FOOD AND DRUG ADMINISTRATION

+ + + + +

CENTER FOR DRUG EVALUATION AND RESEARCH

+ + + + +

ANTI-INFECTIVE DRUGS ADVISORY COMMITTEE

+ + + + +

MEETING

+ + + + +

Thursday,

April 26, 2001

The Committee met at 8:00 a.m. in the Versailles Rooms of the Holiday Inn Bethesda, 8120 Wisconsin Avenue, Bethesda, Maryland, Dr. L. Barth Reller, Chairman, presiding.

L. Barth Reller, M.D. Chairman

Gordon L. Archer, M.D. Member

Dave Battinelli, M.D. Invited Guest

David Bell, M.D. Invited Guest

Joan P. Chesney, M.D. Member

Celia D.C. Christie-Samuels, M.D.,

M.P.H. Member

Alan S. Cross, M.D. Member

Barry Davis, Ph.D. Invited Guest

Steve Ebert, Pharm.D. Consumer Representative

Zachary D. Goodman, M.D. Invited Guest

Ralph Lazzara, M.D. Invited Guest

William M. Lee, M.D. Invited Guest

James E. Leggett, Jr., M.D. Member

Arthur Moss, M.D. Invited Guest

Barbara E. Murray, M.D. Member

Jeremy Ruskin, M.D. Invited Guest

David E. Soper, M.D. Member

Ciro Sumaya, M.D. Invited Guest

Ellen R. Wald, M.D. Member

Thomas H. Perez, M.P.H., R.Ph. Executive Secretary

I N D E X

by Dr. Reller 3

by Thomas Perez, M.P.H. 5

by Dr. Dianne Murphy 8

by Dr. Janice Soreth 12

by Dr. Jeremy Ruskin 17

by Dr. Douglas Shaffer 42

Introduction - Dr. Mindell Seidlin 55

Microbiology - Dr. Andre Bryskier 61

Human Pharacology - Dr. Vijay Bhargava 68

Clinical Efficacy & Safety - Dr. Bruno Leroy 77

ECG Analysis - Dr. Claude Benedict 97

Conclusions - Dr. Mindell Seidlin 112

Question and Answer Session

Clinical Efficacy - Dr. George Rochester 148

Resistant S. Pneumoniae - Dr. Alma Davidson 158

General Safety Profile - Dr. David Ross 165

Hepatic Effects - Dr. Edward Cox 183

Drug-Induced Liver Disease -

Dr. Zachary Goodman 196

FDA Presentation Summary - Dr. David Ross 209

Question and Answer Session 218

Open Public Hearing 255

Discussion of Questions and Vote 259

Adjourn 308

P-R-O-C-E-E-D-I-N-G-S

(8:00 a.m.)

DR. RELLER: Would everyone, please take their seats so we can begin our meeting? Good morning. I'm Dr. Barth Reller from Duke University Medical Center. I'd like to open today's meeting of the Anti-Infective Advisory Committee. We'll begin with introductions going around the table. To my right, Dr. Dianne Murphy.

DR. MURPHY: I'm Dr. Dianne Murphy. I'm the office director for ODE IV which has the antimicrobials.

DR. SORETH: Good morning, my name is Janice Soreth. I'm the acting director for the Division of Anti-Infectives with FDA.

DR. KORVICK: Joyce Korvick, acting team leader Medical Group.

DR. ROSS: David Ross, Medical Team Leader, Division of Anti-Infective Drug Products.

DR. SOPER: David Soper, Medical University of South Carolina in Charleston.

DR. CHRISTIE: Celia Christie from Pediatrics and Infectious Diseases University of the West Indies, Kingston, Jamica.

DR. WALD: Ellen Wald, University of Pittsburgh, School of Medicine, Pediatric Infectious Diseases.

DR. ARCHER: Gordon Archer, Virginia Commonwealth University in Richmond, Virginia.

DR. CHESNEY: Joan Chesney, the University of Tennessee in Memphis, Division of Infectious Diseases and Pediatrics and St. Jude's Children's Research Hospital.

DR. MURRAY: Barbara Murray, University of Texas Medical School in Houston, Adult Infectious Diseases.

MR. PEREZ: Tom Perez, Executive Secretary for the Anti-Infective Drugs.

DR. EBERT: Steve Ebert, Professor of Pharmacy, University of Wisconsin and Meriter Hospital, Madison.

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

DR. DAVIS: Barry Davis, University of Texas, School of Public Health in Houston.

DR. BELL: David Bell, assistant to the director for Antimicrobial Resistance, the National Center for Infectious Diseases at CDC.

DR. GOODMAN: Zachary Goodman, Hepatopathology Division of the Armed Forces Institute of Pathology.

DR. LAZZARA: Ralph Lazzara, University of Oklahoma, Cardiology.

DR. RUSKIN: Jeremy Ruskin, Cardiology, Massachusetts General Hospital, Boston.

DR. BATTINELLI: Dave Battinelli, General Internal Medicine, Boston University School of Medicine.

DR. LEE: Will Lee, UT Southwestern in Dallas, Hepatology.

DR. SUMAYA: Ciro Sumaya, School of Rural Public Health, Texas A & M University Health Science Center.

DR. RELLER: Thank you. We're delighted to have not only our members but guest experts who will be presenting at the meeting and participating in the discussions today. We'll now have our opening statement by Mr. Tom Perez, our Executive Secretary.

MR. PEREZ: Good morning. The following announcement addresses conflict of interest with regard to this meeting and is made a 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 terms 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 with 18 U.S.C. 208(B) full waivers have been granted to Drs. Gordon Archer, Barbara Murray and Ellen Wald. A copy of these waiver statements may be obtained by submitting a written request to the agency's Freedom of Information Office, Room 12A30, The Park Lawn Building. In addition we would like to disclose for the record that Drs. Gordon Archer, Steven Ebert and James Leggett have interests which do not constitute a financial interest within the meaning of 18 U.S.C. 208(A) but which could create the appearance of a conflict.

The Agency has determined notwithstanding these interests that the interests of the government in their participation outweighs the concern that the integrity of the agency's programs and operations may be questioned.

With respect to FDA's invited guests, there are reported interests which we believe should be made public to allow the participants to objectively evaluate their comments. Dr. Jeremy Ruskin would like to disclose that he and his spouse own stock in Pfizer and Merck. He also has received consultant fees from Pfizer and Roche and has lectured for Eli Lilly.

Dr. Barry Davis would like to disclose that he is the PI on an unrelated grant from Bristol-Myer Squibb and he is a consultant on unrelated matters for the steering committee for Merck and Glaxo Smith Kline. He is also consultant on unrelated matters for Pharmacia Corp. Data and Safety Monitoring Board and Pfizer's Data and Safety Monitoring Board for Alliance.

Dr. William Lee would like to disclose that he is a researcher for Glaxo Smith Kline, Bristol Myer Squibb, Roche and Schering Plough. He also has lectured for Roche.

Dr. Arthur Moss would like to report that he has and will consult for Abbott Labs on unrelated matters. He has also consulted on unrelated matters for Eli Lilly.

Dr. Ciro Sumaya would like to disclose that he owns stock in Glaxo Smith Kline and Pfizer. Thank you.

One more thing, actually two more things. In the event that the discussions involve any other products or firms not already on the agenda for with an FDA participant has a financial interest, the participants are aware of the need to exclude themselves from such involvement under exclusion will be noted for the record. In addition, it has come to our attention that some of the materials that FDA typically puts up on the website were not made available till late yesterday evening. They typically go out the day before, hopefully, you know, as much as 24 hours before. We apologize for any inconvenience this may have caused anyone.

We have brought several copies of the background materials with us. They are available for anyone wishing to review them at the FDA table out front in the lobby. Again, our apologizes for any inconvenience this may have caused anyone. Thank you. DR. RELLER: Thank you, Tom. We shall next have opening remarks by Dr. Diane Murphy, who is the Director of the Office of Drug Evaluation IV.

DR. MURPHY: I want to again thank our committee members, our guests, sponsors and everybody who is here to listen to these deliberations because they -- the discussions today will involve topics that this committee in particular has dealt with a number of times. But, it always gets a little bit more precise when we have a product instead of a general topic. We have four issues that I think you will hear themes brought up that, again, don't just relate to this product. One has to do with the development of drugs to treat infections that are caused by resistant organisms and the need to make sure we have a pipeline of products that are being developed and yet at the same time that these products developed are used in a way that does not undermine our ongoing use of them and some people calls this prudent use.

For us what it relates to or where it is particularly important is because the patients who have resistant organisms will be small in number compared to the broad development plan that companies will have, how do we deal with these issues of prudent use of antibiotics when we have broad development plans. And there's a benefit to a broad development plan because it gives us a better picture of what the adverse event profile may be. And that's a benefit to everybody because as you know, the ability to detect signals, a particularly rare signal, is limited by sometimes our preclinical cause and certainly the size of them.

And we also know that our trials are designed in such a way that we often exclude populations because, again, they can only be so large, you need to be able to clearly delineate what's going on in a trial. However, once a product is approved, then that product is used by a population that may not have been included in the studies and appropriate use meaning for the right indication. I'm not talking about off-label prescribing. And that brings us to the issue of enhanced risk as far as what are the -- what are the characteristics of the population that may be taking products that are developed for a more general use.

And I think all of us understand that many of our older populations are on numerous products that may have interactions. And this has been an increasing issue for all the sponsors and for FDA and that happens to be Cytochrome P450 and the interplay between products that rely on these systems and their interactions with other drugs.

Many of the sponsors have attempted to address specific interactions and that's become an increasing part of our drug development request. However, again, populations that will be using this and other products often take multiple products of which it would be very difficult to ask a sponsor to develop every possible combination that a patient may be taking. So we have to look at the potential risks when we're looking at our risk benefit ratio, which is what we always do.

And the last thing that is another focus of this discussion and many of you have heard these discussions previously but we want to have some background information presented to you this morning and that's what we'll call the silent serious adverse events particularly if they're rare but how do we deal with adverse events such as QT and hepatic toxicity. You don't break out in a rash. You know, you don't -- you may not know you're having that complication. How do we deal with these in making our availability of products appropriate and making sure that we have come to the correct risk benefit assessment in our approval and in our availability of these products.

These are complex issues and we really do thank you for being here, for putting time and effort into preparing for this discussion because many of the things that we move forward on are evolving and your discussions are important in helping us expand how we think and how sponsors think about the development of products in this area. Thank you.

DR. RELLER: Thank you, Dr. Murphy. Next we'll have Dr. Janice Soreth, who is the acting Director of the Division of Anti-Infective Drug Products.

DR. SORETH: Good morning. Before I speak a little bit to the organization of presentation for our meeting today, I wanted to first make a quick apology to members of the advisory committee who may have experienced more than the usual difficulty in making travel arrangements. I think recently some of the rules may have changed, if not, in fact, the contract and it came however late, that there were, well, let's just say glitches. We promise you in the future we'll work very closely with you so that this experience is improved upon.

I wanted to talk a little bit today about the charge that is put before you as our Anti-Infective Advisory Committee and that charge is essentially to weigh in on the evidence that you'll see presented to you today by both our colleagues at Aventis as well as here at FDA on whether or not substantial evidence has been submitted to speak to both the efficacy of Ketek (Telithromycin) as well as safety.

In order to strike that balance, we have a couple of issues that you'll hear more about and that this committee has heard quite a lot about in the past decade or so for we have had no fewer than a half a dozen meetings product specific for antimicrobial agents that were requesting indications that included resistant organisms as well as at least three or four meetings in the past five years that have been non-product specific, speaking to the appropriate development scheme or schema for drugs seeking resistant claims and in the same breath trying to strike a balance with that, again, in the interest of the public health for what the prudent use of those antimicrobials would be.

Next slide. In addition to hearing a discussion of resistance previously as well as today, we will also be focusing on safety. I think it a fair statement that the independent FDA analyses that have been conducted on Telithromycin data are in agreement in general with those conducted by Aventis. We know in general that with regard to safety, clinical trials are not really powered to address those events which are uncommon or rare.

Instead clinical trials are powered in the realm of antimicrobial research to address efficacy issues. Should safety concerns arise anywhere in a development program, I think there are basically three options; stop, and that's usually what happens if the adverse events notes are of a very serious nature, like death. Short of that, when we're not dealing with a body count, but either clinical or laboratory adverse events, there are additional options and those include additional studies conducted premarketing that would shed further light on the elucidation of risk management to understand what the safety profile of a product would be for its intended use.

At other times, given a particular development program, one may choose to press on given the data that's been developed and try to put a drug on the market and in the setting of post-marketing have inquiry either in the setting of additional trials or active surveillance so to speak, to understand the use of the product, the safety of the product in the setting of general use. We know that our Phase III trials attempt to tell us that information but at the end of the day we also know that the experience with concomitant drug use and co-morbidities is sometimes a small experience in Phase III trials, even if the Phase III trial is relatively large. The devil is in the details.

Exposure for a product that would target respiratory infections such as you'll hear today is one with potentially very, very broad use with millions of prescriptions potentially written in a single year. Next slide.

What are some of the lessons learned at the agency with regard to safety? I would offer that we have learned through experiences with Mibefradil or Posicor, Terfenadine, Durac, et cetera, that once prescribing patterns are established for a drug, it becomes very, very difficult to change those patterns. Next slide, please.

So to that end, we need to do it right and we need to do it right the first time because again, I think that is in the best interests of the public health and it is also in the best interest of what I think of as the marketing half-life of a product. Next slide, please.

When in need, steal from a better writer. Dr. Temple published a paper in JAMA in 1999 that spoke to alternative means of looking at and gathering safety information for new drug development and I quote from that paper,

and by such it referred to a previous point of modest risk, that which was not necessarily common, somewhere between common -- certainly less than common to rare,

And I would submit to you that for many, if not most of the antimicrobial products that have come before this committee and come before us in the agency, we typically are looking at broad use. Only relatively recently in the setting of development of drugs for resistant bacteria have we begun to talk a lot about prudent use equalling less than very broad use, again, in order to try to preserve those products. Next slide.

And finally again from the paper in JAMA authored by Dr. Temple, "If a question is important", and I would add if a question is important particularly with regard to safety, "answering it may well be worth the cost and the effort".

We're going to hear this morning presentations that are essentially a QT Primer 101 by Dr. Jeremy Ruskin followed by an FDA presentation that will take us through a tour of some of the post-marketing data speaking to Torsades de Pointes and finally wrapping up the morning will be presentations on Ketek from Aventis. We'll break for lunch and then come back and hear presentations from FDA underlining where we, I think, see eye to eye on the efficacy data in general but perhaps not clearly as much on safety and finally wrap up with a discussion and a vote by the committee.

I thank you for your attention and turn the podium back to Dr. Reller.

DR. RELLER: We now look forward to the presentation of Dr. Jeremy Ruskin.

PRESENTATION OF DR. JEREMY RUSKIN

DR. RUSKIN: Good morning. I was asked to give a very basic introduction to this issue of QT prolongation and Torsades and that's what I will attempt to do but let me apologize in advance to the many people in this room who are already well familiar with these data. This is obviously a complex and difficult subject from a regulatory perspective and I certainly don't have any answers to it. What I can hope to do perhaps is help frame some of the discussion that will follow as we hear the data and talk about these issues.

This is an example of the problem that everyone is concerned about. This is a polymorphic ventricular tachycardia occurring in the setting of a long QT interval and in this particular situation it's occurring in the setting of a bradycardia, which was the circumstance under which Torsades was first described. In fact, it was first reported in the setting of complete heart block and we have learned over the years that bradycardia is one of the most potent predisposing conditions to the occurrence of this form of tachycardia.

Now, what I'd like to do in this slide is make an analogy between the occurrence of drug induced Torsades with non-antiarrhythmic drugs and a lightening strike and the analogy is based on the fact that -- it's not going to work. So much for my tricks with PowerPoint. There was text. The text for this slide basically said that this is a rare event, very rare, potentially life threatening however and very difficult to predict in any individual despite the fact that we are aware of risk factors and can define some of those risk factors.

There it is. The issue of QT prolongation and Torsades spans many disciplines and that -- I think that is familiar to everyone who has looked at these data and these include cardiac electrophysiology, pharmacology, genetics, and both clinical practice and regulatory medicine and we'll, obviously, be confronting all of these areas as we talk about the data. What I'd like to do in the next several minutes then is say a few words about the mechanisms of QT prolongation and Torsades, talk a little bit about how drugs effect that QT interval, say a few words about some specific agents that serve as illustrative examples of the problem that we're grappling with, emphasize some of the risk factors associated with drug induced Torsades and also then mention at least a few comments about the clinical and regulatory implications.

This is a partial list of drugs that have been shown to prolong the QT interval. Many of these have been associated with rare instances of Torsades. The list is actually incomplete and I show it just to emphasize one, the very large number of agents that do effect cardiac repolarization and two, the fact that these agents span almost every therapeutic discipline in medicine, or at least certainly many of the major ones including neurology, allergy, cardiology, infectious disease and psychiatry.

There are two major syndromes, if you will, of long QT, the first being congenital which we are not going talk about today except in passing and second the acquired long QT syndromes which is clearly the topic of interest today, the major cause of which at least for the purposes of today's discussion are drugs. But it's important to emphasize that drugs are not the only cause of acquired long QT and this has implications when one talks about multiple risk factors and these include bradycardia, electrolyte abnormalities and certain forms of heart disease including hypertrophy and congestive heart failure in which repolarization is intrinsically abnormal.

This slide is shown just to illustrate the fact that cardiac electrical activity is comprised of a fairly heterogenous group of action potential configurations and durations. But in the normal situation, the heterogeneity exists within fairly tight boundaries and when that heterogeneity is exceeded either as a result of a drug or some pathologic state, we have an environment in which arrhythmias can occur and one form of these arrhythmias is Torsades de Pointes which we'll focus on today. The major interest with regard to today's discussion focuses not on depolarization but rather on repolarization which results from a complex series of events that involves many ion channels. But for the purposes of today's discussion and particularly with reference to drugs that cause Torsades, one can simplify this and this is a gross over-simplification, to a discussion primarily of three channels, the sodium channel which is responsible for depolarization, and then the calcium channel which is responsible for the plateau phase of the action potential and then repolarizing potassium currents, in particular the rapid and slow components of the delayed rectifier current.

It turns out, and this will come up repeatedly, that almost all of the drugs that are known to cause Torsades in fact, effect this channel. And it's also important to emphasize that the duration of the action potential, which is what determines the duration of the QT interval, is a complex inter-play of all of these channels and it is possible to prolong the action potential duration by interfering or inhibiting the outward movement of potassium from the cell.

One could also enhance inward movement of calcium thereby prolonging the plateau phase and if the sodium channels fail to inactivate when they normally should, it's also possible to prolong action potential duration. But the primary mechanism by which this occurs in clinical practice with the drugs that we use, is a result of block of the IKr current. The question of why IKr is so sensitive to a drug inhibition is not completely understood but there are some fascinating data from Dr. Sanguinetti's lab in Utah which are based primarily on structural analyses of the -- of IKr, of HERG and suggest that this channel has a particularly large vestibule in the pore region, that's an intra-cellular receptor obviously, in which potassium is pumped from inside the cell to the outside of the cell.

This pore region and, particularly the vestibule of this pore region, is unusually large compared with other channels and for that reason appears to have a certain degree of non-specificity to it with regard to its capacity for drug trapping. So many, many drugs in fact will effect IKr and this is largely a concentration dependent phenomenon. If you give enough of many, many drugs, you will get an effect on IKr. The question really is what is the potency of that effect and how does it translate at therapeutic concentrations into an effect on the QT interval in humans.

The mechanism of Torsades de Pointes is not completely understood but there are important observations that have come to light in the last five to seven years that suggest that while early after depolarizations which is a focal mechanism and was previously thought to be the primary mechanism, is probably not responsible for the arrhythmia itself, that is the sustained arrhythmia, but rather it is likely due to a form of reentry within the wall of the ventricle muscle. And this is a copy of a figure from a manuscript by Charles Antzelevitch, who has done some seminal work on examining different layers within the left ventricular myocardium and demonstrating that in fact, action potential characteristics in the epicardium, mid-myocardium and endocardium are quite different and that this heterogeneity within the wall of the muscle may be in part responsible for susceptibility to drug induced effects and the occurrence of Torsades.

These are recordings also from a paper from Antzelevitch's lab demonstrating that the mid-myocardia layer, the so-called M cells generally have significantly longer action potential durations than cells within the sub-epicardium and in the endocardium and in addition, these M cells appear to be more sensitive to rate effects particularly slowing of the rate, and to drugs which inhibit IKr than either the endocardium or the epicardium. And remember we talked earlier about the issue of heterogeneity. When one has heterogeneity within the wall of the muscle or any other place in the heart, one has a potential substrate for a re-entrant arrhythmia and this is just a cartoon that created by lifting these action potential figures from one of Antzelevitch's papers, lining them up with the epicardium, mid-myocardium and endocardium and if one gets significantly long enough prolongation within the mid-myocardial layer, such that this area is unresponsive for a longer period of time than either the epicardium or the endocardium, the potential for re-entry exists.

And one of the proposed mechanisms of Torsades is as follows. A ventricular premature beat perhaps due to an early after depolarization may arise anywhere within the heart and I've just placed it in the endocardium. These can arise from the mid-myocardium or from the His-Purkinje system and this ventricular premature beat occurs with timing such that when it conducts into the mid-myocardial layer, it encounters refractoriness because of the long action potential duration in this portion of the ventricular wall, and therefore, it blocks.

It then travels around this zone of block for a sufficient period of time to reach some portion of the mid-myocardium that has recovered and once it does that, it can excite the mid-myocardial layer, reach the epicardium and then descend again through the wall of the heart, resulting in a re-entrant arrhythmia. And there are very good animal models demonstrating that this is capable of producing what we see in clinical Torsades, that is a polymorphic VT varying configuration.

In fact, if one thinks of these as rotors, three dimensional rotors, that may migrate through the wall of the heart, it's not hard to imagine how the kind of polymorphic VT that we see in drug induced Torsades could be produced by this form or re-entry. Now what about the QT interval? The QT interval, as you all know, is an electrocardiographic measurement. It's made from the onset of the Q wave to the end of the T wave and is taken to represent repolarization. In fact, it represents both a combination of depolarization because it includes the QRS complex and repolarization. And it's important to emphasis that drugs or pathologic states that prolong the QRS duration may, in fact, effect the QT interval and if we were going to be purests about it, we would probably measure the JT interval but in clinical practice and drug development, that is not done.

It's also important to emphasize that this is not an easy measurement to make with a high degree of accuracy. The little boxes on those EKG's are 40 milliseconds. The calipers are about 20 milliseconds wide, half a box or a little less than that and there is tremendous variability both in the way that it's measured and also in the measurement itself within any individual over the course of 24 hours. So the techniques that are used to measure QT intervals and to assess the effects of drugs need to be extremely precise and this is not an easy undertaking.

Automated QT analysis has been used in a number of programs. I don't think it's used much any more. We all look at EKG's that show us QT interval measurements and while they are often quite reliable in the setting of normal T waves at physiologic heart rates, they are unreliable at extreme of heart rate and certainly in the presence of abnormal T waves or prominent U waves. In addition, the automated analyses tell us nothing about the TU wave complex morphology, a feature of these EKG's that is extremely important to pay attention to because drugs, in fact, may alter the QT -- excuse me, the TU morphology as their only sign of an effect on IKr and I've seen situations in which a significant QT effect has been missed entirely because of failure to pay attention to morphological changes.

So these EKG's have to be analyzed by people who are expert at doing it and if they're done by technicians, they need to be over-read carefully by cardiologists. This is an electrocardiogram from an entirely asymptomatic young individual who turns out to have a genetic abnormality that is a polymorphism of an HERG, an IKr abnormality, and the evidence of that is in lead two, in which there is a splayed, notched T wave with a fairly large U wave. This morphological change is fairly typical of one form of the congenital long QT syndrome, the one that effects IKr and it's also not uncommon with drugs that have very significant effects on IKr.

This QT-U complex -- this pointer is on way out. If you have a backup, I'd be very grateful. This QT-U complex is distinctly abnormal but if you didn't notice this, and simply looked at the automated analysis, you'd see a QT of 400 and a QTc of 401. In fact, those numbers are wrong and in addition, there is no comment at all -- thank you -- about the morphologic abnormality which is really the only way to detect this problem.

Sorry about that. Thanks. The QT interval is a variable phenomenon. It is highly labile and one of the most important influences on the QT is heart rate. You can see that the QT interval prolongs at slow heart rates and shortens at high heart rates. There's a direct relationship between cardiac cycle length and QT; the slower the cycle length, the longer the QT, the shorter the cycle length, the shorter the QT and in order to make comparisons at different times of day or under different circumstances, one has to use correction formulae that take into account the effective rate on the QT interval in any one individual.

In the vast majority of development programs, and certainly in clinical medicine, the formula that's used is the Bazett Formula which is the QT interval, the absolute QT divided by the square root of the RR interval that precedes that measured QT. And in the general population, this averages somewhere around 380 to 400 milliseconds, probably closer to about 380 milliseconds. The problem is that the Bazett formula is extremely limited. It falls apart at high heart rates and at very low heart rates and it is, in fact, probably the least accurate of all the formulae that are available. It was kind of defined into existence about 70 or 80 years ago based on 39 patients who were reported by Bazett and because it was the first description of this correction, it has stuck, but it is clearly a problem and one that I would at least argue should be abandoned.

There are many other formulae, including Fridericia, which uses the cube root baseline correction formula and a host of linear and non-linear regression formulae that can be used to correct for heart rate. I'm not aware of a perfect way of doing this. This is an area where the state of the art has not yet been defined and it is certain influx. I think the only important point I would emphasize is that the Bazett formula has very major limitations and particularly in the setting of drugs that increase heart rate may result in very misleading QTc intervals, that is over-estimates of the QTc interval.

The formula tends to over-correct, that is give you a longer corrected QT when the heart rate goes up. The normal range for intervals is listed here and for males it's somewhere under 430, females under 450. Clearly prolonged is greater than 450 in males and greater than 470 in females and this gender difference is important, not terribly well understood, but what is clear is that females tend to have longer QTc intervals at baseline and are also generally more susceptible to the effects of IKr blocking drugs and in fact, much if not most of the drug induced Torsades that occurs tends to occur in women and I'll show you some data on that in just a minute. In fact, I'll show it to you here.

These are data from two separate series. This series was reported by Makkar in 1993 in which he analyzed 332 cases of anti-arrhythmic drug induced Torsades de Pointes in which either a QT or a QTc interval was measured around the time of the event. And what he found was that about 90 percent of the patients who were described in these series had corrected QT intervals of greater than 500 milliseconds and of those in whom the QT but not the QTc was reported, about 80 percent had a QT greater than 500, suggesting that the degree of QT prolongation particularly when it exceeded 500 milliseconds, may be important in defining a subset of patients at increased risk for Torsades.

A similar analysis in 189 cases of Torsades with non-anti-arrhythmic drugs turned up virtually the same numbers. That is the vast majority of reported cases had QTc's or QT's greater than 500 milliseconds at the time of the event. There is, however, a potential reporting bias here and one has to be very careful about interpreting these data because we don't know what the denominator is in fact, investigators may be somewhat biased away from reporting cases in which the QT is not markedly prolonged. There's no way to know that with certainty but these are the best data we have.

These are the gender distributions in both series with anti-arrhythmic drugs. Seventy percent of the events occurred in women, 30 percent in men and with non-anti-arrythmics it looks essentially exactly the same and this is obviously important in a drug development program wherein assessing the potential risks of a drug you have to have data on sufficient numbers of women to determine whether there is a gender difference because clearly this is a risk factor for drug induced Torsades.

Now, what about some specific agents or at least classes of agents? I've divided them into two categories and I think about drugs in terms of two categories; those in which the therapeutic effect of the drug is directly tied to its IKr blocking properties and if you don't have the IKr effect you don't have the therapeutic effect and these are the anti-arrhythmic drugs, a least a partial list of the anti-arrhythmic drugs that exert this effect and we understand and accept the fact that Torsades de Pointes is part and parcel of using these agents. Most of them have an incidence of Torsades that is in the range of one percent and sometimes a little higher.

What's more important for the purposes of today's discussion and what creates the regulatory conundrum that everybody is facing these days are the low risk drugs, that is drugs in which the therapeutic effect is entirely independent of the IKr blocking properties of the drug, where in situations in which the IKr blocking effect is an undesirable side effect of the drug and in which the risk of Torsades is very low, that is less than .1 percent and often two orders of magnitude less frequent than that. And these include drugs that you're familiar with, antihistamines, antibiotics, antiviral agents, psychotropics and many, many other agents.

The occurrence of drug induced Torsades de Pointes generally with the latter class of drugs and that's all I'm going to focus on for the rest of this discussion, that is the drugs in which Torsades is an extremely rare occurrence, the occurrence of TdP in that situation is rarely a result of the use of the single agent by itself in a perfectly normal patient. It is more commonly due to the combination of a mild to modest drug effect on IKr in the setting of an effect amplifier. And this is a challenge because it's very hard to study all of these and two, it's extremely difficult to predict them in any population or any one individual over the course of time and these include bradycardia, electrolyte abnormalities, heart disease, particularly hypertrophy or congestive heart failure in which we know IKr is down regulated and action potential durations are long to start with, atrial fibrillation in which irregular cycle lengths predispose to the occurrence of Torsades particularly in the setting of anti-arrhythmic drugs, female gender, undetected mutations in the HERG gene, which codes for IKr. We know now that there are individuals out there who have phenotypically normal EKG's and who are asymptomatic yet who carry mutations in the HERG gene and are susceptible to polymorphic ventricular tachycardia and sudden death and these patients, while probably not numerous, are probably when exposed to IKr blocking drugs, significantly more sensitive than those who don't have these mutations.

High doses of drugs, metabolic inhibitors that result in pharmacokinetic interactions and pharmacodynamic interactions with other drugs that cause IKr blockade. This is just an example of the impact of hypokalemia on potency of IKr blockade. This is from Dan Roden's lab, showing you some examples of -- excuse me, these are concentration effect curves on IKr and let's just look at quinidine for a moment. This would be the IC50, the concentration of quinidine required to produce 50 percent block of the channel and you can see that at eight millimolar potassium, the IC50 is 3.8. This falls to 0.4. That is an order of magnitude change in potency at a potassium concentration of one millimolar. So electrolyte abnormalities can unmask a potentially very serious effect that would not be seen in the setting of normokalemia.

Metabolic interactions are obviously critically important. These were eluded to by -- in the introductory comments by previous speakers. In particular many IKr blockers are either CYP450 substrates and some of them are CYP450 inhibitors and that creates the potential for interaction both in the gut wall and in the liver and terfenadine is a good example of that and I'll come back to that in just a moment. In addition, elimination may also be an important issue particular in the setting of drugs that are renally excreted like Sotalol and Dofetilide. Renal dysfunction may predispose patients to markedly elevated plasma concentrations and increased risk of QT prolongation and Torsades.

This is just a partial list of 3A4 inhibitors to emphasize that they span a fairly broad spectrum of therapeutic classes and with many of these drugs and in particular grapefruit juice, which effects gut 3A4, the effect of these agents on the enzyme systems may long outlast the presence of the drug in the body. This is just a partial list of drugs that cause Torsades. All of them are IKr blockers and I show it just to emphasize that most of them are also either 3A4 substrates or inhibitors. It's an unfortunate combination of properties.

Here's an example of this kind of interaction, Terfenadine, well-known to everybody in this room, a drug with rather modest effects by itself when used at indicated -- at recommended doses of 60 milligrams BID, the effect trough on the QT interval was about 6 milliseconds, at peak about 18 milliseconds when given alone, but in the setting of a potent CYP3A4 inhibitor, there were marked increases in plasma Terfenadine concentrations and profound effects on the QTc in the range of 80 milliseconds. This is a drug that is subject to very significant first pass metabolism. Terfenadine largely disappears by the time its passed the liver and the active metabolite effects of Fenadine is, in fact, the drug with the antihistaminic effect but the QTc liability is related to the parent compound.

And if the parent compound is not metabolized in the gut or the liver effectively and the body sees large concentrations of it, this is the impact. This just emphasizes the difficulty in detecting this kind of risk. Trefenadine was prescribed more than 100 million times during the time that it was on the market. During the course of its development, it appeared to have relatively modest effects on the QTc interval by itself, but in the presence of Ketoconazole, marked increases in QTc, a clear-cut risk of Torsades and an increased risk of sudden death which led ultimately to the withdrawal of this agent.

Another example is Cisapride, which was also recently withdrawn. This is a modest IKr blocker with an effect on QTc, that's under 10 milliseconds. It is a CYP3A4 substrate and in the setting of CYP3A4 inhibitors, patients were exposed to markedly increased concentrations of the drug. Again, very difficult to detect any risk in the development program, more than 30 million prescriptions written for the drug, no arrhythmia signal and a very large data base review or in the NDA, yet between '93 and '99 there were 270 cases of serious arrhythmias of which 70 were fatal reported to FDA and ultimately the drug was withdrawn.

If one simply looks at these numbers, there's an adverse event rate here of less than one in 100,000, obviously impossible to detect in any drug development program and very difficult to detect even with post-marketing surveillance. Let me just come back then to this so-called multiple-hit hypothesis with drugs that have mild to modest effects on the QTc. It requires more than the drug alone to produce problems and we've listed those and discussed them earlier. This is a list of drugs withdrawn because of Torsades de Pointes and again, most of these drugs had relatively modest effects on the QTc when used alone but had some sort of metabolic liability associated with them and they include Trefenadine, Sertindole, Astemizole, Grepafloxacin and Cisapride.

QT prolongation is a significant issue in the overall scheme or drug monitoring. This is a table extracted from a Government Accounting Office report in January of this year which listed 10 drugs that were withdrawn from the market between January 1997 and December of 2000 specifically focusing on evidence of greater health risk in women and, in fact, eight of these 10 drugs had were associated with greater risk in women than in men, but I show it just to emphasize that of the 10, the most frequent cause for withdrawal was Torsades which was the cause in four of these 10 agents. So this is not a minor issue from a regulatory perspective.

With regard to screening drugs in a development program, and I'm going to skip through this very quickly because you will see and hear this data and I think everybody in the room is familiar with them, it is important to understand the effects of all new agents on IKr probably as well as on the calcium and sodium currents and also to assess in vitro action potential duration effects and this can be done in a number of systems. What's critical is that all of these parameters be evaluated over a very wide range of concentrations, preferably around 1000 fold. One can't always do that for technical reasons but that should be the target. These things should be assessed over a very wide range of heart rates and wherever possible, it's important to characterize the effects of metabolites on these parameters as well. A number of other in vitro models have been used including the left ventricular wedge preparation which I described to you, the profused rabbit heart and a host of in vivo models of Torsades including conscious and anesthetized rabbit models and a canine AV block model.

It's important to emphasize, however, that none of the pre-clinical approaches can exclude with certainty some risk of QT prolongation and Torsades and the bottom line is always the QT effect in humans. When one looks at the QT effect in patients or volunteers, we examine and the committee will be looking at these data today, both mean and mean max changes compared with placebo. This is effect on QT and QTc in particular, categorical analyses, that is percentage of patients who have prolongations of 30 to 60 milliseconds or greater than 60 milliseconds and outliers with QTc's of greater than 450, 470 and 500 milliseconds.

What's particularly important and often very difficult and I think this is a major challenge to sponsors, is to focus as much as possible on special populations, particularly patients with various forms of heart disease, hypertension, congestive heart failure and hypertrophy and coronary artery disease. In addition, it's very important that development programs have sufficient number of females in their studies to evaluate the differential effects of these agents on QTc with regard to gender.

Other aspects that one looks at in a development program include the occurrence of Torsades de Pointes which you almost never see in the kinds of drugs that we're talking about, the incidents of other ventricular arrhythmias, the incidents of syncope and the incidents of sudden death. Dizziness is often listed as well. I think that's a highly non-specific symptom and one that can't be used with any degree of comfort to point to the occurrence of a cardiac arrhythmia.

Finally, let me just re-emphasize then the critical importance of pharmacokinetic and pharmacodynamic interactions which can be assessed and then the potential for drug gene interactions which at the moment we are unable to assess. I think we've covered and let me just conclude with this slide which emphasizes that when one examines the potential liability of an agent with a modest effect on QT, a whole host of issues need to be considered, including the pre-clinical finding but most important the QT effects in humans, the adverse event profile, which we've discussed and all of these have to be evaluated in the setting of the therapeutic target. One's tolerance for some risk is likely to be higher in the setting of a life-threatening problem than it is in the setting of a non-life threatening problem.

The relative efficacy of the drug, as well as unique advantages need to be considered as do alternative options that may have a different and perhaps a somewhat better safety profile. And finally, when these are then evaluated a risk benefit assessment has to be constructed and the challenge, obviously here is we're reasonably good at assessing benefit, but in this particular situation not terribly good at assessing risk because we have a measurement, the QTc, which is a long way from the issue at hand, that is the occurrence of Torsades de Pointes and sudden cardiac death.

Let me just conclude by saying that there is no way to exclude risk with agents that have modest effects on the QTc interval. The other side of that coin, however, is that if we eliminated all drugs with some effect on the QTc interval, we would profoundly reduce our therapeutic armamentariums in many areas of medicine, in particular cardiology, infectious disease, psychiatry and oncology.

Thank you.

DR. RELLER: Thank you, Dr. Ruskin, for that telling tutorial. Our next speaker will be Dr. Douglas Shaffer, who will present the FDA post-marketing review of Torsades.

PRESENTATION OF DR. DOUGLAS SHAFFER

DR. SHAFFER: Good morning. It is my pleasure to present results of a post-marketing analysis regarding macrolide antibiotics and Torsades de Pointes completed by Sarah Singer and myself. We used the following outline. First, I will identify the goal and rationale for the presentation this morning.

Second, I will focus on the post-marketing analysis. The majority of the time will be dedicated to a descriptive analysis of the adverse event reporting system or AERS data base. I will present data from IMS Health in an attempt to describe macrolide utilization and finally I will briefly present reporting rate comparisons among the macrolide antibiotics incorporating a negative control. Finally, I will conclude with summary and conclusions.

The goal of this analysis is to systematically evaluate post-marketing data and attempt to provide the advisory committee with a descriptive overview of Torsades de Pointes and association with macrolide antibiotics. Two properties identified in the KETEK advisory committee packet and shared by the macrolide and microbials, specifically Clarithromycin and Erythromycin will be incorporated into results presented today.

First, the pharmacokinetic property, cytochrome P450 3A4 metabolism and second, the pharmacodynamic property of concentration related lengthening of the QTc or corrected QT interval.

While Telithromycin has not been marketed in the United States a post-marketing analysis now is warranted. This is recognized by the following quote for the European Society of Cardiology Policy Conference addressing iatrogenic QT prolongation and Torsades de Pointes. "Of concern is the interval usually measured in years from the marketing of these drugs to initial recognition of their association with QT interval prolongation and/or Torsades de Pointes. It is in the public's best interest to begin considering any potential for Torsades de Pointes sooner rather than later.

Considerable time and attention is warranted to randomize control trial data and drug development. However, we are well aware in the practice of medicine that adverse events often ignore the randomized control trial. It is important to keep in mind this presentation is in the context of the post-marketing setting. I would like to propose the following hypothetical scenario.

Consider a 61-year old female receives a broad spectrum antibiotic for acute sinusitis. The antibiotic undergoes hepatic metabolism and also has the potential to prolong the QT interval. After one drug, one patient, the controlled environment of the clinical trial abates and confounding variables become significant. This hypothetical situation may result in QT prolongation and ultimately a pro-arrhythmic milieu. From here we can postulate three outcomes. By far and most common, there is not pathophysiologic event, that is the patient takes the antibiotic without adversity rarely and as we just heard approximately less than one percent, the patient may experience a non-sustained arrhythmia. Even more rarely, the patient may experience cardiac sudden death. Our analysis today will be focusing on these confounding variables and presenting the patients that fall into the non-sustained arrhythmia and cardiac death categories.

Finally, as introduction to our AERS analysis, I would like to present a representative AERS report. Approximately one-third of our cases presented with a syncopal episode or near syncope. From there, the patient is seen in the emergency room where an ECG is obtained and QT prolongation is documented and ultimately Torsades. We commonly observe three outcomes; first, drug discontinuation and resolution; second, rapid deterioration and treatment, for example, with magnesium or other NCCLS protocol and rarely we do observe death.

I will now present results of our descriptive analysis of the AERS data base. We queried the AERS data base using four individual macrolide drugs as the exposure; Azithromycin, Clarithromycin, Dirithromycin and Erythromycin. We used the preferred term, Torsades de Pointes as the outcome of interest and since Torsades was not coded prior to 1995, we used the ventricular tachycardia for this time period. We included all reports regardless of nationality or routed administration.

In our description analysis our aim was to capture as much data and as many reports as possible. We excluded duplicate reports or reports prior to 1995 without Torsades de Pointes in the text. We systematically extracted pharmacoepidemiological data from each case report. PC SAS was used for analyses.

Our search query resulted in 268 reports being reviewed, 112 were excluded and 156 were analyzed and I will be presenting the details of this 156. Overall Erythromycin accounted for the majority of Torsades de Pointes reports, 53 percent. This was followed by Clarithromycin, 36 percent, Azithromycin, 11 percent. There were no reports of Torsades de Pointes associated with Dirithromycin which may be reflective of its relatively little utilization as I will show later. Twenty-eight percent of reports included an intravenous route of administration and 25 percent of the reports were of foreign nationality, a statistic that is not on this slide.

This is the first in a series of four slides where I will present pharmacoepidemiological data extracted from the AERS reports. For orientation, the variable of interest will be presented in the first column and the corresponding statistic, mean and standard deviation or frequency will be in the second column. Where indicated and appropriated at the bottom of the screen, I will provide the proportion of AERS reports providing information for the variable interest. For example, 93 percent of our cases reported information regarding age, 94 percent regarding gender, far less regarding race, 16 percent and weight was available in 26 percent of the reports.

Given this, the majority of Torsades de Pointes reports were primarily from older female patients. The mean age was 61 years and 70 percent of the reports providing gender were female. While very limited reports provided the data, those providing information, 60 percent were Caucasian and the mean weight was 152 pounds.

Approximately one-third of the AERS cases provided ECG data. Of this, 59 percent identified the QTc. The mean baseline QT in our AERS' analysis was 432 milliseconds. This is within normal limits for females and at the border upper limit of normal for males. The mean event QT were the QT Antecedent two or associated with the Torsades de Pointes report was 594 milliseconds and the mean change was 172 milliseconds. The interval between the initiation of the macrolide drug and the reported event was a mean of four days with three outliers greater than 120 days being excluded.

While we cannot assign causation and our aim is not to assign causation. Fatalities were reported. Fourteen outcomes of the 156 reports ended in a fatality. We also extracted data regarding co-morbid risk and co-morbid disease states. Forty-two percent of the AERS reports included any evidence of cardiac disease the most frequent cardiac disease reported was congestive heart failure, 23 percent. Renal disease and hepatic disease were both less frequent, 11 and 6 percent respectively. Hypokalemia or hypomagnesemia was present on 17 percent of the reports and hypokalemia alone on 15 percent.

Finally, we extracted data regarding concomitantly administered or concomitantly reported drugs and the mean number of drugs concomitantly administered or reported on our case series was four with a standard deviation of three and a range of zero to 15. We also evaluated two mutually exclusive classes of drugs or drug combinations. First, we looked for evidence of a drug interaction focusing on contra-indicated drug interactions using the product labels as guides.

In preview, this in general involves Erythromycin or Clarithromycin and a combination of Astemizole, Cisapride, Penazide or terfenadine. Thirty-one percent of the AERS reports had evidence of this contra-indicated drug interaction. In addition, after this, we further evaluated the reports to see if there was evidence of drugs or drug classes known to prolong the QT interval in uses less similar to that seen in the introduction. An additional 22 percent of the reports included drugs known to prolong the QT interval. This pie chart demonstrates the three resulting categories of AERS reports. In maroon, as I've just indicated, 31 percent of reports met the criteria for a drug interaction or contra-indicated drug interaction. In addition, 22 percent of the reports had evidence of drugs or drug classes known to prolong the QT interval. This leaves basically half, 47 percent, of the reports listing the macrolide drug as the sole suspect.

I would like to further describe the maroon section of the pie chart or the contra-indicated drug interaction section. As I mentioned, we defined this as Clarithromycin and Erythromycin and a combination of four drugs. As you can see, Cisapride accounted for the vast majority of drug interactions noted in our analysis. This was followed by Terfenadine and Astemizole. Two points should be noted.

First, Astemizole specifically is not a drug interaction with Clarithromycin but is with Erythromycin. Second, as just pointed out, these three drugs have been removed from the market. The second portion of our analysis involves data from IMS Health. We queried IMS Health's National Prescription Audit Plus to estimate macrolide drug utilization. We evaluated retail outpatient prescriptions dispensed focusing on oral formulations only since the formulation being considered today as oral.

I will present the data in two manners. First, I will use a figure to describe the representation of annual drug use and second, we will use this data in comparison of relative estimated reporting rate ratios. For this ratio, we consider the reports or numerator and we consider only domestic oral formulation or outpatient reports. Drug utilization will serve as a surrogate analytic population or denominator and we will use cefuroxime as a negative control.

This figure depicts annual macrolide antibiotic utilization with total prescriptions from zero to 35 million on a Y axis and years from 1993 to 2000 on the X axis. As you can see in the light blue line, Erythromycin use has steadily declined since the introduction of the newer macrolides. Clarithromycin, in a dark red or maroon appeared to plateau in 1996 and has a gradual decline since then. In contrast Azithromycin, a drug with five-day dosings, similar to a dosing we may see today has benefited from a marked positive trajectory and has only experienced a potential plateau in the last year.

Dirithromycin is not on the graph and this is due to its relatively low utilization averaging less than 500 prescriptions dispensed yearly. Finally, we proceeded to compare report utilization ratios within the macrolide drugs using Cefuroxime as a negative control. To do this we considered reports only of domestic oral formulation, outpatient origin and used 1993 to 2000 utilization after the drug listed in the first column reports follow in the second column, utilization in the third column and finally the report utilization ratio in the last column. I will focus on the last column.

Clarithromycin has the largest report utilization ratio among the macrolide antibiotics when considering domestic oral-formulation and outpatient reports only. This was followed by Erythromycin and then Azithromycin and Cefuroxime was used as our negative control. We are limited in the degree of certainty that can be placed upon these estimates due to the use of a surrogate analytic population and reporting biases known in the spontaneous system. From this slide, we propose to conclude that Clarithromycin had the greatest reporting ratio among the macrolide antibiotics and this was approximately nine times that of the negative control.

In summary, macrolide associated Torsades de Pointes reports are from primarily older female patients. Concomitant diseases and drugs are prevalent in potentially modifiable risks. Erythromycin overall accounts for most reports. Clarithromycin has the greatest reporting rate when considering domestic outpatient oral cases and accounting for drug utilization and finally Clarithromycin and Erythromycin Torsades de Pointes reporting rates are nine and three times that of Cefuroxime respectively.

Limitations must be addressed in considering results presented. First are those limitations germane to the spontaneous reporting system. These include adverse event recognition and report data quality. Torsades de Pointes is a difficult adverse event to capture. We evaluated each report individually in an attempt to assure reporting accuracy. However, without an ECG rhythm strip or data from telemetry we cannot have 100 percent certainty.

We must also consider the influence of biases. Market time and market environment are biases that should be considered. We did not adjust for secular trends. Rather we propose to let the data speak freely and consider these results in light of biases and the implication they may hold. Under-reporting is a well-known bias with spontaneous reports ranging anywhere from one to 10 percent of adverse events making it to the FDA depending on the severity.

I propose that we should consider a competing bias as well. We cannot overlook the potential bias in Clarithromycin reports due to the Cisapride received. As I showed, Cisapride was the most common drug interaction and overall in our data base accounted for approximately one-fifth of the reports. It is possible among the macrolide drugs that Clarithromycin received relatively more reports. We cannot make inferences regarding missing data and this addresses the specificity of spontaneous reports. An example is Telithromycin. We do not see reports of Torsades de Pointes with Telithromycin. However, it would be inappropriate to conclude that there is no association.

And finally, we are unable to establish causation in this type of analysis. Our goal was to present a descriptive overview. And last, reporting rate estimates are not synonymous with incidents rates. I propose, however, there are several advantages to consider. We systematically proceeded with pharmacoepidemiological data extraction within this class to an extent not previously seen before. This is a cost efficient analysis. In clinical research today we must recognize the need for cost efficiency. Spontaneous reports are a relatively cost efficient means for this type of analysis.

Regarding best available evidence, it is not always possible to turn to the randomized control trial or large cohort study particularly when evaluating rare potentially fatal outcomes. The AERS data base is arguably among the best available for not only single generation but descriptive and qualitative analyses as well.

Finally, we provided a detail analysis of individual drugs in the post-marketing setting. This series of case reports offers a descriptive overview of tangible data unavailable in the clinical trial. In conclusion, Telithromycin, the first of a new class of antimicrobials related to macrolides interacts with cytochrome P450 metabolism and prolongs the QT interval.

Recognition of the potential for Torsades de Pointes should clearly be acknowledged. And monitoring of post-marketing data and development of risk management strategies would be critical if the drug was marketed in the United States.

Thank you. And I would certainly like to take the opportunity to state that our AERS data base is in part dependent on the quantity and quality of reports submitted and we certainly encourage use of this valuable asset. Thanks.

DR. RELLER: Thank you, Dr. Shaffer, for that balanced review. We will now take a short break but please reconvene at 9:35 promptly to begin the sponsors' presentations.

(Whereupon, the meeting went off the record at 9:21 a.m. and went back on the record at 9:38 a.m.)

DR. RELLER: I should like to introduce Dr. Mindell Seidlin, Vice President for Clinical Development of Anti-Infectives Aventis. Dr. Siedlin.

INTRODUCTION OF DR. MINDELL SEIDLIN

DR. SEIDLIN: Thank you, Dr. Reller. Good morning, ladies and gentlemen. It is my privilege to introduce the Aventis presentation on Telithromycin, the first Ketolide. This is the agenda for the Aventis presentation. The introduction will focus on the need for new antibiotics in this era of increasing resistance. Subsequent presentations will detail the mechanism of action and in vitro microbial profile and human pharmacology of the drugs. The clinical efficacy and safety with special discussion on the ECG findings will follow. I will then summarize the unique features of Telithromycin which represents and advance in antimicrobial chemotherapy and address current therapeutic needs in this area.

Clearly, the emergence of multi-resistant respiratory pathogen, particular streptococcus pneumoniae have driven the need for new drugs in this area. Currently, physicians who perceive their patients to be at risk for drug resistant respiratory infections have only one or two classes of drugs to choose from. Availability of new drugs with novel mechanisms of action will reduce the resistance pressure on existing classes. While the emergence of multi-resistant strains is a key driver of medical needs in this area, we must not lose sight of the other elements for successful therapy of respiratory infections. Out-patient therapy is moni-therapy. New respiratory antibiotics must be effective against the full range of pathogens responsible for these infections; common, atypical and intra-cellular.

Ideally, they should be effective when administered with a brief simple regiment that can facilitate patient compliance and minimize drug exposure. Brief regiments which do not lead to misdoses in sub-therapeutic levels may limit further resistance. Recent policy statements by both the World Health Organization and the United States Department of Health and Human Services have included recommendations on development of new antibiotics in addition to judicious use of existing agents.

New drugs for community respiratory tract infections must be effective against the full range of relevant pathogens and must reach sufficient concentrations at the site of infection. Streptococcus pneumoniae is key because it is the most common pathogen but also the one most associated with serious sequelae and bacteremia. Rapid and cidal activity against sensitive strains of the pneumococcus may reduce the likelihood of emergence of resistant strains.

The other typical bacterial species which are associated with these infections include Hemophilus influenza and moraxella catarrhalis, both of which now include many beta lactamase positive strains. Achievement of adequate levels in plasma and particularly for pneumonia an extracellular fluid are important when treating these infections. The importance of atypical and intracellular pathogens is increasingly being recognized. In these inceptions, intracellular levels of drug are key.

Let us now turn to the clinical relevance and impact of resistance. Demonstration of the clinical impact of penicillin resistance was first observed in patients with pneumococcal meningitis. Due to relatively poor penetration of many beta lactams across the blood/brain barrier, even strains with intermediate levels of resistance to penicillin failed therapy in this indication. As the prevalence of high level resistance increased in the late '90's, outcome studies began to demonstrate impact of penicillin resistance on mortality, suppurative complications and other clinical adverse outcomes. It's important to remember in this context that some 60 percent of penicillin resistant pneumococci are also resistant to other classes of drugs.

The first cases of clinical failure of patients with Erythromycin resistance streptococcus pneumonia who were treated with macrolides were reported in the early '90's. At that time, there were few isolates with MICs greater than or equal to 4. Recently there have been an increasing number of reports. All of the reports in the latter part of the '90's have occurred in patients whose organisms have MICs of 8 or more. In contrast to the situation with penicillin and beta lactams, where high plasma levels can be achieved with increasing doses of the drug, increasing doses of macrolides to achieve plasma levels that will cover MICs of 8 or more in simply not feasible.

This slide shows data that was kindly shared with us by Dr. Cynthia Whitney at the CDC. In the left-hand panel, you can see the increasing frequency of Erythromycin A resistance among sterile sites of pneumococcus. In 1999, 20.3 percent of these isolates were Erythromycin A resistant. In the same year 16 percent were penicillin G resistant. The panel on your right shows the frequency of MICs of 8 or more in these same isolates. This demonstrates that not only has the frequency of Erythromycin resistance increased but the level of resistance has increased as well.

Here is the dilemma faced by physicians who must prescribe for out-patients with bacterial respiratory tract infections. Let's take community-acquired pneumonia as an example. Current Infectious Disease Society of America guidelines suggest that if a patient is not at risk for drug resistant streptococcus pneumonia, there are three options; macrolides, tetracyclines or flouroquinolones. If, however, the physician judges that the patient is at risk for drug resistance streptococcus pneumonia, the options reduced to one, flouroquinolones. The lack of options will further increase resistance pressure on this class. Further, there are no options for patients intolerant to that class.

Telithromycin is the first key light. It has excellent pneumococcal activity. This is crucial because in addition to being the most common bacterial respiratory pathogen, it is the one most associated with serious consequences. Telithromycin retains activity against Erythromycin A and Penicillin G resistant strains of the pneumococcus and is effective against all of the key community respiratory pathogens. The pharmacokinetic profile support a brief simple therapeutic regiment which will facilitate patient compliance.

This slides lists the indications that were proposed for Telithromycin. The mechanism of action and in vitro anti-microbial profile will now be presented by my colleague, Dr. Bryskier.

PRESENTATION OF DR. ANDRE BRYSKIER

DR. BRYSKIER: Good morning. It will be a pleasure for me now to share with you our current knowledge on the anti-bacterial activity and the mode of action of Telithromycin new Ketolide. Ketolide will synthesize and design to overcome Erythromycin A resistance within gram-positive cocci. This figure illustrates the structure of a Ketolide. The Ketolide are composed of three parts; the lactone ring, a 3-keto function and a Cll-C12 carbamate residue substituted by a long side chain.

Now, I want to share with you the property issue for this chemical structure. Here you have the cladinose, a natural sugar. If you remove the cladinose and by chemical modification you obtain the 3-keto function. The name came from the 3-keto function the Ketolide. Most important, the property you obtain with the 3-keto function as a following, first, high stability in acidic environment. After six hours of contact of pH1 95 percent of the Telithromycin activity remain.

Second, the anti-bacterial activity against erm-containing strains remain. Inability to induce macrolide lincosamine strepogramine-B resistance. Now, the second part of the structure. The C11-C12 carbamate residue substrated by a long side chain, you obtain the innovation of this compound. The C11-C12 gives you the following; reduced impact of efflux mechanism of resistance; second, enhanced antibacterial activity against gram- positive bacteria; third, govern intracellular accumulation and efflux in phagocytes and most important the mode of action.

I will call your attention of the mode of action. The first to know we are really on the front of the knowledge and the science with the mechanics of action and resistance to macrolide and Ketolide today. Telithromycin inhibits protein synthesis. Second, Telithromycin deplete ribosomes contained in bacterial cells. Let's have a look on protein synthesis. Protein synthesis is the protein produced in bacterial ribosome. Ribosome are constituted by two subunits, small one 30S and a big one 50S subunit. The target for Telithromycinis located on the 50S subunit. One side is the peptidyl transferase site and especially in rRNA place, a subunit of ribosome constituted by protein and rRNA.

As molecular, the peptidyl transferase is constituted by three parts. In 23S rNRA you so-called six domains. Two domains constitute the peptidyl transferase site, Domain V, Domain II and the link with another rRNA, the fifth rRNA. The difference between Erythromycin, Clarithromycin and Telithromycin is where the drug is fixed. When the drug enter in the pocket, here you have a fixation on Domain V through the desosamine, an amino sugar, Erythromycin, Clarithromycin, Azithromycin. Telithromycin also entered the desosamine so Telithromycin is also fixed on Domain V but the difference, there is a very long side chain, a carbamate side chain. This carbamate side chain allows you to be fixed on Domain II.

So the difference, you have a double fixation on the peptidyl transferase sites. Now, the depletion of the ribosomal contained. Inhibition of ribosomal subunit formation, 30S subunit, 50S subunit gives you this big ribosome system. With Erythromycin A, you have inhibition and abnormal protein form within 50S subunit. These proteins are destroyed and you have a depletion of ribosome. Telithromycin also acts on the 50S subunit but the difference, you have a double blockage, 30S subunit is also blocked. And I remind you that the 30S subunit is also the place where the protein synthesis is processed, so that means that you have a deep depletion and a total blockage of your synthesis and the consequence is that the drug is bactericidal. What is a consequence of a double binding?

So first, we explain that the mode of action is due to the C11-C12 side chain. Also overcoming mechanisms of resistance is partly due to this chain. If you have a methylation for instance on the Domain V, Erythromycin is unable to be fixed and the drug is no more active. So you have a resistance to Erythromycin and a cross-resistance with other like Erythromycin and Clarithromycin. Telithromycin could be also blocked in Domain V but a second arm on Domain II and Telithromycin retain activity against Erythromycin air resistance organisms.

Telithromycin is also active against another mechanics of resistance to Erythromycin efflux by another way. The drug is pumped out, when you have a poison in the cells, the cells wants to pump out the drug to survive. The blood is able to pump out Erythromycin A due to a high affinity to the pump, but the low affinity is for Telithromycin and you don't have this problem, so activity retained for Telithromycin. And that's a type of mechanism of mutation on protein, ribosome protein and Telithromycin retained activity for exactly the same problem with the C11 - C12 side chain.

Telithromycin is today the most active drug against pneumococci. On this shot, this work was done by NCCLS methodology and it was done by Gary Doern. And here you have Telithromycin. Telithromycin is more active than Clarithromycin, Azithromycin, Levofloxacin, as seen here, and Linezolid.

There's an anti activity that is a third property of this C11-C12 side chain is not only an anti-activity but also the drug is active against strep pneumonia resistance to other compounds which act on S. pneumoniae. Here you have comparison between Telithromycin and Clarithromycin. It's very evident that even if you have an efflux of macrolide and lincosamine streptogramine-B resistance, Telithromycin retain activity.

We have explored other drugs. When a S. pneumoniae is resistant to Cefotaxime, to Penicillin G, Tetracycline, Cotrimoxazole, Ofloxacin, Telithromycin retained activity and is highly active. So there is no cross-resistance between all these drugs and Telithromycin.

A fourth, very important, among this class of antibiotic it is the first time that we have a drug with a rapid bactericidal activity at an MIC level. So what is important, you have a very quick drop of three or more logs after four to six hours of contact. But at 24 hours with a mode MIC around 0.01 to 0.03 microbial we end up with Telithromycin against S. pneumoniae, we still have a bactericidal activity. So with 800 milligram per day, you cover the 24 hour period and you expect to have a bactericidal activity. All this in vitro data will confirm in vivo in animal model. Disseminated infection, lung infections in mouse and we use Erythromycin susceptible and Erythromycin resistant micro-organisms with different mechanisms of resistance, erm and mef. In a survey in North America only one strain about 2,000 -- out of 2,000 MIC of four, Telithromycin was about 4 so only one strain. So it is today a very rare occurrence.

But Telithromycin is not only a drug for S. pneumoniae. It covers many micro-organisms and mainly all the bugs which are involved in respiratory tract infections; S. pneumoniae, S. pyogenes, H. influenzae, moraxella catarrhalis, and S. pyogenes, Legionella pneumoophila, C. pneumonliae and atypical mycoplasma. The data you have here, I share with you were obtained also by NCCLS methodology.

The other very important point which is related and linked with the section is intracellular accumulation. With Telithromycin you have an accumulation or concentration in the cells in phagocytes with a ratio between 350 and 400 times. But the drug is also eliminated. Forty-five percent of the drug is pumped out in one hour period, so no accumulation in the cells. But accumulation of concentration in cells doesn't mean activity. So bioactivity is very high with Telithromycin demonstrated in C. pneumonaie, bactericidal activity, Legionella pneumophila in many models, S. pneumoniae and other intracellular pathogens. For S. pneumoniae it was demonstrated in the model two that it was the only compound that stabilized the cells, not as a compound we tested. We have tested quinolones, we have tested macrolides, and it is the only one who are able to do that. Even rifampin is not able to do that.

So to sum up the antibacterial activity and the mechanics of action and resistance to Telithromycin; Telithromycin is the first Ketolide, a new class of antibiotics. Telithromycin exhibits antibacterial activity against S. pneumoniae, resistance to other antibiotics. Telithromycin is able to overcome Erythromycin A resistance and achieve our targets.

Important bactericidal activity is obtained with the major respiratory pathogens. Doesn't induce MLSb macrolide, lincosamine, streptogramine -3 resistance. A low frequency of selection of resistance was noted in cellular passages. And it's active against all resistance pathogen with other drug. Thank you. Now I will ask Vijay Bhargava to continue with pharmacokinetics.

PRESENTATION OF DR. VIJAY BHARGAVA

DR. BHARGAVA: Thank you, Andre. Good morning. I will outline the key clinical pharmacology data for Telithromycin. Doses ranging from 800 to 3200 milligrams were given in this program to establish the pharmacokinetic and safety profile of Telithromycin. Safety aspects of Telithromycin in clinical pharmacology will be discussed by Dr. Benedict.

First, I will present the key plasma and tissue characteristics. Second, I will present the disposition profile and exposure profiles when these pathways are blocked either by drug interaction or impairment of an eliminating organ. And third, I will present support for the dose that was used in the Phase III trials.

This slide shows the pharmacokinetics after an 800 milligram single dose and as multiple doses to steady state in healthy volunteers. Absorption is rapid as seen in both cases as seen by the t max. The maximum concentration after single dose is similar to that seen after steady state in concentrations over two microgram per mL are achieved. Trough levels and area under the curve increase upon multiple dosing and steady state was rapidly achieved after the second or third dose.

There is a bi-expediential elimination with terminal half-life of seven to 10 hours. This profile is reproducible and representative of that scene in healthy volunteers and in patients. This slide shows the tissue concentrations of Telithromycin in patients when dosed with 800 milligrams once a day to steady state. In the three tissues here, adequate concentrations were rapidly achieved and are detectible for at least 24 hours. Patient data for the important target tissue epithelial lining fluid is from the laboratory of Honeybourne and Wise where similar data for other drugs has been reported. Levels in the ELF as high as 14.9 microgram per mL were observed. In other tissues high levels of Telithromycin were also achieved.

Regarding other key pharmacokinetic characteristics, absolute viable ability of Telithromycin is high at about 60 percent in both the young and elderly subjects. Protein binding is approximately 70 percent meaning that binding interactions are unlikely. Pharmacokinetics between men and women were similar and no food interaction was observed with this drug.

In the next few slides, I'd like to show you the multiple pathways of Telithromycin disposition. I will also present exposure profiles in population where these disposition pathways can be impaired. After an oral dose, over 90 percent of the drug is absorbed through the gastrointestinal tract. Prior to reaching systemic circulation about 33 percent of the drug is metabolized in the liver or gastrointestinal tract resulting in a systemic bioaviability of 57 percent. Once the drug reaches systemic circulation it can be eliminated by the GI tract or biliary secretion about seven percent, it can be renally eliminated as unchanged drug in the urine, about 13 percent.

Telithromycin can also be metabolized in the liver and excreted as several metabolites that add up to about 37 percent. The metabolites of Telithromycin are equally mediated by cytochrome P450, mainly 3A4 and non-cytochrome P450 pathways. Non-cytochrome P450 pathways are rarely associated with clinically relevant drug interactions or inhibitions. In clinical studies, the non-cytochrome P450 metabolite was not inhibited with Ketoconazole or in hepatic impairment.

Since the exposure of Telithromycin to CYP3A4 isozyme is limited, its potential for increased exposure when this pathway is blocked is minimal. This contrast with other drugs such as Cisapride and terfenadine where CYP3A4 is the primary isozyme for elimination. The effect of Telithromycin when given with or without 3A4 inhibitors on cardiac repolarization will be discussed by Dr. Benedict. Two other important points regarding Telithromycin metabolite to note; first SYP2D6 is not involved in its metabolism and secondly, due to minimal exposure metabolites do not contribute to the activity.

In the next few slides I will show you the exposure in populations where one of the disposition pathways could be impaired. Having established that the contribution of CYP3A4 is limited, we wanted to validate this in clinical studies with several drugs that are known to be potent inhibitor of this enzyme. This slide shows the effect of Ketoconazole , one of the most potent CYP3A4 inhibitors on the pharmacokinetics of Telithromycin at steady state to mimic the clinical situation. Area under the curve increased two-fold. Importantly the increase in C max which may be more relevant to safety was less at about 1.5 fold. As indicated earlier, this contrast with drugs where CYP3A4 is the primary pathway, for example Cisapride levels increased eight-fold and terfenadine levels 16 to 73-fold when they are co-administered with Ketoconazole.

With other potent inhibitors such as itraconazole, we saw a lesser interaction and also very importantly with grapefruit juice, we saw no change in exposure of Telithromycin. Regarding hepatic impairment, the maximum concentration in AUC values indicated no change in exposure after a single dose of Telithromycin. Interestingly the renal clearance of Telithromycin in subjects with hepatic impairment increased about 60 percent when compared to the age and sex matched controls in this study.

A multiple dose study in patients with hepatic impairment has been completed and data recently shared with the agency. Similar to the single dose results, in this multiple dose study where 800 milligrams once a day was administered for seven days, exposure did not change in the hepatically impaired subjects either on day one or on day seven. Increases in renal clearance of Telithromycin were also seen in this study both at day one and day seven in subjects with hepatic impairment compared to the age and sex match control. Similar findings for Clarithromycin have been documented.

Pharmacokinetics were investigated in subjects with different degrees of renal impairment. The Cmax and AUC increased about 1.5-fold in the group with creatinine clearance of 11 to 40 mils per minute and lesser increases were observed with the group creatinine clearance of 41 to 80 mils per minute. Thus, we have now seen that when a Telithromycin elimination route is blocked the risk of increased exposure is limited due to the multiple pathways.

One other point regarding the limited risk of exposure for Telithromycin due to drug interaction is its high absolute bioavailibility of about 60 percent. Inhibition of first past metabolism with Telithromycin will result in less than a two-fold increase unlike drugs with low bioavailability for example, Semvastatin and Cisapride and Terfenadine where changes of six to eight-fold or greater are observed.

Next we will look at Telithromycin in the elderly. The Phase I data are outlined in your briefing document and show that the elderly have a modest increase when compared to the young. Data shown here are from our Phase III study in community- acquired pneumonia patients where risk factors for increased exposure may have been present -- were present, for example, concomitant medication, infections or decrease renal and/or hepatic function.

When comparing the CAP patient under 65 to over 65, we see a 1.4-fold increase in AUC and more importantly, a 1.2-fold increase in Cmax which was lower. The safety of Telithromycin in the elderly will be discussed by Dr. Leroy and Dr. Benedict.

The next few slides illustrate data that was used to establish the dose and regiment for the clinical efficacy trial. Telithromycin was evaluated using the well-known mouse thigh infection model in Professor Craig's unit. The pharmacologically effective dose was similar irrespective of the frequency of dosing. That is, when similar doses were given, divided into three, six, 12 or 24 hours, the outcome was similar.

Area under the curve over MIC and Cmax over MIC were better predictors of efficacy than time over MIC. This indicated that the efficacy of Telithromycin is concentration dependent like that of azithromycin rather than time dependent like that of clarithromycin. Therefore, this model indicated that a once daily dose that would achieve adequate Cmax and AUC values would be efficacious in the clinical studies.

Once the pharmacokinetic, pharmacodynamic data were established, the human dose was chosen so that the unbound AUC over MIC values would be similar to or higher than the AUC over MIC values at the effective dose in mice. Eight hundred milligram given once daily met this criteria. In addition, as shown earlier, the tissue concentrations after the 800 milligram dose in Phase I studies were adequate to achieve the MIC 90 for S. pneumonia. Hence, the doses predicted by the model were validated by levels observed in humans after the 800 milligram Telithromycin given once daily.

A few points need to be made regarding the dose selection in H. flu. Contrary to the S. pneumonia situation, there is no well-validated model for predicting the therapeutic dose for lower respiratory tract infection caused by H. Flu. Non-typable H. Flu is rarely isolated in the blood stream of infected patients. Hence, the drug levels in respiratory tissues are important in the treatment of this pathogen. The ELF levels of Telithromycin after the 800 milligram dosed once a day exceed the MIC value for H. Flu.

Plasma and extra cellular concentration of Telithromycin are higher than those for Azithromycin which is one of the better in vitro macrolide for H. Flu. The efficacy of the chosen dose will be discussed by Dr. Leroy.

In summary, Telithromycin rapidly achieved the targeted plasma and respiratory tissue concentrations. Telithromycin has a well characterized and reproducible pharmacokinetic profile with a high bioavailability. Telithromycin has multiple pathways for elimination and its metabolism by CYP3A4 is limited. It is significantly metabolized by non-cytochrome P450 pathway and also eliminated as unchanged drug, unlike other drugs where CYP3A4 interactions have been an issue.

Pharmacokinetic and pharmacodynamic data were used to support the 800 milligram dose given once daily during the clinical efficacy program. Thank you and I'd like to turn it to Dr. Leroy.

PRESENTATION OF DR. BRUNO LEROY

DR. LEROY: Thank you, Vijay. My subject now is the clinical efficacy of Telithromycin in respiratory infections and I will start with the common element of study design and then cover efficacy in each of the four indications. The dose of 800 milligrams was chosen for all indications. And in this program we also studied the efficacy of short five-day treatment cause for common infections other than pneumonia. This was based on the potent in vitro activity of Telithromycin as well as its high and prolonged diffusion in tissue.

The benefit targeted with the short treatment duration was an improve patient compliance as well as a decrease of antibiotic exposure. In the pneumonia trials, a seven to 10 days regiment was maintained to insure that enrollment will not be biased to want patients with mild diseases. Some key elements of the study design was standardized across studies and indications. There were five study visits and you know in accordance to the FDA drug guidelines for Anti-Infectives, the test of cure was performed at the post-therapy visit between day 17 and day 21.

In studies where five-day Telithromycin was used, a placebo peer at five days was added in order to maintain the blind and the test of cure was performed in the studies at the same time after the start of treatment in both groups. And this was the most stringent approach allowing to capture the early relapses.

This approach was also recommended by the FDA. There were three main analysis populations. The mITT population corresponds to the intent to treat population excluding subjects who did not have the disease or did not receive treatment, the PPc corresponds to the mITT excluding subjects who had major protocol violation or an indeterminate response and this was the primary analysis population in all indications except in tonsillitis/pharyngitis and I will focus on this population during the presentation. Results of the mITT analyses, which are presented in the briefing document, were consistent with the PPc analyses. The protocol population was the primary analyses for tonsillitis/pharyngitis. Let us first consider the efficacy of Telithromycin in community acquired pneumonia. There were three controlled double blind comparative studies; 3001 compared 10 days of Telithromycin with 10 days of amoxicillin high doses, one gram three times daily. This comparator is increasingly considered as the best therapy in countries with high prevalence of S. pneumonia resistant to penicillin. 3006 compared 10 days of Telithromycin with 10 days Clarithromycin given 500 milligrams twice daily and 3009 was performed with Trovafloxacin as a comparator because of its efficacy against resistant strains of S. pneumonia.

This study was stopped before the planned sample size was reached when the FDA restricted the use of Trovafloxacin because of post-marketing safety concerns.

Three open-label studies were also performed, 3000 designed to obtain some pharmacokinetic data in patients with pneumonia. 3009 open-label which was performed only in South Africa, was aimed at gathering additional cases of S. pneumonia resistant to Erythromycin or penicillin and in this study consolidation on chest x-ray was required in all patients at entry. This study was an extension of 3009 with Telithromycin but no subjects of 3009 were included in 3009 open label.

Data from a dose comparison study performed in pneumonia in Japan 2105 will be also presented in agreement with the FDA. In total, more than 1300 subjects were treated with Telithromycin for pneumonia.

Looking at criteria associated with an increased severity, we can see that risk factors for morbidity summarized here for the mITT population were well balanced between Telithromycin and the comparative groups in comparative studies. And turning now to the pool of Telithromycin patients, you can see that the significant numbers of outpatients at risk of complication were included in the program. For example 16 percent of subjects had a fine score of three and a above, 56 subjects had an associated pneumococcal bacteremia. And this to our knowledge is one of the highest number of pneumococcal baceterimia submitted in an NDA for an oral antibiotic.

Therefore, we believe that the upper end of severity expected for outpatients with pneumonia in the community has been well-captured in this program.

On this slide the bar represents the cure rates with Telithromycin in blue and the comparator in gray. And at the bottom of the bars are the study numbers and the comparator used. At the top of the bars are the cure rates and the 95 percent confidence interval of the difference.

Analysis of the PPc population demonstrated equal balance between Telithromycin and high dose Amoxicillin and Clarithromycin with clinical cure rate with Telithromycin of 95 percent in studies greater than one and 81 percent in studies greater than six. In both studies the lower bounds of the confidence interval was well within trial limits.

In 3009 versus Trovafloxacin both treatments gave high cure rates exceeding 90 percent. However, the planned size was not reached for the reason explained earlier. But the results also support efficacy of Telithromycin in this indication. Clinical cure rate in the uncontrolled studies were consistent with the comparative studies and of note in study 3009 open-label where consolidation was required at entry, the cure rate was high with 94 percent in the PPc population.

In the Telithromycin group the clinical cure rate by pathogens for the most frequently isolated organism varied between 87 percent and 95 percent with the highest rate observed for S. pneumonia. Stringent serology criteria were used for the diagnosis of atypical pneumonia, which are detailed in the briefing document. Only patients with no common pathogens were considered for this diagnosis and the cure rate was over 90 percent for these pathogens.

Interestingly, 12 subjects were diagnosed with Legionella infection by serology or antigen soluble and these number of -- these pathogens are less observing of patients than in those hospitalized patients, but still ought to be considered as they represent a threat in the out-patients with pneumonia. All these 12 subjects were cured. This table summarizes the efficacy in the out-patients most likely to develop complications. In the Telithromycin group the cure rate, a test of cure, was high, 90 percent and above in elderly subjects. In subjects with Pneumococcal bacteremia in the PPb population, any subjects with fine score greater or equal to three.

Of particular note is the outcome of serving subject with Pneumococcal bacteremia. Efficacy in this subject may result from the combination of the excelling in vitro activity of the Telithromycin as well as its substantial plasma results and we believe that that differentiates Telithromycin from isolates compound giving low class labels and that it provides a high level of confidence in the treatment of out-patients with pneumonia in the community.

Now we will turn to the outcome in subjects with resistant isolate of S. pneumonia who are treated with Telithromycin. A summary of the results obtained in Western and Japanese studies are shown on this slide. And looking first at all resistant pathogens including both single and multiple pathogen infections, we can see that 19 strains were resistant to penicillin, 16 of which were associated with clinical cure, 25 strains were resistant to Erythromycin, 21 of these strains were associated with a clinical cure and of note, a total of 18 strains resistant to Erythromycin had an MIC greater or equal to 8 microgram per ML and 15 of these strains where associated with a clinical cure.

Results for infections due to single pathogens are given below with similar cure rates to single and multiple infections. If we look at the sub-set of subjects with resistant S. pneumoniae and documented bacteremia the number are limited but still substantial given that these were out-patients treated orally. Seven out of the nine subjects with S. pneumonia resistant to penicillin or Erythromycin were cured.

In one of the two subjects categorized as failure which are counted in all the rolls, and these subjects had an S. pneumonia resistant to both Penicillin G and Erythromycin A, the S. pneumonia was eradicated from the blood with a documented negative blood culture and an improvement of clinical symptoms. But these subjects had a secondary infection with SRAs isolated in urine leading to the prescription of an anti-infective.

In summary, efficacy was demonstrated with seven to 10 days of treatment with Telithromycin in pneumonia due to common and atypical pathogens. Cure rate in patients with pneumonia, with S. pneumonia and Legionella pneumophila which are the two pathogens associated with the risk of morbidity, were excellent. Efficacy was demonstrated in subjects with S. pneumonia resistant to Penicillin G or Erythromycin A and efficacy was also shown in the most vulnerable of patients such as the elderly, subjects with pneumococcal bacteremia and subjects with Legionella pneumophila.

I will now present the results obtained in subjects with acute exacerbation of chronic bronchitis. Two control studies were performed. Study 3003 compared five days of treatment with Telithromycin to 10 days of treatment with Amoxicillin/Clavulanic acid given 500 milligram three times daily and this was performed in subjects with the documented bronchial obstruction by lung function tests.

In 3007 the comparator was Cefuroxime axetil given 500 milligram BID for 10 days and patients were enrolled in these studies with a criteria of exacerbation type one or two. In the PPc population of both studies the clinical cure rate after the short five-day treatment with Telithromycin was equal to the longer 10-day treatment with the comparators, Amoxicillan/Clavulanic acid or Cefuroxime axetil.

In the PPb population which was selected according to strict criteria, clinical cure rate by pathogens with Telithromycin ranged from 68 to 100 percent. In the pooled population for both studies the cure rate was slightly lower for H. influenzae than for other pathogens. This was true also for the comparative treatment. As explained in the briefing document, this lower rate for H. influenzae was due mainly to the lower eradication rate observed in study 3003 in patients with community -- with COPD and documented obstruction.

In 3007 the eradication rate with Telithromycin was higher than with Cefuroxime axetil. For atypical pathogens, the clinical cure rates exceeded 90 percent in the 11 subjects with chlamydia infection diagnosed by serology with a four-fold increase of IGG. Looking at the out-patients most likely to develop complications in this indication efficacy was high in the elderly patients, in patients with one or at least two risk factors and in patients with bronchial obstruction.

To summarize, Telithromycin 800 milligrams given for five days once daily is effective in the treatment of acute exacerbation of chronic bronchitis due to these pathogens in patients with exacerbation requiring antibiotic treatment that is to say with type one or two. Efficacy was also observed in the out-patients most likely to develop complications at trials, the elderly and patients with documented obstruction.

Let us now turn to the acute sinusitis indications. Three studies were performed to support this claim. Study 3002 compared five days and treatment and 10 days of treatment with Telithromycin and this study was performed in patients and all the patients had bacterial documentation by sinus puncture. Study 3005 had three treatment groups, five-day and 10-day Telithromycin and 10 days with Amoxicillin/Clavulanic acid, 500 milligram given three times daily. Finally a second comparative study, 3011, was performed comparing Telithromycin for five days with Cefuroxime given for 10 days and this study also included bacterial documentation at entry.

In the comparative studies equivalence was demonstrated between Telithromycin for -- given for five days and the two goal standard in the Amoxicillin/Clavulanic acid and Cefuroxime axetil. Cure rates after five and 10 days with Telithromycin were also equivalent in study 3005 and in study 3002. Clinical cure rate by pathogens were high and comparable for the five-day and 10-day treatment regimen with Telithromycin for all targeted pathogens in this syndications with rates over 85 percent for S. pneumonia and H. influenza.

This slide summarizes the results obtained in subjects with S. pneumonia resistant to Penicillin G or Erythromycin which are of increasing prevalence in the syndication in the U.S. in our experience. Focusing first on the large population of subjects with single and multiple pathogen infections in the pool five and 10 days treatment with Telithromycin, which have been shown to have the same efficacy for the other pathogens, 11 out of 13 subjects with strains resistant to Penicillin G were cured. Eighteen out of 21 subjects with transference to Telithromycin A were cured and effectiveness was also shown in the five-day treatment group and in single pathogen infections although the numbers -- the experiments involves smaller numbers.

In summary, Telithromycin given at 800 milligram once daily for five or 10 days is effective in the treatment of acute sinusitis due to the main pathogens isolated in this indication. Telithromycin also proved to be effective against S. pneumonia resistant to Penicillin A or Erythromycin A. An important point to make for this indication is that equal balance was demonstrated in two controlled comparative studies between Telithromycin given once daily for a short treatment duration of five days and a standard treatment given two to three times daily for 10 days.

Telithromycin represents therefore, an effective alternative in this indication where the number of antibiotics demonstrating in vitro activity against all the key pathogens I related is becoming limited.

I will now summarize briefly the experience in tonsillitis/pharyngitis due to S. pyogenes. Two controlled studies were performed comparing Telithromycin for five days with Penicillin G given 500 milligram three times daily for 10 days or Clarithromycin given at 200 milligrams twice daily for 10 days. This was the main study performed to support the claim in this indication. Equivalence in efficacy which is the primary end point in this indication was demonstrated between the five day Telithromycin and the 10 day treatment with Penicillin VK or Clarithromycin. Reasons for type of therapy are given in the briefing document and confirmed the equal balance between Telithromycin five days and the two comparators used indicating that the short treatment duration with Telithromycin was not associated with a higher rate of relapses.

To summarize, Telithromycin, 800 milligram once daily given for five days is effective at treating tonsillitis/pharyngitis and is equivalent to the standard 10-day treatment with Penicillin VK or Clarithromycin given two to three times daily respectively. Eradication of S. pyogenes with the short five-day treatment duration given once daily is of particular importance in this indication where in practice compliance to a full 10-day treatment is rarely observed as a clinical symptoms result particularly in the adolescents.

This review has shown that the efficacy of Telithromycin given at 800 milligrams once daily was consistent in 13 Phase III studies across four indications where compared to a broad range of comparators that are well-recognized for their efficacy. A short treatment duration with Telithromycin given once daily for five days was effective in three respiratory tract indications and equivalent to 10 days of treatment with comparators given two to three times daily.

The most recent of efficacy within the treatment duration consistently non-pneumonia indication is important because this short treatment duration may favor better compliance resulting in increased efficacy and also it may decrease the potential to select resistant strains that can be savored misdoses at the end of a prolonged treatment.

In pneumonia, Telithromycin given for seven or 10 days showed excellent efficacy but most of all in this indication Telithromycin was effective in the out-patients most likely to develop complications. In pneumonia it was elderly subjects and subjects with pneumococcal bacteremia or Legionella infections, in chronic bronchitis, elderly subjects or subjects with significant obstruction.

Finally, Telithromycin was effective in patients with S. pneumonia resistant strain to Penicillin G or Erythromycin A with cure rates over 80 percent in pneumonia and acute sinusitis. I will now present the key 50 results of Phase III studies starting with an overview of the adverse event profile and for this I will focus on treatment related events observed in controlled studies. Then I will discuss the serious adverse events followed by key results of laboratory investigations and the review of ECG analysis will be presented separately by Dr. Benedict.

The safety population included all subjects who received at least one dose of study medication and had a subsequent safety assessment. And that's shown here to the left, a total of 3,265 subjects were analyzed with approximately two-thirds of them in controlled studies. Number of men and women were equal and the population of elderly subjects was substantial with 372 subjects analyzed. In addition, 95 subjects are aged 13 to 18 year olds were also included.

This table shows treatment emergent adverse events considered possibly related to the study medication and observed in more than two percent of the subjects. Gastrointestinal events were the most common events observed with Telithromycin with a slightly higher frequency than in the comparable groups in particular for diarrhea and nausea. But most of these events were mild or moderate in intensity and mild cases accounted for most of the difference between Telithromycin and the comparatives. Also added to this table is the rare event of blurred vision observed with Telithromycin in 0.5 percent of subjects and events were generally mild and resolved during treatment. An association with trouble in accommodations of subject, especially in high doses in Phase I studies, points to what is a condition of transient myopia as reported for example, with some of the compounds -- marketed compounds.

Overall the adverse events observed in both Telithromycin and the comparatives were generally mild and moderate in intensity and rate. Discontinuation due to adverse events were low with Telithromycin at about five percent for all events. In both treatment groups, gastrointestinal events were the most frequent events leading to discontinuation. This graph shows the percentage of subjects with diarrhea on each day of treatment, focusing on Telithromycin, Clarithromycin, Cefuroxime axetil and Amoxicillin/Clavulanci acid. Note that the prevalence of diarrhea with Telithromycin is lower than for Amoxicillin/Clavulanic acid which is one of the most widely used antibiotic treatment of out-patients with respiratory tract infections. It is slightly higher than that of Clarithromycin and Cefuroxime axetil.

Considering the distribution of treatment related adverse event by age group, events in both treatment groups were less frequent in the elderly subjects and in subjects age 13 to 18 years old than in the population of young adults. Eleven deaths reported for all treatment groups in the entire development program. In the controlled studies two deaths occurred with Telithromycin and four deaths with the comparatives. There were five deaths in the uncontrolled studies, all in the pneumonia indication. None of these deaths were considered treatment related and the overall rate of deaths in pneumonia patients in the Telithromycin group was around 0.5 percent with corresponds to the expected rate of death in pneumonia in the out-patients. There was no imbalance between Telithromycin and the comparatives in the occurrence of serious adverse events considering all events, all treatment related events and the rate of all treatment-related events was low at .04 percent for Telithromycin and .02 percent for the comparatives as it is expected for oral antibiotics.

In the uncontrolled studies the rate of serious adverse events was similar. In one of the uncontrolled studies a 53-year old male treated for pneumonia in Finland was enrolled with a normal transaminase at baseline and also elevated eosinophilia at baseline. He had a history of diabetes, asthma and three previous courses of macrolide in the previous year. Four days after the end of treatment he had an episode of gastritis similar to what was observed in several members of his family but this episode was followed by fever and transaminase increased with a peak eight days at approximately 1500 unit per liter. The biopsy showed centrilobular process and plasma cell infiltration with those inner fields.

Transaminase returned to baseline levels approximately eight weeks later and then the subject presented a second episode of transaminase increase nine months after the first episode with a peak at around 1300 units per liter. Biopsies performed seven weeks after the peak transaminase showed plasma cell infiltration and fibrosis.

In summary these subjects with transaminase increase at baseline and eosinophilia increase at baseline presented two episodes of transaminase increase with return to baseline after the first episode and a second episode occurring nine months later and to our knowledge there has been no published report of drug induced liver injury or two distant episodes were triggered by one drug intake.

Therefore, we believe it is unlikely that Telithromycin is the etiology of the hepatitis episode observed in this patient. Looking at treatment related hepatic adverse events reported in comparative studies, we can see that they were well balanced between Telithromycin and comparatives with an occurrence of two percent in both groups. Also there was no imbalance in the occurrence of event leading to discontinuation. The effect on hepatic enzymes was also evaluated in detail in Phase III studies and increasing transaminase greater than three times the upper limit of normal as summarized here in controlled studies.

As you can see the rate of transaminase increase was similar between Telithromycin and the comparatives at .05 percent and -- for Telithromycin and .04 percent for the comparatives in subjects with normal transaminase at baseline and 8.5 percent for Telithromycin, 11.1 percent for the comparatives in subjects with elevated transaminase at baseline which were mainly enrolled in pneumonia studies.

In these subjects there were no case with transaminase increase of greater than three times the upper limit of normal and bilirubin greater than 1.5 times the upper limit of normal. These graphs show -- allow a more precise comparison of the different level of increase in transaminase during the course of the study between Telithromycin and the comparatives. I'm presenting the subjects with normal transaminase at baseline were less likely to have confronting factors that complicate interpretation.

And since transaminase increase is frequently observed in pneumonia, we've analyzed the subject in controlled pneumonia studies and non-pneumonia studies separately. We can see a small difference in the transaminase increase above two times the equivalent of normal similar to what is observed with a macrolide but no signal above three times the upper limit of normal.

To summarize, Telithromycin was generally well tolerated with a pattern of adverse events similar to that of macrolides. Frequency of the transaminase events was slightly higher with Telithromycin than with comparatives but within the range expected for antibiotics. The adverse event profile was similar in different age groups and the rate of transaminase elevation was similar to the comparators.

Finally, and most importantly, the rates of serious adverse events and discontinuation were low and similar to the comparators. I'd like now to hand it over to Dr. Benedict.

PRESENTATION OF DR. CLAUDE BENEDICT

DR. BENEDICT: Good morning. I would now present the second part of the safety update, the ECG analysis. Macrolides have been associated with changes in cardiac repolarization. Telithromycin was -- has structural similarities and derived from macrolides. Because of this we performed an extensive pre-clinical and a prospective clinical investigation of the potential effect of Telithromycin on cardiac repolarization and compared it to different comparative macrolides and non-macrolides in our program.

This program was designed in accordance with the EU guidelines and FDA recommendations. Let me now first present to you the pre-clinical data. We performed extensive evaluation of the pre-clinical properties of Telithromycin. This included binding to the different membrane ionic channels, interaction with cloned channels particularly the Ikr or HERG channel but also Kv1.5 and Ikr. We also performed studies in isolated human atrial cells, studies in rabbit Purkinje fibers under different conditions of hypokalemia, different anti-arrhythmic drugs and in the presence of low pacing rates of bradycardia as well as interaction studies with Sotalol and quinidine. We also performed studies in awake animals.

The results of these studies are summarized in your briefing document but here I would like to especially present the data on Ikr or HERG channel. This slide gives the results of four different variables for some of the commonly used antibiotics except maybe Sparfloxacin. The first column is oral dose, the second the peak free plasma concentration, the third, the concentration required to inhibit the HERG channel by 50 percent and fourth, a ratio that relatively ranks these compounds by the amount that is required for 50 percent inhibition which is the plasma concentration that would be achieved.

Please note Telithromycin falls between Clarithromycin and Erythromycin. Please recall from Dr. Jeremy Ruskin's presentation, he said we need to also know not only the effect of the parent compound on the Ikr channel, but also its metabolite. We also looked at the effect of metabolites on the Ikr channel and even at 300 micromolar concentration, the effect was less than 20 percent inhibitory activated.

Let me now turn and present the Phase III data. As sponsor, we feel we have done a large extensive program in Phase III. We have gathered an unusually large number of patients. We have looked at the QT changes. In addition to that, we have also gathered PK/PD relationship in over 1500 subjects. ECGs were performed, pre and on-therapy usually days 3 to 5 when a steady state was reached. As indicated by Dr. Ruskin, it's important to read these electrocardiographic changes carefully. Therefore, it was all read by a single reader who was blinded to the treatment assignment and read in a random order.

There were approximately 1800 subjects in this group. Recall the emphasis or the need to know the information in patients with high risk factors. Our study program had fairly relaxed inclusion criterias, therefore, we were able to capture a large number of patients with high risk factors. In addition, about two/thirds of the way through the program, after safety review, the inclusion criterias were eliminated except for some drugs and congenital QT prolongation giving us a total of about 600 subjects to evaluate which will be presented subsequently.

QT was measured as the longest and the shortest interval from the 12 leads averaged and corrected by the heart rate using the Bazett's formula. To place the data in perspective, we are giving the data here as QTc but your briefing document also has a calculation by other formula which I'll be discussing shortly.

This shows the distribution of the QT measured at entry and on treatment in all of the patients who received Telithromycin. Baseline is orange, treatment is green. Notice very importantly, there is lack of emergence of a shorter, suggesting an emergence of a special population with abnormal QT prolongation. Overall the change was small, one millisecond, in keeping with what Dr. Ruskin has pointed to us, with a large standard deviation indicating a large biological variability of this measure.

This slide gives all the available data, including the comparators for QT change. If you look at Telithromycin for all study -- all subjects in all studies, or controlled studies all the different comparators, please note the change in QT interval is small and with the large standard deviation. In addition, we have also provided QT dispersion which is the difference between the short and the longest interval which are very similar between the different comparators.

Remember we had over 150 patients in whom we collected a PK sample with the electrocardiogram and these were done within one hour of each other. If we look at the change in plasma concentration versus the change in QTc interval, we notice a shallow slope relating the plasma concentration to the QT interval change of about 0.8 milliseconds per microgram per milliliter of the drug. Notice that there is a large capture in values particularly at low concentrations and even the patient who had the highest concentration fell on the line of regression. Since, as Dr. Ruskin pointed out, we need to know what the outliers are, I would like to now show you who these highest 15 subjects are about one percent of the population.

Here are the concentration range for these subjects spanning from 5.2 to 9.9 micrograms per milliliter. Here are the absolute increase in QT, up to a maximum of 435 and here are the delta QTc's spanning from minus 38.8 in a patient with 6.4 micrograms per liter, up to 18 increase in a person with 6.7. Of note, the patient with the highest plasma concentration had only an increase of 8.7 milliseconds consistent with our slope. Clearly we need to know more about the outliers because as we heard from the post-marketing surveillance data, it had a significant number of values are the highest values that are important in addition to what we also heard from Dr. Ruskin about the occurrence of large number of patients with values higher than 500 milliseconds in the studies he quoted to us.

Here we look at the examination of the QTc change by greater than 30 or less than 60 milliseconds or greater than 60 milliseconds, or where there was an absolute increase greater than 454 men or 474 women or greater than 500 for both. Notice in comparison to Clarithromycin, the numbers are the same. I would like to note for you that none of our subjects had an increase greater than 60 milliseconds and an absolute increase greater than 450 or 470.

We will now compare Telithromycin to non-macrolide comparators. Again, the numbers are broken down the same way and there is a small increase in the number compared to the non-macrolide comparators but very importantly, there are again nobody greater than 60 milliseconds and greater than 450 or 470 absolute increase. Probably more interesting and important are the effect of the drug in special population, particularly elderly, particularly those with hepatic impairment, renal impairment, et cetera.

This and the other slide gives the difference subsets that we examined as the number of patients in these categories. Notice that some of the categories had small number of subjects whereas the others had fairly large number to give some meaningful information. Again, I would like to point out to you the absolute changes are small and again, the larger biological variability as shown by the standard deviation. Whether we look at age, hepatic impairment, renal impairment or taking of the concomitant CYP3A inhibitors, we did not see a major change in the QTc.

We will now also examine other variables such as those receiving QT prolonging drugs, hypokalemia, we had 103 subjects with hypokalemia and/or diuretic therapy, those who had cardiovascular disease including hypertension and/or left ventricle hypotrophy congestive heart failure, et cetera, and also those who had prolonged or acquired QTc at baseline. Again, we see a small change in the setting of a large standard deviation.

Of note, patients with acquired QT prolongation had an 80 millisecond decrease on treatment the details of which are given on this slide which looks at all the changes in our study population. Of particular note, patients who had prolonged QT at baseline tended to show a decrease whereas those who had the shortest QT interval tended to show an increase. This is of some importance because often patients who have got acquired QT prolongation could have underlying cardiac disease or cardiovascular respecters and could unexpectedly respond by further prolongation of the QT interval.

Now let me also share with you the occurrence of treatment related adverse events that could be potentially be related to QT interval prolongation. We looked at the various categories and other than for the dizziness the occurrence of other symptoms are very similar. There was a slight excess occurrence of dizziness but most of these cases were mild and were not associated with any change in the QT interval when compared to the comparators.

I would also like to recall for the panel the remark made by Dr. Ruskin that often it's very difficult to interpret the finding of dizziness. Let me now turn to the Phase I program. Recall from your briefing document in our Phase III program when patients received Telithromycin, there was a decrease in heart rate. However, when we went ahead to do studies in our Phase I program involving normal volunteers who often have resting low heart rates, in the 50's or even 60's or 65's, often sinus bradycardia, we saw a significant increase in heart rate with Telithromycin. So in examination the different doses up to 3.2 grams of Telithromycin which is four times therapeutic dose, we also wanted to examine the robustness of correction by the QTc formula as indicated by Dr. Ruskin and to see whether alternate formulas would be required to better interpret the changes in QT interval.

This gives us the available information with respect to the different correction formulas. Here is the change in the QT interval and after correction of the QT interval by Bazett's formula, if you'll look at the relationship to change in heart rate, there is still a strong residual correlation. And an expediential formula mentioned by Dr. Ruskin Fredericia's removes that relationship better but there is still a residual correlation.

Therefore, as it has been done by some of the FDA divisions, we also examined what would be a better formula or exponent to correct this relationship and using our baseline drug free data, we developed an exponent of 0.284 which removes the dependency completely. So for my rest of my presentation I will show you the data as QTm but keep in mind the QTc data and the QTf data are also provided in your briefing document.

If you look at all the available Phase I program, spanning from 800 to 3.2 grams of drug intake, there is a range of plasma concentration up to 7.6 or 7.7 micrograms per milliliter. Again, we observe a shallow concentration relationship to QT change of about 1.01 millisecond. If you were to look at this data by QTc, you would see a slope of about 3.9, showing the effect of heart rate and the better correction by this formula away from the heart rate dependency. Very importantly I would like to next share with you three special drug interaction studies that we performed that is very important to evaluate the safety of Telithromycin; interaction with Ketoconazole, interaction with Cisapride and interaction with Sotalol.

This slide summarized our finding in the Ketoconazole interaction study. Both with placebo and Telithromycin there was a 3 millisecond prolongation in the QTn when the drug was administered to the patients. Again, notice the large standard deviation in these subjects. When Ketoconazole was administered, there was a 10 millisecond prolongation. When Ketoconazole and Telithromycin were given together, although there was about a 1.54 increase in Telithromycin concentration, there was little or no change in the QT measured in this study. I would like you to place this in perspective. For example, with Terfenadine, when administered with Ketoconazole there is a 16 to 72-fold increase in plasma levels and a 60 to 80 millisecond prolongation in the QT interval.

This slide illustrates the finding from out Cisapride/Telithromycin interaction study. Please recall the study was designed to examine the effect of Telithromycin on an increasing Cisapride blood level and its effect on QT. It was not designed as a head to head comparison between Telithromycin and Cisapride. During the placebo period or during the Telithromycin period or during the Cisapride period there is a small change in the QTn of about 1 to 3 millisecond and they were not significantly different from each other throughout most of this period.

But when Telithromycin is given with Cisapride, there's approximately a doubling of the Cisapride plasma level and at peak, there is a 10-millisecond prolongation in the QT interval. Again, I would like to place this in perspective by recalling for you that when you administer Cisapride with Ketoconazole, there is at least an eight-fold increase in plasma level and about a 60 millisecond prolongation in the QT interval.

Next I would like to present to you our finding from the Sotalol/Telithromycin interaction study. The first row gives the Sotalol plasma concentration achieved during the placebo period when Sotalol alone was given or in the presence of Telithromycin. Telithromycin produced a decrease in Sotalol level. If you look at the QTn or the change in QTn, there was 76 millisecond observed with Sotalol whereas in the presence of Telithromycin, it was 58 milliseconds. Since there was a change in the plasma concentration, we also examined the slope to put this in perspective and if you'll look at the different slopes, it was 45 versus 48 not different from each other. So we feel that in the presence of a type 3 mix with Sotalol, Telithromycin does not increase the risk for changes in QT interval.

Next I would like to share with you a special study in special population that was conducted with the -- at the recommendation or suggestion of the FDA to characterize the safety of Telithromycin in patients with cardiovascular risk factors. To our knowledge this is the first such program in an anti-infective area. This involved 24 subjects with congestive heart failure, ischemic heart disease, either vascularized or non-vascularized, non-life threatening arrhythmias or valvular heart disease, et cetera, who were exposed to either Telithromycin 800 milligram or 1600 milligram to either therapeutic dose or Clarithromycin, 500 milligram twice a day or placebo with all subjects receiving all four periods in our cross-over design.

These patients had electrocardiogram and 24-hour Holter monitoring before and after treatment. Here are the changes in plasma Telithromycin concentration on 800 and 1600 milligram and Clarithromycin for comparison. Here are the change in QTn which was not different between Telithromycin 800 or 1600 and here is the Clarithromycin data for comparison. If you look at QTc because the drug had, again, a large effect on heart rate, the changes in QTc are a little bit more prominent but again, in the context of the variability, they remain small.

Of note, there was no evidence of arrhythmias on the Holter monitor in any of these subjects. This slide gives all the plasma concentration data in these 24 subjects in the two different periods for 800 and 1600 milligram of Telithromycin. Again, we note a shallow relationship between QTn or change in QTn and the plasma Telithromycin concentration. Again, if this were to be using QTc, the slope will be higher because of the effect on heart rate. To put these findings in perspective, I would also like to share with you the data from Phase III program where we had more than 280 subjects with underlying cardiovascular disease.

Again, here is the change or the measured plasma Telithromycin concentration and the change in QTc and we notice little or no slope in patients who are -- who will actually receive the drug in clinical practice.

In conclusion, Telithromycin had a wear effect in Ikr channel and importantly in patients with respiratory tract infection who will receive the drug, we observed a small change in QTc of about approximately one millisecond. There was a shallow relationship between QTc and plasma Telithromycin concentration over a wide range. There was no difference in the occurrence or frequency of outliers of QTc between Telithromycin and macrolide or non-macrolide comparators.

Analysis of different at risk populations did not reveal a propensity for enhanced effect on cardiac repolarization. Very importantly, we did not notice any cardiovascular adverse events such as Torsades, admittedly cannot be detected in such a small population but also absence of ventricular tachyarrhythmias, absence of syncope or -- that could be associated with QT prolongation.

As sponsor, we believe that we have defined the risks associated with the change in plasma Telithromycin concentration. The strong Phase III data, the shallow relationship between plasma Telithromycin concentration and change in QTc, the high viability of the drug, the availability of multiple mechanism for this compound including a heart rate that cannot be blocked by clinically available drugs and the compensatory increase in renal excretion in patients with hepatic impairment strongly limits the possibility that there could be an unexpected increase or large increase in plasma concentration of Telithromycin in clinical use and potential for acute cardiac repolarization changes.

I would now like to ask Mindell Seidlin to come and conclude the presentation.

CONCLUSIONS OF DR. MINDELL SEIDLIN

DR. SEIDLIN: As illustrated earlier, there is a clear need for new oral antibiotics for treatment of respiratory tract infections. The prevalence of high level resistance to both Penicillin G and Erythromycin A in the United States exceeds 15 percent now. Resistance to other agents including Cotrimoxazole, Tetracyclines and others have increased as well. Resistance to Quinolones has been reported. The term multi-resistance has now been applied to the pneumococcus. There is increasing evidence that resistant strains of pneumococci are associated with clinical failures and adverse outcomes.

At the same time, it is important that respiratory antibiotics provide effective therapy for the full range of pathogens involved in these infections. The current medical environment is one with increasing numbers of elderly patients and patients with numerous underlying illnesses taking a variety of concomitant medications. In the past, many of these patients might have been hospitalized for treatment of community acquired pneumonia or acute exacerbation of chronic bronchitis. More and more these patients are being treated in the community.

The initial choice of effective oral therapy in these patients is crucial. Telithromycin is the first in a new class of antibiotics, the Ketolides. It has a novel mode of action which two binding sites to the 23 SR&A of the 50S ribosomal subunit. This accounts for its superior activity against sensitive strains of the pneumococcus and retained activity against Erythromycin A and Penicillin G resistant strains. It is also active against the other key respiratory pathogens, common, atypical and intra-cellular.

Telithromycin has a well-characterized and reproducible pharmacokinetic profile. Therapeutic levels are rapidly achieved in plasma, in infected tissue and inside cells. Telithromycin was consistently effective in all analyses in 13 clinical trials in the four proposed indications. Of special note in the studies of community acquired pneumonia, cure rates in the elderly were 90 percent. Likewise they exceeded 90 percent in patients with pneumococcal bacteremia. Cure rates were high in patients with atypical infections and all cases of Legionella were cured.

In the other three indications, the five-day once daily regiment was demonstrated to be as effective as 10-day coursed of comparator regiments despite stringent criteria which could have favored longer duration therapeutic regiments. Efficacy was also demonstrated in community acquired pneumonia and in sinusitis in patients with infections due to Penicillin and Erythromycin resistant pneumococci.

Safety was evaluated in 3,265 patients which included a broad array of ages, underlying illnesses and concomitant therapies. Gastrointestinal adverse events were the most common identified and occurred in a range that is well-recognized in oral antibiotic therapy. Hepatic events and transaminase elevations were uncommon and occurred at rates comparable to those of comparative ages.

Discontinuation of therapy and serious adverse events were rare and occurred at rates similar to the comparators. A thorough evaluation of the effect of Telithromycin on cardiac repolarization revealed a weak effect on the Ikr channel and approximately one millisecond increase in the QTc interval in patients with respiratory infections. No cardiac adverse events attributable to this change were observed.

The effect is similar in magnitude to that of widely used antibiotics. We anticipate that our planned post-marketing surveillance program will confirm the safety profile observed in the Phase III trials. I will conclude by summarizing the benefits that Telithromycin will bring to patients.

It is highly effective against the pneumococcus, the pathogen most associated with morbidity and mortality in respiratory infections. It is active both in vitro and in patients with Penicillin and Erythromycin resistant strains of S. pneumoniae. It is a single agent which is effective against all of the key respiratory pathogens, common, atypical and intracellular. The brief five-day regimen in common infections is likely to enhance patient compliance.

Currently there are few therapeutic options for out-patients with respiratory tract infections who are at risk for drug resistant S. pneumoniae. Telithromycin will effectively and safely expand those options. Thank you.

DR. RELLER: Thank you, Dr. Seidlin and your colleagues for the Aventis presentation. These presentations are now open for discussion and questions directed to the presenters. Dr. Bell?

DR. BELL: The safety presentations were appropriately most focused on the cardiac issues and I guess we're going to hear more about that from the FDA this afternoon but I wonder if one of you would be kind enough to elaborate on the blurred vision issue. I seem to recall that in the tonsillitis studies there were maybe half a dozen people and I think maybe mostly woman who were who had blurred vision versus none in the comparator group and one of the speakers referred to this as transient myopia.

And I guess I'm -- could you please talk a little bit more about that? For example, how transient was it and how do you know that this wasn't a potential harbinger of some more serious neurologic or opthamologic event that just didn't complete? Can you just talk a little more about that?

DR. SEIDLIN: I'd be happy to. You pointed out correctly that most of the cases of blurred vision did occur, in fact, in young women in the tonsillitis/pharyngitis trial. Most of them lasted for a matter of a couple of hours and resolved while the patients were still on therapy. We actually observed blurred vision at high doses in some of the Phase I trials and had the opportunity to conduct opthamologic exams in those patients. In those patients, there were no abnormalities observed in the fundi and in the lens and the retina, et cetera, and the opthamologist concluded that there was a difficulty in accommodation accounting for the blurred vision.

We believe that the mechanism related to this probably has to do with a cholinergic effect of the drug which is generally a mild effect but maybe having more impact on the eye muscle.

DR. RELLER: Dr. Murray?

DR. MURRAY: In the examination in animals of retina or optic pathways, anything pathologically done?

DR. SEIDLIN: There were pathologic examinations of the eye. I'm going to call upon Dr. Miller to detail those for you.

DR. MILLER: We did carry out extensive examinations within the pre-clinical program. This included a peer review of the retinas from the repeat dose toxicity studies in the rats, the dogs and the monkeys and within the monkey one month toxicity study we also measured and recorded electra-retinograms and we saw no adverse effects in any of these examinations.

DR. RELLER: Dr. Lazzara?

DR. LAZZARA: Just in the Sotalol experiments, the Sotalol trials, I'm sorry, in which Sotalol was combined with Telithromycin, the -- you didn't give the dose of Sotalol that was given, 160 BID. And I was -- the delta QTn, that's your QTn correction, that was the QTn on Sotalol versus the QTn at baseline?

DR. SEIDLIN: I'm going to ask Dr. Benedict to come to the microphone so that he can better respond.

DR. BENEDICT: Yes, the QTn product was developed on QTn for Sotalol.

DR. LAZZARA: So it was a mean 76 milli-second prolongation on Sotalol.

DR. BENEDICT: Right, on Sotalol, yes.

DR. LAZZARA: The other point about the Sotalol experiments, was the -- the heart rates would have been then, I guess, fairly low. Do you have any data on what the heart rates were on the Sotalol when the Telithromycin was added?

DR. BENEDICT: I think we did not see any additional change in -- no, there was additional change on top of Sotalol of about four to five beats.

DR. LAZZARA: Decrease?

DR. BENEDICT: Increase in heart --

DR. LAZZARA: Increase with the Telithromycin.

DR. BENEDICT: Right.

DR. LAZZARA: Thank you.

DR. RELLER: Yes, Dr. Moss?

DR. MOSS: Could we get some detail on just how the QT interval was measured, because at least in one of the slides it appeared that one took the longest QT minus the shortest QT or not minus, but averaged the time, the QT interval between the longest and shortest and used this as -- and then corrected for it; is that correct?

DR. SEIDLIN: Yes, that's correct.

DR. MOSS: Do you have any data on simply what was the longest QT, not averaging between the longest and shortest?

DR. SEIDLIN: Dr. Benedict?

DR. BENEDICT: Yes, we have that data available but we provided the QT dispersion and since we did both the averaging for the pre and on treatment, we felt the change would be the same because it's being averaged, but on the whole, to answer your question what was the longest, the longest would have been about 10 milli-seconds more than what would have reported for the absolute QT but in terms of the delta QTc, it would have been no different.

DR. MOSS: That delta QTc would have been no different using the -- your end correction, your exponent that's different from the Bazett or Fridericia?

DR. BENEDICT: For the Phase III program, we presented the Bazett formula QTc correction. So at least in that population, we saw the changes were approximately similar, the same or similar whether we did the averaging or we took the longest.

DR. MOSS: And did you do any corrections for the placebo, that is there is some prior data that I've seen for the various doses, the 800 and 1600 milligram doses, on the QTc Bazett and after -- with subtraction of the placebo and adjusted for the placebo and it was really quite considerable the QTc changes.

DR. BENEDICT: Yes, because the doses ranging from 800 to 3200 milligram were studied in normal, healthy volunteers often with heart rates as low as 60, even some having rates lower than 60 and in this individual there was an increase in heart rate so if we just use the QTc formula we see a slope of about three to four milli-seconds per microgram per milli-liter of the drug but when we appropriately now correct for the effect of the heart rate using QTn formula, the slope is now about 1 milli-second per micro-liter per milli-gram of the drug.

DR. MOSS: Could you give some idea of what the average heart rate changes were?

DR. BENEDICT: Yes, the average heart rate change, I think we have a slide on that. Okay, while the slide is coming up, the average heart rate in the normal volunteers I would say was about approximately in the region of about five to eight beats per minute. DR. MOSS: Right, and let me just ask one final question. I notice there were 95 subjects in the age range of 13 to 18 years. Two questions, were they given the same dosage or was the dosage attenuated for body weight and have you studied any children younger than age 13?

DR. SEIDLIN: The 13 to 18-year olds were all treated with 800 milligrams once a day, so the dose was not adjusted in those teenagers. We are currently carrying out a pediatric program but that data has not yet been submitted to the agency.

DR. RELLER: Dr. Sumaya, did you have a question?

DR. SUMAYA: Yes, it's somewhat related to the latter question. Do the sponsors feel that the potential utilization of this drug amongst various age groups correlates with the amount of studies done in those age groups in both efficacy and in safety? In other words, the ones that are going to use this more have been the ones most studied? Is there some rough comparability?

DR. SEIDLIN: Well, we did cover the age range of patients anticipated to take this drug in marketed use. Whether they are in direct proportion is always hard to say, but we certainly did cover the adolescents, the vast majority of patients between 18 and 65 and a substantial number of patients over 65.

DR. RELLER: Yes, Dr. Lee.

DR. LEE: Yes. Could somebody address the paddock metabolism that's non-P450? Is that glucuronidation, is it -- do we know?

DR. SEIDLIN: Dr. Bhargava?

DR. BHARGAVA: The metabolism of Telithromycin is by several metabolites and one of the major metabolites that's the circulating species is the RU-363 which is loss of the areal grain. And that is the pathway that is metabolized by the non-cytochrome P450.

DR. RELLER: Yes, Dr. Moss.

DR. MOSS: I'm struck by the enormous standard deviations of the measurements. Do you want to provide any explanation that is with mean value of delta QTc of one milli-second and a standard deviation of 21 milli-seconds? I know you touched on this as outliers but that just seems like an enormous variability.

DR. SEIDLIN: We believe that this is attributable to the biologic variability of this parameter and the errors in measurement which are common. There is also a great deal of spontaneous intra-individual variability, it's hard to say but it exists anyway, which is accounted for by this.

DR. MOSS: Was the QT measurement done by manual or was it machine read with operator over-read or what?

DR. SEIDLIN: There was operator over-read for all EKG's.

DR. MOSS: No, but the primary measurement was machine read?

DR. SEIDLIN: I'm going to ask Dr. Benedict to talk about the parameters for reading the ECG's.

DR. BENEDICT: When the electrogram was originally performed at the site, there was a machine over-read for safety evaluation on the spot for the investigator. But then all the electrocardiograms were transmitted to a central reader who blindly manually read every single electrocardiogram looking at all the 12 leads.

DR. RUSKIN: Can you describe the methodology by which they were read? Was this on a standard ECG? Were they computer read at higher speed or how was it done?

DR. BENEDICT: These were all standard 12 lead electrocardiograms at 25 millimeter per second paper speed and using the standard criteria for measuring the QRs interval.

DR. RELLER: Yes, Dr. Lazzara.

DR. LAZZARA: Another question about the hypokalemia trial, can you give us an idea of the severity of the hypokalemia?

DR. BENEDICT: Yeah, we did not specifically do a study on patient population of hypokalemia. We included all patients who had potassium less than 3.5 milli equivalents per liter and we had approximately either about 60 to 70 patients with potassium less than 3.5 or 40 to 50 patients who are receiving concomitant diuretic therapy and that's where we have the data from that group.

DR. LAZZARA: Yeah, but I was curious as to do you have a mean, say what the potassiums were in that group?

DR. BENEDICT: I don't have it right now. I think we can provide it to you later on.

DR. RELLER: Dr. Davis.

DR. DAVIS: Can you say some more about these Japanese studies? They were included and then excluded in this count for the question of resistance.

DR. SEIDLIN: Uh-huh. We only presented the resistance isolates from the Japanese studies. We did not present the overall safety or efficacy from those studies. The study design in Japan was a little bit different from the study design in the U.S. In Japan, the -- there was seven days of therapy whereas in the U.S. there was seven to 10 days depending on which study.

The end of therapy visit in Japan was at seven days and then there was a follow-up visit at post-therapy which was corresponded to our 17 to 21-day visit and that was used as the end point for those studies. The severity criteria for enrollment in the Japanese study were slightly different and none of those patients had blood cultures which is why we had no bacteremias. It may have been why we had no bacteremias in those studies. However, the cultures were all done and the MICs determined by NCCLS criteria and the clinical criteria for cure were quite similar.

DR. DAVIS: You said the severity was different. So the severity was less.

DR. SEIDLIN: Not necessarily less. They actually used a different severity measure in those studies. So they were graded in a slightly different way.

DR. RELLER: The last two questions to Dr. Seidlin were posed by Dr. Barry Davis. Dr. Chesney?

DR. CHESNEY: I had a question about emergence of resistance and I wondered if isolates -- you made any attempt to look at isolates that might have still been present on therapy and after therapy and did you detect any emergence of resistance and then in the briefing document there is the comment, "While exposure to Telithromycin did select for pneumococcal mutants within increased MICs, most remained within the proposed susceptibility range", and I wondered if you could just elaborate on that a little bit for us.

DR. SEIDLIN: To the first part of the question we did not identify any Telithromycin resistant mutants in patients who had been treated in the program. The comment refers to attempt to select resistant mutants in the laboratory by serial passage and we'd be happy to present that data. Dr. Bryskier?

DR. BRYSKIER: So we did -- one study was performed concerning the election of the detection of resistance mutant after serial passages. What we obtained is that we studied Telithromycin. After 44 passages we only selected or obtained three strains, resistant with MIC of two to four microgram per mL with L22 mutations on the ribosomal protein. There work was done in comparison with macrolide Erythromycin and Clarithromycin. We obtained -- we select mutation or mutant after five to 20 passages but the number of these mutant are extremely high and also the level of the MIC we obtained are different. With Erythromycin or Clarithromycin or Erythromycin and also Roxithromycin, we did all the 14 and 15 available macrolides. We obtained MIC above some time 32 or above 32 because there is a difference. So the number -- and so we have tested the frequency, the -- usually mutant will cure after one out of 10 to the seventh micro-organisms, with study is one to eight or one to the nine, so the frequency is low. The number of the mutant we obtained is low after a lot of cellular passages within 40 or more and also MIC we obtained four times the normal MIC or some time we have five times the normal.

DR. RELLER: Dr. Murray, do you have a follow-up question related to the resistance?

DR. MURRAY: Yes, I assume that was with an Erythromycin susceptible non-erm B containing strain. Were similar studies done if this background strain had erm B in terms of mutational frequency to resistance to Telithromycin or increased MICs?

DR. BRYSKIER: We did Erythromycin susceptible of course and also we tested -- or we did also this work with mef and with erm B. So with mef first we obtained also few mutants but increased MIC from all three or all six to all 25. With erm B we are selected also few strains but starting with an MIC, an extremely high MIC, for Erythromycin A. The work was done by Peter Appelbaum and we started with MIC of above 32 and we have some strain with MIC of four. We have one with four and one with all three and as with one. So we obtained few. I cannot say no, also mutation on L22 and that is a very rare occurrence today in clinical setting.

DR. RELLER: Yes.

DR. MURRAY: Could you give frequency numbers as you did for the erm susceptible, one in 10 the sixth, seventh, eighth?

DR. BRYSKIER: Yeah, the work was done by Roland LeClercq in France and presented last year. I can't -- also it's 10 to the eighth or 10 to the ninth. It's very low.

DR. RELLER: Dr. Leggett had several questions related to resistance.

DR. LEGGETT: Not to resistance, just one question to resistance. Regarding a more clinical resistant pattern and regarding your desire for an indication for Penicillin resistant and Erythromycin resistant pneumococci, you stated the data of -- in your MM7 and other places, of 20 to 30 percent resistance and yet in the numbers involved, when you look at your trials of community acquired pneumonia, it's less than 15 percent. Can you explain the discrepancy between what the CDC is showing and what we once again, when we're trying to look at this resistance indication, we can't seem to find them.

DR. SEIDLIN: I think this is a problem that the committee has experienced before with many submissions, that the number of patients with resistant pneumococci captured in clinical trials tends not to be representative of what's captured in epidemiologic studies in the population. I would, however, point out that our rate of identification of resistant pneumococci is relatively high compared to some other submissions. Indeed, we studied some 1300 patients treated with Telithromycin with community acquired pneumonia and had the numbers of isolates that you saw.

In contrast, my recollection is that for instance in the Levoquin submission there were many thousands of pneumonia patients in order to obtain some 16 resistance isolates. So I don't know whether that's a tincture of time with the increase of resistance out there in the world or we were extremely clever at placing our study sites or we were actually able because our enrollment criteria didn't try to select for patients with pneumococcal pneumonia. So I think that, yes, we didn't quite get numbers that were representative in the community but we did get fairly respectable numbers for resistance isolates in this program.

The other point that I would like to make is that the numbers that I cited are sterile site isolates from the CDC and that's, of course, important because those are invasive disease and it's a good conservative number. Most studies that have looked at the incidents of resistance in respiratory secretions have been considerably higher. Numbers 18 to 35 percent have been reported in a variety of studies and in fact, our experience is consistent with that in that the rate of isolation of resistant pneumococci and sinusitis was relatively high.

DR. RELLER: Dr. Cross.

DR. CROSS: With regard to the adverse effects, in your presentation you lumped all the Phase III studies together. I was just wondering in the studies that looked at chronic bronchitis, a population probably enriched in older patients and those with underlying heart disease, was there any difference in the profile of adverse effects in this population?

DR. SEIDLIN: Dr. Leroy, would you present the adverse events in chronic bronchitis, please?

DR. LEROY: There was no major difference. There was -- we can, yes, put the slide on which is the possibly related treatment adverse event in chronic bronchitis patients slide on. So you recall a rate of 13 percent in the presentation of diarrhea and in fact, it's almost less in this indication. It goes with the results that we've presented to you in elderly patients where the rates were a bit lower.

DR. RELLER: Dr. Ebert.

DR. EBERT: I have a similar question related to one of the other adverse effects, that being dizziness. Were there any demographic characteristics of the patients who experienced dizziness that would have predicted that they would see the side effect or was it related to the timing of the dose, for example?

MR. LEROY: I think that for the last question, the relation to the time and the dosing it was well addressed in Phase I, as you know, we've conducted high/low studies in this program which is generally not done. So we've been able to see that the dizziness was related to the dose clearly. At three gram two, there was more dizziness than 800 milligram, and it was following the dosing and the next 10 hours following the dosing.

DR. EBERT: And I was just curious, were those patients generally taken this fasting or were they taking it with meals or did that delay absorption similar to what's been discussed with Trovafloxacin or is delaying the absorption and perhaps, minimizing the dizziness that's seen?

DR. LEROY: Okay, I understand your question. We've not studied exactly this question so I cannot answer exactly to the question. What we know is that the food interaction but Dr. Bhargava could elaborate on that. There was no difference in peaking the dose interaction. The nausea were a bit lower in Phase I when given -- when the product was given with food. We did not see any difference in dizziness but there was not a system of recordation so we cannot answer with certainty to this question.

DR. RELLER: Dr. Leggett, did you have another question?

MR. LEGGETT: Yeah, I have a couple of related questions concerning your proposed break points on page 70 of the book that you showed us in terms of susceptible and resistant, tying that in with the peak concentration levels and the AUCs that you listed and especially if you're looking at the Phase I, the QTc intervals and look at like 5,000 -- I don't remember which slide it was but there were several thousand concentration points.

I notice that very few of them were above four. At least a half or more of your peak concentrations were less than two and yet, you weren't a break point that is four for the H. flu and I'm wondering a little bit about that especially in relationship to the 60 percent efficacy that you showed in, I believe it was the AECB trial. In that regard, I wonder, looking at Bill Craig's data, in his thing on page 222, it looks at the AUC to MIC ratio and the static point is someplace between 125 and 250, it looks like, looking at that trial.

If you take your AUC and divide by the MIC of the dose, to me it looks like your break point should be about .25 or .5 at the most rather than two or four, so I have a question about that. Related to that also, what in his mouse model, what was the peak to MIC ratio that corresponded to that static break point as well? And I say that in regards to many of the peaks with the telithro not getting to 4 and staying at one or two.

DR. SEIDLIN: Okay, there are several points in there and I'm going to try to remember them all so that I can touch on them. The first thing I'm going to talk to is the distribution of plasma concentrations observed. Then we're going to talk about the AUC/MIC for S. pneumo and then we'll talk about H. Flu, okay? Let's see if I can remember all that.

Okay, the concentrations that were shown were not all obtained at peak. We requested that they be taken, I believe it was one to two hours after dosing but there was quite a distribution. The important point there was we were trying to correlate the serum level with the ECG findings and the serum level and ECG findings were within 1 hour of each other.

We can show you, if you like, the distribution of the time points for those levels. So those levels did not necessarily represent Cmax. For some patients, in fact, they did but not for all. All right, now turning to the AUC/MIC for S. pneumo. I think it's important to distinguish S. pneumo from H. flu in this context. Clearly, Dr. Craig's model is a S. pneumo model and it's really a systemic infection model where blood levels are quite important and I'm going to call up on Dr. Craig in a moment to detail those results.

For H. flu there is no good model to predict AUC/MIC and as Dr. Leroy mentioned earlier, this is very much a tissue based infection. It's rather unusual to detect H. flu in the blood. So no AUC/MIC criteria for efficacy against Hemophilus influenza have really been developed from a model based method. Indeed, what we can do is we can look at the MICs of the H. flues isolated in clinical trials and the clinical outcomes to see if there is a correlation. Dr. Leroy, would you like to present that data, please, and then we'll go back to the S. pneumo.

DR. LEROY: If we look for example at community acquired pneumonia, we can see that -- we can see that -- so the number of pathogens both 4 is limited so we cannot conclude on that but we do not have a cut-off point from this data for its influencing community-acquired pneumonia.

DR. LEGGETT: And so as another follow-up question, I was going to ask about AECB but I'll ask it about here. Most of the time in upper respiratory tract infections, H. flu is cleared at least 50 to 60 percent of the time with a placebo. So looking at these rates, I'm a little nervous and I wondered what that placebo rate or if it's been tried, if you can tell me. In my recollection, it's pretty high for acute exacerbations of chronic bronchitis making these numbers of 80 to 90 not necessarily as impressive.

DR. LEROY: Yeah, I think in those patients with community-acquired pneumonia one important point is that some of these patients had relatively severe pneumonia. So they may be cleared. We have also analyzed the fact to say that only single pathogen infections and patients with a single pathogen infection associated with a concurrent gram stain to try to see if there was a difference, and there was no difference.

The number are smaller if we can see these numbers. Certainly influencing community-acquired pneumonia is a question, the causality is a question, that's why we -- the next one, the one with the concurrent strains. Keep going. Okay, in any case, the number for the concurrent and gram stains showed a good efficacy around 80 percent. So we can see this one, slide on, which shows the number with a single pathogen, infection and a single and mixed infection. It was variable between the studies and if we can have the number with the -- we have also analyzed the concurrence gram stain, which is interesting to have, also 9, where we lose probably 20 percent of the single pathogens who had -- who had not a concurrent gram stain that is to say were not gram-positive or gram stain and had also a good cure rate. Yes, slide on.

So these are the single pathogen infections with gram-positive on the gram stain. So when we tried to narrow down to a pathogen that we can consider more causative, we have the same type of rate.

DR. LEGGETT: Can you explain to me the mechanism of bacterial eradication in that highest group of the H. flu where not even the peak levels reached that amount?

DR. SEIDLIN: We believe it's due to the levels that are present in the extra cellular fluid and in the tissue. Dr. Bhargava showed you levels of 14.9 in ECF from Dr. Wise's lab and those probably account for the excellent outcomes.

DR. LEGGETT: But that's at the two to three hour level, right?

DR. SEIDLIN: Right, and they were actually sustained for quite awhile. Okay, shall we go to Dr. Craig?

DR. CRAIG: The drug has significantly different binding in animals than it does in humans. It has higher binding in animals and so we needed to focus on free drug levels. And so when we start focusing on free drug levels, then the AUC to MIC ratio falls further. And then another thing that is quite clear with these drugs is also clear with flouroquinolones is that the white cell has a marked effect on the area under the curve that's required for efficacy.

So that when we look at those strains that we have been able to look at in normal mice or in using special CBAJ mice where we, again, don't have to make them neutropenic, we're getting down to AUC to MIC ratios in the range of about 5 to 10 that's required. And when one takes those kind of values, then looks at the free drug ratio that one sees in humans, then one starts getting much higher break points, probably not up to 4 but up to at least 1.

DR. RELLER: Dr. Ebert.

DR. EBERT: Just a clarification regarding the sinusitis studies; were any of your sinusitis studies double-tap studies so you were able to look at eradication. I'm particularly interested in that because you're saying that the 5 day and the 10-day courses are equivalent and I wondered whether you had any microbiology data to support that.

DR. SEIDLIN: No, we do not. As you know, doing double-tap studies is rather a formidable challenge.

DR. RELLER: Dr. Leggett.

DR. LEGGETT: A follow-up on the sinusitis studies, any ideas about in your comparator groups with the amoxicillin port. In your PPb people there were only like four percent of bacteria that were actually isolated whereas in the other telitrho and the cefuroxime group there were up above 40 percent. Why the variability?

DR. SEIDLIN: That's because the first study which just compared five and ten days of therapy was a sinus puncture study in all patients and the third study there was the option of either sinus puncture on endoscopy. The study you're referring to was a clinical study so there were no bacterial isolation there.

DR. RELLER: Yes, Dr. Sumaya.

DR. SUMAYA: It appeared from the date presented that the resistance strains clustered around those patients that had community-acquired pneumonia as well as sinusitis. Is that correct and also was there any clustering in predictor H groups from the resistant strains?

DR. SEIDLIN: One would expect to see the most streptococcus pneumonia and pneumonia and sinusitis. It is an important pathogen for AECB but other pathogens begin to play a more important role so that's not particularly surprising. Do we have an age distribution of the resistant pathogens? I don't recall that off-hand, but perhaps we could address it after the break.

DR. RELLER: We need to conclude very shortly but I have two closing questions. Dr. Murray has one as well. Dr. Murray.

DR. MURRAY: Actually, I'll ask perhaps that it be answered in the afternoon because you may need -- perhaps you can put the information together on a slide. I do have some concern about resistance emergence in the isolates that have erm B and I think there have been some abstract -- data presented in abstract form that suggest quite a considerably higher rate of emergence or resistance by plating erm B containing strains onto telithro containing agar.

With that background, there are two things that I've noticed that I would be interested in hearing comment on. One was an animal model published in January in JAC showing lack of efficacy against one of the erm containing strains and perhaps there's an explanation for why that was lack of efficacy. Bill may have hit upon it. If there's higher binding in animals but they humanize blood levels but I don't believe they were free drug levels, so I'd like a comment on that. Perhaps that one would be quick.

But what I'd like perhaps to see a slide of is some of the data that were given in here on page 32 about animal results with Erythromycin resistant animals doesn't really allow me to compare either the ED50 -- the way it's written out. I'd like to see a table that says this was the Erythromycin susceptible, the Erythromycin resistant, the two mechanisms and what were the comparators -- comparative decreased in count and I can't pull that out of here. I'm given an ED50 for Erythromycin susceptible, but then for the Erythromycin resistant, I'm just told log decrease. So I can't really compare those internally and if it would be possible, I'd like to see that in a slide.

DR. SEIDLIN: Okay, we will certainly make an effort to get that together for you after the break.

DR. RELLER: Dr. Wald, did you have a question?

DR. WALD: I was just curious, in your high risk populations, patients with liver disease, the elderly and the kidney patients, what was the range of the peak concentration of Telithromycin, because it seems to me there are relatively few patients in each of those groups.

DR. SEIDLIN: So you'd like to see -- remember these are not always peak concentrations but we can certainly show you the measured concentrations in those populations or we can go back to the Phase I data and look at those again. Which would you like to see? Do you want to see the Phase I or the Phase III?

DR. WALD: I wanted to see the peaks.

DR. SEIDLIN: Okay, so that's Phase I. Dr. Bhargava.

DR. BHARGAVA: I think the three that you're ask in terms of the hepatics, we had I said earlier, even the multiple dose situation at steady state almost no change in the peak levels and the peak levels are approximately 2.2 microgram per ML, so that would be the same in the healthy volunteers as well as hepatics.

In the elderly, as I showed in the Phase III, we did have a significant number of CAP patients where we connect -- collected serial pharmacokinetic samples up to six samples per patient and in that again, I was able to show that when you compare the less than 65 patients to the over, there's about a 20 percent increase in the over 65, so it's a 1.2 ratio and I think that last population you asked about was the renal population and in the moderate to severe impairment, we see approximately a 1.5 full change in the Cmax, so it would go from about 2.3 to 1.5 full higher than that.

DR. WALD: I was interested in the range rather than the mean.

DR. BHARGAVA: Okay, those were the means and when we looked at the ranges in the hepatics in fact, the range was very tight. So the outliers, in fact, were -- in our Phase I programs were no more than 6 so the highest, I think was about 6.5, 6.6 in all of our Phase I studies. So it's about, you know, two and a half to three-fold.

DR. RELLER: In the presentation emphasizing activity against Erythromycin A resistant streptococcus pneumoniae, there was an implication of lack of effective methylation efflux mutation and yet a couple of slides later, on M22, there was a shift in the MIC90 in organisms that had -- pneumococci that had these mechanisms. What's the explanation for the shift up in MIC90?

DR. SEIDLIN: There is some shift in MIC90 in the erm containing strains but still well within the sensitive range. Clearly, in erm resistant isolates instead of the two binding sites on the 23SRA there's only one and it's still effective but not as -- the MICs do go up a little bit. For the efflux mutants, the -- as Dr. Bryskier pointed out, there is less affinity for the pump with Telithromycin than with Azithromycin, Clarithromycin, Erythromycin. However, the pump still exists and does pump out a little bit of the drug. So, yes, one does see a slight change in the MIC but it's well within the sensitive range.

DR. RELLER: And secondly, there's discussion of the activity with other potentially effective agents for resistant pneumococci. Could you comment on what data you have for Clindamycin activity versus Telithromycin activity in Erythromycin A resistant strains of pneumococci?

DR. SEIDLIN: Okay, I'm going to ask Dr. Bryskier if he has any data on Clindamycin to share.

DR. BRYSKIER: I want to ask you, do you want to know what's happen when you have a Clindamycin susceptible and resistant strain and phenotype or an erm B phenotype or genotype?

DR. RELLER: Well, I think in clinical terms and in the laboratory that an Erythromycin- resistant pneumococcus, pediatricians, Dr. Wald, others, may comment on this, many of these strains are susceptible to Clindamycin. And I just wanted to know where this agent fits into the whole scheme of things relative to mechanisms of resistance compared with Telithromycin when one looks at what the options are available for therapy in patients who either can't get Penicillin or have resistant strains to some of the other agents.

DR. BRYSKIER: Okay. When you have an MLSb mechanism, for instance, and an erm B or methylase usually Clindamycin is not considered as active so the second is when you have Clindamycin suspectable and I will say Erythromycin resistance, that's an m phenotype. For S. pneumo today when you have I would say mef, that's m phenotype Clindamycin susceptible, Erythromycin resistant for S. pneumo, Telithromycin MIC remain in the range of I would say 01 to 025, but what's most important, there is not really correlation between MIC of Telithromycin and the underlying mechanism of resistance to Erythromycin A for S. pneumo. You can have MIC of 05 and a mef but you cannot install an MIC of 05 with an erm. So there is no real correlation.

I will show you -- may I have the slide? On this slide, that is a population distribution, for the strain, we collected of pre-clinical trial. You can see easily -- we can see easily that whatever MIC you obtain. So 002 up to 1 you can have resistant strain to Erythromycin or susceptible to Erythromycin, so there's no correlation. The same work was done with different gene and also you have no correlation.

For instance, now we know some time you could have a mutation on the 23SRA on the loop, on the peptidyl transferase, for instance we have one strain now with a mutation in A2049 and we still have a good ATt of Telithromycin. So there is no real correlations between the gene, Erythromycin resistance, or susceptibility.

DR. RELLER: We will reconvene promptly at 1:00 p.m. after the lunch break. A quick reminder to the members and guests, there's an area in the restaurant that's been reserved for you, okay, so that you can come back promptly. Thank you.

(Whereupon, at 12:05 p.m. the above-entitled matter recessed, to reconvene at 1:00 p.m. the same day.)

A-F-T-E-R-N-O-O-N S-E-S-S-I-O-N

(12:59 p.m.)

DR. RELLER: Back to the afternoon session. There were a few residual questions, additional data that were left from this morning and I should like to suggest that we handle those after the FDA presentation during the time of the questions and discussion just before the break and then immediately after the break, we'll -- the committee will address the questions posed by FDA.

The FDA presentation will be started by Dr. George Rochester, who will present the FDA's assessment of clinical efficacy of Telithromycin.

PRESENTATION OF DR. GEORGE ROCHESTER

DR. ROCHESTER: Good afternoon. I am George Rochester, a clinical statistician with the Division of Anti-Infective Drug Products. I'm George Rochester, a clinical statistician in the Division of Biometrics III, co-located with the Division of Anti-Infective Drug Products and I will be presenting an overall summary of the clinical efficacy for this application.

We have heard the sponsor's presentation this morning which was quite detailed and our analysis at this point in the game are essentially identical in most aspects of clinical efficacy, so I will not be going over all the details that were presented this morning but just the essential components and making some brief comments.

In order to look at the clinical efficacy, we want -- I want to create for you a basic framework within which we will present our viewpoints and information and characterize the importance of safety as well as efficacy in terms of determining risk benefits analysis for this product. Following my brief presentation, Dr. Alma Davidson will speak to the issue regarding resistant S. pneumonia and Erythromycin resistant pathogens in terms of the indications that are being sought.

And then Dr. David Ross will talk about the overall general safety profile of Telithromycin with specific emphasis on QT issues. Dr. Edward Cox will then follow to talk about the hepatic effects and then we'll have a summary from Dr. Ross again about the risk/benefit profile of this drug.

To outline my talk, I will generally talk about the Phase III clinical data base, mention something about data that was censored from the Phase III clinical trials, clinical or bacteriologic efficacy and my talk will be mostly on clinical and not bacteriologic except the tonsillar pharyngitis indication is actually a bacteriologic end point and followed by an overall conclusion.

The Phase III clinical data base essentially consisted of 13 Phase III clinical trials for four indications; community-acquired pneumonia, acute exacerbation of chronic bronchitis, acute maxillary sinusitis and a Group A beta-hemolytic streptococcus tonsillar pharyngitis.

Essentially, across all indications we've got at least two control trials for every indication. I will be mostly focusing on the information that comes out of the controlled trials rather than the uncontrolled situations and for all these studies, some of them were conducted including U.S. patients but there were no studies that were based solely in U.S. patients, so I did not use the terminology, U.S. studies.

Also there are subtle differences from time to time in terms of populations, maybe the types of patients that are included, even though essentially all CAP patients, enrolled patients with CAP, but we have difficulty in just making -- pooling such results and making all studies being equal into -- being poolable. So from out viewpoint, we will present results with each study just being considered as it is and an overall summary of that information.

Essentially, a few sites were censored based on FDA investigation of clinical trial conduct, quality of data that was received for this Phase -- for the entire application. For the data that was submitted initially, for the first submission which was a year ago, there were four investigators that were censored by the FDA. And these investigators essentially were participants in other studies, other applications, not in Aventis that had some problems in terms of their data quality and data integrity and those investigators were then further looked at in relation to this application.

Those four that were censored accounted for a total of 186 patients and these patients were then excluded from all -- some were excluded from all indications except tonsillar pharyngitis and we excluded them from all our analyses and so did the sponsor and the sponsor agreed with us in terms of censoring this data. Phase III trials, the dosing of interest for three indications; acute exacerbation of chronic bronchitis, acute maxillary sinusitis and tonsillar pharyngitis was essentially for five day therapy and community-acquired pneumonia for seven to 10 days.

All studies had a designated test of cure window and had a similar design structure across all these indications and, of course, the test of cure window could vary in terms of a few days plus or minus, depending on which indications you are describing. But those were always pre-specified in the protocol and essentially were followed.

The primary efficacy populations for community-acquired AECB and maxillary sinusitis are both the mITT and PPc and contrary to probably popular belief, many people kind of assumed that we're only interested in looking at the per protocol -- clinical per protocol population but in fact, we do -- are always interested in the mITT analysis as well. So in my presentation, I will present both of those numbers. The mITT population is defined as all randomized subjects who met disease definition based upon clinical presentation history, bacteriologic and/or radiologic information and received at least one dose of study drug will be included in the mITT population. And this is a modified ITT because patients could be excluded only based upon baseline characteristics. Patients were not to be excluded based upon something that transpired during the course of the trial and that definition was used both by the sponsor and by us.

The mITT as well, that population, allows us to have two clear categories; you're either a failure or a success. There are no intermediate categories into that population. The protocol group, however, includes all mITT subjects minus those with major protocol violations and major protocol violations were always pre-described in the protocol ahead of time.

For community-acquired pneumonia there were essentially three clinical control trials; Protocols I, 6 and 9. The first one comparing Telithromycin to Amoxicillin, essentially had about five percent, 90 percent for comparator and 95 percent cure rate for Telithromycin. The second study, which -- in the protocol population. The second study which compared Telithromycin to Clarithromycin was about equal at 88 percent cure rate. And in Protocol Number 9, which was comparing Trovan, that one had 90 percent for Telithromycin and about 94 percent for Trovan, and of course, the sponsor did explain that that was a study that the -- when Trovan was restricted, that study continued as a single arm study later and so did -- the numbers in this study are a bit smaller.

What we notice is that the 95 percent confidence intervals for both the PPc and mITT were within a 15 percent margin and if we exclude the study 3009, the other two studies fell within a 10 percent margin.

Acute Exacerbation of Chronic Bronchitis; to be included in this trial one of the essential features was that subjects needed to have a documented history of chronic bronchitis and at least in Protocol 3003, there was an FEV1/FVC ratio of less than 70 percent that was used to capture patients with certain severity of illness and these tests had to be made within the previous 12 months prior to enrollment in the study. At a time of presentation, subjects were expected to have increased cough, increases sputum volume, increased sputum purulence and/or dyspnea. And cure was defined as resolution of all signs and symptoms and no subsequently antimicrobial therapy could have been administered prior to the test of cure date. In this study we did notice that even though I'm not discussing details about the bacteriologic efficacy, that in fact, most of the patients here, the most common pathogen was Hemophilus influenza and not S. pneumonia as one might expect from the literature.

In these two trials we see again, both the PPc and the mITT populations that the rates were somewhere ranging from 83 to 86 percent, similar for comparator as well as Telithromycin and the integral bounds in these confidence intervals were within the 10 percent lower bond margin that we've established to declare therapeutic equivalent.

For acute maxillary sinusitis, this is -- we're looking at the five-day of Telithromycin in compared to 10 days and in study 3005 there were two clinical -- two controlled studies here. In study 5 this result is only for the five-day arm compared to the 10-day arm. It was 10 days of comparators. It was a 10-day arm and a 10-day arm also met our confidence integral bounds of declaring equivalence. Both the mITT and the PPc and the mITT populations here showed fairly consistent results in terms of confidence intervals that we expected to see.

One should note, however, in study 3005 the five-day, a rate of just 75 percent consider it the natural history of that disease and that patients were also allowed during this protocol to have concomitant use of -- concurrent use of medications such as antihistamines and antipyretics, inflammatory medications. So, therefore, one should just bear in mind that the 75 percent cure rate itself is not necessarily that great. However, these results seem fairly consistent from among the two trials. In the second study, 3011, in that study the population was restricted to -- by case definition, to include most of the subjects who had at least seven days of symptoms at the time of enrollment while in the first study, that was not clearly the case and in fact, approximately 40 percent to my recollection, of subjects in that study presented with signs and symptoms that were within the seven-day window when one may suspect there is a high possibility of viral infection as opposed to bacterial sinusitis.

However, those differences were equal for both treatment arms. So both populations pretty much had about the same occurrence of both characteristics.

For the Group A beta-hemolytic streptococcal tonsillitis/pharyngitis, we want to make some basic comments, very few comments on this, regarding this indication and that when we're interpreting in the regulatory context our findings of efficacy, simply meeting a statistical criterion that you are within your minimum confidence interval bounds, is a minimum criterion that is usually necessary to meet but it is in no way sufficient in terms of for a drug to win. And there are other considerations that should be borne in mind.

For example, in this indication, penicillin is still the gold standard that we expect as a comparator and the primary efficacy for this is based upon microbiologic eradication and not necessarily just clinical impressions. And any product with an absolute eradication rate of less than 85 percent and the protocol population should not be considered first line therapy. And this is certainly within the spirit of the guidance that has been on the web for some time.

Tonsillar pharyngitis is also a mild diseases. A targeted population is typically children. There are many alternative therapies that are currently available for this indication. There was insufficient evidence of activity against Erythromycin resistance strep pyogenes and the risk benefit ratio must be considered very carefully in terms of when we put a drug on the market whether or not there is truly a benefit that outweighs the risk before it's approved.

Our overall conclusion is that FDA's efficacy analyses are consistent with those of the applicant's computationally and, of course, we do take other considerations into mind when we make conclusions about the utility of these numbers. And that adequate well-controlled trials must demonstrate both safety and efficacy in order for a drug to be approved to market. So I would then like to turn it over to Dr. Davidson, who will now continue the presentation and talk about the resistant pathogens.

DR. RELLER: Dr. Alma Davidson.

PRESENTATION OF DR. ALMA DAVIDSON

DR. DAVIDSON: Good afternoon. My name is Alma Davidson. I'm a med officer with the Division of Anti-Infective Drug Products and I will focus my presentation on the applicant's proposed drug resistant streptococcus pneumonia claims of Telithromycin. This is the overview of my talk which includes at the outset I will present the applicant's proposed labeling for resistance of Telithromycin. Then I'll talk about Penicillin resistant streptococcus pneumonia claim, including a brief review of regulatory history of selected antimicrobial agents which were previously presented before the advisory committee.

Next, I will review the Erythromycin resistant streptococcus pneumonia claim. I will make summary comments at the end of each section. I will be referring to the acronym PRSP for Penicillin resistant streptococcus pneumoniae and the acronym ERSP for Erythromycin resistant streptococcus pneumoniae in my subsequent slides.

This slide displays the applicant's proposed labeling for resistance claims, community-acquired pneumonia and acute sinusitis due to streptococcus pneumonia including strains resistant to Penicillin G and Erythromycin A. Now let us consider Penicillin resistant streptococcus pneumoniae claims beginning with the review of issues discussed by previous advisory committee and the regulatory history of selected antimicrobial agents.

The previous advisory committee considered several issues regarding potential resistance claims. Foremost was the seriousness of the disease, for example, meningitis and bacteremic pneumonia. Much of the previous discussions focused on community and hospital acquired pneumonias. In general bacteremic pneumonia carries a higher mortality rate and is a sign of invasive disease. In addition, documentation of a pathogen in the blood which is a sterile site, add certainty to the diagnoses. It was felt that an applicant should demonstrate efficacy for resistant pathogens in serious disease prior to claims and less serious indications.

Next the strength of evidence to support the proposed resistant claim for antimicrobial agent was discussed. I will review some of the data in the subsequent slides. Another issue is the relationship of the mechanism of resistance -- of the resistant pathogen to the mechanism of action for the agent being considered. These issues will lead to different considerations. For example, the so-called out of class agents such as the treatment of PRSP with quinolones or within class agents such as Agumentin. Finally, what is the impact to public health of such a claim.

For the remainder of my presentation, I will consider only community-acquired pneumonia indication, especially bacteremic community-acquired pneumonia as it represents a serious invasive type of disease. I will now review data which was discussed at the previous advisory committee meetings for Levofloxacin, Moxifloxacin and Linezolid.

This slide reflects the information that the advisory committee considered when reviewing Levofloxacin for the indication of community-acquired pneumonia with a PRSP claim. Within the MDA data base, a total of 250 -- 250 microbiologically documented cases of streptococcus pneumonia community-acquired were studied with a 98 percent success rate. Of this, 15 cases were due to PRSP with 100 percent cure rate. There were 55 cases of bacteremic pneumonia due to streptococcus pneumoniae with 100 percent cure rate. The susceptible cases and the resistant cases both had 100 percent cure rate.

Following from the data we just presented to the agency and the advisory committee, Levofloxacin was granted an indication of community-acquired pneumonia with PRSB claim. The text of the current label follows. "Community-acquired pneumonia due to streptococcus pneumoniae (including Penicillin-resistant strains) MIC value for Penicillin was greater than or equal to two micrograms per mL".

Now, let's turn over to Moxifloxacin. As we can see, the total experience for community-acquired pneumonia due to streptococcus pneumoniae was 89 cases with a 90 percent cure rate. In this application, only one study, Study 140, collected blood cultures from which the streptococcus pneumoniae was isolated. In this study six cases were due to PRSP, 10 cases had documented bacteremia and only one or two of them were due to PRSB. The Moxifloxacin label does not carry -- currently carry a claim for CAP due to PRSB.

Let's turn over to Linezolid. Here the applicant collected a total of 100 cases of CAP due to streptococcus pneumoniae with a cure rate of 88 percent. It is important to note that out of the 33 bacteremic cases, none were due to PRSB. The Linezolid label carries an indication for community-acquired pneumonia and specifically states that it is not to be used for PRSB.

Moxifloxacin and Linezolid were not approved for PRSB claim and the indication of community-acquired pneumonia due to insufficient evidence upon which to base this claim. Now, let's review the Telithromycin data submitted in support of PRSB claims for community-acquired pneumonia. This is a summary of data across controlled and uncontrolled studies. Less than five cases were documented among the comparator group. This slide shows that the total number of patients with documented streptococcus pneumonia isolates regardless of susceptibility was 174 with a clinical success rate of 96 percent. Of this 17 were due to PRSB. There were 38 cases of bacteremic pneumonia with a cure rate of 89 percent, compared to 67 percent among PRSB cases. It should be noted that there were only six case of PRSB bacteremia.

If we compare this experience to Levofloxacin, we note that the evidence is somewhat smaller with lower clinical cure rates among the bacteremic cases, especially PRSB. In summary for Telithromycin treated patients, the overall clinical success rate of community-acquired pneumonia due to streptococcus pneumoniae was 82 percent among 174 cases. Seventeen cases of PRSB and community-acquired pneumonia were documented. The majority of patients had mild to moderate pneumonia based upon fine scores. Two of the failures had severe infections with PRSB and were treated in the hospital setting. The other failure had a moderately severe infection and was also treated in the hospital.

There were six documented bacteremic cases of PRSB pneumonia with a cure rate of 67 percent. Bacteremic failures occurred in sicker populations. They both required hospitalization and additional intravenous antibiotics.

Now, let's switch gears to Telithromycin and Erythromycin resistant streptococcus pneumoniae claim of Telithromycin. This slide summarizes the clinical success of Telithromycin for ERSP claim and community-acquired pneumonia. You will notice that these numbers are identical to the numbers for PRSB cases. However, they are not the same patients. There are about 50 percent concurrence in cases of PRSB and ERSB. Again, the overall number of documented ERSB pneumonia cases was 17 with a cure rate of 82 percent. Likewise, the cure rate among the bacteremic ERSP cases was 67 percent.

Now, since the mef E genotype is the most common type in the United States, isolates carrying this gene will be considered in the subsequent slide. This slide summarizes the Telithromycin and Erythromycin MICs in community-acquired pneumonia with the mef E genotypes. The first column indicates the MICs of Telithromycin. The second column indicates the MICs of Erythromycin and the third column is the number of cases with the mef E genotypes and their data base.

Interestingly enough, as the MICs of Erythromycin increases, the MICs of Telithromycin also increased. You will recall that the applicant's proposed break point of Telithromycin sensitivity is greater than or equal to 1 microgram per mL. Although the number of isolates with the mef E genotypes are small, this observation of increasing MICs Telithromycin may raise the possibility of potential concurrent resistance between Telithromycin and Erythromycin or macrolides.

Finally, summarizing the Erythromycin resistant streptococcus pneumoniae experienced in community-acquired pneumonia. There are 17 documented cases of community-acquired pneumonia with ERSP with a cure rate of 82 percent. All three failures had a concurrent Penicillin resistance. Two of the failures had erm B genotypes. Six bacteremic cases with ERSP had a cure rate of 67 percent. Will cross resistance or concurrent resistance between Telithromycin and Erythromycin clinical isolates occur? That is the question. Dr. Ross will further consider the prospective of risk benefit assessment in his summary discussion at the end of the FDA discussion.

This ends my presentation. Thank you for your attention.

DR. RELLER: Dr. David Ross.

PRESENTATION OF DR. DAVID ROSS

DR. ROSS: Good afternoon. My name is David Ross. I'm in the Division of Anti-Infective Drug Products at FDA. I'm going to speak about the general safety profile of Telithromycin, followed by discussion of its cardiac effects. Dr. Edward Cox will follow with a discussion of hepatic effects of Telithromycin followed by discussion by Dr. Zachary Goodman of drug induced liver disease and then we'll finish with an overview of risk benefit issues.

Let me start by summarizing the Ketek Phase III safety data base. There were 4,985 patients who received at least one dose of Ketek or comparator. Forty-eight of these patients did not have post-baseline safety follow-up information leaving 4,937 patients in the safety data base, 3265 Ketek, 1672 comparator. In the nine control trials, there were 2,045 Ketek treated patients and 1,672 comparator treated patients. In the four uncontrolled trials, there were 1,220 Ketek treated patients.

In terms of extent of exposure, for patients who were treated with the five-day regiment, there were 1,429 patients and as you can see the mean exposure was about five days. For patients who are receiving regiments of seven to 10 days or 10 days, there were somewhat over 1800 patients in this group. Mean exposure was about nine days and for the entire data base of Ketek patients, mean exposure was about seven days. Mean treatment time for comparators was a little under 10 days.

With respect to deaths, there were no deaths in Phase I trials. There were 11 deaths in Phase III trials, 10 of these were in CAP trials, 1 in a tonsillar pharyngitis trial. This was a comparator treated patient who died of acute lymphoid leukemia. There were seven deaths in Telithromycin treated patients, four in comparator treated patients. None of these were directly attributable to drug. Six of the deaths, four in Ketek, two from comparator treated patients were scored as treatment failures.

With regard to primary or secondary causes of death, six out of seven Ketek treated patients who died had a cardiovascular cause, zero out of four comparator treated patients who died had a cardiovascular cause. Most of the CAP deaths occurred in patients who were at high risk for morality, that is Fine Category III or higher.

With respect to serious adverse events, this table shows SAEs in controlled trials, there were 40, that is two percent in Telithromycin treated patients and 41, 2.5 percent in comparator treated patients. The remainder of the table shows serious adverse events that were possibly related to treatment and these are listed here.

In the uncontrolled trials there were four SAEs that were possibly related to drug. These occurred in Telithromycin treated patients and included gastroenteritis, vasculitis, hepatitis and leukopenia. With regard to adverse events in the controlled trials the most common adverse events were gastrointestinal such as diarrhea, nausea, vomiting, dyspepsia, abdominal pain or abdominal LFTs as well as nervous system, headache, dizziness and blurred vision with was a special senses adverse event. For diarrhea, as noted previously, the rate was higher in controlled trials than in comparator and blurred vision has also been noted. These were predominantly younger women in tonsillar pharyngitis trials. In one case, the episode of blurred vision lasted for several days.

Because Telithromycin is metabolized in part by the 3A4 system, it was of interest to analyze adverse events according to 3A4 inhibitor intake as shown on this slide. It should be emphasized that this is an exploratory analysis since patients were not randomized on the basis of 3A4 inhibitor intake. For the most common adverse events, most -- for example, for diarrhea, in general, there was a higher incidence in Telithromycin treated patients who received a 3A4 inhibitor compared to those who did not. In addition, the ratio of incidences between Telithromycin and comparator treated patients was greater when patients received a 3A4 inhibitor than when they did not receive a 3A4 inhibitor.

My pointer is dying here. Let me move on to a discussion of cardiac effects of Telithromycin. I'm going to discuss the in vitro and pre-clinical data Phase I data submitted by the applicant, Phase III data submitted and then finish with conclusions. Let me just briefly speak to one issue which is the use of correction factors. In general, despite its limitations, QTc is the yardstick that we have used to detect signals indicating the potential for malignant ventricular arrhythmia such as Torsades.

Obviously, there's been a lot of work in terms of defining alternative correction factors. It's important to emphasize that these would need to be validated against an appropriate population. So for the remainder of my discussion, I'm going to focus on QTc. Let me start with the effects of Telithromycin on repolarization in vitro and in pre-clinical models. As has been noted previously, Telithromycin inhibits the Ikr channel which is the major repolarization current. The Ki or concentration at which inhibition is half maximal, is 42.6 micromolar. Lower Ki means more potent inhibition. You previously saw data for comparison that Moxifloxacin was 129 micromolar. You can compare this to concentrations seen in Phase I studies, in Phase III studies. The mean serum Cmax was 2.4 micromolar, and this is total drug. The maximum observed concentration in Phase III studies was 12 micromolar.

It's important to remember that these refer to serum concentrations. In a rat study conducted by the applicant, the myocardium had a higher concentration than serum in a ratio of six to one. So concentrations in myocardium may reach those of the Ki. In other in vitro models, Telithromycin prolongs action potentials in isolated Perkinje fibers. At the Ki there was a greater than 75 percent increase in APD90 a measure of action potential duration. It's also important to note that in a controlled in vitro model, Telithromycin potentiates Sotalol induced APD prolongation.

Finally, in a dog model, Telithromycin prolonged the QTc and increased heart rate. After IV infusion, QTc was increased by 30 milliseconds within one minute compared to 17 milliseconds for Clarithromycin. After multiple oral doses, the increase was 27 to 30 milliseconds.

Let me move onto the Phase I studies. Telithromycin showed an effect on QTc increasing it in both young subjects and elderly subjects, despite the entry of these rationally normal subjects not with underlying disease for both rows. The amount of increase showed dose dependence with higher increases at greater doses, 28 milliseconds at 2400 milligrams in young subjects, 19 milliseconds at 2000 milligrams in elderly subjects. All of these increases were statistically significant with respect to placebo.

In a study in changes in QTc in patients with underlying cardiovascular disease, at four hours after dosing Telithromycin at a dose of 1600 milligrams was significantly different than placebo. It's important to note that this peak effect occurred at four hours since the Tmax occurred at around 1.5 hours, plus the peak effect on QTc occurred after concentration peak was reached. I should note that one patient in this study had an episode of syncope. This was not felt to be related to cardiac dysrhythmia.

Results from Phase I studies pooled showed a lot of variability around the regression line but showed that there was a significantly different regression line from the mean, highly significant, with a slope of 3.9. In other words, for each milligram per liter increase in Telithromycin concentration, one would predict a 3.9 millisecond increase in QTc. And it's important to remember that in terms of the range of concentrations that were seen or that were likely to be seen in clinical -- in the real world.

The applicant conducted a study in which Telithromycin -- the effects of Telithromycin alone, Cisapride alone, and Telithromycin plus Cisapride were examined. You will remember tat Cisapride is a 3A4 substrate that has been associated clinically with Torsades and other arrhythmias. As you can see, the curves for -- blue is placebo, green is Cisapride, yellow is Ketek and pink is the combination. As you can see, the changes in QTc for Ketek and Cisapride were comparable. When given together, the two had at least an additive effects on QTc. Because of the metabolism of Ketek by the 3A4 system, it was also of interest to see what the effects of concomitant administration of a potent 3A4 inhibiter, Ketoconazole were on both Telithromycin pharmacokinetics and QTc effects.

When Telithromycin and Ketoconazole were given together, the increase in QTc compared to placebo was about 10.5 milliseconds which was statistically significantly different from placebo. In terms of the pharmacokinetics, administration of Telithromycin and Ketoconazole together increased the Cmax by about 50 percent. The AUC almost doubled. Because of the concentration dependence of Telithromycin's effect on QTc, it's important to understand factors that might effect Telithromycin concentration and therefore might effect Telithromycin's effect on QTc.

It's important to note that Telithromycin has non-linear pharmacokinetics. As the dose goes up, clearance decreases and this decreases our ability to predict what concentration or other pharmacokinetic results will obtain with altered doses or exposures. At a single dose of 800 milligrams, the mean Cmax was around two milligrams per liter. However, the maximum Cmax was over five milligrams per liter and this occurred in the subject with renal impairment. In a multiple dose study of 800 milligrams, again, the mean Cmax was around two milligrams per liter. The maximum observed Cmax was 6.66 milligrams per liter and this occurred in an elderly subject.

In population PK studies in Phase III the maximum observed concentrations were 7.6 to 9.9 milligrams per liter. It is important to keep in mind in assessing these, that these were not necessarily drawn at the peak at Tmax and plus may not represent true peak values. With regard to pharmacokinetics in special populations, elderly subjects in Phase I showed a doubling of Cmax in AUC. In subjects with hepatic impairment, in a single dose study, half life was increased by 40 percent. The applicant has conducted a multiple dose study. The final report from this study has not been submitted to the agency review as of this date but AUC and Cmax appear similar in healthy subjects. Although t-1/2 does not appear to be increased, because we have not had the opportunity to review this study in detail, we cannot comment on the reasons for this discrepancy.

What does seem clear is that renal clearance increases to compensate for hepatic impairment implying the potential accumulation of Telithromycin may occur if there's decreased creatinine clearance in the setting of hepatic impairment.

Finally, in a single dose study in renally impaired patients, in patients with moderate renal impairment, creatinine clearance of 40 to 80 milli-liters per minute. The Cmax was increased by a third, AUC was increased by 42 percent. In subjects with severe impairment, creatinine clearance of less than 40 MLs per minute, the Cmax was increased by 44 percent and the AUC was increased by 59 percent. Recall that these changes occurred despite the fact that renal clearance represents only about 13 percent of the total clearance of this drug.

So let me just summarize the results from Phase I and I want to just say that these analyses are the hard work of my colleague Dr. Jenny Zheng in the office of Clinical Pharmacology and Biopharmaceutics at FDA. Telithromycin showed a concentration in dose dependent increase in QTc. The QTc increase associated with Telithromycin was comparable to Cisapride and at least additive when the two were given together. The increase was enhanced by concomitant administration of a 3A4 inhibitor, Ketoconazole. The concentration of Telithromycin was also increased by a concomitant Ketaconazole administration. PK variability was seen in part due to non-linear pharmacokinetics.

There's the potential for increased exposure with age and renal impairment as well as the potential for increased exposure in hepatically impaired subjects with decreased creatinine clearance. Let me change from the sort of clean world of controlled Phase I studies to the somewhat more murky issues involved in assessment of QT changes in Phase III. It's important to recognize that the clinical events that we're looking for that are associated with changes in QTc are rare. For example, with Cisapride there were no clinical events associated with prolonged QTc in the NDA data base and as Dr. Ruskin pointed out, it was only after a large number of courses had been prescribed that a signal could be detected.

There is substantial variability in measurements of QTc. There are inter-individual measurement differences. There are differences between observers and there are differences for a given individual in measurement either because of biological variabilities such as circadian rhythm changes or for differences between measurements taken by the same observer and there are systems that have been described such as digitizing pads and the like for trying to minimize this sort of variability.

Finally, the significance of changes in QTC may not always be clear. Drugs that are associated with a small change in QTc may still be associated with Torsades. For example, the Terfenadine, which has a mean increase of only six milliseconds in healthy subjects, is well-known to be associated with Torsades because of metabolic interactions. In addition, the increase in risk for a given increase in QTc is not always clear.

Let me mention, moving from these general caveats, let me mention some specific limitations of the EKG data in the Phase III data set for this application. EKG data was not collected on all patients. In controlled trials, there was EKG data from 1,515 Ketek treated patients, 1,276 comparator treated patients allowing calculation of changes in QTc intervals. There were relatively few data from high risk patients and this was in part because of the design of the trials in which there were multiple exclusion criteria that would have left out such patients. For example, in terms of patients with EKG data allowing QTc intervals to be analyzed, there were two Telithromycin treated patients with a baseline potassium of less than three.

There were five Telithromycin treated patients in the EKG data set who were on anti-arrhythmics. In addition, the number of patients with higher concentrations was relatively limited, decreasing the ability to analyze patients -- QTc changes in such patients. The timing of EKGs in the Phase III studies may not have corresponded to the peak QTc effect. As I noted earlier, the peak QTc effect in a number of Phase I studies occurred at four hours. However, the protocols called for EKGs to be obtained at one to three hours after dosing. Thus the EKGs obtained may not have estimated the peak QTc effect.

EKGs were also obtained at different times after dosing further increasing the variability. EKGs were obtained at 25 millimeters per second. A number of studies examining QTc prolongation have used chart speeds of 50 millimeters per second to increase resolution. Finally, there was no data available on serum magnesium, hypomagnesemia is a recognized risk factor for Torsades.

This just shows some of the exclusion criteria for Telithromycin Phase III trials and these included conditions such as long QT syndrome, severe hypokalemia and a variety of concomitant medications. As Dr. Benedict noted previously, a number of these criteria were dropped part way through the development program. However, it's important to note that despite this, there still remain relatively few patients in some of these risk groups. For example, there were only six Telithromycin treated patients receiving Digoxin in the controlled trials.

For other drugs that potentially interact such as protease inhibitors, there were only three patients in the data base who received protease inhibitors. The analyses I'm going to show focus on the controlled Phase III trials. I will not discuss the uncontrolled trials. In order to control for variability, we've tried as much as possible to compare like with like. In the set of patients from all controlled trials for whom EKG data was available to calculate QT changes, on therapy changes in Telithromycin treated patients were two milliseconds. There was a net decrease of .7 milliseconds for comparators.

For the majority of the demographic groups analyzed, changes in the Clarithromycin were greater than those for comparators. We specifically compared Telithromycin with the macrolide used in these studies, Clarithromycin and we compared those trials in which there were Telithromycin treated patients compared with Clarithromycin treated patients, that is, studies 3006 and 3008. For all patients together, the magnitudes of the QTc changes were similar. However, they were slightly greater for female patients, 3.7 milliseconds for Telithromycin treated patients, 2.3 milliseconds for Clarithro treated patients.

For elderly patients, the increases were 5.3 milliseconds for Telithromycin treated patients, 1.6 milliseconds for Clarithromycin treated patients. Because Telithromycin is metabolized by the 3A4 system, we analyzed QTc changes in patients who received concomitant 3A4 substrates as well as 2D6 substrates. Again, I'd like to caution that these are exploratory analyses since patients were not randomized on the basis of co-administration of these substrates. For Telithromycin, for patients who did not received a concomitant 3A4 substrate, the change in QTc on therapy was 1.3 milliseconds. For those who did receive a 3A4 substrate, it was 3.2 milliseconds. Both of these changes were greater than for comparator groups.

For those patients who did not receive 2D6 substrates the increase was 1.4 for Telithro, negative 1 for comparators. Again, receipt of a 2D6 substrate increased the QTc by 5.3 milliseconds for Telithromycin, 0.7 milliseconds for comparators. In patients who received a drug or drugs that were both -- that were 3A4 and 2D6 substrates, the increase for Telithromycin was 6.9 milliseconds, 3 milliseconds for comparators.

We performed the same analysis comparing Telithromycin with Clarithromycin and again, these are studies 3006 and 3008. Again, we saw the same pattern. If patients did not receive a concomitant 3A4 substrate, Telithromycin had a QTc increase of 3.1 versus 2.7 for Clarithro. If there was a 3A4 substrate, the increase was 4.1 versus 2.9. For patients who received concomitant 3A4 and 2D6 substrates, the increase for Telithromycin was 11.5, for Clarithromycin 5.4.

We also examined outliers, focusing particularly on individuals who had increases of more than 30 milliseconds. Looking at all controlled Telithromycin trials, the number of patients who had increases of 31 to 60 milliseconds for Telithromycin was 7.3 percent versus 5.7 percent for comparator. The difference is not statistically significant. A similar analysis for Telithromycin versus Clarithromycin showed 7.9 percent for Telithromycin, 6.8 percent for Clarithro, again, the differences are not statistically significant.

So let me summarize the conclusions from these data. Telithromycin inhibits repolarization in vitro both in the cell culture model looking at the IKR channel and isolated Purkinje fibers. Data from a rat model suggests that the myocardial Telithromycin concentration may approximate the Ki for these effects. In a dog model, Telithromycin significantly increased QTc with both oral and IV dosing. The effect of IV Telithromycin was greater than that of IV Clarithromycin.

With regard to Phase I, Telithromycin increased QTc in controlled cross-over studies and this was a consistent effect. The effect was concentration and dose dependent. It was comparable to Cisapride and at least additive with Cisapride and increased by co-administration of 3A4 inhibitor. With respect to PK variability which might effect Telithromycin concentration and therefore, its QTc effect, Telithromycin shows non-linear pharmokinetics and showed increased exposure in special populations. Finally, in Phase III Telithromycin increased QTc. This was a small but consistent effect. The increase was larger than with comparators, including Clarithromycin and exploratory analyses suggested possible interactions with 3A4 and 2D6 substrates. As an example in trials comparing Telithromycin with Clarithromycin, the mean increase for Clarithromycin with both 3A4 and 2D6 substrates was 11.5 milliseconds compared to 5.4 for Clarithromycin.

Let me stop here and turn the podium over to my colleague, Dr. Edward Cox.

PRESENTATION OF DR. EDWARD COX

DR. COX: Hello. I'm Edward Cox. I'm a medical officer at the FDA. And I'll be providing the agency's presentation of the hepatic effects of Ketek. And just to give you an overview of my talk, first I'll talk a little bit about some of the pre-clinical studies with Ketek and then move on and describe some of the events in the Phase I studies in humans and then move on to the Phase III studies and discuss the hepatic adverse events as seen and then move on and talk about the analysis of laboratory data and then a discussion, a brief discussion of some of the serious adverse events that we're seeing.

And when I get to the point of describing the serious adverse events, I'll ask Dr. Goodman to come up and describe some of the histopathologic findings from one of the patients

-- or from the patient who had a liver biopsy. And then after Dr. Goodman's presentation, I'll come back and just briefly summarize the findings from the hepatic effects.

And first, just to start out, with regards to the pre-clinical studies, hepatotoxicity was seen in rats, dogs and monkeys and this hepatotoxicity was manifested as increases in AST and ALT. Hepatic necrosis was seen in the four-week rat study and hepatocellular hypertrophy and multi-nucleated hepatocytes were seen in some but not all of the pre-clinical animal studies. And our FDA pharmacology/toxicology reviewer had the opportunity to go back and review the original data that was submitted with the Clarithromycin NDA in order to be able to compare the effects seen with Telithromycin with that which was seen with Clarithromycin and the conclusion from the review was that the hepatic effects of Telithromycin were more than what was experienced with Clarithromycin.

And now what I'd like to do is just run through essentially a dose response curve from the Phase I studies in humans. And just to start out, I'll describe the layout of the table here for you. In the right most column we have the Ketek dose in doses ranging from 50 milligrams up to 3,200 milligrams and I've lumped some of the lower doses together here just to compact the size of the table. We also have in the very bottom here, the data for those periods during which subjects received placebo medication. In the middle group of columns here are the results from the single dose studies and then in the right most column are the results from the multiple dose studies. With regards to hepatic adverse events, I'd like to call your attention to this column here which shows the percentage of events that occur for any particular dosing period and if we move from low levels of dose up to the dose of 2,000 milligrams, we do see a clustering of events here at 2,000 milligrams. Then moving onto higher doses, we do see somewhat of a fall-off.

With regards to hepatic adverse events in the multiple dose studies, we see that the events there were infrequent. I will describe some results from one of the Phase I studies, Study 1030 which included eight elderly subjects and who received doses up to 2,000 milligrams and this is the highest dose that elderly subjects received during the Phase I studies.

The study included single doses of 1200 milligrams, 1600 milligrams and 2,000 milligrams and then interdigitated between these doses was a placebo period. The doses were separated by approximately a one-week period of wash-out. There were three patients who achieved increases in ALT and AST with levels ranging from approximately 100 to the levels of 300 with ALT being greater than AST. And these three patients, the first was a 72-year old female, who seven days after receiving a 2,000 milligram dose of Ketek, demonstrated an increase in her ALT and AST.

As part of her work-up for viral etiologies of hepatitis, this patient also had a CMVIGM that was positive. The second patient who in this particular study developed increases in ALT and AST, was a 69-year old male who 17 days after receiving the 2,000 milligram dose of Ketek experienced increases in his ALT and AST.

And the third patient is a 62-year old male who seven days after receiving a placebo which was also, because of the nature and design of the study, 14 days after that patient received a 2,000 milligram dose of Ketek, experienced increases in his ALT and AST and as part of the serologic evaluation for etiologies of hepatitis this patient had an EBVIGM that was positive. The -- the viral serologies that were done when looking at the full set of data available there, do not provide definitive evidence of diagnosis of a viral etiology and I think, you know, one of the other points to be made here is that this could represent a possible drug effect. In such a situation, we'd be talking about a drug effect that would have a seven to 17-day latency period given the chronology of events here.

With regards to hepatic adverse event rates from the Phase III studies, the adverse event rates that were experienced were similar for both Ketek and comparators. The rates for treatment continuation were similar for Ketek and comparators. With regards to serious hepatic adverse events, from the comparative studies, there were two Ketek treated patients who experienced serious adverse hepatic events and one comparator treated patient who experienced a serious hepatic adverse event.

From the non-comparative studies there was one additional Ketek treated patient who experienced a serious hepatic adverse event. With regards to hepatic deaths, here were not deaths that were attributed to drug induced hepatic injury. And now before coming back and talking about the serious hepatic adverse events, I'd like to describe some of the laboratory evaluations that were carried out to -- within the Phase III studies. And I'll focus on evaluations from the comparative studies in patients who are normal at baseline for AST, ALT and T. bilirubin.

I'll look at essentially what are ladders of AST and ALT elevation and I'll only present data from the CAP studies and in the CAP studies there were more AST and ALT elevations at the on therapy and post-therapy visits in the Ketek arm. For these corresponding time points in the non-CAP studies, Ketek and comparator were similar.

And I'll just run through the design of this slide. We're looking at AST changes that occur at the on-therapy visit which is day 2 to 5 in patients who are normal at baseline from the community-acquired pneumonia studies. And we're looking at changes in AST and the ladder here goes from those patients who have values of less than or equal to one times normal and then at intervals increasing up to a level of greater than five times normal.

In this column we have the results for the number and percentage of Ketek treated patients who achieve these levels of elevation, then for the comparator treated patients, similar with the number of comparator treated patients and the percentage of patients achieving that level. And what I'd like to do is to call your attention to this row here where the levels of elevation are between one times and less than or equal to two times the upper limit of normal where we see a slightly greater proportion of elevations in the Ketek treated patients.

Now, again, a slide of very similar design. We're still looking at AST. However, this is at the post-therapy visit, so at day 17 to 21. We're looking at changes in AST and again, the same ladder for Ketek and comparator treated patients and if we look at the percentage of patients experiencing this level of elevation of one to two times normal, we see that six percent of Ketek treated patients achieve this level, whereas two percent of comparator treated patients achieve this level and we see a few events occurring at levels beyond the two times normal category.

Now we're moving onto ALT. We're back at the on-therapy time point of day 2 to 5 and so we're looking at changes in ALT again the same ladder and for Ketek treated patients and comparator treated patients, if we look at the same row that we've been looking at here between 1 and 2 times the normal, we see 11 percent of Ketek treated patients achieving this level of elevation and nine percent of comparator treated patients achieving this level of elevation and then a few events in both arms at higher levels.

Now, ALT changes occurring at the post-therapy visit, day 17 to 21, and again, the same ladder as we've been looking at for ALT and if we look at the level of one times to less or equal to two times normal, we see 12 percent of the Ketek treated patients achieving that level and nine percent of comparator treated patients achieving that level.

And now within your packet there's a loose slide that should replace the one that's currently in there for slide 12. This table looks at combined laboratory abnormalities of ALT, AST and T. bilirubin at the level of greater than the upper limit of normal and less than two times normal and if we look at the right most column, we have the category of lab analytes where we have combined AST, ALT and T. bilirubin and then either ALT and T. bili or AST and T. bili and we have the number of patients achieving that level of elevation for the Ketek treated patients and for the comparator treated patients.

I should note the Ketek treated patients includes patients both from the comparative and the non-comparative studies. And we see we have five events for all three analytes, five for ALT and T. bili, one for AST and T. bili and from the comparator studies, the comparator treated patients we don't have any events there.

And then with regards to combined abnormalities, the late Dr. Hy Zimmerman, in his book, "Hepatoxicity", stated that drug induced hepatocellular injury with overt jaundice is associated with a morality of at least 10 percent. This phrase has come to be known as Hy's law within the agency and as a surrogate for this, within NDA data bases, we often times look at AST or ALT greater than three times the upper limit of normal combined with a T. bilirubin elevation of greater than 1.5 times the upper limit of normal.

And in the Ketek treated arm and the comparator treated arms, there were no patients that met this criteria strictly. Now, there is -- there are some patients in the Ketek arm I'd like to comment on. The first is the patient who has an ALT elevation of 19 times the upper limits of normal and a T. bilirubin of 1.55 times the upper limit of normal. Now, the is patient also had an ALT at 81 at baseline, so he did have a slight elevation and this is the same patient for whom Dr. Goodman will be describing the pathology on the liver biopsy shortly.

There were two other patients who didn't quite achieve the level of elevation of three times and 1.5 times the upper limit of normal but were close. One of these two patients had a mild increase in alkaline phosphatase. And now what I'd like to do is just describe the cases of serious adverse events that occurred during the NDA data base. And one of the reasons that I'm spending some time going over these cases and describing them in some detail is that within the NDA studies, you know, the safety data bases really are not powered to find infrequent occurring events so I think going through these cases may provide us some insights.

The first serious adverse event I'll describe is a comparator treated patient. This patient was a 61-year old male with community-acquired pneumonia a history of congestive heart failure and alcoholism who was maintained on Digoxin. He was treated with Clarithromycin, 500 milligrams po BID for 10 days. He was noted to be jaundiced on day 17 and as part of his evaluation had a CT scan and an ultra-sound examination that showed changes consistent with a disseminated neoplasm thought to be of either hepatic or renal origin.

And I've provided some of his abnormal lab values below. His peak T. bili was approximately five times the upper limit of normal and his alkaline phosphatase was also approximately five times the upper limit of normal. His AST and ALT were normal. And then this is the first serious hepatic adverse event from the Ketek studies, a 76-year old female with community-acquired pneumonia, a history of hypocholestral anemia and hypouricemia who's maintained on Pravastatin and Allopurinol chronically. She received treatment with Ketek, 800 milligrams po daily on days 1 through 6. And then in this table in the right most column we have the laboratory analytes. Next to them, their corresponding normal ranges. And we note that at the time that this patient was enrolled in the study, she had a slightly elevated AST. She receives therapy on day 5. Her AST and ALT are elevated at AST of 295, ALT of 418 with a T. bili and alkaline phosphatase that are just slightly elevated.

On day 6 she stops her Ketek therapy. On day 7 we see her AST and ALT returning towards normal and continuing to do so at her subsequent visit.

The second serious hepatic adverse event from the Phase III studies involved a 19-year old male with tonsillar pharyngitis who had a positive culture for Group A Beta hemolytic strep and no significant past medical history. The patient was treated with Ketek, 800 milligrams daily from days 1 through 5 and then on the evening of day 12, there's a history of heavy alcohol consumption. And the similar design to the table on the last slide, on the right most column the analytes and we see he had normal liver function tests when he began the study. We have the completion of Ketek therapy at day 5, the history of alcohol intake at day 12 and then on day 13 we see the bumps in AST and ALT with the increase in AST being greater than that of what was experienced with ALT and then subsequent resolution over the next few visits.

And this is the case the Dr. Goodman will be describing the pathology on shortly. This is the case of a 53-year old male from Finland with a -- who was admitted to one of the CAP studies who had a history of asthma and diabetes mellitus and he was maintained on inhaled Salbutamol, Fluticasone, Attrovent, Nasonex and oral calcium. There's also a history of Acetaminophen intake that began on day 13 and it's described in the case report forms as the intake of six times 500 milligram tablets of Tylenol over one week. The patient was treated with Ketek 800 milligrams daily, days 1 through 10 and then on day 14 he developed an illness that included fever, vomiting, diarrhea. This was an illness that was similar to an illness that other members of his family were experiencing at the same time. The difference in the patient's illness was that his fever persisted.

Now, I'd like to go through his laboratory studies, again, a similar table. We've added the Eosinophils here at the bottom of the table and we have normal ranges. The normal range for Eosinophils, we don't have a laboratory specific normal range, so we're using the typical normal range of less than 500 cells per micro-liter. And on day 1 we note a mild increase in ALT of 81 and the Eosinophil count of 774. The patient receives Ketek day 1 through 10 and then returns on day 21 for a follow-up evaluation after he's had this febrile illness with persisting fever. He is noted to have an ALT of 354. We see further increase in his Eosinophil count and then on day 24 his ALT achieves a maximum for the course of this particular episode of hepatitis of 1529 and I've also included data from day 35 here when his ALT is down to 518 and this is the maximum Eosinophil count attained during this episode.

As part of the patient's evaluation for his episode of Hepatitis, the patient had serologies for Hepatitis A, B and C that were negative. He went on to have a liver biopsy on Day 29, and Dr. Goodman will describe the pathology there shortly. And then the patient's ALT had normalized by three months.

Now the patient went on and had a second event of Hepatitis, and this was noted approximately eight months after the first event at a routine follow, when the patient had an ALT value of 1331. As part of this evaluation, the patient had Anti-Smooth Muscle Antibodies titers drawn there were positive one to one thousand. The patient was also noted to have an elevated IGG and IGA. With the second episode, Eosinophilia was not present, and a second liver biopsy was also attained for the second episode.

And now I'd like to turn the podium over to Dr. Zachary Goodman, who will describe the findings from the liver biopsies from this patient.

DR. GOODMAN: Well, you'll see that on the schedule I'm listed as giving a lecture on drug-induced liver disease, but I'm really going to focus on the liver biopsy and drug-induced liver disease. And I should say at the start that liver biopsy is not usually done in somebody in whom you strongly suspect drug-induced liver disease. If a patient is on a drug, develops liver test abnormalities, and you stop the drug and the test abnormalities go away, then a liver biopsy is not indicated.

The times that liver biopsies are done is when there's a confusing clinical situation, when it might be one thing or another; or when the diagnosis of drug-induced liver disease is not entertained. And when we're talking about drug-induced liver disease, we're not talking about usually intrinsic toxicity, but an idiosyncratic reaction; and so what do you see in the liver biopsy in somebody who has an idiosyncratic drug-induced injury? Well, it could be anything.

One of the principles and one of the points to be made is that drugs can mimic just about anything that can happen in the liver, anything that can happen in any naturally occurring liver disease can happen in drug-induced liver disease; so when you get a liver biopsy from somebody in whom drug-induced injury has occurred, it can show just about anything. You can have an acute injury, or you can have a chronic injury. And the acute injury can take the form of hepatocellular injury, a cholestatic injury, a mixture of the two, or some sort of vascular injury. And a chronic injury can be a chronic hepatocellular injury; that is it can be a chronic hepatitis, you can have chronic cholestasis, you have granulomas disease, fibrosis or cirrhosis, a vascular injury, or tumors. And so in other words, the drugs can mimic absolutely anything.

So just to give some examples from the area of antibiotics; some examples, tetracyclines typically cause microvesicular fat in a dose related more intrinsic toxic injury, but sometimes tetracyclines can cause chronic cholestasis, although very rarely; and tetracyclines can cause chronic hepatitis, particularly with minocycline.

And one of the principles of recognizing drug-induced injury is that the same drugs tend to do the same thing over again; that there's a certain range of patterns that are seen with each particular agent in which drug induced injury is recognized.

For amoxicillin and calvulonic acid, typically that causes cholestatic injury; but there can also be an element of cholangitis, or it can be combined with hepatocellular injury, or sometimes granulomas. Nitrofurantoin has been around for a long time, long enough to establish a relative incidence of injury; and it's estimated that about one in every 3,000 individuals who takes nitrofurantoin will develop liver injury. And the injury can take the form of various forms of acute hepatitis, of acute hepatocellular injury about 30 percent of the time, chronic injury about half of the time, cholestatic injury in about 10 percent, and other 10 percent is miscellaneous things.

So then the question arises when we've got a liver biopsy from somebody, when should we suspect that a drug might have been the cause; and one answer is easy, we always suspect it since drugs can mimic anything that can happen in the liver. If there's not an obvious other cause, then we always inquire about what drugs the patient was taking. But we're especially suspicious of a drug-induced injury when there's some sort of atypical pattern; that is, something that's not usually seen in the usual range of liver diseases, so such things as a combined hepatocellular and cholestatic injury, that is cholestatic hepatitis that can happen in viral hepatitis. But in a liver biopsy performed in a hospital in the developed world when you see a cholestatic hepatitis, it's much more likely a drug than viral hepatitis.

Granulomas hepatitis; both granulomas and hepatocellular injury, sometimes that happens in zarquoidosis, but it's more likely to be a drug. And especially if we see a hepatitis that has a lot of eosinophils with it, not absolutely 100 percent of the time, but usually it turns out to be a drug. And when you have a really severe injury, and particularly one in which there's zonal necrosis, that's also most often due to a drug.

So let - excuse me - let me show a few examples. Could we dim the lights just a little bit? Does anybody know where the light switch -- for those of you who have been out of medical school for a while, I'll remind you of what normal liver histology looks like.

You'll recall that a liver has portal areas and it has central veins, and the portal areas have portal vein branches, hepatic artery branches and bile ducts, and the blood comes into the liver at the tissue level through the vessels of the portal triades, and percolates through the sinusoids of the liver where the business of the liver takes place by the hepatocytes, and the blood leaves the liver through the central veins.

Now some important points are that the area around the central vein has the blood with the least oxygen and the least nutrients, so it's most susceptible to several types of injury. It's most susceptible to ischemic injury.

The area around the central vein is also where the cytochrome P450 enzymes are localized, so toxic injury often takes -- has its greatest effect in the central lobular area, and something that's not apparent normally is that there's a canalicular network through while bile is secreted and flows in the opposite direction to the blood, is exited through the bile ducts. And the bile is an energy -- bile flow is an energy dependent process; so when there's a cholestatic injury, it also shows up first in the area around the central vein by which - as I've had it explained to me - that's because it has the furthest to go from that area to get out; so that we always pay close attention to the central lobular areas.

Now let me show some examples from some diseases that are drug induced. Here's a liver biopsy. This is a little bit lower power. This is portal area here, a central vein here, and another one here, and another one over here some place. And what we have here is severe necrosis of the liver cells, and particularly concentrated in the areas around the central veins; the areas that are most susceptible to these types of injury. That's zonal necrosis, and when we see that, we always suspect a drug. This happens to be a biopsy from a patient who developed injury after Halofane anesthesia.

Here's a liver biopsy from a patient who had been taking trimethaprinsulfanoxizal. This is a portal area, very high magnification. The rest of the liver tissue is out here, and there was spotty necrosis, hepatitis going on within the parenchyma, but right in the portal area we have a granuloma. These are histiocyes here. Here's a giant cell; and a granuloma has a number of eosinophils.

Now when I see this, the first thing I would ask is well then what medications was the patient taking. Granulomatous disease along with the eosinophils strongly suggests a drug.

Here's a liver biopsy from a patient who was taking amoxicillin and clavulenic acid. There's a portal area down here, the central vein is up here. This is a cholestatic injury. You can see it looks brown; that's because there's lots of bile in the area in the area around the central vein, but there's also other injury going on as well. There's spotty necrosis of liver cells and inflammatory cells within the parenchyma, and inflammation in the portal area.

Now Dr. Cox had asked me if I thought the case that the patient who was taking Telithromycin was reminiscent of what was seen in trovafloxacin. Well, I got this off the Internet. This is the only published picture that I'm aware of trovafloxacin and induced liver injury. There's a central vein here; that's why the quality isn't very good, it looks kind of grainy. But there's a central vein here, and there's an area of necrosis and loss of liver cells surrounding it, and a lot of inflammatory cells, and the arrows are pointing to some of the inflammatory cells. These are almost all eosinophils.

This is another -- a biopsy from another patient who was traovafloxacin. This is one from the files AFIP. There's a portal area here, there's a central area here, and another one over here; and both of these central areas have confluent necrosis and collapse, and a lot of inflammatory cells, and I'll show a higher magnification of that one. It's shown here. The central vein is buried in here some place.

We have confluent necrosis and a very recent necrosis, and all the liver cells are gone, and there are a lot of inflammatory cells; and at higher magnification they're a mixture of cells, not like the ones that were in the picture that I just showed, but there's some granulomatous inflammation. There are histiocytes here. Here's the giant cell. There are many eosinophils, and there are also a lot of plasma cells.

Now this is the first liver biopsy from the patient in Finland who developed the injury after stopping the trovafloxacin -- I mean, I'm sorry; after stopping the Telithromycin, and then had his biopsy at about the time that his eosinophil count was peaking, and there's a central vein underneath all of this, and another one under this. It is reminiscent of the trovafloxacin case though, because there's a lot of confluent necrosis in the central areas. There's also a lot of spotty necrosis all throughout the biopsy. And if we look at this at higher magnification, here's the central vein here, and lots of inflammatory cells; and there are a mixture of cells here. There's histiocytes and almost all of these other cells are eosinophils, which we can see here at high magnification.

Now this is very unusual. I'm not aware of any non-drug related naturally occurring liver disease that really looks like this. It doesn't look like autoimmune hepatitis. It doesn't look like viral hepatitis, so with the history that the patient was taking a new drug that hasn't been recognized as the cause of liver injury, and then seeing this soon afterwards, I would say that this is almost certainly drug-induced; despite the fact that he had abnormal liver enzymes before he took the drug.

Now one thing that helps us recognize whether there was an underlying chronic liver disease is to look at a connective tissue stain; and this is the one that was done in Finland. This is a serious red stain. It's used a lot in Europe, but not much in this country; but collagen stains red, and if this patient had underlying chronic liver disease, we would expect to see some significant amount of fibrosis. But here's a low power of his liver biopsy with a serious red stain. There's a little bit of fibrosis around some of the portal areas; and this is the same one at higher magnification, but really not much - so if he did have an underlying chronic liver disease it must have been extremely mild.

Now the next slide I'm going to show is going to show this biopsy in the same place, but over here we'll see a connective tissue stain that was done on his second biopsy nine months later. And this is using the Masson Stain which is the one that we use in this country, and they used in this case in Finland. The Masson stains collagen blue, so anything that was red over here would be blue over here nine months later; but in addition, we've got all of this other blue here. That's an awful lot of fibrosis that's occurred in the ninth month interval. It looks like he's got a pretty significant chronic liver disease at this stage; and looking at that at higher magnification, there's a portal area over here which has got some fibrosis around it.

This is the central area here, and another one down here, and another one down here. Most of the fibrosis is actually in the central lobular areas, which is the same areas that were injured, that had all the eosinophils nine months earlier. Now that's very unusual.

Looking at the history it looks like this ought to have been autoimmune hepatitis. The patient has hypergymoglobual anemia. He has an anti-smooth muscle antibody, and he's got this liver biopsy but it's not quite typical of autoimmune because of the location of the injury.

Here are a couple of fields. There's a portal area down here. This area appears central where a lot of the fibrosis is. Same thing over here; this is central here, portal over here. And here's the portal area at high magnification.

Now in autoimmune hepatitis the brunt of the injury is usually portal. You have a lot of piecemeal necrosis around it, and a lot of inflammation in the portal areas, but here there's actually more injury going on in the central areas.

Well even so, it -- I think if I were to put one diagnosis on it, it looks like it probably should be autoimmune hepatitis, but it is very curious that it followed in the same areas that the patient had the injury nine months earlier. And another feature of autoimmune hepatitis is that typically the major inflammatory cell is a plasma cell, and if we look at these at higher magnification, almost all of these are plasma cells; no eosinophils. So all I can say is this case is really unique in my experience. I'm not aware of any really well documented case of autoimmune hepatitis that have followed a drug-induced injury, but this I suppose could be the first one.

I think looking at the overall case, I think the first liver biopsy certainly looks like drug-induced liver disease. The second one, I have to admit, I'm not really sure what exactly is going on there.

DR. RELLER: Dr. David Ross will now summarize the FDA presentation before a question and answer period. Dr. Goodman.

DR. COX: Thank you, Dr. Goodman. This is Ed Cox. I'll just summarize the hepatic effects, and then Dr. Ross will be up to provide the further summary. So with regards to the hepatic effects, as seen in the pre-clinical studies, we saw hepatotoxicity in dogs, rats, and monkeys. And then in Phase I, the clustering of hepatic adverse events at the 2000 milligram times one dose that I described. There was no clear dose response for hepatic adverse events in Phase I.

Then if we move on to the Phase III adverse events, there were similar adverse events rates for Ketek and comparators; no apparent drug-induced hepatic deaths. And then with regards to the hepatic serious adverse events from the Phase III studies, there are two that appear to be plausibly associated with Ketek. One of these events was the case that Dr. Goodman just described the pathologic findings on where a patient had central lobular necrosis and eosinophilic infiltration on pathology, and also was noted to have an elevated ALT and eosinophils on day one.

And then we also went through the AST and ALT ladders, which showed elevations in Ketek treated cap patients who were normal at baseline, that were more than what was seen in the comparator treated patients. This was not seen in the studies with patients from studies other than community-acquired pneumonia. And then we also looked at the concomitant combination low level elevations in AST/ALT and T.bili that occurred only in Ketek treated patients.

And now I'll turn the podium back over to Dr. Ross.

DR. RELLER: Thank you, Dr. Cox.

DR. ROSS: I'd like to summarize the -- by giving an overview of risk benefit issues.

This is data that Cynthia Whitney and her co-workers published in the New England Journal in December, showing the prevalence of non-susceptible pneumococci for various antinecrobial, so this includes both intermediate susceptibility, pneumococci and fully resistant. And as everyone knows, we're seeing a steady increase over the late 90's for penicillin non-susceptibility, as well as Erythromycin non-susceptibility. And clearly, we're concerned about this. It represents a public health problem.

I think it's important though for us to try and define as much as possible what the impact of this phenomena -- phenomenon rather is, because that's really what defines what the benefit is for targeting resistant pathogens, and allows us to weigh that benefit against the risk of new agents. And this -- weighing this benefit by determining the impact can be very difficult.

In the retrospective studies that were cited in the applicant's presentation, the one by Feikin did not show an increase in mortality associated with penicillin resistance, unless you excluded all patients who died during the first four days.

The study -- retrospective study by Turett consisted of retrospective cohort, a retrospective population that was -- had a significant number of HIV positive patients. That's not to say that that -- those analyses are not useful, but it does indicate that it can be very difficult to tease out exactly what the impact is of penicillin resistance.

As another example of this, let me cite data that Pallares and his collaborators published in '95 from Barcelona. When they looked at patients with pneumococcal pneumonia, either for all cases or cases with concurrent bacteremia, penicillin resistance -- and here this meant isolates with an MIC of .12 or greater, so it included isolates that were -- had intermediate susceptibility.

Penicillin resistance was not an independent risk factor for mortality in a multivariate regression analysis. Even if you take out conditions that are associated with the infection itself, such as shock or involvement of more than one lobe, penicillin resistance was still not an independent predictor of mortality.

Looking at data from the same publication, if you look at fully resistant isolates; and these numbers are small, you do not see a dramatic difference in mortality for patients treated with penicillin or ampicillin, despite the fact that they have infections with resistant isolates, compared to patients treated with other antimicrobial agents.

This is not to say that penicillin resistance is not a problem; only that we need to define the impact carefully if we're going to define the benefit, and weigh the risk.

In assessing benefits of claims for efficacy against resistant organisms, we need to think about the public health impact. What is the benefit in infections due to resistant pathogens, and if we're talking about out-patient therapy, what is the benefit in out-patients who have mild to moderate disease, who do not require hospitalization. How serious is the infection?

If we're talking about community-acquired pneumonia, are we talking about patients who are at higher risk for mortality, such as Fine Category V, or patients who are lower risk for mortality, such as Fine Category I. Are we talking about a condition with lower mortality and morbidity that's associated with it, such as AECB, or an infection for which the antibiotic treatment effect is less.

What is the mechanism of action? Are we talking about out-of-class -- a mechanism that targets out-of-class resistance, where one would not expect biochemical changes that would result in resistance, or are we talking about in-class resistance, where the mechanism of action may be related to the mechanism of resistance. And if we are talking about in-class resistance, what is the potential for concurrent resistance? In the case of Telithromycin, as Dr. Davidson discussed, there's the issue of Erythromycin MICs rising, with Telithromycin MICs appearing to show a shift at the same time. And then finally, what is the evidence for clinical efficacy versus resistant isolates? What is the weight of evidence in general for infections caused by susceptible or resistant isolates? What is the weight of evidence for infections due to resistant isolates? And what is the evidence for efficacy in invasive disease due to resistant isolates, such as patients with pneumococcal pneumonia with concurrent bacteremia? And we say this for two reasons.

First, as we all know, these are patients who are at higher risk for a poor outcome. Secondly, as Dr. Davidson pointed out, patients with a positive blood culture are those for whom we have greater certainty as to the identity of the true pathogen, compared to patients where we only have a positive sputum culture.

Let me turn to the risk side of the equation. I've shown this data before but this is in a somewhat different format. This shows the change in QTC for Telithromycin when given with ketoconazole, and shows that it's statistically different from placebo. And as Dr. Ruskin pointed out this morning, we need an assessing risk for cardiac repolarization abnormalities to be particularly alert to the possibility of amplifying events that could increase the risk for cardiac repolarization related events.

It's also important to recognize that along with metabollic interactions, electrophysiologic interactions could play a role. This is a list of drugs that are associated with the QTC prolongation or Torsades. A few of these have been withdrawn from the market, but the majority are still in wide clinical use.

Along with the effects that other drugs may have on Telithromycin, we need to consider the effects that Telithromycin may have on other drugs. In Phase I studies when Telithromycin was co-administered with other agents, the Cmax of simvastatin was increased by 433 percent in the area under the curve by 761 percent.

With regard to the pharmacokinetics of the major metabolically active metabolite of simvastatin, the beta hydroxy acid, the concentration was increased by -- Cmax was increased by 1400 percent, in the AUC by 1100 percent.

For digoxin, a drug with a narrow therapeutic index, the increase in Cmax was 70 percent, and AUC 37 percent. For midazolam the increase in Cmax was 162 percent, and the increase in AUC was 511 percent.

What is -- what are the clinical implications of drug interactions? It's instructive to look at the example of Mibefradil which was marketed under the trade name of Posicor. This was a calcium channel blocker that was a potent inhibitor, and still is a potent inhibitor of 3A4, and in vitro inhibited the metabolism of various statins to greater or lesser extents.

It's important to note in the development program for mibefradil, in patients who received mibefradil concomitantly with a statin, no associated adverse events were seen. Mibefradil was approved in August of '97 for treatment of chronic stable angina, and hypertension. The label given on approval carried warnings regarding a number of drugs that could interact with mibefradil because of the concern over cardiac repolarization.

After approval, reports starting coming in of adverse events associated with interacting drugs. Within four months after approval, additional warnings were added to the label regarding use of lovastatin or sinvastatin in particular, as well as use of any statin, and concomitant use of tacrolimus cyclosoporin, along with other labeling changes regarding the potential for bradycardia.

Finally, less than a year after approval, mibedradil was withdrawn from the market. More than two dozen interacting drugs have been identified. There were continued reports of adverse events from co-prescribed interacting drugs. Because of the number and complexity of these interactions, further labeling changes were felt to be impractical. And despite the nature of the indications, it was not felt that the agent had any special benefits relative to other agents. So with those considerations, let me outline some of the safety concerns with regard to Telithromycin.

There is the potential confluence of multiple risk factors. The effect on QTC, the concentration dependence of this QTC effect, and the pharmacokinetic variability that can affect concentration, and therefore affect QTC, arising in part from non-linear pharmacokinetics.

There is the potential for hepatotoxicity as outlined by Drs. Cox and Goodman, that could have an affect on hepatic clearance, and also affect concentration. There's the potential for increase in exposure in elderly patients, in those with hepatorenal disease, as well as potential for exposure with concomitant medications. These factors together could potentially lead to clinically significant changes in QT intervals.

It's important to note that we have very limited data on at-risk groups. There were few patients in any given risk group, and it's also important to note that there is a limited range of concentration data in the Phase III studies. There were relatively few patients who had a concentration of five or over, despite our concerns over pharmacokinetic variability; so we don't really know what might happen in patients with higher concentrations. I'd also mention again that these concentrations were not necessarily achieved -- did not necessarily represent Cmax.

In addition, there's the independent concern over potential hepatotoxicity; an uncertainty regarding how the drug should be dosed in patients in special populations, given altered pharmacokinetics in the elderly and other special populations in the non-linearity of the pharmacokinetics.

There are the potential affects that Telithromycin could have on concomitant medications, such as the statins. And then finally, there is -- even though the -- there were no events associated with QT prolongation, or co-administration of medications in Phase III studies, it's important to remember that absence of evidence in these studies does not equal evidence of absence.

As Dr. Soreth pointed at the beginning, these studies are typically not powered to detect weak signals. And given the potentially wide population exposure of drugs for respiratory tract infections, these needs to be kept in mind.

This is data that Linda McCaig published a few years ago in JAMA showing prescriptions on an out-patient basis for various upper respiratory tract infections in millions of courses. And as you can see, there were in '92 over 80 million courses of antimicrobials prescribed on an out-patient basis; so this is the sort of exposure that we need to keep in mind.

Let me stop here. Thank you for your attention; and my colleagues and I would be happy to answer any questions that the Committee has.

DR. RELLER: Dr. Bell.

DR. BELL: Thanks. I wanted to pick up on your excellent comments about the need for additional outcome data on resistant infections. The CDC Surveillance Project of pneumococcal infections and resistance has just recently started to routinely collect outcome information, but I wanted to also comment on the two studies that you mentioned, the Feikin and Pallares, which I'm sure you've read more recently than I have. But the Feikin study in which you pointed out that the beneficial impact -- or rather the adverse impact of drug resistance on mortality was seen only after the first few days. I think it -- you know, historically my recollection is that even when penicillin was first introduced, the difference in mortality was only apparent after the first days because the idea being that there's a certain percentage of people who come in with overwhelming infection, or they have some debilitated status, and so I think it would be unfair to suggest that an effect limited to after the first few days is -- you know, is questionable.

In the Pallares study -- now my recollection of this also is that his was mainly -- you know, the issue here for pneumococcal resistance and pneumonia is that the microbiologic nomenclature of the break points; this is really an issue it's resistant for meningitis, but in these patients who I think were hospitalized and treated with high doses of penicillin, they could easily exceed what was needed for pneumonia. So of course, those patients they wouldn't see any resistant pneumonia difference, but I think what we're talking about here is out-patient treatment orally of mild to moderate pneumonia. And I believe that the Erythromycin or macrolide levels might not be able to exceed so much the concentrate -- the MICs, but maybe you could comment on that.

I guess I just think that -- you know, I agree with you that we need much better outcome information, but I took away a sense that your -- you know, you may have been questioning that there was adverse outcomes.

DR. ROSS: No. Sir, if that was the impression I gave, I apologize. I think -- first off, with regard to the paper by Feikin, I think my point is not to say that that is a bad study. One of my former attendings is on that paper, so I better not say that.

It is simply to say that I think we need to be careful about drawing conclusions. You're absolutely correct that the original studies on mortality in pneumonia showed that the difference really occurs later, rather than earlier. My point is not to say that there's not an effect shown, because there is. I think one thing, for example, from the paper by Turett that appeared in CID, given the number of HIV positive patients in that study -- and I don't remember that study that well, but the -- one thing you might be able to speculate about is, is that a population that's at particular risk for mortality. So the intent is not to say that these studies don't tell us something useful. I think they definitely do. The ques -- the idea is to be just as precise as possible in the conclusions about benefit and impact that we draw.

DR. BELL: If I could -- just one follow-up; that I think one piece of information that I find a little troubling is that, as was briefly mentioned I guess by Dr. Seidlin in the company's presentation; it's not only that the Erythromycin resistant rates for pneumococci have been -- well they essentially doubled from '95 to '99, from 10 percent to 20 percent, but the median MICs of the most common form here, the M phenotype have increased from four to eight; which means that a large percentage of these isolates are in the range where at least anecdotal treatment failures for pneumonia have been reported with macrolide, so although this is anecdotal, you know, I think it makes us a little nervous.

DR. ROSS: Well, I agree with you. I think that, you know, as I said this is multi-drug resistance in streptococcus pneumonia. Streptococcus pneumoniae is a real public health concern. The point as cited in the data from Pallares was not to say well they didn't find anything, we can all go home because that clearly is not the case.

I think one of the issues that we would like to lay before the Committee is exactly what is the impact of this.

DR. RELLER: Dr. Murray.

DR. MURRAY: I had a question that maybe our GI Hepatology colleagues can help clarify a little bit.

Well, I guess in the case with the hepatitis, I have this concern that with elevated eosinophils, and LFTs or at least one of the parameters that before therapy was given, that I have a little trouble making cause and effect with the drug. But even if we did -- so this is an idiosyncratic sort of reaction that might be seen? I mean, this not a dose related; this is --

DR. GOODMAN: That's right.

DR. MURRAY: Okay. And do you have any sense of how that would compare with other sorts of rates of idiosyncratic reactions either to liver for drugs that are out there, or for anaphylaxis, or some other sort of idiosyncratic reactions that you can't really predict on a dose related basis? I mean, you gave the figures, or somebody did, for nitrofurantoin, but what about -- are there any data for some other drugs?

DR. GOODMAN: That's the only one that I'm aware of that has an estimate. Perhaps Dr. Lee knows of some others.

DR. LEE: Yeah. I mean, I think most drug reactions that are idiosyncratic range from anywhere from one in a thousand for isoniazid, to one in ten thousand for other drugs, to one in fifty thousand for terbinafin, so it's in this range. So if you see a signal in whatever it is, 3200 cases, that's remarkable. And by signal, I mean a case as was described that would satisfy Hy's law that there was an increase in bilirubin and transaminase above the limits that they said.

DR. MURRAY: Are you concerned with the fact that the LFT, one of the LFTs and the eosinophils were elevated before the patient ever got the drug; if indeed the baseline drug levels were drawn.

DR. LEE: I'm not really concerned about that. I mean, it's notable but the numbers were quite small. The eosinophilia was trivial I think, compared to what happened later.

What I think is of concern here, perhaps, is that these -- this occurred late in this one case; and yet it doesn't make it unlikely. It still would be in the realm of drug toxicity, and I think the biopsy was very compelling. The question I had was what were the -- how many late ALTs does the company have, or was in that data? I know there was a -- there were values on Days 17 to 21 I guess, but I think that's key when we look at all these values; is this case unfolded after Day 21.

DR. SEIDLIN: There were two levels that were drawn after the first acute episode where the levels had gone back to 53, and they were at 53 twice several months after the first episode. And then there was a period of time where no levels were drawn before the next routine draw when they were elevated.

If I may, Dr. Willis Maddrey is here and he's reviewed the case with us; and I'd appreciate it if we could have the opportunity to invite him to the microphone as well, as he's quite experienced in drug-induced liver injury, and I could ask him to comment on that. Would that be acceptable?

DR. LEE: Yeah. My question really was in the overall studies, how often did you have late values, how often were you monitoring after 21 days?

DR. SEIDLIN: We only did that if there had been an elevation on therapy. We did not routinely monitor them afterwards.

DR. LEE: Right.

DR. SEIDLIN: Could I invite Dr. Maddrey to the microphone?

DR. RELLER: Please.

DR. MADDREY: This is a remarkably complicated case. As I read this case, and I've had the opportunity to read it on a number of occasions and see the material, I would agree with the concerns that Dr. Goodman expressed early. However, this patient was on a corticosteroid according to the protocol at the time the study started, which should have kept eosinophil numbers down. And the fact that the eosinophil numbers were up, and the amino transferases were up, make it a complicated case to start with.

I would agree that the changes shown on that first box here are surely compatible with a drug-induced liver injury of some type. However, it would be, I think, most unusual for a person to come totally to normal, as this patient apparently did; and then some number of months later show up with another type of liver disease. And Dr. Goodman clearly showed us; I would think this patient does have an autoimmune hepatitis on that second biopsy. The weight of the evidence would suggest that. However, the thing that probably bothers me most is this is the only case like that in this series; and I would be very careful. I realize that that's not -- if this were an absolute straight forward case, as we've seen in a number of other conditions, I would be of more concern than the fact that this is a complicated case for the reasons outlined.

Two different types of liver disease, eight months off the drug before the second type appeared; and of course, no one can say that it's not a continuum, because you don't have all the numbers - but I am concerned of defining this drug based on this one case.

I would comment even further that we have many drugs on the market that would show percentages of elevated amino transferases in the one to two-fold range that would be similar to what's seen here. I would refer to isoniazid. I would refer to most of the non-steroidal anti-inflammatory drugs.

DR. RELLER: Dr. Archer.

DR. ARCHER: I'm very confused about the disparate QTC values from the sponsor in the FDA; and I was wondering if I could get some clarification.

For instance, the interaction with both cisapride and with ketoconazole, the numbers -- prolongation numbers were different in the two. In one case, it actually went down with ketoconzole in the sponsor's numbers, and it was an additive effect in the FDA's analysis.

Second, I'd like the FDA to comment on this first, if you don't mind; and then you can get your shot. The second, in their table with subjects with cardiovascular disease, the sponsor's QTC value was 6.5, and then in the table that the FDA had of elderly patients, their value -- this is with 1600 milligrams was 12, which is double. And I just don't understand how the same numbers were looked at by two different people, and come up with different values. Could somebody help me out with this?

DR. RELLER: Dr. Ross.

DR. ROSS: I think part of the reason for the discrepancy is the use of the different correction formulae. The -- because that's in one case -- in other words, the QT interval divided by the square root of the RR interval, and then there's the use by the applicant of the QTn formula, QT divided by - correct me if I'm wrong here, Dr. Benedict, RR to the power of 4.3. Is that -- so I think that --

DR. ARCHER: Well the ones that I just quoted though, the elderly patients versus the cardiovascular disease patients were both quoted as QTC.

DR. ROSS: I think the other issue that comes up is whether these numbers looked at -- whether the analyses included all ketek-treated patients, or only those from control trials; because on average, the mean change for patients from uncontrolled trials in our analysis was 0.5. It was a decrease of 0.5 milliseconds; so in that -- the difference between that and the ketek-treated patients from the controlled trials was statistically significant - so for that reason we did not combine the two.

DR. ARCHER: Okay. Well -- and the other one about the ketoconazole data. I mean, whatever correction factor you used, their's went down, and your's was additive. Their's is on -- let's see, in there analysis I believe, page 57. And it -- they used a different value, but their's went from 10 with ketoconazole, three with Erythromycin alone, and then down to nine, so it was actually lower than ketoconazole. And your data there was a doubling.

DR. ROSS: In this instance I would -- the heading for this column is QTn, so I would say it was use of the different formula, because in our analysis we used QTc.

DR. ARCHER: Oh, I understand that, but you mean that change in correction actually change the mag -- changed actually the direction?

DR. ROSS: Yes. I would --

DR. LAZZARA: Could I comment on that?

DR. ROSS: Yes, I'd like to --

DR. LAZZARA: About the heart rate changes. The QT direction is heart rate dependent. That's very treacherous and one of --

DR. SEIDLIN: Could you please speak into the microphone.

DR. RELLER: Yes. Dr. Lazzara is commenting, and we'd like to have all of our cardiology consultants weigh in on this very important issue.

DR. LAZZARA: My colleagues will have something to say. The -- this -- the bigger perspective from my stand point of having really scrutinized this problem now for over 10 years since we've been trying to extract this very low level signal of a one to three milliseconds up to six milliseconds, from a noise level that's in the order of 20 to 60 milliseconds. The standard deviations are many several times the signal we're extracting so -- and it's attributed to the biometricians, which I am not, and the biostatisticians, that we have been able to show significance of these small signals. We do in the terfenadine case for example.

But then when one compares the signals, it becomes even more problematic, a 3.6 versus a 2.5. It becomes virtually meaningless. So the question then becomes heart rate correction among the many factors that cause a variability, the heart rate correction is very key; particularly in a drug that changes heart rate. And if a drug amplifies heart rate, it weighs the data against itself in terms of the correction for an amplified heart rate by the QT, by the Bazett as one can see from the graph, will make the change in QT longer. The Delta QT will be greater, so you might end up with a four millisecond, or you'd end up with a one millisecond if the heart rate changed. So my question was going to be, before you brought up this very astute observation, neither group gave us the heart rate changes in any of this data that I could find; so to what magnitude did the heart rates change in the group as a whole, and in elderly versus the -- because it's a very key issue.

All of these differences could be explained by heart range changes and a different correction factor. That slope which is -- and that's been recognized -- by the way, that did not just appear in the company's data analysis, although their group verifies. That's been recognized for a long time. There's this inverse slope when one uses the Bazett which makes for that false correction. Everybody is recognizing what we -- as Dr. Ruskin pointed out, we continue to use the Bazett by force of habit. So do we have any idea what the heart rate change --

DR. ROSS: Well let me separate the answer into two parts. The answer is in terms of what the heart rate change is for the Phase III Studies; no, I think the applicant probably does have that.

I think that for that -- that's one reason that we had concentrated so heavily on the Phase I Studies, in which there was -- for example, the crossover studies where there were conditions where you could try to minimize, but certainly not eliminate variability.

I don't know. Dr. Throckmorton, would you like to say anything, or not?

DR. THROCKMORTON: Well this isn't something that, you know, sort of just appeared with this drug unfortunately, as Dr. Ruskin talked about this morning. I'm Doug Throckmorton. I'm the Deputy Division Director in the Cardiorenal Division, and have the distinct fortune or misfortune as it is to appears sometimes when these issues come up.

It is an uncertainty, and in a sense it's -- we're concerned about it because the changes that are reported are relatively small. Just to remind you, we were able to see effects of -- for cisapride versus placebo without correcting for heart rate at all; those sorts of things. So small effects can be detected here.

Having said that though, it does seem that the correction factors are critical to some of the at least directionality of the differences that have been reported.

I don't think this is the place necessarily to go into that, because the Center has no stated policy. We recognize it as something that needs to be addressed; but also just to remind you that we haven't applied any of these more novel corrections to any of the databases where we know we've had problems before; so it's attractive to believe that if we correct for this inverse slope that you raised, that will make the analysis more sensitive. We don't know that. That's -- we're imposing our own biases on the data to do that.

DR. LAZZARA: Well, no; there's a bias imposed both ways. The Bazett formula makes a correction. It changes the value, and the question basically is if a drug does not affect heart rate intrinsically, then it becomes a non-issues.

DR. THROCKMORTON: My point is that we've applied Bazetts to detect problems in the past. We haven't applied any other analyses to do that, so a conservative approach is to use the Bazetts, recognizing that that may not be the best approach. Now I'm certainly not going to try to defend it, but Jeremy or Art Moss has done a lot of -- have done a lot of work on that as well. They may want to comment.

DR. LAZZARA: But let me contest that though before they -- before I turn it over to them. It's -- it makes for a false change, and a false interpretation if the drug affects heart rate. Now if you apply the Bazett to a drug that doesn't intrinsically affect heart rate, then the changes would not be significant; at least the drug-induced changes. You might have changes that are not correct in terms of random variation in the normal population, but the drug-induced changes would not be significant. But if the drug significantly increases heart rate on a consistent basis, then it would introduce false data, or false changes in the data.

DR. RELLER: Dr. Lazzara, if I understand correctly, you're saying that one issue in trying to understand this is whether the drug under consideration itself has an affect on heart rate.

MR. LAZZARA: On heart rate; yes.

DR. RELLER: And is there any information that ketolides or macrolides intrinsically affect the heart rate itself?

MR. LAZZARA: Well that was my question. They had said earlier that it did, but no values were given; so I guess it's time for the -- someone --

DR. ZHENG: I think I have the information about heart rate increase in the baseline study. Actually, for the baseline study, eight baseline study pooled together, and we can come off model how much heart rate would increase with concentration, if we use the Emax model. The maximum increase in terms of heart rate is 14 beat per minute; but if we use the Concentration 2, which is the mean maximum concentration, after single dose, 800 milligram single dose, the increase of heart rate was predicted to be 7 beats per minute.

DR. LAZZARA: Seven beats per minute, and the Bazett formula would explain away the entire differences, of one versus three milliseconds. She uses the Bazett versus some other formula.

DR. RELLER: Dr. Ruskin.

DR. ZHENG: I think I have another comment about that QT correction. I understand the QTn is generated by the population data; and if you use that corrected formula to apply each study, and if you look at each study separately, you would say QTn in some of the study is over-estimated. In some of the studies, under-estimated; so if you use the population correction parameter, it may not be appropriate for each study.

Actually I did look at the Qtn versus RR interval for each study. Actually, for the study 1045 is the ketoconazole study, Qtn is slight under-estimate, QT prolongation. Same thing for the study 1040 -- I guess 1041 which is the cisapride study; QTn also at high heart rate, under-estimated QT interval. I think it's all because the QTn is generated based on the population data, instead of each study. I mean, placebo data and baseline data for each study, so if you use the population, and then for some of the study is over-estimated, for some of the studies under-estimated. When you pool them together you'll see a flat line.

DR. RELLER: Dr. Ruskin.

DR. RUSKIN: I guess the -- one has to derive the -- this change for a normal population really off drugs so that you know -- in the population -- at baseline; that's the way that the thing would make sense. And so the larger the population the better you would be, but obviously it would be off -- also be better to have a QTn that would apply to the elderly population, to the young population. But if the drug consistently increases heart rate, the Bazett will over-estimate consistently. That's an observation that was made prior to our discussions here, prior to this database; will consistently over-estimate the change, the corrected QT.

DR. ZHENG: Yes. I don't think I'm arguing if the Bazett equation is over-estimated when heart rate is increased. I simply just tell the Committee my observation about the QTn.

DR. RELLER: Dr. Soreth.

DR. SORETH: One point for Dr. Lazzara. You made the comment, does it consistently increase heart rate; and I point you to Study 1049 which was a study of subjects who had underlying cardiovascular disease but had no current infection. They were given various doses of Telithromycin. Those patients did not have an increase in heart rate, so I --

DR. SEIDLIN: those patients, if I might interject, were patients with cardiovascular disease, in contrast to the normal volunteer subjects. It was really the normal volunteer subjects who had the biggest increase in heart rate; and that's when Dr. Benedict showed the data, he used the Qtn in that population because they were the ones who had the biggest heart rate change, in contrast to the Phase III Studies where we did not see much of a heart rate change; actually saw a slight decrease, and that's why we chose to present it in that fashion, and that's how the correction factor was used.

DR. RELLER: Doctor, when you say, Dr. Seidlin, the normal volunteers, was that a placebo controlled assessment?

DR. SEIDLIN: The Phase I Studies were a variety of different studies. Some of them were healthy volunteers. They were healthy young, healthy elderly. There were a variety of others as well.

Perhaps I should Dr. Benedict to the microphone so that he can give you the particulars. The correction factor, however, was always developed on the placebo period, so that was before drug exposure; and that's where the correction factor was generated.

Dr. Benedict, do you want to comment?

DR. BENEDICT: Yes. I think the question from the panel member was outstanding. I think he asked the right question, and I think the response from Dr. Ralph Lazzara is equally important; and I hope Dr. Jeremy Ruskin will make his comments.

I think in terms of full and fair disclosure, I think Dr. Ross did present what they would rather obtain with the QTC value, which is -- I think is perfectly legitimate. I think where we come into conflict is do we consider the fact that Telithromycin had an intrinsic affect on the heart rate?

For example, let me give you the heart rate team that Dr. Ralph Ross has about what happens in the ketoconazole study, can have the heart rate response in the ketoconazole study. You will see that -- no, no, that's 1049. We'll come to that in a minute. We'll answer Dr. Soreth's question in a minute. Let me have the ketoconazole data.

You will see there when you administer Telithromycin there's a heart rate increase of approximately -- no, you have a curve that shows the heart rate response; if you can pull it up I can give the information to Dr. Ralph Lazzara, there's about an increase in heart rate of what -- yes, please.

All right. If you can focus at the bottom here; this is the heart rate response with Telithromycin in this trial, so as Dr. Ralph Lazzara very correctly pointed out, as you administer testimony the heart rate tracks it very nicely, showing that when you start off with low heart rates in the subjects, they started off in about the 60's, they go up to about 70, 72. And here now if you use the QTC as the correction formula, then you are going to sort of show higher and higher QTC changes.

To answer Dr. Jenny Zheng's question, we took both approaches. In other words, we did look at the population approach as well as an individual approach. Since we had placebo data, in other words, data from the drug-free period, we used the drug-free period from this population to derive the exponential for this particular population, and then we applied that uniformly to the placebo, as well as to the cisapride, and as well as to the -- sorry, to the Telithromycin period, and as well as to the ketoconazole period. And that's what the data that we shared with you; so clearly as Dr. Throckmorton very clearly pointed out, these are -- we are looking to what is happening. Clearly, we have a drug that has an affect on heart rate, so in terms of looking at relative changes, we wanted to take to the affect of the heart rate out, and that's what we presented to you as a percentage.

DR. ARCHER: Can I just --

DR. RELLER: Dr. Archer.

DR. ARCHER: Well, this is actually a fairly crucial point because the FDA makes the claim or the bringing up the point that there's an interaction between CYP3A4 inhibitors like ketoconazole and your product, which increases the QTC in their analysis. In your analysis, there is no interaction that increases the QTC, and it's all based on how you derive the QTC or the QTn. What's right?

DR. BENEDICT: Okay. Maybe I can again, if it is okay with you, point to one of the presentation Dr. Ross made towards the end when he summarized the changes.

Notice that when we looked at the changes in the QT interval by different cuts, whether it is greater than 60 millisecond or between 30 and 60 milliseconds, or 20 millisecond, et cetera; where he commented upon the fact that when you look at the comparators and Telithromycin, those individual who had a change or a prolongation in QT interval between 30 and 60, they were comparable to each other. And what he didn't point out is we did not have, or very, very few cases who had a prolongation in QT interval greater than 60 milliseconds.

Let me recall to the benefit the panel the excellent presentation by Dr. Ruskin, as well as the excellent presentation by OPDRA where they said when they looked at the cases of Torsades, on an average there was 172 millisecond prolongation in the QT interval. And of course from Dr. Ruskin's percentage and we heard from his own database, you need to have approximately about a 500 millisecond change, or absolute increase in QT before it relates to it.

Of course, he gave the appropriate caveat about the representativeness, the validity, et cetera, et cetera; but the important thing I would want to share with you is that we are still dealing with a small number, as Dr. Ralph Lazzara mentioned, in a fairly large variability on the population with respect to QT measurement. For example, in one particular individual, within 24 hours a QT can vary by 40, 50 milliseconds.

DR. RELLER: Okay. Now a patient Dr. Ruskin now speaks. Dr. Ruskin.

DR. RUSKIN: Thanks. I just -- a couple of comments. I just want to offer a correction note to something that Dr. Benedict just said.

You implied that there was a change of 500 milliseconds required for drug-induced Torsades based on the data that I showed; and that's not the case. Those data just suggest that many, and perhaps most cases of drug-induced Torsades, when there's a QT or QTc reported, have an absolute QT or corrected QT that is 500 milliseconds or greater. It's not a 500 millisecond increase obviously, and I assume that you didn't mean that.

With regard to the correction formula, I agree with everything that's been said. The problem -- the key issue I think has been framed by Dr. Archer, and unfortunately I don't think there's an answer to it. The question is which is correct, and I don't think anybody knows.

I agree that the Bazett formula when there's an increase in heart rate of 15 or 14 beats a minute, is going to give you an over-correction. You will get falsely prolonged QTc intervals; but the point that Dr. Ross and Dr. Throckmorton made about the fact that all the historical data is based on Bazett is also important, because we have no validation of these other formulae in other databases; so we're to some extent flying blind, and that's part of the conundrum.

The other question that I had for the sponsor is that if you have a drug that is increasing heart rate, but having not much affect on QTC, the absolute QT should go down. And I haven't been able to find any data on the raw QT interval; so is it possible for you to show us some of that, that is uncorrected by any formula, just the raw QTs, both mean data, mean max, and outliers, just for QT interval uncorrected.

DR. SEIDLIN: Okay. I'm going to ask Dr. Benedict to do that. There also is data in an exercise induced model where one can compare QT intervals at fixed heart rates, and we'd be happy to show that data as well. Dr. Benedict.

DR. BENEDICT: Again, that's a good question. I think we can look at what the straight QT does, and at this point in time I don't know whether we have a slide to show the QT data; if we have, can we pull that up. But while the slide is coming up, what I would like to respond is given the fact that there is a heart rate response that affects the QT interval, there have been different measures to try and get away from the effect of QT interval. Would Dr. Jeremy would like to see the data that we have in this regard while we are trying to get the QT data?

DR. RELLER: Why don't you assess the -- Dr. Benedict find, if it's available, the uncorrected QT data, and in the interest of time, I'd like to, while that's being done, pursue two other questions for our cardiology consultants to help the panel out.

DR. BENEDICT: The slide is up. Can I share that with you now?

DR. RELLER: Go ahead.

DR. BENEDICT: If you don't mind. Yes, please.

Okay. Here is the uncorrected change in the QT interval, and here is the slope for this equation, about one millisecond per microgram per milliliter of the drug from all the eight Phase I studies, ranging from concentration 80 to 3.2 grams. And here is the variability in the data, and here is the -- we have not drawn the line but there is the slope.

DR. RUSKIN: Do you have a similar data for heart rate that's graphically the same way?

DR. BENEDICT: I think we -- we will look for it. I don't know whether we have one.

DR. RUSKIN: So this is independent of heart rate then; right? This is the raw data.

DR. BENEDICT: Yes.

DR. RUSKIN: Which would suggest unless I'm reading it wrong, that despite the fact that the heart rate goes up, and I don't know if it's a concentration-dependent affect, but if it is, it suggests that the raw QT continues to go up. Does it?

DR. RELLER: You will need to get up and use the microphone, please.

DR. RUSKIN: Okay. So you're saying that there's a slightly -- the slope slightly less than one, so it's going down slightly with increase in concentration. Okay. And do you have -- it looks to me like no change basically.

DR. SHIN: Can I just clarify --

DR. RELLER: Yes, please. Please introduce yourself at the microphone who you are, and so that we have this for the record.

DR. SHIN: You know, that's --

DR. RELLER: Your name, please.

DR. SHIN: My name is Jun Shin. I'm from DMPK Adventis. That slope -- that plot shows the raw QT changes against the concentration. Actually if you model that with the lead-in model, you will see an active slope, but here we show the positive slope is because we take the heart rate as a co-variant. Therefore, the -- because this drug increase the heart rate, and that slope shows the positive or very minimal; just like the Qtn, whereas the corrector -- heart rate correction, the QTn is about one millisecond per microgram per mil. Here we show same thing; these two corresponding each other. Just don't -- it is not -- there is a heart rate effect confounding here, which is not -- cannot be projected on these two dimension plots.

DR. RUSKIN: Can I get a clarification of that? I mean what you're saying -- don't go away. I mean, what you're saying is that this model already is corrected for a heart rate because you did a regression in which you included heart rate. Is that what you're saying?

DR. SHIN: Yes. That heart rate as a co-variant. Actually we --

DR. RUSKIN: So then there is -- it -- well there's a correction for heart rate with this model.

DR. SHIN: We have our two dependent -- I'm sorry, two dependent variables here, heart rate and QT. And we have the concentration of the independent variable. We model this way; therefore we have that slight positive slope.

If we model the data QT against concentration strength without heart rate, then I guarantee you there is negative slope.

DR. RUSKIN: Can you show us that?

DR. RELLER: Dr. Ruskin asked for the uncorrected QT; and it's available we'd like to see it. If it's not available we must move on for the sake of time because what we have is not what was requested.

Two questions for our cardiology colleagues. The data, whatever its validity that the FDA showed having to do with QTc intervals, and drug interaction were consistent at least by my understanding with the QTc affects demonstration in vitro in the pre-clinical data. In contrast to whatever the validity, the corrected QTCs in this various corrections proffered by Aventis.

Based on what we know from the science, should the pre-clinical and in vitro match up with clinical findings, or do we have a mismatch there? How is the Committee to interpret in their considerations the questions soon to be addressed; these differences? That is, what role should the in vitro and pre-clinical, and should it match up with clinical data?

DR. THROCKMORTON: It should match up. I think maybe the question you're asking is what should we believe about the human data based on the pre-clinical data and the totality of the human evidence. I think the sponsor and the FDA are in concordance that this product affects cardiac repolarization. If the sponsor wants to disagree with that, but I -- we've come to that agreement in the past; so based on the pre-clinical evidence and based on the material in the clinical data set, I believe the conclusion that we've agreed to is that in fact it does affect cardiac repolarization.

The issue is to what degree, and if you want me to sort of philosophize for just a moment about that. Within the Agency, the observation has been that products that have a large affect on cardiac repolarization; that is that at doses used in the clinic or doses that people achieve regularly, there are big affects on mean QT; 25 milliseconds, 30 milliseconds, that sort of thing.

Those drugs are also associated with clinical significant arrhythmias that are readily detectible; that is, we see them in sotalol, we see them in a handful of non-cardiac drugs, but if you're up there, you an anticipate that you will have those risks. That isn't where we are, at least as best as we can tell. Instead we're in a much more difficult place, the place where you found yourself with moxifloxacin not so long ago; which is, as best as you can tell, it affects cardiac repolarization, but to a much smaller degree. And then the issue is can you detect a measurable increase in risk of cardiac arrhythmias in that -- for a drug in that setting.

And what we've been talking about are the ways you try to tease that out. And Dr. Ruskin has spoken eloquently about that this morning, and they're the things that we've talked about. You characterize the dose. You make sure that you understand the potential metabollic interactions, really, really, really well, particularly if you have a compound that's a 3A4 inhibitor, because we know that those are compounds that have gotten us into trouble in this setting in the past. And then you look to the clinical data set, where as is usual, we have not seen any Torsades, but you look at the other things that Dr. Ruskin went over. And then you determine how much uncertainty you have left, and how much you care about uncertainty based on what the other benefits of the compound are, and whether those can be addressed pre-decisional, or whether they need to be addressed, you know, on a post-marketing setting in some fashion. I believe that's where we are.

DR. RELLER: Dr. Ruskin earlier pointed out, as others have, that whether or not it's overly sensitive, the QTC by the set formulation is what has been used before. Has that -- and maybe it's a good thing to have an overly sensitive marker to pick up a warning of potential rare events; but has -- have there been drugs that have past been evaluated with the QTC set correction that have had -- that gave false warnings looking at this from the other side; false warnings that when drugs were put into practice with millions and millions of doses, have never shown a problem?

DR. MOSS: We can answer that question, but there are other issues here too. With regard to the question you asked, probably the one drug or group of drugs, or the calcium channel blockers that have in fact -- are associated with QT prolongation, but have not been associated with any arrhythmia problems, induced arrhythmia problems over a huge exposure, but that's a very specific case. We're talking now about diltiazam type of calcium channel blockers, and berapamil, both of which can prolong the QT interval, but are not associated with any arrhythmias.

Let me clarify one part of this correction for the heart rate. Up until now, over most of the drugs that have been reviewed where a drug has clearly increased the heart rate, people have accepted the correction of moving from Bazett to Fridericia, and based in essence in the denominator, the RR interval you take it at Bazett as a square root, and the Fridericia is a cube root, so you end up with a number that is somewhat smaller in the -- when you divide the numerator by the denominator.

What has gone on here is they've really taken essentially the fourth route, pretty close to it, so that it makes the data look better. The problem is there's no verification in any other population of using this approach, so that's difficult.

The second thing is if the sponsors are going to use t his approach, it seems to me they're obligated to provide detailed information on heart rate, and that was not -- it's not been presented; and that's a very important component. We'd like to know what the degree of heart rate change is. If many of the aspects are in terms of two or three beats per minute, it doesn't really make any difference. And so having detailed information in all of the subsets that they've done would be an important way of interpreting this.

The other thing that has not been fully addressed is that there is a dose response affect in the prolongation of the QT interval by both Bazett and Fridericia. And I haven't seen any heart rate data to say that the heart rate actually changes with the dose of the medication that's given. Maybe it does, maybe it doesn't; and it was the explanation that the visual difficulty was thought to be maybe due to a vagal effect, which is in fact something that you would expect to slow the heart rate. So I think an essential part of this is to see what the heart rate changes are so that one can make some sense out of this.

DR. RELLER: Thank you, Dr. Moss. Dr. Chesney.

DR. CHESNEY: This is very mundane compared to all the elegant discussion that's just been going on, but the patient or individuals who had a QTC greater than 450 milliseconds were excluded from these studies. If this were to be used in everyday practice, would that mean that every patient had to have an EKG before taking this drug? I guess what I'm saying is we don't know what would happen if you had a QTC greater than 450 milliseconds, and most of us haven't a clue of what our QTC is.

DR. BENEDICT: After about two-thirds of the way through the program, approximately about 600 or 700 subjects were included after removing that exclusion, so we do have data in patients who have acquired QT prolongation at baseline, and that's one of the sets that I showed you in my main presentation. They have about roughly 180 subjects who had QT prolongation at baseline who were involved with the study, so we do have some data on those subjects. And they're the very subjects who showed about an 18 millisecond decrease in the QT interval while they went on Telithromycin. Could be regression to the mean, but nevertheless, they did not show an increase.

DR. RELLER: I don't see any other hands raised.

DR. WALD: I have a question.

DR. RELLER: Dr. Wald.

DR. SEIDLiN: Are we going to leave the QTC?

DR. RELLER: We are about to leave everything for a break. Any other questions?

DR. WALD: I have one question.

DR. RELLER: First hear Dr. Wald and then --

DR. WALD: I wanted to ask Alma Davidson, when we look at these few bacteremic cases in the pneumonia study, there are 38 bacteremic patients, but we only account for 35 of them. What happened to the other three; did they have resistant organisms, or susceptible organisms?

DR. DAVIDSON: They had intermediate.

DR. WALD: And how did they do?

DR. DAVIDSON: They were all --

DR. WALD: Were they cured?

DR. DAVIDSON: They were cured.

DR. WALD: Thank you.

DR. RELLER: Dr. Davis.

DR. DAVIS: I had a similar question. When you look at Dr. Davidson's list 17 that are penicillin resistant, but in the sponsor's book it's 16.

DR. DAVIDSON: We had a total of 17, and that's with out analysis and the PPb population. That's because one patient was -- we included one patient who had no MIC, which was reported by the central lab. It was not actually reported by the central lab because the culture died, but this patient had resistant organism by occocil and disk methodology, and so we counted them as one.

DR. RELLER: We will now take a break and convene promptly at 4:00, and we'll launch into the questions for the Committee. Committee vote at 4:00.

(Whereupon, the meeting went off the record at 3:44 p.m. and went back on the record at 4:01 p.m.)

DR. RELLER: We must get to the questions. There were two comments at the break not discussed, that I want to have out in the open for everyone. First, Dr. Lee had a question having to do with hepatotoxicity that he wishes to ask. Dr. Lee.

DR. LEE: Well, I'm still concerned about the likelihood that there's still a hepatotoxic reaction here that is allergic. And indeed in the one sort of famous now 53 year old man, the disease might not have even been elucidated had he not developed this diarrhea with the rest of his family; because in fact, the significant abnormalities were occurring at something like 17 days after initiation of the drug. So I think it's really of concern; remember that he had fever as well as eosinophilia, and this kind of very late reaction.

This drug is going to be used and re-used in these very short bursts, and with a short burst of five days of drug, and then let's say getting toxicity 10, or 15, or 20, or 25 days later; this may present a problem in terms of identification of cases. If it is of an allergic nature, a la Halaphane for example, the second exposure or the third exposure may be more severe; and yet you may not know about it because your first exposure has been -- has missed identification.

DR. RELLER: Thank you, Dr. Lee. And Dr. Benedict from Adventis was going to have one point of clarification; a verbal statement regarding what had corrections applied, what did not. Okay. In the data presented.

DR. BENEDICT: Yeah. The brief statement I wanted to share with the Committee is that in the Phase III Program we did not see a significant change in heart rate; in fact a decline in heart rate as the patients improved for about four beats per minutes. So because of that, to place the QT data in perspective, we presented the QTC information, but the other correction formula was also given in your briefing document.

In the Phase I data, we had an average of about four to eight beat increase in heart rate depending on the dose that was used, with the higher doses producing the highest heart rates. Because of that defect, we used the QTn formula to correct for the heart rate defect in the study. But I would like to remind the group again that in the Phase III Program, expand the concentration up to 9.9 with an increase of only 8.7 millisecond interval in the QT interval.

DR. RELLER: Thank you, Dr. Benedict. And Dr. Ross, I'd like you to verify or confirm that your analyses were done on the -- from data -- from the clinical trials that were control trials only.

DR. ROSS: That is correct.

DR. RELLER: And Dr. Ruskin, that the -- which is the optimal formula to use has not been independently verified or something along those lines; a comment before we go to the questions.

DR. RUSKIN: I don't think I can shed much more light on the issue of the formula than the discussion has already done. I would agree that if there is an affect on heart rate, Bazett will over-correct and that clearly one would be better of with Fridericia. It's a little difficult with this mixed database where there are changes in some areas and not in others to know exactly what to do; so just I think in fairness to everybody, I would just like to in a sentence or two give you my take on this because I think we've talked around this issue a great deal, and not perhaps given the Committee a whole lot of useful guidance. And I still may not be able to do that, but my take on this is that even if you take the worst case scenario which is the Bazett, you're looking at changes with the drug in the range of a little less than 10 milliseconds probably, in a worst case scenario; so I believe that the drug has an affect. I believe there's a concentration dependent affect on QTC, but that it is modest, and not terribly dissimilar from what one sees with other drugs used for these indications.

I can't distinguish very effectively between this and Clarithromycin, although the data is very limited. But my sense of this is that there's a modest affect. It's real, but that it's not in a range where one would expect to see a high incidence of Torsades. That said, it is in a range where there has to be at least the expectation that there is some risk. I don't think one can put a number on that; just as one couldn't for Erythromycin or Clarithromycin. It isn't zero, and it clearly is not in the range of drugs that we know cause Torsades frequently. And I think I pretty much to leave it at that.

I think the decision about what one does is largely based on the potential benefits of the drug in relation to some theoretical risk; and I underscore theoretical.

DR. RELLER: Now while we have a full voting Committee, we must now get to the questions. As the Committee has experienced before, one can discuss, and there's been a lot of discussion. The way they're framed is the way we will take them; so the first question where there are not -- for the record, we had an open public hearing scheduled. There were no scheduled presenters at the open public hearing. Unless we missed someone who should speak up now, the open public hearing is closed.

In going to the question, both the FDA and the sponsor Adventis presented data and its generally accepted that this compound, Telithromycin, is effective in the treatment of community-acquired pneumonia, acute exacerbations of chronic bronchitis, and acute sinusitis. So specifically the question before us, and the pharyngitis/tonsillitis has been taken off of the table because of not meeting the efficacy criteria that were addressed earlier.

So the question to the Committee, and we want a yes or no, coupled with your name; a vote. And we will start at the right with Dr. Cross. And the issue then is not efficacy, but is this drug safe enough to be approved for use for these indications with the data that we have now.

DR. MURPHY: And Barth.

DR. RELLER: Yes.

DR. MURPHY: When -- again under -- in your discussion, because we would like not only your name and a yes or no, but we also would like attention to the A & B as to whether you have an opinion as to whether they have provided sufficient data to warrant a claim, because we think it affects the claim here. In other words, are -- if you say no, are you saying no because you don't think it should be -- that it's proven efficacy for resistant penicillin or Erythromycin too. If you could do that, or do you want to break it out? We thought it was important to roll this whole discussion into one, but it -- again it's up to the Committee if you think it will be easier to break it out.

DR. RELLER: Well, we've gotten into difficulty in the past, in you know, doing it a whole, and then breaking it out, and then breaking it out, and getting bogged down and then not being able to come logically back to the whole. And what we were -- what I was planning to do was to then go back and address these points, and then we would vote again --

DR. MURPHY: Okay.

DR. RELLER: -- swiftly around the table.

DR. MURPHY: Okay.

DR. RELLER: But we could reverse that. I mean, what would give the clearest message for the FDA. That's what we want to do in our advisory capacity.

DR. MURPHY: I think if you could address the resistance issue first; okay? That would be helpful.

DR. RELLER: All right.

DR. MURPHY: As it applies to these indications.

DR. RELLER: Okay. That's what we shall do. So we want -- I will say each person's name and we'll weigh in, so specifically A.

DR. BELL: Can I ask a question?

DR. RELLER: Yes. Dr. Bell.

DR. BELL: This is posed here strictly as an efficacy issue, and I'm just, you know, a guess here; but I wonder is part -- is there some implication in here to what extent is there a public health need for this drug to treat resistant infections that might be weighed against its potential toxicity? Is that something the FDA wants to have discussion on or consider? Because I -- there's some thoughts on that.

DR. MURPHY: Yes. Actually, we didn't want to ask you the classic question; is it safe, is it efficacious, is it safe, and have you addressed them as part. We felt that this has to be a balance of what is the potential benefit, particularly is there a place for the use of this product to treat resistant organisms at all, you know. And for poor resistance, and how do we weigh that against the safety issues which are potentials that we need to deal with.

DR. RELLER: What I would like to do now is to go and we'll ask A; and if a Committee member votes no, then to say right then and there what additional studies would be necessary from that member's view point. We'll do A, which has to do with penicillin resistance streptococcus pneumonia, and then we'll do B, have to do with Erythromycin resistance. And then we'll come back to question number one, and then deal with C, and number two as -- depending on the outcome of the assessments.

Now these discussions -- there's been a lot of discussion. What we will have is we will have a vote, what your opinion is in answer to A, and then what additional studies you would recommend if your answer is no. And you can discuss or make -- you can elaborate on your answer as you see fit.

What we really want to do is to give the best sense of each Committee Member, and then in the aggregate where the Committee weighs in to the FDA for their consideration, given that the regulatory authority rests appropriately with the FDA, and we give our views for them to take under consideration in their ultimate decision. Dr. Cross.

DR. CROSS: Okay. In terms of whether or not there's sufficient data to support the resistance issue, especially in pneumonia, looking at the data on page 30 and 31, there just isn't enough patient information. We have for example, an MM 61. In the case of resistance on MM 60, we're talking about 19 patients with penicillin resistance, and only 25 with Erythromycin resistance. I just don't think that's enough data to warrant that indication.

DR. RELLER: And what additional studies would you want?

DR. CROSS: I think we need more patients.

DR. RELLER: Okay. Thank you. Dr. Soper.

DR. SOPER: This is to A?

DR. RELLER: Yes; A only.

DR. SOPER: Yes.

DR. CHRISTIE: I'd have to say yes, because in previous times when we have discussed this issue we have taken less numbers for other indications.

DR. RELLER: Dr. Wald.

DR. WALD: I don't think that we have sufficient data. For the pneumonia patients we're talking about 17 patients who we think had resistant disease, 14 out of 17 cured for sinusitis, 11 out of 13. I would not feel good endorsing this drug for resistant organisms on the basis of such scant data.

DR. RELLER: Thank you. Dr. Archer.

DR. ARCHER: I would also say no. I was interested in the FDA's comment that there was a discordance even though the numbers were the same between Erythromycin resistance, penicillin resistance, which means that some of the patients with penicillin resistance were erm-sensitive, and yet failed therapy with this drug; which to me just means the numbers aren't big enough. I think if there is truly efficacious benefit, or if there is not a benefit, they just need more subjects.

DR. RELLER: Dr. Chesney.

DR. CHESNEY: I also would say not enough data, and I'm particularly concerned about the bacteremic pneumonias which is the worst case scenario. And I would also support more studies with clearly resistant organisms. I guess that was all.

DR. RELLER: Dr. Murray.

DR. MURRAY: I think I would vote yes. There were seven -- 15 penicillin resistant ones in the levo as I recall, with 100 percent cure rate, 17 in disk with 82 percent cure rate. They had 38 total bacteremic cases which is a lot with an 89 percent cure rate. I think the failures were the same in the two groups of the two bacteremic groups, and of the two of six failures, one of those was actually a cure of the infection to me, with a re-infection. You know, bacteremia clears and five days after you stop therapy, you have staph aureus in the urine and require other antibiotics, so I think that's in the borderline. And six bacteremic penicillin resistant pneumos is not a lot, but 17 is a fair amount. So it's a -- the vote would be yes; although obviously we'd like to see more data, but it's -- I vote yes.

DR. RELLER: I vote no, for two reasons; insufficient numbers of patients, especially those with bacteremia, and also the markedly different in those small number of patients, especially with bacteremic patients, efficacy contrasted with the drug currently approved for resistant pneumococci.

DR. EBERT: Steve Ebert.

DR. RELLER: Dr. Ebert. Yes.

DR. EBERT: I vote no, specifically for the reduced or low numbers I should say of patients in the bacteremic category. And also, I would like to see, although I'm not sure that there is an association between the degree of penicillin resistance and macrolide resistance, but further characterization of the upper limits with regards to penicillin resistance.

DR. RELLER: Dr. Leggett.

DR. LEGGETT: Jim Leggett. I vote no, for all the reasons of the no voters so far. I'd like to reiterate not enough patients, and I also am worried about the fact that penicillin resistant failed even in the absence of Erythromycin resistance, but for yet another reason in addition. If half of the penicillin resistant drugs are going to be Erythromycin resistant drugs, and we do -- the numbers look even less good with Erythromycin, we may see failures because there's also concomitant Erythromycin failure that then leads to failure of the combination drug resistance.

DR. RELLER: Dr. Leggett, to even it out we're going to start with you on Question B. Could you go ahead and comment on, is there sufficient evidence in your view that an infection due to Erythromycin resistant streptapneunae has a negative impact on clinical outcome compared with susceptible strains? And if the answer is yes, has the applicant provided sufficient data to warrant a claim for the treatment of community-acquired pneumonia, owing to Erythromycin resistant streptococcus pneumoniae.

DR. ARCHER: I don't understand that.

DR. LEGGETT: Yeah, I have a question because --

DR. ARCHER: That needs to be redefined. That's not a good question.

DR. LEGGETT: Gordon said this last meeting too.

DR. ARCHER: I mean, what do you mean?

DR. MURPHY: I think we actually had this earlier discussion somewhat about do we -- are we -- do we clearly understand what the impact of this means, and we know it's not good, but --

DR. ARCHER: Do you mean if somebody is infected with a macrolide resistant organism, and gets a macrolide they're not going to do well, and is that important? Is that --

DR. MURPHY: Well, I think the -- first of all, they want me to say this would be the first time we're giving this indication; so that's number one. And number two, then do we understand -- do we feel confident in what we think the clinical outcomes of this mean. Having these resistant organisms and the -- not having clearly indicated drugs for them. Do we think that that is an issue.

DR. RELLER: Actually, maybe I think differently, but it's -- one could change the wording around a little bit. But to me, this question is simply the following; should the drug if it were approved, have a specific indication for the treatment of Erythromycin resistant streptococcus penumoniae. And then I think we should have a comment from each Committee Member that's yes or no; then a comment that we feel that there are data to suggest that if one has Erythromycin resistance among S. pneumoniae that that makes a different clinically owing to the limitations, or owing to anything; limitations of dosing, because there's been discussion earlier that up to a certain limit of penicillin resistance, beta lactams can be used and are efficacious.

Are we dealing with the same thing, that if you just give enough of the macrolide, or are we saying that if you have one that by NCCLS criteria is resistant, we would expect a lower success rate. And then a corollary of that is, clearly there are differences in this compound from Erythromycin itself, but there has been a lot of discussion having to do with the kicking in of the different mechanisms of resistance to Erythromycin, and an assessment from the Committee Member as to whether those trends or changes in your view would affect the assessment or the activity of Telithromycin.

We've heard data from the sponsor that it doesn't, and we've seen MICs that suggest that maybe it's a warning that -- so -- but that's not -- it's not for me to say until it's m vote. It's for you to comment on those two issues; both of which are specifically delineated, comment on the potential for its comment on, and also comment on do you think that Erythromycin resistant organisms do less well with Erythromycin than do susceptible organisms in patients with community-acquired pneumonia.

I think those things are embedded in the concept of whether you think the data is sufficient to get a specific indication with this compound, which is the specific question delineated in B.

DR. MURPHY: Thank you, Barth. You can see why -- we re-wrote this question numerous times and it just got longer.

DR. RELLER: So should it be approved for Erythromycin resistant strains; yes or no? And then comment on the implications of Erythromycin resistance in community-acquired pneumonia, and cross-resistance with the ketolide. Jim.

DR. LEGGETT: Jim Leggett. Several comments; first, I do not think it should receive a specific indication for Erythromycin pneumococcus, but I do think we could -- I would favor wording it somewhat like we did with the amoxicillin in sort of high risk patients, or if it comes across not to worry about it. And for the -- my reasons are this.

We have data on otitis media that Erythromycin resistant pneumococci can do harm. I don't know that we have data about pneumonia per se. Given the fact with otitis media, and the fact that if you don't treat the pneumococcus, 90 percent of the time it's not going to go away. I would by extrapolation, in the absence of data, say the same thing for the lung. That's why I think that the drug will probably be useful; that's on the one side.

On the other side there's no -- not enough numbers. We only got five out of nine responses in -- where we had bacteriologic for blood from ERSP and DRSP together. Data from acute sinusitis to me is much more suspect, because of the 30 to 50, depending on your study, percent better viruses. And I don't know if in the sinusitis area, that I believe those numbers, even though they're 18 of 21, and 9 of 11. I don't know that we can extrapolate from more serious pneumonitis to an acute sinusitis situation; so I put less faith in that acute sinusitis data than I do the pneumonia data.

And then finally, to not get long-winded; I don't think we can increase the macrolides for telithro like we've been able to increase amoxicillin. So whereas, I think of MIC creep for the penicillins and better -- me believing that they're better than the cephalosporins in that regard. We can't push our macrolides. We're at the top; sort of like we are with the quinolones, or at least for the most part.

I don't know about increasing Telithromycin MICs going along with the Erythromycin, and in the absence of data, I'd hate to box you in by giving an indication that you didn't have to back away from, given the pressure to bring drugs to the market, and then the turn around, and then limit and take them off the market. I think you can do that one too many times in the new administration era; so long-winded answer. I'll leave it at that.

DR. RELLER: Dr. Ebert. Dr. Leggett, thank you for your comments on the differences between sinusitis and pneumonia. This question we're specifically voting around the table about the indication for community-acquired pneumonia.

DR. LEGGETT: The first part, B didn't say that. So then I'll just reiterate, I think that we should -- you could phrase it in the high risk groups, but I would not give a specific indication for Erythromycin resistant pneumococcus for community-acquired pneumonia. And in terms of fixing the situation, might consider using a group of patients who would be much more likely to have that, however you find that population; whether it's oral treatment of people who are hospitalized or some such thing to get -- bring up the numbers of pneumococcus and Erythromycin.

DR. RELLER: Thanks, Jim. Dr. Ebert.

DR. EBERT: I'm going to vote yes, but it's going to be a tentative yes. I thank you for your clarification, Dr. Reller, that this is specifically dealing with pneumococcil pneumonia. As was mentioned earlier today, I think there are now an increasing number of case reports of failures for pneumococcil pneumonias with macrolides. Certainly there has not been a controlled clinical study that has looked at this, but I do feel it's an increasing problem.

With regards to the cross-resistance between ketolides and macrolides with pneumococci, again my impression is that's primarily mediating through the erm B mechanism, at least at this time, which seems to be a relatively uncommon mechanism of resistance. However, I am concerned that in the future that may become a problem.

DR. RELLER: I just want to make sure we have this straight. Dr. Ebert, you would recommend that this compound get a specific indication for Erythromycin resistant strains and sntreptococcus pneumoniae.

DR. EBERT: Well again, I think the caveat, much like with penicillin resistant pneumococci would be again in the bacteremic population. I feel more strongly with the non-bacteremic population than I would with the bacteremic pneumonia.

DR. RELLER: I vote no for specific indication for Erythromycin resistant strains and community-acquired pneumonia because the numbers are small. The success rate in that very small number of patients with bacteremic pneumococcil pneumonia was marginal. The effectiveness, and I think the data -- it may work, but the data are simply not sufficient now to warrant that indication.

I agree that resistant strains, if they are truly the pathogens, are unlikely to respond. And although impressive data were presented about the differences between Telithromycin and Erythromycin itself, for those strains possessing one or more mechanisms of resistance to Erythromycin, I'm concerned about the creep. And I'd like to see far more data in cross resistance and follow-up before being comfortable with any specific indication for the treatment of Erythromycin resistant strains; especially given most therapy in this arena is empirical. Dr. Murray.

DR. MURRAY: I think to be consistent I'll vote a weak yes. I think it should be restricted to mild to moderate disease for community-acquired pneumonia. I think caution should be taken, whatever that means, because I'm not completely comfortable with Erythromycin mechanisms not having an effect. There is an increase in MICs with the mef. It's clear, and I think there is the potential for emergence of resistance with the erm. But the same is -- has been true for other drugs in other classes, for other indications. I think it's almost certainly should be better than clari for Erythromycin resistant organisms, but probably is not as good as moxifloxasin or a drug that -- or levofloxasin, I'm sorry, that had a 100 percent cure rate with its 15 penicillin resistant pneumococci. So I think it's probably better than what else may be out there, but not as -- better than some things, but not as good as all.

DR. RELLER: Dr. Chesney.

DR. CHESNEY: I -- excuse me. I also vote no, partially to be consistent; but many penicillin resistant strains are going to be Erythromycin resistant. And since I voted no for that, but I'm also very concerned that this is going to be empiric therapy. And I'm just not won over by the numbers of bacteremic cases that are reported, and they're not the 100 percent response that we did see with Levoquine. And I'm also very concerned about the emergence of resistance on therapy with -- because of the creep which is very impressive. And I would agree that we need more data on cross-resistance, and we just need more numbers.

DR. ARCHER: I also vote no with a heavy heart. Looking at the MICs of this drug, I would -- I'd hoped the clinical data would be better, because I really think we need another alternative to the quinolones, to prevent quinolone resistance emerging. I think we need this drug; but I think we need more data. The data are just not good enough clinically.

I'd also like to say, to weigh in a little bit on the Erythromycin as an indication; the macrolides are one of the mainstays of therapy of community-acquired pneumonia, so therefore it seems to me reasonable that you should consider the major resistance class to those antibiotics. In fact, I think macrolide resistance is more important in many ways than penicillin resistance, because I think the data shows in failures on macrolide therapy that the MICs to macrolides correlate better with treatment failures than do penicillin susceptibility data. And so I think it's really important to have an indication for macrolide resistance.

At our hospital, 50 percent of our -- all of our pneumococci are macrolide resistant. And it's been a huge issue in devising treatment protocols for community-acquired pneumonia about what to use; and we have eliminated the macrolides as a class because of that, rightly or wrongly. And so I think we need more options. I'm just disappointed that this -- the clinical data aren't there yet for this drug.

DR. RELLER: Dr. Wald.

DR. WALD: I would also vote no for the reasons that were stated; that is that there's simply insufficient data. I would say that as an empiric drug for community-acquired pneumonia, I will ultimately vote yes, but we're asking the specific question for either penicillin resistance or Erythromycin resistance. And if we knew those to be the case, or if the patient was particularly high risk, I wouldn't endorse this on the basis of these data.

I must say it troubles me less that the bacteremics break through, because if you suspect a patient has bacteremia you hospitalize them; so this is not for the patient that we look at and think is terribly ill. This is for the patient who has mild to moderate disease.

DR. RELLER: Dr. Christie.

DR. CHRISTIE: Thanks. I'd have to say -- I'd have to cross the bar and say no. I don't think we have enough information here, and I believe we should gather more data.

DR. RELLER: Dr. Soper.

DR. SOPER: To remain consistent, I'll say yes. And I agree that we need more information, but looking at the MICs I'm reassured. I think that there is concern about potential cross-resistance because of the mechanisms involved. I doubt it will be 100 percent penetrable, and I think there is a need.

DR. RELLER: Dr. Cross.

DR. CROSS: Yes. Again, I think that while the data on the pneumonia in the absence of the bacteremia is encouraging, I think that again we still need more data; especially in view of the small numbers with the pneumococcil bacteremia. Although I do agree with Dr. Wald that if a patient were truly toxic from the pneumonia, and we suspect that they had bacteremia, we would hospitalize them and wouldn't use this drug.

In terms of the issue of creep, I would feel -- I don't recall hearing what the maximum serum level of the drug was. In the presentation, it appears that the level of the antibiotic was obtained to correlate with the EKG level, and I'd like to have a better sense of what the maximum achievable concentration in the blood is, in order to interpret the clinical pharmacology with the in vitro microbiology.

DR. RELLER: Dr. Cross. Did I understand your vote no?

DR. CROSS: No. Exactly.

DR. RELLER: So although not unanimous, the Committee weighed in seven to three, not necessarily the same seven each time, against a specific indication for either resistant -- penicillin resistant or Erythromycin resistant pneumococci as a specific indication were this drug to be approved.

So now we'll come back, Dr. Murphy, to question one. And then we'll segway to C, and number two. So today, with the data available and the controversy surrounding electrophysiology and hepatic function, given the risks as best as we can understand them as of the moment; does the efficacy of Telithromycin in respiratory infection support its use all things considered, taking into consideration comments made in the public health interest, et cetera, support its use for community-acquired pneumonia, acute exacerbations of chronic bronchitis, and acute sinusitis. And the reason we're bundling these is that the issue here is not efficacy, but rather does the safety profile the number of patients potentially treated empirically, drug interactions, age spectrum, et cetera. On balance, public health, needs for therapy given that there would not, if the Committee's advice were followed, specific indications for resistant organisms.

On balance, did the merits for efficacy outweigh the potential risk for toxicity? Is that a fair assessment, Dr. Murphy?

DR. MURPHY: Correct. Is the benefit of having this product for these therapies, because we're not arguing -- we're mostly in agreement with the efficacy statements made by the sponsor, are the benefits -- do they outweigh the risks as we know them at this point; again knowing we can't as clearly defined have a precise assessment. We do know certain things, and we know certain things about history; so we're asking you your level of certitude of the benefit versus the level -- which we're fairly certain that it has efficacy. Does it outweigh the risks as defined at this point?

DR. RELLER: Dr. Wald has a comment.

DR. WALD: I just wonder if we do a disservice to bundle them, because the -- I would see the question -- I would see the answer as being different for different of these indications, because if --

DR. MURPHY: Yes, I meant for different indications. Do them -- do each indication separately.

DR. RELLER: Okay. Fine.

DR. WALD: I'm sorry. I thought you meant --

DR. MURPHY: We weren't going to ask the efficacy and safety separately. We're --

DR. RELLER: Right. Right.

DR. MURPHY: Okay.

DR. RELLER: Okay. Well we can do them -- I mean, whatever is going to give the cleanest, clearest message to the FDA.

DR. MURPHY: We do indication by indication.

DR. RELLER: Indication by indication. Now one of the -- Dr. Soreth I believe succinctly put this as approve with further safety studies up front, and with further post marketing surveillance for adverse reactions afterwards.

Ultimately, in Question C and Two, we're going to get to these issues. Would it be an efficient way to categorize these right up front?

DR. MURPHY: Yes.

DR. RELLER: With those options.

DR. MURPHY: Yes.

DR. RELLER: And then the nature of -- I mean if there were restrictions on the front end, rear end, the nature of those could be discussed. But would that be an effective way to deal with this?

DR. MURPHY: Please.

DR. RELLER: Okay. So let's take them one by one. Community-acquired pneumonia, approval, approval with -- or non-approval with some requirement beforehand, or approval with some requirement after approval in terms of post marketing surveillance. So actually it gives the options of saying exactly what you would do if you were in the FDA's position. Alan.

DR. CROSS: Okay.

DR. RELLER: For first -- we'll go around the table with community-acquired pneumonia, and then we'll go left to right for acute exacerbations of chronic bronchitis, and then back to you for the final round for acute sinusitis.

DR. CROSS: Well, I think this is a very useful drug, especially considering the use of the quinolones and the problems associated there. I think the efficacy data presented is convincing, and I think the perspective that Dr. Ruskin added at the end about the meaning of the change in the QT was helpful.

I was also interested in Dr. Soreth's quotation from Dr. Temple saying we really for safety need to see lots and lots of data; and we're in a Catch-22. We won't see that data unless there's more widespread use; and so the best of all possible worlds I would say yes, I would approve this, but I would like to see at least a period in which we have a go at the larger safety data with wider use, so it would be a yes.

DR. RELLER: Thank you. Dr. Soper.

DR. SOPER: Yes, with post marketed -- marketing surveillance for both cardiovascular adverse effects, and hepatotoxicity.

DR. CHRISTIE: I guess I have to say no. I'm concerned about the adverse effects as were outlined today, and especially the cardiac problems. I'm not sure that I have sorted it out in my mind that this drug would be safe.

The other issue too is that if it's given for one, it might be given for all. And I'm also concerned about its use in children, children less than 12 years of age; so as we work through these issues I'd like to see some data presented in the future on children.

DR. RELLER: Dr. Wald.

DR. WALD: I would say yes, with post marketing surveillance. I think that the -- this class of drugs looks like it's very attractive for lower respiratory tract disease. I think that is its niche. I'm hopeful that will show that it is effective against resistant pneumococci and that we'll have increase in usage, but I think as an empiric selection it looks quite good.

DR. RELLER: Dr. Archer.

DR. ARCHER: Yes, we post marketing toxicity surveillance for all the issues raised.

DR. RELLER: Dr. Chesney.

DR. CHESNEY: I would say no, I'm fine with the efficacy, but I'm very concerned particularly about the liver disease. I'm not as concerned now about the QTc issue, but I would favor looking at an additional several thousand patients specifically focusing on liver function tests before, during, after, whatever else it takes to clarify the liver issue.

DR. RELLER: Dr. Murray.

DR. MURRAY: I would vote yes, and I guess I'm willing to accept the toxicity because I think it offers a benefit for some of the resistant organisms.

I am concerned probably more about the potential liver than the cardiac. And also, the comments made about the repeat dosage I think is a good one, that I think people are likely to receive repeated courses, and there are no data on that. And I believe in some of the other toxicity that people that had gotten multiple fluoroquinolones or priatrova as I recall had some -- they were the ones that may have been brought out later on by liver toxicity.

DR. RELLER: I would vote no without substantial pre -- I mean I don't know the mechanics of how this is done, but of the three options Dr. Soreth outlined, I'd like to see more patients enrolled in streamlined clinical enrollments to get the greater experience with this drug, going to the safety issues. Especially because most therapy is given empirically, and although the comments have been clearly heard about the less severe patients who may not have bacteremia at the time the patients presented; we don't know, but those are the ones that I look to because those are the ones I'm certain had pneumococcil pneumonia. The others, there are always questions about them; not questions because of the way the sponsor did the studies, but just questions because of the clinical reality. The ones with positive blood cultures we know had pneumococci, and we've got the organism and know exactly how susceptible or resistant it was.

You know, I see such broad use for these indications, and the potential for cross-reactions in an older population, including older women that I am uncomfortable with the numbers that we have, with the potential warning signals that are there; to give unfettered approval at this time, especially when one takes resistant organisms off the table, which eliminates at this point, the argument about not having other drugs. Dr. Ebert.

DR. EBERT: A question first; should we direct our comments as far as severity of disease at this time, or are we going to defer those with regards to indication?

DR. RELLER: I think we can slice it too much. I mean the indication is community-acquired pneumonia.

DR. EBERT: Okay. I would vote yes, with continued post marketing surveillance for adverse effects, and for hepatotoxicity.

DR. RELLER: Dr. Leggett.

DR. LEGGETT: I'll try not to belabor it. I'm glad I'm not part of the FDA and have to make this decision. I would vote a fettered yes, on whether that's studies before it comes out or studies afterwards; I think that those should be mandated.

I would -- the reason I would vote yes, is I think that the drug does offer some benefit, and it's exactly those patients who are beta lactim and floraquinolone, et cetera, et cetera who might have problems; so I think they do add something, but I am most worried about the 3A4 interactions, and the possible hepatotoxicity. But I think that I view this as much a class action as a specific drug action in a lot of things.

It to me appears that the possible cardiac toxicity is in the ball park with Clarithromycin and Erythromycin. It also looks to me that the hepatotoxicity, except in terms of the LFTs is in that same ball park. I don't know what to make about the idiosyncratic reaction, and the repeated drug uses, but I -- there's also the abacavir story, and repeated use that makes me a little bit worried.

I am particularly worried about not forgetting to do some additional studies that I think need to be done, and that is to clearly -- to better define just exactly what happens to Cmax and AUC in the elderly. Those over 65 are going to have most of the time decreased hepatic and renal function. We can see that with drugs that aren't -- don't even go through CYP such as ceftriaxone, where people over the age of 65 have their drug levels go up several-fold. So I think that we need to catch up on those, and I think we need to make some sort of statement about limiting the 3A4 substrate-inhibitors in a box.

What I would do for the term the possible Torsades de Pointes is sort of do something as with Clarithromycin as in the warning, whatever it's called, that this may happen. But I'm worried without more data with the 3A4 drugs and the higher levels, just exactly we don't know what's going on.

DR. RELLER: Thank you, Dr. Leggett. Could you then start the round for acute exacerbation of chronic bronchitis.

DR. LEGGETT: No. The drug is not needed. The pneumococcus is not a big problem. Tincture of thyme, some steroids and inhalers are the therapy for chronic bronchitis exacerbations. I don't think we need this drug for chronic bronchitis exacerbations.

DR. RELLER: Dr. Ebert.

DR. EBERT: Also no, I agree there are plenty of other options for treatment.

DR. RELLER: I vote no, including the role of the Big Three and Haemophilus being right at the brink in terms of achievable concentrations with the MICs. No.

DR. MURRAY: No.

DR. CHESNEY: No.

DR. ARCHER: No.

DR. WALD: No.

DR. CHRISTIE: No.

DR. SOPER: No.

DR. CROSS: I vote no, and also with regard to this type of infection, we just need more data on what is the effect of repetitive dosing on these votes. I'm sorry; repetitive episodes. I vote no.

DR. RELLER: Thank you. Now Alan, could you start round three for sinusitis, acute sinusitis.

DR. CROSS: Well, I think the data presented in general looks good for -- but I just don't think we have enough data in terms of the Erythromycin and penicillin resistance. It's very hard to get bacteriologic data on this, but what we have overall in terms of the non-resistant isolates, I think it shows efficacy. And again, the only caveat I would have is in terms of repetitive treatment of this episode, and there is the same toxicity issue that we had on the bronchitis, so I would vote a -- what was said, a fettered yes, but I leave it at that.

DR. RELLER: Dr. Soper.

DR. SOPER: Yes.

DR. RELLER: Christie.

DR. CHRISTIE: No, I don't think so. No.

DR. RELLER: Wald.

DR. WALD: I would say no, because I think there are so many other choices for drugs. And while I have to say I'm not really worried, especially about the toxicity of this drug, either cardiac or hepatotoxicity; while we're using it for pneumonia, we can learn a lot more about toxicity and then say it's okay for sinusitis. So until we know that, I'll hold.

DR. RELLER: Dr. Archer.

DR. ARCHER: I agree with Dr. Wald. I'm kind of on the fence, but I think that questions of toxicity in this case don't warrant its use for a relatively trivial infection like sinusitis, so I guess I would say no.

DR. CHESNEY: No.

DR. RELLER: Chesney is a no. Dr. Murray.

DR. MURRAY: I think I would vote no. I think it probably works for the resistant organisms, but I'm not sure the toxicity issues are strong enough to -- or overweighed enough by the need.

DR. RELLER: No.

DR. EBERT: Dr. Ebert. I also vote no for all the reasons listed so far. In addition, the fact that this drug shows some activity against anaerobes concerns me, that by approving this for acute sinusitis, we may be opening a Pandora's Box and be using it for chronic sinusitis as well.

DR. LEGGETT: Leggett. I concur with most of what's been said before, and vote no; because I think the risk benefit analysis is -- weighs more towards the potential risk in acute sinusitis.

DR. RELLER: Okay. Tom Perez has the tallies here, so for the question number one, we have seven to three yes for community-acquired pneumonia, a unanimous no for acute exacerbations of chronic bronchitis, and eight to two no for acute sinusitis.

So let's move now to Question C; that then would logically be what kinds of restrictions in labeling or populations, or specific things to be looked at in additional studies or surveillance, limitations in access or distribution, how would you control these things, Dr. Cross, for community-acquired pneumonia?

DR. CROSS: Well as mentioned, I think that we need to have a study with large amounts of patients, a so-called simple study in which we enumerate the -- a much larger population the risk and side effects. But also on top of that, I think that what has been mentioned is even though it hasn't been discussed here, it ought not be used in a population of children. And even though we have some data on 13 to 18 year olds, I think we really have to note that there really isn't sufficient data in this population to warrant its use.

DR. RELLER: Dr. Soper.

DR. SOPER: I think the provider needs to be warned about this 450 drugs, and I -- whatever we did for amoxi, amoxicilin with the QT, whatever ended up being the final thing there, probably should relate to this one too.

DR. CHRISTIE: We should come to know the potential drug interactions as listed before. We should not use the drug in children, people with previous liver disease, underlying cardiac disease as mentioned before, and we should get larger numbers. And again, I'd like to see data in children.

DR. RELLER: Dr. Soreth.

DR. SORETH: Just a note for Dr. Soper with regard to what we did with moxifloxacin, it's not metabolized through the P54 system, so that's not an issue in terms of concomitant drug use.

DR. SOPER: QTc also. The box, the QTc box.

DR. SORETH: Thank you.

DR. RELLER: Dr. Wald, how would you frame the --

DR. WALD: Restrictions.

DR. RELLER: Yes.

DR. WALD: I mean, I think I would agree with what's been said. Certainly, we're not talking about children today at all, as I understand it; and I would agree that any concurrent use of those drugs that prolong QT should be prohibited, so I think in taking care especially of geriatric or complicated cases, this really means some counseling for the patient to go over an extensive list of other drugs that they might be on.

DR. MURPHY: Could I ask a question about that? Since our experience is that even when we list these contraindications of concurrent drugs, or at least drugs that we want you to ask about; that one of the reasons we, as Dr. Soreth said, have problems is that apparently that is not always paid attention to at the level we would like it to be. So there -- I guess one of the question we have is when you say post marketing surveillance, that's one thing; are you now also suggesting that there are some other models of restricted distribution or use where one has a physician's register with the pharmacist or registries? Is that what -- are you -- so that they would basically have attested that they understand the use of these products, or you're not saying that. I'm just trying to understand if that's what you're getting at or not.

DR. WALD: What other drug do we that for besides Accutane? Is there any other drug?

DR. MURPHY: Accutane was one. Yeah.

DR. WALD: And no other drug?

DR. MURPHY: I mean there's others, Thalidomide of course, but I mean there are some others that are under discussion for doing that. And of course, there are other approaches where, you know, patients have to have tests before they get the drug. That's another approach, but I just -- when you said make sure they reviewed the concurrent drugs, that made me wonder if that's what you were trying to get at; because we do have those systems in place for other things.

DR. WALD: I guess I hadn't thought of something so formal as that, but that might be a very effective strategy; either that or at the pharmacy, I guess that would be the other alternative. I don't know which is a more fail-safe method. Certainly if people have to be licensed to dispense the drug, then I think that would be a very effective way to make sure they reviewed all the potential contraindications.

DR. RELLER: Dr. Archer, how would you go about it?

DR. ARCHER: I agree. I think that all the risk groups, including the risk of Torsades in women as opposed -- higher incidents, and the problems in elderly and all of those things need to be noted, and the same kinds of things that we've done in the past.

I think that -- I don't really think there should be restrictions. I think we need the drug to be used is as many groups as possible so that we can assess the toxicity, and find out if these issues are real. I'm not totally compelled by all the toxicity data. I think it's very -- it was very confusing to me. I think it raised a lot of red flags, but I think this is something that needs to be looked at when the drug is used widely.

DR. CHESNEY: I would I guess have big boxes about drug interactions, and another box about QTc. And my two questions or issues in addition to what's already been said, recognizing that I said no; I don't know if -- how strong a recommendation you can make about getting liver enzymes at the onset of therapy, or making a very complete of pre-existing liver disease before the drug is started, with some very strong wording that the risks are unknown at this point in time.

I also wondered, having heard on Monday or Tuesday about a patient insert in a box. I wonder if that would be a possibility; that every patient who takes this drug is given something that also outlines these things, in case the physician forgets to clarify that.

DR. MURPHY: Well at the end of our labels we have the patient information, so Dr. Chesney, are you talking about the -- what we call the med guide where every -- the med guide, there's a mandate that every patient receive this piece of paper versus the insert which can be part of the label which they may or may not get. So I guess what I just want to ask you is are you -- is that what you're talking about; is a med guide where it's mandated that each patient get it?

DR. CHESNEY: That's -- I would think about a med guide.

DR. MURRAY: Well, I certainly would encourage studying the drug in kids. I think that's where more resistant organisms are, so that's where you really need it.

I'm trying to remember if LFT elevations were excluded before -- it was an exclusion in the clinical trials. I mean, I'm not sure they're in -- I mean, yeah I'm concerned about the possible liver toxicity, but there were no data really to suggest that if we knew they had an elevated LFT, that they were any worse; so I really couldn't see checking LFTs on people before they took the drug. But I do think that some statement about potential prolongation of -- or the prolongation potentially of QTc needs to be -- a statement needs to be made, and perhaps a wording for potential for liver as well.

DR. SORETH: So would that imply, Dr. Murray, that everyone would need a baseline EKG for you to determine --

DR. MURRAY: No. I'm just worrying that it is -- that some prolongation has been noted, but no, I did not mean a baseline EKG.

DR. RELLER: I would -- having voted as I did, I would take a somewhat different approach from my colleagues. And I don't know what -- I don't know enough to know what restrictions with confidence I could place on a label. I'm -- there are -- in applying this drug widely, with the potential for many, many drug interactions, the age issues, the question of older women, the empirical therapy; I favor finding out more about the drug up front, to possibly minimize all of the boxes and warnings after the fact. Because after the fact, or even with the warnings in the boxes, I don't know the data of how closely they're followed, whether they're followed, and what effect they have.

Has anyone ever done a controlled trial of having a box or not having a box, a warning or not a warning, if it affects practice one iota? So consequently, what I would prefer to do is see what would be basically unfettered use of the drug in terms of enrolling patients, but close monitoring or gathering data on those patients to see whether there really is a risk.

Now I realize that if it's an event that's one in a million, it's impractical. But once widely available, I mean the potential market is in the millions; so as a consequence, I'd like to see a lot more than 1,500-2,000 patients. And it's not an issue of demonstrating efficacy any more; it's an issue of focusing on those things that might give a good indication of toxicity when applied more widely for the indication of community-acquired pneumonia. And there are a lot of people out there who are getting treatment. I think we need to observe them in a monitored way prospectively before the drug is released generally, if that's possible to do somehow.

DR. MURPHY: So if I could paraphrase; I think what I heard you say, Dr. Reller, is that it's a large not so simple trial where use is unfettered, as you said; but we have some parameters of monitoring that we would collect before approval, to look at some of these issues.

DR. RELLER: Right. Specifically, for example, you know, let's treat older women with pneumonia and document what drugs they're getting concurrently, and see what happens in those patients.

DR. MURPHY: Just as a -- we have never marketed a product with one -- with a certain label versus having it marketed in a different area with a different label and studied that, so there is no controlled study on that. We do have unfortunately experience on the point that Jan was making; if we don't get it right the first time, we have experiences where we said -- we put in black boxes. We've sent out numerous "Dear Doctor" letters, public health advisories, and behavior does not change, once a practice pattern has set with the original approval; not significantly enough in a number of cases, let's put it that way.

DR. RELLER: Exactly. And you know, the sponsor appropriately and you know, in the context of the controlled trials that were done, had all of these restrictions on who could enter the trial. I'd rather lift the restrictions and enroll thousands to preserve, if I were king, the market for millions. Dr. Ebert.

DR. EBERT: I agree. It should also not be used in young children, but that it does merit study. I would favor, at least initially, placing a warning in the package insert similar to what one might see with Clarithromycin. If the drug interaction data and adverse data were to be borne out as being minimal, that could be petitioned to be removed at a later time.

I feel that it should be promoted only for use in patients who do not warrant hospitalization, so in patients who would be treated on an out-patient basis. And perhaps maybe something a little bit different, but in -- if there are some concerns about this drug being used inappropriately, or in patients who are at high risk, or patients who are receiving multiple other medications, perhaps the policy of how the drug is sampled should be considered, and whether -- certainly there is going to be less recordkeeping, more of a potential for harm if the drug is sampled liberally rather than if it were only available by prescription.

DR. RELLER: Dr. Leggett.

DR. LEGGETT: I sort of won't say what I just said a few minutes ago when I addressed the issue in the other question. But I think we do need primarily to address the toxicity issue, and to sort of concur with Dr. Murphy; I agree that it's better to get it right the first time, so whether that means putting up a road block before allowing its use, or coming up with some new strategy to better look at the way things have been done in Phase IV than has been done in the past for many drugs. I don't know the answer to that.

DR. RELLER: Dr. Leggett, to finish the afternoon session with the last question, if the Committee has not recommended approval, please provide recommendations for additional studies. The only two indications that got a majority of no votes were acute exacerbations of chronic bronchitis and acute sinusitis.

What would you require to consider again those indications? Or one option would say get it worked out for acute pneumonia -- I mean, excuse me -- community-acquired pneumonia. In other words, that's -- those are a couple of the options. What would you do about the two that got a no on balance?

DR. LEGGETT: I would address the toxicity issue. I don't think I had a problem with the efficacy, despite the fact that those are notoriously hard things to know for sure, about efficacy. I think five days will probably work for anything, but that's just what I think; so I -- if it's easier to address the toxicity by looking at community-acquired pneumonia folks, I don't think we would necessarily have to go back and re-do toxicity with AECB and acute sinusitis.

DR. RELLER: Dr. Ebert.

DR. EBERT: My only suggestion would be if those two indications were to be pursued, to aggressively target patients who were infected with penicillin resistant or macrolide resistant strains to really build up the numbers as much as possible there, with the hopes of -- and especially with Haemophilus influenza as well, to try to specifically target those pathogens.

DR. RELLER: I would put the emphasis on getting the safety data before approval with community-acquired pneumonia with a larger number of patients as mentioned; and then that having been settled, one could come back and reconsider expanding the indications. The controlled trials are there for efficacy. I think the place to find the safety information is with community-acquired pneumonia.

DR. MURRAY: If you -- if the toxicity issues are allayed by extensive use in other patients, then the concerns about using it in a relatively minor disease like acute sinusitis are lessened. Similarly, if the increase -- if there's an increase in resistant organisms so that the benefit is great, and the efficacy data is continued, then again our concerns -- my concerns would be allayed. So it's one or the other, either get -- allay my concerns about toxicity, or make the resistance in sinusitis more common and of a bigger issue.

DR. CHESNEY: I agree with everything that's been said. I think -- but I think I would do them both. I would look very closely at toxicity issues that would make us all less anxious; and if it could also be shown that you had many more resistant -- pneumococcil resistant isolates, both Erythromycin and penicillin and that it was very effective, then I think none of us -- I wouldn't have any problem with it.

DR. ARCHER: I agree with those sinusitis statements. I don't think we need any antibiotics for chronic bronchitis period, so I don't think that's anything that I would think they could do anything about. But I think sinusitis, the issues of resistance and toxicity, once they're solved, I think that would be a nice indication.

DR. WALD: I agree that -- again I think the special niche for this drug and for the macrolides is lower respiratory tract disease. If you want to sell it for upper respiratory tract disease, I think you need to show that it's going to be effective for resistant organisms. That will make it special and make it recommendable. And then just as was said, as we establish safety with community-acquired pneumonia, then I think we can use it.

DR. CHRISTIE: I agree with everything that has been said. Specifically, I'd increase the numbers, and make sure that the toxicity is improved, and get enough numbers to look at the resistance data in a better way. Looking at things like drug interactions, ECGs for a longer time, LFTs throughout, and of course, a plug for children; because I still see that once you approve a drug, even though it's not approved for children, it may be prescribed for them, so we really need to pay attention to that population.

DR. RELLER: Thank you. Dr. Cross.

DR. CROSS: Okay. I also agree with doing a lot about toxicity from the pneumonia studies. I agree. I don't see any use for this in exacerbation of bronchitis, and I'd like to see more numbers with the Erythromycin. But I would like to return to the last question, which is in effect related to this, and I don't know whether it's putting too fine a point on it; that in terms of what to put in in a box, that there is a known -- if there's a patient known to have a QTc of greater than 450, that was an exclusion criteria. And you know, it appears to me we don't have enough data on that group, such that I wonder whether or not that also ought to be specified in the labeling which we have on that last question.

DR. RELLER: Thank you. Dr. Soreth and Dr. Murphy, anything else you wish of the Committee; to say to them, or ask of them?

DR. SORETH: Maybe I could just summarize the last few minutes of discussion to make sure I understand it correctly.

I think it a fair statement that as the Committee looked at the indications of community-acquired pneumonia, acute exacerbation of chronic bronchitis, and acute sinusitis; there was for the latter two sinusitis and bronchitis, a consensus, if not unanimity, a consensus that at this point the drug should not be marketed for those indications. This is resistance off the table for the moment, because we already summarized that. And so, to that end for things to go forward, and in order to understand the risk management perhaps more clearly than we do at this point, how the drug would be utilized in a patient population at large, now talking about community-acquired pneumonia.

There were comments made across the table for investigation, not so much in the realm of efficacy, for we feel we understand that in cap for susceptible organisms, but rather further investigation to help clarify and either validate our concerns with regard to QT prolongation and potential liver toxicity, or allay fear that in broad use it could be used wisely, prudently, effectively, and safely. And that to that end, there -- we have mechanisms, obviously, to either investigate that prospectively in something like a large simple safety trial, which I will grant you; I think we do not have a large experience with doing that. More often what we have is marketing a drug and looking post marketing in some way, either through passive surveillance which sometimes makes it difficult to figure out precisely what's going on, or taking a step back and looking at the paradigm and considering an approach that is prospective.

Again, not to delineate anything with efficacy with susceptible organisms for we understand that, or think that we do; but rather to delineate certain safety issues, concomitant drug use, drug interactions, et cetera. And that we will take all of that under advisement and proceed at this point in working with the company to see then which of the various, you know, ways of gathering more safety information is at this point prudent, and in the best interest of the public health.

DR. RELLER: And at the same time in so doing, would accumulate larger numbers of organisms if resistance continues to increase or maintain, that would give the numbers that might enable an accurate assessment on efficacy among resistant organisms.

I want to thank all the presenters for their efforts. I think we have a vigorous discussion on issues that don't have simple answers, if an answer at all as of now, and to the Committee Members for faithfully staying to the end.

Tomorrow's meeting for the Committee Members, a closed meeting begins at 8:30 in the morning in this room.

DR. SORETH: That's a good question. Is it in this room? It is.

DR. RELLER: In this room.

DR. SORETH: And it's informal dress-down Friday, so I hope you packed your jeans, because that's what we'll be wearing.

DR. RELLER: And thank you again. The meeting is adjourned.

(Wherupon, the above-entitled matter concluded at 5:19 p.m.)