Anti-Infective Drug Advisory Committee

FOOD AND DRUG ADMINISTRATION

CENTER FOR DRUG EVALUATION AND RESEARCH

 

"Safety and Efficacy of 1-Day and 3-Day

Dosing Regimens of Azithromycin

Suspension (New Drug Application 50-710,

Pfizer Inc.) for the Treatment of

Otitis Media"

 

 

 

November 7, 2001

Holiday Inn

8777 Georgia Avenue

Silver Spring, MD

 

 

 

Proceedings By:

CASET Associates, Ltd.

10201 Lee Highway, 160

Fairfax, VA 22030

(703) 352-0091

PARTICIPANTS

CHAIRMAN

Reller, Barth, MD

EXECUTIVE SECRETARY

Thomas H. Perez, MPH, R.Ph.

MEMBERS

Chesney, Joan P.MD

Christie-Samuels, Celia, DC., MD., MPH, FAAP

O'Fallon, Judith R., PhD

Leggett, Jr., James E., MD

Wald, Ellen R., MD

Cross, Alan S., D

Consumer Representative

Ebert, Steve, Pharm, D.

FDA

Moledina, Nasim, MD

Alexander, John, MD

Janice Soreth, MD

Goldberger, Mark, MD

CONSULTANTS

Mary Glode, MD

Richard Gorman, MD

GUESTS

Celia Maxwell, MD.

Jane L. Burns, MD

Jan Patterson, MD

Celia Marchant, MD

OPEN PUBLIC HEARING PARTICIPANTS

Michael R. Jacobs, MD

Ron Dagan, MD

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CONTENTS

PAGE

Call to Order, Introductions - L. Barth Reller, MD, Chair 1

Meeting Statement - Thomas H. Perez, MPH, Exec. Sec. 1

Welcome and Introductions - Janice Soreth, MD, Acting

Director, Division of Anti-Infective Drug Products 6

Design of Clinical Trials of Antibiotic Therapy for AOC

Colin Marchant, MD 8

Pfizer, Inc. Presentation:

The Bacterial Pathogenesis of Acute Otitis Media

Edward J. O'Rourke, MD, Prof. of Pediatrics, Harvard

Medical School 27

Azithromycin in the Treatment of Acute Otitis Media

Michael W. Dunne, MD, Vice President, Clinical

Development, Infectious Diseases, Pfizer Global

Research and Development 36

FDA Presentation

Azithromycin for AOM - Nasim Moledina, MD, Division

of Anti-Infective Drug Products 99

Open Public Hearing 114

Discussion of Questions and Vote 151

Clinical Trials of AOM - John Alexander, MD, Division

of Anti-Infective Drug Products 198

P R O C E E D I N G S 8:00 AM

DR. RELLER: Good morning. Welcome to the Anti-Infective Advisory Committee meeting, the principal topic, safety and efficacy of 1-day, 3-day dosing regimens of azithromycin suspension for the treatment of otitis media.

We will begin today's meeting with an opening statement from our Executive Secretary, Tom Perez. MR. PEREZ: Good morning. The following announcement addresses conflict of interest with regard to this meeting and is made part of the record to preclude even the appearance of such at this meeting.

Based on the submitted agenda for the meeting and all financial interests reported by the Committee participants it has been determined that all interests and firms regulated by the Center for Drug Evaluation and Research present no potential for an appearance of a conflict of interest at this meeting with the following exceptions.

In accordance 18 USC 208(b)(3) full waivers have been granted to Dr. Steve Ebert and Dr. James Leggett to participate in the discussions of the new drug application 50-710 for the treatment of otitis media.

In accordance with 18 USC 208(b)(3) general matters waivers have been granted to all participants with the exceptions of Dr. Steve Ebert, Dr. Ellen Wald and Dr. Mary Glode to participate in the general matters discussion of clinical trials of acute otitis media.

Drs. Ebert, Wald and Glode are excluded from this discussion. A copy of these waiver statements may be obtained by submitting a written request to the agency's Freedom of Information Office, Room 12A30, Parklawn Building.

With respect to FDA's invited guests, Dr. Colin Marchant and Dr. Jan Patterson have reported interests which we believe should be made public to allow the participants to objectively evaluate their comments.

Dr. Marchant would like to disclose that he serves as a speaker for Bristol-Myers Squibb; Glaxo, Smith, Kline, Roche and Wyeth Ayerst. He consults for Aventis(?) Biovel(?) Bristol-Myers Squibb, Glaxo, Smith, Kline, the Robert Wood Johnson Pharmaceutical Research Institute and Wyeth Ayerst. He is, also, an investigator on research contracts, grants received from Bristol-Myers Squibb, Glaxo, Smith, Kline and Wyeth Ayerst.

Dr. Patterson would like to disclose that she serves as a speaker for Wyeth Ayerst, Merck and Aventis. She is a consultant to Pfizer, Merck, Fusisawa(?) and Schering Plough. She, also, serves on the Clinical Anti-Infective Advisory Boards for Pfizer, Wyeth, Ayerst and Estraseneca(?).

In the event that the discussions involve any other products or firms not already on the agenda for which an FDA participant has a financial interest, the participants are aware of the need to exclude themselves from such involvement, and their exclusion will be noted for the record.

With respect to all other participants we ask in the interests of fairness that they address any current or previous financial involvement with any firm whose products they wish to comment upon.

Thank you.

DR. RELLER: Thank you, Tom.

Next, I would like to have the members around the table introduce themselves briefly and their position. We will start with Dr. Mark Goldberger at my far right.

DR. GOLDBERGER: Mark Goldberger, the Acting Office Director for Ode-4(?).

DR. SORETH: Janice Soreth, Acting Division Director for Anti-Infectives.

DR. ALEXANDER: John Alexander, medical officer in the Division of Anti-Infectives.

DR.MOLEDINA: Nesim Moledina, medical officer for anti-infectives.

DR. WALD: Ellen Wald, Chief of Infectious Diseases at the Children's Hospital of Pittsburgh.

DR. LEGGETT: Jim Leggett, Infectious Diseases, Providence Portland Medical Center in Oregon Health and Sciences University.

DR. O'FALLON: Judith O'Fallon, statistician at the Mayo Cancer Center.

DR. CHRISTIE-SAMUELS: Celia Christie, Professor and Chair, Pediatrics, University Hospital of the West Indies, University of the West Indies, consultant in pediatric infectious diseases, epidemiology and public health.

DR. CHESNEY: John Chesney, professor of pediatrics at the University of Tennessee in Memphis and Infectious Disease Division.

DR. RELLER: Barth Reller, Division of Infectious Diseases, Director of Clinical Microbiology, Duke University Medical Center.

MR. PEREZ: Tom Perez, Executive Secretary for this Advisory Committee meeting.

DR. EBERT: Steven Ebert, Infectious diseases pharmacist, Meriter Hospital and clinical professor, University of Wisconsin, Madison.

DR. CROSS: Alan Cross, professor of medicine, University of Maryland, Division of Infectious Diseases.

DR. GORMAN: Richard Gorman, pediatrician in private practice and a member of the Pediatric Advisory Subcommittee.

DR. GLODE: Mary Glode, I am professor of pediatric infectious disease at Children's Hospital, University of Colorado.

DR. BURNS: Jane Burns, pediatric infectious diseases, University of Washington.

DR. MAXWELL: Celia Maxwell, Assistant Vice President for Health Affairs and associate professor of medicine, Howard University.

DR. PATTERSON: Jan Patterson, infectious diseases at University of Texas Health Science Center, San Antonio and hospital epidemiologist for University Health System in South Texas Veterans Health Care System.

DR. MARCHANT: Colin Marchant, pediatric infectious disease, Boston University and Tufts University School of Medicine in Boston.

DR. RELLER: Dr. Janice Soreth, Acting Director of the Division of Anti-Infective Drug Products will give an introduction to today's meeting.

Janice?

DR. SORETH: Thanks, Dr. Reller.

Members of the Committee, invited guests, colleagues in industry and FDA, ladies and gentlemen, we have a full agenda today. So, I will keep my comments brief.

First, I would like to recognize several of our Advisory Committee members who are completing their term with us and rotating off the Committee.

I would ask you please to come forward and accept a certificate of appreciation from us. They are Dr. Celia Christie, Dr. David Soper and Dr. Joan Chesney.

Could you step forward, please?

Perhaps Dr. Soper is not here yet, and Drs. Barbara Murray and Wittner are, also, not here, but rotating off the Committee.

We very much appreciate your active participation on the Committee, and we fully recognize that in giving us your considered and critical thoughts you add to already full calendars and long days.

As you rotate off the Committee I would simply add that we welcome your comments on the experience as we are always open to your suggestions in improving the process.

Second, I would just like to comment briefly on today's agenda. We will begin with Dr. Colin Marchant whom we have invited to speak on the dual aspects of clinical trial design in acute otitis media, an issue that we periodically revisit. We will then have presentations from the sponsor, Pfizer and from the FDA on azithromycin on the specific data that were developed for single dose or a 3-day treatment regimen for children with acute otitis media conducted in studies that fully correspond to our current guidance document for developing a drug for acute otitis media.

The open public hearing will then follow and three speakers have come forward to present. After lunch we will resume Committee discussions of and voting on the specific data that speak to the efficacy and safety of azithromycin in various treatment regimens for acute otitis media.

Finally, we will round out the day with an FDA presentation that comes back to acute otitis media clinical trial design issues or lessons learned as it were from half a dozen or so meetings that we have had int he past 5 years on acute otitis media either as a general comment on guidance for design or on specific drug products seeking a claim of acute otitis media.

It is not the intent of the last part of today's meeting to come to any final conclusions. Rather it is intended to serve as a prelude to a future public meeting with an opportunity for all industry, academia and the public to participate where the focus would be on the guidance document itself and whether and how it might be revamped or improved.

Dr. Reller, I turn it back to you.

DR. RELLER: Thank you, Dr. Soreth.

Next we will look forward to hearing Dr. Marchant's presentation.

Dr. Marchant?

DR. MARCHANT: Good morning. Today I will talk about the design of clinical trial and antibiotic therapy for acute otitis media, and my remarks will be of a general nature, and I will not address the particular application before you today.

My remarks are based on our formulation of the issues of clinical trials and I think they are pertinent to the question of comparative trials of acute otitis media.

I won't discuss the obvious issues that we should randomize patients or we should do double-blind trials but rather I am going to focus on the issues of the outcomes we chose in these trials and the implications that result from choosing various outcomes.

Of course, we could use a clinical outcome of symptomatic improvement or a combination of the two, symptomatic and where the clinician examines the eardrum or some combination of these.

The bacteriologic outcome which refers to actually doing a tympanocentesis has shown that there is eradication of the organism from the ear and there can, also, be other combinations of this such as a combined bacteriologic/clinical one where you look at clinical improvement and then do a tympanocentesis to ask in cases where they have found where they have eliminated the organism and then finally is to look at the time of the outcome. It could be during therapy, at the end of therapy or after therapy.

The timing of the outcome is varied, often symptomatic improvement or persistent symptoms have been typically measured at 48 to 72 hours. The trials that did bacteriologic outcome typically were at 4 to 6 days and some trials have looked at symptomatic outcome at that same time. Other trials have looked at end of therapy measuring typically symptomatic outcomes but often measuring events that have occurred throughout this time period and then there is the last outcome which was the test of care where they also consider events that happen after therapy, recurrences here as your outcome.

To decide how they should study acute otitis media in terms of its clinical response in the face of effective therapy or therapy that you would like to be effective placebo-controlled trials are often very instructive as to what the natural course of disease is and may help you in planning these.

Also, studies which followed clinical outcomes with bacteriologic outcomes are also very instructive in telling us how we should plan these trials.

So, I am going to review some of the data in the literature that speaks to this. This is a randomized placebo-controlled trial done in Denmark looking at children primarily older than age 3 with severe earache. They measured a pain score. They were randomized to placebo or penicillin and you can see here that in the first couple of days there is a significant difference in favor of antibiotic treatment. This was done back in the eighties. So, penicillin shouldn't be so shocking, but you can see there is a difference here, but if you will notice out at day 4 or 5 there is no difference between the treatments. If you are asked whether antibiotics are effective here, you come up with the answer that no, they are not.

If you ask the question here, you will come up with the answer that yes, they are. So, the timing in relation to how the disease behaves is very important.

Final placebo controls, one of the ways the administration was to have been done is a trial done at the University of Pittsburgh. The trial was done in the early 1980s, comparing amoxicillin, placebo and various combinations of myringotomy. The Pittsburgh trial divided the otitis media in to severe and non-severe based on high fever and marked earache and symptoms.

One of the outcomes they looked at was initial treatment failure at typically 48 hours after starting therapy.

The group did not feel they could do a pure placebo-controlled trial in the severe group. So, they made these three comparisons, and you can see that amoxicillin and amoxicillin and myringotomy turned out pretty similar results, but the placebo and myringotomy there was a larger failure rate and a difference here depending on whether you make this comparison or that comparison of about 12 to 14 percent.

So, in the severe group you see a spread in that sort of range, between 10 and 15 percent.

In the less severe arm of the trial they compared amoxicillin with placebo. Now, instead of a 10 to 15 percent difference in terms of this initial treatment failure you see a difference of only about 4 percent. So, the severity of the disease that you are looking at will determine what your response rate may be.

Also, look here that they were able to show that amoxicillin has a lower rate effusion at 14 days, but that by 6 weeks the difference is washing away. Why? Because recurrences of therapy occur equally in those that received amoxicillin and those that received placebo and what this should suggest to everybody in terms of a scientific experiment; that is what this is, is a randomized scientific experiment is that this outcome here really has nothing to do with whether you have treated or not treated.

Let us explore that further because this is part of the test of cure outcome. Let us look at studies with bacteriologic and clinical correlations. This is a study by Dr. Leibowitz and Dr. Dagan that was presented at ICAC(?) last September. It is the third of three trials that basically show the same idea though because this has a larger number of patients that previous studies it shows it more closely.

What are we looking at here? This is a group of patients treated with antibiotics with tympanocentesis during therapy and that result was sterile but then on stopping therapy the first week, the second week, third and fourth weeks afterwards there was a clinical recurrence of otitis media that again there was a tympanocentesis to determine what the bacteriology was. This refers to infections as determined by the species, whether it was Pneumococcus or Hemophilus or by pneumococcal serotype or electrophoresis typing of the organisms and for Hemophilus influenzae they either beta-lactamase(?) positive or negative.

What this shows you is that even in the first week the majority of patients had more infections, more relapses of their infection. Those appear at least to be relapses and relapses of the bacteria of pharynx or relapses because the bacteria had persisted in the ear. We have now clinical data to address that situation. The previous data I showed you recurrences occurred at the same rate only in drug and placebo and if you look at a lot of the clinical trials that have looked at this clinical recurrence after therapy, they are the same regardless of the drug that you use. So, here is what you might say about clinical recurrences. They are not reduced by antibiotic therapy. They occur in the patients and new infections are more common than relapses.

So, in clinical trials we are going to be counting as failures the event that don't appear to have any relation to antibiotic therapy.

Let us now move on to the clinical bacteriologic correlations in double tympanocentesis studies looking at patients where the organism is eradicated for the ear versus those where it persists and look at clinical success defined as symptomatic outcome at the time of the second tympanocentesis.

You see that those that eliminate the bacteria have a much higher cure rate than those where you do not. You will, also, notice that there are some here that still fail therapy despite the fact that you have eliminated the bacteria and other studies suggest that those patients have persistent viral infections that might be responsible for these persistent symptoms.

You will, also, notice now that probability of failure clinically is much higher when bacteria persist. This is data that my colleagues and I did in Cleveland since then using a slightly different approach. Dr.Degan and colleagues did another study, again, looking at bacteriologic eradication, failure here in bacteriologic eradication, then at the clinical status not at days 4 to 6 but between days 4 and 6 and day 10, but they are finding the same thing, and that is you are more likely to fail if you don't kill those bacteria by day 4 to 6 than if you have eliminated bacteria from the ear.

So, now, I would like to pursue the implications of this. This is the data from Cleveland. The first two lines are the data I just showed you two slides ago in pie diagrams. If you kill bacteria 93 percent success rate. If you don't, 62-1/2, but in clinical studies where you don't do a tympanocentesis you will, also, be looking at non-bacterial otitis media and a response that right there was 80 percent. You can use these data to calculate the clinical behavior of the disease at various levels of bacteriologic efficacy of the drug which led to us describing the Pollyanna phenomenon.

Here is what it shows. Here is efficacy on this axis. Here is elimination of bacteria from the ear and each line adjoins the corresponding clinical response that you see clinically. A perfect drug looks worse than it is. A drug that eliminates bacteria 90 percent of the time looks worse than it really is, and not all drugs of inferior efficacy, down here at the bottom here, the yellow line which is the placebo rate calculated from Dr. Virgil Highley's(?) study all the less active drugs will look better than they are and in this and subsequent slides I am going to represent drugs that appear worse than they are and drugs appearing better than they are on the right. The predominant effect is to conclude that drugs with poor activity actually work well.

So, now I am off to design issues. Now, we have looked at double tympanocentesis study, single tympanocentesis study measuring clinical efficacy and in the single clinical efficacy you saw on the previous slide. This has very precise implications in terms of the ability of you to do a clinical trial and detect a difference if a difference is there whereas a difference between, well, a difference between say one drug and the other versus a narrow difference it takes fewer patients to show a difference in the clinical trial.

Let us look at that. So, here what I am going to show you is this. If we compare a drug with 90 percent efficacy with drugs that are 30, 40, 50, 60, 70, 80 percent efficacious and lump your different trial designs and use the sample size required to look at that for the bacteriologic outcome less than 100 percent patients are required until you get up to trying to detect a difference between 80 and 90 percent.

If you look at the clinical outcome in bacterial otitis media that is a serotype study now you are able to tell the difference between say 70 percent and 90 percent is approaching 2000 patients and if you look at clinical outcome with no tympanocentesis extremely large numbers of patients are required.

For those of you, I know there is at least one statistician in the audience, all the sample size calculations that I am doing today I am showing you sort of a significance of .05, a power of .9 as the universal method for calculating sample sizes and all sample sizes are for a two-limb trial with half the subjects in each limb, and so now in subsequent trials I am going to try to draw down and simplify the question.

Various levels of bacteriologic efficacy, 90 percent is about as good as you get with most trials in the literature. We will call this a good drug. Here is the placebo related to approximately that of 30 percent and looking at 50 and 70 percent.

So, I am going to focus our attention on those sorts of differences in bacteriologic efficacy.

So, now I am just going to show you exactly the same graph that I showed you before, but now I am just looking at the three different standards compared to a 90 percent drug and I am going to continue to do that.

The only thing I did was I changed the scale here and topped it out at 2000 patients because nobody has ever down a trial larger and there is really no point until somebody does of talking about achieving larger sample sizes.

You can see that the clinical outcomes are way off the chart and not feasible. The argument that I am putting forward and have put forward in the literature has been based on the data from Cleveland that my colleagues and I generated. Perhaps that data is very specific to the particular patients that we studied.

So, now, I am going to look at other data in the literature and make similar calculations. So, when I take the Pittsburgh trial I use these standards. They looked at amoxicillin in the early 1980s and we will assume it is about 90 percent efficacious in terms of bacteriologic eradication. That is what the studies would tell you and here is the effusion for outcome, for example. Here is tap water. That is the placebo group and here is their response rate, and all I have done is proportionately calculated these two numbers in between for the fair drug and poor drug standards.

So, now, if we look at the outcomes in the Pittsburgh trial, what do they look like? Here is the bacteriologic standard. Here is the outcome of severe otitis media. They actually looked at myringotomy but that is using a 14 percent difference, and you can see that with the fair drug you can't tell the difference. With the poor drug you need above 600 patients.

If we look at the middle ear effusion outcome it appears here and if we look at their clinical outcome in non-severe otitis media which was 78 percent of all otitis media you will have a very tough time telling the difference between tap water and a good drug.

So, now we will turn to the test of cure outcome. Here is another one of these Pollyanna graphs,the same bacteriologic outcome here, symptomatic response. Now, when you start to calculate these recurrences against therapy you will see what happens. You progressively make excellent drugs in order that other drugs look worse and worse while your fair drugs still look better than they really are compared to the bacteriologic eradication rate.

I have used a range of numbers here because the recurrence rates are going to vary by the population you look at and your exact time points but that is a pretty wide spread.

What are the sample size implications of that?

There is your bacteriologic standard. Here is your clinical outcome and as you add in the clinical occurrences after therapy in test of cure outcome you just make it harder and harder as you have a higher recurrence rate to show the difference between two drugs.

So, it makes everything look the same. I would now like to turn to some other outcomes. I draw your attention here to the data from Cleveland. We looked at bacteriologic response here and clinical response there whereas Dr.Degan looked at bacteriologic outcome here and clinical response anytime in this interval.

So, during this time after they did the second tympanocentesis the otolaryngologist in this study would evaluate the patients in terms of their symptoms and in terms of the examination of the middle ear, and if they were failing therapy another therapy was described. So, the outcome of failure could occur anywhere from this visit to this visit at the end of therapy. So, that allows us to look at end-of-therapy outcomes particularly the strategy of doing a tap and a tap of failure.

So, here is bacteriologic efficacy, and here we have the bacteriologic outcome in clinical failures the tap and tap of failures. You can see that yes, it has narrowed again but it is a little bit better than just doing a clinical outcome in bacterial cases, and again, this is all using Dr. Degan's data and if we look at the test of cure outcome using approximately a 15 percent rate which is what he found in his studies, no difference between those where the bacteria were eliminated and those where they weren't and a rate of about 15 percent.

Again, you can see it driving down. So, you can see that the choice of clinical trial design is going to determine where your efficacy is set, but it is, also, going to have an effect on the narrowness of the difference in the size of the trial.

So, here are the outcomes with Dr.Degan's trials.

Bacteriologic, a tap and tap of failures study does a little bit better but it requires still quite large numbers at the 70 versus 90 percent, the fair drug standard and again using the test of pure outcome is going to drive the sample size higher.

So, how important are these? Well, I have talked about a good drug, fair drug, poor drug, tap water. Using the relationship between clinical response and bacteriologic outcome if you treat 1 million children, 1 million prescriptions how many children would have persistent symptoms on day 3 to 6 who otherwise would have been asymptomatic if you had a perfect drug if you eliminated bacteria from the ear? Well, we won't achieve that standard, but here is the good drug standard, 20 percent.

Well, the difference between a good drug and a fair drug is 40,000 children per million prescriptions having persistent symptoms because they got a drug that was less effective than a good drug, and if we go down to tap water the difference is 140,000 minus 20,000 which comes out to 120,000 children who remain symptomatic per million prescriptions who otherwise would be better.

So, now to just draw down the see the numbers as they go by you I am going to look first at the tap water standard and study designs using initial bacteriologic diagnosis. How many patients do you need? You can see that the bacteriologic outcome and the tap and tap of failures do pretty well, clinical outcomes less well. Test of cure obviously drives the sample size from here up to here.

This is the number of patients you analyzed because about 25 percent of otitis media is non-bacterial. This is the number of patients you must recruit. So, this is the work that must be done by the investigators, by the sponsor of the trial, etc.

What about the clinical outcomes now, no tympanocentesis, again the tap water standard? The best we can do comes from the data from the Pittsburgh study with severe cases of otitis media and we are a little under 500.

I now go to data I didn't talk about before and that is Dr.Rich Rosenfeld, an otolaryngologist who is trained in epidemiology who did a meta analysis in his book, evidence-based otitis media or something to that effect, and the best outcome he could come up with in terms of its effect on the disease was clinical resolution in 7 to 14 days and using that calculation again we are all seeing a sort of neighborhood, and if we take non-severe cases we are approaching 2000 patients just to tell a difference between a good drug and tap water.

Moving up to the poor drug standard we see now that tap and tap of failures is still feasible but getting tougher, clinical outcomes going up for bacterial cases. The sample size is almost all in the thousands here for the poor drug standard and if we move to the fair drug standard, well, this is potentially feasible with a bacteriologic outcome but this is now getting extremely difficult and the clinical outcomes are really out of sight.

So, I am going to draw some conclusions because most trials are only 300, 400, 500 patients, 100, 200 patients and so forth, most trials that use bacteriologic diagnosis and clinical outcome have been too small to distinguish between a good drug and tap water and all have been too small to distinguish between a good drug and a fair drug.

Most trials using clinical diagnosis and a clinical outcome have been too small to distinguish between a good drug and tap water.

Data do not support the use of a test of cure outcome as a scientifically valid outcome in clinical trials of antibacterial drugs for acute otitis media.

Now, the sort or recommendations from this. Clinical trial data in support of licensure of antibiotics for otitis media should at a minimum at least show efficacy, that is distinguish between tap water a good drug.

I think the FDA is already heading in this direction, but again, I encourage that they should convene experts to examine the design of antibiotic trials for otitis media using scientific data and scientific principles as opposed to professional consensus and revise the guidance for industry accordingly.

The problem with the tympanocentesis studies is basically that they are difficult to do at least in the United States. That difficulty relates to the fact that it is a painful procedure and people are reluctant to do two tympanocenteses and if you look at the literature on how we relieve pain in children for this procedure there are no good studies. There is no good data. There is no standard of care and really in order to do these studies which not only enlighten us in terms of relative drug efficacy but allow us to, also, look at the relationship between the MIC of the organism and eradication and so forth to do these we really need studies looking at systemic or topical analgesia or anesthesia and I suggest that industry might want to become interested in seeing that go on, and because the bacteriologic outcome is not universally accepted and certainly not universally loved mobilize a way to make clinical trials better using clinical outcomes by more sensitive outcomes, enriched populations and these should be investigated in rigorous studies but I caution everybody that large numbers of subjects are likely to still be required if we are going to distinguish between one drug and another.

So, I thank everybody for their attention, and I thank the agency for inviting me to address you today.

Thank you.

DR. RELLER: Thank you, Dr.Marchant for that presentation. That will give new and enlarged insights into the discussion that we have this afternoon on the topic of future approaches to clinical trials for guidance to industry and investigators that we will cover after the discussions this afternoon.

We now turn to the topic of the rest of this morning's deliberation and the content that will be assessed by the Committee in their vote just after lunch and I at this time would like to introduce the Pfizer presentation about azithromycin and treatment of acute otitis media.

Dr. Michael Dunne, please?

DR. DUNNE: Thank you, Dr.Reller.

Good morning. My name is Michael Dunne, and I am an infectious disease clinician that is responsible for clinical development of anti-infective products at Pfizer.

We appreciate the opportunity this morning to discuss with you the data that support the use of azithromycin given either as a single dose or over 3 days for the treatment of acute otitis media.

I will be presenting an overview of the relevant preclinical data followed by discussion of the clinical trial data that supported the use of azithromycin with the shorter courses.

First, we will begin though with a presentation by Dr.Edward O'Rourke. Dr. O'Rourke is assistant professor of pediatrics at Harvard Medical School and a Director of Harvard Medical International.

Dr. O'Rourke will give us an overview of acute otitis media specifically focusing on the influence of Streptococcus pneumoniae and Hemophilus on the presenting signs and symptoms, the clinical course of the disease and the secondary complications of acute otitis media.

Dr. O'Rourke?

DR. O'ROURKE: Good morning. Thank you for the opportunity to speak with you today. I am going to talk about otitis which I think all of the pediatricians in the room understand is the most commonly diagnosed bacterial disease in children and the most common reason that antibiotics are prescribed at all.

It, also,is one of the major reasons that we have problems with antibiotic resistance in pediatrics because of the rate at which this diagnosis is made, either correctly or incorrectly and the treatment habits of our pediatricians. When we talk about tracing therapy for otitis media we certainly want to think about what the reasons are that we are treating and which are the pathogens that we should focus on, and in selecting therapy on a rational basis there are several bits of data that we would like to have to understand how to focus. One would be the prevalence of pathogens, both patients who are untreated and those who are presenting after an antibiotic failure.

We would, also, like to know whether the clinical severity of acute otitis media relates to the pathogen.

Dr. Marchant showed us that some patients have severe disease and may be easier to make an impact on their outcome than those with mild disease.

We would, also, like to know whether untreated infections resolve by themselves and whether that varies by pathogen, where is a risk of complications and whether that varies by pathogen.

We see here a commonly displayed summary of the bacteriological data from Pittsburgh in the 1980s showing that Streptococcus pneumoniae is the most common bacterial pathogen, Hemophilus No. 2, then Moraxella catarrhalis.

In this particular summary of data there were a fairly large number of patients included without bacterial isolates and if we look at data more recently from the same institution we see that in data where there are fewer non-bacterial pathogens in fact Streptococcus pneumoniae is even more prevalent in relation to the other pathogens in otitis media.

If we look at data from the nineties looking at either untreated acute otitis or persistent otitis media, this data from New York we see again a fair number of patients with no pathogen, Streptococcus pneumoniae clearly more prevalent than Hemophilus influenzae, Moraxella or Group A Streptococcus.

In patients with persistent otitis media again Pneumococcus is the No. 1 bacterial pathogen although there are a large number that do not have pathogens isolated. Hemophilus influenza and Moraxella come up second and third.

Now, with this question of what organisms are isolated from patients who have been treated and failed therapy we see that there is a difference between studies that are reporting data from the 1980s and before and those reporting in the 1990s, and fundamentally what we see is that in the 1980s beta-lactamase producing organisms, Hemophilus influenzae or Moraxella were the most likely organisms to be isolated from patients who failed therapy.

Here we see drug-resistant Strep limo(?) and there are some intermediate resistant strains reported but those would likely be different than some of the strains down here in the 1990s where we start to see much higher rates of truly resistant Streptococcus pneumoniae presenting and a fairly stable picture with regard to Hemophilus and Moraxella.

One of the studies that I just alluded to was from Kentucky in the mid-nineties and actually on the prior slide may have been misreported, but here is an example of pathogens isolated from the middle ear after a few days of antibiotic exposure that is those who are failing therapy, those who are symptomatic at approximately 2 to 3 days and therefore being re-evaluated with a tympanocentesis and we can see that overwhelmingly Pneumococcus is the organism that is isolated in that setting, that beta-lactamase producing organisms count for only 11 percent of the isolates in this study from Kentucky.

What about the issue of self-resolution? The classic study here is Harry's study from Alabama in the 1960s and his data show that Pneumococcus actually only infrequently resolves by itself within 3 to 6 days. Hemophilus resolves by itself about half the time. Moraxella more recent estimates suggest probably resolves by itself 70 or 80 percent of the time.

So, we see that Pneumococcus here is the on least likely to resolve by itself.

Are there different clinical syndromes associated with these different pathogens, that is to say are some more likely to be severe than others and should those perhaps be the focus of our therapy?

Rodriguez and Schwartz in the Washington, DC area studied 224 episodes, evaluated with tympanocentesis and clinical scoring, and they found that a characteristic of a proportion of isolates that were found, that is Pneumococcus No. 1, Hemophilus No. 2, Moraxella No. 3. They found interestingly that if you look at the patients with more severe presentation, that is fever greater than 38.3 with either red or yellow bulging tympanic membrane that 41 percent of the cases of Streptococcus pneumoniae fit into this category. Only 1.6 percent of those caused by H. flu and only 2 percent of those caused by Moraxella fit into this category of bulging TM with fever giving actually a positive predictive value of this clinical syndrome of 94 percent for etiology by Streptococcus pneumoniae.

There is no difference in that study in pain score. However, Harry, also, did a study like this back in the sixties, 858 episodes evaluated by tympanocentesis and clinical findings at presentation, found that 34 percent of the patients' middle ear fluids grew Streptococcus pneumoniae, 20 percent middle ear fluids grew Hemophilus. Severe pain was seen in 41 percent of those with Streptococcus pneumoniae versus 17 percent of those with H. flu, fever over 101 in 30 percent with Strep. pneumonia, 13 percent with H. flu, severe pain and temperature greater than 101, 12 percent of Strep. pneumo and only .6 percent of those with H. flu and mild or no pain and a temperature less than 100, 13 percent with Strep. pneumo, 26 percent with H. flu.

So, it is pretty clear evidence that these two organisms while certainly overlapping Hemophilus was more of a pest than a real pathogen in this disease.

Just more evidence supporting the idea that these organisms produce different syndromes comes from evaluating inflammatory mediators. This is IL6 isolated from middle ear fluid by pathogen. Streptococcus pneumoniae is in this first column, Hemophilus influenzae here and Moraxella catarrhalis here. You see clear evidence that higher rates of inflammatory mediators are seen with Streptococcus pneumoniae than the other two organisms.

In fact, in animal studies with Hemophilus influenzae it has been shown the peak of inflammatory mediator with this particular organism occurs before clinical symptoms can be detected, that is the inflammation is already resolving at the time that clinical symptoms are apparent.

Perhaps the major reason that we can justify treating acute otitis media is the prevention of suppurative complications. The treatment of otitis certainly can have some effect on pain, but that is more evident in those with severe disease than with mild disease as Dr.Marchant showed us with mild disease when used to treat 25 people who are going to actually benefit from therapy, but with suppurative complications as a consideration perhaps treatment can be justified.

Pneumococcus if we look at the most common or most severe and important of the suppurative complications of acute otitis media is acute mastoiditis and this is the rank order of pathogens with regard to their rates of isolation in acute mastoiditis. Pneumococcus is No. 1 in virtually every study reported around the world in the last 20 years. Drug-resistant Streptococcus pneumoniae is being increasingly reported in many of these studies although there are some studies that show while Pneumococcus is No. 1 that the rate of pneumococcal resistance is not going up quite as fast as we feared.

Group A Streptococcus interestingly has replaced Hemophilus influenzae as the No. 2 pathogen, remarkable because it is really the No. 4 on the list of causes of acute otitis media. Staphylococcus aureus and Pseudomonas come in in the three and four slots reflecting probably more chronic disease than truly acute, but I think what is notable is that Hemophilus influenzae is reported most often as not occurring at all or in studies that are done looking at children in the 1970s and 1980s up to about 5 percent.

One of the interesting points here, however, is that in the seventies and eighties we were not using the Hemophilus influenzae B vaccine and very few of these studies differentiate the Hemophilus influenzae that they report as being either typable Type B or non-typable disease.

There are occasional reports of non-typable H. flu as an etiologic agent but they are really quite rare. To my knowledge Moraxella catarrhalis has not been reported as a cause of mastoiditis at all. In summary, Pneumococcus i clearly the major focus if we are worried about suppurative complications of otitis media and Hemophilus influenzae is literally reportable as an etiology.

If we consider then that Streptococcus pneumoniae is behaving as suppurative pathogen in this disease acute otitis media what is different about Hemophilus? It is an organism that commonly colonizes the nasopharynx. So it is present in children normally in the day care age group. It is typically thought the non-typable H. flu as opposed to type B. It is typically thought to lack the ability to invade in normal hosts. It is almost never the cause of bacteremia or meningitis for example.

H. flu otitis is more likely to follow viral infection than any other bacterial organism and therefore giving the impression that perhaps even more than with the other organisms it requires some viral infection to damage the host before Hemophilus can impose disease, again, a similar point here that otitis prone children are more likely to be colonized with H. flu reflecting again the fact that Hemophilus influenzae is the No. 1 pathogen associated with otitis media with effusion as opposed to acute otitis media, and as we ponder this, along with some recent data looking at the role of Hemophilus for example in bronchitis, chronic bronchitis, its ability to attach to non-ciliated epithelial cells and to macrophages and remain viable may explain some of the role of this organism in both otitis media and other diseases, and a question comes up whether its intracellular location shields it from some antibiotic therapy or whether simply the fact that it is part of a virtual biofilm relatively metabolically inactive allows it to escape complete eradication and then recur.

In summary if we look at the relative importance of these two major pathogens in acute otitis media since Streptococcus pneumoniae is the No. 1 isolate occurring 30 to 50 percent of bacterial isolates, Hemophilus influenzae the No. 2, 15 to 30 percent in the US, Streptococcus pneumoniae is the No. 1 cause of treatment failure in the 1990s, Hemophilus the No. 2 cause. The rates of drug resistant Strep pneumo are rising faster than the rates of beta-lactamase positive Hemophilus influenzae. Strep pneumo fails to resolve on its own without appropriate antibiotic therapy about 80 percent of the time versus 50 percent of the time for Hemophilus. Streptococcus pneumoniae is associated with a severe acute otitis media syndrome versus a less severe syndrome for Hemophilus and more of a chronic otitis media with effusion syndrome. Streptococcus pneumoniae is an invasive pathogen with the potential to do real damage and long-term morbidity whereas Hemophilus influenzae is a non-invasive opportunist which may cause temporary morbidity and there are virtually no suppurative complications other than an occasion perforated tympanic membrane associated with Hemophilus influenzae in the normal host, and this organism is more likely to recur after therapy.

Thank you.

DR. DUNNE: Thank you, Dr. O'Rourke.

Why should physicians choose to use azithromycin for treatment of acute otitis media? First, it is bactericidal against the pathogens responsible for this disease. It has a pharmacokinetic profile suited for the treatment of infection. It reaches effective concentrations in the middle ear. It is used to sustain concentrations in white blood cells. A complete course of therapy can be given in a single dose and shorter courses of therapy will optimize compliance.

Azithromycin is a well-tolerated antibiotic. The side effects have been well described and are generally gastrointestinal in nature. It is recommended in treatment of children who are penicillin allergic and post approval the oral suspension has been prescribed over 40 million times.

Clinical efficacy equivalent to comparators has been demonstrated first with the 5-day dosing regimen and now extended to 3-day and single-dose therapies.

The single-dose therapy though has additional features of interest. Giving the dose all at once allow for higher peak drug levels earlier in the course of infection. As azithromycin concentrates in white blood cells delivery of the regimen early in the course of therapy takes advantage of the period of maximum neutrophil recruitment to the site of infection. Based on in vitro data entry into white blood cells, uptake per se is increased with elevated temperatures. The single dose maximizes the rate of compliance and minimizes the burden on the care giver.

The focus of today's presentation will be around the pivotal clinical trials that support the use of the single and 3-day dose for treatment of acute otitis media. Before we get to that data, however, we will briefly review some relevant preclinical information.

The development program that generated azithromycin was focused on finding a macrolide that was more acid stable than erythromycin.

That goal was achieved by insertion of a nitrogen into the macrolide ring. The consequence of this nitrogen insertion was to add a second basic site to the compound and it is the positive charge at these two sites here and here that is responsible for the propensity of azithromycin to accumulate within cells.

At extracellular pH a small percentage of the compound exists in a neutral form here. This neutral form is more easily able to traffic through cell membranes.

Once exposed to the lower pH in acidified vacuoles more of the drug becomes proteinated leading to accumulation within this space here. This accumulation within cells that have acidified vacuoles such as neutrophils and monocytes is important in a disease such as acute otitis media, given the infiltration of these cells into the middle ear during the course of infection.

This is an H&E stain of a cross section of the mucosa of infected middle ear. The lumen is here. The mucosa down here.

One can see the infiltration into the mucosa by neutrophils. Presumably it is through this infiltration to the middle ear space by neutrophils that azithromycin is delivered to the site of infection.

It is, also of interest to note the extent of the tissue involvement in this disease process. Evidence that azithromycin accumulates within the middle ear space comes from a number of pharmacokinetic studies in which levels were measured within the middle ear.

In the study by Skaglione, this top study here one can see that the distribution between cells in the extracellular compartment is similar within the middle ear space to what we see in the blood.

In general these levels exceed the MICs for sensitive pathogens responsible or acute otitis media. The in vitro microbiologic profile of azithromycin has been extensively looked at over the last 10 years.

In a recently published study the MIC 90 to Streptococcus pneumoniae for over 4000 organisms was found to be 2 micrograms per ml.

Similarly in over 3000 isolates of Hemophilus influenzae the MIC 90 was 2 micrograms per ml. For Moraxella catarrhalis the MIC 90 was found to be less than 0.12 micrograms per ml.

While MIC 90 for Moraxella and Hemophilus has not changed over the last 10 years the MIC 90 for Streptococcus pneumoniae has been increasing. There are two major mechanisms of resistance used by Streptococcus pneumoniae against macrolides. The first works through an efflux pump. The MIC 90 for 70 isolates of Streptococcus pneumoniae harbored in a FA efflux pump was found to be 8 micrograms per ml.

The other mechanism is coded by an erm(?) d-methylase. Among 65 strains carrying the resistance mechanism for erm-d-methylase, among 65 strains carrying the resistance mechanism for erm-d-methylase the MIC 90 was greater than 128 micrograms per ml.

There has been identified an association between resistance to penicillin and resistance to macrolides. Among 1200 isolates of Streptococcus pneumoniae susceptible to penicillin the MIC 90 to macrolides was 0.25 micrograms per ml.

Among isolates with high-level penicillin resistance the MIC for macrolides increases to 32 micrograms per ml.

Presented here is similar information examined in a different manner. Among isolates susceptible to penicillin only about 3 percent were found to be resistant to erythromycin. For isolates demonstrating high level resistance to penicillin as many as 61 percent were found to be resistant.

The phenomenon of cross resistance is seen across other antibiotics including another beta-lactams F-furoxine(?) trimethaprim(?) sulfa and to a lesser extent with tetracycline.

This list of antimicrobials, also, tracks with age such that isolates obtained from younger children are more likely to be resistant than those obtained from those over the age of 13. For example, 63 percent of isolates obtained from children under the age of 2 were susceptible to erythromycin in this series compared to 80 percent with isolates takes from those over the age of 13.

Resistance to antimicrobials including macrolides has been increasing over the last 10 years. Presented here are data from eight published surveillance studies examining pneumococcal resistance to either macrolides or penicillin, and the penicillin break points are given here in these studies. These red squares here are non-susceptible isolates. The green squares are those that are high-level resistant.

In 2001 in vitro resistance to macrolides is seen in up to 25 percent of isolates with high-level resistance to penicillin in up to 20 percent.

In general macrolide resistance in the US has been associated with the efflux pump mechanism.

These isolates were obtained from a variety of sources. So a range of resistance rates can be seen in any one year surveyed. Generally resistance to community-acquired respiratory tract isolates is higher than that obtained from steroid sites.

Now, focusing on susceptible isolates, in vitro experiments demonstrate bactericidal activity of azithromycin against Pneumococcus.

In this time kill experiment Streptococcus pneumoniae is exposed to azithromycin at 2 and 8 times the MIC. After 24 hours of incubation there has been a 3 to 6 log reduction in the quantity of organisms in culture consistent with a bactericidal effect.

Similarly bactericidal activity has been seen for azithromycin against Hemophilus. In this time kill experiment Hemophilus influenza is exposed to azithromycin at the MIC and at 4 times the MIC. After 24 hours in incubation there has been a 4 and 6 log reduction in the quantity of organisms in culture again consistent with bactericidal activity.

We now move to in vivo data derived from animal models, specifically the chinchilla model of acute otitis media. In this experiment presented at ICAC last year by Franz Bable from Steve Pelton's group chinchillas are infected with non-typable Hemophilus influenzae.

Starting 2 to 4 days later they undergo tympanoscopy as well as tympanocentesis at various time points with bacterial cultures as well as drug levels.

Dosing is started on day zero with either 30 milligrams per kilo or 120 milligrams per kilo orally each day for 5 days.

Drug levels from the middle ear are presented here. Peak levels are seen at day 5. Total middle ear fluid levels which would include any intracellular as well as extracellular drug are 9.6 micrograms per ml on the 120 milligram per kilo dose and approximately 3 micrograms per ml on the 30 milligrams per kilo dose.

These total middle ear fluid concentrations are within the range of levels obtained from children getting a 30 milligram per kilo dose. The extracellular levels from a 30 milligrams per dose are lower than that of the total levels you can see here. These are the extracellular levels. These are the total levels,

Not shown here the serum levels seem with the 30-milligram-per-kilo dose most closely to approximate the serum levels seen with a 30-milligram-per-kilo dose given in children.

Quantitative bacterial cultures from the middle ear were performed. One can see it on day 3 no significant reduction in bacterial counts was observed compared to controls. By day 5 the mean counts on the 30-milligram-per-kilo dose were approximately 2 logs lower than control while those from the 120-milligram-per-kilo dose were 5 logs lower.

Bacterial eradication continued through day 11. Between this slide and the next one there are three points to be made. The first is that at doses at least in the range of what could be seen in children there is a significant antimicrobial effect of azithromycin against Hemophilus influenzae and that effect is observed as a dose response.

The other two points focus on methodology. The second point is that the timing of the culture significantly affects the impression of antimicrobial effect. For example, if day 3 in this experiment was the only time point looked at one could conclude that neither regimen has activity.

In contrast at day 5 one could conclude that the 120-milligram-per-kilo dose is very potent and the 30-milligram-per-kilo dose while not fully clearing the infection has been effective enough to reduce the concentration of organisms by 2 logs.

The third point to be made is done by contrasting these findings with an analysis of the same data looked at in a qualitative fashion. Here is the same information on antimicrobial effects. It is reported as cultures being either positive or negative. Now, one could interpret the day 5 data as reflective of a 50 percent failure rate for the 120-milligram-per-kilo dose and a 95 percent failure rate for the 30-milligram-per-kilo dose, a very different picture of antimicrobial activity from what was previously demonstrated.

So, the sensitivity of the test methodology must be considered in forming an interpretation of antimicrobial activity in these kinds of settings.

We turn now to the clinical program where we will start with an overview of the data at the point of use of azithromycin as a 30-milligram-per-kilo dose now given over 5 days, a 5-day regimen.

These were the pivotal studies that supported the 5-day program. In the two comparative studies patients given azithromycin had a clinical cure rate similar to the children taking the comparative agent. There were two studies that identified pathogens at baseline.

The clinical success rate at day 30 for the 56 children with Streptococcus pneumoniae identified at baseline was 71 percent. With 47 with Hemophilus influenzae identified at baseline it was 64 percent and for the 26 children with Moraxella catarrhalis it was 73 percent. The results of the smaller comparative study were similar.

Of the 975 patients who received azithromycin 7.2 percent developed an adverse event related to drug compared with 23 percent of the 827 children who received amoxicillin/clavulanate. The most common adverse event seen was diarrhea at 12.6 percent for children getting amoxicillin and clavulanate; vomiting and abdominal pain were also seen.

We will come back to these side effects later when we compare them to the shorter courses of therapy with azithromycin.

Based on the animal experiments and pharmacokinetic properties of azithromycin it appeared that the total dose and not the duration of dosing might be most relevant to treatment outcome. Given this background and the ongoing medical need for therapies which could improve compliance rates we undertook a program to reduce the dosing duration for treatment of acute otitis media to under 3 days and then to a single dose therapy.

I would just like to point out that the regulatory implications of this program were not to gain an indication for acute otitis media as that was previously established with the 5-day program but rather to adjust the dosage and administration section of the label to allow for 5, 3 or 1 day of dosing.

Now, for orientation the actual daily dosing of these regimens is provided here. The 5-day dosing regimen is given at 10 milligrams per kilo on the first day and 5 milligrams per kilo each day from days 2 to 5.

The 3-day dosing regimen as given as 10 milligrams per kilo each day for 3 days and the single dose regimen provides the entire 30-milligram-per-kilo dose in a single administration.

The program to study these shorter dosing regimens follows the 1998 FDA guidance on the design of studies for acute otitis media. These guidelines require one statistically adequate comparative study with clinical end points and a test of cure visit 2 to 4 weeks after conclusion of therapy.

In addition there should be one non-comparative trial with tympanocentesis at baseline to identify pathogens of interest and clinical cure rates for these pathogens should be clinically acceptable to find as comparable to a labeled comparator drug.

The program had pivotal and supportive studies. The pivotal studies in support of the single-dose program were study R-0581, a double-blind randomized comparative clinical trial and study 1015, a non-comparative study with tympanocentesis at baseline.

There was one comparative double-blind randomized clinical trial in support of the 3-day dosing. It was agreed that microbiologic data from the single dose program could support the 3-day program given the data already available from 5-day dosing.

There was one Phase II supportive trial that served as a pilot study and it compared azithromycin as a single-dose therapy with 3 days of dosing with azithromycin with ceftriaxone given intramuscularly. Patients had a baseline tympanocentesis in this study as well.

We will start by reviewing the single-dose program. The best place to start to understand comparability between all the studies in this single dose program is to look at who was included in the trials. We will do that by reviewing the inclusion and exclusion criteria for the single-dose program overall.

Now, this is a busy slide, but I will draw your attention to the most important points. First you will see that there are four columns at the top. The three columns on this side are the pivotal studies in the program. This column provides a reference point for the 1998 FDA guidance specifically in this slide on the symptoms consistent with acute otitis media.

All three studies allowed for enrollment of children down to 6 months of age. Ninety-eight percent of the children in study 1015 had either ear pain or fullness. Seventy-seven percent of the children in R-0581 had ear pain. Ninety-three percent had either ear pain or a history of fever and 96 percent of the children had ear pain in study 95001. So, all of these children appear to have symptoms consistent with acute disease.

Presented here are the inclusion criteria for signs of tympanic membrane disease. Over 90 percent of the patients had at least one abnormality in study 1015. The majority had more than one. All patients in both 1015 and R-0581 were to undergo electroacoustic reflectometry and have acoustic gradient angle of less than 70.

Ninety-nine percent of those in R-0581 and 87 percent of the children in 1015 had that finding. The majority of the remaining children in 1015 had a perforated eardrum. So, they couldn't undergo that test. All of the children in 95-001 had a bulging eardrum at entry.

Recent antibiotic use was exclusionary in each study. Though not specifically indicated in every protocol no child was enrolled with typanostomy tubes present or who had otitis externa.

The primary end point was clinical cure. The primary time point for test of cure was day 28. In addition to cure and failure the assessment of improvement at day 28 was collected in study R-0581 and 95-001. As a result the comparison of clinical success which is the sum of cure plus improved was, also, analyzed for these studies at the day 28 time point. This clinical success assessment was made at the end of therapy visit in all the studies and is examined as a secondary end point.

The population of patients analyzed for clinical efficacy at day 28 included patients who took at least one dose of study medication, had a diagnosis of acute otitis media and returned for a visit at day 28.

Anyone who received a concomitant antibiotic for failure was carried forward as a failure. The population evaluated for bacteriologic response included all the clinical efficacy population who also had a pathogen of interest isolated at baseline.

The assessment of cure was defined as complete resolution of specific signs and symptoms of acute otitis media. Following recent guidance for the two more recent studies, 1015 and R-0581 the presence of a middle ear effusion per se did not preclude an assessment of cure. So, some residual signs of effusion did not prevent the investigator from calling that patient a cure.

So, we will focus on the pilot study first. Study 95-001 was the first study designed to assess the activity of azithromycin given as a single dose for the treatment of acute otitis media. It compared the single-dose regiment with azithromycin given for 3 days with ceftriaxone. This was a prospective single center study performed in Costa Rica. The patients were screened and randomized to azithromycin as either a single dose or a 3-day dose or to ceftriaxone. Each of the azithromycin regimens was given with a placebo oral suspension to match the other.

Ceftriaxone was given intramuscularly at 50 milligrams per kilo as a single administration. There was no blinding for ceftriaxone as it was not felt appropriate to give placebo injections to the other groups.

In an attempt to blind this regimen though the physician assessing the clinical response did not administer the study drugs.

Typanocentesis was performed at baseline. Follow-up occurred at typical intervals through day 28.

Sixty-six children were randomized to each regimen. All of these children were available for a safety assessment and all but three were included in an assessment of efficacy at day 28. Approximately one-half of the children had no organism identified at baseline.

The mean age of children enrolled was 2.4 years and roughly 40 percent were under the age of 2. The distribution of demographic characteristics was comparable among the treatment regimens.

The mean duration of symptoms of acute otitis media prior to randomization was 1.5 days for those given azithromycin as a single dose, 2.4 days for those given a 3-day dose of azithromycin and 1.7 days for those given ceftriaxone.

Analgesics or antipyretics were used during the course of the study to a similar extent by children on each regimen. Symptomatic medications for respiratory tract infections were infrequently used throughout the observation period. The clinical success rate at day 14 was similar on each regimen as you can see here.

At day 28, again, there were similar success rates and cure rates, success on this line, cure here. Provided for reference is the 95 percent confidence interval on the difference in outcome between the single dose azithromycin and ceftriaxone.

Clinical outcome was, also, stratified by age. Clinical cure rates at day 28 were similar between regimens for each group. As is typically seen the cure rates for children less than the age of 2 are lower than those for children over the age of 2.

Fifty-seven of the 60 Streptococcus pneumoniae isolates obtained during the study had susceptibility testing performed. Forty-six of those were susceptible to azithromycin and 11 were resistant. All isolates were susceptible to ceftriaxone.

All isolates of Moraxella catarrhalis and Hemophilus influenzae were susceptible to both azithromycin and ceftriaxone. Of patients in whom Streptococcus pneumoniae was identified at baseline the clinical cure rate at day 28 was 85 percent in the 20 patients receiving a single dose of azithromycin, 94 percent for the 17 patients receiving 3 days of azithromycin and 83 percent for the 23 children receiving ceftriaxone.

Of patients for whom Hemophilus influenzae was identified the cure rate was 88 percent in the eight patients receiving a single dose, 69 percent in the 13 patients receiving 3 days of azithromycin and 89 percent in the nine children receiving ceftriaxone. There were only two patients where the Moraxella catarrhalis was identified.

Coming now to safety, 11 percent of children given a single dose treatment of azithromycin had a treatment-related adverse event compared with 9 percent of those given 3 days of azithromycin and 9 percent of those given ceftriaxone.

Similar rates of diarrhea were seen on each treatment regimen with rash more common in the ceftriaxone-treated group and vomiting at 5 percent more common with the single-dose therapy with azithromycin.

There had been a concern that giving the entire dose of azithromycin all at once would result in an unacceptable GI side effect profile. The adverse event rate seen here though was encouraging and supported the clinical development program for single-dose therapy.

We turn now to study R-0581. This study was designed to collect comparative clinical data for the single-dose regimen. It was a multicenter randomized double-blind comparative study of azithromycin given at a single dose and amoxicillin/clavulanate each with matching placebos.

A history and physical exam was performed at baseline. The first dose of study medication was given in the clinic and patients were asked to wait for 30 minutes. Any patient who vomited during that time was redosed. Patients were contacted by phone between days 3 and 5 at which time data relevant to adverse events, compliance and clinical symptoms was collected.

At days 12 to 16 and again at days 28 to 32 patients returned to the clinic where data relevant to adverse events and clinical response was obtained.

One hundred and seventy-five patients were randomized to each study regimen and 173 in each group had a safety assessment. One hundred and fifty-one patients treated with azithromycin and 154 treated with amoxicillin/clavulanate were included in the efficacy analysis at day 28.

The mean age of children randomized to azithromycin was 2.7 years and to amoxicillin/clavulanate it was 3.4 years. Approximately 40 percent of the children enrolled were under the age of 2.

Regimens were balanced with respect to age, gender and race. The duration of symptoms of acute otitis media prior to randomization was 3.4 days for those given azithromycin and 3.9 days for those given amoxicillin/clavulanate.

Approximately 80 percent of the children had a prior history of acute otitis media. Analgesics and symptomatic medications for treatment of a respiratory tract infection were used during the course of the study to a similar extent by patients one each study regimen.

The clinical success rate at day 14 was similar on each treatment regimen as seen here. The clinical success rates at day 28 were, also, similar with a lower limit on the 95 percent confidence interval on the difference of minus 10 percent given here. The lower limit on the difference in cure rates was minus 7 percent.

Clinical outcome stratified by age was, also, examined. The cure rates at day 28 for children above or below the age of 2 were similar for each regimen. Again, as expected cure rates for children under the age of 2 were lower than for those children older than 2.

Ninety-nine percent of children randomized to azithromycin took their study medication compared to 83 percent of those amoxicillin/clavulanate.

Turning again to safety 17 percent of the children receiving a single dose of azithromycin reported an adverse event related to drug compared to 23 percent of those randomized to amoxicillin/clavulanate.

Diarrhea and rash were seen more frequently in children given amoxicillin/clavulanate and vomiting was seen in 4 percent of children on either regimen.

Study R-0581 confirmed and extended the efficacy of a single dose as we had seen previously in the pilot study. We will now review study 1015.

Study 1015 was designed to obtain clinical outcome information according to the pathogen isolated by tympanocentesis at baseline. It was a prospective open-label multicenter non-comparative trial.

History, physical exam and tympanocentesis were performed at baseline. The first dose of study medication was given in a clinic and patients were asked to wait for 30 minutes. Any patient who vomited during that time was redosed.

Patients were contacted by phone at day 5 at which time data relevant to adverse events was collected. At day 10 and again at days 24 to 28 patients returned to the clinic where data relevant to adverse events and clinical response was obtained. The primary efficacy end point was clinical cure at day 28 by baseline pathogen.

Organisms isolated by the local laboratory were sent to a central lab for confirmation. Two hundred and forty-eight patients were enrolled in the study and all of these were included in safety assessments. Two hundred and forty-two patients were assessed for clinical outcome at day 28. Approximately one-half of the patients had an organism identified at baseline.

The mean age of children enrolled in the study was 3.4 years. Thirty-five percent of those enrolled were under the age of 2. The mean duration of symptoms of acute otitis media was 2.5 days and 72 percent of the children had at least one previous episode of acute otitis media.

Analgesics were used during the course of study in 63 percent of the children and symptomatic therapies for respiratory tract infections were used in 19 percent.

The clinical success rate at day 10 was 89 percent. The clinical cure rate at day 28 was 85 percent. This is all patients enrolled. The clinical cure rate for children over the age of 2 was 89 percent and it was 77 percent for those under the age of 2. Again, we see that children under the age of 2 had a lower success rate than the older children.

Eighty-eight percent of 76 children with Streptococcus pneumoniae isolated at baseline were cured at day 28. Sixty-four percent of 44 children with Hemophilus influenzae isolated at baseline were cured at day 28 and all 10 patients with Moraxella catarrhalis were cured.

I would like to focus a little bit more on the isolates of Streptococcus pneumoniae identified in the study.

Of 76 patients with Streptococcus pneumoniae identified at baseline 67 isolates had susceptibility testing performed. Presented here is the number of isolates distributed by the baseline MIC.

The break point for resistance to azithromycin is 0.5 micrograms per ml which would make these 12 isolates resistant.

The resistant isolates sort into two groups, those with an MIC of 8 and those with an MIC of greater than 256 micrograms per ml.

We have some additional data on this group of 12 organisms. There were seven isolates that had an MIC of 8 micrograms per ml. All of these were susceptible to clindamycin and all had the mef-A gene identified by PCR, both characteristics of organisms that have an efflux pump mechanism of resistance.

All five of the isolates with an MIC greater than 256 micrograms per ml were resistant to clindamycin and had the erm B gene identified consistent with ribosomal resistance.

Presented here is the clinical outcome by baseline MIC for 66 patients that had clinical outcome data available. Four of the 12 patients with these resistant isolates failed therapy. You can see there are four right here. We did note that this child here had a resistant isolate recovered on therapy. So, I would, also, like to review this one patient.

This is an outline of the patient's clinical course. He presented with an abnormal reflectometry score in the left ear and tympanocentesis of that ear revealed Hemophilus influenzae and Strep. pneumoniae with an MIC of 8.

On day 4 because of persistent symptoms he came back to the clinic where the left ear had improved. However, the right ear was now found to be involved. Tympanocentesis and culture of that ear revealed Streptococcus pneumoniae but now the MIC was greater than 256 micrograms per ml. The child's therapy was then switched.

In order to determine if this was the same organism, this Streptococcus pneumoniae was the same or not but simply had a higher MIC we did pulsed field gel electrophoresis on these two isolates and that is presented here. This is the strain an MIC of eight. This is the strain with an MIC of greater than 256 micrograms per ml, and these two strains of Strep. pneumoniae were determined to be clonally distinct which raises the possibility that the child's failure on therapy was a consequence of a superinfection with a less-sensitive organism. Because of the cross resistance seen between penicillins and macrolides in Strep. pneumoniae we performed an assessment of clinical outcome by baseline penicillin susceptibility.

Forty of the 65 isolates were susceptible to penicillin as shown in the top left of the table, these 40 over here. Thirty-eight of the 40, 95 percent were assessed as cured at day 28.

Just to note all of these 40 isolates were, also, susceptible to azithromycin. Here in the left-hand column on this side there are 16 isolates with intermediate resistance and 12 of the patients with these isolates were clinically cured.

In the middle nine isolates demonstrated high-level resistance to penicillin and six of the patients with these isolates were clinically cured.

I would like to focus some more attention on these nine isolates now in the next slide.

These nine isolates were analyzed according to their macrolide susceptibility as seen in the bottom table there. Three of the isolates were susceptible to azithromycin and all three of the patients with those isolates were clinically cured.

An interesting thing to note here is that all of the isolates with the erm B phenotype demonstrated high-level resistance to penicillin. So, there was a complete correlation between those two.

Now, I should, also, note that these data are updated from what you have in your briefing document on Page 25. Now, safety. In this non-comparative study 12 percent of patients receiving azithromycin experienced a treatment-related adverse event. The most frequently identified event was vomiting seen in 6 percent of the patients. In this study of single dose of azithromycin the clinical outcome assessed in patients with infections due to the key pathogens responsible for acute otitis media was comparable to that seen with the 5-day regimen.

Okay, so now we will move from the single dose program to the 3-day program, and we will start with study 1014. The purpose of this study was to collect comparative data on the activity of azithromycin given over 3 days for treatment of acute otitis media.

This was a randomized double-blind multicenter study where patients were given either azithromycin for 3 days or amoxicillin/clavulanate at 45 milligrams per kilo per day divided b.i.d. for 10 days with matching placebo.

Children were enrolled down to the age of 6 months. Ninety-seven percent of the children had either ear pain or fullness. The majority of subjects had more than one of the typical signs of tympanic membrane disease. Ninety-eight percent of the patients had an electroacoustic reflectometry exam with an acoustic gradient angle less than 70 at baseline..

Recent antibiotic use and the presence of tympanotomy tubes were exclusionary criteria and no child was enrolled with otitis externa.

A history and physical exam was performed at baseline. Patients were contacted by phone at day 3 to 5 at which time adverse event data and compliance data were collected.

At day 10 and again at day 24 to 28 patients returned to the clinic where data relevant to adverse events and clinical response was obtained.

The primary end point was clinical cure. The primary time point was day 28 at which time a clinical assessment of cure or failure was obtained. Clinical efficacy at day 28 was analyzed for patients who took at least one dose of study medication, had a diagnosis of acute otitis media at baseline and returned for that visit.

Patients who used a concomitant antibiotic for failure were counted as failures and carried forward as such.

Cure was defined as complete resolution of the signs and symptoms of acute otitis media. The presence of a middle ear effusion would not necessarily preclude an assessment of cure.

One hundred and eighty-eight children were randomized to azithromycin and 185 to amoxicillin/clavulanate. All of these children were included in the safety analysis. One hundred and eight-two children given azithromycin and 180 given amoxicillin/clavulanate were included in the efficacy analyses at day 28. Symptoms prior to enrollment for children on either regimen was 1.2 days. Over 80 percent of the children had a previous episode of acute otitis media.

Analgesics were used by approximately half of the children on either regimen at any time during the study and 40 percent used symptomatic treatments for respiratory tract infections.

Clinical success rate at day 10 was similar on each regimen given here. Clinical cure rates at day 28 were equivalent with a lower limit of the 95 percent confidence interval on the difference of minus 5.

Clinical outcome stratified by age was, also, examined. The cure rate at day 28 for children above or below the age of 2 was similar on each regimen. Again, as expected cure rates for children under the age of 2 were lower than for children over the age of 2..

Ninety-nine percent of children assigned to azithromycin completed their treatment regimen compared to 89 percent of those given amoxicillin/clavulanate.

Eleven percent of children taking a 3-day regimen of azithromycin had a treatment-related adverse event compared to 20 percent of those taking amoxicillin/clavulanate.

Diarrhea and rash were seen most frequently in those on amoxicillin/clavulanate. Vomiting was seen to a similar degree on each regimen.

In this comparative study the 3-day regimen of azithromycin demonstrated comparable safety and efficacy to amoxicillin clavulanate.

I would like now to present a summary overview of the four studies that I just presented. As we have seen in the three comparative studies azithromycin was demonstrated to be as effective as the comparator as defined by the lower limit of the confidence bounds and the difference in cure rates.

The similarity in efficacy rates was, also, seen in the more difficult-to-treat subpopulation of children under the age of 2 and the same conclusion can be drawn from an assessment in this subpopulation at the earlier end-of-therapy time point.

An 88 percent clinical cure rate was seen in the 76 children from whom Streptococcus pneumoniae was isolated at baseline, a 64 percent cure rate in the 44 children from whom Hemophilus was identified at baseline and all of the children from whom Moraxella was identified.

Similar results were seen in the smaller comparative study. Presented here are safety data from the comparative pivotal studies for the single and 3-dose regimens as well as the 5-day regimen submitted in the original application.

The comparators have been pooled across all the studies. Treatment-related adverse events were seen in 14 percent of patients given a single dose of azithromycin, 10 percent of those given the 3-day dose and 8 percent of those given the 5-day dose. This compares to a 22 percent adverse event rate in those given the comparator drug.

Diarrhea was the most frequently observed in the comparator regimen. Vomiting was seen in 1 percent of those given the 5-day regimen here, 2 percent of those who got the 3-day regimen, 5 percent of those who got the single-dose regimen and 4 percent of those who got the comparator.

Now, there is a progressively higher rate of patients with adverse events in the 30-milligram-per-kilo dose of azithromycin as the duration of dosing shortens.

This was not unanticipated. The gastrointestinal side effects are dose proportional and the 30-milligram-per-kilo dose on day one with the single-dose therapy delivers more drug than the 10-milligram-per-kilo dose on that same day, but given the shorter duration of dosing on the single dose the likelihood of experiencing related adverse events on subsequent days would decrease. Therefore we thought it was, also, important to consider not just the number of patients with adverse events but the actual number of adverse events accounting for the duration of each event as well.

For reference we see the total number of patients that had an adverse event here. You saw that in the previous slide.

Noted below, however, are the number of related adverse events normalized for patient year of exposure. What one sees is that while more patients getting the single dose had a side effect, mostly on the first day of dosing, the number of event days of side effects was not higher than on the other regimens.

In review of the clinical data there are a few topics that were of additional interest to us and merit some further discussion.

These topics include the effect of vomiting on day 1 on the clinical outcome of children assigned to the single-dose therapy, a comparison of outcome at day 28 for published studies and empiric therapy with acute otitis media with azithromycin as well as a by-pathogen outcome assessment at day 28 and a look at outcomes at earlier time points than day 28, including the use of other measures to assess clinical response.

We wanted to assure ourselves that there was no negative impact on clinical outcome for patients assigned to the single-dose therapy who subsequently vomited. A total of 52 children, 10.7 percent assigned to single-dose therapy vomited at some point during the observation period which goes out to 30 or 35 days after dosing.

Thirty-six of these children vomited on the first day. So, there may be concern about vomiting around the time of dosing on this first day. We assessed outcome for those 36 children and compared that outcome to the group that did not vomit on that day. Ninety-one percent of the children who vomited were cured or improved at day 14 compared to 89 percent of those who did not. At day 28 the rates were 85 percent and 81 percent.

Based on this overall measure of clinical outcome at least of the patients in this program the children who vomited initially were not disadvantaged.

The previous data presented information on a population basis. The data presented here measure absorption on an individual patient basis and are taken from PK studies in normal adult volunteers given a single dose of 2 grams or 3 grams of different formulations of azithromycin, four formulations here in the first study, two formulations here, two formulations over here.

Displayed in the red triangles here are the mean AUC for each group with corresponding standard errors around it. The green triangles are individual results from patients who vomited greater than 2 hours after dosing, here, here, here and here and here, and the yellow from those who vomited within 2 hours. So, that would be these two, this one, these here, here, here.

The yellow triangles highlighted with purple dots are those that vomited within 30 minutes. There are four of those, one, two, three, four.

One can see that the values for patients who vomited fall within the standard error and are as likely to be above the mean as below regardless of the timing of the vomiting post-dose.

These two sets of data provide some assurance that individuals that vomit around the time of dosing are as likely to be clinically cured as those who do not vomit.

We will now examine the clinical response at day 28 for any azithromycin comparative study of empiric therapy for acute otitis media and then look at the clinical outcome at day 28 for patients with either Streptococcus pneumoniae or Hemophilus identified at baseline who were treated with either azithromycin or any other approved therapy for acute otitis media.

Presented here are the 95 percent confidence intervals on the difference in cure rates for all studies either published or presented for regulatory review with outcome data available at day 28.

The studies in yellow here at the top are the two studies that supported the original 5-day dosing regimen. The study here, the 1014 study supports the 3-day dosing regimen. The study, R-0581 supports the single-dose regimen and there are a variety of other published studies that have data available.

Now, in an analysis of studies that compared azithromycin to amoxicillin/clavulanate, so any one of these that had amoxicillin/clavulanate as the comparator totaled about 1800 patients, as well as an analysis of all of these studies with about 2000 patients.

The point estimate of difference in outcome is small, and there are very narrow confidence intervals around it.

Presented here is the 95 percent confidence interval on the point estimate of success. So, this isn't difference in cure rates now. This is just the point estimate of success for antibiotics that are approved for treatment of acute otitis media that have clinical outcomes data at day 28 for patients with Streptococcus pneumoniae identified at baseline.

The data is presented as clinical success. These here are clinical cure down here, and this is the as presented in the labels. For reference the outcome of patients given a single dose or azithromycin is given here. These vertical dashed lines just orient you around the upper or lower limits of the 95 percent confidence interval on that point estimate.

Now, some caution should be taken in interpreting these data. Important demographic variables like age, previous episodes of acute otitis media, other things that are important in outcome are not taken into account in looking at one drug versus another here.

These are simply the data as they appear in the label, but even so these are data that have been used for regulatory decisions and are available to treating physicians.

From these data the outcome at day 28 for children with acute otitis media due to Streptococcus pneumoniae and treated with a single dose of azithromycin is comparable to that of other regimens.

Presented here now are the 95 percent confidence intervals again on the point estimate of success for all drugs approved for treatment of acute otitis media that had data available in the product label on clinical outcome at day 28 in patients with Hemophilus identified at baseline.

The data is presented again as clinical success and clinical cure and again that is as shown in the labels.

For orientation this is the outcome of the single dose treatment here. The dashed lines are the upper and lower 95 percent confidence limits.

Again, no attempt is made to adjust the outcomes by these other important demographic variables. This is just as it is.

Given that caveat it would appear that the clinical outcome at day 28 for children with Hemophilus at baseline treated with azithromycin is comparable to that of other approved therapies.

In addition to the information at the day 28 test of cure the data is collected throughout the observation period that may be, also useful in assessing clinical response. This would include various on-therapy data such as the use of additional antibiotics for failure, the use of analgesics or antipyretics to control symptoms and the results of questionnaires aimed at the patient's impression, the parents' impression I should say of clinical outcome.

Presented here is a survival-type analysis focusing on the time to use an additional antibiotic for failure. As a parent could bring a child to the clinic at any time for reassessment it provides an additional measure of the timing of clinical failure.

If I can draw your attention to the earliest time period here and actually looking over the whole curve as well you can see that there is no obvious difference in the time to antibiotic used for failure between any of the regimens in this pooled analysis of the three comparative studies that I reviewed this morning.

Alternatively the earliest response to therapy could be measured by the use of medications to alleviate the symptoms of disease. Here I present the percentage of patients using analgesics throughout the observation period, in study R-0581 no difference in the use of analgesics was seen and the use of these symptomatic medications had dropped to 2 percent by about day 5.

Again, in study 1014 no difference in the use of these medications was seen suggesting that gross differences in the symptoms of acute otitis media among these regimens is unlikely.

In study R-0581 parents were asked at various intervals to assess how sick overall their child had been. One could see at day 3 to 5 here the parents felt that their children were improving at a similar rate between those given a single dose of azithromycin and those given 10 days of amoxicillin/clavulanate.

Following guidance documents the test of cure visit for this program occurred at day 28. Assessments of the day 10 to 14 end-of-therapy visit though were also performed.

Presented here are the 95 percent confidence intervals on the difference in clinical outcome for all clinical trials comparing azithromycin to a beta-lactam that had an assessment of clinical outcome at days 10 to 14.

There are 5-day dosing studies again. These are the ones that supported the original application, a 3-day study I reviewed this morning and R-0581 here supporting the single-dose therapy. In general one can see that the difference in success rates was small in each of these studies and a pooled analysis of approximately 3200 patients comparing azithromycin with amoxicillin/clavulanate and again here in analysis of almost 4000 patients comparing all the patients in this program one can see that the difference in success at day 14 was less than 2 percent with very narrow confidence limits.

This is for us to examine the outcome of children treated with a single-dose of azithromycin for acute otitis media due to Hemophilus at the day 10 to 14 time point relative to other approved therapies.

Presented here is the 95 percent confidence interval on the point estimate of success for antibiotics that are approved for treatment of acute otitis media and clinical outcome data at day 14 for patients with Hemophilus isolated at baseline.

For reference the outcome of patients given a single dose of azithromycin is up here and again we have the dashed lines that give the upper and lower limits of the 95 percent confidence interval. Again, these are point estimates of success.

The same cautions we noted previously about comparisons between the drugs are important. With those caveats, however, the outcome at day 14 for children with acute otitis media due to Hemophilus and treated with a single dose of azithromycin is comparable to other therapies.

In conclusion then we find that the data presented this morning demonstrate that azithromycin given as a single dose or over 3 days is an effective empiric treatment for acute otitis media.

Adverse event rates were similar or lower than the comparator and are generally limited to the gastrointestinal tract.

We believe that these shorter courses of therapy will optimize compliance while easing the burden on the care giver.

Thanks for your attention. I would be happy to answer any questions.

DR. RELLER: Thank you, Dr. Dunne.

Questions for Drs.Dunne or O'Rourke from the Committee.

Dr.Chesney?

DR. CHESNEY: Would you mind reviewing Slide No. 104 for us again, please?

DR. DUNNE: Main body presentation No. 104, please?

I will go over this again. This is data taken from studies in adults. It is pharmacokinetic data. It looked at different formulations of azithromycin. They were given as 2 grams as a single dose or 3 grams and there are different groups you can see in each study depending on the formulation that they received.

These are the PK parameters that were collected and this particular slide looks at area under the curve. The red triangles here are the means, and there is a standard error bar around each of the means.

The green triangles here, so this one, this one, this one here, here, here, these are individual values. They are individual AUCs for the subjects that vomited at greater than 2 hours, right? Yes, greater than 2 hours post-dose.

The yellow triangles are the AUC measurements for the individual patients that vomited within the 2-hour window, and then we highlighted the ones that vomited even within 30 minutes and that is as you can see there in the purple dots.

DR. CHESNEY: Those weren't levels drawn from the patients though?

DR. DUNNE: I am sorry.

DR. CHESNEY: Were those levels drawn from the patients?

DR. DUNNE: Yes. These are the individual patient's data. So, we have a mean in red and in the individual patient in the yellow or the green to give you a sense of where they fit in and our conclusion from this particular slide here is that even if you vomit after getting the dose of azithromycin you still see to be absorbing it, and you might ask how could that be, you know, you gave them the drug and they are vomiting it up; how does that work?

The first thing to think about is that the amount of drug that is delivered either in this situation or actually in the pediatric trials is very small. It is only 2 teaspoons, 10 cc's. It is actually very hard to bring that back up. It actually lines the stomach wall.

Another thing that is important is that if it is treatment-related vomiting that tends to recur when the drug has reached the duodenum the distal parts of the duodenum. By the time a drug is there, the pyloric sphincter will have closed if you start to vomit.

In other words by the time a drug is getting out to the place where it is causing you to vomit you are already absorbing it and even if you do vomit there is very little actual substance in there to come back up. So, it all kind of fits together when we look at the PK data and the clinical outcome data for us.

DR. CHESNEY: Just one more question along these same lines. Do we have any pharmacokinetic data from children using the single-dose therapy?

DR. DUNNE: No, we don't have any PK data in children per se at the 30-milligram-per-kilo single dose.

DR. CHESNEY: Were any children redosed if they vomited within half an hour?

DR. DUNNE: Yes, there were. There were eight children that were redosed if they vomited within the 30-minute window.

DR. RELLER: Dr. Glode?

DR. GLODE: I had three short quick questions, I think. One would be in the information we were given, the briefing document there is reference to modified intention to treat analysis.

Could you just review what is modified in intention to treat?

DR. DUNNE: Yes. Actually let me put up the slide that shows kind of how we analyzed the patients? They are all kind of the same. So let us put up main slide No. 85, please?

Patients who were analyzed at day 28 had to receive a dose of drug. So, if you randomized and then left the clinic you would not be included.

You had to have a diagnosis of acute otitis media at baseline. I think just about everybody had that, but that was a rule and in this particular study you had to have shown up, actually in all of the studies you had to have shown up at the day 28 visit.

That means that the rule generally was that missing data was excluded from this particular analysis.

DR. GLODE: Thank you.

Then, and please excuse me. You presented so much data that I may have gotten this wrong, but it looked to me like the failure rate in the Costa Rican study was 6 percent for the day 28 overall failure rate and then in your subsequent studies that were done at multicenters in the United States they are pretty consistently in the range of sort of 25 to 26 percent per single dose?

DR. DUNNE: Yes.

DR. GLODE: What do you think is the explanation for that?

DR. DUNNE: It is hard to know. That is a single center. So, within that 25 percent of the overall for the other studies there could have been centers that were higher or lower around that. It is a single center study.

DR. GLODE: My third question is that you said, "And as expected the failure rate in all these studies is much higher for children less than 2 than greater than 2." Why do we expect that?

DR. DUNNE: It has been seen in all of the studies that have used whatever antibiotic in treatment of those children.

DR. GLODE: But what are the possible explanations or what has been excluded? For example, is the rate of recovery of pathogens, bacterial pathogens different? If you do a tympanocentesis on less than 2 year olds versus older than 2 year olds, are we treating more non-bacterial disease in those children? I mean just sort of biologic plausibility I need an explanation.

DR. DUNNE: I think that you, I will just speculate here just for a second. A respiratory tract infection, of course, can occur because of host defense problems in addition to a particular pathogen that has come and taken advantage of that host defense issue. I think the children under the age of 2 are likely to have more host defense issues. The angle of the eustachian tube is a little different. The antibodies they may have to the various bugs are at different levels than the older children. There are different developmental issues which may be playing a role in the outcome of children under 2 rather than over 2. So, it could be other epidemiologic variables which get folded in there as well, day care center attendance, for example, other things that could be important.

So, there is probably a collection of reasons.

DR. GLODE: And has anybody analyzed the tympanocentesis studies to see if the rate of recovery of a bacterial pathogen is different in the younger children than the older children? That would be of interest, and that information is available, I guess, just could be analyzed that way.

DR. DUNNE: Let me just look and see for a second if we have something along that line.

DR. RELLER: While Dr. Dunne is looking up this information, Dr. Marchant had a comment to make.

DR. MARCHANT: There is a study in the brochure that the Food and Drug Administration handed out by Carlin which shows that it is the exact data that you are asking for and it shows that the patients that, in patients whom you fail to eliminate the organism on average younger than those where the organism, sorry. Patients who eliminate the organism are older than patients who fail on average, the mean age.

So, young age is associated with failure to eliminate bacteria from the ear, and there is parallel clinical data in other trials that show the age effect on a clinical basis.

DR. DUNNE: I can just answer the one question. I think we will have to go back and look that up for you specifically. We had outcome data by bug. That is not the same question that you asked. So, we will have a look at that and see if we can bring it back after the break.

DR. RELLER: Dr. Wald had a question.

DR. WALD: Specifically though, Colin, I think that the question is is there a lesser frequency of recovery of bacterial pathogens in children less than 2 years of age, and I think the answer is no.

DR. MARCHANT: Right.

DR. WALD: I think that is the answer to your question that frequency of recovery of bacterial pathogens in children under 2 is not less.

I really enjoyed your presentation. I think you did a very nice job, but there are two issues that really I find perplexing. In every study of the epidemiology of acute otitis media children under 2 are the largest age incidence group. Maybe two-thirds to 70 percent of all children with acute otitis media are under 2. So, I am a little bit surprised that in every one of your studies that this more-difficult-to-treat age group is in fact under represented. That is sort of question No. 1, and question No. 2 is that in the two tympanocentesis studies that were done the recovery of bacterial pathogens was only 50 percent, and I think that is a little bit less than we expect and my suspicion is that when children are selected for tympanocentesis one is very stringent in applying criteria because you really want to recover a bacterial pathogen.

So, that makes me really worry about the studies in which there was no tympanocentesis.

DR. DUNNE: I will start with the first question if I can make sure I have got that clear. Why is there only say 40 percent of the patients in these studies include children under the age of 2 whereas you might expect 50 or 60 percent in the clinical population? Yes, it is an enrollment into studies issue, and it may be that some parents with children under the age of 2 are less likely to enroll their children in a trial compared to what you might see in the community. I can say specifically within the studies, again, enrollment was allowed under the age of 2 and the investigators were free to enroll anyone that they wanted to but these are the data that we kind of got.

On recovery of organisms, I think 50 percent recovery in these types of studies is probably not far off from what is seen typically; certainly in the studies that we have done it is not far from what has been presented.

I wouldn't doubt that in other settings you might be able to get a higher rate of recovery. Say in a clinical practice setting you know the patients and there is more follow-up but the 50 percent mark is about what we have seen in other studies that we have done.

DR. RELLER: Dr. Gorman and then Dr. O'Fallon.

DR. GORMAN; As a pediatrician it is nice to see some recognition of host factors. Knowing we stand in front of the Anti-Infective Committee the microbiology discussion was very elucidating to me, but there are a few other host factors involved, and I was glad to see that those were mentioned at least briefly.

You had a lot of data about failures and successes over and under 2. Was any of that failure or success related to the adverse event and discontinuation of the medication, either yours or the comparator agent?

DR. DUNNE: So, to repeat the question --

DR. GORMAN: The data showed that there was a higher failure rate in children under 2. Was that related to the adverse events and the discontinuation of medicine secondary to adverse events?

DR. DUNNE: For the patients who received azithromycin there was a 99 percent compliance rate. So, we wouldn't have seen failures due to discontinuation of therapy in the azithromycin group. We didn't analyze the data specifically to look at compliance with the amoxicillin/clavulanate arms, for example, to see if that correlated with failure or not. Typically you need fairly large sample sizes to get a sense of compliance giving you efficacy correlations. So, I don't think I can answer that fairly for you, but it wasn't an issue for the azithromycin-treated patients if they altered their drug.

DR. RELLER: Dr. O'Fallon?

DR. O'FALLON: You set me up. My question right from the "get-go" has been about how compliance was defined since compliance is obviously a major issue for this particular application. In those two double dummy type studies R-0581 and 1014 I read trying to figure out just exactly what it meant. They say, "The double dummy," but what we have here in one case is 3 days of active treatment and the other one 1 day of active treatment, and versus 10 days twice a day.

Now, how was that double dummy managed?

DR. DUNNE: You are right. You couldn't assess compliance during the course of the study because it was blinded. You could only do that after we had the data in house and we unblinded the data because the compliance measures that we used for these particular analyses looked at people taking their active study drug.

DR. O'FALLON: That is what I was afraid of.

DR. DUNNE: Yes.

DR. O'FALLON: Now, you are saying, let me just make it the worst case, you have got the one shot drug versus the twice a day for 10 days drug and now were the kids with the one shot getting a pill plus these two dummy pills for 10 days?

DR. DUNNE: Yes, the regimen would be that everybody got that first azithromycin dose either active or placebo right there in the clinic and they would then start that same time point their b.i.d. for 10-day dose.

So, after they left the clinic they were taking their b.i.d. doses.

DR. O'FALLON: Right, and half of them, the ones were getting, okay. So, now, when we talk about compliance, now what are you talking about if the kids on the one shot, the true active stopped taking their dose at 4 days, did they count as being compliant because they had their drug whereas the ones who were on augmentin had to have all 10 days or 8 days of it anyway, 9 days of it?

DR. DUNNE: Yes, that is an important point of the whole presentation. Let me see if I can clarify that for you. The idea of those presentations was to say, "How likely is it that someone will take their single dose versus how likely is it that someone will take 10 days of therapy?" That is kind of the question.

Now, there is a number of methods one can do to get at that. We chose the method of looking at compliance with your active study drug. So, basically the children who were assigned to augmentin retrospectively now we can look back and see who they were. We looked at the number of days that they took their therapy and if you were fully compliant you were fully compliant. Obviously if you were not you were less compliant.

Now, we could do other analyses of course. You could look at the people who were given azithromycin active and took placebo augmentin and see if they took their placebo compliantly. That is another approach. I suspect that because it was randomized you will see a similar outcome.

DR. O'FALLON: You didn't show us that.

DR. DUNNE: Would you like to see that?

DR. O'FALLON: Yes, you didn't show us that.

DR. DUNNE: We will do that for you.

DR. O'FALLON: Because by definition the one drug is going to have 99 or 100 percent compliance.

DR. DUNNE: Sure. You know, it is an interesting question. How do you study compliance? And how does otherwise somebody study that? It is easy to just kind of walk beyond that in our data sets and say, "Well, of course."

Let us look at whether there really is a difference, and I think we see that there is a difference.

DR. O'FALLON: And the parents, that was the other thing. I had two technical questions about what you asked, I mean what you were showing. I couldn't absorb it quickly enough. The adverse events per patient year, what is that?

DR. DUNNE: Okay, let us go into that. We have a few slides to help you with that.

I will flip through my book.

DR. O'FALLON: I didn't write down the number unfortunately.

DR. DUNNE: That is okay. We have it up here. Let us try looking at what exactly it is. First let us look at what I showed and then we will go back and see how did we get that.

Okay, so we will go to the main presentation and that is going to be Slide 99. Okay, good.

The backdrop to all of this is the top line here looks at patients that had an adverse event. So, if at any time after you randomize you had an adverse event you raise your hand, yes, but it doesn't get at the burden of side effects, how many did you have, how many days did you have them? So it is a fair analysis of the subjects with the adverse event. That is fair, but there may be more dimension to this when we are looking at the same total dose but delivered in different ways. So, we attempted to look at something that we actually don't normally do which is total burden of side effects in this particular program.

Okay, now, let us go to the safety slide No. 17 please?

So, this is a kind of a basic sense of what we are doing. This is a patient, for example, who had a vomiting adverse event on day 1 and day 2, nausea on day 1, 2 and 3 and a headache on day 3 and 4.

In that top line of the analysis they get a one. Yes, I had a side effect. We miss all the other burden in here by doing that. So, what we tried to do was basically add up the X's. That is the approach. You look at the number of adverse events they had and the number of days that they had that.

Now, in order to compare one arm to another we had different numbers of patients and we had to normalize it to something. So, we just picked patient years' exposure, but how that works is you have a 30-day observation period when you are in the studies.

So, we just normalize that 30; that is the same for 1, 3 and 5, and then you just normalize that out to a year. That helps us deal with the big N. Okay? Does that help enough?

DR. O'FALLON: Yes, I see the concept, and I think it is a good analysis, I would say. There are different ways to do it, but that is good.

DR. DUNNE: Okay, thank you.

DR. O'FALLON: And finally, one more thing. On Page 54, I guess I wrote the page.

DR. DUNNE: This is of the briefing document?

DR. O'FALLON: No, your presentation, No. 107 and this happened on the other one as well. You keep just saying, "Percent successful clinical outcome," with a confidence interval on there, but I don't know which way the difference is defined.

DR. DUNNE: Okay, yes.

DR. O'FALLON: I cannot tell from this one. Other ones you showed but these don't, and I don't know what that confidence interval is.

DR. DUNNE: I want to make sure I have the right slide for you.

DR. O'FALLON: I have 107 and 108, either one of them.

DR. DUNNE: Let us put slide 107, say, for example? Yes, now, this is very important actually. In the previous tonados(?) there we have the difference in clinical outcome. So, it is the confidence interval on the difference.

DR. O'FALLON: But defined which way, which drug is first and which one is second?

DR. DUNNE: Great. I will go back to that. Why don't we go back to that. Let us just go back to Slide 106 then.

DR. O'FALLON: That one it defines it.

DR. DUNNE: Yes. Okay, so, this is confidence interval on the difference. The right side we favor azithromycin. The left side we favor comparator.

Now, we come to the next slide, 107, and we are switching things on you now.

DR. O'FALLON: That is what I was wondering because this doesn't look good for you if you kept it the same way.

DR. DUNNE: Yes, what this is is the confidence interval on the point estimate of success. So, it is not a difference anymore.

DR. O'FALLON: Oh.

DR. DUNNE: It is just the point estimate of success and there is a certain number of observations in that point estimate, and that gives you a 95 percent confidence interval.

So, we just presented this for data for a quick reference about how this, how our drug would look compared to what else is out there.

DR. O'FALLON: But what is the 80 to 95 percent business? What does that mean?

DR. DUNNE: Okay, yes. Those are the upper and lower limits of the confidence interval on that point estimate. So, here is the point estimate here. It is about 88 percent.The lower limit is down here, 81 percent. The upper limit would be something like 95 percent. So, it is just for reference because otherwise there are lots of lines there. It is hard to find the azithromycin single dose regimen. So, we just labeled it out there for you.

DR. O'FALLON: Okay.

DR. DUNNE: Okay?

DR. O'FALLON: Yes, I will have to look at that.

Thank you.

DR. RELLER: We have a series of questions now. Dr.Christie and then Ebert and Glode and Leggett, and then we have a break, and we can come back if there be questions for Dr.Dunne in the discussion later, I am sure he will be happy to answer those.

So, we will have the four queries on the table and then our break.

Dr. Ebert?

DR. EBERT: My question pertains to the end of treatment assessment. Could you just review the difference between a clinical cure and a clinical improvement as far as the criteria?

DR. DUNNE: Yes. Just to repeat your question, the question is what is the difference between clinical cure and clinical improvement. The protocol gives a lot of leeway to the investigator to make decisions about improvement or cure.

Cure is complete resolution of signs and symptoms. So, they are all gone, but as I pointed out there could be a little sign of effusion left at any of those time points that would not preclude the investigator from calling it a cure.

Improvement is something shy of that. There is not quite resolution to feel comfortable that the child is cured. There may be a little bit of irritability left. There is some kind of sign or symptom which they are not happy is completely resolved, but it is better than baseline, and it is certainly not worse.

DR. RELLER: Dr. Christie?

DR. CHRISTIE-SAMUELS: Thank you. Do we know anything about whether or not the Hemophilus influenzae bacteria were typed and do we know anything about rate of HiB vaccination usage in Costa Rica as compared to here in the United States?

DR. DUNNE: Okay, so, two questions. Was the Hemophilus typed? Was it non-type O; was it type B? I assume that is the question, and the second was about vaccine, I am sorry, which vaccine?

DR.CHRISTIE-SAMUELS: Hemophilus influenza Type B vaccine. Is it used in Costa Rica?

DR. DUNNE: In 1995, I am actually not sure. We will have to go back and check that for you. I am not sure if that was -- that data was not collected as part of the program whether they had had Hemophilus influenzae vaccination or not but we can check with the investigator to see.

The other question was was the Hemophilus typed. No, we didn't do specific typing for Hemophilus as part of the program. Presumably the later cases, the ones in the more recent studies probably had a low incidence of Type B, but we didn't actually do typing on those Hemophilus.

DR. RELLER: Dr.Glode?

DR. GLODE: My question was just about that table 107. You have just explained now clinical cure versus clinical improvement, but what is clinical success?

DR. DUNNE: Clinical success is a combination of cure plus improved. Yes, that is a little regulatory thing.

DR. RELLER: And finally,Dr. Leggett?

DR. LEGGETT: I have a question regarding Hemophilus influenzae and this question of effusions and improved versus cure. Fifty percent of H. flu isolates more or less revert spontaneously to sterility and yet the statement is made on Page 4 of your briefing document and was made here that they are, quote, difficult to cure.

What does that mean?

DR. DUNNE: I think the implication of that statement was not so much that in the short term the organisms could not be reduced in burden or there couldn't be some immediate cure rate, but it is difficult to completely eradicate that organism and the infection due to that organism.

So, in other words the overall success rate as you go farther out to day 28 later seemed to be a little lower for Hemophilus patients than it is for Strep. pneumo patients for example.

So, ultimately as Dr. O'Rourke kind of told us it may be more difficult to ultimately clear away that infection.

DR. LEGGETT: I don't remember seeing in Dr. O'Rourke's presentation, and maybe you can give us the data, was the relapse rate or that incidence of new infections higher for H. flu than it is for Pneumococcus? Is that the proposed -- and then how do we then wrap that around the fact that Pneumococcus is much more likely to cause acute otitis media than H. flu? I am having trouble with those concepts.

DR. DUNNE: I am trying to think of what I can help you with from the data that we collected.

DR. LEGGETT: That is sort of a problem, isn't it?

DR. DUNNE: Yes. I am not sure that we have data within the program to get specifically at the reasons for why it may be more difficult to treat Hemophilus, why there might be a lower overall success rate within the program. I will tell you we can go back and think about that and bring it back to you.

DR. LEGGETT: My question is what caused, what was your definition clinically of failure? Was it that the ear had to be red? Was it just that there was an effusion? That is what I am getting at.

DR. DUNNE: I think we have a breakdown by who failed and who cured and what their symptoms were. Maybe that would help a little bit.

DR. RELLER: We can go at ten-forty-five for the FDA presentation and questions from the general audience we will take care of in conjunction with the public presentation.

Thank you.

(Brief recess.)

DR. RELLER: The FDA presentation will be by Dr. Moledina.

Dr.Moledina?

DR. MOLEDINA: Good morning. I am Dr. Nasim Moledina from Division of Anti-Infective Drug Products and as you must have figured out by now the discussion of topic today is single dose and 3-day treatment of azithromycin suspension in pediatric patients with acute otitis media.

With that as a background I would like to sort of summarize what is currently approved for acute otitis media as a 5-day regimen, not acute otitis media as a 5-day regimen in adults and in children for the indications for Zithromax and in adults for the 5-day dosing regimen it has been approved for acute bacterial inflammation of chronic bronchitis, pharyngitis/tonsillitis, community-acquired pneumonia and uncomplicated skin and skin structure.

In children azithromycin 5-day dosing regimen has been approved in acute otitis media, pharyngitis, tonsillitis and CAP and the indication reads that it is approved for acute otitis media caused by the three most common organisms, H. flu, M. catarrhalis and Strep. pneumoniae.

The dosing regimen is given as 30 milligram per kg total dose given as 10 milligrams per kg on day 1 and 5 milligrams per kg on days 2 through 5.

Basically I would like to give a background on how this drug was approved as a 5-day treatment and you have already heard from the sponsor that on day 30 for this 5-day approval. I would just like to point out the end of treatment data which was on day 11 for the three studies that were submitted in support of this, the original oral suspension.

The first study was study 134 and just to give you an idea, the success rate which now you know what success means, it means cure plus improvement and day 11 was the end of therapy evaluation point where azithromycin and the comparator both had a cure rate of 88 percent.

The 30-day data was already presented by the sponsor. So, I am not going to go through those. When you look at the second study which was a non-comparative clinical and bacterial study the success rate at the end of therapy was 84 percent for azithromycin.

This study, also, had data collected where there was baseline tympanocentesis done and this is for the end of therapy in this column and you look at the patients with Strep. pneumo isolated at baseline, 82 percent were cured compared to 80 percent in the H. flu group and 80 percent in the M. catarrhalis group.

There was a study, 128, which was, also, a comparative clinical and bacteriologic study. This was a study that was initially set up as a comparative study but when our guidance came out the sponsor stopped the study and put all the patients in a non-comparative trial. So, this did not have too many patients enrolled in the study.

For this study when you look at the end of therapy treatment azithromycin had an 88 percent cure rate. The basic data from the study because this was a small study when you look at the Strep. pneumo at the end of therapy treatment end point there were 86 percent of Strep. pneumo cured versus H. flu 82 percent and 7 out of 7 M. catarrhalis were cured.

So, this basically gives you an overview of how they approved the 5-day regimen for otitis media in children. Was, as you must have already found out from the presentation this morning that there is scarce PK data in pediatrics, but when you look at the comparison of exposure among azithromycin regimens, 5 day versus 3 day, basically the conclusion drawn from studies that were submitted to FDA is overall exposure associated with 3-day regimen is similar to the 5 day.

When you look at the 3-day regimen versus 1 day in adults all we can say is overall exposure associated with 3 day may be similar to 1 day, but we can't make any more conclusions than that.

In pediatric patients when you look at the 5 day versus the 3 day you can see that the 3 day may be similar to the 5-day regimen, and we do not have data from 3 day versus 1 day regimens in pediatrics. That basically gives you an overall of what we have already approved.

I will now go to the pivotal studies and go over my slides rather quickly because the numbers don't change. So, I will discuss the 1014 study which was the 3-day study first and then discuss the two 1-day therapy study and touch a little bit on the supportive study.

Study 1014 as you know was a clinical only study which was a double-blind multicenter randomized study comparing 10 milligrams daily dose of azithromycin given for 3 days with a 10 day course of augmentin.

Three hundred and seventy-three patients were enrolled for 28 US sites. The age group for entry was 6 months to 12 years and the mean age was 3.5 years.

I forgot to mention in the 5-day approval the cutoff age point for entry was 2 years and above. So, in that 5-day study none of the children were less than 2 years of age because that was the entry criteria. I forgot to mention that.

When you look at the MITT population and look at end points at end of therapy and test of cure, the success rates for azithromycin were 83 compared to 88 for the augmentin group and when you look at the test of cure with it which was on day 20 to 32 apparently the cure rates flipped and in the azithromycin group the outcome is better than what you see in the comparative group, and I wanted to bring this to your attention.

When you look at the outcomes broken by age I may sound like a broken record, but anybody who is less 2 years of age the outcome for azithromycin is 75 percent compared to 85 in the augmentin group and when you look at children more than 2 years of age the cure rates are a bit higher than what you find in kids less than 2 years of age.

When you look at the test of cure visit you see the same thing. Children who are less than 2 years of age had a lower cure rate than children who were more than 2 years of age.

The second study is a clinical only study, 0581 which is a double-blind, double-dummy, multicenter study comparing a single dose of azithromycin with a 10-day course of augmentin.

In this study 350 patients were enrolled from nine US sites. The age group for entry was 6 months to 12 years and the mean age was 2.7 years.

In this study, also, when you look at the end of therapy and the test of cure the success rate at the end of therapy for azithromycin was 87 percent compared to 88 in the augmentin. So, it was similar and both azithromycin and augmentin had similar performance as test of cure.

When you break the clinical outcomes by age groups, children who were less than 2 years of age had a lower success rate than children who were more than 2 years of age and that was at the end of therapy. This is at the test of cure visit, and you see a similar pattern.

Study 1015 was a single tap open level non-comparative study which had tympanocentesis performed at baseline and 248 patients were enrolled from 22 US and Latin American study sites. The age group for entry was similar to the previous two studies. The mean age in this study was 3.4 years.

When you look at the clinical outcome since this was a non-comparative study azithromycin success rate at end of therapy was 89 percent compared to cure rates at test of cure visit which was 85 percent.

When you break the clinical outcomes by age it is the same story. You see here patients who are less than 2 years of age had a lower outcome than patients who were more than 2 years of age, and this was at the end of therapy. When you look at test of cure you see similar numbers.

This study had tympanocentesis done only at baseline and there were 42 H. flu, 10 M. catarrhalis and 76 Strep. pneumo isolated.

When you look at the end of therapy assessment the success rate which is cure plus improvement 71 percent of H. flu had cured clinical outcome, 100 percent for M. catarrhalis and 92 percent for Strep. pneumo.

When you look at the clinical outcome at the test of cure visit 64 percent of H. flu were cured, 100 percent of M. catarrhalis and 88 percent of Strep. pneumo.

The sponsor gave you a detailed presentation about the resistance, but I am just going to mention that 16 percent of Strep. pneumo isolates were resistant to azithromycin and the success rate at the end of therapy and test of cure was 10 out of 12 and 8 out of 12 respectively.

When you break these two most common organisms down by age, when you look at the outcome at the end of therapy for H. flu patients who were less than 2 years of age the success rate at end of therapy was 61 percent compared to 79 percent for children more than 2 years of age.

In the Strep. pneumo group it did not matter. Kids less than 2 and more than 2 basically had the same outcome. When you look at the test of cure in H. flu patients 53 were cured who were less than 2 years of age, but 72 percent who were cured were more than 2 years of age and for Strep. pneumo the outcome was lower for kids who were less than 2 years of age than who were more than 2 years of age.

So, basically it follows the same pattern.

Study 95001 was what the sponsor calls a pivotal study and was submitted to the NDA as part of a supportive single-step study which was done in one center only which compared a single dose of 30 milligrams azithromycin to 3 days' therapy, 10 milligrams daily with IM ceftriaxone.

Clinical evaluations were made at end of therapy which was days 9 to 19 and follow-up on days 26 to 44. The mean age in that study was 2.5 years and it ranged from 0.3 to 6.5 years. So, this study enrolled younger patients.

When you look at the clinical outcome by baseline pathogen at end of therapy for the two common organisms, as you know the sponsor presented only two M. catarrhalis isolated in the study and both were in the 3-day group. I wanted to concentrate on H. flu and Strep. pneumo.

When you look at the end-of-therapy evaluation the success rate for 1-day therapy for H. flu is 89 compared to 93 percent for the 3 days and 100 percent for ceftriaxone.

When you look at Strep pneumo, 95 percent for 1-day therapy, 83 percent for 3 day and 100 percent for ceftriaxone.

So basically that sort of summarizes all the studies that were submitted for this particular approval for 1 day and 3 day AZ. Just to sort of summarize study R-058 was a single dose comparative study which was a clinical only study, The study 1040 was a 3-day study which was also comparative to augmentin and 1015 was a non-comparative single dose, and it sort of gives you an idea of the cure rates ranging from 75 to about 85 percent, and bacterial outcome which was in the pivotal study where you know the H. flu was like 44 patient who had isolated, 10 and about 76 Strep. pneumo the clinical outcome in those patients who had H. flu isolated at baseline at end of therapy was about 68 percent compared to like lower sixties. M. catarrhalis all 100 percent were cured and Strep. pneumo had a better outcome at end of therapy and at test of cure.

So, this basically summarizes all the efficacy data that was submitted by the sponsor and FDA had a chance to review the source data and verify all the outcomes.

Just to make the discussion more complete we did go and look at the literature and in your briefing package I had included articles by Degan and others where two double-tap studies were done and we particularly looked at this because both the studies used azithromycin. One study used 3-day azithromycin and the other study used 5-day azithromycin, and there were low bacteriological eradication rates for H. flu when tests were done on therapy.

As I mentioned there are limitations of evidence from these publications because FDA has really not had a chance to review the source data from these publications.

One other published study that I didn't put in your briefing package but I handed it out and it should be in your package was a study that was done by Doern and others in Antimicrobial Agents and Chemotherapy this year.

This study looked at all the clinical isolates of Strep. pneumo that were obtained from 1999 to 2000. There were about 1531. This was more like a surveillance study where it looked at all the isolates and susceptibility of Strep. pneumo to several antimicrobial agents, quinolones, cephalosporins, macrolides and there was a whole bunch of them. Since the topic of discussion is azithromycin today from the particular paper when you look at all the clinical isolates of Strep. pneumo 25.7 percent of them were resistant to azithromycin, but when you only took those isolates that were pain resistant which were like 329 isolates 77.2 were resistant to azithromycin.

Moving on to safety, all I need to do is point out a couple of ADRs that are of particular interest to us. As of June 30, 2000 cutoff point 2590 patients in all Phase II to IV studies, this is regardless of indications that they were entered and had received azithromycin and the sponsor presented adverse reactions which were treatment related.

I am going to show you a different picture in which all adverse reactions for all the otitis media studies, the numbers are the same, the denominators. These are the studies that the sponsor presented, also, but they presented the treatment related or what an investigator thought was, the vomiting was due to the treatment. I am looking at all adverse events since vomiting is really subjective.

So, when you look at all the patients with any sort of adverse reaction reported and in vomiting with yellow bile it is the 1 day. The turquoise is the 3 day and the red is the 5 day and this was taken from the original NDA and the comparator. In this study the comparator was ceftriaxone and augmentin.

When you look at this vomiting almost 10.5 percent of patients vomited on day 1 whereas 6 percent on day 2 and a little more than 3 percent on day 3.

When you look at this particular group of vomiting and break it down to the days where these children vomited about 37 of 52 of the patients who reported vomiting on a single dose vomited on day 1 whereas there is only about less than 1 percent vomited in the 3-day group and about half of the patients vomited who were receiving the 5 day and then of course it decreases as you go out with 3, 4 and 5.

This concludes my presentation. I did not really go into detail because we basically did not disagree with the sponsor's data, but I would like to make the following acknowledgements, Dr. John Alexander and Dr. Thomas Smith did the primary review with me and Dr. Makhene, Dr. Soreth, Goldberger, Dr. Gavrilovic, Dr. Ross, and the project manager, Jose Cintron and our statistician George Rochester and the team leader Daphne. Mr. Harold Silver is our microbiologist and Dr. Sheldon. Dr. Charles Bonapace is our biopharmacist and Dr. Pelsor the team leader for biopharmacy. Dr. Andrew Yu is our chemist and Dr. David Katague is team leader, our chemistry team leader.

All of them were good moral support for me and participated in putting this presentation together. Thank you very much.

DR. RELLER: Questions for Dr. Moledina?

Dr. Chesney?

DR. CHESNEY: I had two questions. The first one I just wanted to clarify. You said that the 5-day initial azithromycin studies in children were done only in children over 2 or over 5.

DR. MOLEDINA: Over 2.

DR. CHESNEY: Over 2, and my second question is we have no pharmacokinetic data at all for azithromycin in children?

DR. MOLEDINA: I think they have some data, but Dr. Bonapace can answer that question.

DR. BONAPACE: The data we have in children consists of 5-day and 3-day regimens. The 5-day regimen is 12 milligrams per kilo and sampled only on day 5. We, also, have a 10 milligrams per kilo for 3 days sample on day 3 and a 20 milligrams per kilo also sampled on day 3. So, we do not have any information from the 1-day regimen in pediatrics.

DR. CHESNEY: Were those all in children over 2 years of age?

DR. BONAPACE: Most of the children were over 2 years of age, and there may have been maybe one or more down around 1-1/2 years but most of the children are over the age of 2, yes.

DR. RELLER: Other questions for Dr. Moledina?

Dr. Wald?

DR. WALD: So, does that mean that azithromycin is only approved for use in acute otitis media in children over 2? Is that the current indication?

DR. MOLEDINA: No, the current indication reads 6 months and over.

DR. WALD: But there was never any data evaluated on children?

DR. MOLEDINA: I think there was some data not for otitis media, for other indications that children less than 2 years of age were enrolled and I think when the labeling was done those things were part of the whole package, and I looked at it. I was trying to go through the history and went through the whole 5-day package to see because the label is very clear but you know children less than 2 years of age were not enrolled in this particular indication, but when you look at the indication, you know, it is approved for anybody over 6 months of age, and I was not able to track down the history because the medical officer has left FDA. So, I was not able to find anything but maybe the sponsor can answer that.

DR. RELLER: Dr. Leggett, and Dr. Dunne, if you and your colleagues have any comments on the pharmacodynamics, what data are available, with azithromycin we would appreciate hearing that and Dr. Leggett?

DR. LEGGETT: Question for either you or the sponsors, what are the kinetic data on the single-dose azithro for non-gonococcal urethritis or STDs? Do we have that data?

DR. ALEXANDER: That information was not submitted as part of this NDA, and the dosing of that is, also, different as well.

DR. RELLER: Dr. Gorman?

DR. GORMAN: I would like to follow up on Dr. Wald's question. Have any of the data that has been generated for this presentation, is it a response to a written request under the FDAMA regulations and is this being used as an application for pediatric exclusivity?

DR. ALEXANDER; This is Dr. Alexander. I am actually the pediatric representative for our division, but no, this isn't involved because azithromycin is a pre-1997 antibiotic. So, it doesn't qualify for exclusivity.

DR. RELLER: Dr. Dunne, do you have any comments in response to the queries raised?

DR. DUNNE: We do have data as was discussed for PK in children at 60 milligrams per kilo total given over 3 days or over 5 days. That is the PK that we have in children.

There was a study that was submitted to the agency after the original application in otitis media that looked at children down to the age of 6 months. We don't have that data available on slides today, but that was submitted, and it was used to amend the label to get the age group down to 6 months of age.

DR. RELLER: Thank you.

We will next go to the open public hearing. First the ground rules for this segment. I would like the presenters to disclose their position and any financial involvements that would be pertinent to the data presented.

Secondly, those presenters at the open public hearing can receive questions from the Committee but cannot be queried by the FDA nor the sponsor directly.

The total time allotted, not necessarily to be taken, but allotted maximum is 1 hour. The first presentation is scheduled for 20 minutes by Dr.Ron Dagan who will be followed by Dr. Jacobs and by Dr.Wolfe.

We have had a request for flipping those presentations, and we will do that. I had ranked them by time, not by my listing. I apologize for that. Dr. Jacobs will be 15 minutes and will be first followed by Dr. Dagan and Dr. Wolfe, if he be present after Dr. Dagan's presentation.

Dr. Jacobs?

DR. JACOBS: Thank you, Mr. Chairman, Committee members, ladies and gentlemen. I am Michael Jacobs. I am a clinical microbiologist in the Department of Pathology, Case Western Reserve University in Cleveland and I am, also, Director of Clinical Microbiology at University Hospitals of Cleveland.

I would like to start as far as financial disclosure is concerned that I have received financial support from a large number of pharmaceutical companies, and I provided this listing to the Committee which is available for inspection.

I, also, wish to state that I am representing myself and no one else in my capacity as a scientist and a physician working on principles related to the treatment of otitis media and I have submitted full disclosure of my work related to otitis media in this as well.

What I am going to be looking at today are some of the principles involved in how you study this disease. Many of these have been covered and one of the points that we have been made aware of is that resistance in Pneumococcus and many other organisms has increased remarkably over the last 10 years and you can see here that treatment choices were a lot simpler pre-1990 than they are now, and we now have up to one-third or more of our strains of penicillin non-susceptible and up to three-quarters of the penicillin non-susceptible ones are macrolide resistant as well.

As far as macrolide resistance is concerned you have heard about 40 million doses of azithromycin being given since that drug was approved a few years ago and you can see what this has done to macrolide resistance rates. They were less than 10 percent or around about 10 percent in 1996, and they are now rounding at 25 percent and this has paralleled the usage of macrolides.

Also, just to give you a flavor of what has happened to susceptibility, I managed to get hold of some of the strains from otitis media that Dr. Halley had studied and Dr. Marchant had studied and you can see here that out of 50 isolates of pneumococci up to 1985 that amoxicillin, cefuroxime, cefproxil were very susceptible as were the macrolides with almost 100 percent susceptibility whereas the MICs have now gone up more than 100-fold for many of these drugs and you can see that now we only have between 50 and 90 percent of our strains resistant to this example of drugs.

If you look at Hemophilus we have seen the same issue. Again, I managed to get hold of 50 isolates. You can see there is no shortage of isolates from the current era and you can see again that we have lost a lot of activity, far more strands of beta-lactamase producing. So, we have lost a lot of activity of amoxicillin. Cefuoxine has gone down as well, ceftriaxone/azithromycin and azithromycin have never had much activity against Hemophilus and I will get into that later on as well.

Now, as far as acute otitis media studies in children are concerned I am a clinical microbiologist, not a clinician, but being a clinical microbiologist the role of antibacterials is to eradicate the cause of organisms from the site of infection and as clinicians, Dr. Dagan and many other clinicians who have been involved in this area tell me that that is their goal as well, and if we fail to do this irrespective of clinical response we really are not doing our patients a service.

Now, what are limitations and the problems with outpatient clinical studies and respiratory tract infections? There is a high rate of spontaneous resolution in these diseases that makes it difficult to show clinical differences between agents. These are very different diseases to diseases such as bacteremia and meningitis as examples.

The second point is if there is such a high rate of spontaneous resolution you really need to do bacteriologic outcome studies and these are not often performed due to necessity for invasive procedures such as ear, sinus and lung taps to obtain specimens.

More studies are therefore designed to show equivalent clinical outcome between established and new agents or in other words they are not statistically powered to show differences between agents and therefore if there are inadequacies of agents studied they are often not apparent.

Now, again, you have seen this, and again, I have simplified this going only up to 2000 patients in the study. The Pollyanna phenomenon published by Dr. Marchant and his colleagues and you can see here that when you are doing bacteriologic diagnosis and outcome that unless you have two very good drugs you need very few patients.

When you are doing bacteriologic diagnosis with clinical outcome you need larger numbers of patients and you need the largest number of patients when you are doing purely clinical studies and you can see here that even if you are testing placebo versus a very good drug you need over 500 patients in a clinical study to make this point.

Also, to make this point this database is done predominantly on children under the age of 2 years of age.

Now, let us look at azithromycin and acute otitis media. There are four studies using a common comparator that I could find. The study designs differed. Two of them were clinical diagnosis and outcome. One was bacteriologic diagnosis, clinical outcome. One was bacteriologic diagnosis and outcome.

Patients ages in these studies differed considerably, and there has been discussion on this this morning. The first three studies involved patients from 6 months to 15 years old. Means were 4 to 6 years, while the fourth study involved patients only up to 4 years of age and the mean was 16 months, 1.3 years.

Sample sizes required for these studies to be powered to show differences between agents were determined based on calculations published by Marchant and this is what I came up with.

This is a summary of these four studies. You can see the mean age and the age range is given here. The mean age was not given in the study but was presumed to be similar to these two studies given here, and you can see the differences that I pointed out.

The number were variable. You can see there were fairly large numbers in these two studies. There are very small numbers in the study done here and 143 in this study.

Study designs were these two were purely clinical studies. This was bacteriologic diagnosis, clinical outcome. This was bacteriologic diagnosis and outcome.

P value for clinical outcome was statistically significant only for the study with the lower age group patients. Number of patients needed to show 60 versus 90 percent bacteriologic efficacy is 2000 patients in these two studies. So, you can see that these studies were way under powered to show any difference.

So, it is not surprising that these clinical outcomes were not different. In the bacteriologic diagnosis clinical outcome 800 patients were needed, but you can see the P value was starting to approach significance even with 92 patients, and on this study which was the one done requiring the least number of patients bacteriologic diagnosis and outcome, there was a statistically significant difference in clinical outcome at end of treatment and also in bacteriologic outcome during therapy, and you can see here the 143 patients met the criteria and that at least 100 patients would be needed to show a bacteriologic difference, sorry, a clinical difference based on bacteriologic diagnosis and outcome, and you can see here that if you had placebo you would need even fewer patients and in fact you can see the study sizes of these two studies were hardly able to differentiate any drug from placebo.

Now, the next issue that I am going to address is how can we overcome some of these limitations because it is very difficult to do studies that are powered to show these differences and one of the best approaches that I have been made aware of is to use pharmacokinetic or pharmacodynamic parameters to predict efficacy, and these are parameters that are very familiar to us, pharmacokinetics, serum concentration profile, penetration to the site of infection.

Pharmacodynamics is a relatively new field. It takes the potency of the drug in which we tend to spend a lot of time and looks at the actual way the bugs and drug interact in vivo, whether this is concentration or time dependent and whether the drug has post-antibiotic effects and basically agents can be divided up into two groups, those that are time dependent where the time above MSC of the drug in serum is predictable outcome and those that are concentration dependent where either area under the curve to MIC ratio based on area under the curve in serum or area under the curve to MIC ratio will peak to MIC ratio are predictive of outcome and some examples of this and most of this work is taken from the work of people working in this field like Dr.Craig who was Dr. Reller's predecessor on this Committee and you can see here that in animal model this is a mouse model with Strep. pneumo infection you can see that mortality was reduced to zero once the percent or the dosing interval exceeded 40 percent of the dosing interval in serum for penicillins and 50 percent for cephalosporins.

There is human data which correlates with this very well. These are most of the double tap otitis media studies looking at bacteriologic outcome and you can see very clearly that to get greater than 80 percent bacteriologic eradication as seen with time dependent agents when the concentration of the drug is present for greater than 40 percent of the dosing interval and you can see that this works over a large number of studies, a large number of investigators and also applies to penicillin non-susceptible pneumococci, and if you correlate this with Hemophilus with a spontaneous resolution rate of about 50 percent you can see many drugs have been shown to be no better than placebo and with pneumococcus one study has come up with values that are very close to placebo.

Now, I am, also, showing you this data which you have seen before. This is the chinchilla model of otitis media with Hemophilus, to make a few points about this.

Firstly, when you look at this study here this shows azithromycin being given for 5 days and probably about double the dose equivalent that would be given to humans, but for the low dose which is the 30-milligrams per kilogram per day and the high dose it is clearly way higher than any dosing regimen that has been tested in humans, but you can see that when you follow the natural history of this out that this is no therapy, 100 percent of cultures positive on day 0, 100 percent on day 5 and in the chinchilla you get spontaneous resolution between day 5 and day 10, again very similar to that in humans, about 63 percent.

When you look at the lower dose, but again this is relatively high compared to what we are using in children, you can see that again you can get no change at day 5 but you are getting a significant decrease between day 5 and day 10 which is higher than that of placebo, and as you increase the dose even further you are getting 50 percent eradication at day 5 and almost 80 percent eradication by day 10.

So, there are some differences in this model between what happens in humans as far as natural resolution is concerned and what happens in this model.

When you look at this kinetically again you see the same thing and one of the questions I have about this model is when you take the drug and mix it with the organism in the lab the test tube you get killing very readily, and you saw some examples of that by the sponsor that by day 3, certainly within 24 hours in most cases you get a 3 to 5 log kill with azithromycin in Hemophilus, yet we are seeing absolutely no effect for 3 days, and I have a lot of questions as to why that is happening and what the mechanism is, and I don't know the answer to that.

Now, let us look at some other animal models. This is data that has recently been published in Antimicrobial Agents and Chemotherapy by Muttman et al. This is looking at azithromycin lung and plasma levels.

One of the points made is that drug delivery is very different. This is done intracellularly by macrophages and polymorphonuclear lymphocytes and the tissue level are more important than serum levels, and you can see here that if tissue levels are indeed important then tissue levels would be adequate to treat Hemophilus whereas serum levels would not be, but if you actually look at the experimental data that you get you don't find that is the case.

Looking clarithromycin you get exactly the same parameters. You can see here tissue levels would be even less or slightly more difficult to be able to be effective against Hemophilus and certainly serum levels would not be effective.

When you look at the experiments that these investigators did what they found in fact was that macrolide susceptible pneumococci with both azithromycin and clarithromycin indeed could be treated very effectively based on dosing comparable to that given in humans and they used a whole dosing range here and they got a very nice straight line relationship.

They tested other agents which followed this line which I have removed for clarity, but when they started looking at macrolide resistant pneumococci and Hemophilus they found they had to go to much higher doses than those comparable to given in humans, and the dosing limit for azithromycin and clarithromycin for human dosage is approximately 25 milligrams per kilogram per day. In this model you can see that therefore this model was very effective in treating macrolide-susceptible pneumococci but failed to treatment macrolide-resistant pneumococci with efflux mechanism not shown on the slide because they were off scale up here with ribosomal methylase which has very much higher MICs and, also was not effective against Hemophilus influenzae showing that again as had been demonstrated with many other organisms and drugs that it is the serum concentration of drug that is driving this not tissue concentration.

So, my conclusion from the study was that dosing comparable to dosing in humans azithromycin and clarithromycin were able to reduce inocula by greater than 3 logs in the lungs for macrolide susceptible pneumococci but were not able to do this for Hemophilus influenzae or for macrolide non-susceptible pneumococci with either resistance mechanism.

Let us look at the kinetics of azithromycin. This is a conventional dose, 10 milligrams day 1, 5 milligrams day 2 to 5. Area under the curve is 3 milligram hours per liter formula to get a break point for area under the curve divided by 25 gives you a break point of .1. This is again fine for macrolide susceptible pneumococci, fine for Moraxella catarrhalis but with an MIC 90 of macrolide resistant pneumococci of 8 is not very good for efflux strains or even worse for ribosomal methylase strains and also with an MIC 90 of 1 to 2 for Hemophilus is way above the break point for Hemophilus.

Let us look at what is out there as far as break points is concerned. Pharmacodynamic break points based on pharmacodynamic principles are given over here and these are the ones that have been shown to correlate in animal models and in human data to what actually happens in vivo and you can see for the Pneumococcus the break points that we currently have approved by the National Committee for Clinical Lab Standards are pretty close to the pharmacodynamic break points. However, you can see there are some examples of NCCLS break points for Hemophilus which are higher than that of the pharmacodynamic break points and sometimes in some instances by up to 60-fold and you can see the examples I have given here of clarithromycin, azithromycin, loracarbef and cefaclor and cefprozil. Those stand out as being way higher than the pneumococcal break points or the pharmacodynamic break points, and one of the points about break points is for organisms causing the same infection there is no medical or scientific reason for a break point being higher for one organism than another.

Now, let us look at what happens when you look at susceptibility of organisms at different break points and I am going to go over this quickly in the interests of time, but you can see that when you look at pharmacodynamic break points you get a very different picture with those examples of the break points that I showed you before when you look at NCCLS break points, and you can see in some instances there is almost 100 percent difference.

Is Hemophilus susceptible to azithromycin? According to the pharmacodynamic principles it is not but microbiologically it is regarded as susceptible.

What I have done to summarize this data just to bring out the highlights, these are the drugs where greater than or equal to 90 percent of recent US trends are susceptible at both break points, and you can see here that amoxicillin, amoxicillin/clavulanate and clindamycin actually came in at 89 percent, but I included it anyway are the only three drugs that I could find that fulfilled this 90 percent criterion.

When you look at Hemophilus drugs that do this by both NCCLS and pharmacodynamic break points are amoxicillin/clavulanate, cefdinir, cefixime and cefpodoxime. Those drugs that do this only by NCCLS parameters are cefuorxime, cefprozil which comes in slightly under 90 percent, loracarbef and azithromycin.

Now, let us look a little bit about the specifics of this. As you have seen pneumococci are divided up into three groups, and I think NSC 90 study that was presented earlier by the sponsor showing an MIC 90 of 2 micrograms per ml for Strep. pneumo is not representative of what goes on in the US. That study included US and Canadian data and there the MIC 90 was 2. More studies in the US have MICs in the 8 to 64 range.

With Hemophilus MICs are in the .5 to 4 range, unimodal distribution. You do find occasional strains with Hemophilus with much higher MICs, but what we have learned about Hemophilus in the last few years is that the reason they are much higher baseline than they are for Strep. pneumo is that they naturally have an efflux pump, and you can inhibit this pump and bring MICs down to about .06 micrograms per ml.

Unfortunately the agents we have at the moment that do this are all highly toxic. So, what you find with the Hemophilus is that the MIC 90 is 2 micrograms per ml. That is pretty universally accepted.

With Strep. pneumo it is now 32 micrograms per ml. How does this compare to break points? Pharmacodynamic break point comes out on current dosage to .1 microgram per ml. When you look at pharmacodynamic data in some of these more recent animal models this comes out to a break point of roughly .5 micrograms per ml, and if you look at the dosing where they used doses equivalent to four times current human dosage they can bring this break point up to 4 micrograms per ml which will now include Hemophilus.

So, you certainly can get Hemophilus to be susceptible to macrolides in vivo, but you have to increase the doses enormously.

So, in conclusion antibacterial choices for empiric use in acute otitis media most clinical studies are too small to show clinical differences between agents. Pharmacodynamic parameters correlate with bacteriologic and clinical outcome in animal models and in humans and can be used to select agents with maximum potential for bacterial eradication.

Currently available agents vary significantly in achieving these pharmacodynamic parameters necessary for eradication and few oral agents approved for pediatric use are active against 90 percent or more of current US strains of the three key otitis media pathogens at approved dosage regimens, and finally, bacteriologic outcome studies in children and animal studies have repeatedly validated these conclusions.

Thank you for your attention.

DR. RELLER: Questions for Dr. Jacobs from the Committee members?

Dr. Patterson?

DR. PATTERSON: Thank you for your nice presentation. Dr.Dagan's study that you cited, is that the one where H. flu was the most common isolate which seems to be a little different than the epidemiology we have heard today and wasn't the statistically significant difference in the H. flu infection?

DR. JACOBS: That is correct. The major difference in outcome there was based on H. flu. Dr. Dagan can probably answer this question better than I can, but I think there were a couple of points there. One is the age group that that study was done in was very different from the other bacteriologic diagnosis studies you looked at and many people believe that under the age of 2 Hemophilus is probably more common, and you had a second question, the statistical significance. Because of the MIC distributions we had very few resistant pneumococci there. So, with pneumococci the difference did not reach statistical significance but with Hemophilus with a unimodal population we had enough and that did reach statistical significance.

DR. RELLER: Thank you, Dr.Jacobs.

Dr.Dagan?

DR. DAGAN: I thank the Committee for allowing me to say several things about methodology. I just want to tell you I am not from the United States. I didn't not know that I had to bring with me a declaration of with whom I was contracted but I did send it to the Committee, and you have it on record, and I reconstructed this from my memory.

So, if I forgot something you have that in your files. I have been receiving grants and/or consulting with Abbott, Aventis, and Aventis-Pasteur, GSK, Wyeth, Lederle Vaccines and Lilly Antibiotics, Pfizer, BMS, Lilly Chemopharmaceuticals, Astrozenica(?) and Merck, Sharp and Dohme. I may have forgotten something, but at least I think it shows you that I have quite a wide spectrum of working with various companies.

I thought that first I should answer the question regarding H. flu because I think Michael Jacobs did not answer that correctly. I am sorry, Mike. The study that you have, it was published in PIBJ in 2000. It is a multicenter study including many centers in the United States. When you analyze the data from the States separately H. flu was more common than Pneumococcus.

So, either it is changing with time or it is time and age. In France by the way it is the same thing. We have more Hemophilus. It is not unique only for one study.

I thought that in presentations of FDA and other places a lot of data we probably saw so far 220 or 240 sites and there are lots of outcomes and a lot of data and it is very easy to get confused, and I have been confused for many, many years, and I think I am less confused now after working very hard in the last 5 years on otitis, and I would like to actually change this talk into a second one and try to really show this point which is I think very important to understand when we read results of studies or when we design the studies. These studies might be healthier to your needs.

If you know from preliminary studies the characteristics of your drug you can actually start and design your studies to show whatever you want to show and I think you had a little bit already in the talks and discussion before. I want to highlight that more to make it a little bit more clear maybe.

Now, first there is much outcome to choose, and I am not going to again tell you too much about the Pollyanna. We all know now about the Pollyanna effect issue, but I just want to show that this, if you take this outcome and again I will show you data from our experience where you really see a big, big difference between placebo and the best drug and this is something that can be easily demonstrated and ask a lot of questions, and I want to show you this is as you already know much more accurate because the next big sample size and here you almost can show almost everything, and each one I will show you actually how to play with it in order to get what you really need.

This is a study that we published years ago looking at the cefuorixime-axetil versus cefaclor pharmacokinetics susceptible to penicillin. There was no problem. The two drugs are very, very efficient and both are effective but cefaclor is really much closer to placebo now than to cefuroxime-axetil and with Hemophilus, also, cefaclor is very much close to placebo compared to the other drug. So, here you can show with relatively small numbers of children as you can see, you can show your point in a very highly statistical significance.

This is another study that was not sponsored by the industry, but it was important to look at trimetapolis(?) sulfa because it is still the cheapest drug and we wanted to see what happens if you have susceptible or resistant and here unlike beta-lactams you have most of the time either highly susceptible or highly resistant and so when we get to children with Pneumococcus or with Hemophilus that were susceptible we have 100 percent success rate and in this case when we give, when they were resistant we had placebo effect in terms of bacteriological eradication very clearly showing that for true resistant both Hemophilus and Pneumococcus this drug is the same.

When we look at Hemophilus influenzae now and we look at amoxicillin, there are three studies looking at it and here are the results with double tympanocentesis and with beta-lactamase negative Hemophilus you can see that in these three studies although they don't get exactly similar results they are clearly superior than placebo while for all three studies the numbers are small but very clearly you get placebo effect, again showing that if you have a drug that is susceptible to beta-lactamase then beta-lactamase producing organisms may fit a placebo. Again, very clear results. It is very difficult to argue with those.

You remember the point 12 cutoff that Michael Jacobs presented from the theoretical model. This is a study that we did sponsored by Pfizer a few years ago and published a few years ago, and looking at the bacteriological eradication and actually the only organism that could have an MIC less than .12 was Pneumococcus that was susceptible to macrolide and again we have 100 percent success rate here while those who were not were quite highly resistant and all of them as I said in this case. In a second study that we did we had close results to that.

For Hemophilus the lowest MIC we could get is .5 and as you remember the MIC 50 is one. Ninety is two. So, basically what we got is a success rate in all MICs very close to placebo. So, the only one that it was really shown to be good for is Pneumococcus that is susceptible to macrolides.

So, I think that those examples that demonstrate how you can go with relatively small numbers and get something which is significant and bacteriological proof and to the best of my knowledge otitis media is an infectious disease and the approach should be as for other infectious diseases. Go to the bacteria and demonstrate the eradication by antibiotics.

Now, which end point to choose? You already heard today a lot of discussion and let me just go back to this classical study by Kaleida that we have already mentioned. Now, which end point to choose? You already heard today a lot of discussion and let me just go back to this classical study by Kaleida that was already mentioned before and here these are children over 2 because you didn't use placebo under 2, but what is shown very clearly that at end of treatment she still had very good results with amoxicillin compared to placebo and this is looking at clinical outcome.

However, 2 weeks after the treatment there was no difference at all between placebo and amoxi and at 6 weeks actually placebo was significantly better in term of effusion. This is something which may be strange. It has some biological plausibility and I don't have time to speculate on that. Maybe if there are questions later on, but the point is very clear that if you choose this placebo is superior to amoxicillin.

Now, the whole issue of test of cure, I was surprised to hear that in this very, very important country they call this 1-month test of cure because they have never proved that this is a test of cure. Actually we have enough proof that it is not a test of cure.

If we look, this is a study that was mentioned already by Colin Marchant. If we look, and this is not the only study but this is the most recent one with the highest numbers. One hundred and eight cases that had double tympanocentesis studies. They had first tap with a bug. Second tap bud was eradicated. Children after treatment recur clinically. We go for third tap and now we can compare to the original organism. What we found was that of the 108, 20 percent were no growth. So, they had some viral infection or something or some technical issues. In 56 percent it was a new infection. New infection means either Hemophilus was replacing Pneumococcus or vice versa or another bug, but that Pneumococcus was a different Pneumococcus was proved by serotype or pulse field or Hemophilus proved by pulse field or beta lactamase.

So, basically what we called test of cure is test of relapse or recurrence or bacteriological relapse because in only one-quarter of what we see of clinical relapse it is actually something that comes back and again, as discussed by Colin Marchant before it could be actually that the bug was eradicated but not from the nasopharynx and so far nasopharyngeal eradication or dynamics is not part of what the FDA wants to get for licensure although that might be very important. It might be in the future tested for that, too.

Then we had the Pneumococcus that recurred as Pneumococcus. This is 38 cases, 58 percent, almost two-thirds. It was not the same Pneumococcus and Hemophilus when recurred was Hemophilus in 60 percent almost. It was not the same Hemophilus. So, even when the same bug was there it was not actually the same bug.

So, the test of cure is for me not a real test of cure and maybe that should not be used. Who are the patient's contacts? This is very important, not who are the patients. Who are the patient's contacts? If you really look the patients get the bugs from the contacts. If they have bad contacts, they have bad bugs, and this is a classical study that is hopefully going to be very soon published by Robert Cohen from Paris. He took only this experience of 5 days with this 10-day regimen. You know there is a big issue whether to treat 5 days or 10 days and took kids that are at home not going to day care centers and kids that go to day care centers. Otherwise these kids were comparable, and what you see first this is at end of treatment there were much more failures in those kids compared to those who are at home, but if you look at the 5 versus 10 days here you see a significant difference. Here the difference is not significant.

So, if you have a drug that doesn't do very well but you have still to show 5 days go for this, never go for this one-tailed. So, this is a test of cure which I call a test of relapse, okay? Again, much more relapses in day care centers. We all know that, but we never use that to read the articles, which were the kids from. Not only this, here it shows clearly that these kids who go to day care center 5 days is not appropriate, while here, no difference. So, for kids who stay at home, fine. I never saw that in an article saying that for kids who stay at home you can give 5 days in our study, but if they go to day care center be careful.

Kids who went to day care centers carry much more resistant pneumococci. This was found in so many studies and this is again from that study by Cohen. Not even that, kids who stay at home they may have brothers who go to day care center. Does this make a difference? Definitely. Pneumococcus twice as much carry Pneumococcus if their brother or sister go to day care center so with Hemophilus and Moraxella and even more than twice as much with any bug and this is highly statistically significant.

So, again, you don't get otitis if you don't have the bug in the nasopharynx. So, the idea is take the kids who don't go to day care center and who have no brothers and sisters and almost everything will work.

Age effect, we have heard about, but I want to just bring you something from the literature, very good studies. Some of them were used for licensure. Amoxicillin by Kaleida less than 2 years of age, 12 percent failures with amoxicillin versus only 4 percent above 2 years of age.

Hoberman, amox/clav published I think it was even used for the licensure of this drug in the previous dose and again this is not to compare from one study to the other because there were different outcomes, but within the same study 49 percent unsatisfactory. Look how much you go down with older kids.

Penicillin in Sweden after 2 years almost nothing and cefprozil the same after 1 year almost no problem. So, again, if I have a drug that doesn't show too much of an effect I will take the older kids and make it smaller.

Now, effect of severity, we had that. This is the Kaleida study non-severe disease, placebo, Amoxicillin .5 percent with unsatisfactory results at the end of treatment versus 6 percent, still a difference but not very impressive. You need a huge sample size for that while here you have a very impressive difference.

Again, if I have a drug that doesn't work very well I will go to those type of kids and actually if I go to the clinics and get kids over 2, up to 12 years of age I don't know how many of the pediatricians here treated how many kids, of course. It is very difficult to find them, and if you find them they usually have a little bit of ear pain, a little bit of red ear and you go to the clinic and you need to recruit a lot of kids because you get money per kid. So, you take all this was very mild, and this is in my belief the answer to that question that was asked before about why do we have so many older kids while almost all of our patients are younger kids.

Effects for this factor on acute otitis media, and I want to take all this together. I start to understand now how I can make things according to what I have. I just want to again last time come to this curve again to remind you of one thing. The worse the drug is the better the clinical efficacy is compared to the bacteriological efficacy. So, if I get a very bad drug then it looks much better on clinical, but if the drug is 100 percent I can almost never get really better than 90 percent because of those viral infections.

So, for me the maximum predicted for clinical success in bacteriological cases is about 90 percent. Above that you may be lucky to get a little bit more.

So, now, if I take three drugs or three arms in a study and this will be my last three slides, placebo, antibiotic A that is very, very weak and antibiotic B that is very, very strong I can predict something like that. You know that placebo is about, and I take kids under 2 to start with because these are the kids where you really can do double tympanocentesis. It is not really surprising that when you go for double tympanocentesis you don't go to older kids. You go to younger kids because they have the pus and they need those things and the parents or children will never submit themselves to double tympanocentesis when they have a little bit of pain in the ear with a little bit of fraydrum(?). So, when I take the younger kids I will have here 30 percent placebo, 50 percent efficacy in antibiotic A and we saw already this is somewhere in the middle not very good drug and 90 percent which is one of the top drugs.

We have calculated from Colin's studies that were done together with Colin Marchant and from our own studies that were published before that if you see a bacteriological failure on day 4 to 5 about 40 percent only of those will be clinical failures. The others will still slowly go on and make better at the end of treatment. So, by using this calculation you can see that just for those if you go for clinical results you get 76 percent for placebo, 84 for the antibiotic A and 90 or better for antibiotic B.

Remember this 16 percent or 15 percent is exactly what you find in all the real good studies looking at placebo versus drug and this is what you find in almost all the studies with weak drugs in terms of success rate. They have around 76 percent success rate overall in cases where you don't expect the drug to work very well. This is placebo effect. This is the best effect. So, the effects are not very big. As mentioned before you have 24 million cases in the United Otitis. Ten percent is 2 million cases. So, it is a big effect in general, but not in a study. So, to show the effect here you need as we showed before thousands of cases.

Now, let us assume that we took only the culture positive but 50 percent culture negative as we heard before, this is not appropriate in a study usually but 30 percent this is what you usually find.

So, if you take the clinical cases you have to also assume that 30 percent will be culture negative. Culture negative in my experience are always doing better much faster than anybody else, no matter if you give antibiotics or not. So, they actually are not bacterial where you miss the bug most of the time. They are not bacterial, and so if you include those 30 you already get to over 80 percent here, over 85 percent here and here again you cannot improve anyhow. This is the maximum.

Now, remember this is all under 24 months. Now, if you take kids over 24 months and here i don't put 60 percent as you had before. I just put 30 percent above 2 and of those only one-third above 3. So, it is still younger than most of the studies that have been used by many drugs. I get here over 85 percent. You remember this age effect. Here I get already equivalence antibiotic A and B, but if I really want to show equivalence also as Colin said with tap water then I take those that are totally clinical,just looking at effusion and symptoms, and we all know that effusion and symptoms is most of the time otitis media, but sometimes it is otitis media with effusion just with a viral infection. How can you make really, there is an overlap. The overlap in this case with excellent investigators they only make an overlap in about 20 percent, not more than that. In other places I would say not, but in the United States there are excellent investigators. So, only 10 to 20 percent and here you have equivalence of one in three.

Okay, so, you can already play with this. So, if I summarize this, and in addition to that you heard that I can even make it more influenced by all those cases, avoiding or including those cases that are difficult to treat according to which drug I have.

So, now let us say that I am now a director of a new pharmaceutical company and I have either that drug or this drug. Unfortunately most of the weaker drugs are, also, convenient. Most of the stronger drugs in PKPD are not. I always cite my grandmother who says that a drug must taste like a drug. Otherwise it is not a drug. I don't know or they are injected which is another thing that my grandmother would like, injections, but the point is if I have a drug here and we often have a drug with true bacteriological activity, efficacy as predicted by the PKPD in animal models and others but convenient and well tolerated versus another drug and on the other hand a drug that has bacteriological efficacy which is good by PKPD, etc., but is less convenient and less well tolerated, and I will give you a table how to construct studies to show to sell this one or I am not saying sell in terms of money, sell the idea to whoever needs to approve it or sell that drug.

I call it hand tailored AOM drug studies and this is the weak and well tolerated and the strong and less well tolerated, and what are the characteristics? Study population of course. I will try to have much more of the above 24. I try to take the younger ones that have real problem in spontaneous eradication. DCC, try to avoid DCC children here. Try to take as many as I can here. By using older kids you, also, avoid DCC because 6, 7, 8, 9 year olds don't go to day care centers. Other siblings at home? Try to take single kids, you know, so go to clinics where you have high society type of people. Don't go to the inner cities. They are going to fail. Recurrent non-responsive, avoid those. Put them in the exclusion criteria but actually here try to enrich your population because those are the kids that are difficult to respond. So, enrich them in this, avoid them here. Baseline clinical diagnosis, never go to bacteriological here. Here clinical, we have proven bacterial etiology and tympanocentesis. Otherwise if you go here for clinical then you miss a lot of these things I said before.

Outcomes, tolerability puts much emphasis here because this is a drug that is well tolerated, of course. Try not to talk too much about this because it is not well tolerated. Convenience, the same. Bacteriological efficacy, don't include in your end points. Don't include that. Make it fit your main outcome here, okay, and make clinical outcome, test of cure. It is wonderful. They all go, Colin, your graph shows going down in parallel. I don't think you are right. They converge. They don't go in parallel. They converge at the end. There is almost no difference after 1 month eventually between those who responded well or not and so by going to the end of treatment I can show that everything is like everything, but here I am going to see the maximum difference and for conclusion I will be based on equivalence of clinical outcome and superiority of tolerability, convenience and here will be based on bacteriological and clinical efficacy in difficult-to-treat cases.

Unfortunately, I am not American. I don't deal much with the FDA but my impression as an outsider being foreign and as you know we don't understand a lot of things in the Middle East until you find the situation similar to us and then we think alike.

I think that this is the FDA approach most of the time. So, if I have many drugs that may not be responding to this very difficult criteria, furthermore if I do this, people that are going eventually to take those meta analysis stuff and put them on a slide are going to show my end points inferior to their end points and they are going to say that there are some limitations to it, but I used what I have. When there is some limitation in using what you have you choose, also, what you want to choose. So, the point is here if you use this you are going to have lower end points in your drug and they are going to compare less favorably with this. So, I think that now I give you enough tools to plan your studies according to what you have in hand and my conclusion is with most of the acute otitis media with clinical outcome as currently conducted are virtually guaranteed to show no significant differences between agents, dosing or duration of treatment.

Thank you.

DR. RELLER: Thank you, Dr.Dagan.

Questions from the Committee for Dr. Dagan?

DR. DAGAN: Maybe my English was not clear enough.

DR. RELLER: Dr. O'Fallon?

DR. O'FALLON: As a statistician I just thought that was magnificent. We are always looking for where the sources of variability are. This is beautiful. Thank you very much.

DR. RELLER: Thank you, Dr. Degan.

Is Dr.Sidney Wolfe present?

If not, we will assume that Public Citizen's will not be making a comment at this meeting.

Before closing for lunch there are no direct questions that are permissible. Comments from either FDA or from Pfizer are possible.

Does anyone wish to make an additional comment now?

Dr. Alexander?

DR. ALEXANDER: If I could I just wanted to address a couple of questions that were asked earlier. One of them is related to the Costa Rican site and Hemophilus influenzae B and what we found looking at the bacteriology from the microbiologist was that there were patients who were identified within the trial but only a very small proportion that did have Hemophilus influenzae type B, so that in the three arms there were 5 of 14 patients who were treated with the 3 day who had Hemophilus influenzae type B at baseline. Only 1 of 10 who was treated with the single dose had Hemophilus influenzae type B and 2 out of 9 who were treated with ceftriaxone.

DR. RELLER: Thank you. It is twelve-fifteen. We are exactly according to our agenda. We will reconvene at one-fifteen for discussion of the questions and vote by the Advisory Committee members.

Thank you.

(Thereupon, at 12:15 p.m., a recess was taken until 1:15 p.m., the same day.)

AFTERNOON SESSION 1:15 PM

DR. RELLER: Before beginning our discussion and vote on the issue at hand Tom Perez has an addendum to the meeting statement.

MR. PEREZ: Good afternoon. I stated this morning that Dr. Steve Ebert, Dr. Ellen Wald and Dr. Mary Glode would be excluded from participating in the general matters discussion of clinical trials of acute otitis media. I would like to correct the record and state that in accordance with 18 USC 208(b)(3) general matters waivers have been granted to all participants requiring a waiver. Therefore, Drs. Ebert, Wald and Glode will be able to participate fully in this discussion.

Thank you.

DR. RELLER: Voting today will be the eight sitting members of the Advisory Committee, in addition the consultants, both Drs. Glode and Gorman will be voting members for today's meeting.

Dr. Soreth?

DR. SORETH: Dr. Reller, I wanted to give the charge to the Committee for our product-specific questions for azithromycin in the treatment of otitis media.

The first question, do the data support the safety and efficacy of a single dose, 30 milligrams per kilogram and/or of the 3-day regimen of azithromycin for the treatment of acute otitis media?

We would ask the Committee that in their discussion they comment on the influence and the significance of the following and how they contributed to the Committee's recommendations.

The interpretation of the data from clinical only studies, the interpretation of the data with regard to studies that have microbiology with the tympanocentesis at baseline on single tap, the natural history of the infection and also information from the published literature and how that influences any specific recommendation. Should the Committee recommend approval we ask further if there would be any caveats in the label based on the results of the studies with tympanocentesis at baseline. If the Committee does not recommend approval what additional study or studies and finally, if approval of the single-dose regimen is recommended we ask the Committee for their advice with regard to what we would advise prescribers and patients concerning increased vomiting associated with the use of the single-dose regimen.

DR. RELLER: Thank you. Tom Perez will be recording our votes yes or no. Based on past experience in the discussions I think it would be helpful if we heard from every Committee member the points of discussion that you wish to make regardless of in the end how your vote is recorded so that the importance of the various components that Dr. Soreth has asked us to comment on we now open it up for discussion.

Dr. Chesney?

DR. CHESNEY: I have a real concern about our lack of pharmacokinetic data. In terms of giving this drug only once or for 3 days I don't know No. 1 if it is taken up into hepatocytes. No. 2, I don't know how very young infants will handle this and No. 3, one of my concerns is that children with viral syndromes not infrequently have elevated liver function tests, and I am not sure if these high doses are taken up into hepatocytes how that could influence the course of the viral syndrome particularly since it may be retained for a prolonged half life.

So, I am very anxious about not having pharmacokinetic data particularly in younger children and I don't know if anybody can reassure me on that score.

DR. RELLER: Other comments from the Committee?

DR. LEGGETT: Jim Leggett. Several thoughts occur to me on a purely scientific basis. PKPD has not been shown to work either for the single day or the 3 day. There is some data with the single dose in other uses leading to increased harboring of resistant organisms. That worries me a little, and in terms of AUC to MIC while that may be the parameter of most importance for azithro, that is, also, a little bit experimentally biased by the way and the timing that those parameters are done in mice and everyone else, and I worry a little bit about a single dose study finally running out of its post-antibiotic effects before we get out to the longer time period at least in terms of having efficacious levels, especially in view of the 5-day versus 10-day French data of Dr.Cohen that was presented.

So, in other words if 10 days is better in these sickest patients than 5 days then what does the single dose really do for us in terms of actually treating the patients?

I don't know anything about regulations in terms of whether we should approve the 1 day or the 3 day because a lot of the steps that are in the guidance document appear to have been addressed.

However, I am very much worried about the lack of pharmacokinetics in both the 1 day and the little less than 3 day.

I think that in terms of interpretation of data from clinical only I think that end is where we are beginning to see, at least possibly see today leads to sort of a bio-creep that people are worried about so that we keep comparing ourselves to less and less and less and finally we get down to no better than placebo.

I am, also, a little bit worried about the clinical only, of the difficulty at least I as a non-pediatrician have in making the difference between cured, improved and failed on looking at an ear or in an infant. I think tympanocentesis at baseline to me is a step better. The natural history of the infection looking at the degree of recurrences means that waiting until the test of cure does not seem like a very good idea, and information from the published literature I think of good quality is very useful in terms of placing these things in perspective if we only have active control comparators. It gives us a little bit of an external grounding at least to some degree.

In additional studies I would recommend further kinetic studies. Vomiting with a single dose, I think it has to be stated. I am a little bit worried again with kinetics that those values, the AUCs that were shown in adults for the single dose 2 grams and 3 grams while the yellow and green triangles were within the error margin the AUC in those things said, "Only 0 to 24." With a half life of 68 hours we are still accumulating. I don't know what is going to happen, again to go back to my point, they vomited, it looks fine up to 24 hours, but what happens between day 5 and day 10 when that AUC may not still be within those errors?

DR. RELLER: Thank you very much, Dr. Leggett.

On the points that Dr. Leggett addressed his perspective others who wish to comment on any or all of those and how they influence your thinking on the vote that we will take?

Dr. Ebert?

DR. EBERT: Just one expansion. Dr. Leggett mentioned the issues about AUC and much of that information was developed in animals and that total daily dose whether it was given in single or divided doses showed equal efficacy and again, I share his concerns for two reasons. One is that that data was based on microbiologic response, not clinical response, in other words number of organisms eradicated in the animal models. I have not seen a similar corollary to that in a human infection model if you will and then secondly, again, the concerns about how far can one really extend the dosing interval and still see adequate bactericidal effect.

DR. RELLER: Dr. Chesney?

DR. CHESNEY: I, also, have a real concern about the vomiting in the first day because particularly for infants that comes at a time that they often have high fevers and are already vomiting and not taking anything orally and perhaps dehydrated, and to give them a very high dose of a drug that is conjugated in the liver and to add vomiting on top of that may necessitate hospitalization. So, that is another area that I have some anxiety about.

DR. RELLER: Dr. Wald?

DR. WALD: I think we have a tremendous dilemma here this afternoon because we heard some excellent presentations this morning about all of the parameters that influence outcome in acute otitis media and I think that they really highlight some substantial weaknesses in the studies that are presented today, especially the clinical only studies and especially because the age group that was focused on was older children rather than younger children, and I think that the microbiologic data are weak.

I will restate again a 50 percent retrieval on patients in whom one is very anxious to make a correct diagnosis is very low. Most tympanocentesis studies have recoveries of between 70 and 80 percent for positive microbiology.

So, if a positive microbiology is seen in 50 percent of those patients who I presume are highly selected because you want them to be positive, then I think we really have to worry that the number of children who actually had acute otitis media in the remaining clinical only studies was really substantially lower than 50 percent and so I think these are sort of the facts.

On the other hand these are not standards that we have held other products to, and so, I think that the real focus is do we have a new standard that we use this afternoon or do we create a new standard which I hope will be the subject of our discussions later this afternoon and what do we do with this product at this time? So, I think it is a tough dilemma.

DR. RELLER: Dr. O'Fallon?

DR. O'FALLON: I have concerns, also about the analysis. I mean we didn't talk about the fact that what was called a modified intent to treat is not intent to treat by any definition I have ever seen before. The people who didn't show up for the day 28 or whatever the definitive test of cure whenever that was, the ones who didn't show up were simply left out. They were not involved in the analysis.

Now, that creates, missing data always creates a real problem. If they were, indeed, failures for whatever reason, those success rates, those cure rates that we were seeing are even lower than the numbers that we were given.

In some of the studies those were very substantially missing including ones that were used for labeling; the studies for the 5 day that were used for labeling had very substantial missing data.

I am concerned that maybe what we know ain't so. This really is a worrisome thing. So, I think that the actual success rates are lower than we have been told in these presentations.

DR. RELLER: Dr.Christie?

DR. CHRISTIE-SAMUELS: I have several concerns and some of them were already outlined. However, one includes the representational children are the age 2 years where there is a higher likelihood of failure and I am not so sure that the role of Hemophilus influenzae type B has been clearly outlined.

We heard some data from Costa Rica. I realize it is true that here in the United States the patients are likely to be immunized. We are actually using a very, very young age group and since they are using a drug where you may have resistance to HiB I am sort of worry about that. I am worried that the recovery rate for bacterial pathogens was less than 50 percent and I am worried, I am concerned that the PKPD data was not available, especially for young children and the data that was presented I am not very clear on what was presented but the data that was presented for those who vomited as far as I can understand that data was for adults not necessarily children. So that to me has not been clarified.

The Pollyanna syndrome I wonder how much of that is being shown in the data that was presented here. One of our presenters showed us that efficacy needs to be more than 90 percent if we are going to say that it is significant, and I think the ethical dilemma for us on the Committee is I think the sponsor was given a guidance document which I think they followed, but based on the presentations, the three presentations that we heard from Dr. Marchant and the two public presenters the question in my mind is should we include, since we are now changing the standard should we, as a Committee look at the total picture based on what the guidance, what guidance they were given alone or should we, also, include what we heard today, and I think that needs to be clear to us because if indeed we are to include what the three speakers said, well, then I think I might, well, we will see how the afternoon passes.

DR. RELLER: Dr. Christie has pointed out some of the dilemmas that the Committee faces. One thing I would like to clarify is that it seems to me important that we all understand that whatever vote the Committee takes is not an issue I don't believe of whether the sponsor followed the existing guidelines or not because we must assess the safety and efficacy based on the study requirements that the FDA heretofore has provided as guidance for industry.

At the same time following those guidelines the Committee is asked in their view based on the data presented from the sponsor and from the FDA whether the information available is sufficient to make them comfortable if you will about the safety and efficacy of either a product or an indication or a dosing regimen of an existing product and we will get there eventually.

Dr. Chesney?

DR. CHESNEY: Another issue for me and I would like to reiterate what Ellen and Celia and Dr. Reller have said in terms of how do we make this decision and I think Dr. Reller has clarified some of that, but one of the other concerns for me is that this is not clearly a better drug than what we have. It has a prolonged half life which means that the concentrations are going to decrease and decrease and decrease with time which we have heard over and over again is the way that you develop drug resistance and we already have high levels of resistance to the macrolides and are we only going to be contributing to the problem by approving a drug which is not clearly better than what we already have out there with a number of unknowns of which I think I pointed out that the absence of pharmacokinetic data is another piece to the puzzle.

DR. RELLER: We have heard much discussion from the right side of the, my right side of the table and I am delighted that Dr. Patterson has her hand up from the left side, my left side.

DR. PATTERSON: Okay, I just wanted to make a comment regarding the resistance issue. Based on what we do know about the pharmacokinetics of azithromycin and granted we don't know about the single dose in children, but there is data to suggest that the efficacy would be based on the area under the curve over MIC ratio with this drug as compared to even the other macrolides and certainly the beta-lactam antibiotics, so that with a single dose you know it looked like certainly compliance was better but also it might have, you know, less effect on selection of antibiotic resistance, specifically macrolide resistance than the 5-day regimen if you were just giving it as a single dose, and I wanted to make a comment, too, about the Pollyanna phenomenon which I think is a real one but I think that it is not unique to acute otitis media.

Having been involved in some of the vancomycin resistant enterococci studies when before we had options for therapy we found that all but the or many of the patients except those with very severe infections had some spontaneous resolution, so while I think it is a phenomenon I don't think it is unique to otitis media, and the other issue about the resistance is that I found it interesting that with the strains that have the efflux mechanism which is the most common type of resistance in the United States that six of eight of those patients responded anyway even though the MIC was eight.

DR. RELLER: Dr. O'Fallon and then Dr. Wald.

DR. O'FALLON: One of the things that is making it difficult for the Committee is the fact that we are back to the business of having point estimates. Six out of eight is 75 percent. The trouble is that the confidence interval, any confidence interval you want to choose is going to be so broad with only eight people that it is going to be consistent with response rates running from about 35 percent to practically 100 percent.

So, I worry. I am going to come back and say again, I think it is very important that we see confidence intervals and I want to commend the company for presenting that nice thing that I asked about because I didn't understand it the first time I saw it, but that is the way I think that we should be seeing response rates for any of the ways you define response.

We should be seeing how they line up with everything else and it will give us a feeling for how big the study was and how big the population was.

The problem for me with this whole thing is I don't know how to put this politely; so it is just going to be bald. The guidance as written now rewards sloppy research. If you use a clinical end point without doing any kind of a tap that is the -- and then you throw away the people that just didn't show up at the 28 day you can really manipulate your success rates way up there, and then if you don't take into account the major prognostic factors of the subsets of people that have different responses innately in the disease, if you don't take into account those things, you can make your drug or your agent look good, and I am afraid right now I think the guidance is wearing a black hat in this whole issue today. It was before I came in here and I am even more convinced of it now.

DR. RELLER: Dr. Wald?

DR. WALD: I think that Dr. Patterson really raises another important generic issue for our consideration. In the year 2001, when we heard data that 25 to 35 to 40 percent of S. pneumoniae are resistant, shouldn't it be our expectation that that number of S. pneumoniae isolates in the clinical study will be resistant? Are NC microbiologic data addressing that particularly? I know we had a lengthy discussion about high-dose augmentin around the issue of resistance, but resistance is now an everyday phenomenon. It is not a special phenomenon. It is what we are treating when we treat acute otitis media so that we really have to have an adequate complement of those cases, I think in order to make a judgment, and I have to say that Joan has really highlighted the issue of toxicity and hepatic metabolism.

Why do all these studies start at 6 months? Surely the drug is going to be used in children under 6 months of age. Is that our requirement? Why do folks start at 6 months? Does anybody know?

DR. ALEXANDER: This is Dr.Alexander. I am actually doing a presentation a little later that sort of addresses that a little bit, and from what I was able to research in looking at our previous guidance and the points to consider document and such, they were looking at children that were greater than 6 months of age because of concerns about actually being able to accurately diagnose for the purposes of clinical trials the acute otitis media in those children that were less than 6 months of age, that they had concerns due to, I think some related to the organisms that were identified from children that were less than 6 months of age and some actually related to the physical examination, being able to accurately diagnose infection in that population.

DR. WARD: Just the last issue that I would mention is the one about duration of therapy, and I think that there is a controversy about 5 days versus 10 days and again, do we have any -- and I think the data support the notion that short course therapy, 5-day therapy is disadvantageous in children under 2, and we would expect that because they are more difficult to treat. So, is there any reasonable expectation that the single dose of azithromycin will have a level beyond the fifth day of treatment? Is that known? If a 30-milligram-per-kilo dose is given on day 1 is there anything on day 6?

DR. RELLER: Dr. Gorman?

DR. GORMAN; I would like to echo Dr. Chesney's comments about toxicity. Known tentatively to be intracellularly concentrated it presents some concerns about the young infant in terms of this dosing, and the resulting need for PK and PD data, also, in children less than 6 months or let me say over the entire range in which this will likely be used in the ambulatory setting which would be from birth to the end of the pediatric age range will be likely.

I look at the Pollyanna effect as the fact that you are using this chemical that you call an antibiotic inside a host that has a huge number of other resources that it brings to bear or gets in the way of taking care of this infection, and one of the things that was not clear to me when I read the article, while the Pollyanna effect will make a bad drug look fairly good, it, also, prevents a good drug that is not being used from looking bad.

So, if you take the best known drug, and I will use an old example of an oral antibiotic that was unpalatable that people would take and then throw up and then the patients would not take it again, you would still get the same results.

Things would look good on this data presentation because of the Pollyanna effect which would not even let the best drug when not used effectively, when not used compliantly still look pretty darn good.

I cannot underestimate for my parents, however, how important compliance and tolerability are in the dosing of these agents, and the concept of trucking around from work to day care to home different antibiotic agents becomes an incredible financial burden for these people, and if there was any data that just got them back into day care faster it would be a huge clamor for this medicine much like there is for the injectable medicine in our community, injectable alternative.

Host factors you cannot control for but in any new guidance might be considered might make the studies impossible to do, but you need to start looking at those things as either subclasses statistically as you do your analysis or excluding them from the enrolles so you know who is in the population.

Since these are short studies in terms of each individual's participation you should be able to figure out who is in day care and who is not, who had siblings and who did not.

Lastly on the guidance I wear another hat. I sit as an IRB member on two different IRBs and we would not approve the tympanocentesis study for otitis media under any circumstances. We feel it is not the standard of care. We feel it puts children at risk for a disease which has in 2001 basically no sequelae.

DR. RELLER: This is a great opportunity to hear comments from the Committee after we hear from Drs. Alexander and Soreth about the importance of the single tap studies as regards the data presented in evaluation of the product on the table.

Dr. Alexander?

DR. ALEXANDER: This was actually just a question as a point of clarification for Dr.Gorman's statement. Were you saying that any study with tympanocentesis would not be approved or were you referring specifically to the double-tap studies?

DR. GORMAN: It would be very difficult in the university or community IRB that I sit on to have a, in fact, we have turned down single-tap studies because they appear to offer no scientific benefit to that child, and they pose a risk. So, it adds risk with no benefit, and I have had the opportunity to talk to the chair of an IRB in another large institution who, also, claims that he routinely rejects all tympanocentesis studies.

Now, clearly some IRBs are accepting them because they are still being done.

DR. RELLER: Dr. Chesney, and then we will come to Dr.Maxwell.

DR. CHESNEY: I just wanted to comment on your comments. I could argue very strongly that a tympanocentesis study offered distinct advantages to the child because you know what the organism is. You know what the susceptibilities are and you can narrow your spectrum based on that.

So, just a comment, not immediately relative to the issues at hand.

DR. GORMAN; Our IRB would disagree. It offers advantages to science. It offers not advantage to that child because you are not going to change the drug inside a clinical trial unless you then set up the trial to do that which we have not seen yet.

DR. RELLER: Dr. Maxwell and Dr. Wald.

DR. MAXWELL: I just had an additional question. To my question as to how many children were redosed who vomited in the first half hour the number was eight, but I really wonder how many children vomited 1 hour, 1 hour and 1/2 and 2 hours after and then were redosed, and I don't know if anyone here has that answer.

DR. RELLER: Dr. Dunne may have or will have, and we will come back to him when I see a hand.

Dr. Wald?

DR. WALD: Maybe not to belabor the tympanocentesis discussion, I think there can definitely be benefits to the patient. There is immediate pain relief even though there is some experience of pain at the time of the procedure. I watched one just yesterday afternoon, and then we did have a sample, and we could if we need to make a precise judgment about non-response to that therapy we will have information in our hands which will be invaluable.

So, I think that one can justify it in those terms and some children are, of course, experiencing quite a lot of pain and for those children it probably is a bigger benefit.

DR. RELLER: Dr. Marchant?

DR. MARCHANT: To make a comment in response to Dr. Gorman that adds on top of what Dr. Wald just said about tympanocentesis, the problem is if we don't do tympanocentesis we won't know what is going on at all in drug approval and that means that it will sort of bring to life the Pollyanna phenomenon that poor drugs with poor efficacy get licensed and we don't know that they don't work. We don't figure out that they don't work and out there in clinical practice there are going to be patients that remain symptomatic because we have approved and then as practitioners prescribed drugs that are less efficacious and so to abandon tympanocentesis is to basically put ourselves in the dark in terms of the performance of antimicrobial agents for this disease.

DR. RELLER: Dr. Gorman?

DR. GORMAN: I am sorry I brought my IRB into this. First of all, I don't own this IRB. I just sit on it, and while I don't disagree with the scientific arguments here, you are now in the position of having your own data used against you that has been developed by typanocentesis.

Your predictive value when there is bacteriological cure is excellent in terms of clinical cure. Your predictive value when there is not bacteriological cure is a coin flip or perhaps a little bit better than a coin flip.

So, one would have to argue at least in the data presented in one of the articles that if you didn't have a bacterial cure, a clinical cure was still a 50/50 outcome or roughly a 50/50 outcome, and I understand the other clinical parameters that come in here.

The second thing that your tympanocentesis has shown me to sort of mute what Dr. Wald and you have just said is that often when you retap these people they have different organisms. So, how exactly are you going to then claim you need to retap them each time they have a reoccurrence because I am going to then be in the position I have data that shows me from previous taps that clinical failure doesn't necessarily mean the same organism is present.

DR. MARCHANT: Clinical failure is during therapy when you switch from one organism to the other. Yes, they do occur, but they are way the minority of failures. It is a very small number you are talking about.

DR. RELLER: Thank you, Dr. Marchant.

Dr. Glode?

DR. GLODE: I have really two separate issues to mention. One would be that abstract I gave you. Would this be an appropriate time about unanticipated adverse effects or not?

DR. RELLER: Sure. We want to get all of the discussion out on the table before the vote. Please?

DR. GLODE: Okay. I just wanted to mention this briefly because I don't think it is relevant to the vote today, but it is relevant to the issue of large studies and large study subjects and perhaps unanticipated adverse events, and that is that this abstract that was published in the Infectious Disease Society of America, and I just happened to be at the poster because I happened to be at the meeting whenever that was a week ago, and I happened to know I was coming to this meeting, and it was about azithromycin. So, I was paying more attention than I might otherwise, but it is really interesting for both, I think the FDA Advisory Committee, the FDA and the sponsors, and it is just an article looking at azithro or an abstract exploring in an animal model azithromycin as an immune modulator and in the animal model when you give azithromycin and you follow with a conjugated pneumococcal vaccine you have a very statistically significant effect on the immune response to that vaccine.

That, of course, is very important to me if it had any application to human beings, and we don't know that it does, but someone should look at that and make sure that there is not an immunomodulatory effect in children because I actually personally am more interested in preventive invasive pneumococcal disease than anything else to do with otitis media.

So, I really, you know, would just want that to be explored and obviously this is only in an animal model. So, it may not have any relevance to humans, but is nonetheless interesting scientifically, and so, I just wanted to bring that to people's attention so that it is something people look at perhaps in the future studies.

Then my comment about the information that we were presented with is I keep sort of focusing on the less than 2 year old even though that was not the majority of patients enrolled in these studies but may be the majority of patients with otitis from what we have heard about, and if I look now at day 28 failures with single-dose azithromycin it looks like it is in the range of 24 percent, with 3-day 33 percent, with ceftriaxone 30 percent, with augmentin 49 percent. So, again, I mean I really don't like any of those clinical failure rates in the less than 2 year olds. That is sort of with all comers, and then if you look then specifically for Hemophilus influenzae in the less than 2 year old where it was in the sponsor's study, a 47 percent clinical failure rate at day 28, and I was trying to put that together with the Dagan published study where they concluded that azithromycin was inferior, but in fact they are very similar because that is 42 percent for Hemophilus influenzae clinical failure rate in their study which had a mean age of 16 months. So, it probably fits. I mean probably the data is in fact not discrepant; it is similar. They just reached slightly different conclusions, but it helps me that the data is, in fact, sort of concordant.

So, I have to say that compared to ceftriaxone or augmentin it looks like it is in the same ballpark even though it is a high, if you will clinical failure rate overall, but I guess that is just what we have in less than 2 year olds with otitis media.

DR. RELLER: Dr. Dunne, do you have some additional information for our consideration?

DR. DUNNE: Yes, I wanted to come back up and tell you about the children who vomited within the first 2 hours and how they did with redose.

May I have the first slide?

So, these are patients who received a single dose of azithromycin who vomited on day 1 after this first dose. This population is all the patients that received that dose, 487 children and those who vomited are listed here in yellow. We broke it down by the study from which those children came.

You can see with 30 minutes there were eight. They all occurred on study 1015 and all eight of those were redosed. There were 4 and 10 children four below the age of 2 and 10 above the age of 2 who vomited within 2 hours.

This includes this group, of course. That is why the fours match up here. So, six additional children over the age of 2 who vomited in that 30-to-120-minute interval.

We do have children who vomited, and we don't have the exact time of that vomiting. We know that they were not redosed except for this one child.

Next slide?

Now, how did they do? Okay, we said that there were eight children who vomited within the first 30 minutes. We had outcome data on seven of those eight children under the age of 2. All three of those children were clinically cured at end of therapy and again at the test of cure visit. Of those greater than 2, three out of four were cured. At test of cure one out of four were cured.

The next slide is what happened between 30 minutes and 2 hours. This would have been children that vomited in that interval but did not get redosed. There were no children less than 2 in that category, four out of six both at end of therapy and test of cure were cured.

There was one other point that was made about the what do we do with the missings, and typically you might, well, one way you might handle the missings is by calling all of them failures and seeing what happened to your comparison.

If you could put up that slide, please?

This is the only study that had a significant number of missing and by significant it is in 8 to 10 percent something range. It was R-0581. The other ones had two or three people missing. So, it doesn't really impact much, and this would be an analysis where we set all the missings to failure on each arm, and you see the success rate at day 14 is 80 percent and 82 percent on azithromycin and amox/clav. The lower limit of the confidence bound is minus 11.

At day 28 success rates are 66 and 67. Lower limit is minus 12 on the difference, similar rates for cure. Lower limit is minus 8. These are within a point of the lower limit from what we saw before when we had just excluded the missing. So, we didn't see anything really different. We did sensitivity analysis by setting the missings to failure in all the studies, but as you can imagine it doesn't really change anything very much.

DR. O'FALLON: But what bothered me when I was reading this and doing this was the fact that the Pollyanna effect under those circumstances, you know, for the mix, particular mix of patients, tap water produced 71 percent at test of cure.

So, here we have numbers that are well, I mean percentages that are well under 71 percent.

DR. DUNNE: Yes, I mean I think we don't as you could imagine, the studies that we do are comparative studies to approved agents. So, the issue about comparisons to placebo, that would be a different kind of forum where that would be probably addressed. I think the best that we can do to help people get comfortable that at least azithromycin is as effective as amoxicillin/clavulanate when you start to get the numbers up was on that kind of meta analysis thing.

If we can just show the slide, pick either one, the day 14 main presentation tornado chart there? I will just come back to that here.

That would be Slide 115 in the main presentation. Again, we won't be making comparisons to placebo. We would make comparisons to what we think is an active drug which would be at least amoxicillin plus a beta-lactamase inhibitor and over the course of looking at 4000 patients we see point estimates of difference that are in the range of 1 percent.

So, if people are comfortable that at least amoxicillin is something better than placebo after looking at 4000 patients with these clinical end points; these are clinical end points; this is not the bacteriologic end point, it is at an earlier time point. Once you start adding up all the patients and all the trials which I would say that you probably don't get a chance to see this data very often because it is not often that you see a drug coming here that has already been out on the market for 5 or 6 years and has a lot of experience; so, this is an opportunity to see a lot of patients in a comparison like this and it looks to us like the azithromycin children did as well as the amoxicillin/clavulanate within very narrow confidence bounds.

DR. O'FALLON: When you showed that this morning I looked at it and I saw the azithromycin versus amox/clav as being almost significantly in favor of the comparator. That second one up from the bottom, it almost excludes, yes, the confidence interval almost excludes zero. So, it looks to me like that is even saying that it is worse.

DR. DUNNE: The confidence interval would say that it is within the bounds. So, it is hard to say that it is worse, but if it is different it is less than 1/2 percent.

DR. RELLER: Dr. Cross?

DR. CROSS; Dr. Chesney brought up the issue of giving a single large dose which is accumulated or conjugated in liver. Do you have any data on liver function in the single high dose?

DR. DUNNE: Yes, in study 95-001 where we looked at a single dose of azithromycin, the 3-day dose and ceftriaxone we did collect laboratory exams during that study.

Let me see if I can pull that slide up?

Slide S21, please?

Again, there were 66 children in this study at baseline. We have laboratory exams on most of those children, most but a few. Looking at neutrophil counts ANCs less than 1000 there didn't seem to be any difference between any of these regiments. Total bilirubin, the same pretty much across all. AST and creatinines again, no abnormalities seen in that particular group, and this would be consistent with the sum of experience we have had now with all the patients that have been treated with a 5-day dose.

The regimen 1, 3 or 5 will accumulate to the same degree. The AUCs are very similar in each of those regimens. So, the same total amount of drug will pass through the liver and as people are probably aware a small amount of azithro is metabolized in the liver. It is not a drug that is really active to the P450 system. So, in fact, it doesn't really get metabolized much there.

The experience with 5 day, looking at total dose should be similar to the single dose, and we do have a small patient population here to give you a sense of labs.

DR. CROSS: This is entry or --

DR. DUNNE: This would have been at the day 7 to 8 visit. If anyone had an abnormality later though; they came back with another problem, that would have been picked up, but this would be in the earlier, yes, that is my understanding.

DR. RELLER: Dr. Chesney?

DR. CHESNEY: How old are the children from whom you got that data?

DR. DUNNE: The mean age of the children in that study was about 2-1/2 years with a range fro 1 to about 4.

DR. RELLER: Dr. Ebert?

DR. EBERT: My question is again directed towards the vomiting that occurred in some of these children who received a large single dose. Certainly in adults who are given excessive doses of macrolides one has the potential for at least a reversible degree of ototoxicity that can occur and I am wondering you look at all in any of these children at any impact of the drug on vestibular function or hearing? Could that have contributed to the vomiting or was it thought to be a gastrointestinal phenomenon?

DR. DUNNE: Okay, so the question speaks to the association of macrolides and ototoxicity, and are we in that range, for example, in these kind of RTI doses.

About 3 to 4 percent of adults that receive doses in the AIDS program; so, this was much higher doses, for example, 600 milligrams a day for life essentially, about 4 or 5 percent of those people would develop some kind of a reversible hearing impairment, but the earliest that we really saw any kind of hearing impairment occur in that population, the earliest of that 5 percent was about 10 grams and the median amount was about 40 grams. So, it is well beyond what we see here, and that is consistent with what we know from the 5-day program where we really haven't seen any ototoxicity occur.

We didn't do hearing tests during this program. That is an interesting thing maybe for efficacy end points, actually in otitis media studies, but we didn't do that in any of these programs here.

DR. RELLER: Dr. Patterson?

DR. PATTERSON: I was going to ask you about the vomiting. Was that just a single episode? Were there any instances of persistent vomiting and, also, the patients who were redosed, did hey vomit?

DR. DUNNE: The way we collect the side effect data is by dates. So, if you have an episode of any adverse event we just list the date down there, but on questioning to the coordinators this is generally something that happens one time. There was, and I should, also say that when we look at the mild, moderate, severe assessment of all these side effects 80 percent of the vomiting was mild. There was 20 percent moderate. There were no severe vomitings which we interpret as being persistent and recurring although the investigator is able to assess that for himself, and the other question was did they revomit after redose?

DR. PATTERSON: The patients who were redosed.

DR. DUNNE: Yes, there was one of those eight children who did actually vomit again on the redosing, and then they got amoxicillin, I think after that..

DR. RELLER: Thank you, Dr. Dunne.

DR. ALEXANDER: Related to the vomiting, sorry, Dr. Alexander. Related to the vomiting I actually think that there were two patients that were within the non-comparative study who vomited who were redosed and then one of them, both of them vomited again and one of them or both of them received amoxicillin, I believe. One of them was excluded from the MITT analysis as being considered to have vomited all the medicine that she received, so was treated as though she never received any medication within the efficacy analysis of the study.

The other one, I believe, was included as a failure because of receiving amoxicillin instead.

DR. RELLER: Dr. O'Fallon, you had a comment?

DR. O'FALLON: You see that is the problem with leaving people out. That child, yes, didn't get treated, but couldn't get treated and therefore should be considered a failure instead of left out of the MITT analysis, the one that was left out.

DR. DUNNE: I would have to go back and check on that other patient. I cannot really comment on that right here.

DR. RELLER: Dr. Goldberger?

DR. GOLDBERGER: I thought perhaps before the vote actually occurs it might be useful to make a few comments about this issue that came up about what guidance ought to be utilized in terms of thinking about the voting.

You know, I think obviously as a starting point the first thing to do is to make a determination in terms of the recommendation you are going to provide, is this product safe and effective based upon the current guidance that exists. Concerns have been raised about the amount of pharmacokinetic data and other issues, and clearly the expectation is you will take these matters into account in making your vote.

Then the issue has come up as well about the fact that it would appear as though the current approach to development of products in this area may have certain limitations, and I think it should be obvious to everyone, you know, that we certainly recognize this as well, both in terms of our asking you to comment on some of the things that were important to you as part of question 1 as well as in the fact that there will be some specific discussion about this later in the afternoon.

On the other hand there has been some discussion in the last few minutes about what studies can and cannot reasonably be performed, and I think that one needs to take into account that if one takes a new standard as the de facto standard involving studies in which there is legitimate disagreement as to whether they can be performed that obviously poses certain problems.

In addition, there has been some important PKPD data presented here today that may well be very influential in subsequent guidances, but may require some additional review before one can reasonably conclude that it should become the standard.

So, I think that you are free in the end, obviously to make the recommendation as you see fit. We do, you know, like to understand your thinking, and you should be aware there can be a situation in which our regulatory requirements may require us to think a little more in terms of exactly how to utilize your recommendations given that there may be some difference between what we would like to be able to do in terms of developing drugs and what the current standards are today.

I hope that helps you a little bit in thinking in terms of the voting.

DR. RELLER: Thank you, Dr. Goldberger.

I propose the following. We will go starting at my right with Dr. Wald and ask for since some of the part 2 and the subsets of the questions require this delineation for Committee's advice to agency that we simplify it to single dose 3 day and give us a yes or a no for each and if it is yes for a single day, if you want to add any caveats as Part A delineate please say so, and if the answer is no, any additional studies for single day and then go to 3 day with the same comments and then we will take question No. 2 depending on what the majority view is from the initial vote.

Two yeses or no's and plus or minus comments related thereto.

Dr. Wald?

DR. WALD: I guess what I would conclude is that both the single day and the 3-day regimen compared favorably to the comparator drugs in a study design which would not allow us to easily distinguish a better from a worse drug but that this has, in fact been the standard by which we have judged drugs, and so in that context I would say that the data support the safety and efficacy as we have until now interpreted it, and so that would be a yes, yes, without caveats.

DR. RELLER: Thank you.

Dr. Leggett?

DR. LEGGETT: Yes, for single, yes, for 3 day, no caveats.

DR. O'FALLON: No, for both. I don't think that the evidence we see has established the effectiveness or safety actually of either.

DR. RELLER: Dr. Christie?

DR. CHRISTIE-SAMUELS: I have to say a no for both.

DR. RELLER: Any additional studies, Dr. O'Fallon or Dr. Christie with your no votes?

DR. O'FALLON: Yes, I think that the bacterial studies although the two in the literature are some start, but I think there should be those and then for definitely sure the PK studies in the younger children, maybe even PD but certainly PK in the younger children before we can approve it, because that is apparently the majority of these patients. I think we should see evidence in those patients.

DR. RELLER: Dr. Christie, a part B comment for either or both 1-and-3-day regimens?

DR. CHRISTIE-SAMUELS: PKPD data, larger Ns looking at ototoxicity and hepatotoxicity, children younger than 2 years of age, studies in day care centers and Dr. Dagan's comments, the second to last slide for hand-tailored AOM drug studies. I think he summarized it pretty well.

DR. RELLER: Dr. Chesney?

DR. CHESNEY: I would say no for the single dose because we have no pharmacokinetic data and this is such an appealing dose form. It is 100 percent compliance in the office. It will be used instantly for all ages, and I have problems with that.

I would say yes to the 3 day regimen because it is already being used 5 days, and I don't think 3 days is that much different but only for children over 2 years, again, with absent kinetic data.

DR. RELLER: Dr. Ebert?

DR. EBERT: Yes, for single dose, yes for 3 day regimen. I do believe we need to do a better job of measuring or comparing response in agents earlier in the trial period and not waiting for 28 days to assess outcome.

DR.CROSS: While the data are not optimal I think it would be helpful as has been brought out by other Committee members to have some better data on children less than 2 years and have better PKPD data, I think within the context of the guidelines and its equivalence or at least no difference with comparators I would vote yes for the 1 day and yes on the 3-day regimens and no caveats.

DR. RELLER: Thank you.

Dr.Gorman?

DR. GORMAN; I would vote yes n the 1 day and yes on the 3-day dosage of this form. Caveats is such a broad terminology. What caveats are we allowed to put on the label in terms of the question of tympanocentesis? I am looking for guidance on the caveats that are allowed under your regulatory statute.

DR. SORETH: I think that what you can take into account are issues of age, if you feel that there are data to speak to differences above and below the age of 2. Caveats have at times been put into labels not only for otitis but for other infections as well that some organisms, patients with some organisms recover better or have different rates of success than others.

DR. GORMAN: I would then put in a caveat that this drug not be approved in children under 6 months of age until the sponsor company can provide PKPD and toxicity data in that group.

DR. RELLER: Dr. Glode?

DR. GLODE: I would think that the data do support. I am going to divide up safety and efficacy for just a moment. They do support efficacy of the single dose and of the 3-day regimen compared to the licensed products.

I will, also, support it with regard to safety mainly on the issue of 40 million doses distributed to people of different ages and presumably having been used safely, but I am disappointed that there were not actually more laboratory values for us to look at with regard to safety issues and certainly the absence of PK data was, also, concerning.

DR. RELLER: I would vote no to both 1 day and 3 day based on the paucity of pharmacodynamic data on the 1-day dose and with the 1 day and 3 day where the bacteriologic confirmation is available with one of the studies and the other it is the pharmacodynamic, and I think that the total database though the efficacy by the past criteria is comparable I think the database could be more solid when we go to these new dosage regimens, and it is just not all fitting together for me for either the 1 day or the 3 days. So, my vote is no for each now. I think additional studies are needed before approval of those regimens.

So, what we come out with is six to four for single-day therapy and seven to three for 3-day therapy which enables us to go to question 2 for the single dose regimen. We will come around the table in reverse order, starting with Dr. Glode. What advice on it if any should be given regarding the issue of vomiting associated with the 1-day regimen? This is information for prescribers and patients that presumably would move its way into a package insert, advice for the agency to consider in their final decisions on the issues at hand.

DR. GLODE: Okay, advice for the agency to consider, perhaps a statement that says that with the single dose of azithromycin an increased rate of vomiting has been observed and then I suppose some comment should be made to prescribers about under what conditions and how many times the drug should be recommended to be redosed, and I really with N equals 8, I don't know that I could come up with that. I could make up something. If the patient vomits within 1 hour you may redose times one, but that is as far as I can go.

DR. RELLER: Thank you.

Dr. Gorman?

DR. GORMAN: My recommendation would be more in terms of packaging, much like syrup of ipecac in the past. I would make sure there were two effective doses for the largest child that you would expect to require liquid. So, in ipecac if the first dose was not successful you administer a second dose. In this case it would be exactly the opposite. If you vomited the first dose I would like the packaging dispensed by either the pharmacist or the physician to have enough for the second dose available so that the rescue dose would be available to the parent without another trip to the pharmacy at no added expense.

DR. RELLER: Dr. Cross?

DR. CROSS: I think, also, that there ought to be some comment about the need for redosing if there is vomiting within a specified period of time and on the other hand we don't have enough data on that to actually make that recommendation. So, I think we have to urge the sponsor to acquire that so it is included on the labeling.

DR. RELLER: Dr. Ebert?

DR. EBERT: I agree that there should be some window at which vomiting before that time would require redosing and vomiting after that time may not require redosing. As the previous panelists noted I am not sure that we know the exact time frame for that window, and I concur that if at all possible at no increased cost if we were able to give an extra dose that would certainly save from a consumer standpoint an additional trip.

DR. RELLER: Dr. Chesney?

DR. CHESNEY: I don't think we can make any recommendations. If this is absorbed well in the first 20 minutes and then the child vomits and we give another dose now we are talking about 60 milligrams per kilo as a single dose, and if I was nervous about 30, I am really nervous about 60. So, I don't think we can make any recommendations.

DR. CHRISTIE-SAMUELS: May I seek some clarification please? I have a question. The recommendation is from the Committee, if the Committee recommends that it be approved or is it --

DR. RELLER: I will give you some help, Dr. Christie. I voted no, and consequently I have no additional advice because Dr. Chesney summarized and that was one of the considerations in voting no because I simply didn't know enough about what to do in the 10 percent who vomited, and I think there are sufficient issues regarding safety that without the pharmacodynamic data and the group in greatest need if bacteria would be demonstrated on a tap, and it is not there for me, and it is one of the considerations.

Now, you can comment or not comment.

DR. CHRISTIE-SAMUELS: Thanks for the clarification. I guess I have to abstain since I didn't make the recommendation in the first place.

DR. RELLER: Dr. O'Fallon?

DR. O'FALLON: I, also, did not make the recommendation, but I would like to point out that I don't think there is any evidence or at least certainly not adequate evidence to try to make a decision on this. There is not enough here to make a decision, at least from my perspective.

DR. LEGGETT: I would like to point out the fact that many parts of the world don't even use antibiotics for acute otitis media much in the same age group as these folks. So, the only thing I would say is just note that increased vomiting is seen with the one dose and not say anything about redosing.

DR. RELLER: Dr. Wald?

DR. WALD: I think that we have to let common sense guide us here and that we would certainly give advice or we would let people know that increased vomiting is observed and I would say that it is reasonable that if the child vomits in the first 30 to 60 minutes to repeat the dose.

DR. RELLER: Dr. Soreth, can we provide any additional information to clarify what we want to convey with questions 1 and 2?

DR. SORETH: Yes, Dr. Reller. There were several comments from Committee members and guests with regard to additional studies for PK and PD. Could I just get a little clarification about what those would entail?

DR. RELLER: Dr. Chesney?

DR. CHESNEY: I feel that we need more information at least in the child under 1. I am just pulling that randomly, but I think as I mentioned before this is such an appealing dose formulation that it will be used in 1-month-old infants, and it may be used in 2-week-old infants, and we know that their excretion mechanisms both liver and kidney are not fully mature yet. So, I would guess if one had to limit the age groups for me it would be the lower age group that I would most like to see pharmacokinetic data on.

DR. RELLER: Dr. Burns?

DR. BURNS: I would say, too, that if you are going to put in a caveat about redosing after vomiting that you would have to do PK in patients who had been redosed so you knew whether or not you were getting into a potentially toxic situation.

DR. RELLER: Dr. Chesney?

DR. CHESNEY: I don't know the circumstances under which kinetic data can always be collected, but if it could be done in febrile infants who are being treated who may have viral syndrome where the kinetics might be different than in a perfectly well child who might not be stressed, I think that would be very helpful.

DR. RELLER: Dr. Soreth?

DR. SORETH: Dr. Reller, you, also, made a comment after talking about the pharmacokinetics that you thought additional studies were needed. Did you mean outside the realm of PKPD or what were your thoughts?

DR. RELLER: We have bracketed one single dose and 5-day regimen and there are some extrapolations to the 3 day based on if the 1 day has the studies that involved bacteriologic outcome and then we have the data from the efficacy in use with the 5 days. I don't see the studies being done on each of the regimens is what I am trying to get at, so that the full database for each of the regimens is not the same magnitude number of patients, etc.

The studies are different with the different dosing regimens, the content of the studies available for assessment, the design of the studies.

Dr. Ebert?

DR. EBERT: This will probably be an incomplete answer, but I noticed some comments about PK but not a lot on the PD side of things. I think ideally from a dynamics standpoint we would be talking about bacteriologic eradication as an end point which we really have not seen to date with this regimen to compare single dose and multi-dose regimens with regard to their bacteriologic efficacy.

The secondary issues, of course, would be those of toxicity, whether there may be an association between serum concentrations and adverse effects.

DR. RELLER: That concludes Phase I of the meeting. We will take a 5-to-10-minute break very briefly for a stretch. So, those who are not staying for part 2 can exit and we will launch at 10 minutes of 3 into the discussion of the guidance document.

(Brief recess.)

DR. RELLER: We will start this afternoon's part 2 session with the presentation by Dr. John Alexander and then we will address the question.

DR. ALEXANDER: Hello, my name is John Alexander, and now that we have finished the questions specific to azithromycin we would like to have sort of a continuation of the discussion that we have had already today regarding the clinical trials of acute otitis media, and what I am going to try to present is a little bit of information about our guidance document.

I will start off with a little bit of history and I promise not to go all the way back to 1962 and detail everything in painful detail since then, but just to start out, the Kefauver-Harris Amendments to the Food, Drug and Cosmetics Act were just significant in that after 1962 was really when drug companies were expected to show evidence of efficacy as well as what they had been previously doing which was the safety of drugs, so that it is really in the sixties and seventies where we began to consider, well, how do we show what drugs are effective and actually most of the sixties and seventies were spent addressing what were called DESI reviews looking at already approved drugs and their efficacy.

So, it wasn't until 1977 that we had our first guidance document on the clinical evaluation of anti-infective drugs, and I wanted to pull out these two particular quotes from that.

Within that document there was only about half of a page that dealt with the question of acute otitis media and I thought these two quotes were reflective of the overall tenor of that particular document, one saying that it is necessary to confirm the presence of exudate in the middle ear by pneumotoscopy(?) and needle aspiration to obtain fluid for culture, meaning that patients, they were looking at pneumotoscopy as a method for particularly accurate diagnosis of patients with acute otitis media and the needle aspirate in order to identify those patients who actually had bacteria on culture, and the other quote, in the absence of culture of middle ear fluid no specific claim can be made regarding the effectiveness of any anti-infective drug, and so what we saw after this guidance document in the seventies and most of the eighties for anti-infective drugs that were looking for a claim in acute otitis media was that we had certainly a smaller number of patients that were studied but in those patients that were identified in those trials, they had bacteriologic information on all those patients at baseline.

So, we move forward in time to 1992, and the points to consider document, and this quote should look familiar from the sponsor's presentation this morning that two studies were identified in the points to consider document, one a statistically adequate and well-controlled multicenter trial that establishes equivalence of superiority to an approved drug and a second open study using tympanocentesis at baseline to establish microbiologic etiology, and this is the first point where we were actually listing 25 patients with Strep. pneumoniae, 25 patients with Hemophilus influenzae and 15 patients with Moraxella catarrhalis.

Now, I would, also, point out that in November 1992, in the CID there was a collaboration between FDA and IDSA, also, looking at studies looking at anti-infective drugs and that included, also, a recommendation about studies for acute otitis media that made similar points to this.

So, now, we move forward in time to 1998, and to the draft guidance for industry document that we have.

Now, in that two studies are again recommended, and this should look familiar, very similar to what was in the points to consider document.

Here we just say another trial as opposed to specifying that it can be an open study and otherwise the trials are the same, so, the first, being a clinical only study as we have defined it in your briefing packets from the FDA and the other a single tap study.

Now, for the first trial, the clinical only study the caveats that are made are ordinarily should not enroll children less than 6 months of age and Dr. Wald had brought up this question earlier today, and from what I could tell from looking at those documents there were concerns about the accuracy of the diagnosis of acute otitis media in those children less than 6 months of age and that was part of the reason that those patients were excluded.

There was another caveat in our guidance document related to rigid case definitions and the goal there being to try to create case definitions that would identify within the clinical only studies those patients who are most likely to have a bacterial infection.

Now, in this study it says that baseline tympanocentesis need not be performed, but we do say that tympanocentesis of patients that are judged to be therapeutic failures is strongly encouraged to document potential specific bacterial pathogens that are not adequately treated within the trial.

Unfortunately, what we end up with especially from the clinical only studies is that we rarely, if ever see information on tympanocentesis done in patients who are therapeutic failures.

Now, for the second trial, the micro study, again, post-therapy tympanocentesis is encouraged in patients that are judged to be therapeutic failures. To identify that we want to have at least two investigators in geographically diverse regions for the pivotal studies for microbiology, and there is a quote that relates to acceptable clinical and microbiologic effectiveness of all three organisms in which case if we don't have what we think are acceptable effectiveness against all three microorganisms then for those organisms where effectiveness is in question, we specifically don't list that as, that we can specifically exclude that from the indication where we typically list it as otitis media due to and list specific pathogens, and we could, also, put in a statement that says that it is restricted use and not as first-line therapy.

One of the other caveats in the draft AOM guidance is related to resistance in Strep. pneumo, and we make a statement saying that 25 patients may be insufficient when you are looking at a question of a drug that might be effective due to a resistance mechanism, especially if the resistance mechanism, for instance, penicillin resistance would affect your particular drug like a beta-lactam antibiotic.

Now, I am not planning in going through in detail all the inclusion and exclusion criteria and the guidance document as it currently stands is within the briefing packet that you were sent from the FDA.

I did want to make some points related to the study visits and the way that the draft AOM guidance currently reads there is an entry visit. On therapy at 3 to 5 days there is a strongly recommended visit in order to assess the clinical status of patients at that point. The end of treatment in the guidance is currently optional and the test of cure stands now at sometime between 2 to 4 weeks after entry is when the test of cure visit should be held.

So, it is not out to 30 to 42 days and sort of later follow-up that had been seen earlier. There is a late post-treatment follow-up that is offered as a sort of an optional visit and that is related to the presence of middle ear effusions that you can follow patients out to around day 30 to see whether middle ear effusions persist or not.

For the most part we don't think that beyond that point whether a middle ear effusion is present or not is really related to the effectiveness of an anti-infective drug given 30 days earlier.

Now, in addition to these guidance documents we have had four different advisory committee meetings out of the past 11 that have in some way addressed acute otitis media, and I am going to make points related to each one.

In spring 1997, we had an initial presentation of our guidance document and this was within the context of actually presenting several different guidance documents on several different indications so that the discussion about acute otitis media was about an hour or so worth of 2 days of meetings, 2 or 3 days, but some of the issues that came up during the discussion after the initial presentation of this guidance, one was related to baseline clinical findings and again making the point about concerns about how do we get a case definition that will adequately define those patients who are likely to have a bacterial infection and truly need an antibiotic for treatment especially for the clinical only studies.

There were some comments at that meeting about the timing of the test of cure visit and at that point these were just a few comments that were related, some questioning what is the value of taking a look at the end-of-therapy visit when the 2-to-4-week visit is probably acceptable, some comments that were pushing towards looking at the earlier time point, the day 3 to 5 visit as an important point to look at clinical efficacy of the drugs and that is why the, that is in part why the guidance reads the way that it does, having a test of cure visit at 2 to 4 weeks and making a strong recommendation for an on-therapy visit at day 3 to 5.

One of the other issues that came up was addressing the number of Strep. pneumoniae isolates that were needed in light of the increasing resistance, and that is where the comment that I had made before about 25 patients may be insufficient with Strep. pneumo, but that is not really something that we likely to get into a great deal today.

Then the fourth issue that was discussed was related to follow-up of middle ear effusions out to 30 days, and again I made those comments earlier with relation to the post-therapy late follow-up visit.

Now, in fall 1997, we had a meeting on ceftriaxone for acute otitis media and for the most part the discussion focused on the product itself as you would expect.

There were at that time also comments that were made related to the resistant Strep. pneumoniae and concerns about how to adequately address those patients within the designs of these trials, and there were, also, other comments that were related to study design. For the most part they addressed the issue of the use of pre-therapy antibiotics, studies which excluded patients who had antibiotics more than 30 days or less than 30 days prior to starting the study versus just less than 7 days.

In 1998, we revisited the guidance and the FDA presentation was of changes that were based on the spring 1997 advisory committee where those issues were discussed and leading to what I had presented as what our current draft guidance for AOM is.

At that same meeting there was a lot of discussion that surrounded the letter from the CDC to the FDA and this is a quote from the recommendations of the CDC that since clinical only studies would need to be prohibitively large to detect a difference if one truly existed between two drugs that they were recommending smaller bacteriologically driven studies that would be more effective in showing the effectiveness of a drug.

So, at that point there was a lot of discussion about the different trial designs and alternatives and some of the members of the Committee that were there during that point are here today and hopefully can speak to some of those points.

There was, also, a lot of response from the industry during that advisory committee meeting about the concern of how to do even single tap studies, about the difficulties of enrolling patients in those trials and again, what ended up being left after this meeting was over was the current copy of the draft guidance that you have in your briefing packet.

In January of this year we had a meeting that was on Augmentin ES-600 and I am hoping even more of the Committee was here at that time and will remember that. Again, the discussion focused on the particular product. We, also at that time asked questions related to the timing for the test of cure visit and got some input as to actually looking more towards the end of therapy visit as opposed to the later 2 to 4 week test of cure and there were, also, comments made at that time related to the assessment of bacteriology and clinical failures and how that could potentially be important which leads us to today and again, today's discussion focused on the product itself and it seemed natural from the issues that were being discussed that there are issues related to clinical trial designs that we would like to have some input from the Committee on.

Now, I wanted to make at this point it clear that anything that we do in terms of trying to address the guidance in the future and from comments earlier today as well as from comments at the augmentin ES meeting it seems apparent that we are likely to need to change the guidance document, but any kind of changes would be something that we would probably announce in terms of the Federal Register and open it for comments from the public, looking at different fora to get experts in the field together and probably again take it to another advisory committee where our purpose is to actually discuss the guidance document itself rather than to discuss a specific product.

But given that we do have some issues that we would like you to discuss further and flesh out for us as sort of a lessons learned from today's discussion and these questions are intended to be in a way provocative in order to try to engender some discussion, the first one, should clinical only trials continue or should all enrolled patients have a tympanocentesis at baseline. The second question is should the non-comparative microbiology studies continue; so, should we be actually looking for only comparative data.

A third is should guidance incorporate stratification by age less than 2 years and if so, what proportion of patients should be under the age of 2 years, and what we would like you to consider in your discussion is to comment on the merits of studies with clinical information only, studies with tympanocentesis, the ideal timing of clinical assessments, the end-of-therapy versus a later follow-up and alternative study design, so, comparative trials with microbiology, double-tap studies, placebo-controlled studies with early escape if you can define an appropriate population and we are hoping that what these, what we have outlined here will be issues that the Committee will discuss more thoroughly, and before I stop talking I would just like to acknowledge the hard work of all the review team as well as the contributions of the entire Anti-Infectives Division to putting together the information for this meeting today.

Thank you.

DR. RELLER: John, thanks very much. We look forward to the discussion. I think it will be rigorous and people will make pointed recommendations for your consideration.

Who wants to open it up? We will take question No. 1. Should clinical trials, clinical only trials continue?

Dr. Dagan?

DR. DAGAN: I think that you like to have efficacy data, but you need, also, to have safety data, and if you do very, very thorough studies with microbiology, let us say even with tympanocentesis and you get a relatively small number of kids with very good efficacy data, you still are not going to have enough safety data.

So, I think that for that purpose you need studies with a lot of kids, whatever is your criteria for getting satisfactory safety data. So, I think that this discussion should be separated for efficacy and for safety and since for efficacy you may find yourself eventually hopefully for me at least, I hope that you will be doing less kids but more thoughtfully studied then one would like to divide this into if one needs more kids for safety what are we going to do, just give the drug and see the safety or are we going to design a study that looks for clinical outcome and within that to build on the safety, and in my opinion therefore the first bullet is yes, we should do clinical only trials but as a secondary importance for efficacy but probably important for safety and then, of course, it is answered the same way, should we, the lack of comparative microbiology studies, I guess this is all related because I think this is the wrong, I am sorry, I think this is the wrong sentence because if I understand what you want it is when drug only with no comparison to another drug but the other comparative microbiological studies if you do the ears or if you do azithro or whatever and then with azithro we did see cases; it was not done in double tympanocentesis but you have different cutoffs of MICs and actually these are double-blind studies because when you look at the second tympanocentesis or when we look at whatever you don't know the outcome. You are just going to more to tap. You do the evaluation and how are you going to see what happens and then break it by MICs and you see if you have very sensitive susceptibility, make it susceptible and resistant and you look at the clinical outcome. This is extremely nicely constructed double-blind comparative study but instead of comparing azithro to augmentin you contrast susceptibles to intermediate or resistant and then you can come with what type of cutoff to approve to children.

So, I think this is extremely important and they are comparative, but they don't compare one drug to another, that although they give you actually the most important information, this is up to what level of resistance you should still approve that drug and if you ask it as well of your requirements then the sample size should be calculated and then people are going to come with enough data, not just 1 of 6, 3 of 20 and 1 of 4 or whatever. They should calculate the sample size to answer that question which is extremely important, I believe.

DR. RELLER: Dr.Marchant?

DR. MARCHANT: Notably absent from the questions posed is the question of what differences are you looking for and what sample size is needed to calculate. If the guidance doesn't address that issue then we will continue to look at a lot of small studies, look at a lot of data which never would have shown any difference. An example would be the questions raised whether you should look at 25 pneumococci or more because there is resistance.

Underneath that question is really still the question of sample size. What you question, what is an important difference in terms of pneumococcal resistance, how many resistant serotypes do you need to look at versus how many sensitive; what kind of outcome and all that underneath all that question is the size of the expected difference, the important difference and the sample size questions from that and if this guidance continues to ignore that, then these meetings will continue to on the basis of very small studies that wouldn't have passed the tap water standard to approve drugs or regimens that are not effective. So, that is absent from the mix here.

So, even addressing question 1 you need to figure that into question 1.

DR. RELLER: What do you think it should be?

DR. MARCHANT: Well, I think, so, this is clinical; so, actually those are the same questions here. Well, I think that if you do a tympanocentesis at baseline you have a better chance given any given sample size of showing a difference than you do if you do not do tympanocentesis at baseline, but the fundamental problem is you need to if you are going to do a single tympanocentesis study and look at clinical differences you need an outcome that you are going to be able to show a difference with or you are not doing a scientific study. You are going to stay under the tap water standard.

What is that difference? When you look at the clinical data from placebo-controlled trials the maximum difference you can find between a placebo and a drug is about 15 percent for some end of therapy outcomes, and that is with enriched populations that are highly symptomatic and not just taking all comers.

So, you have to then link not only the sample size questions to the outcome but also to the population you choose to study the outcome in and all those things can be calculated based on samples in the literature, but the question has to be refined more.

DR. RELLER: Dr. Dagan, Dr. Jacobs, yourself and others on the Committee have, there have been some very thoughtful and specific comments made on clinical trial design.

If you were the FDA what would you want done to consider first the efficacy and then as Dr. Dagan pointed out the safety? What numbers would you want? How would you do it?

Let us say you could just unilaterally make the decision. How would you want the data presented?

DR. MARCHANT: I am going to broaden your question, but I am going to get to all of it. The first issue I would do is as follows: Because the analysis that I presented that Dr.Jacobs presented and Dr. Dagan and so forth, I would do a double tympanocentesis study and I would look for a sample size which will detect the difference between a drug that is 90 percent efficacious bacteriologically and one that is 70 percent efficacious and that requires under 300 total recruited patients to do it.

Now, do I like double tympanocentesis studies? No, I don't entirely. I mean it is a painful procedure and as I mentioned in my recommendations we need to look at how to deal with the painful procedure part of it. My colleagues whom I respect in the academic field who, also, don't like tympanocentesis believe that we could show differences with clinical outcomes if we really drilled down and focused on better clinical outcomes, and that would include specifically saying, "How was the child on day 1 and day 2 and day 3 and so forth?" and then summing a cumulative response over time, that kind of approach to get at it. Studies should go on to develop a better clinical outcome because maybe there is something we can do by looking at clinical outcomes. I don't exclude that, but that is, also what we should do. In the meantime we should be doing double tympanocentesis studies if we want to show, find out whether drugs work or not, but I am not saying that that should be forever and ever the approach. There could be new science and better outcomes looked at.

For safety how big should that be? That is, also, the question you are asking. Well, you should take an event like well, okay if I don't like 2 days of vomiting, that that is a side effect that is very troublesome clinically and I want to detect whether my drug makes kids vomit for 2 days and I care that, you know we have to make a judgment here, and I care that if a drug had 10 kids out of 100 let us say that had that outcome.

Then you can calculate the sample size that it is going to take to show a 10 percent increase over the baseline noise that you see in clinical studies with antibiotics, and there are ways to calculate that sample size. Statisticians know how to do that.

So you define, and that is rather asking the total sample size for safety as Dr. Dagan mentioned and on the vaccine side, for example, they are getting interesting events that would occur 1 in 1000 or 1 in 10,000 and then you start driving up your numbers but it depends on how severe they are, but you should be able to create some professional guidance about how big a safety problem you what to look for. Again, it is a statistical question on the safety side just like it is on the efficacy side.

Have I addressed your question?

DR. RELLER: You have got us started because then we can come around the Committee as to whether or not this is in the ballpark, should be more or less, but, John, you had a comment?

DR. ALEXANDER: In order to try to focus the discussion a little bit more on the issue of efficacy one point that I would make is that typically when we are dealing with safety where we are talking about drug exposure what we are looking at is not only studies of acute otitis media, but we have, also, got studies of patients receiving the drug for pharyngitis, studies of patients receiving the drug for multiple other indications that could include young children and so for the purposes of the discussion I think that what we would like to try and sort of focus more on is the issue about really demonstrating effectiveness in acute otitis media.

DR. RELLER: One approach might be that if someone comes with an outcome for which there already is an appreciable experience with other indications that could be included in the presentation but if we didn't have that you may need larger numbers at the outset with a new drug, a first indication but again those numbers could be forthcoming from within the agency and the consultants thereto on the statistical aspects.

So, let us concentrate on the efficacy.

Dr. Cross?

DR. CROSS: I think what we would like to do to handle the efficacy issue is to actually enroll informative patients as opposed to less informative patients. I think that in the last 10 years since many of the guidelines have been written for this and other processes we have learned a lot through clinical research who are the high-risk patients who are the more severe patients. This is certainly true in our last discussion on sepsis. It is certainly true in issues of febrile neutropenia and I think in each of those areas we see in these studies patients who might not be as important in the case of febrile neutropenia folks who have very short durations of neutropenia as opposed to long durations. In the case of sepsis we had the issue last time of patients who were more informative at the more severe end.

I think that Dr. Dagan had a very nice presentation of what we have learned over the last 10 years in terms of which patients are more likely to have difficulty with therapy than others. I think if there is something which we might be able to add in terms of having some mix of severe patients who might be more informative based on a previous experience I think that would be very helpful in evaluating patients.

DR. RELLER: One of the things I would like to comment on in the emphasis on the tympanocentesis and I know that there are different views, that they shouldn't be done and they are very important, and Dr. Marchant has pointed out that smaller numbers, better studied would yield more information and one of the things that struck me over the time span that we have had these discussions on otitis media is the field has shifted, that the decrease in Hemophilus influenzae with immunization and the conjugate vaccine for pneumococci, Dr. Glode pointed out some forthcoming information that may continue on the interaction of antimicrobials and immunization practices and to me the beauty of tympanocentesis and the microbiology and of course it applies to resistance issues as well is then you have a study design in place that can capture the ground that is constantly shifting underneath you whereas these differences clearly get buried with the emphasis on clinical outcomes, particularly clinical outcomes at a time when the reinfection rates are very impressive to me when you look at things too late after the acute events.

Dr. Wald, and because I can't remember whose hand came up first we will just take one round around the table and then I will try to watch for whose hand goes up.

Dr.Wald?

DR. WALD: I think that Hemophilus influenzae type B vaccine has had any impact on acute otitis media, I mean acute otitis media even in the heyday of encapsulated systemic disease was classed almost exclusively by non-typable organisms. So, I think that really hasn't had an impact, and the impact of the pneumococcal conjugate vaccine remains to be seen. I mean there may be a shifting of types but it may not be a very dramatic impact.

DR. RELLER: Granted. I only mentioned that as changes in the microbiology of respiratory tract infections, not necessarily with otitis media itself and as things change if you know what you are dealing with you can handle any change that comes about.

DR. WALD: Right, and so I agree with you entirely. I think that it is very important for us to perform tympanocentesis and I think that groups of children that we should be studying are children with severe disease.

I mean there are tons of arguments these days about non-treatment for acute otitis, but I think what we would all agree on is that severe disease needs to be treated and so I think those are certainly children in whom one can justify diagnostic tympanocentesis without any, I think without too much trouble. I think the double-tap design is another issue and I think it is going to be difficult to get people to do that and I don't think I would submit my child to a second tympanocentesis if they were doing clinically well. I mean that I cannot justify, but I think a diagnostic tympanocentesis for children who have severe disease which means we should load that clinical cart with young children, with children who may have had recurrent infections, with children who do attend day care and we should put in a very stiff requirement for there to be a bulging tympanic membrane which again is going to make our yield higher.

If we were going to consider doing clinical only trials my own bias would be that they only be done in centers who have already proven by virtue of their tympanocentesis studies that they have a yield of more than 70 percent positive bacteriology. I think to do studies in places where they have not proven that they can make a diagnosis, an accurate diagnosis of acute bacterial otitis media is really to waste everybody's time and energy and is not fair to patients.

DR. RELLER: Jim?

DR. LEGGETT: To follow up on that which is sort of the reason I brought up the question during the earlier period, if you can't make a diagnosis of otitis, how do you make a diagnosis of cure of otitis at the end of therapy? That bothered me a lot and as with all these studies.

I agree with many of the things being a PKPD person myself obviously but I would sort of want to go through the handout and just make a few points that I think relate to these questions so far.

The first thing is we don't require double sampling of other infections that I am aware of except maybe endocarditis. We don't do an LP twice. We don't do paracentesis twice. We don't do urinary tract infections, and so we have got to think about that and I think I would approach drug companies with the idea of what is statistically adequate which is the third slide, and so we give them the choice. You can either enroll 3000 people in a clinical only study or you can enroll 100 with a double tap or intermediate number with a single tap. If we do do the single tap I think that what we should do is require tympanocentesis of everybody who is called a failure either at 3 to 5 days or at 10 days which I think should be our major, at the present time, the end-of-therapy trial should really supersede the 1 month which should just sort of be to see if there is still a chronic effusion.

I, also, agree with going away from the 25 H. flu, 25 Pneumococcus, 10 Moraxella and just sort of look at those numbers in terms of what is a statistically adequate sample and that way it allows you to capture any emergent pathogens as well as really emerging resistant pathogens and allows people to look at so now we have 25 percent efflux things and we can go back and look at the MICs and find out what PKPD breakpoint might be.

Then I think in terms of the final thing for the time being it was that 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?

So that should actually probably I don't think require an exclusion unless the antibiotic was sort of like the day before or something, and I will shut up.

DR. RELLER: Thanks.

Any other spontaneous comments on this issue and after we go around the table I want to come back and ask Dr. Dagan his views on double tap, single tap, required taps with failures unless patient or parent refuses.

Dr. O'Fallon?

DR. O'FALLON: Remember that I am a statistician. I am not taking care of these sick kids. I, however, have been the mother and grandmother of such sick kids. So, I know a little bit about what we are talking about here.

One of the things in looking over the data that has really impressed me is the high cure rate in the natural history of the disease. The patients without bugs recover, 80 percent I think was what it was. Patients who even failed the antibiotic therapy recovered, what was it, 62 percent? It was some very high percent. So, in a certain sense the stakes in this game aren't as high as in some others because over half of the kids are going to recover with or without any kind of therapy. So, you can fight with me about that, but that is what I am seeing in the data.

The second thing that is bothering me a lot, I have been listening to you guys, and you are very worried about the promotion of drug resistant bugs, and if we are treating a bunch of kids that don't need it what are we doing to them for their future? Are we being good to them in the long run? Are we being good to the community in the long run? I think there is a big issue here more than just these children.

It is for each of these children because the drug resistance is an issue for each of these children potentially anyway, but it is, also, for the community, and I think we have got to worry about that.

Because placebo has such a large success rate I think that, and because this disease is shifting so fast, look at even the 3 years I have been on the Committee the drug resistance rate has shot up, tripled, quadrupled in just the 3 years I have been on the Committee. So, we have got a fast-changing disease here. Should we, I think maybe that makes an argument for placebo-controlled studies and it was a good idea there about having a back out. If you tap them, if they are a failure, you know, they look bad after 3 days, tap them again, find out what it is, but I think that is an important thing to do. Then you can go on and give them something decent, I mean something that you know has a track record for whatever they. So, I think it is not a bad idea, and I lay it out for you.

I think it is important to do the randomized double-blind and concurrent, the concurrent to deal with the rapidly shifting mix of studies, I mean mix of diseases that we are seeing here.

One thing I would like to see is maybe the FDA could ask these guys now. They treated a whole lot of patients that turned out to have no bugs in their original thing. We didn't see anything about them. We never saw the information about what happened to those guys. I think they had the data in the database. I think it would be important for us to see how did it work with those guys because they just disappeared when they didn't have any of the bugs of interest. So, I would like to see that.

We know that the age groups are important. All who deal with them know that the children under the age of 2 maybe under the age of 1 are different versus 1 to 2 versus greater than 2, react differently possibly.

When we do that sort of thing we say stratify so that you have adequate sample sizes for subset analyses and so I think that type of thing has to be built into the studies for the future, and well, I have said the drug resistance.

So, those are my main points off the top of my head.

DR. CHRISTIE-SAMUELS: I agree with my colleagues. I agree with everything that has been said so far. Another point though, it is true that the FDA develops guidance with regards to antimicrobials for the United States, but as a practitioner who has been here for several years and now practicing in a part of the Third World in Jamaica for the last 2 years I would like to say that the minute something is approved here, it is also immediately marketed and used in other parts of the world, Jamaica for instance.

The problem though is that you can get drug resistance developing in those patients and because airline travel is so wonderful essentially the resistant organisms which develop in another country again become America's problem. So, I think I just wanted to say that we need to be very, very rigorous in what we do here because what you do here impacts the whole world, and then it comes right back to you a few more years later.

DR. CHESNEY: Many incredible issues, and I really, really appreciate that the FDA has asked us to be so open in this session today. I think we haven't had -- we have had discussions before, but I think this is terrific and particularly with the experts we have here.

Just a couple of brief comments. With respect to the double tympanocentesis studies, I think we have to have those because Jim's comment that you only tap failures unfortunately we know from the natural history that some of the failures have negative cultures and some of the successes have positive cultures. So, if we really want to know whether we are eradicating the organism then we have to do double studies, and then your comment, Jim about maybe we don't have to retap, but I think we learned that we didn't have to retap because of the studies where they did retap and told us that things were going to be sterile in 24 or 48 hours. So, then we didn't feel like we had to do it so much anymore.

I think the shift of organisms that Judy raised and that Ron raised in the discussion is really key and we are all very interested to see what is going to happen here. We just submitted a sickle cell center grant and one of the issues was to look at what is going to happen to those patients who are going to be on prophylactic penicillin. They are going to be on azithromycin for their chest syndrome. They are going to get the 23 valent. They are going to get the 7 valent. Who knows what is going to happen in that population, and they may be representing the extreme, but I think we have to know whether we are going to get rid of pneumococci get rid of the resistance problem or whether it is going to emerge in new strains, and this is one of the key ways that we are going to find that out, and I think that that really leads into Dr. Marchant's comment that we really have to decide what is the question that we want to ask because the study that we do is going to be based on the question that we want answered. Do we want to know toxicity in which case it has been pointed out that we need probably lots of patients or do we want to be maintaining change, be advocates for children maintaining a susceptible population and I think that we have to try not to forget that.

So, thank you for, also, bringing that up.

DR. RELLER: Dr.Cross?

DR. CROSS: We spent some time this morning expressing our concern that there was a lack of PKPD data, but I am looking at the guidance for industry and it isn't mentioned there. I am just wondering whether there is the expectation that this is naturally done in the course of things or whether this has to be stated explicitly what the expectations might be especially for different age group populations.

DR. RELLER: Dr. Ebert and then we will come to Dr. Gorman.

DR. EBERT: Obviously as we go around the room it becomes increasingly difficult to add anything that hasn't already been said, but let me just say that at least my ranking of what we have been talking about as far as studies with regard to efficacy would be that at the top of my list would be double tap studies because of the issues that have already been addressed.

Next would be single tap studies at the initiation of therapy to determine what patients are in fact infected and to assess clinical response in those patients but that the clinical response be assessed at the end of therapy as opposed to at day 28.

The third would be where the patients are enrolled based on clinical findings but that taps are done in patients who fail. That would I think probably mimic the clinical scenario most closely, would limit the number of taps and then finally I think probably the lowest on the benefits would be the purely clinical studies and again for the reasons that were mentioned and the higher N that would be required.

The other issue that really gets into I think a little bit of PKPD and also the non-comparative trials would be that as Dr. Dagan mentioned certainly by giving a certain dose and looking at organisms with different MICs we have the ability to do some PKPD analysis, but I think we should, also, consider the possibility at least in early studies of some dose-ranging studies to look not only at a fixed dose but various dose sizes especially because of the variation in the pharmacokinetics in these children as they progress in age.

DR. RELLER: Dr. Gorman?

DR. GORMAN: I think the clinical studies continue to have a role mainly for safety data generation, strictly again because the large numbers you will need to see rare or unusual safety aberrations.

Over the last 3 years I have become a connoisseur of labels which I understand is the FDA's end product. The most interesting label that the FDA has ever written is the label for Ritalin. This is my personal most interesting label which talks about the use of this drug in a total clinical care package, and I would recommend to the FDA to consider in its labeling of agents for otitis media some of the reasons why these agents fail, host factors, age less than 2, day care attendance, siblings so that the expectations for the clinicians as well as for the pharmaceutical manufacturers are there are reasons that these fail and reasons that they do not fail, reasons that they should be more successful which may start to get around some of these other issues that we bat around as we are talking about the microbiology which is so important to this particular group.

I think the entry criteria for any clinical study that deals with microbiological cure rates has to be enriched.

One of my previous mentors did a clinical study where after tapping these patients had a 95 percent bacterial, 95 percent of the ones that he had a 90, 95 percent of the tapped ears had a bacteria identified. Those criteria if that has been implicated in other clinical studies might be adoptable by the FDA as reasons to tap people to do a primary tympanocentesis upon entry into the study.

Children less than age 2 have to be studied, and I think they have to be stratified to be at least 50 percent and perhaps 100 percent because this is the place where the drug is most likely to fail due to the host factors which we have no control over and also the ones in which we have the least PK and PD data and will need to be generated prior to these drugs being used in these populations if they have not already been used. I am sorry, if the drug hasn't been used in these populations previously then the PK and PD data has to be generated for this particular indication.

Placebo controls. I, personally, think placebo controls will never go away, and I hope they never do, Helsinki and whoever else stands between me and my placebo controlled studies.

Even placebo-controlled study of otitis media I think there needs to be very stringent requirement that adequate analgesia and antipyresis be available for children in the placebo arm so that their long-term microbiology can be cured but their short-term pain and suffering can be relieved.

Now, on to disease resistance. I don't share my fellow pediatricians' belief that we should use, I do share that we should use the antibiotics judiciously, but I refuse to sacrifice children on the altar of drug resistance. Less than 75 percent of the antibiotic use in this country is in humans, I am sorry less than 25 percent of the antibiotic use in this country is in humans. The vast majority is in agricultural feed and veterinary medicine. So, now when you go to the antibiotic use in humans, children don't make up the majority of that either. So, if all pediatricians immediately adopted no antibiotics for anything we would probably have a very minimal impact upon the evolutionary pressures on bacteria.

Having said that I support the efforts to limit the use but I won't sacrifice children before chronic bronchitis is not treated routinely, before clinical pneumonia is not treated and before urinary tract infections are not treated, other illnesses that have incredible spontaneous cure rates and no proven clinical efficacy of antibiotics in changing clinical course.

Thank you.

DR. RELLER: Dr. Glode?

DR. GLODE: I think the reason this has been so confounded over the years is that it is virtually impossible to design the perfect study for otitis media. I would post to you even the two-tap issue which sounds great microbiologically is fraught with problems because what about drug A which has no bacteriologic efficacy and the culture is going to be positive when you do it? Of course, when do you do it, 48 hours, 72 hours, 96 hours? Is the fastest working drug the better drug, and if one drug has no efficacy and the next drug has a 2 log kill and the third drug has a 3 log kill unless you are doing quantitative cultures from the middle ear you will not be able to assess that.

So, I just think it is so difficult to actually design the right study for otitis. It may be impossible to do that. I was wondering about the two-tap study or the single-tap study where you then take the organisms that are resistant from the first tap to the agent being studied but by what definition of resistance, a macro standard the PKPD standard that was shown today, but you know I say to the family of that child, "You child had a number of resistant organisms. So your child falls into the category of children that we would like to repeat the tympanocentesis to make sure that this organism which is more resistant is really gone," and maybe you could do that, but I think there will still be a huge argument about when the second tap should be done, and I think that there will potentially be a need for quantitative bacterial cultures which I have no idea if that is possible to do with the small volume of fluid that one gets back.

DR. RELLER: Dr. Burns?

DR.BURNS: I have been impressed today that there seemed to be two groups about which we either are or should be the most worried. The first of these I think has been well addressed, the child under 2 where we see a lot of disease. We see a lot more invasive complications and we see an innate lessened immunity to those organisms that might be invasive such as Streptococcus pneumoniae, and that does concern me that we don't want to neglect that group by any stretch of the imagination.

The other group that seems to me to be important is the group in fact that does have Streptococcus pneumoniae. Certainly any patient who has a positive culture is more worrisome to me than one who doesn't, and it strikes me that it might be reasonable to design trials where we look at specific organisms like say, Streptococcus pneumoniae causes invasive disease; it has an increased risk of resistance, and it is more clinically recognizable according to some of the data that Ed O'Rourke presented.

So, perhaps it would make sense to do that initial tap and then actually conduct the study on those patients who have Streptococcus pneumoniae where we not only expect it to be a higher risk but also expect if we have got a good drug that it would in fact demonstrate better efficacy.

So that would be my thought process to take a very heterogeneous group of microorganisms and then focus on one that we identify as being a specific and real pathogen.

DR. MAXWELL: Not being a pediatrician most of the discussion that I heard makes a lot of sense, but I have a question to the pediatricians. Since it is clear that a good percentage of the children that develop otitis media resolve spontaneously what I want to know is what percentage of those children that resolve spontaneously develop some adverse sequelae, hearing deficits or do they develop anything at all and if they don't develop anything at all then what is the real need to treat everybody if you have not determined that the bug has some potential for doing something somewhere down the line and so I don't know the answer to those questions and would really love to hear because it seems to me that you sort of have a natural history both into your cohorts and now, the way I would look at this I would like to know what the bug is. I would want to go after the bug, treat the bug and then see if I eradicated the bug, but you were saying that even if you identify the bug a kid might resolve spontaneously without an antibiotic and clearly we know that antibiotics are not without risk. So, it would be very difficult for me to answer this question, and I would be ideally bent to having a tympanocentesis at the beginning and at the end but I don't think that that is practical at all, and I would just like to hear the Committee address what happens to those that resolve spontaneously.

What is it that they have?

DR. RELLER: Thank you, Dr.Maxwell. We have multiple, numerous pediatricians on the Committee, and who wants to respond to her query?

Dr. Wald, why don't you start?

DR. WALD: Data have been presented today suggesting that one can identify to some extent the children who are more likely to have one organism than another but it is far from perfect. So, if you are looking at a child who has a bulging tympanic membrane you don't know which one is going to resolve spontaneously and although let us give the benefit of the doubt and say 50 percent do, 50 percent don't, and I think when we prescribe antimicrobial therapy it has a couple of purposes. You know, one is to make children better sooner and I think that isn't something to which we have paid a lot of very careful attention and I think that for those children who resolve spontaneously it is probably true that antibiotics benefit the children who are treated by virtue of the fact that they get a quicker cure than those children who eventually resolve spontaneously and then we hope to prevent complications recognizing that complications are not common but that they rarely exist and that with resistant organisms there is some suggestion that they may be increasing.

DR. RELLER: Your name, please?

DR. ROCHESTER: My name is George Rochester. I ama statistician with the Division of Anti-Infective Drug Products here at FDA, and actually I was associated with this product being presented today.

My comments are not directed to the product specifically in any way but in general I think that our take on the FDA guidance probably needs to be seen as guidance in terms of a minimal standard not in terms of a maximum standard and that in fact I think needs to address the two probably more important issues which are data quality as well as data quantity.

The way I envision that most sponsors seem to approach drug development is in a very simultaneous fashion so that both studies are running at the same time. They get a dash to the 100-meter-yard line without any recognition of real underlying safety concerns there may be that need to be paid attention to.

So the time when the studies are designed and they are sized they are sized with very little information especially when we are talking about the new molecular entity, very little information about what kind of safety end points we are truly interested in and are really truly important. So, probably what I am thinking is that it would be more helpful if we had at least the early bacteriologic studies the studies that are specifically designed to pick up efficacy information early and some preliminary analysis of those studies and then plan the larger probably clinical only studies with more specific interest in safety end points at a later phase so that we plan the studies adequately and we are pairing them to actually detect the things we want to detect.

The other point that I am interested in however that I haven't heard any discussion is that we have some underlying recovery rate for patients that we tap or even in other infectious diseases like community acquired pneumonia. For example, we may do blood cultures for those patients.

There is a certain rate basically even if we think that all the clinical signs or other corroborating evidence suggests a patient may really be infected with a pathogen, we do not always recover that pathogen, just doing one blood culture or just doing one tap and if you do a tap and it came back negative that does not reassure me that in fact that patient may not have a pathogen and that that patient is in fact in any other safer zone than a patient from whom I really recovered a pathogen. I haven't heard any discussion at all about that.

DR. RELLER: We will ask Dr. Marchant and Dr. Patterson if they have anything to say at this point and then we will go next to Dr. Dagan to comment on single, double with failures, etc.

Dr. Marchant?

DR. MARCHANT: I thought I would respond to a couple of things. You asked me what would I do, and I said, "Double tympanocentesis." Ellen, for example, said, "Single tympanocentesis," and gave her reasons.

Single tympanocentesis could be an acceptable way to go with good clinical outcomes but the next step in thinking about it that needs to be addressed is how big a difference are you really going to look for? What difference do you care about; what does the literature on the behavior of the disease tell you that difference is likely to be and then are you prepared to face the sample size required to follow that design?

Again, being somewhat of the devil's advocate or throwing out a little challenge, Dr. Glode said pointed out the difficulties of doing various otitis media studies and very difficult to design the ideal study and so forth, and I would throw down the gauntlet if you will and ask the question are you prepared, for example to just license drugs based on PKPD because it is hard to do a clinical study and have an outcome that is scientific or a data set that is really telling you anything and maybe we should forget doing clinical studies. Are you prepared to do that? I sort of throw that down as the question. Those are the kinds of questions that should be faced if you are going to scientifically approach the question of efficacy and address many of the questions that people have brought up and then the other thing is,Dr. Rochester, I believe, you asked the question of what does it mean when you get a negative culture.

DR. ROCHESTER: Right.

DR. MARCHANT: If you eradicate the organism what does it mean?

DR.ROCHESTER: Well,no, if you did a tap and the tap was negative, you didn't grow a pathogen but that does not necessarily reassure me that there are none at baseline. That doesn't reassure me that the patient did not have a pathogen just because you did not recover.

DR. MARCHANT: You mean at baseline. There is no question that that exists and when you count on electrophoresis looking for the capsule of Pneumococcus as has been done on strong middle effluence Pneumococcus was there. It is not viable and growing now, but it was there in some cases and there is no question that some of those were and there may be a few cases in which there could still be bacteria there that you didn't grow, but we know that the number, an important point is we know that if the bacteria is there and then you don't treat it you don't effectively eradicate over 4 to 6 days that your clinical outcome on average is worse. So, on that side of the equation it is an important thing that you are finding although you haven't proved that one.

DR. RELLER: What I want to do is we will have Dr.Dagan and hear his comments and then we will come back to Dr. Patterson and Dr. Gorman.

DR. DAGAN; Look at those who were culture negative versus culture positive. We actually look at different bugs. Look at interleukins and leukocytes and at clinical response and as a group they had much less suppurative inflammatory response, have much less leukocytes and the respond to nothing very rapidly, at least no difference between the drugs for that group. So, this is for me as a group. Individuals you are right, here we are talking about group. Now, in order to explore that I would just like the Chairman to give me a few minutes. It took me 24 hours to get here, another 24 hours to go back just to hope that you can hear what I have to say.

So, if you can give me a few minutes it would be very, very --

DR. RELLER: Dr. Dagan, we ask you to come to the microphone, please and tell us what you want to tell us.

DR. DAGAN: This is with respect to what was said around the table. I want to start just with some general stuff and I think I was very impressed by some general stuff. I think I was very impressed by the fact that you were the only one to mention that we don't know how to diagnose otitis. So, how can we know how to diagnose cure?

Years ago, at that time SKD did a study with good investigators from all over the world, the United States and Europe, and they wanted to do a study. They put us there, and they brought people from Pittsburgh who showed us a video.

Now, this is the best investigator, recognize that this is the best investigator, showed us videos with otoscopy, whatever. We all failed the test. Every single one, we failed the test, okay? So, we don't know how to diagnose otitis in general.

We miss plus or minus and definitely if you put a needle in, and you get a bug out, plus all the symptoms that is at least the most likely to be acute otitis media.

Now, it goes again for the cure. If you don't know how it is when it is there, you don't know how it is cured. So, the only way is to say that the bug has disappeared.You can add to that leukocyte count, poor inflammatories and things like that that we are looking. So, this is one answer for sure that if you want to be sure that your sample of kids have otitis media, bacterial, take those who had all the things, but after you put the need you, also, get some pus, and it is, also, positive for culture. This is a real case of otitis media in over 90 percent of the cases.

Now, No. 2, the question of, well, I have to read my handwriting which is not that easy.

DR. RELLER: You wanted double tap versus single tap.

DR. DAGAN: I am sorry, but this is all really related because the question is why do we need double tap. This was, also, the question, why do we need double tap. We don't do it in other studies. We have to do it only in studies where we don't know how to do it in a different way. Okay, if you see a big abscess in the skin I know when it is cured, okay, but if you go a little bit wider, I have to do repeated sugar tests for diabetes. I have to do repeated blood pressure tests for hypertension. I have to do repeated cholesterol tests because I don't know how to do it without the test, and I think that if we don't know how to diagnose otitis and how to diagnose improvement the best test is to look at bacteriological eradication. So, that is for me the most important rationale for doing that.

Now, tympanocentesis failures, and here again I apologize not being able to say it gently. The industry needs to bring 600 kids or 500 kids in no time in order to be able to be on time and go against the competitor, and to go with stringent criteria, and you do the best patients you can; you get X amount of dollars per patient if you do it appropriately and then part of doing it appropriately is when you say, "Someone has to tap the child." So, I mean there are so many doctors. Not all of them are good enough, and that is why you see all those 12 year olds or 10 year olds or whatever. They know those kids and then when they see failure they have to do taps, and they don't want to do taps. So, there are less failures because if they say, "Failure" and there is a protocol violation they don't get the money.

I am sorry to say it that strictly, but some of the investigators work for the money, not for the science, and the point is here and this is proved; that is why when you decide to have tap at failures you are going to have older kids. You are not going to have failures, and they don't have to be bothered with the taps, and that is why it is not a good idea. It is very selective cases of taps, and actually if you look at centers always selected if you go to Alejandro Alderman in Pittsburgh he does a lot of taps at failures. You go to other centers, nobody does it. These are not failures. Okay? So, that is a very important issue.

The auditory second tap is you do it or you don't get into the study. That is a different thing. As for talking single or double tympanocentesis I think that we can deduce, we can say that kids that need most the antibiotics, the ones that are under 24 months, etc., if for them antibiotic A is good and you can show that it is at least as good as antibiotic B, it will be good for the reverse of media test, for sure because you just dilute then with more kids and more kids that don't need antibiotics.

So, if FDA comes, and this is a motion for the design, if FDA comes and says, "Okay, you do limited number of double tympanocentesis and then go for single tympanocentesis because we want the safety, etc.," but really instead of doing 500 kids, 6 months to 120 years you do 500 kids under 2. First off, we don't want to see adults if it works under 2 it works adults. Then you are going to have at least results here 10 percent difference under 2 between augmentin and azithromycin as shown by the company. It was not significant. Ten percent would be significant if all the kids that they had were under 2.

So, this is just one way to say that you get the same type of expenses. It costs you the same, but give us patients under 2. We will be happy to extend that to older ages with otitis that are seen, okay? Rather than take above 6 months, mostly above 2 and extend it down, this is for me more dangerous. So, that is the motion in terms of if you do that single, take it only for the young children.

Then we talk about safety and this is not related but this is related to efficacy, nasopharyngeal changes are part of the safety and it is not recognized by the FDA as part of the safety because this is actually more safety than vomiting. Vomiting is tolerability. It is not safety.

Safety is an issue where by treating a child you make his situation or his surrounding situation more dangerous in the future, to be complicated and to have problems, and by changing with drugs these florae and by proposing to that child for the next infection a more resistant flora is a question of safety exactly as now in the vaccine they recognize, WHO started to recognize that changes in carriage by the pneumococcal conjugate vaccine is a safety issue because it will determine the next epidemic or the next infection with Pneumococcus.

So, I think that now some companies have started to do it, and I think that one of the things to be done together with the double tympanocentesis is to look at efficacy or the dynamics of the nasopharynx because it follows the same type of PKPD but in a different site and you have to look at that because some drugs are making some issues, to mention for them junatopin(?) sulfa with the multi resistance, junatopin sulfa increases very impressively the penicillin resistance carriage which means next problem. So, that is another thing.

Now, talking about double tympano and single tympano without talking about what will be your end point is also inappropriate in my opinion. I think that the term "test of cure" for 28 days should be taken out because it is not test of cure. I think this is already proved, and if it is taken out then the FDA has to decide what is a test of cure.

Now, you only can offer with antibiotics what you can offer with antibiotics. The child is prone to infection. The child is going to get whatever they have in the nasopharynx next time and it is going to do more infections. So, I think the test of cure is at the end when you finish giving antibiotics that is what you can achieve with the antibiotics. The rest is not, but you can still follow it, but let us call it test of cure at the end of treatment and let us call recurrence clinical recurrence or clinical relapse rate which is not a test of cure. So, that is another suggestion for that specific issue, and then again, the question was should we go only for Pneumococcus testing. This was double tympano. This was suggestion, okay, and we had a talk today how Hemophilus is so important and H. flu, really doesn't, I mean how Pneumococcus, H. flu really doesn't do much. Well, that did not stand with the data we saw from Pfizer.

From Pfizer they had the highest clinical failure rate with Hemophilus, not with Pneumococcus. So, if Hemophilus is not important then it should go away anyhow. So, the highest clinical failure rate is with Pneumococcus because with Pneumococcus it is resistant, okay, and it goes on and on. Hemophilus is maybe less fulminant than Pneumococcus, but it is there. If you don't respond, you do have clinical problems, and we struggle with now 8000 tympanocentesis results in our, all of us, all the bugs in our hands. we struggle to how do we describe the clinical differences between H. flu and pneumococcal patients and we don't really know how to do it. We don't have any clue how to make it clear that one is different than the other.

So, maybe when we have 20 cases it is easy. When you have 2000 cases you start to see that what you find is this is exactly coming to the confidence interval. When you have 2000 you realize that the 20 cases were purely chance that you have some differences and that is why you have so much debate.

So, definitely H. flu is not a benign bug and therefore if you do double tympanocentesis you probably have to do it for all bugs or otherwise it is not appropriate unless you have specific questions for a specific drug or Pneumococcus and then you need to do it only on Pneumococcus and so I think that I covered what I wanted to say in terms of response, and the point is again I don't see how if you want to get the right answers, I don't see how you can go without doing double tympanocentesis. You have no answers that can be answered by the tympanocentesis. So, you did additional studies, but if you don't do double tympanocentesis in my opinion some of the real important answers are not going to be there, and we are going to get again and again things regarding the inappropriate or semi-appropriate recommendation.

The last point that is related to this is how ethical is it to do all these things, and ethics is really something very relative and it is perception and it is tradition. You do, I mean Pfizer had to present here I think altogether 2000 or 3000 cases. Each kid submitted to a study had a lot of uncomfortable situations. If you can make 10 percent of those cases double tympanocentesis you save a lot of uncomfortable situations to 800. That is something, also, that has to be counted, but in addition Dr.Gorman is ready to accept pharmacokinetic studies.

Now, the tympanocentesis in my center is done by ENT under microscopy and it is 1 second, and you are finished. Blood is taken by my intern and they dig like half an hour before they find the vein. You know how interns do it. They have to learn how to do it. You can't be born with the knowledge how to do it, okay? So, repeated blood tests is so much more painful to the children, but this is always accepted because you know repeated blood tests plus urine and all the stuff that you submitted kids for pharmacokinetics, this is a real torture. I never agree to do pharmacokinetic studies because for many children who stand before my eyes and for hours there and we can't explain to the mother what benefit does the child get from that.

However, the mothers like when we do the double tympanocentesis studies because first it is our standard of care to do one, but in addition to get the best treatment by the best group, to get the attention if they come after 3 days with a failure we already know what is the bug. With multi-resistance this is more complicated now because with multi-resistance stuff we can really change the drug after 3 days knowing what the child had before. We have documented in many children super infection with new bugs knowing that they did not fail. They just had a real problem, and we identified with this the children with severe problem where you don't even finish one study and they get to another infection and this kid now gets followed by a physician.

Talk about safety, we are doing routinely in our center 2000 tympanocenteses a year. Out of this about 15 percent or so are part of the study. The others are not.

In the last 6 years we had 12,000 tympanocenteses and about 2000 in studies. We didn't see even one complication. However, we see every year several complications of acute otitis media that are severe. So, I don't agree 100 percent with what you say. They come with severe mastoiditis. They go to surgery. We get how they call it, synosiventroposis(?). We get two patients in the last 5 years with abscesses, intracranial, all from acute otitis media that did not respond appropriately. So far tympanocentesis is less dangerous for us than inappropriately treated otitis media.

So, I think that again when you measure all those things if you are used to you get different proportion and in my opinion the most of the problem if you submit the whole population to treatment that you don't know what it is it is so much ethical versus doing something in good controlled hands when you see what the child has and you can correct it at a given moment.

My ethics committee thought the last time that I submitted them a study versus azithro that it was unethical to do a study with azithro because we proved it doesn't work very well before, and they agreed finally to do it because it is standard of care still in Israel in other cities, not in our city. So, sometimes ethics is very much a relative thing.

DR. RELLER: Thank you, Dr. Dagan.

Other comments or questions not posed?

Dr. Patterson?

DR.PATTERSON: Yes, I would agree with those who think that there should be more representation of severe disease, particularly as manifested by pain and fever and that would be more associated with pneumococcal infection based on what we know about that compared with H. flu and I think that would be important for looking at rates of DRSP and also suppurative complications and so forth.

I, also, think more young children should be included, at least 60 percent or so. I don't think that those over 2 should be excluded because based on what we heard today there may be some difference in etiology.

Perhaps H. flu is more common now in young children. Maybe Pneumococcus is more common in older children based on some of the things we have heard today and I think because Pneumococcus can cause suppurative complications we don't want to ignore that.

I think the single tap is quite important for bacterial etiology and epidemiology and so a substantial part of the application should look at bacteriology and I think the double tap is difficult to justify in many centers where it is not a standard of care and those where it is I think it is very helpful to have that knowledge, but I don't think it should be required for all the data that is submitted.

The clinical only studies I think may have a role in safety and tolerability. Somebody mentioned today about you know, did vertigo contribute to the vomiting and so there may be something about the particular disease state that could have a role in safety and tolerability. So, I think that could have a role in that.

I think there should be an emphasis on analysis by previous antibiotics within 1 to 3 months and, also, in day care, particularly in terms of the outcome in bacterial etiology and in the clinical only study and single tap study I think it would be important to have a substantial amount of failures with follow-up tap, at least 70 percent or so, and I think a placebo study would be difficult to justify in IRB for those less than 2 years of age.

DR. RELLER: Thank you, Dr. Patterson.

I would like to comment on the numbers by organisms that were presented. My recollection is that the genesis of some of those numbers was when a particular sponsor was interested in an indication for a specific resistance mechanism or a subgroup of organisms that were resistant so that one could not get an encompassing claim that included resistant organisms where the efficacy was buried in the larger numbers as opposed to having to have for example, with tympanocentesis and even double tympanocentesis documented resistant organisms to extend to include specifically resistant pneumococci as an example.

That every study should have some of the numbers mentioned may be difficult as the ground shifts in terms of the distribution of organisms and again, I think that the getting too hidebound with specific numbers and distribution of organisms isn't largely obviated if one has the emphasis on the efficacy studies, the smaller numbers of well-characterized patients with at least initial, if not a high proportion with double tympanocentesis along the lines that Dr. Marchant has emphasized.

Other comments, and we want to make sure, Dr. Soreth and Dr. Alexander that if something has not come out already that you raise it so that we can cover it before concluding our comments.

Dr. Jacobs had his hand up and in the good tradition of this Committee people have been very forthright and outspoken which is to the good, and there is clearly plenty of fodder for changes in the next draft.

Dr. Jacobs?

DR. JACOBS: Thanks, Dr. Reller. I would just like to emphasize the comment you made about distribution of organisms, and I think you need to apply the same standards to each individual organism based on the principles that you have heard this morning from Dr. Marchant and others that you can based on the known natural history of Strep. pneumo. You need a smaller number of Strep. pneumo than H. flu to be able to tell what goes on in a double tap study, and when you have subsets of organisms as you mentioned with resistance you again need to have each subset having the same, having a number that is statistically valid, and I think coming up with arbitrary numbers saying that you need 25 Strep. pneumo, 25 H. flu for the drug to get an indication is not applying any of the principles we have heard about. You need to base it on what type of study it is, what age group and so on and then come up with a projection that is statistically meaningful.

DR. RELLER: Thank you.

Drs. Wald and Leggett?

DR. WALD: That sort of troubles me because it seems to me how many resistant organisms did we hear about today, a dozen resistant S. pneumoniae and a lot of, out of what 100 and they did about 166 bacteriologically proven cases? I don't think we can make a judgment on 12 cases, some of which are split between 1 day and others are getting 3 days. I can't make any judgment about that. I do think we have to have minimum numbers of resistant cases because if a study is done in an area where resistance isn't very high but the drug is going to be used nationwide where the numbers are 35 percent or 40 percent or 50 percent I don't think that we can say that the drug is approved or a good treatment for that particular indication without having had specific experiences.

DR. DAGAN: If you go for otitis media as a design as we know that about two-thirds of the beta-lactam unsusceptible bugs are, also, macrolide non-susceptible bugs, then by having, by the approach of non-responsive otitis media you are going to increase it. You are going to have a little bit more negatives, but those who are going to be positive, are going to be more enriched with the resistance and this is based not on my assumption but on what we have as bacteriology from our studies on non-responsive otitis media, especially that first drug of choice is usually amoxi and the second one is azithro. So, when you go to non-responsive you are going to really get those two to augment the resistance.

DR. WALD: Maybe then we have to say rather than prescribing precise numbers we have to say that it has to in some way mirror what the national figures are since when we license or approve a drug that is the way it is used.

DR. DAGAN: Then be careful because the national numbers that are given to you are of all ages and of all types of bugs.

DR. WALD: They are specific.

DR. RELLER: Actually I know Dr. Wald and I if we vetted this adequately would be in agreement. The concept is that clearly there is a number of organisms that would be required and statisticians would help us if we want a specific plan for a resistant organism, but one of the things that has troubled me repeatedly and it came up in the discussion this morning is that from a clinical standpoint an approved drug will be used in the population at large and if the studies on which we have adequate objective information are done in a population that does not mirror the one that is going to be treated when a drug is approved then we have a problem, and that came out this morning in that we had a disproportionate number of patients over age 2 and had inadequate in my view parameters of resistant organisms where we actually knew what the organism was to be comfortable that we were dealing with a distribution of organisms that is what the real world is in otitis media that needs to be treated currently and the numbers where one knows what was going on are quite small.

Dr. Wald?

DR. WALD: One more comment about the double taps because I think there are very few institutions where practically speaking that is going to be done, and again, I can talk about the discomforts of venipuncture, and you can do other terrible things to children but if a child has no symptoms referable to their ear except some residual fluid then again I think that it is hard to justify that procedure and I will just let it go there.

DR. RELLER: Thanks. Dr. Leggett or Dr.Christie?

DR. LEGGETT: A question was made about placebo controlled trials and we haven't really talked about that very much, but one thought just crossed my mind for the kids who are over 2 or the ones that are most likely going to have these more viral pathogens and perhaps we are less worried about them, one of the ways that we could capture the placebo cure rate or kids over 2, say, would be to do the placebo-controlled trial with an early escape so that if symptoms persisted at day 3 or day 4 we then proceed to that single tap. We then accumulate that as our clinical only trial group for the safety data so that you end up doing a lot fewer of these tympanocenteses and following people along, and I wondered what the pediatricians thought about something like that. That is just spur of the moment.

DR. MAXWELL: That strategy was actually used by Dutch investigators, but what you try is you try to go with not treating a bunch of people and kids and letting them hurt for a while instead of doing tympanocentesis which, also, hurts, okay. I mean you are just trading it off, but that strategy has been used by as I said, the group in Holland and I am not even sure that that would be an easy sell to patients in North America, saying, "Okay, you want to enroll in this clinical trial, and the standard of care is that we treat with antibiotics, but when a child is still symptomatic for 2 days then we will do something?" I think it is a hard sell.

DR. RELLER: Dr.Christie and then Dr. Gorman?

DR. CHRISTIE-SAMUELS: This may be a little bit controversial, but since we are collecting so much information if they couldn't just look at the type of H. flu with regard to Hemophilus influenzae type B and I don't think it would be much of a problem to just collect vaccine efficacy, some information with regard to vaccine coverage, HiB vaccine and pneumococcal vaccine, please?

DR. RELLER: Dr. Gorman and Cross?

DR. GORMAN: To revisit the tympanocentesis, Dr. Dagan presents the local ethics issue which I find very convincing and if in my community there was a group of pediatricians that routinely tympanocentesis some relatively reasonable fraction of their children in their routine clinical care of otitis media I don't think our local IRB would object.

Dr. Dagan was not trying to imply that at all. What I am trying to say is since it is part of your routine clinical care for your patients we would have very little difficulty with you incorporating it into a study, and I hate to sound like a Luddite when I say that our IRB will not approve things like this because it is not the purpose. It is just where is the risk/benefit in our clinical practice for the children that surround us, and last time I was able to listen to another group in another state in the United States that does routine tympanocentesis and if you are going to take routine tympanocentesis that would be an approvable study and when you said that it was local you were absolutely correct. It is a local decision.

The second question I wanted to ask, and I can ask this because I am a simple country pediatrician. Why are there no surrogate markers for infectious diseases that will help you in this case? I have heard so much about amplification of proteins for both DNA and capsid proteins. Why can't you once you know what the organism is after the first tympanocentesis, look for that organism using an amplification technique of a protein that is specific in the blood to follow its cure rate?

DR. RELLER: Dr.Cross has had his hand up.

DR. CROSS: Actually this addresses in part a comment you made earlier about the perhaps clinical outcomes correlating with the MICs and one piece of data in their Slide No. 75 was at least very intriguing that 18 of 25 of the Strep. pneumo which were either intermediate or resistant to penicillin were listed as clinical cures. i mean something like this would argue that we cannot simply rely on in vitro microbiologic data even based on the first isolate, on the first tympanocentesis but would seem to argue that in this situation we would love to know whether or not the organism was there. So, these very limited data would argue strongly for need for tympanocentesis because otherwise you cannot evaluate this small limited data.

DR. RELLER: Thanks.

Dr. Dagan, you had --

DR. DAGAN; I would answer for the DNA stuff.

DR. RELLER: Please?

DR. DAGAN: Actually we did that study in a different way but it answered this question, and it was published years ago in JCM, Journal of Clinical Microbiology. We wanted to see if we can use a PCR for Pneumococcus in the blood to diagnose pneumococcal infections, and what we proved is the following. If you want to do a PCR for Bacillus anthracis, it is a good idea because you shouldn't carry that in your nose or anywhere else but if you take and you do PCR for something that kids carry in their nose continuously and the majority carry in the nose a little bit of DNA is enough to get in the blood and you detect it, and what we showed is we have so many true positives in terms of the PCR but false positives in terms of infection and it was really proportionate to the age and how much you carried, and if we detected carriage we could detect up to 50 percent in the blood. So, PCR is too sensitive for a bug that is routinely carried in the nasopharynx and for what you said I would predict that 18 out of 27 or so would be clinical failures only because we know that of the bacteriological failures you only have 40 percent that are clinical failures. So, I would predict that the majority will be bacteriological failure but only 40 percent would be clinical failure and I would come exactly with that number by the way that you mentioned.

DR. RELLER: Dr. Gorman?

DR. GORMAN: I would, before you sit down Dr. Dagan it strikes me that I can understand screening blood looking for pneumococcus would be very inefficient. However, once you know the specific Pneumococcus couldn't you follow titers looking at your amplification as some sort of surrogate marker for cure watching titers fall during the course of the antibiotic therapy?

DR. DAGAN: Yes and no. You could follow the titers, but you have a lot of assumptions there that you didn't prove, that you have to prove, and you didn't prove. When is the DNA the broadest proportional to the amount of the bacteria that are in the ear and not in the nose? Now, some of the drugs are excellent for ears but not for the nasopharynx, for example, ceftriaxone. Some of them are excellent for the nose but not always for the ears or at least azithromycin is excellent to eradicate susceptible organisms both from the nose and from the ears in terms of Pneumococcus.

So, the point is that you are following the blood, something that is spilled off in another place. So, it could be very good for bacteremia. It could be very good in meningitis; it is easy for meningitis, but you don't know how to measure the amount in respiratory infection of DNA that gets accidentally into the blood.

DR. RELLER: Dr. Marchant?

DR. MARCHANT: There have been a number of issues raised about ethics and IRBs and tympanocentesis, etc., and because I know these issues will be discussed at great length in the future, I think there are some provocative questions to raise, and that is we should, also, ask other ethical questions, other than focusing on the individual child in an IRB situation, but, also, the question of is it ethical to do a study which is not adequate to address the question of efficacy; is it ethical to license, market or prescribe drugs without really knowing how efficacious they are? Those are also ethical questions that we should face

DR. RELLER: Thank you.

Dr.Wald?

DR. WALD: One last thing, I would say that double tap studies, which I sort of lost my train of thought before, and that is I think in some sense the most valuable studies are those that mimic clinical care, and I absolutely concur that a diagnostic tympanocentesis is essential in the bulk of patients who we are trying to learn about.

I think once we start tampering with that TM on multiple occasions we are influencing the outcome, and outcome is one of the things we are trying to judge, and so, I will just sort of keep that in mind that by tampering with the system many times that it makes a judgment about clinical outcome a little bit more difficult, and in the end clinical outcome is still an important thing. I understand bacteriologic outcome, but the things that Dr. Glode said before, you know, when is exactly the best time to tap; is an earlier sterilization substantially different than a sterilization on the fourth or fifth day, and I don't think we know the answers to those questions.

DR. RELLER: Dr. Marchant?

DR. MARCHANT: In response to the issues you raised, Ellen I would say that bacteriologic outcome is only important because the studies that have been done have been shown that it correlates with clinical outcome. If it were not that it would not be an important outcome, but rather draw your attention that it is a surrogate that if you study enough numbers it does predict a clinical response, but that is one of the reasons it is important. It would be unimportant totally if that were not the case, but that is what the data show.

DR. RELLER: Dr. Soreth, have you heard enough?

DR. SORETH: I think we have heard enough at this point, Dr. Reller, and I guess we will take this to another public forum wherein we focus specifically on the guidance I presume at that point with votes making it clear how the guidance should go.

At this point I would say that there is not unanimity but consensus on continuing in some measure having an assessment of clinical response although we have to talk specifics in terms of the size of the trial and the kind of difference that we are trying to detect between a test drug and a control drug.

With regard to microbiological studies, again, I would say that there is more consensus on a single tap study, tympanocentesis and baseline and still a fair amount of controversy or lack of agreement with regard to an actual requirement for double-tap studies.

Is that a fair assessment?

DR. RELLER: What I have heard today on this issue is that the agencies and alternately the sponsors and most importantly patients and practitioners would be better served by having studies of higher quality where we know what one is dealing with microbiologically and clinically and those two are complementary and that one is kidding one's self if one can make any honest with realistic numbers, any honest assessment of efficacy in the absence to tympanocentesis, and that the clinical studies have their complementary role and are important in adequate numbers especially when there are no data from other patient populations for assessing safety of a compound and that the pharmacodynamic information is of interest, but is not a substitute for clinical trials and microbiology and in patient populations where drugs may not act the same at different age groups that it is important to actually have the pharmacokinetic, pharmacodynamic data from those age groups to properly assess safety, as well as whether things all fit together in terms of efficacy, and I think that summarizes what we have concluded or the consensus view that we discussed today.

Lastly, I would say that in this issue of a study done well that adequate numbers is the best thing to do in the public interest. Quite honestly the numbers required to do that that are dependent upon design we need expert help from our statistical colleagues to pin those numbers down and those numbers will be different depending on the quality of the study done and in general the better the study the smaller the numbers required to show the targeted differences which are a judgment call.

Have I missed anything, Dr. Marchant?

DR. MARCHANT: No, i think you have covered it pretty well, and I would just like to reinforce your last point. The clinical outcome of persistent symptoms during therapy, if you look in the literature from single studies, single centers you come up with a much lower sample size estimate of what is required to demonstrate that outcome than if you take the meta analysis that Dr. Rosenfeld used pooling multiple studies from multiple time points and multiple centers. It is just not as precise and as soon as you spread the whole game around to multiple different studies or investigators you are going to get less precision, and you are going to have less ability to show anything.

So, that is another wrinkle in the mix here of the difficulty of doing these studies, but a difficulty that you could overcome if you really want to.

DR. RELLER: Thank you, all for your contributions. The meeting is adjourned.

(Thereupon, at 4:50 p.m., the meeting was adjourned.)