ATDEPARTMENT OF HEALTH AND HUMAN SERVICES
FOOD AND DRUG ADMINISTRATION
CENTER FOR DRUG EVALUATION AND RESEARCH
ANTI-INFECTIVE DRUGS ADVISORY COMMITTEE
Thursday, July 11, 2002
Marriott Washingtonian Center
9751 Washingtonian Boulevard
L. Barth Reller, M.D., Chair
Tara P. Turner, Pharm. D., Executive Secretary
David M. Bell, M.D.
Alan S. Cross, M.D.
Steven Ebert, Pharm.D. (Consumer Representative)
Mary P. Glode, M.D.
James E. Leggett, Jr., M.D.
Judith R. O'Fallon, Ph.D.
Jan E. Patterson, M.D.
Julio A. Ramirez, M.D.
Ciro V. Sumaya, M.D.
Ellen R. Wald, M.D.
P. Joan Chesney, M.D.
G. Scott Giebink, M.D.
Robert M. Nelson, M.D., Ph.D.
Vernon M. Chinchilli, Ph.D.
Ron Dagan, M.D.
Alejandro Hoberman, M.D.
Colin D. Marchant, M.D.
George H. McCracken, Jr., M.D.
Jack L. Paradise, M.D.
Michael E. Pichichero, M.D.
Coleman Rotstein, M.D.
Renata Albrecht, M.D.
Rosemary Johann-Liang, M.D.
Mark Goldberger, M.D. M.P.H.
John Powers, M.D.
George Rochester, Ph.D.
Thomas Smith, M.D.
Janice Soreth, M.D.
C O N T E N T S
Call to Order: L. Barth Reller, M.D. 4
Introduction of Committee 5
Conflict of Interest Statement: Tara P. Turner, Pharm.D. 8
Guidance Development: John H. Powers, M.D. 11
Development of Antibiotics for Otitis Media:
Past, Present and Future:
Janice M. Soreth, M.D. 22
Design Issues in Antimicrobial Treatment
Trials of AOM:
G. Scott Giebink, M.D. 41
Experience with Tympanocentesis:
Clinical Diagnosis of AOM:
Michael Pichichero, M.D. 59
Double Tympanocentesis Studies:
Ron Dagan, M.D. 76
Limitations of Clinical-only Studies:
Colin Marchant, M.D. 100
Study Designs for Acute Otitis Media Trials:
What Can Each Design Tell Us?
C. George Rochester, Ph.D. 118
Lesson Learned from Past Approvals:
Thomas Smith, M.D. 134
Study Considerations: Recurrent/Treatment
Rosemary Johann-Liang, M.D. 148
Open Public Hearing
Michael R. Jacobs, M.D., Ph.D. 168
Jack L. Paradise, M.D. 187
Summary and Charge to the Committee:
Renata Albrecht, M.D. 192
Committee Discussion and Vote 203
P R O C E E D I N G S
Call to Order
DR. RELLER: Good morning. I am Barth Reller and I should like to call the Advisory Committee meeting to order.
We have an exciting agenda with multiple presentations, multimedia. It is very important that we adhere strictly to the schedule to enable full discussion of this important topic - Clinical Trial Design for Studies of Otitis Media.
This is coming to fruition of a great deal of work that has been done by many individuals over the years. To help us adhere to the schedule, Dr. Tara Turner, our executive secretary, will be having a light system that will quietly but firmly give the speakers notice when there are two to three minutes left depending on the length of the talk, two minutes for the short talks and three minutes for the 15- to 20-minute talks.
We will see the light, you will see the light that will go yellow, when it is time to send up red when your time is up, and a short period thereafter, the floor will open, and like the Mozart opera, there will be a display and disappearance.
We will begin with an introduction of the committee members and starting on my far, far right, actually, the consultants and the committee members, and to the far right, Dr. Pichichero, the name and affiliation.
Introduction of Committee
DR. PICHICHERO: Michael Pichichero, professor at the University of Rochester Medical Center and practicing pediatrician, Elmwood Pediatric Group, Rochester, New York.
DR. MARCHANT: Colin Marchant, Pediatric Infectious Disease at Boston University and Tufts University.
DR. HOBERMAN: Alejandro Hoberman, Department of Pediatrics, University of Pittsburgh School of Medicine, Children's Hospital.
DR. DAGAN: Ron Dagan, Professor of Pediatrics and Infectious Diseases, Ben-Gurion University, head of the Pediatric Infectious Disease at Soroka Medical Center in Beer Sheva, Israel.
DR. GOLDBERGER: I am Mark Goldberger from the Office of Drug Evaluation IV, FDA.
DR. POWERS: John Powers, Office of Drug Evaluation IV, FDA.
DR. ALBRECHT: Renata Albrecht, Division of Special Pathogens and Immunologic Drugs, FDA.
DR. SORETH: Good morning. I am Janice Soreth. I am the Division Director for Anti-Infectives at FDA.
DR. SMITH: Tom Smith, medical officer in the Division of Anti-Infectives at the FDA.
DR. JOHANN-LIANG: I am Rosemary Johann-Liang. I am the medical officer at the Division of Special Passages.
DR. NELSON: Robert Nelson, Children's Hospital, Philadelphia.
DR. GLODE: Mimi Glode, Pediatric Infectious Disease, University of Colorado, Denver.
DR. BELL: David Bell. I am Assistant to the Director for Antimicrobial Resistance in the National Center for Infectious Diseases, Centers for Disease Control and Prevention in Atlanta.
DR. TURNER: Tara Turner, Executive Secretary for the committee.
DR. RELLER: Barth Reller, Division of Infectious Disease at Duke University Medical Center and Director of Clinical Microbiology there.
DR. PATTERSON: Jan Patterson, Medicine Infectious Diseases, University of Texas Health Science Center, San Antonio.
DR. WALD: Ellen Wald, Division of Pediatric Infectious Diseases, University of Pittsburgh School of Medicine.
DR. SUMAYA: Ciro Sumaya, School of Rural Public Health, Texas A&M University System Health Science Center.
DR. GIEBINK: Scott Giebink, Professor of Pediatrics and Director of Infectious Diseases, Director of the Otitis Media Research Center, University of Minnesota Medical School.
DR. O'FALLON: Judith O'Fallon, statistician at the Mayo Clinic Cancer Center, Rochester, Minnesota.
DR. CHINCHILLI: Vern Chinchilli, biostatistician, Penn State Hershey Medical Center.
DR. CHESNEY: Joan Chesney, Pediatric Infectious Disease, University of Tennessee, Memphis, College of Medicine.
DR. RAMIREZ: Julio Ramirez, Chief, Infectious Diseases, University of Louisville, Kentucky.
DR. EBERT: Steve Ebert, Pharmacy and Infectious Diseases, University of Wisconsin, Madison.
DR. LEGGETT: Jim Leggett, Infectious Diseases, Oregon Health Sciences University.
DR. CROSS: Alan Cross, Infectious Diseases, University of Maryland, Baltimore.
DR. ROTSTEIN: Coleman Rotstein, Infectious Diseases, McMaster University, Hamilton, Ontario.
DR. McCRACKEN: George McCracken, Infectious Disease, University of Texas Southwestern Medical School.
DR. PARADISE: Jack Paradise, Department of Pediatrics, University of Pittsburgh School of Medicine and Children's Hospital in Pittsburgh.
DR. RELLER: Thank you. It's an exciting day. Dr. Soreth and colleagues have assembled what is recognized, it's like a Who's Who in Otitis Media in the world, if not the universe.
Dr. Turner will read our Conflict of Interest statement.
Conflict of Interest Statement
DR. TURNER: The following announcement addresses the issue of conflict of interest with respect to this meeting and is made a part of the record to preclude even the appearance of such at this meeting.
The Food and Drug Administration has prepared general matters waivers for Drs. Joan Chesney, Jan Patterson, Julio Ramirez, James Leggett, Steven Ebert, Ciro Sumaya, and Vernon Chinchilli.
A copy of the waiver statements may be obtained by submitting a written request to the Agency's Freedom of Information Office, Room 12A-30, of the Parklawn Building.
All other participants did not report any financial interests relevant to today's meeting; therefore, waivers were not necessary to permit their participation.
The topic of today's meeting is an issue of broad applicability. Unlike issues before a committee in which a particular product is discussed, issues of broader applicability involve many industrial sponsors and academic institutions.
The committee members and invited guests have been screened for their financial interests as they may apply to the general topic at hand. Because general topics impact so many institutions, it is not prudent to recite all potential conflicts of interest as they apply to each participant.
FDA acknowledges that there may be potential conflicts of interest, but because of the general nature of the discussion before the committee, these potential conflicts are mitigated.
We would like to note for the record that Kenneth Brown, M.D., is participating in this meeting as an industry representative, acting on behalf of regulated industry. As such, he has not been screened for any conflicts of interest.
In the event that the discussions involve any other products or firms not already on the agenda for which FDA participants have a financial interest, the participants' involvement and their exclusion will be noted for the record.
With respect to all other participants, we ask, in the interest of fairness, that they address any current or previous financial involvement with any firm whose products they may wish to comment upon.
I have a brief announcement. Dr. Kenneth Brown will not be able to join us today. He is ill.
DR. RELLER: Thank you, Tara.
We will begin the presentations with Dr. John Powers speaking about guidance development.
John H. Powers, M.D.
DR. POWERS: Thanks, Dr. Reller. I am really privileged to be the first one to get to test drive this trap door that is underneath my feet, so in case I fall through the floor, you will know why.
Today, I would like to welcome the members of the committee, our guests and consultants, the members of the audience, and our colleagues at the FDA.
Most of you were here for the advisory committee meeting that we held for two days back in February of this year, in which we dealt with some issues related to non-inferiority trials or deltas in antimicrobial drug development, and on the second day we talked about development of antimicrobial drugs for resistant organisms.
At that time, we stated that that meeting was the first in a series of meetings that we were going to talk about related to antimicrobial drug development. So, here we are today fulfilling that promise, talking about acute otitis media.
We really see this again as part of a continuing discussion and we plan future advisory committees to talk about other guidances, as well, as well as to continue the discussion about otitis media.
We are also planning a workshop, cosponsoring that with the Infectious Disease Society of America and the Pharmaceutical Research and Manufacturers Association in the fall of this year.
The divisions that deal with antimicrobial drug products in the FDA are the Division of Antiviral Drug Products, Anti-Infective Drug Products, and the Division of Special Pathogen and Immunological Drug Products.
All three of these are subsumed under the Office of Drug Evaluation IV and as part of the Public Health Action Plan dealing with antimicrobial resistance, the office has been given additional resources to deal with antimicrobial drug development and resistance issues. I am honored to be the lead medical officer to move those initiatives forward.
In an attempt to move this process of guidance development forward, which was started by Dr. Lillian Gavrilovich [ph] when she was the acting head of Anti-Infectives, and then since moved forward by Dr. Renata Albrecht, and also in order to provide some internal consistency with the kind of guidance that we offer to drug sponsors, and also we like to promote some interactions both within and outside the FDA.
Why do we have these guidances in the first place? Well, these guidances are really not regulations, they are not absolute requirements, but actually they are very helpful both for us within the FDA and also for drug sponsors.
In terms of the drug sponsors, they provide an outline for the scope of data that they need to show the efficacy and safety of their drug products, and we often heard from industry that they want to know the kinds of things that we are looking for.
These guidances are also helpful within the FDA to provide some internal consistency in the kinds of guidance that we offer to drug sponsors. Over the years, there have ben several iterations of these guidances, and Dr. Janice Soreth will talk to you this morning about how each of these guidances has impacted on the development of trial design in acute otitis media.
All of these guidances are based on the best available science and regulatory knowledge at the time they were written, but one of the things that makes medicine both fun and challenging is that the state of our knowledge is constantly changing.
So, why revise these guidances and why talk about them now at this point in time? Well, obviously, there are those changes in the knowledge of infectious diseases, and since the 1992 guidance, there have been several meta-analyses published on the effect of antimicrobial therapy and the natural history of acute otitis media.
Also, even since the 1998 guidance, the Agency for Healthcare, Research, and Quality has published an evidence report again relating to the natural history and the impact of antimicrobial therapy on acute otitis media.
Also, over the years we have seen a change in the resistance patterns of the common organisms associated, not only with acute otitis media, but also with many other infectious diseases.
Finally, there have been advances in the science of clinical trials. Both the FDA, PhRMA, and European and Japanese regulatory agencies have participated in the International Conference on Harmonisation in an attempt to bring some global consistency to how we develop antimicrobial drugs.
Also, over the years, this committee has discussed several of the clinical trials related to acute otitis media, and we have learned some lessons from those which we now need to incorporate into our future guidances.
Each of the guidances is arranged in a similar way and covers these important points in design, conduct, and analysis of trials. They talk about the definition of the disease and how to actually diagnose it, the study characteristics, the inclusion and exclusion criteria for that particular disease which again includes diagnostic criteria, but also defines the populations of interest for that particular disease, the drug and dosing regimens used in that particular infection, the evaluation of patients, and the timing and definitions of the outcomes, and finally, statistical considerations.
That is an awful lot to talk about in one single advisory committee, so what we are going to try to cover today, related to acute otitis media, is not all of these points, not to say that the ones we won't cover aren't important, but just given the time constraints that we have today, we are not really going to touch on specific drugs and dosing regimens, and although the statistical considerations are very important, we hope to touch on those at a future meeting, and not specifically to discuss statistics per se today.
The first thing we are going to talk about is definitions of disease. Obviously, it is important that the terms that we use are specified, so that the results that we look at are comparable across trials.
In the AHRQ evidence report, they examined almost 3,500 clinical trials in acute otitis media, and their conclusion was that the basic definition of acute otitis media used in many of those trials varied considerably.
Also, the definition of disease is important when we talk about particular subsets of patients, for instance, children with recurrent disease and treatment failure versus children who are experiencing their first episode of acute otitis media.
It would be appropriate to analyze these populations separately if the cure rates were radically different in children across those groups, or, as we have heard from this committee before related to the development of fluoroquinolone drugs for pediatrics, if it would be appropriate to limit the use of those drugs to appropriate patient populations.
The second thing we would like to talk about are study characteristics, what do we learn from different types of trials, and Drs. Dagan, Giebink, and Marchant are going to talk about this today, as well as George Rochester from the FDA.
When we talk about superiority versus non-inferiority trials, one of the main things we deal with again is that issue of the non-inferiority margin. In the non-inferiority trial, we need to know the benefit of antimicrobial therapy over placebo in order to be able to set that margin.
That actually brings up the issue of the role of placebo-controlled trials in allowing us to determine that given that there is still a significant controversy about the actual magnitude of the benefit of antimicrobial therapy in acute otitis media.
Placebo-controlled trials have been done in Europe, and I put this trial up here by Damosieaux in the British Medical Journal that was published in the year 2000. This trial enrolled children with a clinical diagnosis of acute otitis media and also looked at clinical outcomes, and it did enroll children who were between the ages of 3 and 24 months of age. So, these have been done in places other than the United States.
When we look at inclusion and exclusion criteria, again, we are defining patients who actually have the disease, and one of the issues that we will talk about today--again, some of our consultants will bring this up--is the issue of clinical trials which use only clinical diagnostic criteria versus the value of baseline tympanocentesis in defining children who actually have bacterial otitis media.
Also, we can use the inclusion and exclusion criteria to define specific populations of children. The population of children most likely to have acute otitis media is those kids between the ages of 6 and 18 months, therefore, what is the role of data in children who are over 2 years of age and how can we use that in applying it to all children with acute otitis media.
Also, Dr. Rosemary Johann-Liang will talk today about evaluating patients who failed prior antimicrobial therapy or prophylaxis, and mostly those kids have been excluded from prior trials, and should we be looking at them today as a separate indication.
Here is the big issue when we talk about enrolling children who may not have a disease which is amenable to antimicrobial therapy. If we look at the top bar and the bottom bar, let's just say that is Drug A versus Drug B.
If we just say for the sake of argument that 80 percent of kids in a particular trial get better either because they have viral disease or self-resolving disease, we then look at only the population of interest only comprises about 20 percent of the trial.
There is a difference between the two drugs, and say in the 20 percent of interest, 15 percent of kids get better in one arm of the trial, but 10 percent get better in the other.
If we then look at the overall cure rates in that trial, for the top drug, the overall cure rate would be 95 percent. For the bottom drug, the overall cure rate would be 90 percent. Therefore, the difference between the two drugs that we would examine in this particular trial would only be 5 percent, driven primarily by the large number of children with viral or self-resolving disease.
On the other hand, if we just do the percentages in the population of interest, the cure rate in the children for Drug A would be 75 percent, and the cure rate in the children for Drug B would be 50 percent. So, the first point would be the cure rate would be much lower, but the other point is that the difference between the two drugs would be orders of magnitude larger, namely, 25 percent in this particular example.
So, enrolling children in the trial who may get better spontaneously or who do not have bacterial disease has a huge impact on the outcome of the trial.
Lastly, we look at microbiologic and clinical outcomes, how good is the correlation between bacteriologic and microbiologic outcomes, and some of our consultants will talk about that today, as well as the role of the second tympanocentesis in determining differences in microbiologic outcomes and evaluating efficacy for resistant pathogens.
One of the other things we would like to talk about today is the timing of assessments, should we still be looking at some fixed endpoint or should we look at something like time to resolution of symptoms.
The last thing is what actually defines a clinical cure and how do we measure it.
Finally, even if we wanted to do the perfect trial, the issues are: What are the barriers to doing that, that are practical issues, what are the barriers to performing tympanocentesis, are placebo-controlled trials practical in the United States, how acceptable are these procedures to patients and parents, and can we perform trials more efficiently while still getting useful data from those.
This has been a multi-person effort from folks at the FDA. I would like to thank all the people that have contributed to this, some of whom you will see speaking today, as well as our support staff without whom this would not be possible at all.
Thanks very much.
DR. RELLER: Dr. Soreth will now talk about development antibiotics for otitis media, past, present and future.
Development of Antibiotics for Otitis Media:
Past, Present and Future
Janice M. Soreth, M.D.
DR. SORETH: Good morning. I would like to add my thanks and my welcome to committee members, invited guests who are experts in the field, members of academia and industry, and consumers who may possibly be in the audience, as well, and to my FDA colleagues.
Let me state at the outset that although we probably could have done a better job in having a multinational group here representing the mavens in otitis, I think we have done a fairly decent job in inviting the mavens in the field.
There are a few others who I wish, on retrospect, I had been able to invite, but I guess at the end of the day, there is only so much time and so much money.
As Dr. Powers said, this won't be the last meeting we have on guidance development in general or in the furthering of the guidance document for developing an antimicrobial for otitis media as we fully expect that we will get additional written comments to a docket, whose number I will give you later, so that if your thoughts at this point are not at a point where you wish to speak them at a microphone, you still will have ample opportunity to make written comments to us here at FDA and submit them, so that we can review them and, as is fitting, incorporate them into whatever the next iteration of the guidance is.
I feel especially privileged to come before you today because I was an otitis-prone child, and I think it had a tremendous effect on my development as a child and as an adult, because for much of my childhood, I don't think I could hear very well, so muffled is my perception, my memory of what it was like to learn language in between many, many bouts of otitis media.
I am convinced that at some point I had that bioactive membrane that Dr. Giebink spoke of some years ago where the middle ear cavity is not just filled with pus, but probably has a pseudocolumnar epithelium that is secreting gunk all the time, which intermittently gets infected. That is quite a challenge for any anti-infective to go after and probably different from the garden variety not often happening acute otitis media.
I have also given birth over a decade ago to an otitis-prone child, so I have a special vested personal interest in this field, as well as professional interest, despite the fact that I turned out to be an internist, and many of the pediatricians in this from Pitt were my mentors, for I am also a Pitt alumna.
With that as background, then, let me go to the next slide.
We had, starting in 1977, written guidance, which I will briefly go through, followed by a period of formal silence, that is, nothing written between '77 and '92, although there was not silence in the office, there was not silence in speaking with academicians in the field or with industry. A lot went on, and I will try again briefly to summarize what is now oral history, anecdote, et cetera, and I will count on my other senior colleagues in both special pathogens and anti-infectives to keep me on track and speak up if I misspeak and misremember.
In '92, the Anti-Infectives Division came out with a Points-to-Consider document, talking about many different indications and infections, and how to go about trying to get a claim for them in one's antimicrobial product development, and that dovetailed in '92 with a contract that we had with probably some of the folks in this room and others, with IDSA, in writing, and what you call this depends on where you are. If you are in the FDA, you call these the IDSA guidelines, and if you are in the IDSA, you call them the FDA guidelines.
Finally, in 1997 and '98, we took another hit at coming up with an iteration of a draft guidance on many different infections including acute otitis media, and in '97 and '98, brought the guidance document formally then before this committee, whose composition in '97 and '98 was different because members do rotate off and go back to doing what they always do, or doing it in addition to this, I should say, because we appreciate that you have very, very full schedules and full professional lives, so again our thank you for your full participation in today's proceedings.
I say "back to the future" because I think, as you will hear today, we started in the seventies with a paradigm under which all children underwent tympanocentesis if they entered a trial with the clinical diagnosis of acute otitis media. We are going to read this at that. Some may think that is overkill, others may think it is the only way to go if you want to understand the microbiologic etiology of this infection.
So, questions that will come up throughout talks and certainly in the discussion this afternoon will focus on whether or not we need to return to a paradigm in which we have tympanocentesis for all, and then if we develop that further, should it be tympanocentesis at baseline only or tympanocentesis at baseline combined with a look, a further tympanocentesis, a repeat tympanocentesis on therapy, does that make scientific sense, is it ethical, and so forth. I am sure these will come up in our discussion, I hope they will, should all failures be tapped, is that practical, can you often do it, what do you do with a child at 3:00 a.m. on the eve of Christmas, Hanuka, New Year's, whatever, and it's virtually impossible to get it done.
One of the pivotal trials that we have held or recommended to companies to do is something we refer to as a "clinical-only" trial, a trial in which you make a clinical diagnosis of acute otitis media, and we will argue about whether or not that is an easy call, a difficult call, or something in between, are those studies serving us at this point.
I fully believe that any guidance document written, is written with the best of intentions in mind, and while some issues that we discuss today may appear to be Monday morning quarterbacking, I think that it is true that guidance documents and such provisions are written with, at that time, the best thinking in mind, the best of intentions, the most efficient way to get at testing a hypothesis and coming up with answers that are good for the public health, but as Dr. Powers said, our knowledge changes over time, at least we hope that it does, and we hope that it improves over time and maybe that is why we call it the practice of medicine, hoping that at some point, we really will get it completely straight and right and perfect.
Is there a role for placebo-controlled trials? I was happy to hear a report on NPR, because that is where I get a lot of my information, that when put to scrutiny, arthroscopic surgery in adults compared to placebo is no better, at least if you believe the data that have just come to light, and may be deleterious.
Is it time, is it neat and right at this point to consider the "P" word, placebo-controlled trials, in the context of studying patients, primarily children, many of them under 2, who have acute otitis media or who have an otitis media even if it's somewhere between acute and chronic
Regardless of the paradigms that we talk about in a given clinical trial development program, are we talking one trial, multiple trials? I don't expect that all of these ideas will be developed in today's proceedings. We do have a full agenda, many, many speakers with many things to say, and just note parenthetically that this will be one of a number of discussions in a public forum that we will have on this document, on guidance development for acute otitis media and going forward.
So, what is the bottom line, what do we know, what do we need to know to conclude that a drug works and it is safe for children with otitis media?
What are some of the stats? Well, over 25 million visits for otitis media yearly, and that is just in the United States, and we know that we are part of a global community, so there are millions more out there in other countries, accounts for 1 out of 3 pediatric visits, and I have done my share to contribute to that number.
By 1 year of age, upwards of 60 percent of children have at least one episode of acute otitis, and 17 percent more than 3. By 3 years of age, 80 percent have had more than one episode of otitis, one or more, and 46 percent, greater than 3 episodes.
It is a spectrum of disease. I am a lumper, not a splitter, and I see things along a continuum starting out with garden variety acute otitis media where there is pretty much a normal middle ear cavity and pus in it versus changes in histopath--I always wanted to be a pathologist, but I didn't get there for a number of reasons--changes in the histopath that bring you over to a fundamentally different protoplasm in that patient, bioactive membrane, and do we lump all of these children together in a single study, if we do, should we be cognizant of that and come up with schemes in which we stratify to understand the effect of a drug in different subpopulations, and then do we power it to be able to look at.
So, anything that we say today, as much as may not have the time to get into all the nitty-gritty for the statistical plan and considerations, that is not to say that we are not cognizant that that is an incredibly important part of clinical trial design.
I am looking at some of our dear statisticians smiling at me, because I think that we have to recognize that anything that you might recommend to us today has definitive implications for clinical trial design sample size.
If you are talking about non-inferiority margins, necessarily, delta determinations, the "D" word, and we ma not get to the specifics of numbers today and what do you recommend and what do we think, et cetera, but at some point in the discussion, it will rear its head. It is going to grow arms and legs, and it will be in front of us to deal with.
Likewise, the implications of clinical trial design, non-inferiority, placebo-controlled, whatever, have implications for the whole economic side of the house, one that we don't often talk about, but is obviously a very important part of the business of drug development, for if we take a tack or accept a recommendation that at the end of the day, completely skyrockets by orders of magnitude what is costs to do a clinical trial, I am afraid we won't get it, because there is only so much money that a corporation or NIH, or anybody, has to put to the study of anything.
So, we have to have a balance between what is I think right in terms of science and regulation, and the good of the public health, because ultimately, we are taking care of pediatric patients, patients in general, at the same time that we are cognizant of the fact that there is, by and large, corporate development of new anti-infective compound, not individuals working in their basements or in their garages, and that if it is much more profitable to develop, and practical and doable, to have a cardiac drug used forever and ever by a population or a drug for Alzheimer's that there is only so much money in the pocket that can be devoted to the study of any given entity and that common sense and practical issues also necessarily come to play.
So, let's go back to 1997, to the guidance on acute otitis media. The number of trials in this guidance were not addressed, but there was a case definition that spoke to having clinical evidence of acute otitis media or evidence of inflammation of the tympanic membrane and middle ear.
The guidance document recommended or required that in both studies, you have a tap at baseline, and it went to say that a second tap was desirable to obtain data on middle ear fluid concentrations and the promptness of bacteriologic eradication or cure.
Endpoints were both then clinical and microbiologic, and while the document did not specifically address a test of cure, it did recommend a four-week follow-up period.
I think at times, people and groups, and so forth, went back and forth on this. If you have a tap on therapy, you know it is sterilized, end of story, you don't have to worry about it anymore, you just need to see the patient at the end of the treatment course and no longer.
Others have argued, no, you really need to look at the patient for several weeks beyond that period of time, so that you can see whether or not the effusion resolves to make sure the child doesn't relapse, and certainly in the setting of an active control trial, that you can compare even longer term what happens even if at the end of day, you want to argue, but that is not really drug effect, you can't hold the drug's feet to the fire, so to speak, four weeks out, five weeks out, six weeks out.
Again, I expect this to come up in our discussions and be further developed.
The 1977 guidance concluded that in the absence of culture of the middle ear fluid, no specific claim could be made regarding the effectiveness of any anti-infective drug.
In the eighties, as I mentioned, there was no new formal guidance on otitis, and what I am going to give you now is what we talked about in the corridor, anecdote, the lowest level of evidence of what was going on, but in our internal discussions of acute otitis media, we really talked quite a lot about the requirement to perform or the heavy recommendation to perform tympanocentesis on every child enrolled in a trial.
From what I remember from those discussions a decade ago, and admittedly, my memory is not what it used to be, but what I remember from those discussions is that we often heard from colleagues in industry and others that the procedure was not that easy to do and was not well known by many, many pediatricians, many family practitioners, the very folks who were taking care of these children, and that it was much more involved than a venipuncture.
That may be incorrect. I am just telling you what we heard that I think caused a fundamental shift in paradigm that led to what we came out with in '92 from IDSA or from the FDA. Too few were really trained to do it or do it well. It seems to be slowing down enrollment in trials, hampering enrollment in trials, and that the cost was going up in requiring that every child have a tympanocentesis.
So, was there a better way to design these trials, better, without costing the patient anything, better for the efficiency of doing a trial, and at the same time, in that better way, not give up the opportunity to know whether a drug works or not.
The 1992 points-to-consider then said that two trials should be conducted in investigating a drug and its treatment effect on acute otitis media. One could be a clinical-only study in which no tympanocentesis was necessarily performed at baseline to establish equivalence to an approved product, and that a second trial that had both clinical and micro endpoints would be done with, at a minimum, a tympanocentesis at baseline.
The case definitions should be rigid. This is an important point, because I think, at least my understanding of what was going on back in the early nineties, was that we thought we really could come up with a rigid case definition, a look to the TM, a set of signs and symptoms that would be virtually pathognomonic for acute otitis media mediated by bacteria.
I see Dr. Pichichero is smiling because I think he is going to give us information that is other than what I said, that it is at times maybe more often than not, not such a straightforward call. It is probably why I went into internal medicine. Those little structures were so little, you know, sometimes it is really hard to tell is this acute otitis media with effusion, is it otitis media with effusion with a child who is sick otherwise and has something else going on, but not a bacterially-mediated otitis media, could we have been in error that we thought this was so straightforward that we could say rigid case definition, this child has a bacterial-mediated acute otitis media, no need to do a tympanocentesis?
The 1992 points-to-consider strongly encouraged--oh, my gosh, is that red light going on, the yellow, I have another minute and a half, okay, I will move faster--tympanocentesis was strongly encouraged in patients who were therapeutic failures at any point in the trial, and the endpoints, as I mentioned, both clinical and micro. Test of cure wasn't specifically mentioned in terms of the timing.
The open micro study should establish acceptable outcomes in, you know 25 patients with H. flu, 25 patients with Strep pneumo, and 15 with M. cat.
By and large, this dovetailed with what was published in the IDSA FDA guidelines.
One other think I want to mention about the '92 document and then to move on, the 1992 points-to-consider document stated that the micro trial could be uncontrolled, could be non-comparative, and the interpretation of that almost exclusively was is non-comparative, so once we say something can be some way, it probably will be, so we have to be very careful what we ask for because we know we will probably get it.
In '97 and '98, then, when we revamped the guidance document and took it before the Advisory Committee, we again spoke of two trials, a micro study, which could be non-comparative, but should have more numbers in it, and a comparative clinical trial. Again, case definition, please let's tighten it because I think we were certainly beginning to appreciate at that point that it wasn't maybe so very easy to have a rigid case definition, that there was a lot of wiggle room and a repeat tap to be considered day 3 to 5 as a critical measure of treatment efficacy, perform tympanocentesis in all failures, primary efficacy endpoints being clinical at the test of cure and pathogen eradication.
This test of cure, we have talked about a lot in the past five years in product-specific meetings, and the consensus at the last couple of meetings, when we have talked about Augmentin ES, or talked about azithromycin, short course treatment, I think the consensus that we have is that when we are looking at test of cure from the clinical perspective, we should define that closer to the end of therapy, and that still do a look several weeks out as another measure outcome, but the test of cure be closer to that last pill that is taken for clinical.
Further recommendations that came from the committee were to enroll more patients under 2 years of age, because as we look back at different products in development over the past 15 years, some had few or no children under the age of 2, very striking, so I am sure we will talk about that at length, and gain much more experience in this era of resistance.
Increase the number of patients under 2, I have said that, and we just skip forward.
Timing of assessment of clinical outcome. Primary endpoint, I have mentioned, again, the recommendation to encourage that those who fail, have another tap, whether it is the second tap or the third or whatever, and that the most informative tap would be baseline to understand what was the etiology, and then a consensus that on-therapy taps could tell us a lot.
Whether or not that has to happen all the time, some of the time, again, I am sure we will get into.
Experience has told us that this can be a difficult clinical call in some hands, and that even when the inclusion criteria are tight, we have experienced in looking across many different drug development programs that some investigators bat .800, 80 percent of the time they have a positive culture, and others are batting .200, 20 percent of the time they are getting a positive culture, so something is going on.
So, back to the future. We want to revisit the case definition, is it strict, is it strict enough? Trial design considerations, I have really already talked about, as well as endpoint and timing of assessments.
I think at some point in our discussions today, we will revisit the issue of placebo-controlled trials because what we want to understand is not only does a drug work in this disease, but the general question of what is the role of antibiotics development in acute otitis media.
I want to hold up for a moment as I walk off before the floor swallows up in Don Giovanian fashion--thank you for that opera reference--the management of acute otitis media and evidence report by colleagues at the Agency for Healthcare Research and Quality, it was actually done on contract to a group in I believe Southern California led by Dr. Michael Marcy [ph] if you have not read this, and I think many of us have not.
On the FDA side, I want to thank Dr. John Powers and Dr. Erika Brittain for bringing this to my attention. It is quite a comprehensive report that is certainly I think teaching us a lot about what we thought we knew and what we do know.
The key question is again what do we need to know, what constitutes substantial evidence that a novel antimicrobial drug works and is safe for children with acute otitis media or some other variety of otitis media.
With that, I will stop and I will turn the podium back over to Dr. Reller.
DR. RELLER: Thank you.
Dr. Scott Giebink will now speak to Design Issues in Antimicrobial Treatment Trials of Acute Otitis Media.
Design Issues in Antimicrobial Treatment
Trials of Acute Otitis Media
G. Scott Giebink, M.D.
DR. GIEBINK: As we are getting started, since Dr. Soreth went back in time, it is unfortunate that Medline searches only go back to 1968, because what this group needs to know, and I want to put in the public record, is that we are all indebted to a physician/scientist at the Mayo Clinic, between 1958 and 1962, who conducted four separate clinical trials of acute otitis media, comparing antibiotics we wouldn't consider today, but put the whole issue of AOM design on the table, and that now retired Professor of Pediatrics is Gunnar Stickler [ph], who had maintained a life-long interest in otitis media, and really brought us out of the dark ages into the era of clinical trial design for otitis, and as we have heard already, there have been lots of innovations and we refinements to that over the years, but it really started with those publications in 1958 to 1962.
It is worthwhile going back and looking at some of those for some of the early thoughts on design.
Well, I wanted to pick up a few design issues now, to just basically put them on the table for your consideration, and I thought I would start by saying the obvious, that we really have three ways we look at outcome in these otitis media trials.
The one that has been mentioned is a bacteriologic cure, which is basically defined as sterilization of middle ear fluid, eradication of the original pathogen, and that obviously requires an on-therapy tap, just as a second urine culture would require in urinary tract infection.
We have issues that Dr. Dagan is going to talk about later that relate to eradication of organisms versus growth suppression of organisms, that I think to be considered by the committee.
The second, of course, is clinical cure. Dr. Soreth just mentioned that. This is the resolution of clinical signs and symptoms. For reasons that should become apparent over the next couple of hours, the test of cure is really too obscured by issues of relapse and reinfection to be useful in measuring otitis media outcome, so as Dr. Soreth mentioned, moving that test down to end of treatment makes a lot more sense, and I will say a bit more about that in a moment.
Finally, Dr. Craig has put into the literature, as have others, issues of pharmacokinetics and pharmacodynamics, and our use of the kinetic parameters to describe an expected clinical and bacteriologic outcome, and the parameter that seems to be holding up over time or at least the last half-dozen years is this parameter Time over MIC.
There, the issue I think that needs to be more considered is whether we can really rely on Plasma Time over MIC or should we be talking about Middle Ear Fluid Time over MIC, and this will get into some of the characteristics of chronic otitis media I will mention in just a moment.
Now, the design issues, the four specific design issues I would like to comment on here in the next 15 minutes are some of the issues around the double tap design and using that in a non-comparative setting. I know that Dr. Dagan is going to amplify on this considerably.
I have already mentioned that the sub-issue there is the timing of the second tap and the related issue, the question of eradication versus growth suppression.
The second issue that I would like to show you some data on is the issue of enriching subject populations in clinical trials for the infection with penicillin resistant and multidrug resistant Streptococcus pneumoniae, PRSP.
The bottom line is I will show you that the risk factors for PRSP infection are those very same risk factors for recurrent and chronic otitis media, so that by enriching, by definition, you change the subject population, and then you have questions about generalizing data results from such a trial back to the whole population at large.
It has become clear, certainly in studies we have done, in studies Dr. McCracken has done, and several others, that these PK parameters that we talk about are valid in a particular patient, but they are incredibly variable. There is a tremendous variation in PK parameters.
The otitis media pharmacokinetics probably don't relate very well to the murine models where these PK parameters have been used extensively, and one of the big issues is most of the PK studies in humans are single-dose studies, and single-dose studies don't measure drug accumulation over time. We know that that is a factor in the middle ear.
So, using PK and PD parameters as surrogates of clinical effectiveness or bacteriologic effectiveness, I think is problematic.
Finally, I would like to show you evidence that otitis media severity at entry correlates, not only with clinical cure, but also with bacteriologic cure, and these issues have rather large implications for sample size determinations in clinical trials.
It is a fact that we have actually known for a decade, but has not been strongly considered in most trial design.
I am going to show you the exact same numbers Dr. Soreth just showed you, that there are 24 million visits at least as of about 1995, in the United States for AOM, a tremendous burden in very young children.
The reason I put in, in half of the 7 to 12 million cases of pneumococcal otitis, 25 to 40 percent are now resistant to penicillin, the reason I put this number up there is to emphasize that very small differences in treatment response have an impact on millions of children, so when we talk about 3, 5, 7, 10 percent differences in outcome, we are talking about 2, 3, 4 million children, and we shouldn't lose sight of that fact as we make these decisions, which seem very small in terms of percentage response, but very large in terms of number of children affected.
I would like to use the data from Pittsburgh, and Dr. Paradise and Drs. Wald and Hoberman can amplify on this later. This was a study led by Phil Kaleida at Pittsburgh in the late 1980s, early '90s, looking at a placebo-controlled trial of AOM.
I remember sitting on the opposite side of the table with Ellen and Jack in the early 1980s as this was being designed, going through all the ethical questions about placebo-controlled trials, and I am delighted to hear that it will come back on the table here for discussion, because I think it is time to do that.
Let me just make a point about enrollment severity, the severity of otitis media at entry. These are the bacteriologic data from that study that show that there is a tendency of a difference in the bacteriology of mild versus severe AOM. You will notice that the incidence of pneumococcal otitis in the severe group is almost twice that of the mild group, and the Hemophilus-infected ear is lower in the severe group than in the mild group. I believe Dr. Dagan is also going to talk about his recent fairly large experience with Hemophilus otitis when he talks.
So, there is a difference in the bacteriology, I believe, of mild and severe otitis media.
In that study, the older children, I believe the age cutoff was 2, were given placebo treatment compared to
amoxicillin, the younger children had myringotomy if they had severe otitis.
If we lump together the placebo and the myringotomy groups, we see that there is a 92 percent spontaneous resolution rate that we have already seen in the mild group and the 76 percent spontaneous resolution rate in the severe group. Those differences were significantly different, but small compared with amoxicillin.
Now, what happens when you put together the bacteriology and the clinical response?
That is what I have done in this rather jumbled slide, but I think you can follow me through here. In the first line of the mild and the severe group are the percentages we just saw two slides ago of those different bacteria isolated from the ears.
On the second line of both groups are the spontaneous cure rates that were described 30 years ago by Virgil Howie in his studies in Huntsville, Alabama, and have been large correlated by other placebo studies since then with bacteriology, that there is a spontaneous cure rate of about 20 percent with pneumococcal otitis, about 50 percent with Hemophilus, about 70 percent with Moraxella catarrhalis, roughly 30 percent in the mixed groups, and, of course, 100 percent when there is no growth in the middle ear fluid.
So, multiplying the first and second line together, you see the bacteriologic cure rates that would be anticipated, and on the far right are the total cure rate adding up that row, 63 percent bacteriologic spontaneous resolution or cure with mild AOM and 50 percent with severe AOM. That delta of 13 percent is one of the deltas that we speak of.
If you remember on the previous slide, the clinical cure rates from the Kaleida study. Here is a 92 percent and a 76 percent, a delta of 16 percent. So, these deltas are very similar, but the magnitudes, as the statisticians can comment later on, are quite different, and these differences have big implications for sample size and are an issue that I think need to be discussed further in the day.
The group that Dr. Marchant was a member of in Cleveland a number of years ago, led by Susan Carlin, most recently summarized all of their experience with clinical and bacteriologic outcomes, and demonstrated that if you compare clinical with bacteriologic outcome, the clinical status failure or success predicts about 93 percent of the bacteriologic responses, and it misses about 63 percent with a specificity of only 37 percent, 15/40.
The cells I think of interest really are these cells, this one and this one, and that is the discordance between the bacteriologic and clinical response, and you might ask then why is there bacteriologic success in the absence of clinical success.
A couple of the reasons for this are the presence of persisting bacterial and host inflammatory mediators in the middle ear, which we know continue the inflammatory process after organisms have died, and concurrent viral infections that may be related to or have nothing to do with the middle ear bacterial infection, but cause what is interpreted as a clinical failure.
That constitutes about 6 percent of the total pie, and then we have about 9 percent bacterial failures with clinical successes. Why does this happen? Perhaps it's because we have low-grade pathogens in the middle ear or these pathogens are growing more slowly because of inhibitors in middle ear fluid. This gets to the issue of bacterial suppression in double tap studies.
Other reasons for persistent symptoms during treatment, in addition to concurrent viral infection, is obviously that the organism continues to grow either because of noncompliance with treatment or resistant organism, or because the drug does not distribute into the ear, and I am going to comment on that in just a minute. Dr. Soreth mentioned the continuum of otitis media, and I will show you why I think that is critically important.
The persistence of inflammation after organisms have cleared and then the very rare case of immune deficiency that impairs the response to clearing those organisms, those are all reasons that symptoms may go on during treatment related or unrelated to continued bacterial presence and emphasize why a clinical outcome is so problematic in this disease.
The group in Finland has probably, along with Tasni Chalmatri's group in Galveston, have done a lot in the last decade to tell us about respiratory viral infection in otitis media.
It has been clear for a long time that respiratory virus play a major role in acute otitis media, and in addition, the studies, particularly in Finland, show us that in the absence of bacterial isolation from the middle ear, respiratory virus play a large role, as do absence of any pathogen in the middle ear causing the clinical diagnosis of acute otitis media, and since all of these ears were tapped, this 16 percent had middle ear fluid.
I think that represents the host clearing the organism by the time the needle is put into the ear, but you will notice in each one of these bars, Pneumococcus, Hemophilus, Moraxella catarrhalis, that there are ears with both pneumo and respiratory virus and without. So, respiratory virus play a very important role in this disease.
This diagram may be one of the most important summaries of otitis media pathogenesis that I could show you because it demonstrates how heterogeneous this population of otitis media really is.
The acute uncomplicated acute otitis media, which perhaps every child gets before they go to school, at least 80 percent get this disease documented in medical record studies, these days has very few suppurative complications although we do occasionally see mastoiditis still today. We just had a child with facial nerve palsy due to mastoiditis last week. So, these issues do continue to occur, but they are much less common now than they were 50 years ago.
The difficulty in designing a clinical trial is that we have this conundrum of a clinical mixture of AOM and chronic otitis media with effusion, shown in the green here, most of which in young children is mucoid otitis media or "gunk," I think Dr. Soreth called this, is these glue ears, and many of these ears are becoming acutely infected and appear to be AOM, but, in fact, pathologically, are chronic OME with a superinfection, and studies are starting to demonstrate that drugs distribute more poorly into the chronic OME ear than they do the AOM ear.
Then, we have children that go on to those nonsuppurative sequelae that include hearing loss, as well as pathology of the middle ear.
So, when we enrich a subject population for recurrent otitis media or for penicillin-resistant pneumococcus, we are creating a study cohort that is not representative of uncomplicated AOM, and yet, the indications go back to that uncomplicated AOM population and one has to ask the question is this a valid extension of those studies.
I have just put side by side here for you, fairly well accepted risk factors for PRSP on the left, and for AOM treatment failure and recurrence on the right, and you will notice that there is a tremendous similarity, antibiotics within the last month, in the case of treatment failure, any AOM diagnosis within the last month, recurrent or persistent AOM for PRSP, recurrent and persistent sinusitis, as well, infection during the winter or spring for PRSP, obviously is an AOM risk factor, too, young age, young age at the first otitis episode, daycare center attendance, which Dr. Wald's studies demonstrated clearly.
For treatment failure, not necessarily for PRSP, bilateral versus unilateral disease. So, when we select for treatment failure or PRSP, we are getting both.
This is a figure that I extracted from Dr. Wald's study that demonstrate this very clear increased incidence of OM complications of common upper respiratory infections based on daycare size from the home care, group care, to the center care group in the children that are less than 1 year.
In more recent studies, the day care center A, B, and C, all in the same community showed extension of a multidrug-resistant Type 14 pneumococcus that spread across the community through these daycare centers.
It was not detected in general pediatric practices. So, daycare centers serve as a reservoir for transmission of organisms that cause AOM and these penicillin-resistant pneumococci to show you the impact of multiple risk factors.
This is a study that Dr. Daly, with our group, did back in the mid-1980s, taking just three risk factors - bilaterality, daycare, and otitis for more than four weeks at entry in this AOM epidemiologic trial, looking at the percentage of children that had OME persisting six weeks later, and you will notice if they had none of these risk factors, a third of them had persisting OME.
If they had all three risk factors, two-thirds of them had persisting OME. So, risk factors are very important in identifying this subset that have persisting disease.
Dr. Dagan is going to say a lot more about carriage rates, but it is all about these very young children.
The rates of pneumococcal resistance by drug are shown in the figures that you have. They are all significantly greater rates of resistance, these are susceptibility rates, are lower rates of susceptibility in ear infection compared to eye, respiratory, blood, and central nervous system.
And younger children have lower rates of susceptibility or higher rates of resistance than older children.
Finally, pneumococcal conjugate vaccine selects those very serotypes that are carrying the resistance genotype, at least today, and you will notice here that the seven types contained in the Wyeth-7 valent conjugate vaccine include the most frequent resistant types including two types that are closely related to serotypes in the vaccine with fairly high resistance rates not found in the non-vaccine types, indicating that routine pneumococcal conjugate vaccine, I believe is going to have a significant impact on the early childhood rates of PRSP, and it is going to make the design of studies for enrichment with PRSP very difficult in the next few years if we can get enough conjugate vaccine in the pipeline to immunize all of these children.
So, in conclusion, I would again emphasize, as the previous speakers did, the importance of controlling enrollment in these trials and call your attention to the fact that bacteriologic and clinical cure rates are very tightly related to these clinical definitions.
The importance of end of treatment cure, not test of cure, at 25 to 30 days.
The issues with enriching for PRSP that we have just finished talking about, and, finally, the issue of pneumococcal conjugate immunization and its anticipated impact on PRSP prevalence in young children, all issues for us to consider.
DR. RELLER: Our next speaker is Dr. Pichichero. The presentations have been wonderful, although each drifting into the red zone. We will pick up the time one way or the other, so think about it either eating into lunch or eliminating breaks.
Experience with Tympanocentesis:
Clinical Diagnosis of AOM
Michael Pichichero, M.D.
DR. PICHICHERO: Thank you, Dr. Reller.
As I mentioned in my introduction, I am blessed or privileged depending on your religious viewpoint, to continue to practice primary care medicine half-time, as well as spending the other half of my time at an academic medical center. As such, tomorrow morning at 8 o'clock, I will be seeing patients once again as one pediatrician in a 10-pediatrician private practice group in Rochester, New York.
In sitting at my desk, I calculated that as a pediatrician, I have looked at in excess of 100,000 ears over my 20-year career, and that number will continue to climb. Many of my patients are the children of physicians or nurses, and many of them are on clinical trials.
I have participated in over 150 clinical trials, about 20 of them involving tympanocentesis, and this year, for the first time, we intend to attempt a double tympanocentesis trial at the encouragement of my former student, Dr. Ron Dagan, who was a fellow in our training program, so student became teacher. I don't know how it will go in that patient population, but we are going to give it a try.
My presentation will have three components. As Dr. Reller implied, it is a multimedia presentation. The first part will be a 12-minute video demonstrating a tympanocentesis procedure by myself on one of my patients. Then, I have a mannikin, and I am going to actually perform a tympanocentesis for the committee on an infant mannikin, live. That will take two or three minutes.
Then, I am going to show a video, which we produced in collaboration with the Pittsburgh group, Dr. Hoberman and Kaleida, on otitis media diagnosis.
These three pieces of teaching material are used in workshops which are taught around the country since 1999. Faculty of those workshops include Dr. Giebink and Dr. Marchant, and we have now trained in excess of 3,000 primary care providers in the tympanocentesis procedure through these workshops. Less than 10 percent of them went on to actually do tympanocentesis as a routine in their practice, as we do in our practice in Rochester.
If we could roll the first video.
DR. PICHICHERO: Hello. I am Michael Pichichero of Rochester, New York, and I am going to be performing a tympanocentesis procedure on this young man, 4-year old Nicholas.
Tympanocentesis procedure, which we perform in our office every day, has a series of indications, so these are met in all of the children to one degree or another. Tympanocentesis in our office is performed when a child is toxic in their appearance in association with acute otitis media. We also will perform a tympanocentesis if the child has a very bulging eardrum to the point where we anticipate it is going to rupture spontaneously anyway.
We also perform a tympanocentesis in the highly febrile patient, which would be acute otitis media with fever over 102 degrees Fahrenheit orally in the teenager or young adult, or over 104 degrees Fahrenheit in the young child, such as Nicholas.
We also would perform a tympanocentesis on the patient who has been unresponsive to previous antibiotic therapy. There is some discussion whether we would perform the procedure after a single failure of first-line therapy, such as amoxicillin or trimethoprim-sulfamethoxazole, whereas, most every physician expert in otitis media would agree that following failure with first-line therapy, such as amoxicillin, end of failure with a second-line therapy, that in this circumstance, a tympanocentesis can be very helpful to determine whether there is a pathogen present, and if so, what is the pathogen and what would be the preferred antibiotic therapy for that isolated bacterial species.
The benefits of tympanocentesis include immediate relief of pain in the crying child who is suffering from the pain of a bulging tympanic membrane, we can provide instant relief, as acute otitis media really is an abscess of the middle ear space.
We can determine whether the infection is a bacterial etiology or if it's a viral etiology, and if it is of a bacterial etiology, we can perform sensitivity testing in order to determine whether the organism will be killed with traditional first-line agents or whether a second-line agent would be preferred in this circumstance.
Tympanocentesis has the benefit of draining an abscess, which we know is therapeutic in and of itself, and last but not least, we feel that tympanocentesis can improve a physician's diagnostic accuracy.
Nothing is more self-educating than to diagnose acute otitis media, perform a tympanocentesis, and find that the ear tap is dry, the patient never had acute otitis media.
Also, we think that it is very beneficial if you perform a tympanocentesis and no bacteria are isolated, then, no additional antibiotics are necessary, and that can be very beneficial, as well, in avoiding the unnecessary overuse of antibiotics.
For the tympanocentesis procedure, we typically do not provide any anesthesia, we don't put the children to sleep. Some offices do give a medicine called Versed, which is taken orally, and then the child becomes very sleepy, but then they have to remain in the office for an hour or so before they are completely recovered.
Other times we will give a child some tylenol with codeine and a little valium mixed in to make the child relax, but in the case of Nicholas, his eardrum is so bulging with infected fluid, actually, I think he is going to feel relief rather than pain when we perform the tympanocentesis.
It is like opening a pimple or a boil yourself. When you open it up, it actually feels better, and you don't even feel the needle go through. So, that is what we are anticipating with Nicholas.
So, the first thing we will do is we will lay him down, make sure he is completely still with something we call a papoose board. My nurse will hold his head firmly, and then we will look into his ear with the Welch-Allen otoscope, not this one which you are used to seeing me examine him with, but rather we use this otoscope because it allows me to put the needle through, and I can still see through this mirror, so I am watching the whole time exactly what I am doing, so I put the needle exactly in the spot I want in the eardrum, so that there won't be any damage to his eardrum.
To do the procedure, we take a needle that looks like this. It's a spinal needle, and I bend it and hook it to a syringe, and then, as you see in this picture, the needle will be inserted through the ear canal until it touches the eardrum, and then we will suck the fluid off of the middle ear space in order to--the needle will be inserted through the ear canal until it touches the eardrum, and then we will suck the fluid off of the middle ear space in order to culture it and in order to drain that middle ear abscess.
There are some potential rare or hypothetical complications from tympanocentesis. Certainly, you would expect the possibility of some bleeding because we are going to put a hole, a tiny hole through his eardrum, and some pus and fluid may come out of the eardrum puncture site which I create with the tympanocentesis needle.
That should stop in a day or two as the hole heals over. Usually, three days after a tympanocentesis is performed, you can't even tell where the hole was.
Now, if the child is not properly restrained and they move their head about in the middle of the procedure, then, there are other possible complications where the little ear bones behind the eardrum could be scratched or injured, and there are blood vessels back behind the eardrum, and they could be scratched or injured, so that is why it is necessary for us to restrain your child and hold him very still during the procedure. The only real risks are when the child moves very suddenly and very unexpectedly, and they are not properly restrained.
A critical element to the tympanocentesis procedure is proper immobilization. Here, Mrs. Koon will put Nicholas into our papoose board, and my nurse Julie will secure him into the papoose board.
We usually do allow the parent to remain in the room throughout the procedure to reassure their child, and we will papoose children up to the age of about 4 or 5 years of age. After that, it may not be necessary to papoose the child, but in all cases, we require an assistant to restrain the child at the arms, and a second assistant who will restrain the child at the head.
We then will remove the spinal needle from its container. We use a 20-gauge. Other physicians who practice tympanocentesis recommend an 18-gauge needle. Of course, the stylet is removed and then the sterile syringe is attached to the spinal needle, and then the needle must be bent at a 45- to 90-degree angle, approximately one-third from the hub.
So, I will bend it thusly, and this depends on your own comfort level and how you hold your hand during the procedure, but in all cases, the needle must be bent, but the precise angle according to your own comfort zone.
Now, we maintain sterility by keeping the sheath over the spinal needle tip until we are ready to proceed with the actual procedure. Visualize the tympanic membrane. I am proceeding now down through the canal. I am right at the tympanic membrane, everybody takes a breath, and there we are.
We suck back the fluid, pull out, and we are finished. It is as quick as that.
So, we are going to perform the tympanocentesis procedure. The speculum is inserted. We visualize, we ask the nurse assistant to pull back on the pinna. We now insert the needle through the speculum, through the ear canal, get in good position, we are ready, and, pop, we are through. We draw the fluid. We pull out and we are done.
For needle placement, here is a normal ear for orientation. The preferred location for the tap, interior quadrant, where the light reflex is. An acceptable alternative is the posterior/inferior. It is essential to completely avoid the entire superior half of the tympanic membrane.
Now, here is an image of an abnormal ear bulging with infection. Again, the preferred location for the tap is the anterior-inferior quadrant. An acceptable alternative is the posterior-inferior quadrant. it is essential to completely avoid the entire superior half of the tympanic membrane.
Following the tympanocentesis procedure a decision is made regarding antibiotic selection. This can be guided by gram staining of the tympanocentesis material showing gram-positive or gram-negative bacteria, and then specifically directed at the pathogen and penicillin-susceptible versus resistant pneumococci, if isolated, beta-lactamase positive or negative, Moraxella or Hemophilus, as isolated. So, you can do directive therapy.
Of course, follow-up is necessary. We usually see the children back in three weeks, sooner if the bleeding or fluid persists beyond a day or two, or any alarm on the part of the parent.
[End of video.]
DR. PICHICHERO: Okay. That is a tympanocentesis. If we could have the lights up, please.
DR. PICHICHERO: This is a baby mannikin. We train pediatricians how to do tympanocentesis with this mannikin. The manikin is loaded with a disk. The disk looks like this. This is four tympanic membranes. You will see that in the top half of the tympanic membrane, when it is in the right position, will be a red dye. If the needle goes in through the red dye, you fail the test.
In the bottom half, you see a yellow pus. If the needle goes into the yellow pus, be it anterior or posterior, you have had a successful tympanocentesis. If you put the needle too far, you get a blue dye. This is to indicate that you have now hit the posterior--you have hit bone, periosteum bone at the posterior aspect of the middle ear space.
So, when we test our doctors, it is very easy. You have either got red dye, yellow dye, or blue dye.
This mannikin is engineered for the disk to go into a position, so that when it slides into the head, it is has the proper angulation and anatomical position of a real child.
Here is the otoscope that you saw in the video. You turn it on. I have already pre-bent my needle, and I won't maintain sterility today. So, you look into and you locate your anatomy. I can see the red dye and the yellow dye, and then go into, I progress down, I puncture, withdraw the fluid, and come out, it's that fast.
We can do another one. Rotate the disk one-quarter turn. In it goes. Again, put my light on. You see it. The red is the top, the yellow is at the bottom. I go in, puncture, draw the fluid, and come out. It's as quick as that.
I personally performed a little over 1,000 tympanocentesis. I have not had any major complications. I have had a few patients with minor complications like the hole stays open for more than two or three days. I had one where I did hit the posterior wall, and the bleeding was sufficient that there was blood that stayed in the middle ear space for a little over a week, which made me quite nervous, but resolved on its own thereafter.
I have had the privilege of polling some of the major tympanocentesis centers. Some of those people are in this room. Over 10,000 tympanocentesis in primary care, no major complications reported by any of those in the survey.
I am now ready to show you another video. We will need the lights down.
There will be a lot of discussion about the causal diagnosis of otitis media. This video is shown during our workshops, and it has taught me a lot and the other faculty a lot about what we should know a lot about.
This video was developed in cooperation with Drs. Hoberman and Kaleida at the University of Pittsburgh, and we are very grateful for their cooperation. They actually took video with an otoendoscope. It is a lot like a laparoscope that you put in the ear, and you simply take pictures, and they have made some beautiful pictures, and during our course, we show examples of sclerosis, atrophy, retraction pockets. We won't have time for all of that today. I am just going to show you four ears.
The first two are examples. Here is an example of a normal tympanic membrane. You will notice that all the wax has been removed, and here is an easy to-and-fro movement, which occurs with pneumatic otoscopy properly performed when there is an air-filled middle ear space.
Here is the light reflex. Here is the malleus. Our participants actually vote and we record their diagnosis. You will notice that this eardrum is gray in color, it's in a neutral position, that is, neither bulging nor retracted.
It's translucent. You can see right through it including seeing the malleus, and it has a nice normal landmark, notably light reflex in the malleus. This would be a null effusion, a normal ear diagnosis.
Here, in our second example, this is acute otitis media. It is a bulging tympanic membrane filled with pus, limited mobility. Only with positive pressure do you get a little bit of backward movement of the tympanic membrane; with negative pressure, it is so bulging, it can't bulge further.
This is what you might consider a severe acute otitis media, which includes a smattering of hemorrhagic area on the surface of the tympanic membrane. You will notice that it's kind of a mixture of red and white or yellow. It's bulging, it's opaque, you cannot see through it.
There is some mobility, but only with positive pressure, and the diagnosis would be yes, an effusion is present this is acute otitis media.
Now, if you were examining a child and you saw this ear, what would you think? Now, you are getting to look at this ear for 20 to 30 seconds. All the wax is gone, the mother is not breathing over your shoulder, the child is not screaming. What is the diagnosis?
Well, we could argue about that amongst ourselves, but although there is yellow fluid, this eardrum is retracted. We know that from the anatomical position of the malleus. You have got some air fluid levels; 88 percent of ENT physicians say that this is otitis media with effusion, but only 40 percent of pediatricians think it is otitis media with effusion. What is it?
Here is ear number two. Again, all the wax is gone, all the time in the world to look at it and think about it. The average pediatrician looks at an ear for less than two seconds. You are looking at it for 30 seconds. Usually, the cerumen blocks more than 50 percent of the view. What did you think that was? Did you notice the bubbles? Eighty-two percent of ENT physicians thinks this is OME, about 60 percent of pediatricians think it's OME. What do you think?
Do you want to see another one or have you seen enough? One more. The chairman says one more.
I am going to show you the ear. This is going to be a good one, Barth, because watch the malleus. At the beginning of the video, the tympanic membrane is gray. Then, the child starts screaming, and the eardrum turns red, first, a blush down the malleus, then, the whole canal turns red. By the end of the video, everything is red, but at the beginning of the video, everything was gray.
So, when Dr. Giebink and Sylvan Stool and other leaders tell us that color is the worst--there it is, it's gray, folks, but watch a child cry, watch the blush of the capillary bed down the malleus. Here is comes, boom, and then the whole eardrum turns red. What is that, is that otitis media? It's red. It's red.
No, it is not otitis media. There was not even effusion behind that tympanic membrane. That child had a retracted tympanic membrane, probably had a cold or an allergy, and there is nothing wrong with that ear.
That's otitis media again, by the way. Those white flecks are epithelial cells on the surface of the tympanic membrane, pealing off from the heat of the infection. The eardrum is so bulging that when you puncture it, pus explodes out of the tympanic membrane, and the child stops crying on the table from the relief of pain.
Thank you very much.
[End of video.]
DR. RELLER: This has been choreographed by Dr. Soreth. Dr. Ron Dagan perhaps is the only person who would be willing to follow Dr. Pichichero.
DR. DAGAN: If you think I am going to dance, I am not.
Double Tympanocentesis Studies
Ron Dagan, M.D.
DR. DAGAN: I was asked to talk today about the bridging between double tympanocentesis and clinical outcome studies.
This is, as you see, a very bulging eardrum, and just to remind all of us, a double tympanocentesis means that before treatment, we do one tympanocentesis, as you saw now, and we take it for culture, and during treatment, and usually, after 3 days to 5 days, because this is really the middle, but at 72 hours of treatment, this is day 4 to 6, if this is day 1, and then you do another one, and you take for culture.
The double tympanocentesis means that we are going to see whether the organisms that exist here, they disappear on the second tympanocentesis, and then you can compare to drugs or compare to virus MICs or whatnot.
Now, I have a series of seven questions that I have tried to see whether we get answers in terms of bridging, and the first question, of course, in acute otitis media, is there any difference between drugs in regard to bacteriologic eradication of day 4 to 6.
I am not going to show all the slides from all the studies, but I wanted to bring a summary from the recent studies, and this is, as you can see, cefaclor, cefuroxime-axetil, amoxicillin or amoxiclav at the regular doses, ceftriaxone, one dose, azithromycin, 3 to 5 days, trimethoprim-sulfa, ceftriaxone, 3 days, Augmentin ES-600, and the gatifloxacin. These are the recent studies that we have data for.
If we look at placebo, you remember that 84 percent of the 3 to 4 days is still persisting, so this is percent of persistence, and you see they all times were quite nice days, whatever you gave had eradication that was significantly better than placebo. Hopefully, after the pneumococcal vaccination, we will see something more similar to this.
However, the situation with the resistant era, is that you can see all drugs are affected somewhat, and you can see really much differences between the drugs. You can see that for those who are beta-lactams, when you have penicillin, no susceptibility for the macrolides or for the trimethoprim-sulfa, et cetera, you do see much difference between the drugs nowadays in eradication of the nonsusceptible organisms.
We don't have data on quinolone nonsusceptible pneumococci yet, but I think that in a year or two, when we meet, I will bring you probably already resistant quinolones because that is the way it will go if the quinolones will be given to children.
As you can see, there are drugs, such as cefaclor here, ceftriaxone one dose, and azithromycin as presented of the macrolides, trimethoprim-sulfa where really are not very much different than placebo in terms of eradication of the organisms. Others are sort of reasonable, and others may be good.
If you look at the Hemophilus, remember 50 percent eradicated around these and 50 percent persist, and you can see again excellent drugs versus not so good drugs. Cefaclor is not too far from placebo. Azithromycin is in the range of placebo in terms of eradication rate.
You have some sort of acceptable. With trimethoprim-sulfa, if it's resistant, it's not eradicated, but only 30 percent are resistant, so you still have some good results. Here, you can see better results, but there is a big variety.
Now, with Hemophilus, when you have beta-lactamase, then, you have amoxicillin, of course, because then amoxicillin is placebo when you have beta-lactamase.
With Hemophilus, there is much more experience that can be drawn from the past, because in the past, the differences in Hemophilus today and in the past are not as big as with pneumococcus.
So, I took all the studies I could find, which is about 35 or 36 studies all together with double-tap tympanocentesis, and compared to placebo, and you can see that there are two groups of drugs, one group that is ranging from excellent eradication rate to reasonable eradication rate, and this is the number of studies done, not necessarily by our group, but all groups all together, and these are what I think is not too acceptable, cefaclors are frozen of the beta-lactams and the macrolides.
So, you can see that really there is a big difference between drugs in terms of potential eradication exactly on the same timing.
The second question is can double tap studies determine an MIC concentration cutoff, above which a given drug is not bacteriologically efficacious, because now you get for licensure sometimes application which is hooked to an MIC.
This is again one of our first studies looking at cefaclor versus cefuroxime-axetil. These are the placebo eradication rates as found by Howie in the past. Remember that both drugs are good for pneumococcus that are susceptible to penicillin.
If you see a nonsusceptible, and this time we really didn't have resistance only to immediate you could see that both are effective, but you can see one drug that is more effective than the other, and this is sort of the gradual increase in MIC, you can find some cutoffs. Hemophilus is not relevant to this question here in this.
Now, with trimethoprim-sulfa, for example, since MIC of 0.5 is considered to be the cutoff, we wanted to see whether MIC of 0.5 really is associated with eradication failure. You can see that for both pneumococcus and Hemophilus here, you have 37 cases, 100 percent eradication with trimethoprim-sulfa, while if you have above MIC of 0.5, basically, for pneumococcus and for Hemophilus, you have a placebo.
So, again, I think I mentioned last time with a question that will come whether we need placebo studies, we have some placebos here that we don't really need to give placebos, they are as good as placebos for eradication.
With azithromycin, there is now a study where we did from 3 days and 5 days, and pharmacokinetic/dynamic calculation predict a 0.25 or less than 0.25 actually than MIC, below which you should see a response, and above which you should not see a response.
These two studies actually show that for pneumococcus that is susceptible to macrolide, you do have almost 100 percent response, while if it is above that, which is usually above 2, because you don't have really intermediate values, this is basically placebo rate of eradication.
For Hemophilus, there is no Hemophilus of less than 0.25 MIC for azithromycin. Up to 4, it is susceptible, but you can see that for both studies, 3 and 5 days, at 5 days, even one at 0.5, which are not the majority of the cases, you have basically placebo eradication rate, the same goes, of course, if the MIC is higher. They are all acting about the same.
So, definitely, here, in this case, and the previous slide, you could see that there is an MIC where we can really measure above which you are not going to see good results.
This is the Augmentin ES study that was published. The data, you can see that again this is penicillin MIC, and this is pneumococcus, and the majority had an MIC of 1 or less, and you have 100 percent eradication, but you start to see increasing failures with MIC, and as far as I know, the FDA did not approve it for MIC maybe because of this.
For Hemophilus again, 0.5 or less, you don't have all those failures, and then you start to see more and more failures, and we need a little bit more cases to know where it is starting to be unacceptable, but definitely with a double tympanocentesis, you can go down to talk about MICs and for which MICs you start to see problems in eradication.
The third question is, of course, is there a relation between bacteriologic eradication on day 4 to 6 and clinical outcome? I think this is maybe the most important question.
We have two studies actually, only two studies that looked at this because in order to look at this, you need to do double tympanocentesis and to be able to follow clinically, otherwise you cannot correlate those.
One is a study that was mentioned by Carlin, et al, and the other one is ours. This is the cases where you did eradicate the organism. You start with positive culture in cases you eradicate the organism, you see that there is about--no, I am sorry--you don't eradicate. This is culture-positive, about 40 percent would be clinical failures.
If you eradicate the organisms, you get less than 10 percent clinical failure, so I think there is no argument that most of the clinical failures will be those for whom you did not eradicate the organisms after 3 to 5 days. This is very clear.
But if you really want to see how the children feel, you have to start to use some scoring. This was the scoring we use, giving from each one of those from zero to 3, and this was evaluated by an independent ENT who did not know what the children were receiving and what the organism was.
The maximum score is 15, the minimum is zero, and if you look on day 4 to 6, and you try to see how the kids feel by scoring, this is the culture-negative, this is the children that responded to treatment bacteriologically on that time, and you will see that 45 percent have zero or 1 score, and very few have 4 or more. This would be equal or above.
Those who are still culture-positive, you see the difference, a highly statistically significance, very few with zero to 1, and one-third above 4. So, the children in this group definitely feel better than children in this group, and this is the group where you eradicated the organism.
So, there is a correlation, there is no doubt about that, between bacteriological eradication and how you feel after a few days, and how you feel at the end of treatment and whether you fail or not.
The fourth question is can we determine by double tap studies if an organism is not important in acute otitis media?
This is very important actually, because there are some authorities and some manufacturers and some clinicians who think this H. influenzae is not important, and the Hemophilus is relative or absolutely is going to be more important after pneumococcal vaccination than it is now.
There was already data to show replacement of Haemophilus influenzae that replaced some of the vaccine type that disappear in the Finnish study, and definitely now since the vaccine types are going to be reduced, you are going to see maybe less of pneumococcal resistance, but more Hemophilus, so this is a very important question.
One of the studies that was done by our group and presented at ICAAC a year and a half ago was looking at the regular bug that we have now in Israel, which is not very different from what you had in the States before starting with vaccination.
Pneumococci were mainly penicillin-nonsusceptible, Hemophilus with about a third that were beta-lactamase-positive, very few Moraxellas, 43 patients, 56 bugs receiving what is recommended in the status of first liner, 80/kilo amoxicillin, and 13 failed with 16 organisms.
You can see that now we have very few pneumococci that are penicillin-susceptible, the susceptible went away, and you see very clearly that you have now lots of beta-lactamase, and actually, if you look at 13 here, 8 here, so basically what you have got is again this spontaneous eradication of the beta-lactamase production, which is still placebo effect and most of the beta-lactamase not producing went away. Most of the children have beta-lactamase producing organisms.
So, it depends on what drug you have, but with amoxicillin, Hemophilus is definitely a very prevalent one. The question is whether it causes any symptoms.
We have now quite experience with reading, giving a scoring. This is more objective without ear tapping, giving a scoring by the ENT that sees the child before the first attempt, which means it is sort of a blind reading because you don't know what the organism is going to be when you tap, and you score the child.
We have now about 1,000 cases like that where we can start to summarize those. You can see we have 762 that are culture-positive, 240 that are culture-negative, and the mean score is here, and the culture-negative, of course, has lower score despite the fact that all are involved as acute otitis media cases.
Now, if you look at the organism, this is a negative, this is mixed pneumococcus and Hemophilus, pneumococcus alone, Hemophilus alone. You see the numbers are quite big, and you see that the culture-negative has a different score than the culture-positive, and because of the big numbers here, the P is significant between Hemophilus and no growth, and really not different from the others. If anything, this is a little bit higher, not significantly.
So, basically, you can see that Hemophilus really does not have a different score when you see the child, when you look at the tympanic membrane and the fever than pneumococcals when they come to you, and I think that these big numbers really makes it more accurate than the few small series that characterize 1 or 2 or 5 or 20 patients.
Even more important, if you look at eradication, what happened to the score after you give antibiotics. This is just an example of a score that was given before antibiotics. This is day 4 to 6, another score, and what is really important is the delta, and we want to see whether the delta is the same if you did not eradicate and did eradicate pneumococcus and Hemophilus.
What you get is first when you eradicate, the organism is gone, the second test, you see quite a nice big delta, which is no difference between Hemophilus and pneumococcus from mixed infection.
When the organism was not eradicated, the delta is much smaller, and again, not different between those three, and definitely it means that if you did not eradicate the organism, Hemophilus is as bad as pneumococcus, but remember that within 3 to 5 days, you have more eradication of Hemophilus compared to the pneumococcus, so all in all, there will be more cases that will look better with Hemophilus than pneumococcus, but the 50 percent of where you do not eradicate the Hemophilus, are going to look as bad as pneumococcus, and think this is proof that Hemophilus is not negligible at all in otitis media.
This is the next question. Can we bridge between double tap studies and studies with clinical outcome? This is the main question coming from the previous questions.
I use here an example of the one study of the previous dose of amoxicillin or amoxiclav, or the regular dose, if you will, of 45 mg/kg compared to azithromycin, 5 days, and here you see the placebo rate again of eradication, and you remember that, or I am not sure I showed it, but basically, the results were that you have 87 percent eradication rate.
Now, this is not persistence, this is eradication rate with Augmentin and 40 percent with azithromycin, and all together it is high statistically significant. Now, this is bacteriological eradication.
If you look at this, this is what I show now, this is bacteriological eradication of pneumococcus alone. There was some difference, which was not statistically significant, and the overall bacteriological eradication rate was significant.
But when you look at the clinical outcome now, you could see that here, there is no difference, significant difference in clinical outcome. Here, here is a significant difference in clinical outcome, and all in all, you have here 16 percent difference, which is statistically significant clinical outcome.
Now, by doing the double tympanocentesis, this is the clinical outcome, but by doing the clinical tympanocentesis, you know that the main difference that accounts for this 16 percent between the two drugs is coming from the Hemophilus eradication, not really much from the pneumococcus.
So, by doing this, and then doing clinical studies, you are going to see that the clinical studies don't say much different that we have here, but you can know that this is not because of pneumococcal problems, but because of Haemophilus influenzae issues.
Now, if I take this, again, I am surprised, I am the fifth speaker or so, and nobody mentioned yet the Pollyanna phenomenon, but this is what I call--I don't call it anymore Pollyanna phenomenon--I call it the Colin Marchant drum, because this diagram was shown first by Colin Marchant.
Remember, this is the eradication rate after 3 to 5 days. Placebo is very low. One hundred percent is best. You heard from Scott that even if you have 100 percent eradication, you are not going to see 100 percent clinical response here. With placebo, you get up to 70 percent clinical response just because some and most of the organisms go away within 10 days.
The difference is that small here. I take the data that I showed in the previous study, and I try to put them here. So, for pneumococcus with Augmentin, a regular dose, you have about almost 90 percent clinical success rate, which is well located.
With azithromycin, you have 80 percent. It might be a difference or not, it is an issue of sample size, but they are located in the upper 50 percent. If you look at the amoxiclav, Hemophilus, it is here, 87 percent bacteriological eradication, it is well located here.
If you look at the azithromycin, it is located basically in the range of placebo.
So, this is what we saw with clinical response. This is what we saw with bacteriological response. This is just to show you how we are--and I am trying to take this diagram, it's the bridging diagram for clinical studies--and try to see what happens if I put clinical studies on that.
In order to choose that, I took the only one, the only FDA meeting I was in was the previous one, which now it says 7/11, the other one was 11/7, in November, for licensure of one dose and three doses of azithromycin, and I took data from clinical studies that have one tympanocentesis, that were obtained, but clinical outcome, it was obtained by the people who wanted to have the best results because this was shown, this was presented by the people from Pfizer, and you can recognize those slides from work you can download from the Internet.
What you can see here is--I didn't find the slide with macrolide-resistant pneumococcus, but with penicillin-resistant pneumococcus, I could find one slide and, of course, the more penicillin resistant you are, the more it is enriched with macrolide resistance.
They showed, the point was that even if you are susceptible intermediate resistant, although you have a little bit lower response, you still have quite a nice response for all three.
Well, if I put this again in this, what I find here is that penicillin, pen-susceptible pneumococcus has 95 percent success, which is the best you can have, you cannot have better than that, which really is concordant with what we found in our studies on azithromycin.
If you have penicillin, it is immediate, which is enriched with macrolide resistance, you already drop to not too nice results, and if you have penicillin resistance, which is even more enriched, you actually are within the placebo range.
So, with the same drug, in the same study, clinical outcome only, if you put it here, you actually find a very nice distinction although the sample size is not sufficient, but if I bridge it with a double tympanocentesis study, then, this drug should not be approved for macrolide-resistant or penicillin-resistant pneumococci in the States why it was approved.
Now, if I take the Hemophilus versus pneumococcus data for 3 days, 1 day, this is post-treatment, this is EOT, this is after 28 days, and you can see that there is a difference between Strep pneumoniae and Haemophilus influenzae, the same here, the same here.
Again, take those here, and I show 3 days pneumococcus 94 percent, excellent; one day pneumococcus, about the same. It is not a comparative study between those two, so I am not sure that you can deduce anything with the difference especially that it's a small size, but it might be that there is a difference between one day, but they are excellent, both of them are excellent.
This is Hemophilus 3 days, this is Hemophilus 1 day. I think again, what they showed basically is that for pneumococcus, they get an excellent drug, if it is not pneumococcus that is macrolide resistant; for Hemophilus, in my opinion, it should not be approved because it falls into the placebo range.
The next to last question is how do double tap studies help in understanding the best timing for clinical outcome?
We heard end of treatment versus test of cure.
You remember this? Basically, you have still a way to go until here, and we heard about the otitis-prone children, and many of these children are otitis prone, so what happens here?
The FDA elected until now to look at test of cure here, and if there is a clinical relapse, to put it as a failure. What we have documented, and again, about 800 kids we have positive culture here, negative here, and we have clinical relapse, and the question that Scott was asking, do we have really eradication or it is just a suppression, and you get it back here.
Of those kids, we have 108, of the over 800, that came with a clinical relapse, and we were able to do a tympanocentesis, 30 tympanocentesis, and see what happens compared to the previous bugs.
In 20 percent of the clinical relapse, there was culture negative. In 54 percent, it was totally new infection with a different bug. That means, and I will tell you a second what it means. Only in 28 percent, it was a different organism.
So, the real bacteriological relapse was only 28 percent, the majority just reflected the child's otitis-prone nature.
Even if you had a pneumococcus that was replaced by a pneumococcus, when you do serotypes, you find that the majority are not the same pneumococcus. If Hemophilus is replaced by Hemophilus, the majority is not the same Hemophilus, so even if it's the same organism, sort of, it is not the same, it's a new infection.
So, definitely, what I can say, that if you think about EOT versus TOC, definitely, what reflects more is EOT and not TOC, and I think this should be taken into consideration. Again, without the double tympanocentesis, you cannot determine the third one, of course.
The last question. Are the patients that are studied in double tap studies different than those in purely clinical studies?
Because the question is how can we extrapolate from them, and my question is do we need to extrapolate from them, and I will tell you why I ask this.
First of all, yes, they are different. As Scott alluded a little bit to you, in order to be able to get kids for double tympanocentesis, they usually have to be less than 2 years. Older kids are less cooperative despite the nice kid that Mike was showing. I believe this child was deaf, because you heard all those things about the tympanocentesis, and he was smiling.
Tympanic membrane bulging plus pus is not the rule for every single child with otitis, but these are the ones that we take really to tympanocentesis.
Positive culture, you only take the ones with positive culture, and also, as Scott said, they are enriched for more complex acute otitis media, so, of course, they are different kids than the rest, but in my opinion, these are the ones who need antibiotics.
You cannot extrapolate to the majority of kids that get antibiotics because those are diluted by older kids, mild disease, those who don't have otitis at all, and other things, and if I have to really say I don't want them to reflect what is usual to get patients given antibiotics, I want them to reflect the ones that need antibiotics, and I think therefore, these are the appropriate patients to study despite that they don't reflect the rest.
So, in conclusion, double tap studies clearly demonstrate a considerable difference between drugs in regard to their ability to eradicate the pathogens with 3 to 5 days.
Double tap studies can determine an MIC concentration cutoff above which a given drug is not bacteriologically efficacious.
Bacteriologic eradication within 3 to 5 days and clinical outcome correlate.
Double tap studies demonstrate that Haemophilus influenzae is an important pathogen in otitis media.
We can bridge between double tap studies and studies with clinical outcome.
Double tap studies help in understanding that the best timing for clinical outcome determination is EOT rather than TOC.
The patients that are studied in double tap studies are those who need antibiotics more often than patients enrolled in purely clinical studies.
DR. RELLER: Thank you very much, Dr. Dagan, for a succinct, focused delivery and an early arrival.
I should like to have our 10-minute break now. We will begin promptly at 10:50.
DR. RELLER: Dr. Marchant.
Limitations of Clinical-only Studies
Colin Marchant, M.D.
DR. MARCHANT: Good morning. First of all, I would like to thank Dr. Soreth and her colleagues for inviting me and for allowing me to speak.
I have spoken several times before and I am not going to repeat all of that, but perhaps take it a little bit further. As you can see, I am from Boston University, and the teaching hospital affiliated with Boston University is Boston Medical Center.
We have had a number of talented chief residents, but one in particular had an unusual talent. He was an amateur cartoonist, and during grand rounds, conferences, meetings, he will sit with a piece of paper and draw cartoons pertinent to what is going on.
On a morning like this if he were in the audience, he would have at least six cartoons floating around the audience making cryptic comments about what had gone on.
So, I am fortunate to have some of his cartoons and I am going to use them loosely as a metaphor as we talk about some of these things.
Here is his first cartoon. Some people can't see the forest for the trees.
The next cartoon, some people can't find either the forest or the trees. Maybe the cell phone will help.
Some people get lost in the forest.
And some people find a path through the forest, and it is our task to find a path through the forest here, of all this data and all these ideas, all these concepts, et cetera.
This is the slide that Dr. Dagan already showed you. What is important about this earlier was raised the question what is the correlation between bacteriologic outcome and clinical outcome.
Well, the answer, we have facts. We have two studies, and they both came up with the same answer. It may not be the correlation you wanted to see, but this is what the data shows, and in addition to the comments and the details that Dr. Dagan mentioned, the importance of this data is that it validates the bacteriologic outcome. The bacteriologic outcome would not be important if it didn't result in a better clinical outcome when you kill those bacteria, eliminate them from the site of infection compared with when you don't.
This data, specifically the Carlin data, but the Dagan data could be used in the same way, leads us to the Pollyanna phenomenon where excellent drugs look worse than they are, and poor drugs look better than they really are, and then that shows us that there is a very narrow difference at the clinical efficacy level between one drug and another.
That leads to the next issue, which is the sample size issue if we do the double tympanocentesis, we don't need a lot of patients.
Notice, this is in thousands. If we do a single tympanocentesis with a clinical outcome, we have trouble telling the difference between a 90 percent effective drug and 70, we are getting near 2,000 patients here, and if we do clinical-only studies, then, at this end of the graph, we can't really do a study of 15,000 or so patients or particularly to see fine differences, but we even need hundreds or thousands of patients just to do that.
The next issue that comes from this same data is the issue raised by Dr. Giebink, and that is, he said that because there is so many millions of children treated with otitis, we need to think about it because there is just such a large number, and this data allows you to calculate what that burden is with various levels of bacteriologic efficacy, and obviously, the perfect drug, there aren't going to be any children who have persistent symptoms on days 3 to 6 who otherwise would have been better, but even at 90 percent, there is going to be 20,000 per million, 60,000 per million, 100,000 per million, 140,000 per million, and so this data allows us to put some numbers of what is the cost of not finding out whether a drug is efficacious or not efficacious.
So, I am going to cover some design issues. I have put up here that they are all important, yes, they are all important. This is just my preference, order of the day, if you will, but they are all important.
In the IDSA guidelines, they covered these general issues, that trials should be randomized, double-blind, should measure compliance, et cetera, et cetera, but the place that the guidance has fallen down, in my judgment, is where the issues are otitis media specific, which means you have to go to the data on otitis media to get properly designed studies for industry.
I noticed in Dr. Powers' talk that statistical issues will be talked about later, but I say you can't divorce yourself from the statistical issues, you can't divorce yourself from the sample size issues because the sample size is so affected by the outcome, because the sample size, in fact, is affected by the patient selection factor, and the sample size, if you use poor diagnostic criteria and put a lot of non-otitis, we saw Dr. Pichichero's illustrations, it is really not always easy if we don't have good diagnostic criteria, then, we will also drive up the sample size, decrease the power of our trials, so even when we spend time talking about these, they all have sample size statistical implications, and we can't get away from them.
So, I am going to talk in the next few slides about four trials designs - a double tap, a tap at entry to the trial where you then do a tympanocentesis on the clinical failures, a tap at entry with clinical outcome only, and then clinical criteria at both entry and by outcome.
This is just for reference because these slides are in your handout or end up on the web site or what have you. These are the statistical parameters used in the tables that I am going to show you.
Here, in this table, I am showing you if we compare a drug that is very good, 90 percent bacterial efficacy versus tap water or placebo at 30 percent, we look at the number of patients we have got to recruit, the number of taps we are going to do, the number of patients that we analyze.
At this lowest level, we see that the double tap study shows us was small numbers, but also notice that amongst the three studies with the tap designs, we also do fewer tympanocenteses. Yes, they are repeated on the same children, but actually is fewer tympanocenteses that are done.
Now if we do these calculations for a poor drug, the numbers are going to rise in each column. We are now close to 100 with the double tap study. We are close to 300 with a tap and tap of failures, we are already over 1,000 with the clinical outcome.
This relationship remains the same, fewer tapes in the double tap than the tap and tap of failures and respectively the initial tap only. Then, when we get up to a 20 percent difference, and that is equivalent to 40,000 children per millon remaining symptomatic at the time of this second tap, who otherwise would have been better, that difference is going to take you near 300, near 1,000, and up at 4,000.
So, sample size clearly depends on the outcome, the population, even diagnostic criteria, and the minimal standard should be that the trial is large enough to have shown that an antibiotic that was no better than placebo, that it, in fact, was efficacious, so the sample size should really be that large.
Let's think of the high jump. In the high jump, you jump over a bar. If you jiggle the bar and it shakes, it may fall off. When you do a trial, you would like to be jumping over the bar, but, in fact, most of the trials have been ducking under the bar. When you look at the result, the bar is still standing, but you didn't jump over it, you ran under it.
In the November 11th meeting, people went under the bar. So, how large should it be? I am suggesting that perhaps a 20 percent difference in bacteriologic efficacy might be the standard. So, we need to move on. We need to find a path through the woods, if you will, so we need some recommended guidance for industry, so I am going to propose some for consideration in the sample size area.
One of the main, as I look back at previous guidance, at the IDSA guidelines, et cetera, one of the big problems has been that guidance was based on general principles, on expert opinion, and not by going back and saying what does that data say, what does the best data tell you about how the disease behaves.
Sample sizes, if you are going to calculate them, should not be based on assumptions or expert judgment, but based on data, and there is data in the literature that you can use from previous trials to make more informed projections of how you base your sample size.
I have already said that we need to at least exceed the tap water standard proposed the 40,000 children is what we should look at, and we also need to consider the power of subgroup analyses for specific pathogens if we want to look at those.
Previous guidance, the 1998 one had I think arbitrary 25 pneumos, 25 Hemophilus, 15 Moraxella. Where do these numbers come from, how are they powered, what is the chance of showing them, are they going to show anything by looking at those?
So, now let me shift to the outcome, which, of course, is linked to sample size, but the outcome should be directly meaningful like is the child better at 72 hours or 48 hours, as used in the Pittsburgh Kaleida study. That is a meaningful outcome, and the bacteriologic outcome is only meaningful because it has been validated by the data I showed you earlier, the data Ron showed you, the data that Scott showed you earlier.
The outcome should be objective or at least reproducible. The outcome should be sensitive, that is, it has to be an outcome that is affected by antibiotic therapy, and there is data in the literature to tell you what outcomes have been affected by antibiotic therapy, and it should be timely. You have got to measure it at the time point when it is, according to the data, affected by antibiotic therapy.
So, we have already pushed back the test of cure thing as being incorrect. We are now getting closer to the end of therapy, and the end of therapy guideline appears to me that it came from the general guidance in the IDSA recommendations or guidance, and not otitis specific, but just as a general principle that it is at the end of therapy that we are interested in, but many of the outcomes in otitis media, in fact, happen earlier, and if we are going to actually measure them, we need to measure them when they happen, and not at some time later. I have already been vigorous in looking at that issue.
So, if we take the four designs--your handout is incorrect here, it is incorrect here, as well, and this should read increases as you go from double tap to clinical outcome, the sample size increases--but the other important point is we get more information as we climb this order.
Dr. Dagan has showed you that if you do double taps, you can find out what MIC it takes to or what the relationship is for a specific drug and organism and MIC. Pathogen eradication rates, you can only get those if you tap the ears, and then there is the emerging area of PK/PD data, and that has become clinically relevant because Dr. Craig correlated the double tap outcome studies with the serum concentrations and MIC's of organisms, and Dr. Jacobs, in the public session, I believe is going to amplify that.
So, one of the values of going up this hierarchy is that we find out more, it teaches us more, it will help us go in better directions to manage these children.
So, here are the recommendations I would make. We should do double tap studies, and they are preferred for the reasons that I have just mentioned, and a tap and tap of clinical failures is an alternative that if large enough, will also provide useful information.
If clinical outcome studies are going to be done other than symptomatic response, which, of course, will require thousands and thousands of patients, we need to use outcomes that are validated, that are against the clinical response to the clinical outcome.
I didn't spend a lot of time on that, but this was mentioned. Dr. Dagan mentioned it, what should we say we should do for the recommended guidance on population selection and enrichment.
It is these enriched populations, the young, those that fail treatment, those with prior antibiotic therapy in daycare, that are most challenging, and we need data. Clinicians want to have data on how our drugs behave in those groups. We should include those, not exclude them.
Diagnostic criteria, I am just throwing these up into the mix. Yes, they should be symptomatic otitis because that's our goal, is to relieve those symptoms. Yes, we should use some good diagnostic criteria, the kind that have been championed by Dr. Paradise and others, and the other issue that Dr. Soreth has raised, that some folks doing these studies bat 80 percent and some bat 20 percent on their bacterial isolation rate, and those batting 20 percent, we are not sure what disease they are studying most of the time, and we would want to do better.
I have deliberately left the ethical issues until later because I think if we are going to stay out of the woods, we have to think through the science first and then ask the ethical questions, because the ethical questions aren't show-stoppers.
If the ethical questions were so large, we wouldn't even go here, but they are not that large, they are important, but they are not show-stoppers. So, think through the science first, we will get further, and then let's move on to the ethics, and there is more than one ethical question, there is broad ethical questions, as well as focused ones.
Of course, is it ethical to perform tympanocentesis, is it ethical to perform double tympanocentesis? Those are two important questions, but these other questions are important also.
Is it ethical to license, market and prescribe drugs without knowing that they are efficacious? Is it ethical to duck under the bar? Is it ethical to perform drug trials in humans that will not yield scientifically valid data? I suggest no, they aren't.
With regard to tympanocentesis, the question in part is, well, the question about the ethics of it, we haven't really heard from anybody that there is a significant permanent damage.
Dr. Pichichero talked to you about the case of the blood behind the eardrum for a week that made him nervous, but healed. It appears to be a fairly safe procedure, and every day in our country, otolaryngologists do a more extensive procedure. They put tympanostomy tubes in the ear, which stay there for months. They perforate the eardrum, and although there are issues of scarring, and so forth there, many eardrums heal and tympanocentesis is very much a lesser procedure than that.
So, it is primarily the pain of tympanocentesis that is the objection here.
So do the benefits outweigh the risks? I believe this to be true, and therefore, I believe that the benefits of the knowledge gained from properly done studies that are going to give us answers, do outweigh the risks.
That, of course, is a judgment. However, tympanocentesis is still a painful procedure, and in order to move guidance for industry forward, in order to move forward clinical trial design and to get it right, to see a path out of forest and not stay back in the woods, we need to do something else, and that is we need more efforts to find ways to make this procedure less painful and less objectionable.
Currently, it has been pointed out that many practitioners don't do tympanocentesis, and this is true, it is really a very small number of people that do this procedure, many more could, but when something is not familiar with people, they tend to fear it and many of the objections to tympanocentesis come from those who are really not that familiar who fear it, who aren't experienced, not solely, but in many cases. We have more work to do to do that.
So, in summary, then, I offer up for consideration some, not a complete list, it doesn't cover all the issues, but some things that we should offer as a guidance for industry.
One last comment. Dr. Soreth mentioned the problem where you have to be worried about making these things too expensive for industry, and I think that's right, but the first and foremost duty we have really is the public, and in this case, the public is the children, and it's all about how many have ear pain as a result of what our decisions are.
That is what we need to do first. Industry, they are business people, and what they do is they negotiate. That is very much part of their culture and part of what goes on in business. So, when they tell you it's too many, it's too much, it's too expensive, that is part of their negotiating position.
So, you need to judge them by their behavior, and when they stop coming around proposing new drugs for otitis media, then, we will know that we have gone too far in coming to high standards, which are going to get us the data that will help us make clinical decisions and license-effective drugs for this indication.
DR. GIEBINK: Colin, could I ask a quick question of fact here? Okay. On your first recommended guidance slide, the last bullet says, "If clinical outcomes other than symptomatic response are to be used as outcomes, they should be validated."
Do you mean externally validated, internally validated, validated against tympano? I just would like a definition for that word.
DR. MARCHANT: Let me give by example. If you were to propose acoustic reflectometry or tympanometry, or the appearance of the drug on otoscopy as important outcomes, then, those important outcomes have to relate back to what the child care is about, which is whether it hurts or not, just as the bacteriologic outcome has been shown to be important in terms of whether there are persistent symptoms or not, that is the validation that I would be speaking about there, or any other new measure that somebody came up with.
DR. RELLER: Dr. Soreth.
DR. SORETH: Very briefly, a point of clarification for Dr. Marchant. I did not say that we need to be worried about the cost, but rather cognizant that any particular set of recommendations for clinical trial design has implications at the end of the day for cost, and it is just one of many, many factors that are taken together as we are all on the same page about caring for the public, in this case, caring for children who have acute otitis media, and that at times, not necessarily for otitis, but that at times, one can conclude that a set of recommendations in the ideal world are best, but that in the practical world, sometimes cross some line of practicality and doability.
That was really my only point, that in some measure, it is also part of the overall complex equation of what can be done, should be done in an ideal world or in the real world, and that was my only point.
DR. MARCHANT: I didn't mean to put any words in your mouth, and really, what I did, was I extended the issue that you raised with my own view of it is what I did.
DR. RELLER: Dr. Rochester will present for the FDA, Study Designs for Acute Otitis Media Trials: What Can Each Design Tell us?
Thanks, Dr. Marchant. We will have much discussion later on all of the important issues raised and perspectives given.
Study Design for Acute Otitis Media Trials:
What Can Each Design Tell Us?
C. George Rochester, Ph.D.
DR. ROCHESTER: I am George Rochester. I am a mathematical statistician in the Division of Biometrics III, and I am co-located with Division of Anti-Infective Drug Products.
The purpose of my talk today is to discuss the topic briefly, study designs for acute otitis media, and what can each design tell us.
I would like to begin with just a couple of opening works in the sense that when we start thinking about acute otitis media, as well as any other kind of infectious disease, we must have some clarity about what exactly is the question that we want to answer with our study.
Until we have clearly articulated our hypotheses and ensure that we are going after the correct populations that we are studying, we tend sometimes to go amiss in terms of the value and interpretation of what we get out of each study. So, I want us to bear that in mind as I move through these.
The outline of my talk essentially will address three main areas. One will be the role of tympanocentesis, which I abbreviate as TAP, and will speak of as TAPS in acute otitis media trials, and then advantages and disadvantages of each design, and I will speak primarily of two types of designs, the superiority design in which we will refer to placebo-controlled, and the non-inferiority design, which has been the design that we have used mostly in the last probably decade or so.
Acute otitis media represents a spectrum of illness, and I think that has been nicely described by other speakers already. In order to demonstrate the efficacy of a new drug, one needs to provide both clinical and microbiological proof of efficacy.
We must be cautious. We need to guard against post-hoc subset analyses as proof. We have all been confronted with a situation where at the end of a trial, when our data has been analyzed, we get this kind of ah, oops, I think I should revise my protocol here in order to restate my hypothesis for what I wish I had studied now that I have found something.
We have seen where people do become very enthusiastic and very excited because we have seen something that looks really wonderful in a small group of patients that we didn't otherwise anticipate when we started the trial.
I get excited about that, accept that in the context that that generates a new hypothesis that I would like to see studied in a future trial. It may offer certain important reassuring information, but it is not enough for me to call it solid clinical or microbiologic proof.
Then, we want to also guard against extrapolating to populations not directly studied. In the era of evidence-based medicine, where we want to really provide a good, solid foundation upon which to make medical decisions, it is imperative that we understand that having completed a study, having generated the data, that we are very careful when we make extrapolations to populations we did not actually study.
Now, those extrapolations need to have solid scientific pinnings and underpinnings for what we are doing. The temptation is very easy to just say, well, we have studied, you know a group of children from age 5 to 12, and that's just as good for the ones that are under 2, I don't see any reason why not, pain is pain, and so on, and so on.
These generalizations, really, one needs to be careful and very cautious about that.
The current state of affairs, what is the evidence that we are getting now, that we are looking at in terms of a dossier for registration. We tend to get a clinical-only study, comparative in nature, non-inferiority in design, in which we are comparing a new versus a standard therapy.
Dr. Dagan's statement, this actually nicely concurred with our thoughts on this, back and at the November 2001 Advisory Committee, where he said, "Most of the acute otitis media trials with clinical outcome as currently conducted are virtually guaranteed to show no differences between agents, dosing, or duration of treatment."
I would like us to think about this within the context of Dr. Marchant's ethical framework that he just provided, that if we are going to make a study in which we really do not have a real high probability of successfully answering our question, that may call into question our ethics in human trials.
Then, we get another study, which is a baseline bacteriology study, some baseline bacteriologic information at study entry, followed by a clinical outcome at some later time point, usually at end of therapy or at some test of cure, which we might agree on should be different.
That is often non-comparative although not required to be non-comparative, but we often see that people take the path of least resistance.
Why do some trials fail to detect differences among treatments? Well, for one, differences among these different treatments may, in fact, truly not exist. These drugs probably are not different.
We also may have the issue of "noise," and noise in statistical jargon probably means kind of all these things that are confounders that you are probably not controlling very well, things you are not measuring very well, imprecision in terms of how you are carrying out your study.
Sources of noise in AOM studies include enrollment of subjects without bacterial infection at baseline, an example, they have got viral infection, or they probably just have some sort of situation in which, for example, effusion leads to diagnostic confusion.
We have got loose case definitions. We have seen a situation where you have spontaneous resolution even with a bacterial infection, and we have just heard about the tympanocentesis, for example, that it, in and of itself, has some therapeutic value.
So, we are not even sure, that we may go in, perform a TAP, pull out fluid. We have nicely cleansed this nice little pocket of pus, and maybe that, in and of itself, has some clinical benefit to the extent that we are now attributing that benefit to a drug, I am not sure.
Determination of treatment response includes both subjective components, as well as objective components, but the subjective components, in fact, may be subject to significant inter-rater variability.
So, strategies for handling noise would include designing placebo-controlled trials, and for differences observed in the placebo-controlled trial, we know that we can say we have demonstrated a clinical benefit.
We may also have a non-inferiority trial in which we could have either a baseline TAP, which reduces noise in terms of at the diagnostic phase, and we may have a repeat TAP, which actually reduced some noise, as well, in terms of our outcome assessment.
Should TAPS be performed? I think we have heard many other speakers address this issue.
Placebo-controlled trials, in general, will provide clear evidence of clinical benefit, but if you add TAPS to a placebo-controlled trial, then, it does add efficiency to the trial.
Baseline TAP is probably a little bit more critical if we are thinking of the non-inferiority design where "noise" sometimes may lead to a false proof of efficacy.
Then, a follow-up TAP in which we have bacteriologic outcome becomes more objectively determined.
The optimal time and number of TAPS to perform may need further research. I have heard several speakers may use timing of day 3 to 5, some people say day 4 to 6. We do know that if we tap probably too early, it may not be as helpful to differentiate differences between drugs; if we tap too late, it may not be ethical, the children are actually cured, their fluid has gone away, they are fine, they are happy, and so on. People do not feel that may be a good to tap.
However, tapping all failures has always been encouraged, it seems, in all the guidances I have read, however, there is also a difference between clinical trial and clinical practice.
What I have seen in a lot of the studies that come to us for review, is that physicians sometimes forget the difference between practice and a trial. A clinical trial is an experiment in which a protocol has been designed and agreed to, and should be followed.
It ensures uniform documentation and it ensures that we can interpret our data with a certain rigor. In clinical practice, however, a patient appears to a health care provider for care, and that care means that physician has a wide latitude of discretion in the way the patient is ultimately managed.
If, in a trial, you have a protocol and the investigators are not following the protocol, it actually becomes very difficult in order to really interpret and understand the information.
A single TAP at baseline. You have got bacteriological diagnosis and a clinical outcome assessment, that is the standard trial we have been talking about in a non-inferiority setting.
The baseline TAP ensures that patients in he primary analysis have baseline pathogens. It is better than having no TAPS, but the bacteriological outcome is presumptive if we are going on a clinical outcome assessment to determine success or failure.
In practice, failures do not usually get follow-up TAP regardless of what the protocol specification is. A non-inferiority with baseline TAP may allow a wider non-inferiority margin which leads to a smaller sample size, and Dr. Marchant did speak about sample size actually quite nicely, so I won't go further into that.
Repeat TAPS provide objective bacteriological outcome. Blinding in this situation is not as critical for the bacteriologic endpoint, but it is essential to reduce bias during study if the clinical outcome is the ultimate goal.
Study is successful, though, if efficacy is shown at both the microbiological and the clinical assessment time points.
Fundamental question regarding the utility of a microbiological endpoint. Bacteriological endpoint is a surrogate and the correlation with clinical endpoint sometimes may be less than satisfactory given current data.
So, I think until we are really certain of whether or not we can truly predict the clinical course or the ultimate clinical outcome of this patient from the bacteriologic data, bacteriologic endpoint, then, it needs to be seen as probably a co-primary kind of information with the clinical outcome.
I am not sure if we are at the point in the literature where we can say we can substitute one for the other.
Much uncertainty still remains about the bacteriological endpoint.
The Agency for Healthcare Research Quality Evidence document, published in 2001: Management of Acute Otitis Media, makes the following quote that I find very useful:
"There is still a need to adequately address he role of antibiotics in the initial treatment of acute otitis media in children compared to placebo or observational treatment especially in terms of various influencing factors such as age and otitis-prone status.
"Close monitoring of patients in these studies with a priori plans for appropriate intervention should allay any concerns about suppurative complications and should also be a focus of research."
So, when we are talking about any trial in a pediatric population, children fall within a group that we consider vulnerable populations who deserve significant additional protections.
So, whether you are doing a placebo-controlled trial or a non-inferiority trial, it is important that we have an ethical framework, such that children are monitored carefully and all strategies that are important to protect them from any harm is actually in place and followed.
The randomized, double-blind, placebo-controlled trial is kind of what I am thinking of when I say placebo-controlled trial, and that is the gold standard. It is efficient and easy to interpret, it provides direct evidence. We may consider a three-arm trial in which we have a new drug, a standard drug, and a placebo.
We want to have certain features of blinding, randomization, all of which ensure that we are minimizing the bias that can be present during study conduct, and, of course, the placebo helps in terms of giving us direct estimate of the treatment benefit, and the placebo-controlled information is what becomes the scientific foundation on which to plan future trials.
So, advantages and disadvantages are that the placebo-controlled trial will provide clear evidence of a clinical benefit. If TAPS are added, it will improve the efficiency of the trial and provide direct bacteriological information and obviously may help with a smaller sample size than a non-inferiority design. Once we add TAP into the design, it also improves upon efficiency of one that wouldn't have had a TAP.
A disadvantage would be that one treatment group is untreated, and that could be taken two ways. You may say one group did not get treated, they ultimately could have not reaped the benefit that it could have otherwise had if it turns out to be useful, but they also were not exposed to any of the toxic effects that they could have experienced on drug, so to some extent, that could be an advantage or a disadvantage. If no TAPS are done in the placebo-controlled trial, certainly an additional microbiological study would be necessary and preferably in a comparative study.
Non-inferiority trials. You are comparing a new drug against a standard. Your estimate of the treatment benefit will depend intricately upon knowing the benefit of the standard over placebo.
Efficacy here is indirect and is demonstrated only if we actually knew that the control itself would have had a benefit over placebo. The choice of non-inferiority margin will depend upon microbiologic rigor, as well.
Advantages of this one include acceptability, all patients get treated, so parents probably may sign up for this one more readily. It does provide some comparative clinical information.
But I couple of the disadvantages I want to point out are that bacteriologic infection may not clearly have been established at baseline if you have no baseline TAPS, and over time, the magnitude of the initial benefit of the control may not be maintained.
So, this one may not give us real good assurance that the new drug could actually beat placebo.
In a non-inferiority design with a baseline TAP added, then your additional advantages would be that you have better microbiologic diagnosis, setting your non-inferiority margin becomes a little bit easier, but a clear disadvantage is that determination of efficacy is still indirect and relies upon clinical judgment, because the outcome is being measured as a clinical response.
Certainly, with a repeat TAP, we now can assess two endpoints. We can assess a delta for a micro, which is our overall microbiologic response, we can assess for clinical response, and certainly a combination of clinical and micro endpoints would be what we would call a successful trial.
So, which design to use? If you want to demonstrate absolute efficacy, and a placebo-controlled is your design, if you want to demonstrate absolute and relative efficacy, then you can consider a three-arm trial in which you can compare new drug to placebo, new drug to the old drug. We get relative efficacy and, of course, we have a placebo arm there.
Now, if the magnitude of the advantage of the active control over placebo is known for the primary endpoint, then, we could consider a non-inferiority design, and the ICH E-10 gives us some advice probably on how to consider setting those non-inferiority margin.
The basis idea is be conservative if our historical information is poor or if it is not relevant. Do not extrapolate beyond the strength of your data.
So, what does each design really tell us?
In a placebo-controlled setting, we know that the new drug beats the control and so it shows a clear clinical benefit among the patients studied.
If have a non-inferiority design and with no TAPS, then, all we are saying is a difference in clinical success rates is less than some non-inferiority margin delta that we set.
If we have a baseline TAP, then that difference is within the delta, but with patients with baseline pathogens.
If we have one in which we have repeat TAPS, then, we have an observable difference in both a microbiologic endpoint and a difference in the clinical endpoint.
So, in summary, TAPS do improve the efficiency of AOM trials. Repeat TAPS provide objective microbiologic information in which to judge not only the subjects who are successful at the end, but it also helps us to understand why subjects are failing.
Placebo-controlled trials are efficient, easy to interpret, provide direct evidence, and the non-inferiority design, microbiologic rigor can improve the quality of those trials if the benefit of the standard over placebo is known.
Then, we come to the real question, when we are setting all these studies up, what it is we really are interested in, is the microbiological or the clinical endpoint more desirable to patients, what it is that we really, truly are interested in at the end of the day? So, bear that in mind as we proceed with the discussion for today.
I just want to thank the other members in our Division of Biometrics III, who contributed to this presentation.
DR. RELLER: Thank you, Dr. Rochester.
Dr. Smith. Lessons Learned from Past Approvals.
Lessons Learned from Past Approvals
Thomas Smith, M.D.
DR. SMITH: Thank you.
In this presentation, I am planning to use some examples from recent approvals to highlight specific areas of the current draft guidance where we have had problems and where we would like to get the committee's advice as we prepare to make revisions.
The current draft guidance speaks of two clinical trials. The first one of these is a statistically adequate and well-controlled multicenter trial that uses rigid case definitions with specific subjective and objective diagnostic and effectiveness parameters clearly defined.
We have heard from Dr. Pichichero's presentation and from some of the other speakers today of some of the difficulties with these rigid case definitions and the fact that the diagnosis is not always so easy to make.
In these studies, baseline tympanocentesis need not be performed, and as a result, in fact, most of the trials that are submitted to us are clinical-only trials. Tap of failures is strongly encouraged to document inadequately treated pathogens.
Again, the taps of failures are rarely performed in studies even though the guidance recommends it and, in general, the protocols that are submitted also strongly encourage the tapping of failures.
The second trial is a tympanocentesis trial. The guidance actually is silent on whether this trial should be comparative or non-comparative and, as a result, most of the trials that are submitted are non-comparative in design.
These trials should establish acceptable outcome in at least 25 patients with Haemophilus influenzae, 25 patients with Streptococcus pneumoniae, and 15 patients with Moraxella catarrhalis.
Tap of failures is strongly encouraged. Again, even though baseline tympanocentesis is done in this studies, failures rarely get tapped.
This is an example from our most recent approval, which was actually for a labeling change in which the applicant very closely followed the recommendations of the current draft guidance and submitted as the two major trials, a clinical-only trial and a non-comparative tympanocentesis trial.
The clinical-only trial was a double-blind, double-dummy, randomized trial that enrolled 350 patients from 9 United States sites. The ages of the children eligible for the trials were 6 months to 12 years, and 60 percent of the children turned out to be over 2 years of age.
The clinical outcomes from this study are presented here. I have shown both the end of therapy and test of cure results. Although the current guidance uses the test of cure, which is at day 28 to 32, in this study as the primary outcome, the committee recently voted unanimously that the end of therapy clinical outcome was of greater value.
These results are typical of most of the clinical-only studies in acute otitis media in that you have high end of therapy success rates, which are somewhat lower at the test of cure visit. The other thing to notice here is that there is no difference between the drugs. There is a satisfactory confidence interval around the treatment difference.
The second trial submitted as part of this package as a tympanocentesis trial, which was an open-label, non-comparative trial with baseline tympanocentesis. 248 patients were enrolled from 22 U.S. and Latin American sites.
The ages of the eligible children were 6 months to 12 years, and in this study, 65 percent of the children were over 2 years of age with a mean of 3.4 years. Fifty-one percent of the children who had tympanocentesis had positive cultures.
Clinical outcomes by pathogen are presented here, and I simply presented them for the end of therapy visit. The overall success rate at the end of therapy was 89 percent, which is consistent with what was seen in the earlier study that was presented.
For the individual pathogens, the point estimates for successful clinical outcomes ranged from 71 percent for Haemophilus influenzae to 100 percent for Moraxella catarrhalis.
These data were presented before the Advisory Committee in November 2001, and there was a great deal of discussion that was generated. Much of it centered around the limitations of clinical-only trials, the fact that you are relying on a clinical diagnosis of otitis media, and that this necessarily includes a lot of patients who do not have bacterial disease.
There were issues raised with the microbiologic data, questions about some of the point estimates presented and about the non-comparative nature of this data. There were comments made also concerning the age distribution of the patients and the fact that the population in this study was not representative of the population where the incidence of acute otitis media is greatest.
Finally, there were several calls from the committee members for the revision of our draft guidance.
A couple of months later, in the Pediatric Infectious Disease Journal Newsletter, there was a comment by Drs. Nelson and McCracken to the effect that, "The supporting studies for these two regimens have shortcomings, similar to studies of other therapeutic agents in acute otitis media. It is time for the FDA to establish strict criteria for conducting clinical trials in patients with acute otitis media if a new antibiotic is to be approved for therapy."
"Such clinical trials should include a predominance of children younger than 2 years, a tympanocentesis at diagnosis to establish etiology, a repeat tympanocentesis at 4 to 5 days in a subset of patients to establish bacteriologic cure or a repeat ear tap in patients who are considered clinical failures, and follow-up evaluation at 10 to 14 days as the primary clinical endpoint."
I think the example of this recent approval raises a couple of the major issues that we would like the committee to address in the first question for discussion today. These issues are the value of comparative studies with diagnostic tympanocentesis, and these studies might be single tap, double tap, or some combination, and also the issue of the future role of clinical-only studies.
Now, another area of the study considerations of the current draft guidance talks about the listing of pathogens, and it states that pathogens listed in the label should have acceptable eradication rates.
These rates are not otherwise defined in the guidance. It does state that if a product fails to have acceptable clinical and microbiologic effectiveness against all three major pathogens, it should be listed only for those it has eradicated. This would take the form of a restricted listing as not a product for first-line therapy.
This restriction is based on the empiric nature of treatment and the need for first-line therapies to be effective against all common pathogens.
I have here a couple of examples of pathogen labeling in which products have not achieved approval for Streptococcus pneumoniae.
This first one is for otitis media caused by Haemophilus influenzae, Moraxella, and Group A Streptococci. The clinical study section states that the response rate of Strep pneumoniae to this drug is approximately 10 percent lower and that of Haemophilus influenzae or Moraxella catarrhalis approximately 7 percent higher than rates of these organisms to the active control drugs.
The second label is for a product, which again is approved for acute bacterial otitis media due to Haemophilus influenzae, Moraxella catarrhalis, or Group A Strep. There is a note here that although this drug used empirically was equivalent to comparators in the treatment of clinically and/or microbiologically documented acute otitis media, the efficacy against the pneumococcus was 23 percent less than control. Therefore, this drug should be given empirically only when adequate antimicrobial coverage against Strep pneumoniae has been previously administered.
The clinical study section of this label contains a table showing bacteriologic eradication rates for the pneumococcus for this drug of 65 percent versus 88 percent for the active control.
This example demonstrates two important points. First, it shows some of the problems with restricted labeling in situations in which a drug is approved when it lacks acceptable efficacy versus all three major pathogens.
This is labeled as a second-line drug which is indicated empirically for treatment failure only when adequate coverage against the pneumococcus has been previously administered.
In an era of increasing pneumococcal resistance, however, many formerly adequate therapies no longer are adequate and the treatment failure population for whom this drug is prescribed actually had a disproportionate share of resistant Strep pneumoniae compared to the general acute otitis media population.
I think the second important point from this example is that it demonstrates the importance of having comparative rather than non-comparative microbiologic data in evaluating pathogen-specific efficacy.
Among the issues for discussion then related to the microbiology of acute otitis media are whether it is important for a drug to demonstrate efficacy against all the major otitis pathogens in order to obtain approval, whether per-pathogen efficacy should be demonstrated using comparative as opposed to non-comparative data, and whether it is feasible to have objective criteria for the inclusion of individual pathogens in the label.
I would like to talk briefly about some issues with inclusion and exclusion criteria. The draft guidance states that among the inclusion criteria, clinical-only trials ordinarily should not enroll children less than 6 months old.
There is no recommendation, however, in the guidance about the actual distribution of the children in these studies. This lack of guidance has resulted in several instances of submissions that contain unrepresentative study populations.
I have a couple of examples here of products where one product in the 90s, although the tympanocentesis study that was submitted had 44 percent of the children under age 2, the large clinical-only study had less than 20 percent of the enrolled population that was under age 2, and had a median age of 4 1/2 years.
We have other approvals from the 90 of products where another product, as part of the package submitted, one clinical-only and two tympanocentesis trials, all of which enrolled only children from 2 to 15 years of age.
Even the most recent supplement that I have described for you in the two major studies that were submitted, 60 to 65 percent of the children were over 2 years of age.
We have heard from the speakers today and from previous committee meetings that when you consider that the peak incidence of acute otitis media is between 6 and 18 months of age, the fact that these children have lower rates of successful treatment, it seems that we should be considering whether the future guidance should include some type of recommended age distribution for future trials.
Under exclusion criteria in the current guidance, children with tympanostomy tubes, children with acute otitis externa are excluded. Recent systemic anti-infective therapy for clinical-only trials, children treated within the 7 days prior to enrollment are excluded, and for clinical and microbiologic studies, children receiving systemic therapy 3 days prior to enrollment are excluded.
The guidance also recommends exclusion of children who are receiving antimicrobial prophylaxis for recurrent otitis media. I think we have heard today and particularly for studies in which baseline tympanocentesis is going to be done, and you will have bacteriologic confirmation of the etiology of the acute otitis media, that it certainly seems reasonable to allow for the inclusion of these children in acute otitis trials.
The issues for discussion then related to these inclusion/exclusion criteria issues are the age distribution of children enrolled in trials and whether there are other methods of capturing populations of greatest interest, where the exclusion criteria, as I mentioned, the issue would be to permit enrollment in clinical/micro studies of recently treated patients and patients receiving prophylaxis.
The final topic regarding recurrent guidance, and this, the committee has already voted on, is the timing of outcome assessments. The current guidance recommends study evaluations at entry, on-therapy, which is 3 to 5 days into therapy, there is a visit strongly recommended.
The end-of-treatment visit is actually optional in the current guidance, and the recommended test-of-cure visit is 2 to 4 weeks after study entry with an optional late post-treatment visit.
The current guidance uses, as the primary endpoint for both clinical and microbiologic assessments, the test-of-cure visit at 2 to 4 weeks after entry.
The committee recently has voted on this issue, and in regards to clinical outcomes, the committee unanimously voted that the relevant clinical test of cure is at the end of therapy, with the later follow-up visit, meaning the one that we currently use as the test of cure, being an important secondary endpoint.
Furthermore, in studies that contain a repeat tympanocentesis component to assess microbiologic response, the committee voted that the most informative repeat taps were on therapy, followed by those obtained at the time of clinical failure.
In summary, then, regarding the general indication of acute otitis media, we would like to get the committee's comments during today's meeting, and we would also appreciate other comments in the form of written comments to the docket, regarding some of these issues here - the value of comparative studies with diagnostic tympanocentesis, the role of clinical-only studies, how best to demonstrate efficacy against all the major pathogens, and issues regarding the inclusion of pathogens in the label.
Changes in recommendations for the age distribution of children who are enrolled in these trials, and limiting the exclusion criteria to permit enrollment of recently treated patients, and patients who are receiving prophylaxis for children who are in tympanocentesis studies.
The next speaker will be Dr. Rosemary Johann-Liang, who will be talking about design issues for studies targeting acute otitis media in special populations, particularly as it relates to recurrent otitis media in kids with treatment failure.
Recurrent/Treatment Failure AOM
Rosemary Johann-Liang, M.D.
DR. JOHANN-LIANG: I am delighted to speak before the committee one more time, although today's topic is very different from yesterday, and I had the pleasure of being the last hurdle before all of us and lunch.
Today's topic is on recurrent and treatment failure acute otitis media. As we consider revisiting the current guidance, we have heard quite a lot this morning about clinical trial designs.
I would like to draw your attention now to the types of children who will populate these clinical trials. Specifically, we will be discussing the proposal for an additional indication that will study the population of children with recurrent and/or treatment failure acute otitis media.
I will be following this outline. The relevant sections in the current guidance will be first shown, then, the rationale and proposal for change will presented. This will be followed by the discussion of definitions and the types of trials for the indications. I will end with some issues we hope will be included in the committee's discussions this afternoon.
The 1998 draft guidance taken after the 1992 Points-to-Consider lays out study considerations for one all- comers indication of acute otitis media. There is no differentiation of different populations, however, there are exclusion criteria and they include the following: children who have received systemic anti-infective drug product in the previous 7 days prior to enrollment in the clinical-only study, systemic anti-infective drug product in the previous 3 days prior to enrollment in the clinical micro study, and patients receiving antimicrobial prophylaxis for recurrent otitis media.
Various beta-lactams and macrolides have been approved thus far under one indication by studying all-comers population with these exclusions.
You all have been telling us that changes need to take place to the current guidance. Of the various recommendations for change by the committee, these are a few of the advice we have heard regarding populations to study.
Dr. Leggett's statement from last year's November meeting - "There was a thing about not being able to use antibiotics within the last 7 days of the last month. I think that would be another way to actually enrich the resistant population because isn't that who we have the trouble with, the more severe illness and the more resistant pathogens?"
Dr. Wald's comment - "I think that groups of children that we should be studying are children with severe disease."
You have also heard Dr. Giebink and the other experts this morning so wonderfully discuss population issues.
So, in thinking about this in picture format--and I would like to ask for your indulgence at this point, all my PowerPoint diagrams are conceptual in design, and not proportional and not drawn to scale--we have the all-comers population for acute otitis media in the large green oval.
You are telling us that the recently treated population should not be excluded, in fact, they should be perhaps studied more in depth. You are also telling us that the population with severe disease should be especially studied.
What is the driving force behind these proposals for change? I think you will all agree with me that the underlying factor is resistant pathogens, specifically, at this point, PRSP. PRSP is a critical factor for otitis media disease in general, but a problem of greater magnitude in these subpopulations.
Rising to meet the challenge of resistant pathogens in otitis media disease, drug development programs are already ongoing. I would like to spend the next several slides briefly reviewing with you the lessons we have learned and are continuing to learn from looking at these examples of drug development programs.
I want to share with you this morning two examples, the high-dose formulation of Augmentin and the development of fluoroquinolones in pediatrics.
High-dose Augmentin, the 14 to 1 formulation was presented to this committee in January of last year. As you are aware, the high-dose formulation was developed with PRSP in mind, and enrichment strategies were used in its clinical trials to maximize patients with bacterial disease especially PRSP.
However, the restricted subpopulation that this formulation is currently labeled for was not prospectively defined and therefore not the defined population studied during development.
How does this label currently read? It says, "Augmentin ES-600 is indicated for the treatment of pediatric patients with recurrent or persistent acute otitis media, characterized by the following risk factors: antibiotic exposure for AOM within the preceding 3 months, and either of the following - age less than or equal to 2 years, daycare attendance."
This recurrent or persistent indication was inserted post-development following this committee's advice that this 14 to 1 formulation should be differentiated from the 7 to 1 formulation, and should not be used for routine acute otitis media.
The lesson learned here was that the population that the indication will be labeled for needs to be pre-defined.
Next, I would like to walk you through a time line of a series of recommendations by this Anti-Infective Drugs Advisory Committee on the development of fluoroquinolones in pediatrics.
The story starts in 1989 where the committee recommended that mainly due to safety concerns, fluoroquinolone development in pediatrics should be restricted to older children with severe underlying diseases of cystic fibrosis and cancer needing therapy for gram-negative resistant pathogens.
The committee met again in 1993 regarding this matter and recommended expanding the types of diseases and age, but again unanimously voted that this class of drugs was not for investigation in routine indications.
By 1997, there is a change. There was again the recommendation to continue the pediatric study of these drugs for severe indications, however, the committee began to discuss the development of these drugs to treat the sick subpopulations of generally well children due to the increasing emergence of gram-positive resistant organisms.
This is a statement by Dr. George McCracken from that committee meeting. "The fluoroquinolones could then be evaluated in hospitalized pediatric patients with community or hospital-acquired pneumonia and possible middle ear or sinus infections caused by resistant pathogens, PRSP, i.e., recurrent or persistent otitis media."
Currently, the development of fluoroquinolones for use in pediatrics is ongoing, and it is not just for severe indications, but also for the sicker subpopulations in routine indications, such as acute otitis media.
One example is the gatifloxacin development program, parts of which were presented at the 41st ICAAC last year. That sicker subpopulation within the acute otitis media being studied with gatifloxacin is called recurrent and/or non-responsive otitis media.
Clearly, the committee has pointed out throughout the time line that I have just presented to you that fluoroquinolones are not for study in routine cases for routine indications due to the safety issues especially the arthrotoxicity, the fact that many other alternative drugs are available for routine use, and the worry of more resistance if this class of drugs are to be used widely in pediatrics.
To summarize what we have heard from you and the lessons learned regarding populations for study in acute otitis media, you have told us to enrich the populations for study for better yield of patients with bacterial disease especially those with PRSP, and that this may be accomplished in part by studying the subpopulation of children with recurrent and/or persistent disease, and not to exclude children recently exposed to antibiotics.
Furthermore, you have told us that the drug development programs geared towards treatment of resistant pathogens, especially PRSP, should not be pooled together for study in routine use. This is due to safety issues at the individual level and the judicious use of drugs to curb more resistance at the public health level.
All in all, what we have learned is that this not for routine subpopulation of acute otitis media need to be precisely defined as we move forward in developing drugs for resistant pathogens.
So, bringing together all that you have told us through multiple advisory meetings, we have a possible solution to propose. The proposal is for an additional indication termed recurrent and/or treatment failure acute otitis media.
This is a population-driven concept. I think it is fair to say that we would all agree that the child coming into the office with an occasional episode of acute otitis media is a distinct entity in comparison to the child that is constantly in the office with multiple and frequent episodes of acute otitis media requiring repeated and cycling of therapy.
The proposal for change then is that we go from the one all-comers indication that is currently in guidance to two indications relevant to the targeted populations, one for routine acute otitis media, and one for the recurrent treatment failure AOM.
This would, in turn, facilitate drug development programs by pre-defining the appropriate populations for clinical trials. For example, a regular dose beta-lactam being studied here for routine AOM, going on to be labeled for this indication at the time of approval, while high-dose formulations or fluoroquinolones being studied here, will eventually be labeled for the indication of recurrent treatment failure at the time of approval.
It is also possible for a drug without particular safety or resistant pattern concerns and having necessary efficacy parameters, may be able to pursue both indications concurrently with data from both programs complementing and supporting that overall development program.
A simple illustration of this concept may be as follows: In choosing the clinically distinct populations as the basis for separating out the indications, we will be able to clinically distinguish the population that will be studied under routine acute otitis media here in the large pretty pink color from the recurrent and/or treatment failure disease here on the smaller green oval.
The resistance factor will overlap both populations, but will have a greater overlap for the not-for-routine indication.
With that change in general concept in mind, let's spend a few minutes on defining the elements of the additional indication, so that as we revise the guidance, we can reflect the consensus that was reached on this concept and be precise with our definitions. Defining exactly what we mean by the terminology used will provide a clear channel for communication by all interested parties and avoid confusion.
First, the definition for recurrent. Are we correct in hearing from you that recurrent otitis media should be part of the not for routine population for study? The generally accepted and used definition for recurrent AOM is shown here: greater than or equal to 3 episodes of AOM over the last 6 months or greater than or equal to 4 episodes of AOM over the past year.
This population of children includes children with various underlying and predisposing factors to acute otitis media including young children with anatomical immaturity. Clinically, this definition would encompass the children thought of as a distinct entity.
Microbiologically, however, when the literature is carefully scrutinized, this population defined exactly and precisely, as shown here, may not have significantly higher rates of PRSP when compared to age-controlled children with routine AOM.
Next, the definition of treatment failure. Are we correct in hearing from you that children recently treated with antibiotics or early treatment failure should not be excluded from clinical trials for AOM, but rather be studied vigorously since this is the population that microbiologically appears to have higher rates of resistance?
One definition, then, one definition that we may be able to propose here is this. During therapy: No improvement observed in signs and symptoms of acute otitis media after at least 48 hours of antibiotic management, or post-therapy: Presentation with signs and symptoms of acute otitis media within 7 days of completing a course of antibiotics for acute otitis media.
This definition is inclusive of the accepted definition of persistent acute otitis media, signs and symptoms continuing on the third day after start of therapy, while being exclusive of the time point beyond 1 week after end of treatment, where it becomes very hard to differentiate reinfection from new infection.
Now that we have proposed some definitions for what the elements of the new indication might be, I want to clarify what the new indication is not synonymous with.
Some terms that we have been seeing in recent protocols that are used as names for subpopulations of otitis media are: difficult-to-treat otitis media, otitis-prone children, hard-to-treat otitis media, and children "at risk."
For example, we have been seeing protocols wanting to study the hard-to-treat or difficult-to-treat acute otitis media with high-dose formulations or fluoroquinolones that has the listing under the inclusion criteria of less than or equal to 2 years, daycare attendance, or 3 or more siblings, et cetera.
This would mean that even with the first-time otitis, just by being a 6-month-old infant, that infant will be exposed to drugs like fluoroquinolones, for example, which I don't think is what anybody wants at the moment.
These listings are then not the elements of the proposed new indication, but rather enrichment strategies to yield patients with bacterial otitis media especially PRSP otitis media for both indications.
Again, the two distinct populations are shown here in the pink and green ovals with the resistance factor overlapping both populations. The enrichment groups are overlaying both indications and the resistance factor.
Now, taking into account all the definitions that have been discussed, I would like to walk through a series of possible scenarios in the next slide.
This is an illustration of a hypothetical AOM drug development schema. Having used enrichment strategies to increase the chance of having a patient with bacterial otitis media, a 6-month-old infant in daycare full time is identified.
If this baby is in the office with his first episode of AOM or has now grown to be a 9-month-old and is having a second episode of AOM, for both of these scenarios, the infant will be studied under the indication of routine AOM, enrolling in drug trials seeking first-line therapy.
If this baby has treatment failure OM meeting the predefined definitions or is now a 12-month-old and is always in your office because this is the fourth episode of acute otitis media, this infant will be studied under the indication of recurrent/treatment failure AOM, enrolling in drug trials seeking not-for-routine therapy.
So, I have presented to you what you have told us about targeted populations and have laid out for you our responsive proposal of relevant indications corresponding to appropriate drug development programs that can move forward with prospectively defined populations that needs consensus on precise definitions.
In the next slide, I would like to highlight some particulars about the types of trials that would be part of the drug development program for this additional indication.
We would be looking for well-controlled single or double tap tympanocentesis trials with non-inferiority or superiority design with pathogen-specific diagnosis by tympanocentesis at entry.
For single tap studies, the primary outcome assessment will be clinical at end of therapy, and for double tap studies, the primary outcome assessment will be on-treatment micro and end of therapy clinical.
I might mention here that if the claim for PRSP is being sought for the label, it may be particularly valuable to include a double tap trial in the drug development program.
These two types of trials may be supplemented by empiric or actual use therapy trials to increase the safety information for the product. This type of trial is particularly encouraged for new molecular entities, drugs with specific safety issues, or drugs with limited safety data and should be inclusive of children with various underlying conditions.
Non-comparative double tap trials may be another supplemental study in cases where efficacy data on a specific organism, for example, needs more support. Relevant studies from "other" indications may also provide supplemental information.
Finally, I would like to show a broader schema for our considerations regarding the types of trials for acute bacterial otitis media overall.
As we consider revisiting the guidance on acute otitis media, this is a summary overview of our overall proposal.
We have heard from you that acute otitis media should be studied in a microbiologically-driven, comparative manner with populations enriched to yield the patients having bacterial disease under the indication of routine acute otitis media, drug development programs for regular beta-lactams, macrolides, et cetera, or new drugs can proceed.
The types of trials for study in this indication would include single tympanocentesis trials, double tympanocentesis trials, placebo-controlled trials, and other supplemental studies.
Under the indication of recurrent and treatment failure otitis media, drug development programs for high-dose formulations, fluoroquinolones, or other new drugs can proceed.
The types of trials for study in this indication include single tympanocentesis trials, double tympanocentesis trials, empiric therapy safety trials, and other supplemental studies.
We have arrived at this overall conceptual proposal in response to your recent recommendations for change by incorporating what you have told us and the lessons that we have learned.
We would like to turn this proposal back to you now for discussion and further advice. Some items for discussion are listed here for you. We would like to know if you agree with the definitions for recurrent AOM, the definitions for treatment failure AOM. Do these two groups fit the population to pre-define for "not-for-routine" drug development programs with PRSP emphasis?
Is it reasonable to have these two groups be placed together in the new indication?
Are the types of trials for this indication appropriate? Can you suggest any other types of studies?
Thank you so much for your attention and we look forward to listening to your discussions this afternoon.
DR. RELLER: We have had a packed and informative morning. It is just after 12:15. This is the plan for the afternoon with a reward for promptness and punctuality.
At 1:15, we reconvene. There will be a 20-minute open public hearing. Dr. Jacobs is the only speaker. When you look at your schedule, that would bring us to 1:35. Thereafter, if you take one-half hour off all the listed times, we will finish at 3:30 p.m. Stick on schedule and we will be done at 3:30 for the people meeting the commitments for flights including international ones.
[Whereupon, at 12:15 p.m., the proceedings were recessed, to be resumed at 1:15 p.m.]
DR. RELLER: We will begin the second half with the open public hearing, actually presented by a colleague of all of ours in the field, Dr. Michael Jacobs from Case Western Reserve University.
Open Public Hearing
DR. JACOBS: Mr. Chairman, committee members, advisers, guests, I am pleased to have this opportunity to give my thoughts on this complex area and while I will be giving you a lot of information, I will try and make the points that I want to make clear about the issue of what the problem is with respiratory tract infections and using antibiotics, and validity of evidence for using those.
One of the points I wanted to make is that otitis media is a very good example and we probably have the best data on respiratory tract infections for otitis media, but most of what I am going to say applies to other respiratory tract infections as well.
Now, one of the big limitations we have with respiratory tract infections is there is a high rate of spontaneous resolution that makes it difficult to show differences between agents.
Bacteriologic outcome studies are not often performed due to necessity for invasive procedures, and you have heard a lot about those. Most studies are therefore designed to show equivalent clinical outcome between established and new agents, and what that means is that if there are inadequacies of agents, they are often not apparent.
I found this slide that Dr. Soreth showed very interesting and very informative. In the absence of culture of middle ear fluid, no specific claim can be made regarding the effectiveness of any anti-infective drug. This statement was in force in 1977, and this was a very important year for me because that was the year I started working on the pneumococcus and found the multi-resistant pneumococcus, and I hope that we can go back to this statement.
Now, some of my objectives are to define pharmacokinetics and pharmacodynamics because this gives us a basis for predicting the activity of most antibiotics certainly against extracellular pathogens, and if we just look at these basic parameters, we can see where many of our problems are.
I want to show you how we can correlate pharmacokinetic parameters with outcome of infection, show examples in animal models and in humans, and apply these to otitis media.
Now, we need to be able to accurately predict efficacy. We need newer dosing regimens, we need newer antimicrobials, we need revised susceptibility breakpoints, and we need statistically valid clinical studies, and many of these points were discussed extensively this morning.
I am going to try and bring this into focus by looking at what pharmacokinetics and pharmacodynamics do for us, and basically, you are all familiar with oral ingestion of a drug. We talk about oral drugs, and the drug is absorbed through gastrointestinal tract, distributed through the bloodstream, and this is where we can conveniently measure drug concentrations and kinetics, but we must not lose sight of the fact that what we are looking at is the actual effect of the drug in the extracellular compartment of tissues.
However, what is driving the concentration of drugs there is the concentration in serum, so that even though the serum concentration doesn't correlate with what is going on in tissues, it does drive what is going on in tissues certainly in instances where you have acute inflammation. This is why blood concentrations are so important, not only in antibiotics, but in many areas of therapeutics.
Now, looking at the serum pharmacokinetic profile of a drug, we can measure this very conveniently, we can time it, whereas, measuring this at the site of infection is very difficult and very difficult particularly to do over time.
As you can see here, we can look at various parameters, the concentration of the drug present for various percentages of the dosing interval, we can look at the peak serum concentration, we can look at the area under the curve.
For time-dependent agents, time above MIC correlates with outcome. For concentration-dependent agents, either area under the curve to MIC ratio or peak to MIC ratio.
For beta-lactams, this needs to be 25 to 35 percent of the dosing interval for penicillins and 35 to 40 percent for cephalosporins. The presence of neutrophils decreases this by a further 5 to 10 percent, and free drug levels of these drugs therefore need to exceed the MIC for between 35 and 50 percent of the dosing interval to produce maximal survival.
This is showing an animal model, as you have all seen this figure of Dr. Craig, and I would like to acknowledge Dr. Craig and the other key people who work in this field for teaching me about this area.
You can see here it shows the value for cephalosporins, and I have tried very extensively to apply these principles to respiratory tract infections and also see if I can find examples of where these principles don't work, and I can find very few.
For concentration-dependent agents, it is the area under the curve to MIC ratio or the peak to MIC ratio. From the data that I have seen, either of these parameters works equally well.
This again shows the animal data at 25 to 30 ratio for immunocompetent animals.
At dosing comparable to dosing in humans, looking at a rat pneumonia model with both pneumococcus and Hemophilus published last year, azithromycin and clarithromycin were able to reduce the inoculum for macrolide-susceptible pneumococci, but not for macrolide-resistant pneumococci with either of the common resistance mechanisms, the efflux or the ribosomal methylase, and it could not do this against Haemophilus influenzae either.
This is another study showing the same thing, and I am quoting directly from the paper. "This is a chinchilla otitis media model. After administration of azithromycin at 30 mg/kg as single daily doses in our chinchilla model of experimental otitis media due to non-typeable Haemophilus influenzae, we were able to achieve levels in serum and AUCs approximately twice those observed in children treated with the dosing regimen given, and concentrations in the middle ear fluid comparable to those found in children, as well.
"Our observations provide evidence that current doses of azithromycin administered to children are likely to have a modest antibacterial effect on otitis media, characterized by a reduction information density of infection, but not eradication of infection. Maximizing the dosing of azithromycin in children has the potential to improve the microbiologic outcome."
However, I also want to point out that even going to 4 times this dose, which would be equivalent to about 8 times the dose we give in humans, the high dose still did not eradicate Hemophilus from the ears in 15 percent of the animals.
Looking at human data, Dr. Dagan has shown you this data in different format, and you can see here that when you get to above 40 percent of the dosing interval, you get greater than 80 percent bacteriologic eradication. Note also the cluster of Haemophilus influenzae around about the 40 percent point here. This is not 40 percent eradication, this is spontaneous resolution of disease. These are drugs with no activity against Haemophilus influenzae. Similarly, this point here of 20 percent is a drug with no activity against pneumococcus.
There is very much less data in sinusitis, but when this data is available, it shows exactly the same thing.
This is a very interesting study that was done on community-acquired pneumonia, predominantly in patients treated with intravenous levofloxacin. In 134 patients, predominantly with pneumonia, you can see here how well the PK/PD correlated with outcome.
When these parameters were optimal area under the curve to MIC ratio greater than 100 or peak to MIC greater than 12, then, there was almost 100 percent clinical and bacteriologic success. This is based on clinical outcome in these patients. There was only one patient judged to be a clinical failure. This patient was not a bacteriologic failure.
When your parameters were below those which have been shown to work in animals, in other words, area under the curve to MIC ratio of less than 25, then, there was a 43 percent clinical failure, and the successes were due to spontaneous resolution.
When the values were between these, you got an intermediate value of 12 percent clinical failure, so you can see this is one of the best, although one of the few, pharmacodynamic studies ever conducted in humans, and it shows how well these parameters correlate.
When you take these parameters and for beta-lactams and macrolides, you then determine--and Dr. Dagan discussed how to do this--the microbiological, the MIC breakpoint, this is what you come up with, values between 0.1 and 2 mcg/ml depending on the mechanism of action and the actual concentrations you get with these drugs.
However, when you look at what the regulatory agencies have come up with, this shows the same PK/PD breakpoints, you can see for the pneumococcus, these values as of the year 2000 were changed, and are very similar to those that are predicted, whereas, those of Haemophilus influenzae with the exception of cefixime are all considerably too high and are based on four clinical studies that were not adequate to show differences.
When you look at susceptibility of our pathogens, you see that these agents vary considerably in achieving pharmacodynamic breakpoints, and if you believe that these pharmacodynamic breakpoints are correct, then, would believe that this information is correct, and you can see here there are very few agents that cover the majority of all three of our major pathogens in otitis media and other respiratory infections.
You can see, in fact, if you go by this, our choice for empiric therapy in both primary disease, as well as recurrent disease or complicated patients is really pretty limited, and we have a great need for new drugs.
Hopefully, the situation with the pneumococcus, we are expecting the resistance to decrease because of the vaccine, but we started to see evidence of this, but we don't know how extensive this is going to be. We don't know whether replacement is going to be by susceptible pneumococci or by Hemophilus, and we don't know whether replacement organisms are going to develop resistance.
Just to mention one point also which disturbed me considerably when I was hearing many of the presentations this morning, I was hearing many of the speakers refer to PRSP, penicillin-resistant Streptococcus pneumonia, as an acronym for drug-resistant organisms in these respiratory tract infections.
To me, that is a very bad terminology and particularly when you are discussing non-beta-lactams to try and describe a drug being active against a totally different class where you have resistance, to me, that makes absolutely no medical and scientific sense.
If you want to use a macrolide for pneumococci, you need to use it for macrolide-susceptible pneumococci. There is cross-resistance with beta-lactams, but macrolide resistance is the reason for the macrolide working or not working in those agents, not the penicillin resistance.
Now, let's look at a couple of drugs in more detail, starting off with amoxicillin-clavulanate. As you saw from the data Dr. Dagan showed, it has activity against Hemophilus, but its activity is pretty close to the breakpoint of 2 mcg/ml, and by extending the dosing regimen to the new dosing regimen, if we have such, is 90 mg/kg of the mass component, you can bring the concentration that you are going to achieve up to 4 and possibly even 8 mcg/ml.
The way I have colored these graphs is the green area shows you the pharmacodynamically achievable breakpoint, the yellow area that can be achieved with higher doses, and the red area are strains which you would expect to be in the resistant range.
When you look at the pneumococcus, when you go back to strains that we had 20 years ago, they were all at 0.03 mcg/ml or less, but now 30 to 40 percent or even more of our strains have higher MICs, but you can see that amoxicillin still covers the majority of our pneumococci.
When you look at Moraxella catarrhalis, almost all of these would be lactamase producers, so we need the clavulanate, but again, those are all well within pharmacodynamically achievable concentrations.
You can see that the breakpoints we have with H. flu are maybe a fraction too high, but otherwise the breakpoints are correct.
Looking at cefaclor, not a very active drug against Haemophilus influenzae, and, in fact, as Dr. Dagan showed you, acts as you would expect a placebo to in otitis media. As far as bacterial eradication, not a very good drug even against penicillin-susceptible pneumococci, and not a very good drug against Moraxella catarrhalis.
When you look at the breakpoints that we have for cefaclor, the pneumococcal breakpoint is reasonably correct, the H. flu breakpoint is totally incorrect.
Looking at cefuroxime, reasonable, pretty similar parameters to amoxicillin-clavulanate against Haemophilus influenzae, fine against penicillin-susceptible pneumococci, but doesn't cover the nonsusceptible strains because of dosing limitations, and also not a very good drug against Moraxella catarrhalis.
Again, you can see the Haemophilus influenzae breakpoint that we have is too high.
Cefprozil, not a very good drug against Haemophilus influenzae, and if you remember, Dr. Dagan showed you bacteriologic outcome with Haemophilus influenzae, which was very poor, not a very good drug against penicillin nonsusceptible pneumococci, and not a very good drug against Moraxella catarrhalis.
You can see here the breakpoint for pneumococcus is reasonably correct, that for Hemophilus is way too high. Also, just to make the point that no one has official breakpoints for Moraxella catarrhalis.
Looking at cefixime, very good drug for Haemophilus influenzae, okay for penicillin-susceptible pneumococci, it doesn't cover nonsusceptible pneumococci, and is adequate for Moraxella catarrhalis. Breakpoint, there is no breakpoint for pneumococcus, the breakpoint for Hemophilus is correct.
Looking at macrolides, azithromycin, pharmacodynamic breakpoint is 0.1 mg/ml, covers virtually no Haemophilus influenzae, and as you saw from the bacteriologic outcome studies, acts accordingly.
About 30 percent of our strains are macrolide resistant, and we see two resistance mechanisms, the efflux strains which have MICs in the 4 to 16 range, and you can see even these are nowhere near the MICs you need for being able to treat this organism, and obviously, the strains with ribosomal methylase are way out of any kind of reasonable range, but even these strains here, you can see it is not surprising that you don't get any response with these strains here even though the MICs are not very high, and, in fact, they are fairly similar to those in Haemophilus influenzae. They are way above the breakpoint.
Even going to 4 times the dose of azithromycin, as I showed you in experimental animals, has great difficulty in covering Hemophilus. So, the breakpoint you have for pneumococcus is too high, but it doesn't make much difference because we don't get many pneumococci in that range. That for Hemophilus is way too high.
Clarithromycin, very similar, very poor against Haemophilus influenzae, covers only macrolide-susceptible pneumococci, however, does cover Moraxella catarrhalis. The breakpoint again for Hemophilus, much too high.
Clindamycin, a drug that is not often talked about and is difficult to administer in children because of taste, but is used in some patients, again is well known it doesn't have Hemophilus activity, but its activity again Hemophilus is no worse than macrolides, and also is only active against pneumococci, but it is active against pneumococci with the efflux resistance mechanism, so as opposed to macrolides, which cover 70 percent of pneumococci, clindamycin covers 90 percent of them. The breakpoint is correct.
An experimental drug, telithromycin, I am mentioning because that is one of the next drugs on the horizon for use. It has already been approved in Europe. Its Hemophilus activity is very similar to that of azithromycin potency-wise, but again, the pharmacodynamic breakpoint has now been fairly well established to be 0.5 mcg/ml, and this makes Hemophilus pretty much resistant to this drug. Some strains will come up as intermediate if you use one as the intermediate range.
Pneumococcus, it does have an advantage over macrolides and clindamycin even though it is in the same group, it does seem to be active against all resistance mechanisms at the moment, but there is a lot of potential for resistance to emerge. With Moraxella catarrhalis, it is also active. The breakpoint that has been approved pharmacodynamically, is 0.5, and that was the breakpoint, in fact, that was approved in Europe.
Doxycycline is not applicable to pediatrics, but in my remarks, if you remember, I said were going to be applied to all respiratory diseases. It is fairly commonly used, but not much is known about it, and it is not a very active drug against Haemophilus influenzae even though there is no specific tetracycline resistance mechanism.
It has greater potency against pneumococcus, but we have about 25 percent of strains that are resistant, and it is active against Moraxella catarrhalis.
Going on to the quinolones, as these are now starting to be used more in pediatrics and some of them are being tested, starting off with ciprofloxacin, one of the original quinolones, very active against Haemophilus influenzae, but inadequate activity against the pneumococcus, but I note that it is still approved for pneumococcal infections to this day in its product insert, and also very active against Moraxella catarrhalis.
With the quinolones, there is no breakpoint problem. The breakpoints are all correct.
Levofloxacin, MICs against Hemophilus remain extremely low, better MICs against pneumococcus in relation to the breakpoint, so that all strains or pretty much all strains are susceptible. We only have a few percent of strains that are resistant, currently less than one, and also highly active against Moraxella catarrhalis, also no problem with the breakpoints.
Looking at trimethoprim-sulfa, an old drug, but one that was mentioned several times this morning, what is not well appreciated is that approximately one-quarter of the strains of Hemophilus are resistant to trimethoprim-sulfa, and probably slightly more than that of pneumococci are also resistant, and also Moraxella catarrhalis is intrinsically resistant to trimethoprim-sulfa, as well.
So, again trimethoprim-sulfa is not nearly as useful as it was 10 or even 20 years ago.
So, my conclusions are antibacterial choice for empiric use in respiratory tract infections, most clinical studies do not show clinical differences between agents. Pharmacodynamic parameters correlate with bacteriological and clinical outcomes in animal models and, where we have the data, in humans.
These parameters can be used to select agents with maximal potential for bacterial eradication, and currently available agents very significantly in achieving these parameters.
Going back to the 1977 statement, I want to make the following statements. We need new FDA guidance on AOM. Do we admit there is a problem? Do we admit that we were right in 1977? What does it take to fix the problem, and hopefully, that is being addressed today.
Will we fix the problem? I certainly hope so. And when will this be achieved? I think that is a crucial point because some of the discussion that we are having today goes back, in fact, to 1977, and a lot of it, in fact, goes back to 1998, and not much has happened between 1998 and now.
Thank you for your attention.
DR. RELLER: Thank you, Dr. Jacobs.
I would next like to call upon Dr. Jack Paradise for some supplementary comments to this morning's presentations.
DR. PARADISE: These are just a few random, not necessarily connected thoughts that I had about what was discussed this morning.
The evidence report on acute otitis media that was issued with the sponsorship of AHRQ, I think is based in many instances on studies that I think are questionable in terms of methodology, and I think many of the studies that were included were studies in which diagnostic criteria were not satisfactory, much too loose, and allowed for the admission of children with OME or perhaps even children without otitis at all.
With respect to tympanocentesis, the point was made earlier today that tympanocentesis may, in itself, be therapeutic, and if that is the case, and I think we don't know with certainty whether or not it is, but if it is the case and it seems likely to be true, then, incorporating tympanocentesis may have one of two effects in a clinical trial.
One effect, and the likeliest one, would be to blur the distinction between efficacy of the two drugs being compared, but another possibility, a little more far-fetched, would be the possibility of enhancing apparent effectiveness through interaction, because how tympanocentesis affects an infection due to Hemophilus may differ from that of how it affects an infection due to pneumococcus. The group in Denmark under Dr. Toser's [ph] direction has recently shown, and I think it has been shown in other studies, that there are distinct microscopic differences between changes in the epithelium when the infection is due to pneumococcus as compared with Haemophilus influenzae.
The issue of double taps and ethics, I certainly would agree that no research is justifiable that doesn't stand a reasonable chance of producing new information no matter what the activities of the research consist of, but I think that it is questionable.
The issue of ethics was raised earlier, and I think it is questionable ethically to perform a painful procedure on a child who is doing well symptomatically and who is apparently improving in all respects from a clinical standpoint, and I believe irrespective of my opinion on the subject, that my experience with our own review board suggests that they will not tolerate that as a study procedure.
Colin's comments that fewer tympanocenteses would be done, I think is entirely accurate if studies were restricted to double tap studies, overall, fewer procedures would be done, but all of the procedures that were done in a tap failures-only study, which is what I would personally like to see as the usual type of study, all of the children who got tympanocenteses initially at baseline and then, if failure occurred, would stand a chance of benefitting from the procedure itself, and I think that is the issue rather than the number of procedures that are done.
On another vein, I think that it may be artificial to try to dichotomize patients into two categories, those with ordinary garden variety AOM and those with persistent or recurrent AOM.
First of all, I think persistent and recurrent may be different animals in some cases, and, secondly, I think there is such a multitude of variety of presentation of children with otitis, that a child on one occasion may have a mild episode, a sporadic episode that you think is not likely to be problematic, but then, in fact, turns out to be persistent or problematic, and one is dealing with histories based often on information that is of questionable reliability. Lots of studies have shown that parental recall is not necessarily adequate for demonstrating what actually has happened with children.
So, I would be inclined to have studies be fairly inclusive of children with bona-fide acute otitis media and to collect as much information as possible about their past histories and particularly about the degree of severity of the episode, which could be greater in a clinical rating scale similar to the one that Ron used this morning or using other parameters, as well.
One last point, and that is, it seems to me that the emphasis has been on bacteriology and on the organism, but in categorizing children as likely to have resistant organisms or not, it is also important to take into account the host.
Children vary a great deal I think in their susceptibility to the disease and in their ways of responding, and the problem may not always be a resistant organism, but rather the child who anatomically or immunologically is performing less well than his peer with the same infection.
DR. RELLER: Thank you.
The open public hearing has been closed, and we will now hear from Dr. Renata Albrecht with a Summary from the FDA and Charge to the Committee.
After, we will have the discussion, may or may not have a break, and vote.
Summary and Charge to Committee
Renata Albrecht, M.D.
DR. ALBRECHT: Thank you. I think I will address this group from the podium, so that I may advance my slides. Before I begin, let me apologize. I made these summary slides during the lunch break, and therefore, I do not have copies of them. However, I believe they will be posted on the FDA web site should anyone need to gain access to them.
My responsibility is to provide a summary and a charge to the committee. As I do that, I stand here feeling truly privileged having been able to listen to these august group of presenters that we had today, both the distinguished external consultants and truly even our FDA colleagues.
I think I am quite humbled by the expertise in this room on this topic, and I feel I have got a daunting task to try to summarize that, but I will give it a try, and cover some of the issues that we would like to have you deliberate on, and a couple of questions that we would specifically like you to vote on.
With that, and at the risk of introducing yet another term today, I have inserted the word "bacterial" into the indication of acute otitis media, and I have done that intentionally really to focus us on the fact that within the Divisions of Anti-Infective and Special Pathogen and Immunologic Drug Product Divisions, we are responsible for regulating the drugs for bacterial infections, as well as some others, and it is bacterial pathogens that are responsible for otitis media and the morbidity associated with it that have been discussed today. Parenthetically, we acknowledge that some of these bacterial etiologies do cause self-limited disease.
I have mentioned that the drugs that we are reviewing do involve treatment of bacterial pathogens and equally, importantly, as has been included in several presentations today, the product labeling that is written as a result of review of these drugs, does include the listing of bacterial pathogens for the indications.
Dr. Giebink reminded us that viruses also contribute to morbidity in otitis, however, those are generally self-limited and we certainly have not yet had any drugs to treat viral otitis.
Let me go ahead and talk to some of the categories that were covered today, as Dr. Powers indicated during the opening presentation this morning.
The first of these, the diagnosis of acute bacterial otitis media, as we have heard, the current guidance talks about clinical signs and symptoms as the basis of diagnosis. It talks about using a strict or rigorous case definition. I think we have heard what may or may not be some of the limitations of using a clinical diagnosis.
We have also heard about the use of tympanocentesis at baseline to establish the diagnosis of a bacterial etiology of otitis media, and actually, I guess the third bullet, if you will, is perhaps the diagnosis can best be established by using a combination of both clinical and tympanocentesis results to make the diagnosis.
Let me turn to another category that has been discussed, which is endpoints, and I have added timing as part of those endpoints, and, in fact, parenthetically say this is relating to the baseline characteristics that were documented, because again I think what we recognize is that we look at the endpoints and compare them to the baseline to come to the conclusion of whether the child, in fact, did get better or did not as a result of the intervention be it treatment tympanocentesis or some other management.
The endpoints that we have essentially used consistently before '77, since '77, and today, are clinical, and the one that I think we have heard repeatedly recommended is probably the one that we should be focusing on, is the end of therapy assessment. I have put in parentheses that we actually don't mean the last day of therapy, we tend to be thinking in terms of 2 to 7 days after the last dose. Again, those dates could vary depending on the drug used and the half-life and perhaps other parameters.
I have used a softer font to just remind me to mention that an on-therapy clinical assessment has not been used rigorously. I think we recognize clinically it is used to make a decision whether a patient is responding to therapy or not, but from a regulatory perspective, it has not been a major evaluation time point.
However, perhaps we might consider whether in the future we could use it to assess time of resolution of patients as supported by some of the information that Dr. Dagan presented to us this morning.
An endpoint that I think has been a topic of much discussion today, that we might consider looking at a new way of evaluating is the microbiological endpoint. What we have done in the last decade or so as far as the microbiological endpoint is in clinical-only studies, we didn't have it in studies where a tympanocentesis was performed at baseline. We would look at the clinical outcome and extrapolate that the organism was eradicated if the outcome was successful, and the organism was presumed to be persistent if the outcome was not successful.
I think we have heard that there may be limitations to that kind of interpretation. I think the newly proposed say of looking at microbiology that we are hearing or have heard actually in several Advisory Committees and again today, is the possibility of using a tympanocentesis on therapy, and this would be day 3 to 5. Some have suggested 4 to 6 days, or 48 hours into therapy, to be able to actually compare the pathology of the otitis at baseline and on therapy, again, just for sake of discussion, is perhaps one of the options to get both of these.
We have heard today about populations. The studies over the past several decades have focused primarily on patients with acute otitis media, and the drugs like in the penicillin, cephalosporin, macrolide classes have been developed for that indication.
You have heard today the proposal that we consider recurrent acute otitis media and treatment failure, also sometimes referred to I guess as persistent or nonresponsive otitis media, as a separate category.
For example, I think we have seen studies looking a fluoroquinolones for these kind of indications and also high-dose beta-lactams. As was brought up earlier today, I think one of the reasons to consider this is, is this a population likely to predict patients with PRSP or otherwise resistant organisms. A corollary of that is whether this would be a way to encourage a more limited use of agents that we would feel should be reserved for treating organisms that are resistant.
Then, we heard several presentations about clinical trial designs. The two that have been discussed are active control, normally, a non-inferiority design although I think the possibility could exist that one could even do a superiority design in an active controlled trial, and then placebo-controlled studies.
Putting these three elements together, clinical trial design, diagnosis, and endpoints, I have tried to summarize sort of the categories of studies that could be done, and I was going to say in the interest of time, let me skip them, so that I think I will get an opportunity to go over them during the questions, as I read those.
Let me just mention one thing. We have been talking clearly about the science of otitis media and treating children, but as regulators, I just did want to mention that there are certain constraints under which we operate, and the rules and regulations that are relevant in this particular context is the Code of Federal Regulations, Title 21, 314.126, which defines adequate and well-controlled studies. These are relevant because our approval of drug products should be based on adequate and well-controlled studies. The different choices allowed us are placebo-controlled, dose-ranging, no treatment control, active control, or historical control.
There is another part in this section that I thought is also important, which is that these adequate and well-controlled studies should be conducted in patients who have the disease, and in quotations is the definition of that, which is that, "the method of selection of subjects provides adequate assurance that they have the disease or condition being studied."
In this case, we would assume that what we are looking for is patients with acute bacterial otitis media in contrast or just to compare them to patients who may be managed clinically as patients with otitis media.
I just wanted to briefly then refer back to some of the remarks made by Dr. Powers earlier this morning, about the practical issues that face us. I think as we talk about tympanocentesis, this is not the first meeting that this topic has been brought up, the question is really what are the barriers to performing tympanocentesis in clinical trials in the United States.
Another issue or question is that are placebo-controlled trials practical in the United States at this point in time. We have heard from Dr. Rochester and others that sample sizes could be smaller if placebo-controlled studies are undertaken.
How acceptable are these procedures to patients and to their parents? Can we perform trials more efficiently while still obtaining useful data?
With that overview, let me go ahead and turn to the questions.
The first question before us is:
Should a comparative trial incorporating tympanocentesis be required--and that word I think is used in context of what the Code of Federal Regulations requires that we do adequate and well-controlled studies--should it be required for demonstrating the effectiveness of drugs for acute otitis media?
As you deliberate this question, I think we would like you to keep in mind some of the topics that have been discussed today including clinical-only studies, single tympanocentesis trials, double tympanocentesis trials, and placebo-controlled trials.
Consider also how predictive is a strict case definition of clinical otitis media to the pathogenesis of a bacterially documented acute otitis media. Consider also the relative value of comparative versus non-comparative studies and the use of tympanocentesis in these.
The second question we would like you to consider is whether you agree with the proposed definitions for recurrent otitis media and treatment failure in otitis media, and that this actually represents a separate population that we should study.
As you deliberate this question, consider whether the use of this definition is helpful in identifying patients who are more likely to have penicillin-resistant Streptococcus pneumoniae or perhaps other resistant pathogens, as well.
Consider the likelihood of differences in treatment response in this population versus the general population, and consider this as possibly a means to suggest that agents developed for this population might not be used in the same wide range of patients as drugs developed for acute otitis media.
I am sorry, we have provided for you the definitions that Dr. Johann-Liang reviewed with you earlier.
The final question is: Do double tympanocentesis trials have a role in demonstrating effectiveness of drugs for general otitis media, acute otitis media, and all-comers, or for the subset of patients or the population of patients that have recurrent or treatment failure in acute otitis media.
In considering these questions, consider the timing of assessments, both clinical and microbiologic. Consider the importance, the relative importance of clinical and microbiological assessments.
Consider the ability of the on-therapy tympanocentesis results to predict clinical outcome, and whether practically, there are adequate sites within the U.S. and other parts of the world to perform the double tympanocentesis studies.
If we are so fortunate as to have time, you could perhaps also give us some advice on alternative methods of clinical outcome assessment, such as I mentioned earlier, the time to resolution, the expected activity against the major pathogens, the role of other results, such as PK/PD, in vitro susceptibilities, age distribution within placebo and active controlled trials, and other factors, daycare attendance, prior antibiotic use, exclusion criteria, and seasonality.
DR. RELLER: Thank you, Dr. Albrecht.
Committee Discussion and Vote
DR. RELLER: In the ensuing discussion, I would like to encourage all of the persons at the table, both voting and non-voting consultants and guests, to express their viewpoint. There is much expertise here. Some individuals we have not heard from as yet. This is your opportunity, as well as responsibility, to speak up.
Secondly, to get a vote on these questions, I think it may work well to have a discussion of the subcomponents, then hearing that, which will be captured for the record as has been delineated earlier, some of you have seen remarks from past meeting portrayed on the slides, captured going back even decades, so don't be intimidated that you will be quoted in perpetuity.
On the other hand, everything is not captured in the vote alone, but also the discussion is captured for the Agency's consideration in carrying forth the next steps.
Then, on the specific questions 1, 2, and 3, we will actually have a show of hands to see how strong the consensus is on the individual questions posed.
Then, we will conclude with some additional discussion on the important but secondary fine points that Dr. Albrecht alluded to at the end of her discussion.
Dr. Dagan had a couple of points of clarification in terms of terminology, so we are all talking about the same thing.
DR. DAGAN: There are four points where people don't always mean the same thing, and I think we have to have it at least very clear. When you say "post-therapy," I mean some drugs are given for 3 days, some are given for 10 days, some are given for 5 days.
That is the point that I want to raise. My opinion is that we have to have one time for everybody because if you start to give 3 days, you don't want to be in fewer than 10 days, so probably 10 days would be the time when you end therapy by definition, even if you give 3 days.
That could be discussed or not, but this is the point where we have to at least know that it might be controversial.
The second point is day 4 to 6 or day 3 to 5, it depends how you actually start to count. In our studies, and this is came to 4 to 6, we counted the first pretreatment day, I mean the first day of involvement is day 1. Now, we want to test the second tympanocentesis after 72 hours at least, so that is why it comes day 4 to 6, which is after 3 to 5 days of treatment.
So, I don't think that day 3 to 5 is appropriate if day 1 is the first day. So, that is another clarification. We want to have 72 hours of treatment before we assess bacteriological outcome.
Then, people have used PRSP as a synonym to antibiotic resistance, Strep pneumonia, which is inappropriate. It is RSP. If you give macrolides, you really want to look at macrolide resistance, and if you quinolones, you want to look at quinolone resistance.
A secondary question could be penicillin resistance and how you promote those. So for the summary slide, one of the summary slides uses PRSP, but should be actually resistant Strep pneumonia, not PRSP.
The fourth point is that when you do one-arm, say, Augmentin high dose, whatever, gatifloxacin, it could be still a comparative study, it depends what is your question.
If you establish a drug that is appropriate, penicillin-susceptible or whatever, pneumococcus is susceptive to that drug, you still do a comparative study, actually, it's a double-blind sort of study, because you don't know what is going to grow there, comparing the resistant organism to the established already treatment of the susceptible, so it could be still a comparative study, and then you have to site it appropriately, but it could be something that sounds like one arm, but it could be actually a very nicely non-comparative study looking at exactly cutoff of MICs and all others.
So, not necessarily you don't have a comparative drug, it's a non-comparative study, and that is another point that I wanted to make.
DR. RELLER: Thank you.
The first question, should a comparative trial incorporating tympanocentesis be required for demonstrating the effectiveness of drugs for acute bacterial otitis media?
Let's have then the discussion on the bullets below that would enable us to vote on this question, in essence, the centrality, if that is the conclusion, or complementary, what is the positioning of tympanocentesis in the regulatory requirement for rigorous, adequate clinical trials.
Discussions in the context of the bullets and the relevant issues.
DR. GIEBINK: Dr. Reller, I think that we should be aware that if the committee accepts the FDA's suggestion that the word "bacterial" is inserted into the title, then, we can skip over this bullet, because there would have to be tympanocentesis for middle ear culture, and we would be automatically then accepting the 1977 guideline that absent a middle ear fluid culture, no claim could be made regarding the effectiveness of the anti-infective.
So, I think that point that was made by both FDA speakers slipped in, and perhaps we should decide are we measuring antibiotics and developing indications for their use in clinical otitis media or in bacterial otitis media.
This gets to the issue of do you do scientific studies of antibiotics for the treatment of a particular infectious disease, or do you try to replicate clinical practice. I will express my bias right now for the former, and not the latter, because you can't get to clinical practice unless you have done the scientific study.
So, I favor doing tympanocentesis to define the bacterial nature of the infection, so that we can then measure the outcome, and we will talk about double taps later on. I have some other thoughts about that.
DR. RELLER: I purposely slipped that word in to get exactly what you hit on, because from the discussions presented, if other phenomenon, apart from bacterial infection, are self-limited, then, no matter what you were saying about acute otitis media, how would you know what category you were in without a microbiological confirmation of either the presence or the absence of an agent.
Additional discussion. Dr. O'Fallon.
DR. O'FALLON: I think that there are some other ethical issues that haven't actually been expressed here explicitly, they are implicit, but they have been bothering me throughout this whole two or three years that we have been at this.
We have underneath this the fact that there is a large percentage of patients who are misdiagnosed as having --well, they don't have bacterial otitis media. Something like 25 percent is the data that we are seeing from these guys.
So, if there is no tap upfront, what we have is a bunch of patients who are being treated with something that isn't going to do them any good, and I think there is an ethical issue there.
Secondly, there is the ethical issue of we have been struggling with the creation or enhancement of the fast development of resistance, and if we are treating people with antibiotics that don't need them, I think my understanding is that that is going to increase the development of resistance.
So, I think that those two issues really need to be addressed when we try to argue that it is not fair to do the taps on the children.
Also, if we don't do a tap, there is inability to identify the subsets that would benefit from that particular treatment. We have seen a lot of information that has been given here. I think that is a very important scientific and again ethical issue, because what we realize is that with so many children or in so many people, but mostly children, who have these acute otitis media, that we are looking at thousands, hundreds of thousands of people that are going to be treated based on these studies, and if we don't get the answer right, that is going to have a tremendous impact on the future of treatment of an awful lot of people.
I think this is a big stakes' game that we are dealing with here, and we need to get the answers right.
If we do a single tap versus a double tap, there is talk about it is not fair to the patients, but as they pointed out, I thought that was a very interesting thing. Because of this Pollyanna effect, if you do the double tap, you use so many fewer patients that actually, the number of taps administered is fewer. You are tapping fewer kids if you do a double tap study.
So, if you are going to argue on the tap business, I think that that is an important piece of information to think about, again, as a critical issue.
I do think, again, placebo or not, the question is what are you trying to do. If you are trying to prove the effectiveness of a new drug, if you are trying to show that is has any activity, then, it really should have a placebo.
Again, the fact that 25 percent or better, even the ones that have pathogens there, treatable pathogens, the fact that some high percentage, 75 to 80 percent of them are going to resolve without any treatment means that we are not being unfair. It is not like they have leukemia or something.
The ethical issue of not treating them, of giving a placebo, is not the same as it is, say, in a leukemia study. So, I think there is an ethical issue there.
I think that I will quit there because I have other points, but I can't read them.
DR. RELLER: Dr. Bell.
DR. BELL: I want to congratulate Dr. O'Fallon. She has said probably better than I could exactly what I wanted to say. I totally agree that for the clinical studies in the future, clinical diagnosis at entry is not acceptable. I am in favor of tympanocentesis at entry. We have to make sure these antibiotics work, not just for the patient, but also to minimize the selective pressure that is exerted on favoring antimicrobial resistance.
I would be very interested in seeing placebo-controlled trials and I would also just to say we need to know which bacteria are in the ear and whether they are sensitive to the antibiotic being studied, because it could turn out in the future that those incidence rates of bacteria etiology might change, and we need to know that information for that drug.
So, I just want to totally agree.
DR. RELLER: Dr. Nelson and Dr. Pichichero.
DR. NELSON: What has impressed me here is the devil is very much in the details of all of this information. Personally, I found the most helpful presentation to me and sorting out from the perspective of someone who chairs an IRB is Dr. Pichichero's presentation of benefits, risks, and the like.
I would like to present what I see as perhaps a way of getting through the forest. It would bother me if we started tapping kids, which everyone says we diagnose poorly, in order to use that as an enrichment strategy to make sure that we have the right group to go into a study.
The thought that occurred to me is given the 80 percent response rate and the percent viral etiology, that you could argue quite convincingly that a three-arm placebo trial is appropriate for a clinical diagnosis and a clinical endpoint, and that you actually could use that as a first phase of an enrichment strategy defining failure, which could be defined in a way similar to the indications for doing a tympanocentesis that was presented by Dr. Pichichero.
You could also have an arm that goes in, which could be into that second phase, which could be a severity of illness, toxic, bulging, febrile child, perhaps other things, to where they would go in immediately to the second phase, which would be a double tap comparative trial.
I think the whole issue of the efficacy of the tap itself, I think raises an interesting question of how you would design that, does that really mean it's an add-on trial of antibiotic on top of tap, or should you have a tap alone arm.
So, in listening, I think I would hope ultimately that FDA would begin to list 50, 51, 52, 53, and 54 as part of their regulatory constraints besides just the desire to have good science.
I think there is equipoise if you are in the clinical setting, and so that fits in 50-52, which is the direct benefit. If the tap is being done by someone who has done 1,000, and teaches others to do it, it looks to me like it fits in a minor increase over minimal risk, and the second tap wouldn't necessarily have to meet a constraint of providing benefit.
The first tap would provide benefit, but I would also be worried that those taps would be done by people without sufficient expertise unless they are privileged or certified in some way to be able to perform it.
I have never done one, and I work in an intensive care unit. I have never actually seen one done until today. So, you know, it is not out there being used a lot.
So, I guess to summarize, what I began to sort of think about is a way that the Phase I, the clinical diagnosis could be enriched by not using the tap, but by using basically a three-arm, randomized phase to get into it, and then, at that point, take the non-responders, the recently treated, and the severe ones immediately that could bypass that first phase and put them into a double tap trial at that point.
A company that wants to do that through both phases could end up potentially with labeling either for the general indication or for the specific limited indication, because their motivation obviously is to want to have the general indication, so it might be able to kill a bunch of birds with the same stone.
DR. RELLER: Dr. Pichichero, Dr. Chesney, and Dr. Sumaya.
DR. PICHICHERO: I wanted to comment on the notion of placebo-controlled trials before there gets too much of an enthusiastic endorsement of that idea by the committee or the FDA.
Several of you are quoting a rate of 75 or 80 percent placebo response rate. As Dr. Paradise briefly alluded, if you look at the actual studies of placebo-controlled trials, there are not many. The entry definition of otitis media is so vague or nonexistent that I would question whether many of those children had otitis media, and if they did, whether they had otitis media fusion rather than acute otitis media.
So, I think at this time we really don't know what the placebo response rate of children with otitis media might be, but my suspicion is if you use a bulging tympanic membrane as the single most important criteria, it is not going to be 75 or 80 percent spontaneous cure.
Secondly, it was mentioned in Dr. Rochester's presentation that if we were to do placebo trials, we would need careful follow-up. Does that mean that you are going to follow the patient and see them every day, and even if you do, how many cases of meningitis or mastoiditis that you pick up early would be a tolerable level in the United States?
In my own practice, one would be intolerable. Therefore, as an investigator in the field, I would be very reluctant to participate in a placebo-controlled trial to ask my patients to accept a placebo in what I think is bona-fide otitis and to accept a risk in my community that I would cause one child to get meningitis or mastoiditis.
My last point about the tympanocentesis is that in my experience, this procedure causes no more pain than a venipuncture, which we do routinely, for example, in vaccine trials multiple times to children, and what I see in terms of the amount of pain that it induces, the amount of change in heart rate on the pulse oximeter, the amount of times it takes a child to recover, in experienced hands, it is the same as a venipuncture.
Those are my points for the moment.
DR. RELLER: We will stick with our rotation, so that everyone gets a chance. Then, we will come back to Dr. Marchant. Dr. Chesney.
DR. CHESNEY: Just three brief comments. Dr. Paradise's comment that we have all wondered about whether a tap is therapeutic, and if we limit our studies to those involving tympanocentesis, that is not going to apply to the real world, because most people are not going to do taps before they treat otitis media, so can we really extrapolate studies that involve tympanocentesis to the real world.
The second point, I think we do need tympanocentesis studies for sure, and again we have all talked about this now for a number of years, particularly now that we have the pneumococcal vaccine, because we really don't know that PRSP is going to persist.
We don't know that other strains are going to pick up the resistance organisms, so I think we really don't know what the future of otitis media with respect of PRSP is, I don't think, or RSP.
The third point that I wonder about is if we really do placebo controls, and a year down the road that child turns out to have hearing deficit or developmental delay, where are we going to be at that point legally, and do we know enough about the relationship between acute otitis media, no treatment, and hearing and developmental delays following that.
DR. RELLER: Dr. Sumaya, then, Dr. Marchant.
DR. SUMAYA: I am reaffirming what Dr. Pichichero has just said, because I was very worried about the discussion on the 70, 75 percent, up to 80, of spontaneous resolution, because I think it is very unclear how that relates to specific pathogens in the ear, whether they are viral or if it's a pneumococcus or whatever.
The other part of it was the complication rate that may be related to that, and, thirdly, is clinical manifestations that the child may have on day 1, 2, 3, 4, that may be different depending on the pathogen.
Secondly, on the tympanocentesis, again, I would refer to what he just said. When an RFP eventually comes out, I would assume that there is going to be some very good requesting of experienced people who do tympanocentesis, because I think in experienced hands, it is a simple procedure; in non-experienced hands, I wouldn't go for that.
I am in favor of the tympanocentesis on entry.
DR. RELLER: Dr. Marchant.
DR. MARCHANT: The first thing I would like to do is comment on the issue of placebo trials. I think there are some placebo trials in the literature that are instructive. None of them are completely ideal. The one that was done in Pittsburgh is an interesting case in point. There is also some European trials that were reasonably well done, although typically on selected populations.
In terms of what do the best placebo trials say or what do the meta-analyses of placebo trials say about the response of otitis media antibiotics, it is that children get better a little bit faster perhaps by a day, a day and a half, if you use antibiotics than if you don't. There is a benefit.
So, if we are going to do a placebo-controlled trial, then, we are going to withhold a therapy that might have some benefit to that child. I think that, yes, as Dr. O'Fallon pointed out, they are going to get better, most of them, fairly quickly.
That doesn't preclude doing a placebo-controlled trial, but I would want to see a placebo-controlled trial that was going to really teach us something new, and not just be a placebo-controlled trial for the purposes of new drug B that we are now testing for licensure, but rather that if we are going to do the placebo group, that we do it to identify a group of patients that don't need treatment because they are milder or something of that sort.
The other thing, I know I sound a little bit like a broken record, but it keeps coming up, and that is, if you do a clinical-only trial at entry with a placebo arm, et cetera, you just drive that sample size issue back up again, and it always needs to come back into the conversation.
Those are my comments.
DR. RELLER: Dr. Dagan.
DR. DAGAN: Again, going for a second to the placebo issue, it is very nice to talk about placebo, but if you really read those articles, and I promise you I read them much more than once, all the placebo except the one that was done by Howie, which I use all the time as my reference point, they have limitations because they don't want to put into that study, patients that are going to be actually in danger if they get placebo.
So, the one that you cite, of 1 day out of 14, or whatever, took away those with high fevers, took away those with real bad bulging, took away those who were looking a little bit more sick, et cetera, so eventually, you will come down to those who don't really need antibiotics, then, it is only 1 day out of 14, so the real placebo study that enrolls all patients with otitis including those who need antibiotics the most is nonexistent for the moment except Howie's study.
So, I think this is one very important point. I don't think the Ethics Committee will approve us, or ourselves will approve ourselves, to do a study with placebo without a priori ruling out those who need antibiotics the most. So, this is a point, and I think therefore, it is not feasible to do it.
The other point is talking about Dr. Nelson's remark. He works at the ICU, and he rarely sees these because this is not an ICU procedure, this is a very benign procedure, you don't do it in ICU.
Actually, my ICU people, it is very difficult to convince them to take blood cultures, they are so busy doing the big stuff. This is the small stuff. A second blood culture in study to go to ICU to get it, which is totally benign, they always forget to do it.
My point is that Dr. Paradise mentioned the word "dangerous." Now, we were talking about this last time, and people have talked about this time. I don't think it is dangerous, I didn't see any real complication out of the dozens of thousands of tympanocentesis we do in our center, and it is one center for the whole region. If we had this complication, we would have seen them.
I think that the point again and again and again, otitis media is a disease with complications, inappropriate treatment is procedure with complications. Giving drugs that act similar to placebo is much worse than placebo because they promote resistance, and therefore, knowing what you do is the most ethical thing, and therefore I don't see any danger of doing tympanocentesis.
I think this is a very, very important point, and I would like everybody who says it is dangerous to justify why he or she says it is dangerous.
The last point that I want to make is that Ethics Committee, like all of us, like the FDA, they are subject to continuous education, and what is ethical and what is not ethical, 10 years ago it would be unheard of.
Dr. McCracken is the editor of his journal, sent me back a case report of quinolones in children with a letter, which I keep, saying that he would never publish a study on such drug that would never be used in the United States.
So, what is ethical now and what is not ethical changes. If FDA thinks this is the most appropriate study to do, together with the Advisory Committees, and together with experts, it is going to become ethical slowly but surely as people would never believe that we can move the FDA to do again bacteriology, and it is now moving.
Don't take a snapshot of what is ethical now, convince the Ethics Committees how dangerous it is to treat without knowing what you do, and therefore, it will become slowly but gradually ethical.
DR. RELLER: Dr. McCracken.
DR. McCRACKEN: Well, one thing I have learned in medicine, never say never, so I don't think I said never in that letter I wrote.
The comments I want to make, obviously, my bias is well known, it has already been on the screen a couple of times, but it harkens back to the very simple principle, what is an antibiotic for. Why are we giving an antibiotic for otitis media? It is not a decongestant, it's an antibacterial, and if you are going to be evaluating a drug for otitis media, for meningitis, which happens to be my real love, you have got to know whether it works, and you don't know unless you can show bacteriologic eradication. That is the only thing that drug does.
Now, it has secondary benefits obviously, but its primary benefit is only for the eradication of that organism either with the help of the host or not, but it is the eradication.
So, not only am I in favor of an initial tympanocentesis for the reasons stated, and I think Scott did a good job in doing that, but I am still in favor of a second tympanocentesis at least in a substantial subpopulation to demonstrate exactly what this drug is doing, does it eradicate it in a timely fashion.
This is true in many bacterial diseases - sepsis, meningitis, otitis, urinary tract, it is all the same. You just have to pick the time when you want to demonstrate that.
Two other points. It has been stated that tympanocentesis probably may improve outcome, and that could be true to a certain extent, but I just remind you of Ron's study that he showed already, with azithromycin versus Augmentin, the regular formulation of Augmentin where both groups got double tympanocentesis, and yet there is still a difference, both in the clinical scores and in the outcome.
So, if it does help improve symptoms to whatever the modest degree might be, it doesn't obscure the clinical outcome, which is very important.
The final point about placebo. Mike Pichichero is concerned about meningitis, and I will just say that in the British Medical Journal placebo study, published in 2000, in the placebo group was a case of meningitis, so it is not a far-flung possibility.
DR. RELLER: Dr Wald has a comment, but just to follow up on that point. It seems to me this issue of placebo and the requirement for tympanocentesis are related. Accepting, I think all would agree that the rare, but potentially devastating complications are related to bacterial infection, the abscess in a closed space, I mean it was described earlier.
So, the need for placebo, it seems to me is related to showing a difference, tap water, if you have got a study population that is so diluted by people who don't have the real thing, that you might come up with not being able to show a difference, but if you have a tympanocentesis, and you know where you are to start with, as Dr. McCracken has just mentioned, one can show differences in efficacy of agents that would otherwise be obscured for the reasons that Dr. Marchant has emphasized earlier.
DR. WALD: I would comment on that, as well, the need for placebo-controlled trials, and that is, that there is tremendous enthusiasm now among physicians, as well as the lay public, to not treat acute otitis media. Now, there don't seem to be too many of those folks in this room, but there is a tremendous enthusiasm for a no-treatment policy, and I think it is essential that we show, in fact, that this is an acute bacterial infection that benefits substantially from the antimicrobial therapy, and the only way that we can do that is with a placebo-controlled trial that is very tightly monitored.
Although I share concerns about meningitis, I would say two things along those lines. One, there has never been very good evidence that otitis leads to meningitis. They occur in some patients together, but I think that one leads to the other is not clear, and that we were never in a better position to do this study than we are now because of the availability of pneumococcal conjugate vaccine, which is really going to protect the meninges of the majority of children who we will be studying who have been immunized.
In fact, you could make it a requirement that anybody who entered a placebo-controlled trial had received the pneumococcal conjugate vaccine, and I think that would provide a lot of protection.
I think there is a general consensus in the room that tympanocentesis is appropriate for a lot of patients who are going to be studied, and it is essential, I think, in terms of establishing the microbiology, which is an ever-changing phenomenon.
I would like us to require that when investigators submit cases or when industry submits cases, that there be a certain minimum level that investigators achieve in order to enter patients into those studies, you know, whether that be a 75 percent positive culture or an 80 percent positive culture, I think we need to insist on some minimal level, and that those very same investigators who achieve competency at that level, be the people who can do clinical-only studies where we know that they have established their expertise in the diagnosis of acute otitis media.
Just one word about double tap studies. It is true that you will tap fewer children and do fewer taps if you do double tap studies, but you will not be benefiting all the children when you do that.
When a child is symptom-free on the 4th or 5th day of therapy, I think it is very hard to ask permission of that patient to tap that child, whereas, at least at the entry points, there is some thought that every child who undergoes tympanocentesis will benefit from that procedure.
DR. RELLER: I don't know who was first here, but Dr. Marchant and Dr. Soreth.
DR. MARCHANT: I am having a little trouble understanding here. If we withhold antibiotics from children in a placebo-controlled trial, and they have symptomatic otitis media, even the mild variety, such as the one in the Kaleida trial, I think we are using a study design which is going to result in more discomfort and more pain for those patients.
So, my earlier comment was motivated we need to learn something good from doing such a trial because if we are going to have a trial that has more discomfort or pain, we should at least get something scientific about it.
On the flip side of that, when you talk about tympanocentesis, the second tympanocentesis, in my mind, is justified even in an asymptomatic patient because we are getting the data that we need to know whether the drug is going to work for all those children out there, and for that reason, it is justified.
Now, there are other design approach, tap and tap of failures, but it has other implications in order to get that information, but I think there needs to be some consistency about how much discomfort we are going to design into trials and for what benefit for patients, and be clear about what those are.
DR. RELLER: There will be additional discussion about double tap. We will be voting on whether tympanocentesis is essential for any trial that would claim to show efficacy for the treatment of acute otitis media.
Dr. Soreth, Dr. Leggett, Dr. Ramirez, and Dr. Nelson.
DR. SORETH: I think a comment that pertains either to active control trials or to the prospect of a placebo-controlled trial is that in the development of a novel compound for any infection and this one, acute otitis media, we can't forget that part of the equation involves safety.
So, whether we have an active-controlled trial with some standard agent that we feel we know a lot about or a placebo-controlled trial, we can't assume that the new drug is completely safe or safe enough, so part of what we might get out of a placebo-controlled trial is information about safety, and similar information can come in an active-controlled trial, but we can't assume that we have all the data to say absolutely the way to go in every case of acute otitis media is antibiotics because we know that a day's difference is the end-all and be-all.
It may be, but I don't know that we have enough data to say that definitively, it is, so we can dismiss completely placebo-controlled trials as an issue.
I think I had a second point, but my thought train may have been derailed.
DR. RELLER: We will hear from Dr. Leggett and Dr. Ramirez. It is very important for the continuity, the togetherness of the session that we all here to the end, so after comments from these two, and there will be time to have throwbacks to some of these issues because they are all interconnected, we will vote after these two comments on Question 1, move on to Question 2.
DR. LEGGETT: I had two questions to bring up along the lines of the tympanocentesis and the single or double, in the sense as follows. If we are going to try to include folks who have recurrent otitis or who have recently received therapy, and therefore, are more likely to have the more severe disease, and we tap them, and because they have just been on antibiotics, the tympanocentesis is negative, what do we do about that?
The second question is presumably one of the purposes of the guidance is to improve upon some places where the FDA recognized that there were some problems as in the recent azithromycin case. Without a double tympanocentesis study, I would like to hear some comments about how we avoid doing the exact same thing again.
Those are two questions.
DR. RELLER: Dr. Ramirez.
DR. RAMIREZ: I just want to make a general comment regarding the first tap. It was already mentioned, it seems to me that the clinical symptom of acute otitis media involved a large number of patients that may not have the disease or less number of patients may have a viral infection, and when we design clinical trials for infectious diseases, we never say, okay, I want to see what happened with these antibiotics against meningitis, because we don't take the meningitis syndrome and try an antibiotic, because we know there are plenty of patients who have a viral meningitis.
We always design antibiotics for bacteria meningitis, we don't discuss antibiotic for chronic extravasation of chronic bronchitis, we discuss antibiotics for acute bacterial extravasation of chronic bronchitis.
I think that this is supposed to be a discussion of antibiotics for acute bacterial otitis media. Now, how do we know if the meningitis is bacterial? We put a needle, and we figure out is this a virus or is it a bacteria.
I think that we have the possibility to make a microbiological diagnosis, it is not just to define the etiology, it is to define the disease because we don't put needles in the lung to define if the patient has a bacterial pneumonia, because then the complications are a bit high, but otherwise, if we all of a sudden find a way to put a needle in the lung without complications, we would put needles to figure out what is there.
I think that was already explained clearly by the experts that this is a very simple procedure, and if have the possibility to eliminate all the known bacterial cultures of otitis media, to me it is a no-brainer that if I decide on a study to study acute bacterial otitis media, I need to make the right diagnosis at least in a significant number of patients.
Now, where we are mixing those, we discussed yesterday in this committee, one thing is a clinical trial for the right indication, and the other thing is clinical practice. Now, we know that what we get approved here for these 20 percent of acute bacterial otitis media is going to be used in the other 80 percent that have viral disease, but this is a different discussion, because this is because the general practitioner, it seems to me, they use the clinical syndrome for diagnosis, they are not going to be doing the tap.
The antibiotic is going to be overused in some patients with viral otitis media, but I don't think that we are going to be able to prevent this unless we have a very, very simple way to define these are bacterial or viral with a needle, and this is why we have an overuse of antibiotics, but still, it is going to be justified overuse from the clinical point of view.
To me, to define that antibiotic that is well expressed, it needs to kill a bacteria. This is the only thing that we ask for the antibiotics. First of all, we need to figure out is there bacteria there.
DR. RELLER: Thank you.
From the voting consultants, Drs. Chesney, Giebink, and Nelson, and the current members of the committee, a vote. We will start to my right.
Basically, should the FDA require a study that incorporates tympanocentesis, not necessarily as the only evidence, but as one criterion for the approval, looking forward, of a drug that would be claimed to demonstrate efficacy in the treatment of acute otitis media?
Dr. Nelson, yes, no?
DR. NELSON: I have not heard enough information for me to vote, and I had a specific question which I wanted to ask to get that information.
DR. RELLER: Excuse me?
DR. NELSON: The question I was going to ask if you said we would vote before, I wanted to ask to get the information so then we would vote, so I am happy to abstain and wait, or whatever, but I am not going to vote yes or no based on what I have heard.
DR. RELLER: Okay. So, that's an abstention.
DR. GLODE: Yes, I think a comparative trial incorporating tympanocentesis should be required.
DR. BELL: Yes, I think that tympanocentesis should be required initially. I do not believe it should be required for follow-up. I am not sure it is ethical. I think too many parents will not consent, and it will make subjects too hard to enroll.
DR. RELLER: Thank you, Dr. Bell.
I vote yes, I think we need to know what is there to be able to judge efficacy.
DR. PATTERSON: I think tympanocentesis studies should be the standard. I think many of those may be single tap studies, which should be accepted because they will be done over a wider geographic range, and I think we need the information about the microbiology and susceptibility over a broad geographic range.
The double tap studies will be useful in subsets in centers where those are the standard of care. Placebo trials, I have some concerns about. Even with the Prevna [ph], which is I think of interest, would we be selecting then for less severe disease, making pneumococcal disease less common in this group, and therefore, less sick or severe population.
The clinical-only studies in the setting of safety or placebo trials, which I have a little discomfort with, and I am going to throw in age distribution. I think at least 50 percent should be 6 to 24 months.
DR. RELLER: Thank you.
DR. WALD: Yes.
DR. SUMAYA: Yes.
DR. GIEBINK: Yes.
DR. O'FALLON: Yes, and I enthusiastically endorse what Dr. Patterson said.
DR. RELLER: Dr. Chesney.
DR. CHESNEY: Yes, also without qualification.
DR. RELLER: Dr. Ramirez.
DR. RAMIREZ: Yes.
DR. EBERT: Yes, although I think that we need to be clear on entrance criteria for patients to enter a study involving a tap.
DR. RELLER: Dr. Leggett.
DR. LEGGETT: Yes, a comparative tympanocentesis trial should be the pivotal trial. I wanted to address one of the other points we were supposed to, and I haven't heard yet, about the non-comparative versus comparative.
If we use non-comparative data, it should be used for gathering more safety data or for boosting the N for efficacy purposes, but I think that is where we can incorporate PK/PD things with MICs to give us more information about the breakpoint while we are doing the trial.
DR. RELLER: Dr. Cross.
DR. CROSS: My answer is yes, but since I didn't have the opportunity to make a number of comments earlier, I will take this opportunity to say that if we do encourage comparative trials of drug A and drug B, it seems that we almost have to invite either a placebo trial or ask the FDA to come up with a response if drug B is 70 percent and drug A is 90 percent, what happens in terms of judging the 70 percent of there is no placebo, will the FDA accept that for approval, that is, is that 70 percent drug sufficiently effective for approval even though it is inferior to another approved antibiotic.
So, I think that the question is kind of in a way tied into the issue of placebo-controlled, and in terms of addressing the point of a placebo control, that Dr. Dagan made, I mean I think it is really incumbent upon us if we include placebo-controlled, that we would have to really tighten up the clinical definitions in a way that would really incorporate who are the people who were excluded out of all those other "placebo-controlled" trials.
Then, lastly, in terms of the issue of double tap, I would like to return to an issue raised by Dr. Giebink, where he showed the cells, and there were certain people who were bacteriologically cured, but were clinical failures.
I think by doing a double tap in those patients, it really affords us the opportunity to say are there any inflammatory media that may have resulted from the bacteriologic cure which may account for the clinical failure, which may at least lead us into other therapeutic areas.
DR. RELLER: Thank you.
Question 2. Does the committee agree with the definitions below of recurrent acute otitis media and AOM treatment failure, used to identify a separate population of patients for study?
There are some additional things that we are to address in the discussion, but the two definitions are listed below, and I think it would be helpful to take these individually.
So, first of all, does the committee--and maybe a brief discussion on this--does the committee feel comfortable with, feel it is appropriate to define recurrent acute otitis media with the numbers given, that is, 3 or more episodes of AOM over the last 6 months, and 4 or more episodes of AOM over the past year?
Those have been used earlier in slides from the experts in the field. Are these pretty well accepted? Do they need to be modified?
Yes, Dr. Hoberman.
DR. HOBERMAN: One additional comment. There is two different groups of children that will not be included if those two definitions are used. One is children that have early infection during the first six months of life, but we can argue whether it should be nine months, may not have had enough time because they did not live through the previous winter to have declared as otitis-prone, so an early in life otitis media would probably be similar to more than 3, and children that have had an otitis media within the previous month might be at a similar risk as somebody that had 3 episodes over the past 6 months or 1 year.
So, those two additional groups of children may enrich the population at risk.
DR. RELLER: Dr. Paradise, Dr. Giebink.
DR. PARADISE: I would just want to add the qualifier of documented episodes, because it has been our experience, and that of many other people, that situations don't always pan out as they had been forecast.
DR. GIEBINK: I would feel more comfortable with a definition that embraced the high risk and the low risk child, and to not try to wordsmith the definition of high risk at this kind of a setting.
The beauty of the schema that was proposed by Rosemary is that this is the exact scheme that came from a CDC consensus discussion about five years ago, published by Scott Dowell [ph] in George's Journal.
So, it is a scheme that is being used now in clinical practice, and in terms of meeting the pragmatic threshold for industry to develop trials, it has a relatively large hoop to jump through.
So, I think that high risk and low risk, good idea. I share Dr. Hoberman's worries about age and number of episodes, and I think that just needs a lot more discussion to define what is a high risk episode.
DR. RELLER: Dr. Marchant.
DR. MARCHANT: In terms of the reasons why in a single episode, a child will not do well, one is resistant bacteria, which is mostly related to prior antibiotic use and daycare exposure, thereby prior antibiotic use by their mates in daycare.
The other one is young age. Dr. Paradise earlier talked about the host and the clear factor we have that predicts bacteriologic failure and clinical failure in Pittsburgh trials and double tap trials, and so on, is young age, and so the age factor, if you are going to enrich a population in terms of their risk for not doing well on antibiotics, age is an important issue, and you can cut it at 1, you can cut it at 2, or 18 months, or whatever, but that is a factor.
The recurrent otitis media definition per se, I believe that it is enriching the population mostly because those kids have already been on a lot of antibiotics, and maybe daycare, et cetera. On its own, I am not aware of it being a predictor for poor response inside a single episode of acute otitis media, so I am not sure it, on its own merits, is critical here, and I would be interested in the other folks that know the otitis literature, what their comment would be.
DR. RELLER: Dr. Wald and then Dr. Ramirez. I have asked Dr. Johann-Liang to bring up the definition that Dr. Giebink alluded to, because to the extent that there are vetted definitions that might not as a necessarily definitive statement, but close to the mark, it might save us a lot of time if we have a target for trying to reach some degree of consensus.
DR. WALD: I agree with the definitions, but I think that sort of categorizing children according to risk is more helpful, however, most children, the peak age incidence for acute otitis media is under 2, and we know that age is a risk factor.
So, most of the children that we will be entering into these studies, by definition, have an important risk factor. Although some of them may not attend daycare, we know that that is an increasing trend among U.S. children, and even those who don't attend daycare, go to church on Sunday morning, in the play group, or they go to mother's exercise class, and they are in a play group, or they go to McDonald's once a week, and they are in that little playground.
So, I think that a daycare equivalent is almost universal, as well. I think most children are in the high risk category, and maybe what we want to create is a low risk category for children who are over 3 or 4, and who never had an episode of otitis media before, but the majority of children are really in a high risk category.
DR. RELLER: Dr. Dagan.
DR. DAGAN: Some risk studies looked at daycare center versus age versus previous antibiotic treatment, and they found each one to be independent risk factor, so what you say is correct, but probably if you go every day for 5 hours together with kids, it is different than if you see them on Sunday morning for whatever, 3, 4 hours at the play group.
So, I think that so far, the evidence tells us that each one is independent, and if you have all the 3, you multiply each risk by the other, and you get enormous risk. So I still think that this should be taken into account as for risk factors.
The other point is that when we take our 1,000 cases with double tympanocentesis, and we look at those who have, first, otitis media or at least did not have otitis media in the last three months, or those who have clinical nonresponsive/recurrent otitis media, and you look at the MIC of the bug, this is the number one thing that counts, and not the previous episode in terms of bacteriological eradication.
If you don't see lower bacteriological eradication, what you see is, in general, you have lower bacteriological eradication because you select for more resistance, but if you break them by MICs, you actually find exactly the same. Not only this, even if you have mixed infections, each of the bugs behave according to what they were supposed to behave according to the MIC.
So, I think that if you look at bacteriological eradication, it doesn't really matter. There are two slides that you want us to consider in this Question 2. One is in relation to whether you do to groups or one group, and I think that bacteriological eradication, what counts is the MIC and the dose of the drug.
For clinical responses, for the second group that has recurrent, relapsing, et cetera, they returned immediately, during treatment, to get to the next complication, then, the clinical outcome is going to be worse in one group than the other.
So, if eventually, this group here decided they want double tympanocentesis study, and look at bacteriological outcome, it doesn't really matter which kids to take, and we have the evidence, and I can send these tabulated.
If you look for clinical outcome, it makes a lot of difference if you accept this--I am not sure you need two groups, but you need to analyze them separately.
DR. RELLER: Dr. Glode.
DR. GLODE: I don't see that there is two distinct populations, and I think it is very confusing to have them as two indications, so I am doing the study for group 1 indication, but not group 2, because it's a spectrum.
Recurrent otitis media, as an enrichment issue, is just a selection for people who have gotten antibiotic courses. So, if they got it for sinusitis, then, they are in group 1. Because they didn't have recurrent otitis, they are still going to have a higher risk of resistant pneumococci.
So, I think you can look at that by having the bacteriology and analyze that way, and I just think this is, I don't know, more confusing and suggests that there is two distinct populations when I don't think there really are.
DR. DAGAN: I think this was invented because some drugs are intended not to be given to all children. Nobody mentioned that, but this is the main justification for me to put it in two groups.
If I don't want to give quinolones to every child, only to those that don't respond, then, I take this group and study this group as an indication for the specific study in order to limit the drug, not in order to get the better information.
DR. GLODE: Then, you do that as your Phase II of your bacteriologic failure, and you have no other choice, and so you must go now to this less safe antibiotic, but I think to use it as an excuse for testing those kinds of drugs is also a wrong reason to make these two groups.
DR. RELLER: I think we are making some progress here. The last two comments, and then we are going to have a vote, and maybe, given what is heard, I mean we will see whether people think this is crucial to have this, not that it couldn't be incorporated, but crucial to have it, or is it the real issue is part (b), namely, treatment failure and what might be appropriate approaches there.
Dr. Bell, Dr. Ramirez, and then we must take a stand on 2(a).
DR. BELL: Drs. Giebink and Dagan are correct in that this was developed some years ago by a group, some of which are in the room, that CDC convened to try and identify episodes where second-line treatment was not needed, at least empirically.
I guess the question is does this refer to clinical trial designs only, or is it what I think is the intention is the practical use by a practicing pediatrician, who is not going to do ear taps, and this is a nice, convenient category, and the clinical trials, the pharmaceutical companies might find it attractive to have this admittedly rough distinction. I am inclined to support it.
The final comment I want to make is that I guess I was a little surprised to see fluoroquinolones appear on the FDA slides and be kind of mentioned glibly as options. I think that requires a lengthy discussion in its own right, and I just would hate to see a message go out that that is a done deal.
DR. RELLER: Dr. Ramirez, do you have a comment?
DR. RAMIREZ: Yes. If I remember right, we discussed this a lot. In this committee, we discussed the amoxicillin-clavulanate. The idea was trying to enrich the population. If you have an antibiotic that you want to get approval for penicillin resistant Streptococcal pneumonia, you don't want to get 1,000 children and get only 10 penicillin resistant. The idea was just study in a specific group that we call an enriched population, that you have a very high chance that you want to get penicillin resistant Streptococcal pneumonia.
I would say that for a drug company that is looking for this indication, for PRSP, then, this may be a good possibility for them just to select these populations.
Now, this is different to say that because in this population, you have the greater chance for getting pneumococcal resistant, but as already mentioned, because all children or most children with this disease are less than 2 years of age, I would not use the same criteria to say to the clinician, now, you have a child with this. Without these risk factors, the penicillin resistance is not going to be there, because by definition, these are disease where penicillin resistance is going to be prevalent, and if one of the risk factors is less than 2 years of age, it is going to be very difficult to make an algorithm for clinical practice to say you have these risk factors, use second line, you don't have these risk factors, use the first line, because you have to put less than 2 years of age as a risk factor, and he is going to read the first line only for patients that are 3, 4, 5 years of age.
I think the intention here is to separate populations for a study, to identify patients with higher risk for penicillin resistance, then, I would say yes for these, but not for empiric use of antibiotics, you know, first line, second line, as it seems to me that was the intention of the presentation.
DR. RELLER: One has heard some of the major points of discussion, so the vote is do you agree that it is important to differentiate into high risk, low risk, or are these particular categories not necessarily limited to those, in other words, to differentiate the population, and perhaps that is less necessary, although it could be part of the analysis if one has a tympanocentesis and knows whether you have got the organism in the first place, if these are tools to get at the surrogate for knowing what you have, because of the likelihood of having a bacteria.
We are going to start over here this time. Alan, do you think it is crucial to incorporate these or it is part of trial design, but not essential to separate them into two categories, whatever the definition?
DR. CROSS: I am not a pediatrician, but from the discussion I have heard, the frequency is not sufficient to define the population at risk, and high risk/low risk has its problems for what we have heard.
It seems to me the most logical is that if we truly want to focus on the resistant population, that after our tympanocentesis, of the failures, those are the folks who are the most highly enriched for failure by definition, and would be a good population to study the antibiotics, which we don't want used for initial therapy.
So, the answer is that I don't think we have enough information to simply use the definitions as proposed here, and I think that the best information will come from the double tap studies.
DR. RELLER: Thank you.
DR. LEGGETT: I am not sure that the use of these two definitions per se will help us delineate well enough to make it worthwhile to industry or anyone else, especially if our guidelines are now going to be tympanocentesis at the baseline and the inclusion of lots of kids under 2. So, we will have so much overlap between the kids under 2 with everything we have heard about all the other things that is going to happen, that these are no longer going to be very useful.
DR. RELLER: Dr. Ebert.
DR. EBERT: I think overall I believe that the age group under 2 should be a broad focus regardless of other risk factors, that even in simple, uncomplicated cases of otitis media, increasing the percentage of children that are under 2 would be useful.
Having said that, I think that using other factors, such as recurrent infections, may be of benefit because they may enrich the likelihood of having more resistant organisms, and I think they also parallel in many ways the clinical stepwise approach that many physicians take to treating recurrent cases, that you tend to up the ante, if you will, as far as the types of antibiotics that you are using with recurrent cases.
DR. RELLER: Thanks.
DR. RAMIREZ: I agree with the definitions. I think that these plus other risk factors can be used to identify patients that are more likely to have penicillin resistant Streptococcal pneumonia for clinical trials of enriched populations. We are looking for this indication.
DR. RELLER: Dr. Chesney.
DR. CHESNEY: Could you restate the question again?
DR. RELLER: Basically, the question is do we agree with these definitions, and we are taking them in two parts. The way I interpret it is should clinical trials, the patients necessarily be categorized as being recurrent or non-recurrent, or is the population that you really want to study, the under 2's, the ones that are at higher risk because of daycare, the children under 6 months of age, in effect, that this becomes a component, but not a critical one that really you are talking about studying the patients who really have the disease, but this as a tool to get there alone is either not enough or is too restrictive, I mean however you want to look at it.
Basically, I know how I am going to vote. I am going to vote that it is not crucial. I was convinced by Dr. Glode's comments and Dr. Wald's comments earlier.
DR. RAMIREZ: More than the two definitions, I would like to see what are the risk factors for penicillin resistant Streptococcal pneumonia, and mostly because we know that having two or three risk factors is different to having one. I would like to see at least risk factors for otitis media produced for penicillin resistant Streptococcal pneumonia, and then incorporate in the trial, and then you can see these, you have a population with five risk factors or three or none of the risk factors. These may help.
DR. CHESNEY: I think I understand the question, and I think my answer is no, and what I think it is saying is would we break down this into a separate population, and my answer would be no to that, but I think it is a much more complex question in terms of when you rephrased the question, you complicated the issue for me even more, because you brought in age.
I am not sure that I wouldn't use age in some way, but to be very concrete, my answer is no, I wouldn't use recurrent acute otitis as a discriminating factor.
DR. RELLER: We are actually in agreement. I mean you want to use age as an additional thing, Dr. Hoberman brought that up, as well, and I think that this is not sufficient to identify the patients that you want to study, or that is an adequate separator, if you want to look at it that way.
DR. O'FALLON: As a statistician, I have to answer as if I were your statistician working with you on developing a study. After listening to the discussion here, what I would say is the factors that you have identified, I think age has to start out as being the most important one.
So, where I am going is we are going to go for stratification. Okay. Statisticians do that. I would say we are going to have to be able to stratify the population. How that will be done is a whole discussion but the principle is we have got to stratify by age to start with.
But it seems to me, listening to your discussion, there ought to be something like has this patient ever had antibiotics before, and so there will be a class of patients that have never had antibiotics before. That is one group.
Then, there is the group that have. Now, there seems to be levels of that, and how you break that down, it sounds like that is a topic for discussion that you guys have to duke it out, but it sounds like that there ought to be some sort of a prior treatment history factor.
So, I think that they sound like the two things, an age factor and a prior antibiotic therapy factor that ought to be involved, and this one isn't it.
So, I vote against this one.
DR. RELLER: Dr. Giebink.
DR. GIEBINK: I will tell you what I believe, but I don't know, given the question, whether to say yes or no.
DR. RELLER: Well, we are actually more interested in the comments and what you believe than a yes or a no.
DR. GIEBINK: Let me tell you what I believe. As long as the trial includes entry tympanocentesis, the whole business about enriching for antibiotic-resistant organisms is moot, because it will be addressed.
So, just leave that aside. There is a concern about heterogeneity of the subjects with regard to ear chronicity. So, I do believe you have to stratify for ear chronicity and probably the best parameters are recurrence and age.
So, I would stratify based on recurrence and age, and I would leave the rest of this aside, and not try to enrich for resistance.
DR. RELLER: Thanks.
DR. SUMAYA: Again, I am not totally clear on the question, but what I was interested in is in having some identification of the patients that would be a proxy of sorts for a complicated case, and so recurrence and treatment failure fall under that category, and there could be others.
I would use that as my indication of why I would favor a second tympanocentesis. This would be the subgroup that I would be in favor of having that done, because I am not favorable to doing a double tympanocentesis on all who would enter a study.
DR. RELLER: Dr. Wald.
DR. WALD: Essentially, I agree with Scott. I think that it is important to collect all the information, such as age at first episode, number of occurrences, recent antibiotic use, attendance at daycare, and then either take that into account by stratification or in your ultimate analysis. I don't think we need a separate study for those children.
DR. PATTERSON: I agree with Dr. O'Fallon and Dr. Giebink that some stratification of high risk versus low risk would be very useful to physicians in delineating the role and hopefully conservation of broader spectrum antibiotics.
DR. RELLER: I agree.
DR. BELL: I agree with Dr. Giebink that as long as ear taps are required for entry, this is moot, and so these people don't need to be targeted. I do think that we want to be sure that there is a sufficient group of penicillin non-susceptible or other drug resistant organisms in the study population to draw conclusions on them, but if the ear taps are done, this doesn't need to be required.
The only final comment is that these are common clinical problems, and somehow in the guidance to physicians, these concepts might be useful, because they are not going to do ear taps routinely.
DR. RELLER: Dr. Glode.
DR. GLODE: I don't think we need these separate groups and having companies go for separate indications. I do think that one could modify their exclusion criteria to eliminate the issue of not including children who have had recent antibiotics again if you want to enrich.
So, I favor stratification on the front end and analysis on the back end, and the microbiology.
DR. RELLER: Dr. Nelson.
DR. NELSON: In listening to this, I guess I would support if the goal is to move to riskier antibiotics that would be stronger and therefore deal with issues of resistance. It would concern me that you have narrowed your population, and a bacteriologic diagnosis, to narrow that population would be important.
I would like to clarify what I think was a misinterpretation of my earlier remarks. The reason why I felt I could not address the tap was because I hadn't heard clear diagnostic criteria for what acute otitis media is.
If, indeed, it meant bulging eardrums, I would have no problem with that. The difficulty I have is the bouncing back and forth that is going between what pediatricians do in their office, which we are all admitting is haphazard, and what actually happens in a trial.
I think the reluctance of IRB to deal with this issue is that when someone says can you tap acute otitis media, they are thinking of what happens in the pediatrician's office, and to the extent the tap is used to compensate for faulty diagnosis, I think that is a problem.
To the extent the tap is used in a narrow population defined by good criteria, that is not a problem.
DR. RELLER: Thanks for that clarification.
In my positioning the microbiology, I mean there is a clinical presentation, an examination that Dr. Pichichero went over, and others, and then there is the tap, which is the only way to establish etiology in what has been a targeted for clinical trial definition of who would be appropriate for tap in the first place.
DR. NELSON: But the key there is the skilled diagnostician who says this is an ear worth tapping, which is what I heard in his presentation as opposed to this is maybe otitis media and an ear worth treating with a drug that might not be any better than tap water or placebo, and not tapping and not going into trial.
DR. RELLER: I think we are actually in agreement and related to some of the remarks you made earlier about the IRBs, there are additional requirements that all of us face in terms of minimal training to participate in NIH grants and other things. It seems to me that clinical trials--and I think everybody in this room would agree--are far more complex that meets the eye, and if you do not have appropriate training and education to do whatever is necessary to participate is a clinical trial, you have no business gathering data on those patients because it is just going to end up with stuff that is devilishly difficult to interpret in the end.
So, it all comes together in terms of people who are appropriate candidates for entering into study in the first place by the criteria that have been discussed, having them entered by people who know what they are doing in clinical trials, and know what they are doing for procedures that might be required for an objective assessment.
We must go on because the down side, I mean we have tried very hard to have everyone have an opportunity to speak, but we are going to start losing members unless we have at least some comments on all three questions.
When the turn comes around, anything that people want to say that they missed before, that will be the opportunity.
Dr. Nelson, to finish Question 2(b), treatment failure. There is a definition of treatment failure that has been put forth here, and I would like to ask you and around the table do you agree with this as a definition that would be acceptable, not final necessarily, but is it a reasonable definition of treatment failure, and if you would change it, how would you change it.
DR. NELSON: I will confess that this is probably not in my area of expertise, but I will just say that I was impressed by the correlation between bacteriology and the symptom scores that I think were presented earlier from some of the clinical studies, and whether 48 hours was sufficient to see those changes or not would be an open question, but it would look to me like you could potentially use some of those symptoms, if you will, appropriately. The signs, I will defer.
DR. RELLER: Thanks.
DR. GLODE: I think one has to distinguish between clinical treatment failure and bacteriologic treatment failure, so in Dr. Marchant's study of the 40 bacteriologic failures, 62 percent were clinical successes.
So, I think it is very important, so treatment failure, you will have to ask me whether--I want bacteriologic failure or success I think is my definition.
DR. RELLER: We will get into this a little more with the double tap issue, but basically, if a child had persistent symptoms after 48 hours or 72 hours, whatever you want to say, or had it all over again within 7 days after finishing treatment, is that a child that, in general, in the context of a trial, that you would want to know whether the organism was gone or not gone.
DR. GLODE: That would be a clinical failure, which then would raise the question of--
DR. RELLER: Trigger a microbiological confirmation.
DR. GLODE: Yes, which may or may not be a bacteriologic failure.
DR. RELLER: Exactly.
DR. GLODE: Right.
DR. RELLER: These are basically, if you want to get right down to it, that if you were a double tap believer, would these be children that, at a minimum, you would want to re-tap?
DR. GLODE: Except I would change 48 to 72.
DR. RELLER: Thanks. That's exactly what we want to hear. I mean what you would do.
DR. BELL: I agree. I don't have anything more to add.
DR. RELLER: The 72 hours has been mentioned earlier. I think that is what I would do is I would give them 72 hours, and by that time, it should have done what it is going to do or not.
Dr. Dagan mentioned about what to call day 1, and it is sort of like tertian malaria. I mean it gets very confusing. There is 40 hours between the cycle, but day 1 is day 1, day 2, day 3, it is actually only 40 hours between, so 72 hours of treatment it would be if you start at day 1, and then day 4.
DR. PATTERSON: I agree.
DR. WALD: I think there are two issues. There is no improvement by 72 hours, with which I agree, there is worse at any time, so if a child deteriorates in 24 hours, that's a failure.
I don't think I would call it a treatment failure, I would call it an early recurrence for what you are calling post-therapy, because that could be anything. It could be a brand-new infection. So, it's a second early infection.
DR. RELLER: Would you like to know microbiologically what the status is?
DR. WALD: I would.
DR. RELLER: Good.
DR. GIEBINK: I would use 72 hours after initiating treatment for the during, and 1 to 5 days after the end of therapy.
DR. O'FALLON: This is hardly my area of expertise. What I am hearing, I agree with what you have said before, and I just want to make the comment that I was not all that impressed by the correlation between the clinical and the microbiological points.
That is very debatable from a statistical point of view, and it needs more discussion.
DR. CHESNEY: I like Dr. Wald's comment of worse at any point, and I would defer the 48 to 72 hours to the experts. I also would agree with Dr. Giebink that 1 to 5 days after completing the course.
DR. RELLER: Dr. Ramirez.
DR. RAMIREZ: I agree. In most respiratory infections, we use 72 hours. We need to give at least 48 to 72 hours to the antibiotics to start having some killing or bacteria decrease to see clinical response in at least 72 hours, I want to take the chance now to go back to the prior question, because I think that some members of the committee are missing or at least I consider that the enriching population, what we discussed here before, was that yes, you want to do a tap, eardrum tap.
Then, you say, well, I look for the resistant organisms, but you have in the population, 20 percent of resistant pneumococci, and 50 percent of otitis media is caused by pneumococcus, and then you have 10 percent of all the bacterial otitis are going to be resistant pneumococci.
They will say to a company go ahead, do 100 taps to get the 10 percent resistant pneumococci. What we are saying is that in enriched population, we are saying we have these inclusion criteria, if you don't meet this inclusion criteria, you don't get into the study.
Then, we are going to need probably 30, 40 taps to get this. They were worth doing, we increased the population, not to enroll 100 patients, again, only 10 patients for the study, just from those 40, I get 10 patients for the study. To me, the idea of enriched population in clinical trials looking for penicillin resistant is very valid.
DR. RELLER: Thank you.
DR. EBERT: I agree with the treatment failure during therapy being at 72 hours or after 72 hours of therapy. As far as post-therapy, I am reading within 7 days as meaning 1 to 7 days after therapy, and I will defer to the experts whether it should be 1 to 7 or 1 to 5.
DR. RELLER: Dr. Leggett.
DR. LEGGETT: Ditto.
DR. RELLER: Dr. Cross.
DR. CROSS: And the same.
DR. RELLER: Question No. 3. Do double tympanocentesis trials have a role in demonstrating effectiveness of drugs for general AOM, for recurrent/ treatment failure AOM?
Then, you can see all of the related issues about timing, relative importance of clinical and microbiology assessments, et cetera.
I think lest we lose some members, it is now 3:30. We can continue on as long as there is a healthy number. There has much discussion, some allusion to this before, but let's start with you, Alan.
Dr. Cross, what do you see as the role, if any, for double tympanocentesis trials for demonstrating effectiveness of drugs?
DR. CROSS: I think they are essential. I think we saw some data early on that showed a very good correlation on some limited data, on clinical outcome after doing studies with a single tap. Perhaps at some point in the future, we will reinforce that data. If after an initial tap, a patient does well clinically, we might not need a second tap, but that's in the future. We still have to firm up that correlation. I think it's essential we do double taps.
The timing of the second tap, whether it's during therapy or at the end of therapy, I am not sure. We heard positions at both ends, that perhaps end of therapy is better than during therapy, obviously unless a patient is worsening.
I am not sure if there are any other issues in this last question that you would want us to address.
DR. RELLER: Thank you. This is great. The comments like Dr. Cross has made for or against, and then the additional discussion points we had scheduled until 4 o'clock, so let's go around on Question 3, the central issue about double tympanocentesis, and then we will fit the remainder of the discussion in the time allotted, and then that's it for this meeting.
I think from an optimist's standpoint, that there has been clear demonstration of the Agency's commitment to pursue and revisit these issues for however many times and however many decades it takes to get it as close to right as possible, and to revisit it to keep it right.
DR. LEGGETT: In talking about this, I go back to the question I had before, how are we going to avoid another azithromycin problem without some sort of confirmation that it actually works. So, whether you call it a double or a single, and then with failure as long as you actually see somebody at day 4, or whatever it is, and decide at that point to do the double tap or not, I will leave to the experts and people arguing with IRBs, but we need to have some confirmation that the drug is actually working against the bacteria, against what it is supposed to be doing.
DR. RELLER: Dr. Ebert.
DR. EBERT: I think double tympanocentesis does have a role. I am very strongly in favor of second taps in patients who have clinical failure based on the data that Dr. Giebink presented, as many as 50 percent of those patients will have a positive culture.
I am also supportive of double taps in a smaller number of patients where you may still see clinical response, but still looking for recurrence or persistence of the organism, but I a hoping that that will not need to be as large of a patient population as the primary descriptor of where you have just a single tap.
I am hoping that our earlier suggestions of assessing clinical response at the end of therapy as opposed to at a later time point, will help us to delineate some of the issues that Dr. Leggett mentioned.
DR. RAMIREZ: I think it was mentioned in the presentations that the use of an antibiotic, that you may decrease the inoculum of bacteria to the point that the patient clinically respond, but without clinical cure.
In these group of patients is when we may see a relapse. I think that is going to be necessary to ask, that we are asking the antibiotics to kill the bacteria, it is going to be necessary to have repeat taps in as many number of patients as the statistician requires to see if there is any difference between one antibiotic and the other.
I totally agree with Dr. Dagan regarding the education of the IRB, because if we are convinced that a poor antibiotic that doesn't stabilize the middle ear, is going to be on schedule with relapse, and repeat the tap is going to be necessary, and repeat tap is no good even though the patient may be doing clinically better, still is going to be an indication to see if this antibiotic is really going to prevent relapse.
It may be even beneficial for this patient, and be beneficial for the future to see what is the best antibiotic to use for otitis media. I don't think there is an ethical issue to repeat a tap when you are really defining what is going to be the best antibiotic that you need to use in this disease.
DR. RELLER: Dr. Chesney.
DR. CHESNEY: Could I have another day or two to think about this? Let's see. Double tympanocentesis trials, I feel definitely have a role in both (a) and (b). The timing of clinical assessments and of microbiologic I think should be between that 48 to 72 hours, and I think the clinical should be obviously end of therapy and even beyond that.
On-therapy tympanocentesis in a child who is clinically improving, I think that is what we are all having trouble with, and that is the one that I feel like I would need more time for, but I think we probably do need to do some number who are clinically improving.
In order to answer the third bullet, which is can we predict clinical outcome based on the on-therapy tympanocentesis, so to me you would have to do double studies in order to answer the third bullet, and the fourth issue is I don't think there would be any problem finding enough study sites in the United States.
DR. RELLER: Thanks.
DR. O'FALLON: Yes, obviously, I think that the double tap is essential. Now, my reason is a little different. Everybody is worrying about the ability of the tap to predict the clinical response, which is important, but I am more worried about the clinical response being used to predict the microbiological one, and I am not impressed with the--well, let's put it this way--I am impressed with the misclassification rates between the success and failure in those two endpoints. I think you had better go back and take a look at them and see if you really think that is such a good idea.
So, yes, I think double taps are needed in both kinds of studies.
DR. RELLER: Dr. Sumaya.
DR. SUMAYA: I would favor the double taps for the treatment failures of acute otitis media. Presumably this would occur at around 72 hours after initiation of therapy.
I would also advocate for a tighter clinical evaluation at entry and then at 72 hours and probably at the end of therapy, as well, and very interested in the scale that is used, but more particularly in the criteria that are used within the scale of clinical assessment and if it could be made into a semi-quantitative type of an assessment, I think would be of value.
I am not in favor of a double tap in general acute bacterial otitis media unless there is some type of complication.
DR. RELLER: Dr. Wald.
DR. WALD: I certainly agree with doing second tympanocentesis in any treatment failure, and while I think the microbiologic data on repeat taps, even where there isn't treatment failure is of interest. I like to look at microbiologic data, and I think it teaches us something.
I don't think it is essential for judging outcome in the majority of patients because, in fact, there is a reasonable correlation between the bacteriology and the clinical outcome.
Some of the differences that we may see in children who are bacteriologic failures and clinical successes may be a function of the fact that we don't stop there beyond day 4 or 5 or 6, in fact, we continue treating the majority of those patients until day 10, and by that time, they may be a bacteriologic cure.
I think when we look at the data that exist, we need to look at that precisely were they children tapped on day 4 or 5 or 6. I think we are going to see differences according to the duration of therapy.
DR. PATTERSON: I think double tap studies have a role in efficacy studies as a subset in some centers where they are routinely done. I don't think the efficacy studies should be exclusively double tap studies because I think we need probably a broader geographic range for pathogens and susceptibilities for where those might be done.
I think that in single tap studies, the second tap is useful for therapeutic failures particularly with regard to the resistance issue and how to direct the use of broader spectrum agents.
DR. RELLER: I think there is an important role for double taps. Perhaps the only exclusion would be a patient who at the appropriate time of follow-up, who is doing well, and on examination by an experienced investigator, is so fortunate to have no evidence of the signs and symptoms that caused them to be enrolled in the study in the first place.
DR. BELL: I think double taps are nice, but in terms of an FDA requirement, I do not think they should be required for the patient who is clinically improving. For treatment failures, I want to see the information. Whether it should be required, I guess I would like some more input from people who have done these studies as to how feasible this is and how much information it provides, but I would very much, I would like to see it for treatment failures. I don't think it should be required for people who are improving.
DR. RELLER: Dr. Glode.
DR. GLODE: I think they do have a role and I agree with what most other people have said here, that for treatment failures they have a role, and I think in a small group of children, the double tap studies are also important. If you don't do them, it looks to me from the information provided you will overestimate the efficacy of the drug if you believe in bacteriologic eradication.
Now, it could be as Dr. Wald said, that if we were doing quantitative cultures, we would find that when you tapped them on day 3, they are still positive, but it's a 2 log kill, and that is why they are a clinical success, but in the absence of that knowledge right now, I think in a small study that there should be smaller studies of two taps.
DR. RELLER: Dr. Nelson.
DR. NELSON: I will give an IRB answer to this. First of all, I think we all need to be better educated about the ethics of our pediatric rules in addition to IRBs as well. In a treatment failure, I would presume the tap is potentially of benefit, so that doesn't sound to me like that would be terribly controversial to do a second tap.
In a child who has already had the first tap, if we had that population appropriately defined, what you would need is the tapping down by someone with the skill to be able to argue that it is only a minor increase over minimal risk.
It certainly is an experience that is reasonably commensurate--this is the language from the regulations--with that child's experience because they just had one, four or five, six, seven days ago.
But then the other threshold is it has to be of vital importance for understand or ameliorating the child's condition, and I have heard a mixed message on that point, some saying it is vitally important, others not so sure, particularly for the children that are improving.
So, from my point of view, I would remain agnostic on that vital importance, but if you want to convince your IRB, that is what they have to be convinced that it is, in fact, vitally important and that may demonstrate the variability from institution to instruction depending upon what the investigators actually believe ought to be done for that population.
DR. RELLER: Thanks. We have 15 minutes or so for additional discussion, and I would like to pose a question related to Dr. Nelson's important comments.
For Dr. Pichichero, Dr. Hoberman, Drs. McCracken, Paradise, others, would we more often see clinical failures, that is, the symptoms that were microbiological successes, or if you had double taps, the flip side of that, because there was perhaps if quantitatively done, it would be a decrement, but not enough, and when the treatment is completed, if it's one of the courses that is longer in treatment, that it would eventually improve, and what about the issue of the proportion of children in the population that goes to daycare, I mean the higher risk patients, of the probability of having fluid that can be tapped at 3, 4, 5 days into--let's just assume that it is an effective drug, how long does the fluid last, and is there something to tap safely.
DR. PICHICHERO: On a number of the items you just voted on, you didn't ask the opinion of the consultants before you voted. I just wanted to give a few sobering facts.
Regarding the diagnosis of otitis media, for example, to define recurrent otitis media, you rely, as Dr. Paradise alluded to, it was a correct diagnosis in the past. Physicians, pediatricians who come to our CME course and see that video miss the correct diagnosis 50 percent of the time. When we have taken the course abroad, they miss the diagnosis 65 percent of the time.
So, Dr. Nelson's comments about a skilled operator to do the tap are well stated, and similarly, skilled people to make the diagnosis. The practicality is today that many of the centers enrolling children are not like our center, they are referral centers. They rely on diagnoses coming in to them for the background history, which may or may not be reliable, and I would submit that they are not reliable.
Dr. Chesney said there should be no problem getting such a number of sites. I was recently at an investigative meeting, two of them, in fact, which called for a double tympanocentesis in the protocol design.
There were 30 sites sitting approximately in each of those audiences. Three of those sites were in the United States, 27 sites were outside the United States. Dagan was at both of them. The other sites, which I chatted with Dr. Hoberman about, many from Latin and Central America, they have never done tympanocentesis double tap, so I don't know whether they are going to do it or not.
I don't know about their diagnostic capability. I don't know whether they have a certificate from outcomes management or somewhere else that they are killed in tympanocentesis, and I have a lot of concerns about some of those issues the practicality of what you might be about to mandate here.
I think double taps definitely need to be done, but I am concerned about those issues of accurate diagnosis, and for me, the ear needs to be bulging, and we don't know so much about symptoms. We don't know whether that ear tugging really means they are in pain or not, if they are irritable. Children get irritable, but if that ear is not bulging, it is not otitis media in my opinion, and if it is bulging, it still is otitis media, and it deserves to be tapped because there will be pus there, and there are two papers to say that more than 90 percent of the time, if they have not been on an antibiotic, you will get bacteria.
DR. RELLER: Drs. Chesney and Hoberman.
DR. CHESNEY: Just to make a correction. I didn't mean to imply that there were plenty of centers already set up, but I think that I am already planning to send all our house staff to your course and including all of the general ambulatory faculty, and I think that if this came out as being a requirement, then, we would become skilled at a technique probably we should all be skilled at.
Maybe that's your fault for making it look so easy.
DR. RELLER: Dr. Hoberman.
DR. HOBERMAN: I could not agree more with Dr. Pichichero with regards to the accuracy of diagnosis. I think we went by the definitions of otitis media, and they were not addressed today but they need to be more stringent than what you had as stringent in the last draft guidelines.
The repeat tympanocentesis in the case of clinical failure, I absolutely agree with it, and there has to be some limitation. It needs to happen at the end of treatment basically, but there is no need to repeat a tympanocentesis at day 25 if the child is failing because the odds of that being related to the antibiotic treatment that was used, it is nil.
The other key point is that after recent visit of Dr. Nelson to Pittsburgh, the IRB has become very, very stringent with regards to the criteria for a repeat tympanocentesis, and I think I heard--I wasn't at your presentation, but I watched the video, and I agree with the concepts that were raised there, but one thing came up which was the 13 tympanocentesis.
I was asked the question based on this protocol, whether there was going to be a 13 tympanocentesis. There should not be 13 tympanocentesis in any child. Either they get re-tapped at day 4 to 6 and the criteria that we are debating with the IRB are bulging of the tympanic membrane of 2 or 3+, or 1+ plus ear pain. Those would be the instances in which we may be allowed to repeat a tympanocentesis, of course, in anybody that has clinical failure, but not in a child that is failing at 28 days.
So the point, and you raised the question today about greater than minimal risk with no prospect of benefit to the patient, which will put us in the category 3 that requires vital importance and hoops that nobody will be able to jump over.
We still feel like the repeating tympanocentesis is greater than minima risk, but of prospect of benefit to the patient if we identify children with bulging of the tympanic membrane at day 4 to 6.
DR. RELLER: Dr. McCracken and then Dr. Bell.
DR. McCRACKEN: The point raised by Ellen, and Dr. Nelson actually also, about the rate of kill of bacteria and whether, at 3 or 5 days, it is sterile or at least nothing grows because you can't be completely certain that it is not suppressed and would grow, or whether at 8 days, it would be okay, too.
Well, there are several things about that. First, the rate of bacteriologic kill is different than the rate of eradication. Time to eradication is one thing, rate of kill is yet another, and where the comes into focus, and hasn't been done yet, but I think Ron has started to do this, is to determine the concentration of bacteria, because we know in meningitis if you have 108 organisms, and your kill is the same as for two drugs in 105, it is going to take longer. Time to eradication depends on those two factors.
However, when you look at the data and at bacteriologic eradication at 3 to 5 days from several of the studies I have already mentioned, it does correlate with clinical outcome and the argument has been with the macrolide, while they may not tell it 3 to 5 days, but they probably do at 7 to 9 days. Well, that could be because no one is going to be probably tapping at that time in the normal child, but nevertheless, the positive culture at 4 days correlates with a poorer clinical outcome, both in score and just globally.
This has come up with meningitis, too, they say why do you do 18 hours and not 30 hours?
Renata and I have talked about this. I think 30 hours is the way you do it, so that you get away from the impact of the higher concentration in some children, because by 30 to 36 hours, that has dissipated, and I suspect by 3 to 5 days, it has also.
DR. RELLER: Thank you.
DR. BELL: I was happy to hear Dr. Pichichero's comments because underlying my hesitation has been the concern that double taps, although scientifically justifiable, practically, just may not get done, and we don't have anybody from the pharmaceutical industry commenting on that here, but I am cognizant of Dr. Soreth's comments this morning about something to the extent that, you know, we have to be sure that what--I am paraphrasing it--but if we set the bar too high, then, the studies won't get done, so I think we have to keep that in mind.
DR. RELLER: Dr. Soreth.
DR. SORETH: I think an important experience that we discussed here in January of 2001, was that of GlaxoSmithKline's trials with Augmentin ES, the 14 to 1 formulation in which double taps were done in children who had, on the first tap, penicillin resistant Strep pneumoniae.
As I recall--and Dr. Winn can correct me--there were between a dozen and 2 dozen centers all told within those trials, and I think roughly half were in the United States, perhaps, and half not. I mean there were a goodly number of centers that were U.S. based that were inexperienced hands, and the percentage of positive cultures at the baseline tympanocentesis was quite high.
When I look back over many different applications that we have had in the past dozen or 15 years, there is a great range, low to high, of even in what I would submit to you on paper would be a tight clinical case definition.
They checked off the box that said bulging TM, they checked off that box. I don't know what that child's eardrum looked like because that data we don't get, we don't ask for pictures as yet, but perhaps we should.
But the box is checked off that there is a bulging TM, the box is checked off that we did acoustic reflectometry. The boxes are checked off that there is an effusion there and that the child meets the definition of AOM, presumably ABOM, and not OME, and then when you look across centers, at the rate of positive cultures on that baseline tap, it might be as low as 20 percent or as high as 90 percent even with the tight clinical case definition, so there are limits to who tight you can make it.
There probably are things that we could do in terms of assessment of one's level of training, expertise, competence, so that if you were batting .200, maybe you shouldn't be an investigator in these trials and that maybe you bat something at a minimum to be such a learned investigator.
DR. RELLER: Drs. Hoberman and Marchant.
DR. HOBERMAN: There are ways of getting pictures of tympanic membranes. We are using those systems and Dr. Smith has copies of the computer system we are using to capture pictures every time we enroll a child in an acute otitis media trial.
With regards to they are batting too low or batting to high, I absolutely agree that the batting high should be the ones entered in patients in clinical trials.
On the other hand, with regards to encouragement of re-tap of clinical failures, when clinical trials, pharmaceutical companies, quote, end quote, "encourage" investigators to do it, it doesn't happen, so there has to be some mandated proportion of children that have a clinical failure that need to be retapped.
I frequently encounter our site and a few other sites being the only sites as part of clinical trials or retapping the clinical failures. So, i would suggest a 75 percent of clinical failures if we want to learn something about it, will need to be retapped as part of the design of the study.
DR. RELLER: Dr. Marchant.
DR. MARCHANT: I think Dr. Nelson's concern about training is well taken. At our hospital, we got our pediatric ER physicians to learn tympanocentesis, and we had each on of them do a minimum of 6 taps in the OR while the patient was under anesthesia, as the otolaryngologist put in tubes as a way that they became competent, and there are available practical ways to get people to be competent in the procedure and that can deal with the issue that you raised.
DR. RAMIREZ: May I ask a question?
DR. RELLER: Yes, Dr. Ramirez.
DR. RAMIREZ: I get the feeling that sometimes we are thinking that in a clinical trial, we cannot go beyond clinical practice because it is unethical, because it seems to me that the second tape is never, unless it is a failure, is never clinical practice, but this make a definition of unethical, because when we do clinical trial for sinusitis, we require a tap. I would never tap any person with sinusitis when we see the patient in the office.
We do always clinical type things that go beyond clinical practice, and if we want to see what is the base antibiotic to treat an infection, we may need to repeat the tap even though it may not benefit these children, but, yes, sometimes you discuss with the patient, you are doing a Phase II/Phase III antibiotic study, you don't even know the antibiotics are going to work.
You may say to a patient, well, you know, this may not work, may work, but it may not benefit you, but in the future we are going to know what is the baseline antibiotic, and this is not just for you, it is for Dave, for future patients. I don't see why this would be such a big ethical issue.
DR. RELLER: Thanks.
DR. NELSON: I think going beyond clinical interventions can be appropriate. The issue is, is the risk of going beyond roughly similar to the risks of the kinds of procedures that child would experience otherwise.
The regulation specifically restrict exposing particularly a child to risk for others on that basis, but from what I have heard, it sounds to me like in experienced hands, tympanocentesis fits with something that could be done when it is not only clinical indicated for benefit, but the issue of experience and context is crucial to that decision.
DR. RELLER: Thank you. Dr. Soreth, in the time allotted, we have tried our best to address the issues put to us. I think the points that have just been made about the standards of a clinical trial with appropriate design to demonstrate efficacy and safety within the confines of independent IRB review, adhering to the highest ethical standards that have been talked about.
In my view, from summarizing the discussions, if you want to look at it, the bar needs to be raised, I think there are concerns about the stability of the bar and passing under it in the past, and this is coupled with a higher caliber of criteria as well for clinical investigators who are capable of carrying out the trials and adherent to all of the requirements including a rigorous review by institutional review boards.
The potential end result of that is greater confidence in drugs that would be approved for specific indications by the FDA, in general use by practitioners that they would do what they are licensed to do.
My final query, and this is for a future meeting , is what within the regulatory process would enable the Agency to reconsider looking backwards for drugs that may be approved now for indications that in our heart of hearts, we have grave questions about whether they do what they say they do.
DR. SORETH: I can hazard an answer.
DR. RELLER: Dr. Soreth, you got us--and colleagues--got us all together. You get the last word and then we will conclude the meeting.
DR. SORETH: Quickly, we will publish in the Federal Register the appropriate docket number to which anyone and everyone is invited to send in written comments. I don't want to give you the previous number, because that may not be the best way to address this.
We will publish it in the Federal Register and we will also put it on the web site together with slides and transcript from today's proceedings.
Secondly, to try to answer your question about what do we have within the regulatory framework to address, that which we approve, at one point in time maintaining being safe and efficacious in current times, and I think there are a couple of mechanisms that we have and a couple of databases to try to answer that.
We have with all of the caveats attendant to it a postmarketing spontaneous reporting system for adverse events for any and all drugs and for vaccines. That includes coding such reports for drug lack of efficacy for infections that one may get as a result of taking an antibiotic, in other words, there are codes and queries that you could do of this spontaneous reporting system and marry that information to usage data in a crude attempt to try to get a denominator to understand within a given drug, across drug class, within a given drug class or across drug classes, et cetera, whether or not something that used to work, might not still be working.
Perhaps more rigorous than scientific, we have theoretically surveillance data that tell us with current isolates and current antibiotics and old antibiotics what theoretically should still be covered and what ought not to be covered from isolates and from real people who have real infection.
I think that we are trying to be diligent in our efforts to embrace those two real big databases and get our hands around them to try to answer the simple question that you raised, which on inspection, is actually rather complicated.
DR. RELLER: Thank you. The meeting is adjourned.
[Whereupon, at 4:00 p.m., the hearing adjourned.]
- - -