, and not separate from drugs the importance 20 of a diagnostic to define the illness early enough 21 so you can do treatment trials versus empiric and 22 prophylaxis trials and so you could also do 228 1 prevention studies. So, we tried to put that whole 2 aspect into perspective. Diagnosis, prevention and 3 treatment is the triad that we try and fold into 4 just about all of our approaches. 5 DR. POWERS: Dennis, could I ask you a 6 follow-up question because you and I have talked 7 about this before a couple of times? That is, what 8 would the level of interest be? It seems to me 9 like a whole third of the NIH budget now is for 10 bioterrorism-related issues. But what would the 11 interest be for common diseases like this, like 12 acute exacerbations of chronic bronchitis and acute 13 otitis media, and doing some of this work on what 14 seem to be the more mundane but, yet, the reasons 15 why people actually get antimicrobials for the vast 16 majority of folks? 17 DR. DIXON: Well, we do have investigators 18 submitting applications for just those kinds of 19 issues. 20 DR. POWERS: So, if this group were to put 21 together those kinds of studies and submitted them 22 to you, that would be something you guys would be 229 1 interested in taking a look at? 2 DR. DIXON: Yes, within the constraints of 3 the mechanisms and I am sure that is going to be an 4 issue given the 500,000 dollar direct costs 5 limitation for many of the mechanisms across the 6 NIH. But we have infrastructures in place, such as 7 the clinical trials networks, that have dedicated 8 resources to which these questions can be brought. 9 DR. DRUSANO: Along those lines, can 10 PhARMA contribute to the funding of the trials so 11 they are supporting something that is going to 12 benefit everybody in PhARMA? Is there a mechanism 13 for that? 14 DR. DIXON: Well, they do in a number of 15 ways, George. That is one of the examples that 16 some of the people in this room know through the 17 bacteriology and mycology study group. The issue 18 of resources which came up in John Rex' discussion 19 is that the resources are going to be limiting to 20 federal agencies as well as to families and to 21 companies and the balancing priorities. We have 22 been able to manage that through the bacteriology 230 1 and mycology study group by funding the 2 infrastructure and PhARMA bringing the product 3 forward and paying the per capita. Those trials, 4 by and large, do not go forward unless there is 5 industrial partnership. 6 DR. FLEMING: The question was asked who 7 should participate in validation. I would hope 8 that everybody would be involved in validation. I 9 think it is not a simple thing of we are going to 10 take on this one project and when this is completed 11 we will have validation. You have given, John, 12 some examples in acute otitis media and the 13 clearance of the bug, and the problem there was 14 some people who didn't have clearance, a number of 15 them still had clinical success. 16 In essence, the totality of the mechanisms 17 of what influence outcomes aren't being fully 18 captured simply by at least our quantification of 19 clearance of the bug. We could be underestimating 20 the overall ability to achieve success in acute 21 otitis media. We could be overestimating it in 22 inhalation anthrax where clearing the bug isn't 231 1 going to be enough. 2 In essence, it takes a much richer 3 understanding and the arguments that you have been 4 giving about looking at the kappas and the 5 correlations, that is a step. We might call it a 6 necessary but not sufficient step. A classic 7 example I like to give, and I have referred to this 8 already, is mother to child transmission of HIV. 9 We know, in fact, that the essence of what matters 10 there is viral load and there is an extremely 11 strong correlation. In fact though, women who have 12 higher viral loads have lower CD4 counts, in 13 addition to also having a much higher risk of 14 transmission. If you look at her CD4 count, there 15 is great correlation with her risk of transmission 16 of HIV, but it is not the causal mechanism. If I 17 intervene before labor and delivery with IL2 I can 18 spike her CD4 240 cells, 250 cells, and it is not 19 going to do anything about transmission. If I 20 intervene with ART and reduce her viral load I may 21 not influence her CD4 count very much but it is 22 going to have a lot to do with transmission. It is 232 1 an example of John's drug A versus drug B, or 2 intervention A versus intervention B. IL2 is going 3 to look a whole lot better than ART on CD4 but it 4 is a whole lot worse in achieving the intended 5 benefit. 6 I have to understand the causal mechanisms 7 here. It is a myriad of issues here. I need to 8 understand the correlations. That establishes 9 plausibility. That gets my foot in the door. If I 10 have a strong correlate it is at least a potential 11 surrogate. To go beyond that I need three things. 12 I need to understand as clearly as possible what 13 are the mechanisms by which the disease process 14 influences the outcome. And, is my specific 15 measure, if it is a microbiological measure, is it 16 capturing the essence? Many times you might be 17 right but you are not measuring it the right way. 18 CD4 count is not a very good surrogate in HIV, not 19 because the essence of the status of the immune 20 system isn't highly relevant to whether you are at 21 risk for AIDS-defining events, but it is a noisy 22 measure. 233 1 Maybe in the acute otitis media setting 2 what we really need is not just a dichotomization, 3 clearance or not, but in essence a much better 4 sense of the nature of the clearance. 5 Secondly, we have to understand quite 6 clearly the plausibility that an intervention that 7 is sufficiently potent to achieve its intended 8 mechanism--how can we be so confident that it 9 won't, in fact, induce unintended mechanisms? 10 So, those are tall orders and it is in 11 essence, in my view, the reason that, in truth, 12 there are very few validated surrogates. However, 13 there is a continuum. How well we understand that 14 influences the likelihood of reliability. 15 Ultimately, from a data perspective, to validate a 16 surrogate should not in essence be based on whether 17 the pregnant woman with HIV, the higher her CD4 18 count, the lower her risk of transmission--that 19 correlation is great; my kappa would be terrific, 20 and in no way is that a valid surrogate because it 21 is simply correlated with what is truly causal. 22 What I really need is an array of studies 234 1 that will establish what is the level of effect on 2 the marker and what is the level of effect on the 3 clinical endpoint to show that there is, in fact, a 4 prediction or a correlation in those measures. 5 That is why raised the question of who should be 6 involved in validation of surrogates, we all have 7 to be involved because the ultimate validation of 8 the surrogate is going to depend upon our increased 9 understanding of the biological processes of the 10 disease process and the mechanisms of action, 11 unintended as well as intended, of the 12 interventions, and then having an aggregation of 13 trials that measure effects on both to ultimately 14 see whether or not we have the level of prediction 15 that we are hoping. 16 DR. DRUSANO: First of all, what I meant, 17 Tom, was not to throw all out the information 18 because I guess I had always intended that all of 19 us should do this with all the points of view that 20 we bring to it. But what I don't want to see is, 21 you know, drug company A validates this thing, 22 doesn't really publish it well enough and then drug 235 1 company B comes in and does the same stupid trial 2 and doesn't use the validated measure. My view 3 that I was trying to get across is that there 4 should be a funding mechanism, and I was glad to 5 hear Dennis say what he said--a have a funding 6 mechanism where the FDA can have input, NIH has 7 input, IDSA has input, that all the people that 8 really need to be there are at the table and the 9 thing starts the validation process. And, it may 10 take a year, two years, three years. Viral load 11 wasn't validated in a year; it took a decade. But 12 when we validate it, hopefully, then it is out 13 there for everybody to see and every company picks 14 up that way of doing it. Again, the level of the 15 trials rise and we don't have these issues. 16 DR. POWERS: George, you make an excellent 17 point. If you read the oseltamivir trial, it says 18 in there we used a validated patient-reported 19 outcome tool; data on file at the company. 20 Therefore, no one else can use that validated tool. 21 We have thought about that. We are trying to move 22 forward with trying to at least get off the ground 236 1 patient-reported outcomes in sinusitis and AECB 2 because we want to get these into the public domain 3 where everybody can use them, and it would be great 4 to have partners to be able to do that. 5 DR. EDWARDS: Yes? 6 DR. SCHENTAG: Jerry Schentag, Buffalo. 7 Tom, you have given us a very comprehensive 8 description of what is necessary to validate a 9 surrogate like microbiologic outcome for example. 10 By the way, I believe it applies to clinical as 11 well. The problem, of course, is that that is a 12 discouragingly long list that requires a long 13 evolution of the process and it occurs without 14 consideration perhaps of the fact that you could 15 very easily validate a microbiologic surrogate if 16 you just simply did all of those studies in an 17 animal, if you infected it and treated it, and then 18 you have all that information available as prior 19 data that goes into that. In essence, that is what 20 we have done with antibiotics over the years. We 21 have pretty much been through that whole business 22 of validating the microbiologic endpoint versus 237 1 drug, versus concentration, separated out with a 2 resistant organism will kill you if you don't treat 3 it with the right thing, etc., etc., etc. 4 What I am trying to figure out is if, in 5 fact, any of that is useful here. I mean, there is 6 a big disconnect right now between that because if 7 we have to start over, assuming a micro. endpoint 8 is potentially not useful and even clinical 9 endpoint is not useful, which has been our course 10 over the last five years or so, maybe we do a 11 disservice to all our hard work that has gone into 12 validating these endpoints in other systems besides 13 human. How do we use it? How do we bring it 14 together? 15 DR. FLEMING: If I understand the essence, 16 you are saying if we do carry out studies in 17 animals, for example, where we can look in a 18 controlled manner at what is the effect on the 19 microbiologic marker and what is the effect on the 20 intended clinical endpoint-- 21 DR. SCHENTAG: Correct. 22 DR. FLEMING: --I do view those to be 238 1 useful supportive evidence. The question is are 2 they the ultimate answer? Well, the answer to that 3 is, again, the validation of the surrogate depends 4 so heavily on understanding the mechanisms by which 5 the disease process influences the clinical 6 outcomes and understanding the intended and 7 unintended effects of the intervention. The 8 question is are those mechanisms the same in 9 animals and humans? Are the effects of the 10 intervention the same in animals and humans? To 11 the extent that they are, this is going to be very 12 valuable information. How uncertain are we though? 13 In many cases we are uncertain--certain enough or 14 confident enough to appropriately use this to enter 15 into studies in humans and then to through the 16 Phase 1/2/3 steps. Can it in some way though 17 actually be used in a more reliable manner than 18 that? My answer to that is only if we can provide 19 persuasive arguments that these mechanisms in 20 animals are reliably representing the mechanisms in 21 humans. 22 DR. SCHENTAG: Well, we do have situations 239 1 where we have pretty close concordance between 2 micro. outcome and clinical outcome, like 3 life-threatening infections, resistant pathogens, 4 that sort of thing. I submit that maybe that is 5 the place where we might argue that it is easiest 6 to make the leap between what we have learned in 7 animals and what we can learn in humans. I agree 8 completely with what has been said, that in 9 diseases where it is not clear that the bacteria is 10 the cause of the disease in the first place, like 11 bronchitis for example or sinusitis, it is almost 12 impossible to do a validation. That is because I 13 think the clinical outcome is just as bad or as 14 weak a surrogate as the micro. outcome is in that 15 situation. We don't know. We don't know if it 16 causes the disease in some ways, so how are we 17 going to be sure clinical outcome is the result of 18 an antibiotic? 19 John made that point earlier. He said the 20 micro. outcome is always a surrogate. I think 21 sometimes the clinical outcome is a surrogate and 22 the micro. is the primary endpoint. You know, it 240 1 is going to depend on the causality that you can 2 link between your antibiotic effect and your 3 disease resolution, and we have that in some 4 diseases and not others, but that is the place 5 where you can use animal data as supportive 6 perhaps. I don't know; I am just fishing here. 7 DR. POWERS: I think what you are getting 8 here is awfully tough to ask a chinchilla if it has 9 ear pain, or to ask the mouse whether it is 10 coughing or not. So, I think there are some 11 limitations to asking some of those symptomatic 12 resolution questions in an animal model. 13 But let me go back again for a second 14 because we seem to keep getting confused on this 15 point. If you look at what we showed up there, the 16 clinical endpoint is how a patient feels, functions 17 or survives. That is the gold standard. You can't 18 say when it doesn't function well, then the micro. 19 becomes the gold standard. If the clinical outcome 20 doesn't function well because of some issue it 21 means we need to find a better way to measure the 22 clinical outcome. It does not automatically 241 1 validate the microbiological outcome. I think that 2 is a real important point to get across. 3 DR. EDWARDS: Dennis, I think yours is 4 going to be the last comment before we go on. 5 DR. DIXON: Thank you, Jack. I would just 6 like to come back to the example of aspergillosis 7 because it built on the example that John Powers 8 made earlier about the workshop that was held with 9 FDA and NIAID a year ago on endpoints in clinical 10 trials and the importance of fever as a marker for 11 aspergillosis. After a long discussion of that, I 12 am not sure we answered the value of fever and it 13 was still undetermined. But what came out of that 14 is that we need a better way to identify the 15 presence of fungus, which is what we know will be 16 involved in the outcome or at least the instigation 17 of the disease for aspergillosis. 18 So, one of the ways you can approach that 19 is through animal models, and we dedicate resources 20 specifically to diagnostics for aspergillosis, not 21 just yes and no but the degree of infection. So, 22 that is set up to have a model of a small animal 242 1 and a large animal, something like a mouse and 2 something like a rabbit, for both quantitative and 3 qualitative tracking of the infection so that it 4 can be brought back into the clinical setting, put 5 into one of trial infrastructures to see if the 6 correlation pertains from mouse to person. 7 We would like to be able to do that for 8 every single disease, but I am pretty sure that we 9 can't all at one time, but there are approaches in 10 using tools in animal models and refining tools 11 from the basic research forefront that can make 12 their way through to help with answering some of 13 these questions. 14 DR. EDWARDS: I am going to take the 15 prerogative here to tell you how we are going to 16 proceed, that is, I have two people who urgently 17 want to add comments before we conclude and I am 18 going to give them a total of five minutes between 19 the two, Tom and George. Then we are going to go 20 into the next session. We have intentionally 21 decided not to try to end at 3:30 today. We are 22 planning to end sometime around 4:30. So, we are 243 1 not just going to go on and on; we are going to 2 have a fixed ending point that, hopefully, is going 3 to be acceptable. 4 This is such an important area, the whole 5 area of the surrogate markers, and we have so few 6 opportunities to get together to discuss these 7 kinds of issues that we really felt it was 8 critically important that we try to get as far as 9 we possibly can. So, let me ask Tom and then 10 George to just very briefly comment and then we 11 will go on. 12 DR. FLEMING: And I will try to keep it 13 from one to two minutes of those five minutes. We 14 didn't answer question three, and I wanted to at 15 least allude to it. What are the issues for 16 obtaining validation of a surrogate necessary for 17 fulfillment of Subpart H? 18 It is important just to remember that 19 Subpart H is intended as a way to get earlier 20 access in serious disease conditions to 21 interventions that are promising. The marker that 22 we use doesn't have to be established to be 244 1 reliably predicting clinical endpoints, clinical 2 effects. That is what we have been talking about, 3 how complicated that is. It just has to be 4 reasonably likely, the concept being if it is, get 5 the agent out there while you then validate. You 6 are still held to having to show the same level of 7 evidence to validate. 8 It is a little sobering though to see how 9 that process is working, and if you are interested 10 I would encourage you to go back and look at the 11 transcript from the March 12th and 13th, 2002 12 oncology drugs advisory committee where they 13 reviewed 8 of the 12 accelerated approvals that 14 were the first 12 that occurred in oncology from 15 1995 to 2000. The average time between the 16 accelerated approval and projected completion of a 17 validation study is 10 years. Why? It is just 18 really hard to randomize patients to controlled 19 studies once you have given the agent an 20 accelerated approval with the impression that it is 21 effective. 22 Secondly, one of the reasons it is great 245 1 to work with industry is because of their sense of 2 urgency. Is that the same after they have an 3 accelerated approval? I would suggest that it 4 appears that it isn't. 5 Thirdly, in 3 of those 8 cases the first 6 validation study was done and it was quite negative 7 but the agent is still out there. Why? Because it 8 is extraordinarily complicated to pull the agent. 9 So, there is, in fact, I think a great 10 deal of uncertainty about what do you do when you 11 get an accelerated approval, do the validation 12 study and it is not persuasive. In essence, if you 13 are going to leave the agent out there after a 14 fairly non-persuasive validation study, then I am 15 arguing that accelerated approval is tantamount to 16 full approval. If it is, then why should we be 17 using criteria that are less rigorous than full 18 approval criteria? 19 So, the motivation for Subpart H is 20 compelling. The realities of the complexities of 21 being able to complete validation studies in timely 22 ways, and taking proper action when they are done 246 1 is highly complicated. 2 DR. EDWARDS: Thank you. George? 3 DR. TALBOT: I just wanted to raise the 4 general question for the agency and all of us to 5 consider as to whether these discussions are not as 6 fruitful as they might be because we are locked 7 into a terminology which is no longer relevant 8 completely. I am thinking about the indications. 9 We always talk about acute otitis media; we talk 10 about acute bacterial sinusitis; we talk about 11 acute exacerbations of chronic bronchitis as though 12 these were unitary diseases. We know they are not 13 and people have mentioned that, but maybe it would 14 be easier to move forward if there were more 15 precision in exactly what it is we think needs to 16 be treated. The corollary to that could be that it 17 would be easier to define the potential benefit of 18 a surrogate marker. For example, in acute 19 bacterial sinusitis one could say that if one 20 enrolls patients on day one the bugs are all going 21 to go away probably in 99.5 percent of those 22 patients and the patients are going to get better 247 1 if there were bugs to begin with. So what? 2 On the other hand, if you take patients 3 who are culture positive on day one and are still 4 culture positive and symptomatic at day seven, and 5 then treat those, I would submit to you that you 6 could perhaps prove more readily that in that 7 instance the eradication of the bug would prove to 8 be a very good surrogate marker for subsequent 9 clinical improvement. 10 In conclusion, let's be sure that we don't 11 lump when we need to split. Let's think about how 12 that should be reflected in the indications that 13 are open for regulatory approval, and let's think 14 in these studies about how to define the specific 15 subpopulations where you could expect benefit for 16 getting rid of the bugs, as George said, and where 17 you might find a valid surrogate marker. 18 DR. POWERS: I think Dennis brought that 19 up. We need better diagnoses of some of these 20 things and then selecting the correct patient 21 population is a big issue. That was sort of 22 three-quarters of what we talked about at the 248 1 October, 2003 advisory committee. But how would 2 you find that right population? You have to do the 3 study. 4 DR. TALBOT: You do, but just in terms of 5 guidance and so forth, you need to think about that 6 catch-all indication of AECB. Maybe there should 7 be some brain-storming about how to better define 8 some of those things. It is not an issue for 9 complicated UTI perhaps or for acute bacterial 10 meningitis, but for these other things I think it 11 is really germane. 12 DR. POWERS: Right. You don't have that 13 with HIV because you know the person has HIV before 14 you treat them, and we do have this issue with a 15 lot of acute bacterial diseases. 16 DR. EDWARDS: Excellent points. We are 17 going to take a ten-minute break now and then we 18 will get started and, again, we are shooting for 19 4:30. 20 [Brief recess] 21 DR. EDWARDS: We will ask Tim Henkel to 22 begin the discussions with his presentation on 249 1 bloodstream infections. Tim, I think we are ready 2 to go. 3 IV. Discussions about Bacteremia as an 4 Indication/Issues with Clinical Trials 5 with Endocarditis - Industry 6 DR. HENKEL: Thank you, Jack. 7 [Slide] 8 I am going to do something a little bit 9 different than we have done before. I am going to 10 talk about one particular study, as you can see 11 from the slide, a Phase 2, randomized, controlled 12 trial of dalbavancin versus vancomycin for 13 catheter-related bloodstream infections. George 14 Talbot has stepped out of the room, but I think we 15 might subtitle this splitting when we should be 16 lumping instead of vice versa. 17 I want to talk just briefly about a little 18 bit of the regulatory background, the guidance 19 documents in operation when we started this study; 20 just a bit of history of the indication. I will do 21 that fairly briefly because I think David will 22 touch on that later. Then I want to spend some 250 1 time talking about the design of the study, the 2 criteria used for getting patients in, and then at 3 least operationally the results, what happened 4 these criteria to try to conduct a clinical trial 5 today. What were some of the limitations; what 6 were some of the challenges? What I won't do is 7 talk about safety and efficacy results of the 8 study. Those will be presented at a scientific 9 meeting in the coming weeks. 10 [Slide] 11 So, let's go back a little over a decade. 12 In 1993 the Anti-Infective Drugs Advisory Committee 13 recommended elimination of the bacteremic sepsis 14 indication. In 1998 AIDAC had a discussion of 15 catheter-related bloodstream infections as an 16 indication. But draft guidance followed in 1999. 17 As of 2004, no drug has been approved for 18 catheter-related bloodstream infections. So, let 19 me review the entirety of the published literature 20 controlled trials for CR-BSI. 21 [Slide] 22 It is a single study, Sam Raad. It is a 251 1 study with Synercid versus vancomycin, published in 2 The European Journal of Clinical Microbiology and 3 Infectious Diseases. It was actually initiated I 4 think following the advisory committee but before 5 the issuance of the daft guidance. It is a study 6 of 39 patients, upon analysis, 16 of whom did not 7 have gram-positive bacteremia. So, there is not a 8 great deal of information out there. But this was 9 the context in which we started the study and put 10 this design together. We certainly did have a 11 number of discussions with the agency about the 12 design because of that background, even though it 13 was a Phase 2 study. 14 [Slide] 15 Let me go through a few of the pertinent 16 features of the design. It was Phase 2, as I said. 17 It is randomized, controlled, open-label with 18 vancomycin as a comparator at the standard dose. 19 We looked at both clinical and microbiological 20 entry criteria for the study which I will describe 21 a little more later. 22 The patients, in order to be entered--the 252 1 idea was catheter-related bloodstream infection 2 but, of course, upon entry you don't know all the 3 information you need in order to make that 4 determination so I will come back to that in a 5 minute. The primary endpoint of the study was the 6 global response at follow-up, that is, the combined 7 clinical and microbiological response. So, you had 8 to be a success both clinically and 9 microbiologically to be a success in the study. 10 This was actually in the modified intent-to-treat 11 population. The sample size initially planned was 12 about 60 patients per group, and we planned 13 descriptive statistics with 95 pc confidence 14 intervals in this Phase 2 study. 15 [Slide] 16 In order to be entered into the study a 17 patient had to have a documented gram-positive 18 bacteremia, that is, at least a single positive 19 culture in order to be entered, or empiric 20 enrollment based on signs and symptoms such as 21 elevated temperature, hypothermia, an elevated 22 white blood cell count or leukopenia, tachycardia, 253 1 tachypnea or transient hypotension. In fact, very 2 few patients were enrolled on empiric grounds and 3 most were enrolled with at least one positive blood 4 culture for a gram-positive organism. 5 [Slide] 6 In terms of exclusion, patients could not 7 have received more than 24 hours of prior 8 antibiotic therapy that was effective against a 9 gram-positive organism isolated. So, those were 10 reviewed on a case-by-case basis. They could not 11 have an alternate focus of infection identified at 12 the time of enrollment in the study. They couldn't 13 have had recent Staph. aureus bacteremia from a 14 source other than a central venous catheter. They 15 had to be patients who, it was anticipated, could 16 be treated with two weeks or less of antibiotic 17 therapy. The creatinine clearance needed to be 18 greater than 50 ml/min simply because appropriate 19 studies for adjustment doses hadn't been completed 20 at the time this Phase 2 study was started. They 21 couldn't be severely neutropenic, that is, an ANC 22 less than 100 for more than 72 hours. they 254 1 couldn't be on chronic immunosuppressive drugs and, 2 of course, you couldn't have documented resistance 3 to either of the study drugs since it was a 4 randomized trial. 5 [Slide] 6 There are a number of microbiological 7 methods in the draft guidance which were 8 implemented in this study. We tried to look at 9 catheter cultures where they were available. We 10 tried to look at time to positivity of catheter 11 cultures versus peripheral cultures, again where 12 available. We looked at insertion site exudate 13 cultures if they were present. Then, in addition, 14 for certain organisms, which I will elaborate on, 15 we looked at the antibiograms as well as pulse 16 field gel electrophoresis to confirm identity of 17 pairs. 18 [Slide] 19 Let me just go over the outcome 20 definitions. They are binary. Success means you 21 were improved such that you didn't need additional 22 antimicrobial therapy. If you were failed you had 255 1 persistence of signs and symptoms and additional 2 therapy was required. In terms of the 3 microbiological it was fairly straightforward, you 4 could culture the pathogen isolated at baseline or 5 not. 6 [Slide] 7 There were several categories of infection 8 identified up front. Definite catheter-related 9 bloodstream infection was defined as one of the 10 following: Either a positive peripheral blood 11 culture plus a positive semi-quantitative catheter 12 tip culture; a positive quantitative lumen wash 13 culture or a positive hub or tunnel exudate 14 culture. You could also have more than a 5-fold 15 increase in the concentration of organisms of an 16 identical pathogen from a central venous catheter 17 versus a blood culture drawn peripherally, of in 18 centers where this could be done, you could have a 19 more than 2-hour difference in the time to 20 positivity for the peripheral culture versus a 21 centrally drawn culture. 22 They are fairly rigorous definitions and I 256 1 should say that these are slightly more relaxed 2 than in the guidance for catheter-related 3 bloodstream infections. 4 [Slide] 5 We had a second category of probable 6 catheter-related bloodstream infection. For Staph. 7 aureus, if you had a single positive peripheral 8 blood culture in the absence of any other source of 9 infection, you could be considered probable CR-BSI. 10 For all other organisms you had to have at least 2 11 blood cultures positive for the identical species, 12 at least one of which was drawn peripherally. 13 [Slide] 14 The study was initiated at 34 centers in 15 North America, with an enrollment period of 17 16 months. Less than half of those centers actually 17 managed to enroll a single patient, and 2,639 18 patients, the vast majority of whom had at least 19 one positive blood culture with a gram-positive 20 organism, were screened and 75 patients were 21 enrolled. So, I will save you the trouble of doing 22 the math. That is about 2.8 percent of the 257 1 screened patients. 2 It makes several points. First of all, no 3 surprises--gram-positive infections in patients 4 with central catheters are common. It is easy to 5 identify them; it is hard to enroll them in a 6 clinical trial. 7 [Slide] 8 The reasons for the screening failures are 9 shown here. In 30 percent of the cases there is 10 inadequate culture data, and this is not the lack 11 of the first positive culture; this is the lack of 12 the second culture, the lack of the availability of 13 a catheter tip, a lumen wash, or the absence of an 14 exudate culture. In many cases those are actively 15 discouraged at clinical sites around the country. 16 Very few centers are able to look at time to 17 positivity of blood cultures, and relatively few 18 sites do semi-quantitative blood cultures as well. 19 So, it is extremely difficult to get 20 patients into a study and meet those definitions of 21 catheter-related bloodstream infections. In 20 22 percent of the cases there was prior antibiotic 258 1 usage that excluded a patient from enrollment. 2 This doesn't comment on the number of patients who, 3 because of other infections that developed during 4 the study, received an antibiotic, for example, the 5 patient who developed a urinary tract infection 6 with a gram-negative organism but was treated with 7 a drug that had gram-positive activity. In 20 8 percent of the cases renal insufficiency was the 9 reason given. There were additional foci of 10 infection identified in 13 percent of patients. In 11 9 percent there were mixed gram-positive and 12 gram-negative infections, and neutropenia in 6 13 percent. So, you can see quite readily that under 14 the current guidance this is a challenging 15 indication in which to do a study. 16 [Slide] 17 Let me just sum that up by saying this is 18 a common disease. Gram-positive bacteremia in 19 patients who have a catheter in place is 20 exceedingly common. That is no surprise. Patients 21 who meet the definitions of catheter-related 22 bloodstream infections, however, are very rare. 259 1 This is a heterogeneous population. There are 2 patients with coag. negative staph; there are 3 patients with Staph. aureus; a smaller number of 4 patients with enterococci. 5 The inclusion/exclusion criteria used for 6 the study--when you exclude 97 percent of the 7 patients screened, I think it is not a stretch to 8 say the generalizability of the results is highly 9 questionable. What can you do with that? I think 10 the patients who were studied were well 11 characterized, however what is a clinician to do 12 with that information if you don't know how it 13 pertains to the population you are treating every 14 day? 15 Other issues--I mentioned the 16 microbiological methods required to defined this 17 entity of catheter-related bloodstream infections 18 are not standard of care in most places, and they 19 are not the sorts of things you can impose upon a 20 clinical laboratory in order to do a clinical 21 trial. 22 There is no approved comparator, which 260 1 might become an issue for a Phase 3 study. That 2 would certainly be an issue for discussion with the 3 agency before anyone would go down that road. 4 So, in my view, a Phase 3 study with the 5 current design is really not feasible. I do not 6 see how it could be done. If one wanted to use a 7 brute force approach, open hundreds of centers and 8 screen tens of thousands of patients, at enormous 9 cost, conceivably it could be done but you would 10 have to question what the results would tell you. 11 I don't think they would be generalizable. 12 So, what I would like to urge the group is 13 to consider alternate approaches to bloodstream 14 infections. I don't want to imply that catheter- 15 and device-related infections are not important. 16 It is a common clinical problem and there are many 17 things we don't know--which patients require 18 treatment? Who could simply have a catheter pulled 19 as opposed to needing treatment? Which patients 20 can be treated through an infection? Are there 21 patients with coag. negative staph. you can 22 effectively treat without removing the catheter? 261 1 What is the appropriate duration of therapy? So, 2 these are very important clinical questions but not 3 answerable with this kind of study design. 4 So, as I said at the opening, I think this 5 may be a case where lumping rather than splitting 6 might be the way forward. I would urge the group 7 to think more about other ways to look at 8 bacteremia in general. I know we will hear from 9 David comment on whether we can look at bacteremia 10 as a stand-alone indication. I think we have to do 11 something completely different than this if we are 12 to evaluate new drugs for bacteremias. Thanks. 13 DR. EDWARDS: Thank you very much, Tim. 14 Ralph Corey, from Duke, will extend the discussion 15 now. Ralph? 16 IDSA 17 [Slide] 18 DR. COREY: Let me first acknowledge Vance 19 Fowler whose data I am stealing. Vance and I 20 started a group called SAVAGE, a dramatic name, in 21 1994 when he was an intern. I had good ideas; he 22 loaded the trucks. Now he has the good ideas and I 262 1 make the lectures. 2 [Laughter] 3 [Slide] 4 Let's talk about clinical trials. I stole 5 this slide from David Ross. You know, talking 6 about issues of bacteremia as an indication, he 7 clearly pointed out that bacteremia is obviously a 8 multifaceted disease and if you clear Staph. aureus 9 pneumonia with bacteremia, you clear the pneumonia 10 and you are probably clearing the disease most of 11 the time. So, having an indication for Staph. 12 aureus pneumonia with bacteremia is, I think, 13 breaking this down way too far. However, there is 14 a middle group here that is very, very important 15 that doesn't have an obvious source of infection. 16 [Slide] 17 In 1998 as you have heard, the 18 Anti-Infective Drug Advisory Committee had a big 19 meeting. I read through 200 pages of transcript of 20 this meeting, and this is the way I summed it up. 21 Truthfully, there were a lot of great ideas, a lot 22 of the people are here, and they said bloodstream 263 1 infections of known tissue origin--you know, the 2 cure of the origin depends on the cure of the 3 bacteremia, and those are examples. 4 [Slide] 5 However, the bloodstream infection of 6 known catheter origin, removal of the catheter, one 7 to two weeks of antibiotic leads to cure. But what 8 was interesting is that Staph. aureus, 9 enterococcus, Candida, coag. negative staph. and 10 gram-negative rods were all catheter-associated 11 bacteremia. Can they all be lumped together? I 12 think one of the things we want to do is talk about 13 why they shouldn't be lumped together but perhaps 14 the trials can be. 15 [Slide] 16 The conclusions became guidelines for 17 trials of anti-infective agents in patients with 18 documented catheter-associated bloodstream 19 infections, and there are some problems. 20 [Slide] 21 First of all, as Tim just said, we have 22 this huge problem with requiring proof of catheter 264 1 infection. The catheters are in the trash bag by 2 the time you are trying to enroll these people. 3 That is one of the problems. All the problems that 4 he mentioned are big ones. 5 One of the key points that we started with 6 in 1994 is--and I didn't believe this, this second 7 thing--that the origin of bloodstream infections 8 defines the complication potential. Does it? 9 Basically, if you have a catheter-associated 10 bloodstream infection, does that mean it is benign, 11 especially standard staph? If you give them two 12 weeks of therapy they are all cured; you don't have 13 to think further? That is wrong, and we have data 14 to show that is wrong. Overall, there are other 15 markers that have to be added to that to help you 16 define the uncomplicated group. 17 This goes to the third point of the 18 guidance combines virulent and less virulent 19 organisms. Is there treatment for Staph. aureus 20 and coag. negative staph? We don't even know if we 21 need treatment for coag. negative staph. 22 catheter-associated bloodstream infections. Can 265 1 you just pull the catheter? Probably in 99 percent 2 of the people you can. We haven't been able to 3 define that one percent. In Staph. aureus, can you 4 just pull the catheter? No one in the room would 5 do that. With candida, would you do that? 6 The final thing is this allows approval 7 for more benign diseases, catheter-associated 8 disease, if it well defined, without subsequent 9 requirements obviously for trials in 10 life-threatening disease. That is not the agency's 11 problem but it is sort of all of our problem. Are 12 we testing the drug in a population in which it is 13 not going to be used? 14 [Slide] 15 Let's look at these five organisms. 16 Everybody in the room knows this chart. I have 17 sort of never seen it put together like this, so we 18 put it together. Staph. aureus is a rare 19 contaminant. It does cause septic shock. But the 20 interesting thing is column three, "metastatic 21 infection." It loves to cause metastatic infection 22 and, obviously, you can look at the resistance 266 1 column. 2 Coag. negative staph.--often a 3 contaminant. It doesn't cause many metastatic 4 infections. That rate is going up now. Actually 5 in our endocarditis group of almost 2,000 patients 6 we have 100 cases of native valve coag. negative 7 staph. endocarditis. I wouldn't have believed it 8 in a thousand years. So, the bug is getting worse. 9 Enterococcus? Enterococcus is sort of a 10 wimp; it is just a wimp. It loves to get resistant 11 but it often doesn't cause disease. Occasionally 12 it can though. 13 Gram negative rods, we all know they don't 14 cause metastatic infections usually. Then candida, 15 and then you have to break this out in the normal 16 host and the non-normal host. Obviously, you have 17 to break out coag. negative staph. with devices in 18 and with devices out b it loves devices. 19 [Slide] 20 So, you really have to split down to what 21 you want to study. We wanted to study Staph. 22 aureus first. The reason I show this slide from 267 1 The New England Journal, by Frank Lowy is that it 2 just shows all the receptors that Staph. aureus has 3 on its outer coating that bind to things. It binds 4 to collagen. It loves to cause metastatic disease 5 and that is the point of this picture. 6 [Slide] 7 I want to walk through this because this 8 is data that is really unique. This comes out of 9 the SAVAGE group. Out of about 1,300 patients 10 then, there are about 1,700 now enrolled. These 11 are all the patients with Staph. aureus bacteremia 12 at Duke Hospital in the last ten years, 13 prospectively identified and entered. 14 There are 368 patients with uncomplicated 15 Staph. aureus bacteremia, any source, all-comers; 16 no metastatic infection identified at that time, 17 time of entry; treated for 14 days. The cure rate 18 was 63 percent. Recurrence rate was 6 percent in 19 this group; attributable mortality, and this is our 20 best guess, is about 6 percent; and other mortality 21 was 20 percent. Then, there are still a few that 22 we don't have all the data on for that particular 268 1 characteristic. 2 If you keep going and you say, okay, I 3 want this same group but I want it 4 catheter-related, and the way we defined 5 catheter-related is we couldn't find another source 6 and the guy has a catheter in. That is it; not it 7 was inflamed; not that you could get pus out of the 8 tunnel; not that you did differential blood 9 cultures--none of the other, just a simple clinical 10 diagnosis; we can't find another source and the 11 patient had a catheter in. 12 Now all of a sudden, the cure rate goes up 13 to 75 percent. The recurrence rate goes down; 14 attributable mortality goes down; overall mortality 15 goes down. 16 [Slide] 17 We take this a step further because we had 18 an algorithm saying what is uncomplicated, which we 19 will get to. We said, as Sam Raad had suggested a 20 long time, maybe if they defervesce in 72 hours 21 that really helped define a group that was 22 uncomplicated, and if they had a follow-up blood 269 1 culture we figured that at 24-48 hours after the 2 start of antibiotics and after pulling the line, if 3 they had a positive culture then that, again, 4 defined a group that was the uncomplicated group. 5 They were going to do well. We give them two 6 weeks; they are not going to get endocarditis; they 7 have not gotten staph. infections and, indeed, we 8 are up to 81 percent cure rate, with a recurrence 9 rate that is 3 percent, and all three of these were 10 re-infections, like in a person who has more lines 11 in or is an IV drug user. The attributable 12 mortality is way down and overall mortality is also 13 down. 14 Finally, we are down to the last group 15 where we have all the others, catheter associated, 16 defervesced within 72 hours, negative follow-up 17 culture and a negative echocardiogram, and now our 18 cure rate is 88 percent and our recurrence rate is 19 3 percent, and these are re-infections. Basically, 20 95 percent of these guys were cured if they didn't 21 die of another disease. 22 So, we recommended if you are going to do 270 1 short-term therapy, we started out with just one 2 week, and we recommended TEEs on everybody whom you 3 were going to treat. But the trouble is that that 4 breaks the group down to about 35 patients and the 5 cure rate again goes into the 90s and the mortality 6 is lower yet. 7 [Slide] 8 So, what do we gain by all this? This is 9 a picture just looking at 370 patients to 65 10 patients. You lose patients as you go down, as you 11 more accurately define, there is no question. But 12 when you are down at the bottom you have a 13 well-defined group, well characterized, with a cure 14 rate of over 85 percent. 15 [Slide] 16 So, we proposed new definitions, and this 17 has actually been discussed by several people in 18 the room--John, David--and saying uncomplicated 19 bloodstream infection health care associated, and 20 that includes the outpatients and nursing homes 21 with IV catheters, etc.; no signs or symptoms of 22 metastatic infection; follow-up blood cultures, and 271 1 this is the strongest predictor, the follow-up 2 blood culture. If it is positive you are in deep 3 trouble, yes, you have problems. So, if you get a 4 positive blood culture 24-48 hours in--this has 5 actually been published just a few months ago. 6 Defervesce within 72 hours; and then an echo 7 showing no valvular disease, basically you just 8 have trivial insufficiency or stenosis; your valves 9 are good. Removable focus obviously has to be 10 removed; no hardware in place and non-neutropenic. 11 That is what defines it. We can do trials, you 12 know, adding these bottom ones back in. 13 [Slide] 14 The complicated ones are follow-up blood 15 culture positive despite removal if intravascular 16 catheters; no signs and symptoms of metastatic 17 disease upon enrollment except uncomplicated IE. 18 You can say, hey, I am going to take all-comers; I 19 will take the guys who got vertebral last year when 20 they show up. If you have bone data, that is 21 another way of doing it but then you have to have a 22 lot more data up front to make sure that you are 272 1 not going to get into trouble. Again, no hardware. 2 The length of the fever may be either less 3 or greater than 72 hours. Defervescence may get 4 greater or less than 72 hours and echo is not 5 necessary before enrollment in this group. 6 [Slide] 7 Inappropriate, we thought, were signs and 8 symptoms of metastatic disease on presentation; 9 removal focus--that is one thing that we feel very 10 strongly about; you have to remove the focus; and 11 hardware. We have real good data that if you have 12 a hip in or knee, you have about a 25 percent 13 chance of seeding the hip or knee if you have 14 Staph. aureus in the bloodstream. That is 1,000 15 patients from Duke in New Zealand. There is also 16 data on prosthetic valves; it is about 25 percent. 17 So, hardware in place is something you have to 18 really think about, whether you want to include 19 those in a trial. Neutropenia I think defines a 20 whole different group of people that we have to 21 deal with, with data that they don't do as well. 22 [Slide] 273 1 I am not going to go through this, what is 2 significant and which hardware you have to remove. 3 I think that you don't have to remove all hardware 4 and you have to define it fairly well. For 5 instance, plates and screws are not at high risk. 6 Knee joints are probably the highest risk and hips 7 are probably second. 8 [Slide] 9 The present guidance for a drug--and this 10 is my interpretation so you guys can correct me, 11 beat on me, or whatever--for a drug for MRSA that 12 is coming on the market to really handle MRSA as we 13 all hate it and vancomycin, we are all worried, is 14 not powerful enough, is to start out standard skin 15 studies, a randomized, double-blind study of skin 16 and skin structure, and then go on to Phase 2 and 17 do two randomized, double-blind studies of 18 uncomplicated bloodstream infections, and these are 19 catheter-associated, preferably superiority--I am 20 just reading it out of the guidance from 21 '98--followed by a randomized, double-blind study 22 of complicated bloodstream infection. 274 1 Assumption one is that step one predicts 2 success in step two. Does that happen? And, does 3 step two predict in step three or failure in step 4 three? Do we have to go step-wise? This is 5 something that makes all of us a little nervous. 6 We like to check it out in easier infections and 7 make sure it works. Do we need to do that? 8 Assumption two is that completion of step 9 one leads to step two. Why? I can do a skin study 10 and stop and say it is a good drug; use it. 11 [Slide] 12 So, our proposal is that we change the 13 definitions first to uncomplicated bloodstream 14 infections and complicated bloodstream infections. 15 Like Tim was talking about, we think about new 16 trial designs. That doesn't mean we change them; 17 we just think about them. 18 [Slide] 19 One of my first thoughts is--and I take 20 full blame for these--that a Phase 2 randomized 21 trial in skin infections, followed by an 22 open-label--and actually, I have spent a fair 275 1 amount of time with our Ph.D. statisticians at Duke 2 looking at what is the downside of this--50 3 patients open-label. Now, open-label causes a 4 problem in this because you may just say, hey, I am 5 going to pick a 26 year-old healthy guy and I am 6 going to get to my cure rate of 85 7 percent--open-label study with a new 8 anti-infective, and I reach a target value that is 9 comparable to the value that we get with what we 10 consider good care. If you reach 85 percent, you 11 go on to your large Phase 3 trial in both 12 complicated and uncomplicated disease combined. If 13 you don't reach 85 percent, you back off and have 14 to do a double-blind study in Phase 2. That is one 15 thing. 16 [Slide] 17 Middle of the road, do your skin study and 18 then go straight to a large open-label, randomized 19 study of all the patients with Staph. aureus 20 bacteremia and with frequent adjudication. You 21 want to adjudicate every five patients, or you want 22 to have a DMS look at every two patients, it 276 1 doesn't matter, whatever makes people feel 2 comfortable. But go rapidly. Combine the two; 3 don't split them out. 4 [Slide] 5 Finally, the head of our clinical research 6 institute is a cardiologist, Rob Califf, who is a 7 little bit outrageous but has done a lot of big 8 trials, and one of the things we were talking about 9 is when we have a heart attack drug we don't test 10 it in stable angina; we test it in heart attacks. 11 You sort of say, well, gee, I am going to take this 12 drug and I am just going to test it in the people 13 that it is going to be used in. I think that we 14 have to think this way. Whether we look at each 15 patient separately and make sure we are doing the 16 right thing, but I think that otherwise what is 17 happening is we are using drugs and testing them in 18 the clinical arena by people who aren't testing 19 them; they are just using them and that is what is 20 happening. 21 So, I think that you start the study of a 22 new anti-MRSA drug, as an example to bloodstream 277 1 infections, taking all-comers and looking at the 2 results. Obviously, you have to have your basic 3 package and obviously this doesn't take any 4 consideration of what you have to have on the side 5 for safety. For instance, you have to have a 6 safety database of--I don't know--2,000 patients, 7 whatever, but you can do easier trials. You can do 8 skin trials looking at safety of the drug and do a 9 large pivotal trial with both complicated and 10 uncomplicated disease. That is all I have and we 11 can talk about this. 12 DR. EDWARDS: Thank you very much, Ralph. 13 David Ross, from FDA. 14 FDA 15 DR. ROSS: Well, I feel like I have the 16 hard part here because after two really fascinating 17 talks I have to try to make the regulatory issues 18 interesting. I just promise I won't talk about the 19 Code of Federal Regulations. 20 [Slide] 21 What I would like to do is just very 22 briefly review the regulatory history of 278 1 bloodstream infection and bacteremia; talk about 2 defining anti-infective indications; and then talk 3 about primary bloodstream infection as an 4 anti-infective indication. In terms of defining 5 things, my sister is an attorney and she told me 6 once that the only thing she ever learned in law 7 school is why two situations that are exactly alike 8 are actually completely different. So, with that 9 in mind, I think we should just keep in mind that a 10 lot of what we do in terms of defining indications 11 is saying who is a describable group of patients. 12 [Slide] 13 If you look in the PDR there are ten drugs 14 indicated for bacteremia or septicemia or both. 15 All of these were labeled prior to 1992. Having 16 looked at some of the reviews, it was actually 17 quite interesting. The labeling was really based 18 on either very non-specific or completely 19 unspecified clinical manifestations that really 20 pooled a few patients from different studies, for 21 example, pneumococcal pneumonia, urinary tract 22 infection and so on. These NDAs included 279 1 bacteremia that was associated with focal infection 2 at a defined anatomic site, as well as bacteremia 3 that had no known origin. As I mentioned, patients 4 were pooled from multiple trials of multiple 5 indications. 6 [Slide] 7 This was not a particularly satisfactory 8 situation for anybody. Nobody really knew what we 9 were studying. In a 1993 advisory committee 10 meeting there was presentation of data from a large 11 NDA data set that suggested in that particular data 12 set that bacteremia did not affect outcome. The AC 13 discussion concluded that bacteremia is associated 14 with infection at a primary anatomic site; that 15 patients who have a systemic inflammatory response 16 syndrome in bacteremia form a very heterogeneous 17 population. It really was unclear in terms of 18 making a distinction on the basis of bacteremia in 19 this data set if bacteremic SIRS patients did worse 20 than non-bacteremic SIRS patients. 21 [Slide] 22 So, the conclusions at that time were that 280 1 bacteremia per se was less important than the site 2 of infection in terms of classifying infections and 3 defining indications. The study of bacteremia or 4 septicemia or what was then proposed is bacteremic 5 sepsis as a separate indication was not feasible 6 given the heterogeneity of the patient population, 7 and I will talk some more about that in a minute. 8 However, it was felt that because this was 9 information that was important to prescribers that 10 labeling should include bacteremia in the context 11 of site-specific infections, for example, pneumonia 12 with associated bacteremia. 13 [Slide] 14 I said this to some extent when I gave a 15 variant of this talk back in '98, why should we 16 reconsider this? Well, we all know there is an 17 increased incidence of bloodstream infections due 18 to resistant pathogens and without an identifiable 19 primary source. There is data suggesting that when 20 you can identify the primary source for a 21 bacteremic infection the outcome is worse. 22 But I also think it is important to think 281 1 about some issues as far as how we measure the drug 2 effect in bloodstream infection and first and 3 foremost is patient heterogeneity. Getting to this 4 lumping versus splitting question, there are the 5 differences in natural history for different 6 pathogens. Obviously, there is a much different 7 attributable mortality for bacteremia due to Staph. 8 aureus than that due to coag. negative staph. 9 There are differences in the epidemiology 10 associated with various pathogens and their 11 virulence. Finally, for the purpose of clinical 12 studies--let me distinguish that very clearly from 13 clinical practice--we need to define a discrete 14 clinical syndrome that we can describe in labeling 15 and use as a basis for studies. 16 [Slide] 17 Let me just give two different patients 18 here. This gets back to the question of is 19 bacteremia per se important. We obviously think it 20 is but it is not the only important thing. These 21 are two patients from a recent NDA, a 28 year-old 22 woman with category 1 community-acquired pneumonia, 282 1 and a 52 year-old man who falls in category 5. 2 The 28 year-old woman is bacteremic. I am 3 sorry, my graphics didn't come out. There wasn't a 4 question mark about whether those cultures were 5 obtained; they were, in fact, obtained. For the 6 first patient they both showed penicillin 7 susceptible Strep. pneumo. In the second case only 8 the sputum culture showed Strep. pneumo. But even 9 though one patient is bacteremic and the other is 10 not, their predicted mortalities on the basis of 11 the Port study are very different. 12 [Slide] 13 Furthermore, in terms of heterogeneity, I 14 think it is very important to recognize that not 15 all bacteremias are the same. Again, these are 16 real patients. This was a 76 year-old man whom I 17 helped take care of a few years ago at the VA, down 18 in the District. This poor gentleman came in with 19 Staph. aureus all over the place. He had it in his 20 lungs, he had it in his CSF and he had it in his 21 blood. The health staff did a great job of trying 22 to save him. They got an MRI while he was in the 283 1 unit. I still don't know how they did that. But 2 he died. 3 Another patient who had a graft through 4 which he was getting hemodialysis also had Staph. 5 aureus bacteremia, had a follow-up because culture 6 that was positive, but this patient did okay. He 7 had a negative echo; negative Doppler. We weren't 8 quite sure what to do with him. We weren't quite 9 sure what infection he had so we gave him four 10 weeks of IV vanco. and he did great. 11 But you can see in the context of a 12 clinical trial that these two patients are very, 13 very different. In terms of describing them in 14 labeling, it is hard to combine them. 15 [Slide] 16 So, right now we have site-based 17 anti-infective indications, and these are defined 18 in the Points to Consider document which we still 19 do refer to from time to time, although I will 20 leave out the delta discussion. 21 Infection at a specified body site due to 22 a specified susceptible microorganism is how we are 284 1 currently defining indications. The reason we 2 chose this is that this really accounts for 3 differences in drug efficacy for infections at 4 different sites. As a very simplistic example, we 5 think that meningitis is more difficult to treat 6 than skin and skin structure infections for a 7 variety of reasons. 8 It allows us to enroll patients more 9 accurately, or colleagues in industry to enroll 10 patients into adequate and well-controlled studies 11 and demonstrate efficacy. Then, in terms of 12 telling physicians and other providers who can 13 benefit from the drug, it allows us to describe the 14 drug efficacy in labeling. 15 [Slide] 16 The essentials of an anti-infective 17 indication are that it has to be a recognized 18 disease or condition, or an important manifestation 19 of a disease. In practice, that means it has to be 20 a definable syndrome with specific clinical 21 manifestations, diagnostic criteria and therapeutic 22 requirements. It has to be possible to demonstrate 285 1 drug treatment effect in trials using clinically 2 relevant endpoints, and this effect has to be 3 describable in labeling in the form of an 4 indication and usage section. 5 [Slide] 6 I have shown this slide before but just to 7 take some kind of simplistic ideas, UTIs are an 8 indication; dysuria is not. Osteo. is an 9 indication; an elevated sed rate is not, and so on. 10 Prophylaxis against postoperative infection is 11 indication, whereas decrease in skin colonization 12 is not. 13 [Slide] 14 In terms of the relationship of different 15 indications in bacteremia, I think it is important 16 to recognize--I sort of stole this slide back from 17 Dr. Corey, so forgive me--at any rate, I think it 18 is important to recognize that we have this group 19 here that is undefined and right now the question 20 is what to do with them. It is also important to 21 recognize that just because you can get a drug in 22 here, into the bloodstream, it does not necessarily 286 1 mean that you are going to be equally successful 2 getting it into a vegetation or getting it into 3 consolidated pulmonary parenchyma, and so on. 4 So, I think it is important that efficacy 5 in treating bacteria in the blood be combined with 6 evidence with efficacy for other tissues. I think 7 extrapolating from simple bacteremia to 8 tissue-based infections may be difficult. 9 [Slide] 10 Let me just talk briefly about the 11 distinction between bloodstream infection versus 12 bacteremia. The former term I am using to describe 13 a clinical entity if it, in fact, exists, whereas 14 the latter is a lab entity, a lab finding. We have 15 certainly recognized in labeling candidal 16 bloodstream infection or candidemia as an accepted 17 individual, which is frequently primary without an 18 identified underlying infection, and it can be 19 defined as a clinical syndrome in trials and for 20 purposes of labeling. 21 In contrast, E. coli bacteremia is not 22 accepted as an AI indication. It is virtually 287 1 never primary. There is a variety of potential 2 sources, for example, the GI or GU tracts, and 3 there is a wide variety of clinical manifestations 4 depending on that underlying infection. 5 [Slide] 6 I think any discussion of bloodstream 7 infection and associated bacteremia also has to 8 address the issue of whether we can use bacteremia 9 as a surrogate endpoint, and we have been talking 10 about this particular finding all day, that as you 11 increase the dose of clarithromycin you are more 12 effective at reducing bacterial load but less 13 effective at keeping patients alive. That is not 14 the same, for example, as Staph. aureus 15 bacteremia--different population; different but. 16 But if we want to move to Staph. aureus it has been 17 shown that addition of gentamicin to a regimen for 18 endocarditis sterilizes the blood more quickly but 19 it does not improve mortality. So, again, I think 20 it is important to be cautious in terms of what the 21 meaning of bacteremia is as an outcome measure. 22 [Slide] 288 1 Let me finish we some questions. First, 2 is there a distinct patient population with primary 3 bacterial bloodstream infection that can be 4 identified at the time of randomization into a 5 clinical trial? 6 If so, how should patients with BSI who 7 demonstrate evidence of metastatic infection 8 post-randomization be classified with regard to, 9 first, whether the BSI was primary or secondary at 10 the time of randomization and, two, how should they 11 be considered in terms of outcome? 12 Finally, in terms of definitions, what 13 combination of clinical manifestations, diagnostic 14 criteria, including exclusion criteria, therapeutic 15 requirements and clinical endpoints would define 16 primary BSI due to a specified bacterial pathogen 17 as a unique individual in labeling? Thanks very 18 much. 19 Discussion 20 DR. EDWARDS: Thank you very much, David. 21 We are going to open these presentations for 22 discussion now. Could we put those questions back 289 1 up? Any comments? George? 2 DR. TALBOT: My first question is, is 3 there a consensus that catheter-related bacteremia 4 is not a viable indication? 5 DR. EDWARDS: John, I think you are going 6 to have to answer that. 7 DR. POWERS: Well, we certainly agree 8 there are problems in studying that. George, you 9 and I talked about delta at the last meeting and 10 that is the question, what is the treatment effect 11 in those people? Ralph brought this idea up if you 12 are studying a less severe disease. 13 So, there are a couple of issues here I 14 think. One is the specificity of the diagnosis. 15 The most common organism you are going to find in 16 catheter-related bloodstream infections is 17 coagulase negative staph., which also has the 18 highest rate of contaminants. I thought you were 19 going to show that. Dave had a slide on this and 20 it actually showed the high rate of contamination 21 with that organism versus Staph. aureus. 22 The second thing is, which is sort of the 290 1 point you brought up at the last session, how do 2 you pick the right patient population, and it is 3 really hard for Staph. aureus to be able to do 4 that. Then, if we are going to do non-inferiority 5 trials here, which I assume would be what we would 6 be discussing, what is the treatment effect? Ralph 7 brought up that we don't know if you just take the 8 catheter and don't do anything for Staph. 9 epidermidis. 10 Then the other point is Tim's. Why should 11 we care that it comes from a catheter versus 12 something else? Then, the last thing is when you 13 look at these meta-analyses on catheter-related 14 bacteremias and diagnosing them, what is the gold 15 standard for comparison to those things? There 16 isn't one. So, is 15 CFU off the tip of a catheter 17 really the god standard or would we actually look 18 at that? 19 So, there are a lot of issues related to 20 that that I think become very problematic that we 21 have thought a lot about since that guidance came 22 out. But this really points out an issue with 291 1 guidances, and that is that now we don't think that 2 this is such a great idea what do we tell people 3 that might be in development for this particular 4 indication? 5 DR. TALBOT: I think you have highlighted 6 a lot of important scientific issues. I think, 7 just trying to speak from the IDSA perspective, if 8 we are here to speak about what could encourage 9 drug development for anti-infectives, it seems to 10 me that what we have at the moment isn't doing that 11 because there has been, to my understanding, no 12 indication granted for these. I don't know how 13 many trials are in progress but, clearly, what we 14 have now, it seems, isn't working for anybody. I 15 guess I would venture to say that. 16 So, the choices are either to dramatically 17 revise the guidance for catheter-related 18 bloodstream infection to the extent that it would 19 be clear that it would be feasible and highly 20 pragmatic to perform the study, and that the 21 results would be clinically meaningful. As Tim 22 pointed out, how can you generalize from one 292 1 percent of your screened population? So, that is 2 one choice. The other choice is to say, well, we 3 gave it the best shot; let's look at another model 4 of bloodstream infection, say, with staph. and try 5 to do better there. Either way, I think from 6 IDSA's perspective would be a great step forward. 7 DR. POWERS: I think one of the other 8 issues is that we are sort of granting indications 9 already that include patients that have Staph. 10 aureus bacteremia. So, daptomycin in complicated 11 skin and soft tissue trials include people that 12 have methasone-resistant Staph. aureus bacteremias. 13 The linezolid trials for hospital-acquired 14 pneumonia included people like that. And, are we 15 actually raising the bar by requiring people to 16 have positive blood cultures in those settings? 17 Right now what we have done is looked at 18 subsets of people who are bacteremic. The way we 19 have used it really is to ensure the specificity of 20 diagnosis, that we know that those people who have 21 a positive blood culture really have the disease in 22 question and that helps give us some certainty. 293 1 So, the question here is what information would be 2 gained by doing separate trials of Staph. aureus 3 bacteremia, and would that add anything either to 4 clinicians' knowledge about how the drug is 5 effective in these particular settings, and would 6 it be something that would be palatable for folks 7 in industry? Would it help them or would they 8 perceive it as here is yet another thing we have to 9 do? 10 DR. TALBOT: If I could just comment on 11 one of the things about bacteremia with complicated 12 skin, and I would like to hear what Frank, Tim and 13 Barry have to say, but I wonder, having seen a 14 number of trials, just how representative those 15 data are. My impression is that the vast minority 16 of patients have bacteremia and they are only in 17 there because in the setting of an investigational 18 drug trial they slipped in somehow. The 19 investigator didn't really think they were that 20 sick, or what-have-you, and then they show up and 21 they are doing okay so they are kept in. But I am 22 not sure that that is answering the question about 294 1 how the drug does in bacteremia. If you want to 2 answer that question I think you need to set out to 3 study it. 4 DR. EDWARDS: Yes, David? 5 DR. ROSS: Let me make two points, one 6 about the issue of catheter-related bloodstream 7 infection and the other about sort of the 8 development strategy if one was going to have BSI. 9 First off, Tim outlined very, very nicely 10 I think the difficulties with this sort of study. 11 There are two things I think that drive this, and 12 everyone is aware of this but let me just restate 13 them. First off, these are non-inferiority trials 14 in general because nobody feels comfortable saying 15 we shouldn't treat these patients with coag. 16 negative staph. in their blood if we think it is 17 real, even though we don't know the magnitude of 18 the treatment effect. But, as John pointed out, 19 coag. negative staph. is more likely to be a 20 contaminant than if you have a peripheral blood 21 culture. So, in order to try and reduce the noise 22 in these studies and make sure that where two drugs 295 1 look similar it is because they are similar, not 2 because there are a lot of patients who don't 3 really need treatment. The CR-BSI guidance was 4 written to try to be very strict, but it is a 5 problem. 6 I am grateful to Tim for sharing the 7 experience with this trial because we all have had 8 the experience of rolling our eyes when we are told 9 there is a consult or there is a patient with one 10 positive blood culture for coag. negative staph., 11 how long do we need to treat him for? You know, we 12 all say, well, you don't. But in the context of a 13 clinical trial it becomes much more difficult, more 14 expensive to say we have to prove that it is not 15 real. That is number one. 16 Number two, just to get to the issue of 17 package, sequence and that sort of thing, I think 18 our intent with the guidance, the original guidance 19 was that catheter-related bloodstream infection and 20 complicated skin and skin structure infections 21 share some pathophysiologic features. One of our 22 thoughts was that those two could support each 296 1 other, not necessarily having to do them I think in 2 a sequential way. In terms of rational drug 3 development you might not need to show efficacy for 4 cSSI before going on to catheter infections. You 5 could actually proceed simultaneously. Potentially 6 you could see the same sort of strategy with 7 primary BSI of unknown origin, that it could be 8 supported by other indications being developed at 9 the same time. 10 DR. COREY: David, you gave the examples 11 that, you know, these people are vastly different, 12 the young woman with pneumonia, etc. But the same 13 thing happens with other diseases that are lumped 14 together. You have a young man with single vessels 15 coronary disease who has an MI and you treat him 16 with a 2B3A and you have an 80 year-old who has had 17 three MIs who has no left ventricle and has an MI. 18 the mortalities of those two groups are 19 phenomenally different and yet they are treated 20 under MI studies. I don't see why we are doing 21 that. 22 We have to test these drugs in the 297 1 population in which they are going to be used. We 2 are trying to think here how we can get there 3 quicker and not lose safety. I don't want to hurt 4 or harm a patient; nobody here wants to harm a 5 patient and, yet, we do want to get new drugs on 6 the market and available to us before vancomycin 7 becomes totally useless. 8 DR. ROSS: I absolutely agree with you. I 9 think that is a very good point. Let me point out 10 that those two patients were actually enrolled in 11 the same trial under the same criteria, and I think 12 that is absolutely appropriate. I think, you know, 13 we have a validated prediction rule for saying how 14 different these patients are, and is that because 15 the patients are different or because the drugs 16 that they are being treated with are different? I 17 think that in the case of other conditions 18 associated with bloodstream infection, with 19 bacteremia let me say, it is certainly valid I 20 think to be able to enroll people with the same bug 21 who have very different prognoses as long as we can 22 define at baseline how they are liable to do so we 298 1 can separate the underlying protoplasm from drug 2 effect. 3 DR. COREY: One of the problems with at 4 baseline is that that is one of our big problems 5 with a trial like this. You enroll them before you 6 know what their prognosis is. If one of your big 7 criteria is the follow-up blood culture, then you, 8 at that point, can start defining them as 9 complicated or uncomplicated bacteremia. But if 10 take all-comers I think the cut point really is at 11 14 days. If you take all-comers with Staph. aureus 12 bacteremia and you treat them, at 14 days you are 13 going to know whether you are going to stop at that 14 point because there is no evidence of metastatic 15 disease and the patients respond, or whether you 16 are going to go on and after that point new 17 metastatic infections become complications. So, if 18 you do take an all-comers trial, it is fairly 19 complicated to set up but it can be done. 20 DR. ROSS: If Dr. Fleming has any thoughts 21 on this I would be grateful for your thoughts on 22 this. If we are defining patients on the basis of 299 1 a post-randomization event, since randomization is 2 frequently the major defense against entry of 3 statistical biases, how do we deal with that? 4 Obviously, we can prespecify the analysis and try 5 to deal with things that way but if you have a 6 relatively small trial where there are imbalances 7 between treatment arms, how do we deal with that? 8 DR. FLEMING: If you want to characterize 9 a subgroup for example, a subset of your patients, 10 ideally you would like to have that information at 11 time zero. Obviously, you have to have that 12 information at time zero at time of randomization 13 if you are going to use that for your eligibility 14 criteria. 15 What is critical is to be able to 16 establish that whatever measures you are actually 17 collecting post-randomization were in no way 18 influenced by the randomization. So, if you are 19 collecting information post-randomization that is 20 helping you to better characterize your patients, 21 if you want to use that for subgroup analyses you 22 have to be persuasive that those factors weren't 300 1 influenced by the effect of the knowledge of the 2 intervention to which the patient was assigned. 3 DR. ROSS: For example, just to take a 4 ridiculous example, let's suppose you have an 5 unblinded study and you are more likely to sample 6 the patient's blood post-randomization based on 7 treatment assignment, that could be an example. I 8 mean, that is obviously a ridiculous example but 9 that is one theoretical way-- 10 DR. FLEMING: Then you couldn't be 11 confident that the integrity of randomization would 12 be preserved. 13 DR. ROSS: But if, on the other hand, you 14 had, let's say, a standardized protocol for 15 collecting blood cultures post-randomization at 72 16 hours and no excuses accepted, otherwise the 17 medical student who is supposed to get it gets an F 18 for the rotation, then that would be okay in terms 19 of having post-randomization classification of 20 populations. 21 DR. FLEMING: Yes, you could have a blood 22 sample taken at baseline to identify gram-negative 301 1 sepsis, or something, and actually not determine 2 the results of that until after randomization. The 3 problem I have with that though is that the 4 practicality is that you are getting a result in a 5 population that isn't defined or isn't known at 6 time zero. Ideally, what I want to do is a trial 7 that will represent real world; that will allow me 8 to assess what is the effect in a population based 9 on time zero information that I would know in the 10 real world. 11 DR. POWERS: I think one of the other 12 issues gets back to our previous discussion about 13 surrogate endpoints, and that is the differential 14 effect between drugs. We know from data in terms 15 of data like vancomycin versus nafcillin in terms 16 of time to clearance of because cultures that there 17 appear to be differences across those drugs which 18 do not appear to translate into clinical outcomes. 19 So, it would be hard to imagine--even if you 20 specified in the protocol, clinicians aren't going 21 to change what they are going to do if they see 22 that blood culture remaining persistently positive 302 1 in one group of patients versus the other. 2 DR. COREY: I think we have to be careful. 3 That data all comes from Erol Sandy's group and 4 that was a combined trial between the University of 5 Virginia and UCSF, and it was mostly IV drug 6 abusers and the mortality in that study was 7 incredibly low. The chances of finding a 8 difference with a week of gentamicin was almost 9 impossible. They just didn't have the negative 10 outcomes to do that. That is the only study I know 11 of where we don't show a benefit even though we 12 show only a one day improvement in clearance. It 13 was a 6 day to 5 day improvement in clearance. So, 14 with a small improvement in clearance and no 15 outcomes or very few outcomes--you know, these are 16 young men with right side endocarditis' they didn't 17 die. 18 DR. POWERS: But how many times have you 19 heard vancomycin isn't as good a drug for Staph. 20 aureus bacteremia as nafcillin, based on that 21 information? So, there is a belief out there-- 22 DR. COREY: Yes, but that comes from a lot 303 1 of different data. That comes from the data on 2 right side endocarditis with nafcillin with gent. 3 where you can treat for two weeks but if you use 4 vanc. and gent. you tend to fail at a high rate. 5 That comes from looking retrospectively and in our 6 database prospectively at patients with MSSA and 7 the considerable number who got vanc. because, you 8 know, they are on dialysis, and looking at 9 outcomes. There is a lot of data to support the 10 feeling that vanc. isn't as good a drug. There is 11 not a prospective head-to-head trial because nobody 12 will fund it. 13 DR. EDWARDS: David Shlaes? 14 DR. SHLAES: Actually I agree with Dr. 15 Corey that you could, in fact, define a trial based 16 on Staph. aureus primary bacteremia very 17 reasonably. But the other thing to understand is 18 that when you talk about real world there is 19 nothing more real world in hospitals than primary 20 bacteremia. This is probably the most common 21 reason why people are treated with antibiotics for 22 nosocomial infection in hospitals. You get a call 304 1 in the middle of the night; a patient has a fever; 2 you worry about catheter-related bacteremia and you 3 treat the patient empirically. If you look at 4 antibiotic use in hospitals in the United States 5 and basically globally, 70, 80 percent is empiric 6 use and this has to be one of the most common 7 reasons for empiric use of antibiotics. So, there 8 is nothing more real world than this issue. 9 So, what I would submit is that you could, 10 in fact, construct a trial along the lines proposed 11 by Dr. Corey with reasonable inclusion and 12 exclusion criteria, but then the question is would 13 you be willing to take--going back to one of Dr. 14 Goldberger's suggestions--the data on the more 15 virulent pathogen, which is almost never a 16 contaminant, Staph. aureus, and apply that or 17 extrapolate that to a disease due to coag. negative 18 staph. which, of course, is a much less virulent 19 pathogen? So, that would be something I think to 20 consider. 21 DR. POWERS: That is one of the things 22 when we brought that up in March of 2003 that the 305 1 advisory committee did not feel comfortable with, 2 and that is extrapolating from different 3 microbiologic causes and, clearly, there are 4 differences, as Ralph showed, between Staph. 5 epidermidis and Staph. aureus. But the real 6 question there would be could we use that data from 7 Staph. aureus to support some other indication 8 where Staph. aureus is very common, like 9 complicated skin infections? So, when we were 10 talking about this internally we were thinking 11 maybe bacteremia with Staph. aureus would not stand 12 alone as an indication by itself if that was the 13 only study you wanted to do for drug approval, but 14 perhaps it could give us some important information 15 as supportive information for other diseases where 16 Staph. aureus may be more common, like 17 hospital-acquired pneumonia, complicated skin 18 infections, etc. to try to use it in a supportive 19 role. 20 There are some other issues here though. 21 The post-randomization thing is not minor, and we 22 have come up against this in empirical therapy 306 1 trials in both the fungal and antibacterial realm. 2 How do you define baseline infection versus 3 breakthrough infection? So, you get into the trial 4 and on day three you pop up with your big phrenic 5 abscess. How do I know that wasn't there at the 6 start versus this is progression of the disease on 7 therapy which one would consider a failure? 8 But let's get back to the issue of 9 supportive information. Would that be something 10 that would be helpful for people, at least 11 extrapolating from Staph. aureus to other diseases? 12 DR. EDWARDS: Anyone want to answer that? 13 Bill? 14 DR. CRAIG: To me, the important thing for 15 Staph. aureus bacteremia is that we are also 16 treating potential endocarditis. So, that is an 17 area that I think almost needs to be studied along 18 with bacteremia so that somehow you can put the 19 endocarditis patients in there to make sure that if 20 you are going to treat severe bacteremia you are 21 also going to be effective against endocarditis, 22 whether you can add them in as an open part or 307 1 somehow get them into the mix, because at many of 2 your smaller community hospitals, if they get a 3 staph. bacteremia, they are not going to 4 necessarily have the techniques to go ahead and 5 rule out endocarditis. So, I think you need to 6 have that included somehow into the clinical trial. 7 DR. POWERS: Bill, you read my mind. I 8 have written at the bottom of my page is this a 9 tacit approval for endocarditis? So, I think that 10 becomes something we are very concerned about, that 11 when people have bacteremic patients and they can't 12 find something, they assume--why do we treat those 13 people for four weeks? They assume it is 14 endocarditis and treat them. So, we have done 15 something like this with candidemia where we took 16 people with primary bloodstream infections with 17 candidemia but looked at the patients that had 18 organ disease within that trial, and that gave us 19 some comfort that, gee, this drug looks like it 20 works for hepatosplenic candidiasis. We did not 21 require people to go out and study 50 cases of 22 hepatosplenic candidiasis but we looked at that 308 1 subset within there. 2 The two subsets here we want to see are 3 both right-sided and left-sided endocarditis 4 because those do not appear to be the same disease 5 in terms of outcome and prognosis. 6 DR. EDWARDS: John Rex? 7 DR. REX: I wanted to make a comment about 8 the thing that I did in my previous life, which is 9 study candida bloodstream infections, and how that 10 actually has some parallels to this debate because 11 99 percent of what we have just been hearing has to 12 do with Staph. aureus. With candidemia you 13 actually are in a situation where you almost never 14 know why it is in the blood. Right? So, we don't 15 have these debates in the fungal world about is 16 this candidal pneumonia with candidemia or is this 17 candidal osteomyelitis with candidemia. So, you 18 might call it primary though truly it is never 19 primary. The candida did not materialize via a 20 transporter beam into the bloodstream. It got 21 there via a catheter or via some hole in the gut or 22 some other thing. The same thing is true of Staph. 309 1 aureus. It did not materialize in the blood; it 2 got there via some mechanism. Your problem though 3 is that you don't know what mechanism drove it 4 there. You can eliminate some of the obvious 5 suspects. The patient doesn't have pneumonia; the 6 patient doesn't have a UTI; the patient doesn't 7 obviously have osteomyelitis right now, isn't the 8 right age for bacteremic osteomyelitis. 9 So, what you are left with then is this 10 group of unknown primary or catheter. Then you ar 11 sitting there, staring at this patient, in bed with 12 a catheter in the vein and you say to yourself, all 13 right, is it the catheter or not? I have looked at 14 hundreds of cases of candidemia and I have looked 15 at catheter cultures until it drove me nuts. What 16 I conclude from all that is that I don't know for 17 any given patient--patient in bed 12 has a catheter 18 in place and the catheter has to come out; there is 19 no question, it has to come out for all the 20 patients, but for this particular patient I can't 21 tell you but I know that on aggregate the patient's 22 catheter has to come out. As a consequence, I 310 1 think there is a disease here, as David Shlaes 2 said, this is it. This is the disease you have to 3 treat. You are staring at a patient in bed 12 and 4 you have to treat him. 5 From my standpoint, there are really three 6 groups of candidemic patients. They are people 7 without a catheter; they are people who have a 8 catheter and somebody pulls it; and they are people 9 who have a catheter and somebody doesn't pull it. 10 And, I can show that there are differences in the 11 way that those three groups behave, more or less, 12 on an aggregate basis but for individual patients I 13 don't know which is which. So, I think you 14 actually have to treat them. Maybe you stratify 15 across the three if you are enrolling them into 16 your trials but even then it is very hard to do 17 that because somebody says they are going to pull 18 the catheter but then it turns out they don't, and 19 it is all very complicated and very painful and 20 irritating. 21 So, there is definitely an entity that 22 must be studied because it is a disease that a 311 1 clinician has to treat, and that is really what I 2 want to say. Can Staph. aureus bacteremia without 3 a probably source other than a catheter--I have 4 just made up a label indication for you--is that 5 something that by itself should get a drug 6 approved? I don't think so. I think you ought to 7 have something else to go with it. But by itself 8 is it a thing that I want to know about? Yes, it 9 is definitely something that I want to know. I 10 want to know what happens and I want to know 11 something about what happens when I do and do not 12 remove the catheter. 13 So, that is the end of my sermon on this. 14 It is a very hard area and Tim's story about having 15 to screen 2,600 patients to find 70 in whom they 16 proved that it was the catheter--you know, I think 17 that is a bold maneuver. I thank you for that data 18 because that tells the rest of us that we 19 never-ever, ever want to attempt to do that. 20 [Laughter] 21 And, I thought candidemia was bad enough. 22 But at least of you screen 2,600 patients with 312 1 candidemia you will enroll 200 and be able to do a 2 study. I have done that. So, it is possible to do 3 that study. But here you are telling us that you 4 can't really do it and you have just shown us that 5 if you take the group of patients who have this in 6 their blood with Staph. aureus, you have a high 7 relapse rate. You have an opportunity top measure 8 a serious disease and, thus, in the three studies 9 that I helped do for candidemia the first study was 10 about candidemia. For the second the disease was 11 candidemia and its complications. That is what it 12 really I think all boiled down to, it was "and its 13 complications." When you get down to a low relapse 14 rate, a low secondary event rate then you are 15 actually feeling pretty good about what you have 16 done. 17 DR. EDWARDS: David, let me ask a 18 question. It is now 4:30. We are basically at 19 about the beginning third of this discussion in my 20 estimation and we need to make a decision as to 21 what we are going to do about that. Keep going? I 22 am sorry, I promised 4:30. Why don't we shoot for 313 1 another 15 minutes? Okay? Then we will use that 2 as a cutoff. 3 DR. POWERS: Can I sum up what we have 4 heard so far and then we can get to the parts we 5 haven't covered? 6 DR. EDWARDS: John, you know, I have been 7 planning to do a summary and I am so completely 8 incapable of doing that at this point that I would 9 love to have you do that. 10 DR. POWERS: About this part of the 11 session, anyway, I think what we heard was that 12 primary Staph. aureus bacteremia without another 13 source could be an indication that supports other 14 disease entities in which Staph. aureus would also 15 be common, things like complicated skin and 16 hospital-acquired pneumonia. That database of 17 Staph. aureus bacteremia would include patients 18 with both right- and left-sided endocarditis 19 because we are essentially treating a lot of people 20 like that and some amount of information within 21 there in well-characterized cases, with a positive 22 echo, that we are sure have endocarditis--because 314 1 this is what Dr. Goldberger has been talking about. 2 We don't think the placebo cure rate in those 3 people is very high. So, that would give us some 4 really good, helpful information that the drug is 5 actually effective in those diseases. And, I think 6 it is problematic to extrapolate that to Staph. 7 epidermidis because they are not the same kind of 8 organism but it could be useful in extrapolating to 9 complicated skin and hospital-acquired pneumonia, 10 etc. as part of a package. 11 Now, what I think we have left to 12 discuss-- 13 DR. EDWARDS: John, excuse me, I just need 14 to ask one question about that. When you say 15 support-- 16 DR. POWERS: This means one study in this 17 bacteremia plus endocarditis group would support 18 one study--that is a really good question. What 19 does the word support mean? 20 [Laughter] 21 You know, what I said last week when 22 somebody came in with this, I said you have a tent 315 1 pole and no tent because you have to have something 2 to support. It doesn't mean you can get 3 hospital-acquired pneumonia with no 4 hospital-acquired pneumonia trial. It means doing 5 one study of hospital-acquired pneumonia and that 6 would go together in terms of each of those linking 7 to each other. Dr. Goldberger I think has used 8 this idea of, you know, in the Civil War they used 9 to stack the rifles up against each other to hold 10 them up together. That is what we are looking at, 11 each of these things supporting the other but there 12 has to be something there to support. That word 13 gets used very loosely--"I have no cases and no 14 isolates but here is my other stuff; I should get 15 approval for this completely different indication." 16 DR. TALBOT: I wanted to comment on the 17 support issue also from a slightly different 18 perspective because, just speaking about what I 19 heard, I heard support from the clinicians, over 20 there, for Staph. aureus bacteremia in one of its 21 variants and with adjustment for what you know at 22 the time of enrollment. I heard that that in and 316 1 of itself would be a valid indication. When you 2 commented about that supporting something else, 3 that actually causes me concern because I am not 4 sure I would want to go into what would be, of 5 necessity, a Staph. aureus bacteremia/possible 6 endocarditis before knowing a lot about the drug, 7 such as did it really work in a well-performed 8 development plan for complicated skin with a lot of 9 staph? Do you know anything about staph. if it is 10 in the urine, etc.? So, I actually would be 11 thinking about that as happening only if you have a 12 very convincing demonstration of success in other 13 indications and if you have a very convincing 14 rationale for dose selection for endocarditis based 15 on PK/PD, animal and clinical data. 16 DR. POWERS: This is a great dovetail for 17 tomorrow's discussion and it is what Dr. Schentag 18 brought up. There are well developed rabbit models 19 for endocarditis. It this a place where an animal 20 model would be helpful in terms of giving some 21 proof of principle to do you have any shot of being 22 effective in endocarditis so that you could use 317 1 that before you go forward in a Phase 3 trial, and 2 the dose selection issue as well. 3 DR. TALBOT: Well, the experts in that can 4 debate that but, frankly, when I was a practicing 5 clinician I wouldn't want to enroll my patient who 6 might have endocarditis in a study with a drug that 7 was going to its Phase 3 studies. 8 DR. COREY: I disagree with that, George. 9 I think the animal model is a good model. The 10 rabbit model is a good model. There is also Claude 11 Carbone's group in Paris, looking at penetration of 12 carbon-14 labeled drug into vegitations. So, we 13 can get some really good idea of penetration, and 14 we have every other group, from heart attacks to 15 oncology, where we are testing our drug in patients 16 with metastatic cancer or with cancer, and we test 17 them in heart attaches and, yet, we are sort of 18 afraid to say, okay, we are not going to test them 19 complicated bacteremia and endocarditis until we 20 have all these other things behind it, and I don't 21 think that is right. I think that leaves us right 22 now with no drugs approved for endocarditis, yet we 318 1 are treating endocarditis. So, the only people 2 testing them are the private practitioners, like 3 me, sort of saying, well, I will give this drug but 4 I have no trials to back me. 5 DR. TALBOT: I just want to comment. 6 First of all, as I said, I am not a current 7 clinician and I am not an expert on endocarditis so 8 I appreciate and respect your comments. I guess I 9 was coming more from the sorts of issues I have 10 seen. Certainly, a drug with potential should be 11 studied in this indication. The only question I am 12 posing--maybe that is a better way to put it--is 13 when in the development process it is appropriate 14 to do that. For some drugs, with George's Monte 15 Carlo and so forth, maybe it is fine but I guess-- 16 DR. DRUSANO: No, it is not. 17 DR. TALBOT: But I guess thinking about 18 some of the development programs I have seen, for 19 sure, I would want there to be as an iron-clad dose 20 rationale as possible; a totally consistent 21 preclinical PK/PD basis. There is an issue of 22 safety. This is a very ill patient population. If 319 1 you are talking about going directly into Phase 3 2 with a compound with a limited safety database from 3 Phase 2, I would be concerned about unexpected 4 safety issues which could then have a very 5 unfavorable impact, regardless of bacteriologic 6 efficacy, on the outcome of patients in that study. 7 So, those are the questions. The challenge is to 8 move into it as quickly as possible and I agree 9 with your premise that that should be the goal, but 10 just have a couple of caveats about how quickly you 11 do that on the basis of the information. 12 DR. ROSS: First off, I want to endorse 13 Dr. Rex' statement. There are no normal blood 14 flora; the Staph. aureus doesn't just get there. 15 One question I would like to ask, and I am not 16 asking anybody for any commitments to come in for a 17 pre-IND meeting, is what would people say is the 18 level--if there were a Staph. aureus BSI indication 19 and guidance of some sort on how to go about 20 developing drugs for that, what do people think 21 would be the level of interest on the part of "big" 22 pharma, "small" pharma, whoever, in terms of 320 1 developing drugs for that indication? 2 DR. EDWARDS: Who would like to respond to 3 that? Frank? 4 DR. TALLY: I thought when we first 5 discussed this in '98 and, actually, at an internal 6 FDA-industry meeting we also discussed it, and I 7 thought at that time there was of a lot of interest 8 in the pharmaceutical companies and in our company 9 for that. We were looking forward to doing that. 10 When we saw the criteria for the 11 catheter-associated infection we took a different 12 tack at that point in time, for many of the reasons 13 that Tim just outlined. Having experienced that in 14 a couple of other studies, it is a problem. So, 15 took the route of going a little bit in the 16 opposite way because there is an indication for 17 endocarditis. So, we took the route to go 18 endocarditis bacteremia, and then you can get it. 19 That is a tortuous route to go, and we are 20 in right in the middle of a study, as you well 21 know, and we have screened 3,000 patients to get 22 100, about double the amount that you have. They 321 1 are tough studies to do. But it is the type of 2 thing when you do complete it, if it is successful, 3 that is very important to the treating physicians, 4 and that is what they are asking us for. 5 DR. EDWARDS: Dave Shlaes, would you mind 6 commenting on that question? 7 DR. SHLAES: I think you are going to find 8 more interest from the smaller pharmaceutical 9 companies who are more concentrated in the hospital 10 market right now than from large pharmaceutical 11 companies who have traditionally been more 12 concentrated in the community market. That doesn't 13 mean that this will change but I would guess that 14 that would be kind of a generalization industry 15 sort of view. 16 DR. EDWARDS: Thank you. Yes, George? 17 DR. DRUSANO: I would like to kind of go 18 back to Dr. Corey's statement. I can only speak 19 for myself. I think, first of all, an animal model 20 is a necessary but insufficient condition to 21 support going into something as serious as 22 endocarditis. That is number one. 322 1 Number two, I think you have to be very 2 careful about the animal model that you use. You 3 know, people shouldn't use a mouse eye infection 4 model when the target is derived from mouse eye 5 infection models to do a PD assessment and say, oh, 6 we can hit that; we should probably go right into 7 endocarditis. So, I think you have to be very 8 careful about the database that you generate for 9 preclinical-clinical bridging. So, that is number 10 one. 11 Number two, I can only talk about our IRB 12 and our IRB, if you tried to go right from a Phase 13 1 environment into an endocarditis trial, you would 14 just get laughed out. I think, again, a necessary 15 condition is to have some reasonable expectation 16 that the drug actually works in lesser infections. 17 We learn all the time, but I do remember back in 18 the '80s we had cefonicid when euthanasia was 19 indicated. That thing got approved but we didn't 20 know enough about protein binding at the time and 21 we had a lot of endocarditis failures with that 22 drug. I would, sure as heck, before I went into 323 1 something like endocarditis want to know for the 2 patient's sake that it really worked. 3 DR. COREY: But did the previous skin 4 trials predict that there were going to be 5 endocarditis failures? I think the problem is you 6 are assuming that studies in skin and skin 7 structure infections are going to help us predict 8 what is going to happen in the serious infections, 9 and I don't think they are. I go back again, the 10 fourth time I am mentioning it, to we test drugs 11 for MI prevention in a highly lethal disease in 12 patients who have MI. We do not take those 13 sequentially through a series of other diseases 14 that don't exist. We do our preclinical as best we 15 can. We get a drug, we test it. We find out that 16 our bugs are highly susceptible. We find out it 17 penetrates into the right area. We look in our 18 data for drug levels in humans as Phase 1. Then we 19 go test it in diseases that we really care about. 20 I remember being overseas and we talked to the 21 French about skin studies, and they just laughed at 22 us because we don't need another drug for skin 324 1 studies. We need a drug for bacteremia. 2 DR. POWERS: Ralph, you are getting to a 3 great point. So, it is okay that we give the drug 4 to people that don't need it, and then what do we 5 know about how the drug works. I think this 6 dovetails greatly into what we are going to talk 7 about tomorrow, that is the utility of what happens 8 in Phase 2. We see a lot of people going from 9 Phase 1 to Phase 3. Nobody wants to take it from 10 the test tube and give it to somebody for 11 endocarditis. What is the role of small Phase 2 12 pilot trials where we could actually get some proof 13 of principle that would make people feel 14 comfortable and go forward? 15 DR. DRUSANO: That is a different story. 16 I mean, at the end of the day you have to have some 17 idea that the drug works. In point of fact, in 18 cefonicid trials there were some hints that the 19 complicated skin and skin structure trials didn't 20 quite get to exactly where you wanted to go, and 21 before you go into something like meningitis, like 22 endocarditis, you know, the lesser infections are a 325 1 necessary, again, but insufficient condition. If 2 they don't work in complicated skin and skin 3 structure it is unlikely to work in endocarditis. 4 Whether you say it as a function of I have to see 5 some well-controlled trials or I have to see some 6 good, well-designed translational Phase 2 trials, 7 that is fine; you know, six of one and half a dozen 8 of another and we are talking basically about the 9 same thing. I would be happy with that. But I 10 think before you go into things like endocarditis, 11 like meningitis you have to be very clear that your 12 drug actually does work. 13 DR. EDWARDS: Thank you. George, while we 14 are talking about dovetailing to tomorrow, we are 15 going to do that at this moment. If I could just 16 have one second though, tomorrow we really have our 17 work cut out for us. The intensity of the areas of 18 discussion are, if anything, more so than they are 19 today. I wanted to just thank everyone for the 20 extremely high quality presentations and extremely 21 thoughtful discussions, and I think we have all 22 learned a tremendous amount from what has gone on 326 1 today and have a lot to build on from here on. 2 I am not going to try to summarize today 3 at this moment because of the hour and, besides 4 that, tomorrow you are likely to get a higher 5 quality summary than you are right now. So, I am 6 going to conclude. We resume at nine o'clock 7 tomorrow morning. Again, thank you all very, very 8 much. Thank you. 9 [Whereupon, at 4:45 p.m., the proceedings 10 were recessed, to resume on Friday, April 16, 2004 11 at 9:00 a.m.] 12 - - -