8:34 a.m.

Friday, July 20, 2001


















CDER Advisory Committee Conference Room

5630 Fishers Lane

Food and Drug Administration

Rockville, Maryland 20857




Charles B. Jordan Professor

Head, Department of Industrial & Physical Pharmacy

Purdue University

1336 Robert E. Heine Pharmacy Building

West Lafayette, Indiana 47907

NANCY CHAMBERLIN, PHARM.D., Executive Secretary

Advisors and Consultants Staff

Center for Drug Evaluation and Research

Food and Drug Administration (HFD-21)

5600 Fishers Lane

Rockville, Maryland 20857

GLORIA L. ANDERSON, PH.D., Consumer Representative

Fuller F. Callaway Professor of Chemistry

Morris Brown College

643 Martin Luther King Jr. Drive, N.W.

Atlanta, Georgia 30314-4140


University of Puerto Rico

School of Pharmacy

4th Floor, Office 416

P.O. Box 365067

San Juan, Puerto Rico 00935-5067


PRESIDENT, Boehlert Associates, Inc.

102 Oak Avenue

Park Ridge, New Jersey 07656-1325


Professor Emeritus of Pharmacology and

Toxicology and Therapeutics

University of Kansas Medical Center

3901 Rainbow Boulevard

Kansas City, Kansas 66160-7471


Professor of Pharmaceutics

Department of Pharmaceutics

School of Pharmacy

State University of New York at Buffalo

Buffalo, New York 14260

ATTENDEES (Continued)



Department of Pharmaceutical Sciences

School of Pharmacy

University of Southern California

1985 Zonal Avenue

Los Angeles, California 90033


Associate Professor of Pharmaceutical Sciences

College of Pharmacy

The University of Michigan

Ann Arbor, Michigan 48109


Department of Pharmaceutics

School of Pharmacy

Medical College of Virginia Campus

Virginia Commonwealth University

Box 980533, MCV Station

Room 450B, R.B. Smith Building

410 North 12th Street

Richmond, Virginia 23298-0533




Executive Director, Center for Drug Studies

Medical College of Virginia

MCV West Hospital, Room 12-410

1200 East Broad Street

Virginia Commonwealth University

Richmond, Virginia 23298

ATTENDEES (Continued)



Acting Chief

Section of Nuclear Magnetic Resonance Laboratory

of Membrane Biochemistry and Biophysics

National Institute on Alcohol Abuse and Alcoholism

National Institutes of Health

Park 5 Building, Room 150

12420 Parklawn Drive

Rockville, Maryland 20852


Section Chief

Section of Fluorescence Studies

Laboratory of Membrane Biochemistry and Biophysics

National Institute on Alcohol Abuse and Alcoholism

National Institutes of Health

Park 5 Building, Room 114

12420 Parklawn Drive

Rockville, Maryland 20852




Head, Division of Clinical Pharmacology & Toxicology

University of Toronto

Room 8229

The Hospital for Sick Children

555 University Avenue

Toronto, M5Q 1X8 Canada


University of Kentucky, College of Pharmacy

401-A College of Pharmacy Building

Rose Street

Lexington, Kentucky 40506-0082


Professor, Chair and Associate Dean

for Research and Graduate Programs

Department of Pharmaceutical Sciences

College of Pharmacy, Health Science Center

University of Tennessee

847 Union Avenue, Room 5

Memphis, Tennessee 38163

ATTENDEES (Continued)



Professor of Physiology

University of Colorado Health Sciences Center

4200 East Ninth Avenue

Box C240, Room 3202-2

Denver, Colorado 80262




Vice President, Biopharmaceutics

Eon Labs Manufacturing, Inc.

227-15 North Conduit Avenue

Laurelton, New York 11413




Alza Corp.

1050 Hamilton Court

Menlo Park, California 94025




Review Chemist

Office of New Drug Chemistry



Office of New Drug Chemistry


Senior Clinical Pharmacologist/Biopharmaceutics Reviewer

Office of Clinical Pharmacology and Biopharmaceutics



Office of Clinical Pharmacology and Biopharmaceutics

ATTENDEES (Continued)



Deputy Director

Division of Pharmaceutical Evaluation III

Office of Clinical Pharmacology and Biopharmaceutics


Review Chemist

Office of New Drug Chemistry





Product and Process Characterization

Gilead Sciences, Inc.

650 Cliffside Drive

San Dimas, California 91773


Executive Vice President, Government Affairs

Elan Corporation

Suite 350

1140 Connecticut Avenue, N.W.

Washington, D.C. 20036


Executive Director

Preclinical Development


One Research Way

Princeton, New Jersey 08540




by Dr. Nancy Chamberlin 8



Introduction and Background -

by Dr. Larry Lesko 10

Overview of Draft Lactation Studies Guidance -

by Dr. Arzu Selen 18

Nonclinical and Clinical Methods to Determine

the Amount of Drug in Breast Milk -

by Dr. Shinya Ito 27

Drug Transfer into Milk:

Clinical Methods and Issues -

by Dr. Patrick McNamara 35

Committee Discussion 45



by Dr. Christine Swenson 72

by Dr. Gerard Jensen 75



Introduction -

by Dr. Mei-Ling Chen 84

Overview of Liposome Drug Products -

by Dr. Francis Martin 88

Pharmaceutical Equivalence: CMC Issues -

by Dr. Arthur Shaw 113

Bioavailability and Bioequivalence:

Biopharmaceutics Issues -

by Dr. Kofi Kumi 123

Committee Discussion 136


(8:34 a.m.)

DR. BYRN: While we're getting ready, let's introduce the guests to the committee. On my right is Arzu Selen.

DR. SELEN: Good morning.

DR. BYRN: Good morning.

Larry Lesko. Then on my left is Shinya Ito and Patrick McNamara. The rest I think are all accounted for. Oh, yes, and Dr. Peg Neville also. I was just introducing the speakers, but there's another. So, we have five guests. Thank you very much for coming.

I'd like to call the meeting to order and read the conflict of interest statement.

DR. CHAMBERLIN: Thank you.

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.

Since the issues to be discussed by the committee at this meeting will not have a unique impact on any particular firm or product, but rather may have widespread implications with respect to entire classes of products, in accordance with 18 U.S.C. 208(b), all required committee participants have been granted general matters waivers which permits them to participate in today's discussions.

A copy of these waiver statements may be obtained by submitting a written request to the agency's Freedom of Information Office, room 12A-30, Parklawn Building.

With respect to FDA's invited guests, Dr. Patrick McNamara, Dr. Frank Martin, and Dr. Leon Shargel have reported interests which we believe should be made public to allow the participants to objectively evaluate their comments.

Dr. McNamara would like to disclose that his employer, the University of Kentucky, has received research funding from FDA and NIH for studies concerning drugs in breast milk.

Dr. Martin would like to disclose ownership of stock in Johnson & Johnson, Imclone System and Alkermes. He is also a consultant to Target Protein Technologies, and he is employed one-third of the time with Johnson & Johnson Alza.

Dr. Shargel would like to disclose that he is employed by Eon Labs Manufacturing Company.

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

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

DR. BYRN: Thank you very much, Nancy.

We can begin the clinical pharmacology group discussion with Larry Lesko who will provide an introduction and background.

DR. LESKO: Well, good morning. It's a pleasure today for me to be here to discuss what I think is a very important topic, perhaps one that's under-appreciated, and that is the drug transfer in breast milk to infants. It's important because, when this occurs, the infant is in a state of rapid development so it becomes a critical situation, particularly in the hospitalized mother who may be breastfeeding or in the neonatal period, in the first 30 days of life where drugs in breast milk can have an impact on development of the child.

Now, we've had probably more than 500 years' experience with the fact that breast milk is perfectly suited to nourish infants. You see reference to it in our culture going back to the 15th century in paintings, more recently in stamps. You can find statues, and very recently a report from the HHS that talked about public health goals for the next decade.

If you go to that HHS document, section 16, which talks about maternal/pediatric health, there's a section there that deals with breastfeeding, and the goal in the United States is to raise the percentage of mothers that breastfeed from the current number to 75 percent in the early days following birth. Further, that report targets 50 percent of mothers for breastfeeding within the first 6 months. So, it's a substantial public health goal, and in light of that, we feel we need to know more about drug transfer into breast milk.

Let's talk about demographics a bit. We're not talking about an orphan population, a small population here. 61 million, the number of women between 15 and 44 years of age. 4 million, the number of newborn infants. 65 percent, the fraction of infants who breastfeed in the hospital shortly after birth. 2.6 million. This is the number of potential recipients of unwanted drug residues that might come from therapeutics in the mother and transferred into breast milk. We can compare this population -- maybe we want to call it a special population -- to other populations we study routinely in drug development, renal patients, hepatic patients. Those numbers are not as large as this group.

Now, new parents want to give their babies the very best in nutrition and they have choices to make about breastfeeding or using commercial formulations. Many choose breastfeeding. In fact, the American Academy of Pediatrics has authored a couple of articles in some of the FDA journals. FDA Consumer, for example, in September 1998 had an article by the American Academy of Pediatrics. They talked about a lot of things but the fact that human milk is made for human infants. It meets all their specific needs.

They say in that article that we have a very scarce amount of information on the transfer of drugs. Oddly enough, commercial formulas for infants is closely regulated by the FDA. In contrast, we have no FDA guidance or regulations that pertain to drugs that might appear in breast milk and be fed to the infant.

Now, when it comes to medications for the breastfeeding mother -- let's say she has a chronic condition, diabetic condition, epilepsy, hypertension, maybe short-term problems like infections -- a big decision has to be made about therapeutics and what we're going to give to that mother. We have to weigh, on one hand, the benefits of the medication for the mother, which are obviously substantial. On the other hand, we have to weigh the risks of medication to the infant. What may be safe for the mother may not be safe for the infant.

And where do mothers frequently go? They frequently go to where a lot of us go for health care information when we can't find it in the package insert. We go to the Internet. This poster is from a Breastfeeding in the Information Age Week that is going to begin this year, next month, August 1st, to try to promote communication about drug transfer into breast milk.

I went to one of the sites to see what I would find in the area of drug transfer in the breast milk, one of the more respected sites, the one on perinatology. And this is only one of many. This one is for professionals as well as educated lay people. I started at the front alphabetically, and I went through a few drugs. I want to share with you some of the things I found.

Look at what it says. Acyclovir, excretion into milk is concentrated. Albuterol, excretion into milk is negligible. Aminoglycosides, most excreted into milk. And this one, excreted into milk but affects on infant are unknown. They may be of concern. And then the last one, very common, caffeine, excreted into milk, but acceptable when not used excessively. These were the drugs for which data was available. And you imagine there were other categories of drugs that had no information or were contraindicated. But this was by far the biggest chunk of drugs in this reference place on the Internet.

Now, when you look across all of these drugs and pay attention to all of those statements, because there was a last column on this Internet web page, which told the mother about the safety of these drugs if she was breastfeeding. What was amazing, despite that information, is that all were rated compatible with breastfeeding. So, if you're reading that, to me it was very confusing about whether these drugs were safe, and I'd have a bit of a problem that they're all compatible, given the statements that appeared on the web page. So, this went on for hundreds and hundreds of drugs, and at the end of the day you walk away saying I don't know a heck of a lot about it.

Well, what happens when you're faced with that kind of uncertain information? Well, a couple of things will happen. The mom will stop breastfeeding, and that results in some detriment to the infant in terms of the benefits of feeding, ranging from nutrition to protection against disease states, and there are a lot of down sides.

The other problem is the mom decides I'm not going to take the drug. There's a risk there because in particular if this is a chronic condition, there may be an aggravation of that disease state.

Now, I think it's interesting, as we talk about drugs in breast milk, that we turn our attention to another source of milk, the bovine mile, and the fact that the dairy industry for a long time has had a systematic system in place to monitor the drug transfer into breast milk. Cows get antibiotics. Cows get a lot of things for therapeutic as well as nutritional reasons. But the FDA Center for Veterinary Medicine has been concerned about this for a long time and has set up a procedure to monitor drug transfer. Antibiotics are one class, and there are systematic screenings in place to determine the safety of milk with regard to drug transfer. CVM has put out protocols for the measurement of drug in cow milk, and in many ways we could do the same thing for human breast milk.

With that introduction, let's talk about why we're bringing this to your attention for your discussion today. We feel that there's a public health responsibility, given the size of this population, as a regulatory agency to do something about the paucity of information in the area. We would like to convey ways to identify and reduce barriers related to medications, which may keep women from initiating or continuing to breastfeed their infants.

We feel that the major barrier is the absence of reliable data, the absence of comprehensive studies on drugs in breast milk. We've gone through our NDAs. We've failed to find well-designed breast milk studies in any of our NDAs. If they're there, they're very rare. We do see some information. Frequently it's incomplete, almost to the point where we feel uncomfortable putting any information into the label. So, what the patient and the physician gets then is a label without any information.

We'd also like to encourage improvements in the science of drug development so that somewhere in an efficient, informative, and cost effective way this information is obtained during the course of drug development so that we can use it for conveying the information to the public.

We'd like to see data on the transfer of medications in the breast milk with some hypothesis about the potential risk to infants. We'd like to include this information in our product labels related to breast milk, not unlike the information we currently are including in labels with regard to pediatric patients, with regard to pregnancy. This is a very similar issue and a problem.

And we'd like to empower women and their physicians to be able to make these rational choices about benefits and risks, to weigh the drug therapy against the risk to the infant, to weigh the benefits of breastfeeding against the benefits to the mother. You need information to do this, and we don't think it's a big stretch to get the information.

We're going to proceed with this discussion through three presentations. Dr. Arzu Selen will talk about the current initiative within the center to develop a framework for a guidance for industry on determining drug transfer into breast milk. The methods to do this I think are very much within our reach. In many ways, they're clinical pharmacology issues.

We're going to hear from Dr. Ito who has considerable experience in this field and in particular the mechanisms of transport in mammary tissue and getting drugs into breast milk that way.

Then finally, Dr. McNamara, who has also conducted extensive research in this area, and he'll touch upon some things.

What we're looking for here is a hierarchy of methodologies that could be used in drug development to gather the information that we need. These may be in vitro methods. They may be in vivo methods. And there may even be drugs that we can take off the table and say that we don't need information for these drugs because we have a high degree of certainty that they don't transfer into breast milk.

If they do transfer into breast milk, we have some questions about the magnitude of the clinical effect of that drug. How do we estimate that? Is there a threshold level below which we can feel safe about that exposure in the infant?

So, these are the type of issues that hopefully we'll get a good discussion of today. Thanks.

DR. SELEN: Good morning. This morning we would like to get your thoughts and your views on drug transfer into breast milk. We want to specifically talk about the methods, the in vivo methods and also in vitro methods.

So to talk about these issues that we would like to discuss with you, I'd like to go over some of the background material. Now, the outline for this talk this morning is, following Larry and after my talk, there will be Dr. Ito and Dr. Pat McNamara. Before we go into their presentations, I will highlight some of the areas that we would like to discuss with you in terms of the key points and the key elements of the questions, and then we'll go to the questions and open it for discussion.

So, this guidance that we referred to is the Clinical and Nonclinical Studies for Drug Transfer into Breast Milk. We fondly refer to it as the Lactation Studies Guidance. There's a big guidance working group. Actually it's a very efficient group of individuals. There are 16 core members. Members are from CDER, CBER, which is the biologics, CVM, and we also have a member from the Office of Women's Health. So, it's a huge, big group but, as I said, a very efficient group, and because of their energy and input, we achieved a lot in a very short period of time. And I'm looking forward to continuing to work with them, as well as our supervisors, Drs. Lesko and Sandy Kweeder. It's a privilege.

Now, the driving force of this guidance, what's behind this guidance, is really in this slide. I just want to take you through the figure. This bar chart has four clusters. The first three clusters represent data, and of these, the light colored bars are the percentages of babies that are breastfed at the time of birth or close to birth. The dark colored, the pink colored bar is the representation of the percentages of children that are still being breastfed 5 or 6 months after birth.

Now, this chart illustrates that there's an increased awareness and acceptance of breastfeeding because we could see that over the years, starting from 1980, 1997 and 2000, the percentage of babies that were breastfed increased from 35 percent to 46 and to 65 percent. So, there's an accepted interest on the benefits of breastfeeding.

But what else is happening? If you look at the solid colored bars, that's the percentage of babies that are still breastfed 5 or 6 months after birth. We have 14, 20 and 16 percent. So, the mothers are not continuing to breastfeed. Really that outlines an issue, as Dr. Lesko was presenting, that there's a serious concern that there's a lack of information that people feel the choice. They either continue breastfeeding or they do not take their medications. There's a conflict.

And this is the area that we have to improve on because the fourth cluster, which is the Healthy People 2010 Goals, says, as Dr. Lesko also pointed out, that at the time of birth or close to that period, there will be 75 percent of babies that will be breastfed, and of those, we still hope that by 5 or 6 months after birth, 50 percent will be still breastfeeding. This is a significant increase.

And if we're looking for such an increase, then we're looking for a ways and means to close the gap in information. So, we have to have the science supporting this objective and, in addition to that, of course, providing information in a way that the mothers who read the prescriptions, read the package inserts, can really utilize the information, and can continue breastfeeding and the level of information is communicated at the level that is clinically meaningful.

So, I think this is the most important slide of my slides, and this is really the background, the driving force of the guidance, why we want to have this information.

Now, objectives of the guidance are, of course, along those lines. We're looking to get information on the amount of drug and/or significant metabolite in breast milk as a percentage of maternal dose, or if there's a therapeutic infant dose, as a percentage of that dose. We just don't want to end the information at that point. In addition to providing the percentage of dose, we also want to make a clinically meaningful recommendation for the mother. So, if it is 2 percent or 3 percent or 5 or 10 percent, what does it really mean? Can she continue taking the drug or is she going to take it at a certain time? So, this is the type of information we would like to include in the guidance.

Now, as I mentioned, this working group efficiently went through a lot of information and literature and spent quite a bit of time looking at two types of studies essentially, two major groups of studies. They could be clinical or nonclinical.

Under the nonclinical group, which is on the left-hand side, you can see that there's the area for the mathematical methods, which is what we also call the log-phase distribution model. These are the calculations that one can utilize drug characteristics to estimate the amount of drug that will be in milk. Following that is another approach which is the animal studies, or another approach is the bottom one, the in vitro methods where the mammary cell lines can be utilized to determine or estimate, depending on the various parameters, what percent of drug is going to be in milk. Of course, there are other approaches such as equilibrium dialysis. So, there's a huge series of nonclinical methods.

In addition to that, on the right-hand side, there are the clinical studies that can be conducted, which will include studies that will be conducted in the breastfeeding women, in the lactating mother, just the mother alone, and then the second group will be only the babies who will be given the milk that will contain the drug. Then in the very last box on the right-hand side, it will be both the lactating mothers and also the breast milk-fed infants. So, the data will be collected from those patient populations.

Now, we're repeating. The usefulness of measuring drug and/or significant metabolites is one of the topics that we want you to discuss with us this morning. We need your views on this. And then, of course, the methods. How reliable are those methods and what are the limitations is going to be the other item.

Now, this is a fairly complex area where we talk about drug transfer and exposure in the infant. We're looking at pharmacokinetics not only in the infant, not only in the mother, but also at the mammary cell level. There are the kinetic changes that will occur. The drug may undergo metabolism. There's the transport. So, in assessing exposure in the infant, we have to have information in all of these areas and a better appreciation, and then the size, at what level this is useful or its clinical meaning.

Now, the drug transfer into breast milk. I oversimplify this in a way. Dr. Neville, you presented to us on the 18th. There are clearly many subsets of these, but if we're going to look at the basic breast drugs, it's facilitated diffusion, diffusion, or active transport. So, one of the things in here is sort of an easy way to get a handle of is the diffusion. For this one, there are many parameters that are related to drug physicochemical characteristics. With utilizing dose, there are publications that show that one can estimate how much drug gets into breast milk.

So, although this is a simple and maybe sort of a soft approach, it does have some value. So, we also want to discuss this and see your views, what percent of drugs really go by diffusion and can we utilize this tool to estimate and the value of the tool as a first step or maybe not so valuable. We'll discuss that.

If we're going to go with this log-transformed phase distribution model, it includes information on pKa, log P values, octanol water partition coefficients, or protein binding, which all of this information is readily acceptable, and it's not really difficult to get a handle on. But again, maybe we'll make some assumptions on the way, and can we really accept all of those assumptions? We'll discuss those as well.

So, essentially one of the parameters we'll also bring up for questions and discussion is milk-to-plasma ratios. This is the amount of drug in milk to amount of drug in plasma. In lactation studies, there are a lot of publications that report this number, milk-to-plasma ratios, and there are also issues with the methodology. It's a single point or a comparison of the area under the curve values. In any case, whichever methodology -- of course, let's work with the best method, which is the ratios of the area under the curve -- we can assess the value of this parameter then. Can we utilize this and how far can we utilize it?

There are publications, as I mentioned before, that look at the maternal drug concentration as C average, uses the milk-to-plasma ratio, and then milk intake, which I have an equation at the bottom of that slide, which is somewhere around 150 mls per kilogram per day. This is essentially for a child, depending on the body weight of the infant. The bigger the body weight, the more milk the infant is going to ingest. So, it reflects that.

And if we are going to work with Dr. Ito's exposure index model -- and he's here also. He will refer to it to some extent, and I think Dr. Pat McNamara will also discuss it. His equation looks at a more adjusted value for the clearance in the infant. So, in the first equation, it really doesn't have a component that relates to the infant, but the exposure index has the clearance of the infant as a denominator. That utilizes that ratio. So, the information is normalized in terms of the infant's clearance of the drug, which is one of the very important components because the drug may be cleared at a faster rate in the mother, but at a very slow rate in the infant. So, it will become a very important consideration, and I'm looking forward to these discussions.

So, essentially I just want to highlight key components of the questions and after Drs. Ito's and Pat McNamara's presentations, we'll go back to these questions and discuss them.

The first question is really based on the importance of measuring drug and/or significant metabolites in milk. Is this information important? In what cases is it? Again, the first question continues. These are things to be kept in mind while you are listening to the presentations because we're going to come back to them.

So, we're looking at what type of methods to utilize. Can we use some information as estimates?

Further on, what parameters can we use to assess safety risk in the infant? This is very, very important because we just don't want to end up with a number that says it's 5 percent or 10 percent, but put that into context, what does it mean clinically.

Following that, question 2 deals with the diffusion in some ways because we talk about the M/P ratios and the log-transformed phase distribution. So, we're saying if you were going to use the log-transformed phase distribution equation or a model, could it be an acceptable first model? And what percent of drugs are transferred into breast milk by diffusion? And that's fairly important because is this a tool that can accommodate most drugs.

Further on, we want to talk about, of course, the potential of actively transported drugs, and are there screens that will help us to identify those?

Finally, the third question deals with M/P ratios. M/P ratios have some limitations and advantages. Let's work with the one that has the best method, best calculations based on the area under the curve. After that, are there other methods? Are there other approaches? Can we also consider approaches such as utilizing only milk data? That's based on information comparing milk-to-plasma ratios.

So, with that, I would like to turn it over to Dr. Ito. After their presentations, we'll go over the questions. Thank you.

DR. ITO: Good morning. I'll be brief.

I have four discussion points today. The first is I would like to discuss why we need data. The second point is what kind of data we need in terms of the drug excretion in breast milk. The third point is I'm going to describe briefly drug transporting proteins in the mammary gland. Finally, I will summarize my thoughts about this issue in terms of what kind of research should be done in a drug development process.

First of all, why do we need data? As Larry said before, the uncertainty about the information compromises breastfeeding, which has tremendous benefits in the infant. I'm going to describe this using the antibiotics and PTU as an example.

Also, if we have data, we can identify certain drugs or groups of drugs which we can adapt therapeutic drug monitoring to individualize our management plan. Lithium is a good example.

Also, if we have data, we can identify contraindicated drugs in breastfeeding.

First, the benefits of breastfeeding. Breastfeeding can reduce many different diseases, especially infections. The infection rate goes down. Diarrhea, pneumonia, bacteremia, otitis media. There are many epidemiological studies. Not only that, this is not good news for people like me who were not breastfed, but cognitive function can increase.


DR. ITO: There are many studies. Especially the key paper is Lucas. On average there is probably an 8-point difference in IQ, which is half the standard deviation of IQ in the general population.

However, if the data are not there, the breastfeeding is compromised. Number one, we did a study to look at the compliance of lactating women who were prescribed antibiotics. If the information about risk assessment of this issue to the women is kind of equivocal, they don't comply with the study. So, we found that out.

The second, PTU. I characterize it as labeling the issue, but it's a chronic medication to treat hyperthyroidism. But again, even if there are data, if the physicians are not aware of the data or the physicians receive negative imprinting from the original labeling of the drug, breastfeeding is compromised. I will give you an example as PTU.

Now, the amount of PTU excreted into milk is less than .3 percent of the therapeutic dose on a weight basis. If this is less than 10 percent of the therapeutic dose, the current wisdom is that it's not a big deal. This is the case for PTU.

On top of that, even if the mothers received PTU and breastfed the infants, the infants' thyroid function is not compromised. So, we have pharmacodynamic data here. Based on that, most experts believe it's all right.

CPS, the bottom, is the Canadian version of PDR. Even this year, the 2001 version, it still says it's contraindicated.

Then what's going to happen? Less than half of the women taking PTU start breastfeeding. This is our data. Look at the control. In the Toronto area, the breastfeeding initiation rate is around 80 percent. So, a tremendous decline in the initiation of breastfeeding in women receiving PTU.

We wondered why. We asked them, and they said those who breastfed while taking PTU said the physician advised them to breastfeed. That's good news. What about those who didn't breastfeed? They said the physician also told them not to.

We surveyed all the endocrinologists in the Province of Ontario, and we found out about half of the physicians don't believe PTU is all right in breastfeeding. So, I think it's negative imprinting. If the labeling had been clear that PTU is all right in breastfeeding, probably it wouldn't have happened.

Everything is in your handout. I changed a little bit some slides and they're are a little bit different. But to save time, I'll skip this.

What kind of data do we need? What do we need to know? We need to know the infant exposure level, how much drug the infant will be exposed to if the mother is breastfeeding while taking drugs. So, to estimate that, we need to know the actual dose of the drug in milk. It's called the infant dose. Or we can express it as percent weight-adjusted maternal dose. It's a percentage of the mother's therapeutic dose. As I said, if it's less than 10 percent, currently we believe it's all right.

To estimate that, we need to measure the drug level in milk. Secondly, we probably need the information about infant serum drug concentration because clearance of the drug in infants is quite different from adults. We may need some pharmacodynamic endpoints if possible.

Also, we can utilize the exposure index. Ours was just briefly mentioned. I'll come back to that later.

Also, we need to assess the effects of maternal drugs on milk yield. Some drugs can decrease milk supply, and that might compromise the breastfeeding.

To understand the transfer mechanisms, probably we need to have an index such as the M/P ratio. It's not crucial but it will be very helpful.

The exposure index is a concept to understand the determinants of the infant exposure level. As you can see, the M/P ratio times 10 is the coefficient which is milk intake, expressed as milliliter per kilogram per minute times 100, because this is a percentage index, divided by infant clearance.

Those are the drugs which have a very high exposure index, and that fits actual observation. This is a conceptual index; however, it fits the observation. So, if we can derive those things in a newly introduced drug to the market, we may be able to standardize our assessment.

Now, the mammary gland has a carrier-mediated systems, active transporters or drug transporting proteins. The clinical implications are there may be some drug interactions in that area. There are not much data on that, and maybe down the road, potential intervention is possible to decrease further the drug transfer into milk.

Even if there are transporting proteins, net transfer may or may not deviate from a diffusion model. That's something we have to consider when we apply the diffusion model to estimate the drug transfer into milk.

I will just focus on the organic cation transporters. Milk is a little bit acidic than plasma. So, the cationic drugs are ionized and entrapped in milk.

On top of that, there are at least several drug transporting proteins for organic cations. So, the excretion of cationic drugs into milk sometimes exceed what we expect from a simple diffusion model. In this area, Dr. McNamara's group contributed quite a lot.

Now, if you look at the organic cation transporting proteins, they're P-glycoprotein, other organic cation transporters, OCT1, 2, 3, N1, and N2, and so on and so forth. Probably by the end of today, I think there may be others.

Now, we checked the expression of those transporters in the human mammary gland. I will come back to P-glycoprotein later. OCT2 is not expressed. However, OCT1, as you can see; MCF12A is a human mammary epithelial cell line. HMEC is a myoepithelial cell line. Actual mammary tissue. Same thing. OCTN1, OCTN2. OCTN2 is a carnitine transporter, which is an essential nutrient for the infants for the energy metabolism and lipid metabolism.

P-glycoprotein, which is a multi-drug resistant protein, is actually expressed in the mammary gland. As you can see, panel b, the surface -- it's confocal. But plasma membranes of the human mammary gland cells express P-glycoprotein, and we don't know yet what kind of contribution P-gp has in overall drug transport in the in vivo situation.

Using the MCF12A in vitro cell model, we can demonstrate, for example, typical organic cation uptake saturation curve. As you can see, carnitine uptake can be also characterized using this in vitro model, and saturation can be also demonstrated.

Using this model, we can try a lot of drugs to derive the IC50 value for the inhibition of carnitine transport. Cimetidine, TEA, choline, guanidine. We are now doing a panel of drugs to look at the relationship between IC50 values of probe compound transport in this model to actually in vivo derived M/P ratio to see whether we can apply this technique to estimate the drug transfer into milk in the in vivo situations.

So, in summary, I think this is what I think in my view we should do. I just took the liberty of naming it levels. Level 0, preclinical study. Physicochemical model, in vitro cell model, animal models should estimate drug excretion into human milk in the in vivo situation. Based on that, that will give the ethical framework to go on to clinical studies.

In the first clinical study, level 1, in my mind probably we should recruit lactating but non-breastfeeding women, for example, women who are weaning breastfeeding. Then we can check a lot of pharmacokinetic parameters, and then detailed clinical studies can be done.

Based on that, then that will increase our confidence level to go on to the actual clinical study using the actual breastfeeding dyad.

So, level 0, as I said, various models to estimate the in vivo drug excretion into milk in humans.

Level 1, using lactating, non-breastfeeding women we can build detail up from pharmacokinetic studies. How detailed? That's a point of discussion.

Level 2, we can go on to the actual breastfeeding dyad to check the dose-milk concentration relationship to estimate variations in the population. Serum concentration of drug in the infant or some pharmacodynamic endpoints.

I think I will stop here.

DR. BYRN: Thank you very much, Dr. Ito.

I think we should wait until Dr. McNamara finishes and then we can have questions for both of you. It's also a very good way to present a lot of material in a brief amount of time. Thank you very much.

DR. McNAMARA: Thank you. Given the time, I'll go quickly through these. You have the slides in front of you, so you won't need all of the information that I'm going to talk about. I want to thank Larry and Arzu for the invitation.

I'm going to talk a little bit more about the clinical studies in terms of the design. I sort of looked at the questions that Arzu sent out and sort of then tailored the talk to address a couple of those issues.

This is one variation of that same relationship in terms of looking at serum concentrations in terms of the dose exposure. This is just my version of that same equation.

Again, in terms of what's the important point, I think it's what concentrations are we going to achieve at steady state. I think most of us would agree that it's probably the chronic dosing situation rather than acute dose that we're concerned more about. So, an average steady state in the neonate is a function of the dose derived from milk and the clearance mechanisms, and I'll get to some of those issues in a minute.

One of the questions was a single time point versus area under the curve approach, milk concentration versus M to S and M to S and neonatal concentrations. Certainly somebody who's not here who's done a lot of work in the area, Dr. Wilson, has talked about the time-dependent milk-to-serum ratio and has been looking at drug distribution into any tissue or other area. There is a potential for a time lag, and he has cited several examples in clinical studies. Here I'll present one study that we had in cimetidine in rabbits that we were using at the time as an animal model.

Here's the blood concentration and here's the milk concentration. This just gives you an example of what the milk-to-serum ratio could be if you picked one point in time to take that, and you see it graphed here where the milk-to-serum ratio could vary anywhere from less than .2 to 15 depending on what point in time you picked that sample. Whereas, the area under the curve ratio is here, somewhere around 1. When we did infusion studies to steady state, indeed they came up with a value of around 1. So, I think this speaks to the issue that one should look at area ratios rather than single time points if one is going to use that milk-to-serum ratio. Again, there are examples in human literature as well.

Milk concentration versus M to S. I think that milk concentrations, while they're sufficient for estimating exposure, I think M to S gives us a better value in terms of getting some idea of the kinetics that are present in lactating women, which may be different. Some insight into mechanism, which is something that's near and dear to my heart. Then it gives us some additional information where we might look at overall modeling of that drug distribution into milk.

M to S versus neonate concentrations. Obviously, neonate concentrations would be very valuable, but there are logistical and ethical issues that make some of these studies maybe difficult to carry out.

I was asked to talk a little bit about some of the models that are out there. I'll touch briefly on some of the physicochemical models, a little on animal models, and some on cell culture models.

There have been efforts to model this for a long time. As you can see, Rasmussen back in 1958 and 1959 talked about the unbound distribution model, trying to look at drugs -- that should say unionized -- where they looked at the pH partition hypothesis and then the various variations of that where we start to account for other things, the fact that drugs interact with proteins in the milk, the fact that drugs can partition into milk fat. There have been various models. It was mentioned the Atkinson and Begg model, a log-transformed model, and one of the papers that I believe was in your binder was one on the neural network where these individuals looked at a variety of components trying to predict M to S, with a number of things.

This is my favorite. It's an equation that we used a lot. It says the milk-to-serum ratio can be accounted for in terms of the ionization, the differences in protein binding where this is the fraction unionized in serum and the fraction unionized in milk. Likewise, unbound in serum and milk, and then a whole-to-skim milk partition ratio.

This sort of breaks it up into the ionization difference. Because the pH of the milk is slightly lower, then you get the possibility of cations being trapped in the milk.

Protein binding. Usually what we see is there's more extensive binding in serum than in milk. Hence, the milk-to-serum ratio tends to be lower for more highly bound drugs.

Then for very lipophilic drugs, the question of partitioning into lipids comes into play which may boost that milk-to-serum ratio very high, especially if we start talking about some more lipophilic drugs, the amiodarones and maybe even pesticides, insecticides that have very high partition coefficients.

This was that neural network modeling, and I simply have transformed the data. Here is the predicted, and here it is on a log-log plot. One of the problems with this is that many of the drugs of interest lie down here, and it's hard to see whether they're predicting or not. In a log-log plot, you can see that a little better. It does a pretty good prediction of those values, although there are some significant outliers that you'll see here. But this might be a good place to start in terms of no data at all.

Cell culture model. This is some data that we did in our lab with help from Peggy Neville who developed the CIT3 model which is a murine model for studies. They form a nice monolayer and you can look at flux studies. Here we look at one drug that we think is actively transported, nitrofurantoin, and you see there's a basal to apical difference here over apical to basal lateral, suggesting a transport process. That transport process is saturable, and you can inhibit with dipyridamole. And if you look in vivo in rats and look at the influence of dipyridamole on nitrofurantoin M to S, you also can see an inhibition. So, again, there's an active transport process. It is inhibitable.

The CIT3 cells are quite valuable, but it is a murine cell system. Now, one thing that that cell line doesn't do is actively transport cimetidine, and cimetidine is one of those other compounds that is actively transported. So, why that is, we're not quite sure yet.

Animal models. My lab and others have used the rat, as well as the rabbit, as animal models. This is some work that Frank Kari did with Peggy Neville looking at nitrofurantoin. Here you see predicted based on that binding and partitioning of .3, and when they actually observed the milk-to-serum ratio, they see something that's quite a bit larger than that, suggesting an active transport process.

I'll quickly go through this. This is, again, milk-to-serum predicted based on this model versus observed, either in rat, human, or rabbit. And those are conducted either at steady state or by looking at the ratios of areas. We started with the rabbit, liked it because it was easy to work with. You get lots of milk, do multiple time points, but then found an article by the folks at NIEH where they looked at cimetidine and saw what looked like active transport in the rat. But they did a single time point. We thought we were better kinetics people than they are, so we did cimetidine in the rabbit and saw no active transport in the rabbit, and then said, well, see, they just did it wrong. Then we said, well, maybe it's species.

So, we decided to do the rat ourselves, and did infusions to steady state, and lo and behold, the rat does actively transport cimetidine. You see the predicted value off the line of identity.

So, the rabbit doesn't look like it's a good model. Now, in my discussions with Peggy, it may be because they don't form tight junctions, and maybe Peggy can comment a little bit on that.

But the rabbit, in terms of predicting an active transport component for cimetidine and nitrofurantoin, did a pretty good job of predicting that. We've done also acyclovir and there is some literature evidence that also suggests acyclovir is accumulated at concentrations greater than predicted by diffusion.

So, the rat is a pretty good model in terms of mechanistically predicting something. Now, if you wanted the rat as an animal model to tell you what the human M to S is going to be, that won't work because the rabbit and the rat have concentrations of lipids and proteins that are much higher than human milk. Also, the pH, at least in our hands, tends to be slightly lower. So, if you're looking for an animal model that you can get an M-to-S ratio that's exactly the same as it is in humans, you won't find that, but we found the rat to be useful mechanistically.

Active transport issues. I'll show you a little bit about clinical evidence. I'll sort of skip the carriers since Dr. Ito presented some of that.

This is data that we generated in cimetidine in humans, looking at different size doses, and I'll just focus your attention down here. This is M to S observed, looking at the ratios of those, versus predicted of 1. So, we see something about six-fold greater. Again, animal studies in the rat that showed saturability, inhibition.

This is something we just published that looked at nitrofurantoin in human milk. We see here a ratio consistent with what we saw in the animal studies both in Frank Kari's work and some that we did, a considerable accumulation of nitrofurantoin as well.

We are also looking at transporter gene expression and looking at that as a potential way of identifying what candidate genes there might be. We've found a number of genes, not just in the cation family, some that are negative. Dr. Ito and I can talk about this one since he sees it and we don't. But we're progressing down this, and I think the next is to do protein gene expression studies to look to see if those drugs are transported by these carrier systems.

Then I think we need some sort of a database that would identify which drugs are substrates for these carrier systems. Then we might have an idea of what to expect in vivo.

Neonatal exposure issues. I'm going to go through this quickly. It was covered quite a bit before this. Again, to save time for questions, I'm going to talk about this.

Obviously, developmental patterns vary with regards to clearance. The most varied is the cytochrome P450 system, and some phase II reactions are inefficient at birth.

Here is some in vitro data that was gathered looking at human microsomes and looking at functional activity as a percent of adults based on a milligram protein. Here you see CYP-1A2, 2C. This was protein levels. 2D6, 2E1, and 3A4. You see these protein levels and functional levels start out very low and progress upwards. So, clearly in the past we've talked about putting up one value of clearance. I think what you should realize is that clearance varies as a function of developmental age here.

I'll skip this one. This one was simply to show you that timing dosing versus when you nurse is sort of a non-issue in terms of trying to avoid the peak concentrations of exposure. I think one needs to think in terms of an overall steady state exposure and not trying to time nursing to miss the peak of drug levels. That's just not going to happen.

Again, this table is a little busy, but the real point here was that if you're looking at percent of dose exposure, you look at this number. Interestingly enough, something we don't think about but actually the maternal clearance, to a certain extent, defines what that percent of dose will be; that is, the lower the maternal clearance, the higher the dose exposure is actually for the newborn.

But if you're interested in concentration ratios, that is, the neonate-to-maternal concentration ratios, it is indeed a function of that neonatal clearance. So, higher exposures of the neonate are a result of either higher M-to-S ratios, lower clearance, either maternal or neonate, and questions about what the first pass effect may be or bioavailability may also be something one wants to think about in terms of exposure.

So, in conclusion, most drugs -- and I would say all drugs -- are going to be present in milk. It's only a question of whether we can measure them with their analytical sensitivity. So, it's not a question of whether they're present or not. They all will be.

Many of them can be predicted based on their physicochemical properties, governed by diffusion. You'd expect the unbound, cationic, lipophilic drugs to be higher M-to-S ratios.

There are some transporter issues. We're just now finding out which transporters are present, which may lead to an accumulation or an M-to-S greater than what we can predict.

I think the real issues here are neonatal exposure. The thing that seems to be missing the most is the neonatal clearance and then whether that concentration that we ultimately do achieve actually results in the pharmacologic or toxicologic response in the neonate.

I think that's it.

DR. BYRN: Thank you very much.

DR. SELEN: So, now we'll go back to the questions.

DR. BYRN: We have them on our handout if you want to go ahead that way. It's up to you. It would be very nice to have them up there.

DR. SELEN: Since everyone has the questions, perhaps we can go to the questions. They were in the handout.

Let's look at the first question, question number 1. I would like to get your thoughts on this one. Now, the first question is asking if it is important to estimate or measure drug and/or significant metabolites in breast milk. I'll open that question to you.

DR. BYRN: Thoughts of the committee?

DR. LEE: I think the answer is yes. Right?

DR. BYRN: Yes. We're saying it's a no-brainer.


DR. SELEN: So, it looks like we're going to move through the questions very quickly. The first one was the answer was a sound yes.

Part A says, for what type of drugs do we need this information on the extent of drug transfer into breast milk? Are there certain drugs that we don't want this information on?

Yes, Dr. Venitz.

DR. VENITZ: I think there are two ways of looking at it. One is what do we know, what can we predict in terms of their potential extent of delivery. And that seems to be able to be predicted based on some of the stuff that we heard today.

The second approach is how likely are they to be administered to women who are lactating, and what are the potential consequences.

So, we can kind of triage how important the information is and, further down the list, what kind of data would you require in order to make the decision.

If the drug is unlikely to be administered to lactating women, if the consequences are benign, then in vitro data or predicted based on physicochemical characteristics might be sufficient. If, on the other hand, like PTU, the consequences could be disastrous, you might require what you call I guess a level 1 or level 2.

DR. SELEN: Thank you.

DR. BYRN: One other idea might be to have some kind of flow chart that would take you through these decisions, a decision tree, flow chart.

DR. SELEN: I see the members of the committee are raising their hands on that one. We went through several decision trees. It's very close to the hearts of many.


DR. LESKO: Can I interrupt here? I think in this idea of the hierarchy and the framework for ethical studies, I would think we wouldn't want to advocate studies that we feel are unnecessary to get this information. So, looking at this question another way, can you think of ways in which you can take drugs off the table and be confident that certain pieces of information would suggest that these drugs are not going to be a problem, therefore I won't go any further?

I don't know if what Dr. Venitz laid out is all of the criteria one might think about. Certainly the clinical consequences are one thing. Potential for use may be another. For example, could I conclude that if the milk-to-serum ratio is less than a certain value, I might take that off the table for consideration in doing a clinical study. If I conclude that the drug is not absorbable in neonates, for example, large molecules, aminoglycosides, I could take that off the table.

Can you see the value of an approach like that, or is that perhaps taking some risks that would not be acceptable?

DR. JUSKO: Larry, what you indicated is largely a great deal of common sense. But I'm not sure you would exclude drugs only on the basis of a low milk-to-plasma ratio because of differences in potency and differences in clearance, exposure in the infants. It's too simplistic to do it that quickly.

DR. DOULL: I guess I have that same kind of concern. Larry, in the introduction you said if you can prove the drug doesn't get into the milk, fine, take all those drugs off. And Dr. Venitz said, well, if it has a great therapeutic ratio, you might take a bunch of drugs off for that reason.

I think the guidance needs to recognize that it needs to be a case-by-case decision rather than blanket. Dr. McNamara, you mentioned pesticides, and I'm thinking of the Food Quality Protection Act which is blanket-issued for pesticides and was a factor of 10. You have to say something about the susceptibility of the infant to the agent. Therapeutic index may not blanket-predict for both the mother and the infant. So, I think the argument for a case-by-case analysis rather than a blanket kind of approach should be part of the guidelines to ensure that we don't really make the same mistake that we made with pesticides.

DR. BARR: I'd like to thank the speakers for what I thought was an incredible review of an awful lot of information, a lot of concepts in a very clear and concise and comprehensive manner. Thank you.

I wanted to get back to this issue. It seems to me that the biggest unknown we have in most cases is what this real exposure rate is to the infant because we just don't know how they metabolize those drugs in most cases and all of those factors. So, that's number one. We almost have to go back and say how do we collect that information. That, of course, is the biggest mystery.

So, it means that we really have to be very cautious I think in how we view any kind of transfer into the milk, particularly for drugs that may have consequences, significant pharmacologic consequences.

One of the other factors that I wanted to ask about that I didn't see up there is if you have drugs that are relatively lipid soluble, of which the membrane may not be the rate-limiting step, then you get into kind of a Renkin dialysis method in which the actual milk flow may be the rate-limiting step. So, this single point ratio may not be valid and may be, in fact, milk flow dependent. If you had low flow versus high flow, it may change. Have you got any data on that?

DR. SELEN: Well, that's a very good point, but I think most of the literature I have seen works with the concept of like six feedings per day at 150 mls per kilogram per day. Now, if you're saying if the baby is more frequently fed. But, of course, they'll be ingesting more milk and they could be getting more drug. But I don't have an answer for you. I don't think so.

DR. BARR: Not just the volume, but just actually the blood flow. In other words, you've got plasma flow on one side of the membrane and then you've got milk flow, which depends upon the milk production rate. And that may alter, I think, the number for drugs that are fairly lipid soluble in which the membrane is no longer --

DR. SELEN: Dr. Neville is an expert on this, but I'm not quite sure I can see the milk flow to be as fast or as rapid as the plasma flow.

DR. NEVILLE: So, it's different than, for example, if you're looking at the kidney. Milk accumulates in the breast until the baby actually feeds, and then it's pushed out all at once. So, basically it's a question of how fast does the drug -- and Patrick does this much better than I. I'm a physiologist. How fast does the milk transfer across the mammary cell? Where does it distribute? Is it in the milk fat, which makes it a very different system from any other because 4 percent of the human milk is fat. If the drug is lipophilic, that makes a difference.

In terms of the amount of milk produced per day, on average at 1 month it's about 600 mls per day, and at 6 months it's about 800 mls per day. So, we're not dealing with huge differences there.

One of the points I wanted to make at some point -- and since I've got the microphone, I'll make it -- is that a group of infants that we must consider very carefully are the premature infants. These are very small infants. It's becoming very clear that these infants need human milk. The anti-infective properties, the brain development properties of the polyunsaturated fatty acids -- that 8 percent change in intelligence actually came from premature infants, not from term infants. There are some real issues with premature infants. Then, of course, you have the problems of metabolism compounded enormously. So, in designing therapeutic guidelines, I think this particular group really needs to be taken into account.

DR. SELEN: I think the premature infants is a great point because also, like Dr. Lesko was presenting with the 4 million babies being born, of that 4 million, 11.5 percent is premature infants. This is just in the white population. If you look in the black, it goes up to 16 percent apparently. So, it's a serious number. We're looking at 400,000 or more per year.


DR. MEYER: It seems like one approach, rather than thinking of drugs you can take off the table, which is always difficult because you can come up with examples why nothing should be removed from the table, is to work first on those drugs where it would be very important to know whether they're transported or not.

DR. SELEN: Thank you.

DR. SHARGEL: I presume these studies are going to be for drugs going for an NDA submission for new drugs. Since women are already taking drugs that are already on the marketplace, how much information is going to be tried to be gained from those drugs that are already marketed, being consumed, realizing that there is a lack of knowledge on many of these drugs? Is there any attempt to provide an incentive to obtain this information?

DR. SELEN: Incentive is a different question, but at least I can answer one part that doesn't deal with the incentive.

You were noticing that in the chart that a big percentage of mothers discontinue breastfeeding. So, if they're taking a chronic medication, they have to be on it. It's already approved. I think they still deserve to have the information to use the medication properly. So, we would rather that they have the information and they continue taking their drugs. I think it's critical that this information is available not only for new drugs, but also for drugs that are out there.

In terms of incentive, that becomes a different issue that Dr. Lesko might want to address.

DR. LESKO: I was just going to suggest, because of timing -- and Steve is watching the timing of the discussion, I think we should turn the discussion over to Steve to sort of make sure we stay on track, at least to moderate the discussion.

DR. BYRN: We can do it together, if you want, Arzu. I think it's good if you stay up there because you have a little bit more knowledge than I do about the field, but I'll proceed with the discussion or try to summarize what people are saying. So, we'll do it together, sort of like a talk show.


DR. BYRN: I think we've got a pretty good answer or some ideas for item A here. The ideas are that we look at drugs that are dangerous, and that would be one approach. The other approach would be to look at a case-by-case basis but try to take drugs off the table on a case-by-case basis. Is there anything more on item 1A?

DR. BARR: Going back to Marv's statement I think really makes a lot of sense. It seems to me that the priority ought to be to really look to those drugs which are already on the market which are widely used, likely to be used by women and may have consequences. Make up that list first. I think that would be the place to start.

DR. BYRN: So, look at the risk and -- go ahead, Marvin.

DR. MEYER: Drugs that women are on that they can't get off of. Anticonvulsants, for example.

DR. BYRN: Let's go ahead to the next one. Go ahead, Arzu. You just couch it. We'll discuss it, and then I'll summarize.

DR. SELEN: So, if we need this information, when would it be appropriate to estimate or when shall we be collecting data? Of course, the type of studies are the in vitro, nonclinical studies, or clinical.

Dr. Jusko.

DR. JUSKO: It would seem that the FDA would be able to encourage pharmaceutical companies to carry out animal studies to determine milk/plasma kinetics of drugs. Dr. McNamara suggested that the animal data is sort of confusing in that lipid and pH differences exist between animals and humans. But I would encourage the development of animal scaling principles in conjunction with the kinetic principles that you have been using in order to develop a way to convert the animal data into predictable human parameters. That way one could mine the great deal of information that one could get from animals and use that for assessment of potential human exposure.

DR. BYRN: I also like Dr. Ito's idea of using women that are weaning children because there you could get probably quite a bit of good data. You'd have to work fairly quickly, but you could get good data and there would be no exposure risk. I don't know how feasible that is, Dr. Ito. Is that feasible to do that in general?

DR. ITO: I think so. Of course, there are some limitations such as they are weaning, so it's not quite really a physiological state. However, it's going to be a good starting point, probably as good as good animal data.

DR. BYRN: Dr. Neville?

DR. NEVILLE: There's another population that I think, with proper organization, might be very good for studies, and these are the mothers of premature infants who are pumping milk for their infants. Some of them make a good deal of milk so that they can store up for a couple of days and have some extra milk for a double. So, I would encourage people to get in touch with their neonatal nurseries. That actually might be a population where you can also look at the premature situation.

DR. BYRN: Are there any other ideas on this one?

DR. BARR: Just one comment. It seems to me that if you were to set up one or several centers in which you obtain women who are weaning who would be willing to serve as a milk donor, not necessarily weaning, because you can continue the production of milk for a long period of time if one chooses to do so. And if you were to get a cohort of women who were willing to do that in a center in which several drugs could be done in succession, take those which are most important and put them out, it seems to me that some of these could be done fairly quickly.

DR. BYRN: Should we go to the next question?

DR. SELEN: So, what parameters can be used to assess the safety risk in the infants? Let's say the drugs that get into milk or they're predicted to get into milk. What are the parameters that we can utilize? For example, like a certain percentage is acceptable or not.

DR. JUSKO: I think you need a hierarchy of information as you've been discussing. It's perhaps easiest to get the percent of the maternal dose that gets into breast milk, but that's not as important as having milk/plasma ratios that you can factor in with maternal exposures and different dosage levels. But then that's not as important as having the infant exposure index and having additional information about potential toxicity in the infant. You may not be able to get all of this level of information at one time, but all should be part of a composite body of the data.

DR. DOULL: I agree. That's the case-by-case argument.

I do object to using both words "safety" and "risk." Safety is a yes/no question; risk has no bottom. What you're talking about is toxicity.

DR. SELEN: Yes. Risk in terms of it's a safety risk.

DR. DOULL: But in order to define that for an individual drug, you need to know whether the toxicity comes from the kinetics or whether it comes from the dynamics. Those are safety questions which can be answered. It's just the two words together that disturbed me.

DR. LESKO: One of the things that Dr. Ito presented was the exposure index as a way of combining the factors that would influence exposure in the infant, and a suggestion was made of a 10 percent cutoff. Presumably below that, one would feel relatively safe; above that, one would be perhaps concerned.

The other part of the exposure concept, I noticed, in the slide was a bit of variability in it. For example, lithium had 2 to 30 percent variability.

It sort of gets me to the question of variability. If clinical studies were deemed to be important in this area, obviously there are factors that will limit the size of those studies. I wonder if people that have conducted these studies can comment on what they feel would be the logistical aspects of it and the number of subjects or volunteers that would have to come into a study to try to get some data that would be credible.

I guess the other part of that is the exposure index. Is 10 percent something that people feel good about?

DR. BYRN: Dr. Ito, can you talk about the variability in these studies?

DR. ITO: I think if we know a certain drug has quite a variability in terms of the exposure level to the infant, that tells me at least that we need to monitor, individualize the approach. So, I think to me that's good enough. At least we can tell that there are huge variabilities in the drug excretion to milk. I think that's good information to have.

DR. BYRN: So, you're saying we could do maybe a rather small study. If it's tight data, we're fine. I guess you would recommend the 10 percent level. If there's a lot of variability, then we know there's a problem and there's going to have to be monitoring.

DR. ITO: Right. That would be my approach. 100 percent exposure index is actually the same as a therapeutic dose to the infant, and 10 percent is one-tenth. So, I'm quite comfortable with 10 percent as far as dose-dependent effects are concerned.

DR. BYRN: Is the committee comfortable with 10 percent? That's a key thing I think if we could say we're comfortable or not.

DR. BARR: I don't feel comfortable with 10 percent mainly because I don't really know what it means.

I think the problem is that we have a given dose and we're assuming that we know something about the relative toxicity of that dose to a neonate, to an infant. In most cases, we probably don't know that simply because we don't even know it for pediatrics, let alone neonates.

If we do this project, which I think we certainly should -- I think it's very necessary to do -- it means that more drugs will be used by women who will be nursing presumably, that we will be telling them that it's going to be reasonably safe. So, there almost needs to be a second phase in which that actually is monitored sometime in a clinical way once that begins to be done, particularly for critical drugs.

DR. BYRN: Now, are you willing to use the 10 percent level for that second phase? In other words, if it was below 10 percent, you could say it's presumed safe and then do a second monitoring, or do you think a second clinical trial should be done?

DR. BARR: Well, I'm not sure I have the information to make that judgment. I think 10 is a reasonable arbitrary number, but I think it ought to be considered on an individual basis. If we have, for example, a drug which is very essential to a woman but may have some toxicity to the woman -- one of those critical drugs that we're talking about -- then I think that would have to be evaluated with all the knowledge that's known at that point in time.

DR. MEYER: Steve, I support Bill. I can't pick a number based on a 20-minute presentation, and I don't think I could pick a number if I heard a 30-day presentation because everything is going to be different.

We haven't talked much about intra- or inter-subject variability. If the woman is on a drug that has a 30-fold inter-subject variability, what's the infant's variability? Is that comparable? Does that go up? Does the infant tolerate more drug as the woman tolerates more drug, or are the receptor sites growing like the metabolizing enzymes are changing during those early days? There's no complication? What about drugs where a woman starts on a drug after the first 10 days of life? That infant is going to respond, according to the one slide, differently than if she starts on the drug or is taking the drug on day 1.

I might feel comfortable in picking a number if the drug is used, say, in a newborn center and you could say, well, the infants there take this drug routinely and tolerate such and such a dose. I'd have a feeling that might be safe, but there are an awful lot of unknowns out there that really deserve careful consideration before I'd put it in a label, okay to take.

DR. BYRN: We probably now need to go much faster. So, let's go ahead.

DR. SELEN: So, essentially this is dealing with the diffusion model. So, if you go to 2A, using the model such as the log-transformed diffusion equation, which incorporates PKs and log P's, and that information, protein binding, now would we consider that as a useful first step?

DR. BYRN: Everybody is saying yes.

DR. SELEN: Okay.

So, then the following question is what percent of drugs can we estimate -- this is going to be an approximation -- are going to be transferred into milk by diffusion?

DR. JUSKO: That seems to be a major research question. Drugs that are actively transported. The number of those needs to be evaluated much more extensively.

DR. LEE: Yes, I agree. I think the question I had is the pattern pretty similar to kidney?

DR. SELEN: Can you repeat the question, Vince?

DR. LEE: Yes. I was just wondering whether or not the process of secretion of drug into milk is like secretion into urine.

DR. McNAMARA: I don't think we have enough data on that. I think there are some examples like cimetidine and probably nitrofurantoin that would suggest that it looks like that, but you can find other examples of drugs that are excreted into the kidney by active transport processes that don't seem to have that same pattern in milk. So, it's not something that you can equate one to one in terms of the numbers of drugs. I'd say that the percentage is small. There's probably a handful, but it's based on how many drugs have been studied, which is also not a large number.

I think the 10 percent number -- I'm going to get back to that earlier point. I think the question really has to do with the clearance mechanisms in the neonate, and if one can anticipate, based on what we now have to know in terms of a new drug, is this drug cleared by one mechanism, is it predominantly renal, and do we know something about that in the neonate, or is it predominantly 3A4 or 2D6, or how is it predominantly cleared? The more pathways there are for clearance, the better chance that that drug will be cleared to an extent, on a body weight basis, more like the adult, than if you were depending on one particular clearance pathway and that one happens to be undeveloped.

DR. BYRN: So, the answer to B is we need more research. It's a research question.

And C?

DR. SELEN: So, if we were going to look at active transport, what type of approaches could be possible screens, reliable screens?

DR. JUSKO: I think what's very nice is that the physicochemical principles provide the first screen because of the great degree of predictability based on pH and pKa and such. Then when the predictions for the models are not confirmed by either animal data or human data, one then sees the probability that there's some additional transport mechanism. But fundamentally these things need to go hand in hand, more research into investigation of transport mechanisms, which will then identify additional drugs that may be transported by those mechanisms. In turn, evaluation of the toxic drugs that women may take.

DR. BYRN: So, we're hearing that C is obviously a research topic.

Yes, Larry.

DR. LESKO: Question 2 more or less pertains to methodologies that we characterize as in vitro based on concepts of physicochemical characteristics and so on. For several drugs -- I think Dr. McNamara showed nitrofurantoin and some others -- we have pretty good data.

I guess my question is, would we derive more information from these studies if we, in fact, included "internal" standards in the procedures, in other words, put in drugs we know about in terms of M/P ratios and drugs we know about in terms of drug transfer into milk, and then use those as reference points to assess the relative risk of the new drug that we might be talking about? It's a way of interpreting the data instead of trying to interpret it in terms of an absolute risk. What do people think about that notion, and could people see that as a framework for moving forward on these types of studies?

DR. BARR: I think that's an excellent idea. It really brings up the issue. Most of the industry now spends a fair amount of time determining permeability by a variety of in vitro methods, and they characterize them exactly that way. You standardize your system, make sure that it's working well relative to known ingredients or products.

It would seem to that this is an area that in the NDA process ought to be looked at. It's something that would be obtained routinely. This would be the place where you'd like to collect that information. For those drugs that are likely given to women who may be nursing, it would seem to me that this would be a reasonable thing to ask in the IND process.

DR. BYRN: Let's go ahead.

DR. SELEN: The third question is with the M/P ratio, and we want to discuss the advantages and limitations. Of course, as Dr. Pat McNamara illustrated, there's a difference between if it's a single point or an area under the curve comparison. If we just work with the area under the curve comparison, the best approach, then what are the advantages and limitations?

DR. BYRN: Area under the curve. Are there thoughts on that? Bill?

DR. JUSKO: It's always better to get more data whenever possible. So, again, there's a hierarchy. Maybe in some women one can only screen to get an M/P ratio, but whenever possible a full profile should be obtained.

DR. BYRN: I think that's a general consensus of the committee.

DR. MEYER: But the caveat is whenever possible.

DR. BYRN: Right.

DR. MEYER: These women have other things to do than get stuck 12 times in a 24-hour period. And they certainly don't want an AUC done on their infant.

DR. SELEN: So, the M/P ratio would be in the mother. That's the intent. However, do you see any limitations with it, in addition to being stuck for 12 hours.

DR. VENITZ: Well, you're assuming that you're measuring all the active moieties. Maybe you don't measure the metabolite and it's the metabolite that does something untoward to the infant. Right now we are talking about areas under the curve of the active moiety.

DR. SELEN: The intent is it's the drug and/or significant metabolites because we're interested in that.

DR. VENITZ: Known metabolite, right? You're asking about limitations. That's an intrinsic limitation.

DR. SELEN: Yes, good point because it might be a different metabolite. So, the value of this may not be really pertinent for the baby.

DR. VENITZ: Right.

Another kinetic limitation would be, do you have dose proportion kinetics in terms of the maternal pharmacokinetics? So, a single dose might not predict what's going to happen at a higher level.

DR. BYRN: Larry?

DR. LESKO: I was going to go a little bit beyond these three questions, but it's relevant. Assuming that a sponsor develops this information during the course of drug development and provides some information to fill the gaps that we've been talking about, what do members of the committee think about how to transfer this information to knowledge within the label? There are a couple of possibilities.

One is obviously to just put descriptive information, say, in the clinical pharmacology section of a label. People may or may not read that or be able to interpret it.

There's another way and that is to interpret the data as we might, say, drug interaction data that comes out of drug development.

Do people have thoughts on what they see as the most effective way to transfer information to knowledge so that it gets out to the clinician and to the patient so that they can make some sense of it in making decisions? What would be the format for that communication of knowledge?

DR. BYRN: Ideas?

DR. SHARGEL: Larry, I think I'd approach that as a marketing kind of thing, looking at labeling in general. I think you have a group looking at revising labeling, just recently a guidance, and how labeling is reviewed by practitioners, pharmacists, and others. So, I would take that to a different level than this committee.

DR. MEYER: And you can't beat the Internet for disseminating to patients.

DR. BYRN: Are we done? Bill, one more comment.

DR. JUSKO: Yes. The NIH, the Women's Health Initiative, and probably the FDA have a program ongoing where there are going to be RFPs issued to solicit more extensive pharmacokinetic/pharmacodynamic drug efficacy studies in pregnant women. It would seem like this whole initiative should also be connected to that one since it's the logical final stage to study this question.

DR. SELEN: In fact, it is. There's one individual from that group from the Office of Women's Health. In the interest of time, I didn't want to go into the background and the details of this, but there is a big initiative, as Dr. Lesko has mentioned I think at one point in time, about the pregnancy labeling and all of these are the subcomponents.

DR. BYRN: I think we should conclude this session.

DR. SELEN: There's one more.

DR. BYRN: Okay, there's one more question.

DR. SELEN: There's the last one, and this is the last one. What other approaches would be acceptable? Sometimes we hear points made such as instead of obtaining milk-to-plasma ratios, just obtaining milk-drug concentrations. Will that be adequate or do we want to normalize it with exposure in the mother by obtaining plasma data? And what other approaches do we think might be useful?

DR. JUSKO: I think we've seen that you get much more mechanistic information by having the milk/plasma ratios. But once again, sometimes only one may be obtainable, but it would be better to get more comprehensive information whenever possible.

DR. MEYER: What's the reproducibility? If I took the six feedings and measured a drug concentration in that total 150 mls, how much variability would I have throughout the day and night? Would a concentration tell me anything, or does it vary by a factor of 2 or 3 or 4?

DR. SELEN: The point you're making is a very good one because there are so many changes in the milk composition that affect the amount of drug in milk. So, like we mentioned in the draft guidance, it's important to collect all of the milk. There's a difference in the fat content in the foremilk versus hindmilk, as Dr. Neville can also elaborate on. So, depending on how you collect it, there's going to be variability, and depending on when you collect it, there's going to be variability. So, I think it's very important in these studies to collect all of the milk at all collection times and then having a sample from that, because the foremilk versus hindmilk -- this is published information. The big difference is in concentration of a lipophilic drug.

Does it address that adequately? Or Dr. Neville might wish to add more.

DR. NEVILLE: The sampling of milk. You don't just go take a milk sample. There are some very standard ways to do this, but it has to be done right. I don't think this is the place to talk about that.

The other thing that ought to be considered is are there drugs that have effect on milk yield? It's very clear that estrogens have an effect on milk yield. Estrogen-containing contraceptives at high doses definitely have an effect on milk yield. At low doses, I'm not so certain of the data. But there very well may be other drugs, particularly drugs that interfere with the hormonal mechanisms that regulate lactation that may affect milk yield as well. While that isn't a purview of this particular group, it's something that in the long run we really need to work on if we're going to have women getting starting breastfeeding even properly.

DR. BYRN: Other comments?

(No response.)

DR. BYRN: I think we are done. Thanks very much for the presentations, and I thought we had a very good discussion.

Let's break until 10:30. We have two presentations in the public hearing. Then we will try to start the liposome discussion at 11:00 if we're able to.


DR. BYRN: We'll begin the open public hearing. We have two presentations in the liposome area. The first speaker is Dr. Chris Swenson who's going to make a presentation on liposome drug products, the importance of supramolecular structure.

Just as a comment for the committee, these two speakers are going to present important issues about liposomes that they think the committee should be aware of. That's the purpose of these presentations.

DR. SWENSON: Well, thank you. You just made my introduction for me. My name is Chris Swenson. I represent Elan who are involved in discovery, as well as drug delivery, including liposomes. I just wanted to make a brief presentation today and also make you aware that Elan is willing, indeed eager, to assist the committee in any way they can on these subjects.

I wanted to talk about the importance of supramolecular structure of lipid-based and liposomal drug products. The supramolecular structure can affect the biological properties. Therefore, I think understanding the physical as well as the chemical characteristics of these types of drug products is essential during process development, scale-up, and manufacturing, and in establishing appropriate release specifications.

Abelcet I'm going to use as an example. This is a lipid formulation of an amphotericin B, which is an antifungal drug. During its development, we looked at a number of physicochemical characteristics. We evaluated morphology by microscopic techniques. We evaluated the homogeneity of these lipid-based suspensions by density gradient techniques. We characterized the complexation, the nature of the complexation, between the lipid and the drug by spectroscopic techniques, as well as a biological assay, which was hemolysis in vitro or red blood cells. The supramolecular organization was evaluated using differential scanning calorimetry, NMR, and both small- and wide-angle x-ray diffraction.

By using these techniques, we were able to devise a model for what the real organization of the molecules in this drug product were. We found that the amphotericin B alternated with the phospholipid -- and here the phospholipid is blue; the amphotericin is yellow -- in a cylindrical structure with the hydrophilic face of the amphotericin facing towards the inside. These cylinders were actually interdigitated membranes with a length of -- actually that should be 25 Angstroms, not .25, which is about the half the width of a normal bilayer. That's because this is an interdigitated membrane. Then these complexes then associated to form a larger membrane of associated complexes.

Understanding this supramolecular structure gave us the ability to control our manufacturing process, but also to establish appropriate quality control tests. On the left-hand side, these are the normal sorts of tests that you would use for a parenteral pharmaceutical. On the right-hand side, are those tests that are specific for lipid-based or liposomal products, as well as those that have a supramolecular structure, and therefore you have to consider things like particle size and the nature of the complexation and the drug-to-lipid ratio.

We were not the only ones to recognize that formulating amphotericin B with lipids resulted in a drug that was less nephrotoxic and had an enhanced therapeutic index. There are three products marketed in the U.S. that are based on amphotericin B-lipid interactions. The Fungizone is formulated with a detergent, deoxycholate, but Abelcet is a large, ribbon-like complex. AmBisome is a small unilamellar vesicle, and Amphotec is a small, disc-like complex. So, they're all very different.

And this is borne out by their pharmacokinetic properties. Abelcet has a much greater clearance than Fungizone, whereas Ambisome's clearance is much less than that of Fungizone, and Amphotec is in between.

So, the supramolecular structure, in addition to the lipid composition, affects the biological properties of these drug products, and I think should be considered when you're considering the pharmaceutical equivalence and the bioequivalence of these products.

Thank you.

DR. BYRN: Are there any questions for Dr. Swenson?

Actually I have one very brief question. Obviously, these are solution liposomes, so they're dynamic. Things are moving. Is that correct? How fast does an amphotericin molecule, if we could sit on it, move from one liposome to another or move --

DR. SWENSON: When these are in aqueous solution, as they are in the bottle, they don't move.

DR. BYRN: They don't equilibrate.


DR. BYRN: Interesting.

DR. JUSKO: I have one question. I assumed that what you're presenting for pharmacokinetics represents the total quantity of drug, both free and in the formulation, which is generally the problem with these products. You can't make the separation?

DR. SWENSON: That is absolutely correct.

DR. BYRN: Our next speaker is Dr. Gerard Jensen from Gilead Sciences, and he's going to make a presentation on liposome therapeutics.

DR. JENSEN: What I wanted to do today is just contrast the role of process and material quality control versus formulation. Most of the literature on liposomes is dominated by formulation dependence of properties. I wanted to highlight the process of manufacturing of them is, in many cases, of equal importance.

Similar to the previous speaker, we're speaking of the third dimension here. We're looking at the chemistry of multiple components, physical assembly of many thousands of molecules. There are elements of that assembly that are critical: size and the distribution of size, the level to which the drug is entrapped or encapsulated in the species, and related to that is the structure. So, if I have a drug molecule, is it in the interior solubilized, is it in the interior precipitated, is it in the membrane, that sort of thing.

This is a table of stress and consequences, and actually they're not meant to be paired up, but on the left side are the things that we do to liposomes, filtration, refrigeration, freeze-drying. Brownian collisions result from their natural motion in the bottle. IV administration, that sort of thing. And then on the right are consequences that, depending on how a liposome is assembled, can be the result of those stresses.

The usual way that this liposome technology is represented involves if I need to make a new product, I want to have reproducibility of that product from lot to lot. I want to maintain the therapeutic index enhancement, whatever that is, whether it's on the efficacy or on the toxicity side, and I need to have a stable formulation. I need to have a commercially viable shelf life. Again, most of the formal literature describing these situations involves composition, lamellarity. Is this an SUV, a small unilamellar vesicle, or is this a large liposome or that sort of thing?

What we'd like to emphasize, though, just as important is how it's all put together, material quality and characterization, and I'll give a few examples.

Again, going to the literature, this is a paper that's only three years old basically reviewing liposome science. They're showing a couple of pharmacokinetic plasma half-life curves. In the white triangles, we've got a so-called conventional PC:cholesterol liposome. The red circles are a trace in this article where they're illustrating the effect of putting this polymer coating on the outside. The implication is that without that polymer coating, conventional liposomes wouldn't survive.

But going back through the history of our own company, many years ago we had an imaging agent called Vescan and the yellow squares are a rendering of what the blood stability of those particular liposomes were. Those were also simple, conventional PC:cholesterol liposomes of very similar composition to those cited in the article.

More recently, we've seen a clinical development product, MiKasome, which has a 100-plus hour terminal half-life, and the other two traces I've shown is Doxil, which is the long-circulating peg-coated liposomal doxorubicin, and the yellow boxes are a research formulation of the same drug with no peg on the outside. I don't mean to imply by this that those two are equivalent, but I do mean to show that within the range of so-called conventional liposomes, with the same composition, you can get very different biological stabilities based on how they're made.

Another area of interest involves characterization of liposomes. I did mention earlier particle size determination. What I've shown here is a very common looking size distribution that you might get from any of the commercially available dynamic light scattering instruments that are used to control liposomes, median particle size. Those instruments give you many reported parameters, but the only one that has a validatable precision is the mean and median particle size, and it has a precision of about 3.5 percent.

However, we know that liposomes are a distribution of sizes and that there is a heterogeneity of size. The real question is how are we sensitive in these techniques to change in that distribution and most importantly detection of small subpopulations, for example, of larger particles.

If you reprocess the data on a linear scale, you get a more realistic picture of what we're looking at. So, the squares are a linear scale rendering of size distribution based on volume weighting. What you can see quite clearly to larger size is a tail. The importance of that tail can, just for example, be in two different areas. In one case, it's been shown long ago that liposomes of greater than 100 nanometers will, for example, be accumulated by Kupffer cells in the liver, and those smaller than 100 nanometers won't. So, having a difference in this tail, in terms of the number of particles that are in there, can give you a different biological response just from that.

Then the other plot here, which are the open circles, is a shell to interior volume ratio. So, this is the ratio of the lipid shell to the interior aqueous base. Across the size distribution, you can see that for small liposomes, they're dominated really by the lipid portion, and for larger liposomes, they're dominated by the aqueous portion. So, that can affect how the drug is held in a liposome. And there are many other consequences of size distribution. I just wanted to point out a couple of them.

So, what techniques do we have to study that? This is the median diameter, which I mentioned was the validatable, precise value, as a function of spiking with large liposomes. So, for a 50 nanometer liposome, if I'm spiking 230 nanometer liposomes into it, I don't see much happening in median diameter. Even some of the less precise, but more sensitive passing diameters, so 90 percent and 95 percent passing diameters, you can have up to 5 percent larger particles in there and not really notice that in your measurement.

So, at Gilead years ago, we took advantage of turbidity and we developed a proprietary method for screening liposomes. We call it normalized quantitative turbidity. It takes advantage of the r to the sixth dependence of light scattering on size in the range 50 to 300 nanometers. What I want to point out here is the first phrase here, which is man versus machine. Experienced liposome folks will look at a bottle and they'll be able to tell you whether there's a tail in the distribution or not. What we wanted to do was be able to quantitate and validate that kind of measurement. So, that's what we've done with this assay.

This is an example of two preparations of liposomal product, DaunoXome, both of them over 15 months' shelf life exhibit stable median particle size diameters. One of them is a commercial lot and one of them is a development lot that was identical in formulation but had different processing parameters. This is what this normalized quantitative turbidity is doing. So, in the case of the commercial lot, it's stable as a rock for 14-month period that we're looking at here, but in the case of the other lot, which had different processing parameters, we start to see the growth of larger particles in the tail.

There are other characterizational tools that may be available. In each case for each product, you need to evaluate whether there is or is not value in them. You certainly don't want to use them all for every product, but these are some examples.

This is an intermediate. So, from the processing point of view of liposomes, you usually have to prepare a lipid intermediate, which is the combination of lipids and sometimes drug. This is a differential scanning calorimetry. It's basically a thermal melting of those liposomes, identical formulation, but very different structure resulting from that. And then that in turn leads to different properties of the resulting liposomes.

This is a cell-based assay we developed around the liposomal amphotericin B product AmBisome. This is a very simple thing, essentially a titration of amphotericin B in rat blood and looking for potassium release after incubation. We're comparing this to Fungizone, which is the detergent formulation. So, we see quite a shift in concentration needed to induce potassium leakage, and this is a measure of how tightly held the drug is. We developed this into a quality control assay, which has a 9 percent RSD and a good correlation to lethal dose testing, which is what we used to do.

But I want to illustrate the point of process versus formulation. I want to show this slide. Essentially on the x axis we have a K50, which is a 50 percent potassium leakage parameter derived from that assay, and on the y axis, we have an LD50, which is the corresponding lethal dose. These are three available commercial products of amphotericin B, and this is the commercial AmBisome product.

These lots here were made again with different processes by identical formulation. This shift here is basically the result of those different processing conditions. Chemically those formulations are identical. And I want to emphasize that those differences would not be evident in a PK analysis because since the free drug is so toxic way down here on this end of the scale, these differences are the result of far less than a percent of the total amphotericin B behaving differently in those formulations. If you have a drug that's beginning to push up the toxicity curve like this, tiny amounts of drug that is not entrapped in the same way are going to give you some significant consequences biologically.

So, just to finish I want to say that we're not trying to say that making liposomes is magic, and we also want to emphasize that it's not all in the formulation. But it does involve high quality, well-controlled components -- I didn't go into that, but that's a key issue -- precision assembly and rigorous quality testing and control.

This is just an example. This is 10 years of AmBisome production going back to about 1990. Over the last 100 or 150 batches, for median size we have an RSD median size that's the same as the assay precision. It is an achievable reality, but it does require a significant investment of time.

DR. BYRN: Any questions for Dr. Jensen? One question.

DR. ANDERSON: Under characterization, you had ESR. This is the first time I've seen anyone put that up there. What were you looking at?

DR. JENSEN: ESR can sometimes be used. It's a technique where you use spin labels and you can put them in the membrane. Sometimes if you have a drug that's supposed to be encapsulated in the interior of the liposome, you can determine whether or not some drug is in the membrane as well by looking at the response to a spin probe.

DR. ANDERSON: So, you tag this.

DR. JENSEN: You can tag a lipid or that kind of thing, yes. We've never used that in a quality control setting, though.

DR. BYRN: Thank you very much.

I think we should go ahead now with our formal program on complex drug substances, liposome drug products.

I'd like the special participants to come up and sit at the table, and these would be Klaus Gawrisch, Burton Litman. Okay, they may be stuck in traffic or something. They're not here yet. We are a little bit of ahead of time.

Mei-Ling, should we go ahead and start?

DR. CHEN: Sure.

DR. BYRN: Okay, we'll go ahead and start, and our participants hopefully will join us in process. We are a little bit ahead, which is unheard of, so we can accept their late arrival.

DR. CHEN: They may show up after 11:00 I think.

Good morning, everyone. This session will be devoted to the discussion of liposome drug products.

Liposome drug products, as you may know by now, represent a unique class of dosage forms that has been developed in the past two decades.

So, what are liposomes? Liposomes are microparticulate lipoidal vesicles that are used as a carrier for improved delivery of a broad spectrum of therapeutic agents, and these may include chemotherapeutic agents, imaging agents, antigens, immunomodulators, chelating compounds, hemoglobin, and others.

Liposomes can be given by various routes of administration. It could be delivered by intravenous, subcutaneous, intramuscular, topical, or pulmonary route of administration. But most drugs that we have seen so far are for intravenous administration.

In the Center for Drug Evaluation and Research, we have a coordinating committee that deals with scientific and technical issues related to complex drug substances, complex dosage forms, or complex reagents used to manufacture drugs. This coordinating committee is currently co-chaired by Dr. Yuan-Yuan Chiu, who is sitting here to my left, and myself.

As you may know, under this coordinating committee, we have a liposome working group that is involved in the policy and guidance development for liposome drug products. This working group has recently prepared a draft guidance for industry on the submission of new drug applications for liposome drug products, which is currently going through internal review in the agency.

This is the cover of the draft guidance. As reflected by the title, the guidance talks about chemistry, manufacturing, and controls, human pharmacokinetics and bioavailability, as well as labeling information. The document, however, doesn't provide corresponding information for abbreviated new drug applications, that is, generic drugs.

The key issues that are not addressed in the draft guidance are related to the equivalence comparison in the area of chemistry, manufacturing, and controls, CMC, and bioavailability/bioequivalence. These are the core issues for this advisory committee discussion today. Dr. Shaw and Dr. Kumi from the working group will present these issues later on, so I will not get into the details now.

The agency has, in fact, broached these issues on liposome drug products to a public workshop in April of this year. The workshop was cosponsored by the American Association for Pharmaceutical Scientists, AAPS, FDA, and USP. The workshop focused on ensuring quality and performance of sustained and controlled release parenterals that included liposome drug products.

The participants discussed critical formulations and process variables in order to develop necessary specification/characterization that assure product quality and performance. They also discussed bioavailability, bioequivalence, and pharmaceutical equivalence. However, no conclusion was reached at the workshop regarding the appropriate approaches for demonstrating the sameness between two liposome drug products.

So, to continue the AAPS workshop discussion, we are bringing the topic to this committee for your consideration. What we would like to do today is to share with you the advances in pharmaceutical technology, the unique features of liposome dosage forms, and some of the regulatory concerns for these drug products.

This is the agenda for today for this session. After my talk, Dr. Francis Martin will give you a brief overview of the liposome drug products with an interesting example comparing two liposome products with the same drug substance, doxorubicin. Dr. Martin is from Alza Corporation, and he's the key person who's involved in the development of Doxil, doxorubicin liposome injection. He has over 25 years of experience in liposome-based systems. I personally thank him for his participation today.

Following Dr. Martin's talk, our FDA staff, Dr. Shaw and Dr. Kumi, will present the CMC and the bioavailability/bioequivalence issues respectively.

Today actually we are also expecting two experts from the National Institutes of Health, Dr. Litman and Dr. Gawrisch. I had hoped that they would be here by now. They will join us for the discussion.

The topics for discussion, after all the presentations, are shown on this slide. The main purpose of today's discussion is to share with you the general information and regulatory issues. Recognizing the complexity of the questions and issues involved, we will not be seeking specific advice from this committee today at this time, and we would like to come back to the advisory committee sometime at a later date for further deliberations after we've conducted more research and investigation.

Finally, I would like to take this opportunity to thank all the committee members, our speakers, guests for your time and effort in helping the agency to address these regulatory questions so that we can move forward in the area of liposome drug products. Thank you.

DR. BYRN: Thanks very much, Dr. Chen.

DR. MARTIN: Thank you, Mei-Ling, for inviting me to speak today on this issue that's near and dear to my heart, liposome drug products. What I would like to do is give an overview and, as Mei-Ling had mentioned, an interesting example, interesting comparison. I have to make the disclaimer that the proposed classifications and observations I make today are those of my own and don't necessarily reflect the opinions of others.

Traditional drug delivery systems, DDSs, have been designed really to control the input rate of drugs into the central compartment. If you look at this list of drug delivery devices on the left, oral devices, patches, pumps, implantable depot formulations, inhalation formulations, these are really designed to control the input rate into the central compartment, which then controls the input rate into tissues and presumably controls or influences the pharmacodynamic effect.

Drug delivery devices are able to affect this input rate constant and therefore affect the entry of the active ingredient into the central compartment and presumably into the tissue compartment. So, this is sort of the basis of the simpler drug delivery systems.

Now, liposomes represent a drug delivery system that actually enters the central compartment, because I'm going to restrict my comments to intravenously administered liposomes. So, they introduce an additional compartment. Any pharmacokinetic model, one has to consider now the volume within the liposome. The liposome itself enters the central compartment with the drug on board and distributes to tissues and can distribute differentially to tissues depending on the formulation, and I'll describe that in a moment.

The kinetics then describing these distributions must include liposome-specific rate constants as well as drug-specific rate constants. Just to confuse you all -- it certainly confuses me -- this is a diagram of a liposome entering the central compartment. Now, this drug could leak from this device or the device could be taken up by a tissue, such as the mononuclear phagocyte system, or it could be taken up by another tissue, and then the drug released from the device within the tissue. So, it gets very complicated in terms of interpreting pharmacokinetic information, and this is just one example.

What I'm going to try to do is describe sort of retrospectively my take on the evolution of liposome design, and some of this is retrospective but it falls into interesting categories. So, based on selection of drug and the clinical indications, technology families have evolved in liposomes. The existing products I would maintain represent members or species of these families. The reason I'm trying to categorize this is I think it will be helpful in terms of drafting some guidelines on how to see whether or not these members or species within the same family are equivalent in different ways.

These are the four liposome intravenous products that are approved. AmBisome you've already heard about. DaunoXome you've heard about. This is daunorubicin in a liposome. Doxil, which is doxorubicin in a liposome. And Myocet, although it is not approved in the U.S., it is approved in Europe and there's a lot of information around on it. It also includes doxorubicin.

So, what I would like to do, since these are comparables, in that the drug is identical, is I would like to Doxil and Myocet in terms of the formulation and some of the pharmacokinetic parameters and then propose that liposomes be classified based on pharmacologic behavior.

One possible classification would just be a vehicle where the majority of a drug in the liposome is released immediately upon introduction of the particle into the central compartment. So, this is less of a carrier. It's a carrier into the blood stream only, and then the drug and the liposome part company. And there are possible safety advantages relative to other vehicles, such as cremophor, less hemolysis, for example. But there are also infusion reactions associated with liposomes. So, the tradeoff here in terms of safety is uncertain, but there are certainly examples in development of what I would call vehicle formulations of liposomes.

Then another possible classification would be liposomes that are designed to be taken up by the MPS system or the RES system. They're synonymous. These are the macrophages which reside primarily in liver, spleen, and bone marrow. MPS liposomes that are designed for uptake into the MPS do have advantages, the most important of which is a safety advantage because they avoid peak levels, and they form a depot within the MPS cells. The drug then reenters the central compartment, but does so at a rate which avoids peak levels, yet maintaining, more or less, the AUC of the free drug administered at the same dose. And clinical benefits, in terms of safety, have been documented for these types of liposomes in the form of both Myocet and AmBisome.

Then the other classification I would propose is liposomes that are designed to avoid the MPS by surface modifications. You want to, in this case, keep the drug on board the liposome because what you'd like it to do is distribute to the tissues and not necessarily to the MPS tissues but to other tissues. So, this is a design feature that has led, I think, to an interesting spectrum in terms of the rate at which these particles are taken up by the mononuclear phagocyte system. So, these are the ones that would be designed to be uptaken by the RES and these to be avoided.

And there are some intermediate ones here that have been described already. DaunoXome, for example, has a half-life of a couple of hours. There's another one under development with vincristine that has a half-life in that same order. These others have half-lives on the order of minutes. And as I'll show you in a moment, Doxil has a half-life on the order of several days.

So, this would be my family tree in terms of the evolution of intravenously administered therapeutic liposomes: the MPS uptaken ones, those that avoid the MPS, and the vehicle.

The way this has all worked out over the years is that design strategies have been introduced. For example, let me take the ones that are designed for MPS uptaken. They are just made large without any surface modification, and they are taken up rapidly by the RES or MPS.

In terms of avoiding the MPS uptake, two paths have been taken over the years: the pure lipid path, that is, make liposomes out of just lipids, but design them in such a way that they avoid RES uptake as long a possible. There are two versions of these, different lipids that are very solid lipids and small particles, and one is in development, a vincristine product, another one, the NX211 is in development, and DaunoXome I believe is a product of this pathway. Surface modified, polymer modified. There's one example which is Doxil that I'll talk about more.

And one vehicle formulation is under development with Taxol.

Now, these families, if one thinks of liposomes in this way, will help in terms of issues of equivalence and bioequivalence and pharmaceutical equivalence because comparing a product of one of these to another, as I'm going to show you in a moment, may not be very useful. But comparing members of the same family or the same lineage may be helpful. So, let me get to the comparison.

This list is sort of a redundant list of what was discussed at the AAPS/FDA/USP workshop. Just what were the critical influences on pharmacology of liposomes? Well, the drug, the size of the lipid it's made from, and surface properties.

In terms of the drug, the class of drug, the drug's intrinsic properties, clearance, its toxicity side effects, these are all pretty obvious things. Where's the target site? And how the drug is encapsulated depends a lot on the chemistry of the drug. So, the drug itself has important effects.

The liposome size has important effects. As was just mentioned by the earlier speaker, distributions in sizes are usually the case with liposomes. There are these outliers sometimes on the edges that have to be considered for safety and other reasons. And very large liposomes can actually cause micro-occlusions in the lung and the brain. So, these things have to be carefully examined.

Also, with respect to targeting to tissues, there are windows in terms of extravasation. If one wants a particle of these sizes to extravasate into tissues such as tumors, there is a window of opportunity there. If they're too large, they will not enter. The size of these windows in tumors now has been probed and there's some information about the dimensions of the defects in these capillary beds that will permit particles to extravasate.

The structural lipid is important. Including cholesterol in the liposome is important. The fatty acids that the phospholipids carry are important in terms of their phase behavior and so on. So, these are all things that will influence the pharmacology of the liposome.

Surface properties, whether it's a naked lipid or whether or not it is polymer coated. Does it have a surface charge or not? If it does have a surface charge, what is the charge density? All of these things do have effects on the interaction of these particles not only with each other but with formed elements in blood, with proteins in blood, because there's a lot of electrostatic interactions that go on.

What I would like to do now is compare the two doxorubicin products for which there is the most clinical information, Myocet and Doxil. What I will do is go through quickly the design features, morphology, pharmaceutical properties, the format the products are in actually as products, pharmacokinetics, and then a few observations about the comparison.

Myocet's size and its loading battery was selected to maximize the payload, and it has a fluid lipid matrix, a very simple phospholipid cholesterol matrix. This product is designed to rapidly be taken up by macrophages. This is to avoid peak levels and attenuate the toxicity associated with doxorubicin, namely cardiotoxicity. This creates this depot from which the drug reenters the blood stream, mimicking a slow infusion, which was the objective because slow infusions were known to reduce the cardiotoxicity of doxorubicin. Yet, it does maintain the plasma and tissue AUC, comparable to similar doses of the free or the conventional doxorubicin.

Myocet is about 180 nanometers in diameter. You can read the lipid components here. One thing I would point out is the very high lipid-to-drug ratio, so there's a lot of drug per particle in this particular system. The drug is loaded by a nifty pH gradient in the pharmacy, and when the drug enters the liposomes, it forms these very interesting organized fiber bundles shown here in an electron micrograph. So, these are the Myocet liposomes and you can see the drug has formed a precipitate here in the form of fibers, very organized fibers within the liposome itself. The formation of these fibers in some cases deform the liposome from its normal spherical shape into what I call the coffee bean shape.

The product format for Myocet is a three-vial system. So, it comes to the pharmacy as a three-vial system. The contents of the vials are listed here. This is from the product labeling in Europe. One is just doxorubicin. The other is the empty liposomes and the other is a buffer. Instructions are included for reconstituting this system. Those are listed here. You reconstitute the doxorubicin as you would normally reconstitute doxorubicin. You turn on your heat block because a heat block is required, and then adjust the pH of the liposomes in the liposome vial by just shooting in the buffer from the third vial, so this creates a pH gradient low inside the liposomes and high outside the liposomes. Then the last step is shown here, shake vigorously. The drug enters the liposomes and the loading is done in the pharmacy. So, this is quite a bit different than some other systems.

And this is the reason why because once the drug is loaded, it will start to come out of the liposomes over a few-hour period. So, this is just showing the release of the drug from the Myocet liposome at two different pH's at 37 degrees. You can see that, in just a few hours, going from 100 percent down to 20, 30, 40 percent of the drug comes out of the liposomes. This is why the drug cannot be supplied as loaded. It must be loaded in the pharmacy and used before there is much release from the liposome.

This is the pharmacokinetics, recently presented at ASCO, of Myocet. Shown here is the liposome encapsulated doxorubicin, the total doxorubicin, and the metabolite, doxorubicinol. And I think this is an interesting plot, so I'm going to spend just a few minutes on it because I think it tells a very interesting story.

First of all, the investigators worked out a way of separating encapsulated from free doxorubicin in plasma. This is not a difficult thing to do with a little column separation technique. What you can see is during the initial period here, as the drug is being cleared, it's still in the liposome. So, the drug is not bioavailable at this point. It is being cleared in the form of liposomes. When you look at the tissue distribution, at least in animals, where it is going is to the RES cells. You can see after about 24 hours, the free drug, the liposome encapsulated drug, and the total drug part company indicating that the drug now is separated from the carrier. Very likely this is drug that is reentering the central compartment from the macrophages. Moreover, the appearance of a metabolite pretty early in this whole story would suggest that there is some release of drug immediately from the liposomes which enters tissues and then is metabolized and reenters the central compartment.

So, it's a complex sort of a composite of reactions that's going on. Clearance of liposomes. There is some leakage of drug from the liposomes, uptake by the MPS cells, destruction of the liposome, reentry of the drug into the central compartment. So, you can see it is, from a pharmacokineticist's point of view, a very complex system to interpret.

These are the pharmacokinetic parameters of Myocet, and I would just point out the differential between the half-life of the encapsulated doxorubicin, which is just a few minutes, versus the half-life of the total doxorubicin. So, this disconnect suggests that the drug and the carrier are parting company.

Again, Doxil's size and payload and loading was also selected to maximize loading. The size was selected also to be smaller than the Myocet liposome because of the potential for it to circulate longer and to extravasate. It is pegylated; that is, the surface is coated with polyethylene glycol. And it was intended to passively accumulate in tumors by extravasating from blood vessels in tumors into the interstitial spaces of tumors. And the lipid matrix was designed for plasma stability, yet to break down in the tissues. So, this was a delicate balancing act.

These again are the attributes of the product. It's about 100 nanometers in size. The lipid components are listed there. You'll notice it has a lower, about half, drug-to-lipid ratio than the Myocet product. And this is loaded using an ammonium ion gradient instead of a hydrogen ion gradient. But the drug precipitates in the liposome similarly to the Myocet liposome, as I'll show you in a moment.

This is from the product labeling. Doxil is supplied already to go in a vial, ready to be diluted.

This is a cross section view of the product. It's about, as I say, 100 nanometers in diameter. The drug is inside the liposome precipitated in the form of a sulfate salt of doxorubicin. The polymer is loaded onto the outer surface. It actually is on the inner surface as well. It's not illustrated here, but there's a dense layer of polyethylene glycol on the surface and there's a single membrane between the external medium and the drug.

This is an electron micrograph of the Doxil liposomes, and you'll see similar features to what you see with the Myocet. The drug precipitates in the form of a striated gel inside the liposome. There is some deformation from the spherical shape. This we believe is a critical feature to keep the drug in the liposome.

So, we now have a precipitated drug, and this is not unexpected. What we do is we load it in a way where we have a sulfate salt inside the liposome, and as the doxorubicin enters, it forms doxorubicin sulfate which is insoluble in water, so it precipitates. This keeps going until all the drug is in the liposome, and it stays nicely in the liposome by virtue of this loading method.

This is the stability of Doxil in an in vitro release at 37 degrees to show that this product does not release in vitro so that it can be shipped as an already loaded liposome.

This is the pharmacokinetics of Doxil, also recently presented at the ASCO meeting. I would point out that this is over now a 30-day period. So, the time axis here is different than that which I showed with Myocet. Shown here is the liposome encapsulated doxorubicin curve, the unencapsulated doxorubicin curve, that is, the actual drug that is free of the liposome. Now, this is not further fractionated into what is protein bound, what is not protein bound. This is the total amount of drug that is not in the liposome. Then this is the metabolite, doxorubicinol.

The half-life for the liposomal encapsulated drug is about 2.5 days. The free drug has a similar half-life, and the metabolite doesn't begin to appear for a few days after administration. So, a different time frame here in terms of the circulation of this particle.

These are the pharmacokinetic parameters. I think I'll skip that.

The tissue distribution of Doxil then provides these long plasma residence times, slow uptake in the RES, and preferential accumulation in tissues with compromised or what I would term chaotic vasculature, such as tumors. These are some micrographs recently presented by Rakish Jain in Nature showing, indeed, the chaotic nature of some of these vessels. These are in colorectal tumors, human xenographs in animals. For example, the red staining here is the endothelial cells. The yellow are the tumor cells in one of these colon tumors, and you can see gaps in the side where there's no red staining. In a normal vessel, of course, you would see endothelial lining the entire surfaces of the capillary. In fact, there are tumor cells that form part of the wall of vessels in these kinds of tumors.

This is an electron micrograph of a similar vessel showing the strange architecture of the endothelial cells with these bridges across the capillaries and these gaps that are shown here by the arrows. These gaps are on the order of about half a micron. So, these holes that are present, at least in this kind of tumor, are about a half a micron in diameter, which relates to this size window issue I was alluding to earlier. If your particle is too big, it simply will not pass through these defects.

So, the whole idea of one of these tissue targeting liposomes is to have the particle physically extravasate through these gaps into the interstitial spaces of the tumor.

This is evidence again from Rakish Jain. These are vessels within a living rat actually. This is a window overlaying a tumor in a living animal. These animals had been injected 24 hours earlier with fluorescent liposomes, and these are the vessels within the tumor. These patches of fluorescence indicate extravasation of the liposomes that were injected 24 hours earlier, to point out that the distribution is rather focal. There are areas where there are many liposomes that have extravasated and areas where there are few. So, this is not a homogeneous process within the tumor. It appears to happen where new vessels are sprouting. So, where angiogenesis is taking place is the weakest area and has the most defects, and that's where these particles are extravasating.

I'll just show some clinical evidence of this, again to make a point about the movement of these intact particles into tumors. This is a Kaposi's sarcoma patient. I would point out a lesion on his left leg and thigh because in the next slide I'm going to show serial gamma scintigraphy of this patient after he received Doxil liposomes that didn't contain doxorubicin but contained indium-111 chelated inside the liposome. I'd also point out this lesion on his left foot.

This is the patient again at 4, 24, 48, and 96 hours after administration. If you note, first of all, in the first time point, there's a large blood pool. This is mainly circulating liposomes, still in the major vessels in the chest and so on. You can see a little bit of activity in the bladder. This is the unencapsulated indium that's immediately eliminated in the urine, but very little uptake in the target lesions we're looking at.

The uptake begins to be seen about 24 hours. This is the lesion in the foot, and then 48 hours and 96 hours you see the peak uptake in these indicator lesions in the foot, in the calf, and in the thigh. Indeed, the activity here, the specific activity in this lesion is similar to the spleen, which is an RES organ. So, there is uptake in the RES, but the activity density is similar in the lesion as it is in any normal organ.

Now, the point of this slide -- there are several take-home messages. Number one, this process is slow. You see that it takes days for these liposomes to percolate into these lesions, and KS lesions are particularly leaky.

The other thing it's important to note is these represent intact particles. So, what you're seeing here are the movement of intact liposomes into these extravascular spaces because if the indium were released, it would immediately be eliminated and you wouldn't see it. It would be eliminated in the urine. So, this represents intact particles.

This will come back to a point I will make at the end, that the disposition of liposomes in tissues is very important. Following them in the blood is also important, but following them at a tissue level is I think equally important for these tissue targeted type liposomes.

This is another patient from the same series showing another histology. This is a patient with a very large cavitating lesion in the upper right lobe of the lung. You can see some RES uptake at this time point. This is 96 hours after administration. This is the posterior view. There is also some uptake in the ilium, probably representing bone marrow uptake.

But this is the tumor. The MRI of this tumor showed that it was a cavitating, necrotic center. So, you see less activity in the center, which makes sense. You see most of the activity on the periphery, again suggesting that where the angiogenesis and growth of the tumor is taking place is where these particles are most likely to extravasate.

Another patient in this series was a head and neck cancer patient. Again, 96 hours after administration, you see a very nice blood pool image of the liposome circulating at this time point. The primary tumor is up here in the head. This is the head now rotated 90 degrees. You see the primary tumor under the tongue and a positive reactive node in the neck.

What's remarkable about this image is if you integrate the activity represented in these two lesions, it represents 5 percent of the total dose of radioactivity injected 96 hours earlier. So, although this is a passive process, if these liposomes are around long enough, you can get a significant amount of a dose that you inject in a peripheral vein into a target tumor.

The dosing and safety profiles of Myocet -- again, the same drug -- are also different. The recommended dose for Myocet is 60 to 75 milligrams per square meter at a 3-week interval. It is approved in combination with cyclophosphamide, so this is the usual dosing regimen for a cyclophosphamide/doxorubicin regimen. The dose-limiting toxicity is the same as doxorubicin. It's myelosuppression.

Now, Doxil is different. The recommended dose is lower than that of Myocet, and there's a reason for that because at higher doses, single doses, we see mucositis and stomatitis as the single dose-limiting toxicity, and the multiple or cumulative toxicity is a skin toxicity called palmar-plantar erythrodysesthesia, or PPE.

There's always the good and the bad. The good part of this tissue distribution is that you can get these particles into tumors. The bad part is it also redistributes to other tissues. One of them happens to be skin, so again getting to the point that the fate of these liposomes at the tissue level is increasingly important, especially for these long-circulating particles. I think this complicates the matter for the regulators and the pharmacologists here to provide guidance in terms of equivalence.

The conclusions for this comparison are these two products are identical drug, but they have been derived from different families of liposomes, different loading. They have different pharmacokinetics and tissue distribution patterns, different safety profiles. They're certainly not bioequivalent, and my comment would be they're apples to oranges.

So, in summary here, based on the selected drug and the clinical targets, these families have evolved, and it may have not been very prospective, but nevertheless this is the way the families have fallen out. I think it's important to think of them in these families because then I think it narrows the view somewhat in terms of the guidances that might be drafted in terms of comparing members of the same family because I think that's going to be easier.

They represent a complex system that does more than control input rate. The liposomes actually carry the drug into the central compartment and then into tissues. So, new assay approaches I believe are needed that reflect the disposition of liposomes and the release kinetics of the encapsulated drug in tissues, as well as the plasma pharmacokinetics. The plasma pharmacokinetics, even if you are careful getting encapsulated, unencapsulated, and all that, is useful, but it's not enough particularly in terms of demonstrating bioequivalence.

So, I will stop there and thank you for your attention.

DR. BYRN: Questions for Dr. Martin? Marvin?

DR. MEYER: Do you have evidence that if you took two different liposomal products that differed in some fashion, however, and they had identical encapsulated and unencapsulated drug-plasma concentrations, they could still have different tissue concentrations or site of action concentrations?

DR. MARTIN: Yes, I do. In fact, I have an example of one. We have taken Doxil, the commercial product, for example, and modified it slightly. This is work underway at the University of California at San Francisco where they're interested in active targeting of the particles. So, they've attached to Doxil an antibody. They have carefully looked at pharmacokinetics and tissue distribution, and there is no difference between Doxil and the targeted Doxil. And the only difference is the targeted Doxil has about 20 molecules of the antibody on its surface. So, there is no difference in the tissue distribution or pharmacokinetics.

Yet, there is an improvement in the antitumor activity in animal models that overexpress the receptor to the antibody. So, they've looked carefully at an EM level, what's happening to the liposome once it gets in the tissue. With Doxil, the liposome enters the interstitial spaces of the tumor and it just sits there. It's not taken up by epithelial cells. It sits there and is opened up eventually over time because the lipids hydrolyze and the drug is released. It's the drug that has to then enter the cell. With the case of the targeted Doxil, the entire particle is internalized by the cell by virtue of the ligand that's on the surface.

So, there's a perfect example of my point, that at the tissue level, the disposition of the particle can be critical, and how we measure that is going to be very, very difficult. But it's an example of why it's a complicated system and why, I think to your question, they can appear to be identical in every other respect, but their fate at the tissue level can be quite different.

DR. BARR: Do you have suggestions on how this tissue level ought to be measured?

DR. MARTIN: Not really. We've looked into things like microdialysis, for example. The problem is the disposition is different in different tissues. For example, the tumor might be different than the liver might be different than the skin. They're different cells. These liposomes are taken up by some cells and not by others. So, you get differential uptake in tissues and differential disposition in tissues. So, what tissue are you going to look at? You can't be poking microdialysis probes into every tissue. It's not going to be useful. Not really. I think it's an area that's open for development.

DR. BARR: If you have two plasma curves in which you characterize both the liposome and the total drug and found that those two plasma curves were in fact identical, would you expect there to be any differences in tissues? Or is that possible?

DR. MARTIN: There can be, yes. You're talking about now the total drug and the encapsulated drug. Well, in the case of Myocet, for example, they part.

DR. BARR: I'm assuming that you're looking at the same family of liposomes, rather than trying to compare them across classes. I understand the difficulty there.

DR. MARTIN: I don't have an example of different tissue distribution. If you have the same pharmacokinetics, the tissue distribution does seem to be similar, but what can be different, as I mentioned, is the fate at the tissue level. So, that's the complication because that could lead to a differential pharmacodynamic effect.

DR. BARR: In other words, within product bioavailability you wouldn't see that as a problem in terms of looking at bioequivalence within the development of a single product.

DR. MARTIN: Right.

DR. BARR: The problem would be coming if you were looking in terms of bioequivalence between a reference and another product.

DR. MARTIN: Correct. I think pharmacokinetics can be very useful to demonstrate product equivalence, physical chemical equivalence, the tendency for it to leak, for example, in blood or not. I think that would be very useful. But for bioequivalence, the ultimate activity of the drug, at least in the case that I cited, is different, and it can be positive or negative. So, that's where the complication is; it's a bioequivalence issue.

DR. BYRN: Vince.

DR. LEE: One quick question. Frank, how would the commercial source of a given phospholipid affect the pharmacokinetics?

DR. MARTIN: I don't see that as an issue. They're just chemicals. Lipids that form the liposomes are defined chemical entities, at least in the products we're talking about here, at least for the Doxil product. It's a defined chemical so that you can introduce specifications so that you're sure each time you have the same molecule. That's just chemistry. I don't find any problem with that. That's the simple part I think.

It's when you assemble these things into this structure and have all these other features, polymer coating or not, that's where you get into the issues of demonstrating comparability among liposome products.

DR. BYRN: Thanks very much.

DR. MARTIN: You're welcome.

DR. BYRN: Our next speaker is Dr. Arthur Shaw who's going to address pharmaceutical equivalence, CMC issues.

Art, before you start, I'd like to welcome our guests, Dr. Klaus Gawrisch and Dr. Burton Litman. Thank you very much for coming.

DR. SHAW: Thank you. I want to thank the committee. I also want to thank Nancy in particular and the IT people who got these slides loaded early this morning.

I am a member of the liposome working group, and Dr. Zhou is away and I was asked to prepare this talk. I want to thank the other members of the committee and also Dr. Diane Burgess, who is visiting the FDA on sabbatical, who helped us with some of the information and some of the slides.

So, you say liposome and I say liposome. We should be glad that some of the original names for these weren't adopted. They were first described in the literature, so far as I know, by A.D. Bangham, and they were called Bangosoms for a while.


DR. SHAW: Also called a smectic mesophase.

So, a liposome definition is a microvesicle composed of a bilayer of lipid amphipathic molecules enclosing an aqueous compartment. This is to be distinguished from a micelle where you essentially have a monolayer.

Liposome drug products -- I'll call them LDPs -- are formed when a liposome is used to encapsulate a drug substance, and the drug substance can be, as we have seen, either within the aqueous phase or sometimes, if they're particularly lipid soluble, within the lipid bilayer.

Here's a diagram that Dr. Burgess sent us, and you can see this is essentially a single lamellar vesicle. You can have multiple layers, and the polar heads point towards the aqueous phase and the lipid tails, the fatty acid tails form a lipid bilayer.

There are a number of reasons to make LDPs. You can do this for targeting or site-specific delivery. You can do this for extended release, delayed release. Those of you in the PK business know the difference between extended and delayed release. And also for internalization to promote the intracellular delivery of the drug.

Now, as we have heard, liposomes and liposome drug products present some unique challenges. One is the question of characterization of the drug product, looking at the physicochemical characteristics. These are some of the examples. And the biopharmaceutical characteristics which have been discussed by the previous speakers and will also be discussed by Dr. Kumi.

One of the important points that needs to be kept in mind, and this is in line with one of the questions earlier, is the influence of the lipid composition on the properties of the membrane. The permeability and stability of the liposomes can be influenced by the rigidity or the stiffness of the lipid bilayer, which in turn is determined by the size and shape of the liposome, if you have a radius of curvature or a less radius of curvature, and by the lipid composition.

Purified lipids, when they are formed into liposomes, demonstrate a property of a phase transition, a gel-liquid phase transition, which is usually marked by a sharp transition temperature, essentially a melting temperature, which you can see on differential scanning calorimetry. This Tc is affected by the fatty acid side chain, the degree of unsaturation, the chain length, and the polar head group. DSPC, distearoylphosphatidylcholine, and DMPC, dimyristoyl and dipalmitoyl PC, liposomes all have different Tc's and the only difference there is in the length of the side chain.

So, the issue for concern, in terms of CMC, is demonstration that a liposome drug product is the same when there are manufacturing changes, and this includes changes by the same manufacturer and when there is a new manufacturer.

Before I get into some of those issues, I wanted to address some of the classification of liposomes. We've heard some of this before. There's the size and lamellarity, the MPS uptake, and the coating. I want to emphasize these classifications are not mutually exclusive.

Dr. Burgess provided us with this slide. You can have small unilamellar vesicles, or SUVs, and I'm trying to think of a way that we could make an acronym that comes up "minivan" but we haven't come up with that.


DR. SHAW: You can have large unilamellar vesicles, multilamellar vesicles, and so-called giant vesicles.

The traditional way to prepare liposomes is by drying down lipids into a thin film and then adding an aqueous solution. You shake it and you get an opalescent liquid or a cloudy liquid. That usually is multilamellar vesicles. If you then sonicate that, you then get usually small unilamellar vesicles, and that is an opalescent liquid. Large unilamellar vesicles and giant vesicles require different techniques for preparation.

The other aspect of classification is the MPS uptake, the reticuloendothelial system. You can have targeting and avoidance and also coating. We heard the example of antibody coating.

I want to emphasize that the definition of an active moiety is the drug responsible for the therapeutic effect. The lipids, in the case of liposome drug products, the agency has determined are functional excipients. So far, all approved liposome drug products use drugs that are already approved in other dosage forms. At the moment, we don't have any new molecular entities that are coming in fresh as liposomes.

So, the questions that we want to address are what information we need to demonstrate that a drug product is the same when there are manufacturing changes and the manufacturing procedure may determine the clinical behavior even if the components are the same and the finished product meets the same specifications. That is an issue we need to address. We don't have definitive information on that.

LDPs are unique. I can't think of any other drug product where you have specific characterization of the drug product. One of the things we are addressing is the question of characterization because LDPs are a new kind of dosage form. Characterization, similar to what we do for a drug substance, may include tests that are not necessarily part of the specifications, for instance, lamellarity and particle size or charge. From what you've seen with some of these, the particle size is in fact part of the specs. And some of the characteristics can affect clinical performance.

So, how can the changes in manufacturing be assessed?

Pre-approval changes, that is, changes for the same manufacturer during development, for instance, during scale-up or site transfer during development. We need a way to assess these changes.

And post-approval changes. If the product is approved and the market increases, they need to scale up even more, get new equipment, how do we assess those changes?

The manufacturer has extensive experience with the process and the process controls and the specifications and the characteristics. If you have a new manufacturer who is looking at the same product, same lipids, same drug, how do we assess these changes? Because remember that that new manufacturer does not have the same experience as the original manufacturer. They may actually use a completely different process.

So, some of the factors that can affect the sameness are the raw material, the manufacturing process and controls, and the storage and reconstitution.

There was a question Dr. Lee asked about the lipids. If you have lipids from natural sources, such as egg lecithin, you need to set your specifications to control the degree of the lipid composition and degree of unsaturation, which can vary from supplier to supplier. If you are using strictly synthetic lipids, that's much easier to control, but those have a tendency to be quite expensive and the question is whether you can get the liposome with the properties that you want from commercially available lipids.

Factors that can affect sameness in the manufacturing process are how you form the liposome, the removal of residual organic solvents, the removal of free drug, encapsulation control, control of liposome size and distribution, and of course, scale-up and economic feasibility during scale-up and possibly site transfer.

Storage and reconstitution. The liposomes can be stored either frozen or as a freeze-dried powder. Usually people don't store liquid liposome preparations for any length of time.

Now, depending on how you reconstitute your liposomes, that could affect the particle size and size distribution. If you remember, the directions for reconstituting Myocet were very careful in specifying a temperature range, shaking, particular buffers. So, those are questions that need to be addressed when you develop a liposome drug product.

I discussed earlier the degree of fluidity or stiffness of the membrane. That can affect the release of the drug, the size of the liposome, and the physicochemical properties.

The other questions that we have to address are how changes in particle size can affect targeting, uptake, and the clinical safety and efficacy. The in vivo stability of the whole liposome is important for targeted liposomes because they need to remain stable in the plasma until they get to their target.

Here I have a list of the approved liposome drug products. We recently approved a drug called Visudyne which is a drug for treatment of macular degeneration which is very carefully labeled that it is not a liposome, though it does contain a substantial quantity of lipids.

With that, I'd conclude my talk and thank you for your attention.

DR. BYRN: Questions for Dr. Shaw?

DR. GAWRISCH: Liposome preparations are notorious for not being always 100 percent reproducible. From the preparation procedures that you described, I would imagine that even for the approved products, you don't get always 100 percent the same. I'm wondering how much variability is permitted before you lose efficiency. Has that been investigated?

DR. SHAW: Those are questions that have come up during the review of the drugs, and I can tell you I didn't review these drugs. But they are questions that have been addressed. That is a matter for intense negotiations as to how to set the specs and how much process control there is.

As I indicated about characterization and specification, one of the questions that we often have with complex drug products, not your normal tablets, et cetera, is how much information do we need to set a specification because a specification has to be met for every lot. Characterization does not necessarily have to be met for every lot. Eventually if a drug product becomes a compendial product, the specs for the drug product have to be met by anyone who wants to copy it.

One of the issues that we have to be concerned with is are the specs sufficient to show sameness from batch to batch and from manufacturer to manufacturer or for changes within the same manufacturer. I don't know offhand of any cases in which a product met specs and then failed in its clinical efficacy. Information is hard to capture. It would have to be a fairly catastrophic failure.

DR. GAWRISCH: I guess it would be much easier to answer such questions if it would be known which properties of these particles are critical for the uptake and if you would know the details of the uptake process.

DR. SHAW: Right. That actually is part of the review process. Unfortunately for you and the world, we can't disclose that information. That has to be disclosed by the firm. Dr. Martin has presented some of the properties which are very important for the targeting and uptake of their products and what's in the literature, for instance, for Myocet. But in terms of setting specifications, that is an issue that has to be handled on an application-by-application basis.

I can tell you that going from what we used to do in the lab making liposomes and what has to be done industrial is a huge gap. That's part of the reason why it's taken so long for many of these to come to market.

DR. BYRN: Dr. Chiu?

DR. CHIU: I would like to add to the review process about approval of liposomes. It's like many other biological products. We pay a lot of attention to the process itself, also to the in-process controls. So, there are specifications set for the in-process controls.

Also we have multiple clinical batches used in clinical trials. We look at the variability of those batches and those things placed into the setting of specifications for the in-process control, as well as the final product release test and the stability test. So, therefore, we take into account multiple factors to make sure the batches will be consistent over time.

DR. BYRN: Thank you very much.

Our next speaker is Dr. Kofi Kumi, who will cover biopharmaceutics issues.

DR. KUMI: I guess it's still morning. Good morning.

I'll be trying to present our current thinking on the biopharmaceutics issues surrounding evaluation of the bioavailability/bioequivalence for liposome drug products.

First, I want to give you a background in terms of how the regulation defines bioavailability and bioequivalence. "Bioavailability means the rate and extent to which the active ingredient or active moiety is absorbed from a drug product and becomes available at the site of action."

For bioequivalence, the regulation defines it as "the absence of a significant difference in the rate and extent to which the active ingredient or active moiety in pharmaceutical equivalents or pharmaceutical alternatives becomes available at the site of drug action when administered at the same molar dose under similar conditions in an appropriately designed study."

So, when you are considering liposomes, the current thinking is that the key factors we need to consider is how the active drug is released from the liposome drug product and how that becomes available at the site of action.

The regulations give us some idea about how you go about determining or evaluating bioavailability and bioequivalence. In descending order of accuracy, sensitivity, and reproducibility, the first one is the blood/plasma/serum drug concentration measures, and this is what I will be discussing whether we can use this for the liposome products. Then you have urinary excretion. You have in vivo pharmacological effect. There's also here well-controlled clinical trials, in vitro test, and there's this catch-all phrase of anything that is deemed appropriate by the FDA.


DR. KUMI: So, our current thinking is, for you to be able to determine the bioavailability and bioequivalence of a liposome drug product, you first need a sensitive and specific assay. By sensitive and specific assay, I mean here an assay that can differentiate in vivo the encapsulated from the unencapsulated drug product.

We know that the release of a drug from a drug product affects the overall pharmacokinetics because if you release immediately, the pharmacokinetics is going to resemble that of the free drug, or if it's still encapsulated while it's still in the blood stream, it's going to reflect the pharmacokinetics of the carrier.

Also, another key question here is that you need to demonstrate the in vivo stability. Our current thinking is that you need to do a pilot study. This will be a single-dose study where you will separate and measure both the encapsulated and unencapsulated drug. If the drug remains in the circulation substantially in the encapsulated form and the ratio of unencapsulated to encapsulated remains constant, then we think you may be able to consider the liposome drug product as being stable in vivo.

As has been discussed earlier by the earlier two speakers, one way of classifying liposomes is based on the MPS system. Again, the MPS is synonymous with the RES, or the reticuloendothelial system. You have those that are designed to be taken up by the MPS, those that are designed also to avoid the MPS, and as Dr. Martin mentioned, those that kind of fall in between.

I think Dr. Martin has mentioned some of these already, but as I stated, some are designed specifically to be taken up by the MPS system. They usually have relative short duration in the systemic circulation. They are taken up some in minutes, some over a very short period of time, in comparison to those that are designed to avoid the MPS which circulate for a longer period of time.

Generally, when they are taken up by the MPS, the free or the unencapsulated drug is released back into the systemic circulation. The PK of this type of formulation is dose-dependent. We believe there is some saturation when they are being taken up at certain levels. PK, as has been mentioned several times, is affected by particle size, charge, and other physicochemical properties.

Again, the other general class we would like to consider is those that are designed to avoid the MPS. As I stated earlier, they remain in the circulation for an extended period of time compared to those that are taken up. They are preferentially taken up, or you could consider them targeted to specific sites other than the MPS. So, what Dr. Martin was talking about, the families of these lipids, we may consider them those that are taken up and those that aren't taken.

The PK here is dose-independent usually. And they also are affected by the composition of the liposome. Charge may affect some in some cases, and other physicochemical characteristics may also affect this.

So, what are the key issues and the questions we would like to discuss? Can we do business as usual? We usually do use plasma concentrations to determine bioequivalence/bioavailability. Can we do that with the liposome drug products? So, is blood/plasma/serum concentration of drug adequate to determine the BA and BE in view of the fact that the liposome drug products may or may not be stable in vivo, the residence time of the liposome drug product in the blood or serum or plasma may differ, and different liposome drug products may be designed targeted to separate sites or different sites, for example, to the MPS or to tumor cells?

I'm going to try and go through a flow chart, but first we are making a very big assumption over here. The assumption that is being made over here, you are considering a family of liposome drug products, and that also you can somehow determine the sameness of these drug products which results in functional similarity. We do recognize that you may not be able to do that. These are our current thinking and they're kind of what we are suggesting now to get your opinion.

First, let's take the drugs that are designed to be taken up by the MPS. The first key question you have to answer over here is that is this drug stable in vivo and prior to be taken up. We want to know that is it released or dose-dumped before it's taken up or it's taken up while the drug is still encapsulated in the liposome formulation. If your answer to that question is yes, then could you use the total drug concentration to determine the bioavailability or bioequivalence of this drug product? If the answer is no, is doing clinical trials using PD measures, pharmacodynamic measures, such as biomarkers, or clinical endpoints, safety and efficacy, the only method in which you can determine the equivalence of these drug products or not?

The other class that we will consider -- and this is a little more complicated -- is the ones that are designed to avoid the MPS system. Again here you have your liposome drug product. You specifically have designed it to avoid the MPS system. A key question here is that can you use plasma concentration in determining bioequivalence and bioavailability. As I said, if we find out that, yes, you can do that, then I think we still need to know why is it still circulating in the blood, is it stable, or is the drug going to be leaked out before it gets localized at the site of action or it gets through the leaky vasculature and gets localized at the site of action, or is the drug going to be released and the empty liposome is what is going to get over there?

Then if you answer that it is stable, the drug remains in the liposome and it gets localized at the site of action, then can maybe you can use total drug concentration. If no, then can you measure separately the encapsulated and unencapsulated drug and use that to determine bioavailability and bioequivalence? If we cannot do that, again then can we use clinical trials with PD measures, such as biomarkers, or safety and efficacy to determine equivalence of these products?

This is just to put both slides together and also to point out a few things. Again, I have liposomes that avoid the system. Again, the first question you ask is can we use that, and then also we want to know whether the drug is stable and gets to the site of action, is the drug still in the liposomes or it gets released before.

Then you have the MPS uptake. The key question there is, is the drug stable in the liposomes before it is taken up? Is it still in the liposome system or is it released?

We do recognize, again as I mentioned earlier, there's an intermediate. Our current thinking is that those will be handled on a case-by-case basis depending on how you design your liposomes. So, the mode and site of action and other factors such as the release characteristics and so on.

I would like to recognize and thank a lot of people who have contributed, especially Dr. Mei-Ling Chen, Barbara Davit, and Dr. Arthur Shaw, and other members of the liposome working group. I couldn't list everybody's name over here. They work closely with me. Everybody here with their questions, their comments, their suggestions, which came in very helpful when I was preparing for this, thank you very much.

DR. BYRN: Thank you very much.

Are there questions for Dr. Kumi?

DR. JUSKO: A critical factor in this categorization is MPS uptake, and I wanted to know whether that is something that can be measured unequivocally or is it obtained by inference basically? Long persisting liposomes have avoidance and the other way around.

DR. KUMI: If I may repeat your question. You want to know whether you can measure the encapsulated and unencapsulated unequivocally. Is that what you were asking?

DR. JUSKO: Yes. You need to categorize by MPS avoidance. How do you know that exists unequivocally? Is it something obtained by inference by the plasma residence time or is it directly biologically measured?

DR. KUMI: I'm not sure how to answer that.

DR. MARTIN: I can give you our experience. It is both. Clearly you do animal studies always before you get into people, and there you can easily see which organs the liposomes are distributing to.

But a cautionary note, because we've seen examples of very short-circulating liposomes and then been surprised to find out they're not in the liver where you'd expect them to be, but could be in the lung, for example, the reason being that they've aggregated or something and they've been caught in the capillary bed of the lung. So, it is an empirical, I think, exercise to determine what is the principal organ distribution in terms of this classification scheme? You can't determine it a priori I'm afraid.

DR. LEE: Can you do that in vitro?

DR. MARTIN: It has been done in cell culture, but it's not very informative. You can take peritoneal macrophages, for example, and look at their uptake rate, and it's true that if they're coated with polymer, it's somewhat slower, but it's not very informative.

DR. LEE: Why is that?

DR. MARTIN: It's a misnomer to think of the mononuclear phagocyte system as one kind of cell. These are highly differentiated cells. They range from very large to very small, and they have different functions. So, some take up particles of a certain size. Some take up particles opsonized with certain proteins and so on. So, to isolate peritoneal macrophages, which are the easiest ones to isolate, and use them as a surrogate for the entire mononuclear phagocyte system is not appropriate. You couldn't gather enough of a variety of these cells to determine the uptake rates, I don't think. I don't think it would be very useful.

DR. BYRN: Other questions? Marvin?

DR. MEYER: When you say stable in vivo, do you have a time frame there? Is that until the next dose, a couple of hours?

DR. KUMI: Usually if you look at the MPS uptake, you have a distributive phase. If within that phase for that, I would say usually within an hour you should know whether the drug has been taken up, whether for that phase, what you have is stable or what is still encapsulated. So, it will differ from the class of liposome that you have. If you have the low-circulating liposome, obviously that you have to go for a longer period, three days. So, you won't know for three days whether you have that stable or that's still encapsulated.

DR. MEYER: I'm not sure, under the arm where it's MPS uptake and it's not stable -- so presumably there's some circulating unencapsulated, as well as encapsulated -- why you couldn't use that rather than go to some less precise measure such as PD or clinical.

DR. KUMI: That was made from a safety point of view because what I want to make sure is we are not just releasing all the drug and dose-dumping the drug.

DR. MEYER: But that should show up as very high unencapsulated concentrations. Therefore, two products wouldn't be equivalent if one had high unencapsulated and one had lower.

DR. KUMI: I understand what you are saying, but I guess our point of view is that we don't want it to be a safety concern. Like amphotericin, they give it in very low doses in the free form. In the encapsulated, you give 5 milligram per kilogram versus the conventional which is .6 milligram per kilogram to 1 milligram per kilogram. We don't want a case where you have everything released at the same time and then cause a safety concern.

DR. MEYER: I understand that.

DR. KUMI: That is the approach we're taking to be on the conservative side.

DR. MEYER: One quick question. Have you seen the data that Dr. Martin referred to from UCSF?

DR. KUMI: No, I haven't.

DR. MEYER: That was most troubling.

DR. KUMI: Yes. I haven't seen that, about the differential distribution within the tissues? Yes.

Again, I'll reemphasize that we are making a big assumption that you can demonstrate sameness and that will translate into functional similarity. That's a big assumption. We do recognize that you may not be able to do that.

DR. BYRN: Thank you very much. Oh, Frank?

DR. MARTIN: One follow-on question to the one that was just asked. In terms of the stable in vivo prior to uptake, setting a parameter there, for example -- if you can measure, which you can, the encapsulated versus the unencapsulated, where would you make the break there? 20 percent, 50 percent?

DR. KUMI: That's still under discussion. We've wrestled with that. We haven't come to a conclusion. That's still under discussion.

DR. BYRN: Before we go into the discussion, we have a brief correction that we want to get on the record.

MS. PENDERGAST: My name is Mary Pendergast. I'm an executive vice president of Elan Corporation, and thank you for indulging me.

Our drug, Myocet, was discussed by Dr. Martin, and we would like to state our concern. We talked to the committee staff before this meeting, and we were told that Dr. Martin would only speak on critical formulation parameters. We were not told that he would be speaking about our drug. The slides that were posted about his speech on the website did not mention our drug, so we had no idea that our drug was going to be discussed today. And we did not come prepared to discuss the drug in the kind of detail that Dr. Martin presented. Indeed, we were told we would have five minutes.

But he made many statements about the drug, and while I'm sure he did his best to be accurate, there are several substantive concerns we have with the statements he made.

So, I would ask the committee -- since our drug is under review by the FDA and we would not be able to discuss this kind of information in public session, we would ask the committee and ask the FDA to consider letting us present to the committee in closed session before the committee makes recommendations to the FDA because we would believe that those recommendations have the possibility of being different, if you could hear information about our drug that either was not presented or was presented somewhat inaccurately today.

Thank you.

DR. BYRN: I can assure you that this committee, as we said at the introduction, is simply dealing with general scientific issues and that we will not delve into the specifics of any drug product that was mentioned today in our deliberations. So, I don't think you have a problem. I suppose there is a mechanism by which you can make appeals to the FDA to make a presentation, but as far as I'm concerned, the committee is only going to look at this in a broad, general scientific way, and so I don't think that that material would be used in any manner that would cause problems to your company.

But we can proceed and then you can contact the agency later, if you believe that's not correct.

MS. PENDERGAST: Thank you very much. We do believe that there are some broad, general scientific principles that you may wish to consider with respect to whether, for example, the MPS families are as distinct as was presented. Thank you.

DR. BYRN: Okay. Thank you very much. We'll take that all into consideration.

I think we can go ahead with the discussion and break it into two parts. First, topic one is pharmaceutical equivalence issues. I would just point out most of these questions are in Dr. Shaw's presentation, and let's spend a few moments on that. We have a total of about 30 minutes. So, let's spend some time, maybe say 10 minutes, on pharmaceutical equivalence and then 20 minutes we can continue on the BA/BE.

So, issues on pharmaceutical equivalence that people would like to raise? I can just direct you to some of the issues in Dr. Shaw's presentation, slide 8. An issue for concern in the pharmaceutical equivalence is demonstration that a liposome drug product is the same when there are manufacturing changes. Comments by the committee?

DR. SHARGEL: Actually that's the one issue that I had flagged as well in terms of SUPAC changes. Does the agency have some general guidelines to approach the manufacturers who may be scaling up or changing the formulation or changing the site?

DR. CHIU: As Mei-Ling has told you, our guidance on liposome does not include post-approval changes. We wrestled with this issue for a long, long time. Then we decided we just do not have enough information and the knowledge to really address this special type of products. So, our guidance will only talk about an original submission.

The reason we are here is because we are seeking your input to help us to address the post-approval changes. The current SUPAC guidance, that is product-specific, will not really address this type of product. Therefore, to answer you question, the agency has not had any position how to address technical positions on how to address this type of change.

DR. BYRN: Just to go on, I have the impression that the process that's used is very critical to the performance of the product, and I also assume that some of that is proprietary. Intellectual property is involved in that part of it. So, I assume that this is a very difficult matter. I don't know whether Frank could confirm that or not.

DR. MARTIN: I can confirm that it's evolved a long way from mixing it up in the flask and that sort of thing.

I think the innovator has a tremendous advantage because access to intermediates and to all of the process development information related to changes that were made and extremes that were tested during the development.

But it's a sequential process, multiple steps and so on, and each step I think can be isolated. When changes are made, assay methodology can be applied that's even more advanced than as was suggested by the speaker from Gilead where higher order assays are used to assess changes in particular steps. So, I think it is a complicated process, but I think when you break it down into its components, if it's a one-step change, a new filter, new this, new that, I think that's just routine sort of pharmaceutical science. But if it's a whole new process, then it's a different story.

DR. CHIU: In the meantime, I need to add we do have a CBER/CDER guidance called comparability. That guidance addresses how you deal with changes, small or big, with respect to complex biologic systems. So, there's a hierarchy system. Simple change, you may only need to do simple analytical testing. A very complex change, you may need to repeat your clinical studies. So, there's a grading. In the meantime, we're using the principal layout in this comparability guidance to address post-approval changes.

DR. BYRN: Just to go on, on 13 and 14 there's a discussion, especially of these tests that we were talking about.

A question I have, even in a certain, say, wet granulation manufacturing process, it may not be well known what the critical process steps are. How well known are the critical process steps in this area? Are they pretty well known? Or even in cases where we're using sophisticated tests, still we're not sure they're measuring a critical parameter or not. What's the state of the art? Maybe our guests could comment on that too.

DR. MARTIN: I think it's pretty well understood. By the time you've gone through the NDA submission, you understand your process pretty well. I think the issues arise when you change manufacturing site, for example. So, now everything is changed. Or some major piece of equipment or something, or some major scale-up.

DR. BYRN: What's a major scale-up in this field?

DR. MARTIN: You can make clinical supplies at the 10 liter scale and commercial product, at least for our product, is made at something like 400 liters or something. So, that's 40-fold. There was something in between. There was a 100-liter I think in between. So, that's the scale at least for now.

DR. LITMAN: It is very critical that there be good control on the original products because if you're using natural products, how those natural products are prepared, how the cuts come off the purification columns all are going to influence the properties of the end liposome. If those have any variation in them, then you can expect the end product to have a variation also.

DR. BYRN: So, raw material, just as in any other raw material, is absolutely critical.

DR. LITMAN: Well, I think the point is that egg PC is not a defined molecular species, and where you get it and how they do the preparation is going to determine what the acyl chain composition is and the subsequent properties in the liposome.

DR. BYRN: Could I ask a question? Neither of the guests were here. I don't want to get way off in the details of this. We were talking about an amphotericin product and I was asking specifically what the dynamics in that product were. I was told that there wasn't a lot of mobility in those systems. Is the coating what's preventing the mobility?

DR. LITMAN: No. It's the lipid composition.

DR. BYRN: The Tc of the lipid.

DR. LITMAN: Right, the Tc, the amount of cholesterol that's in the system. Whenever you put cholesterol in, you get a very immobile or rigid system.

DR. BYRN: Now, if you do NMR mobility measurements, you still see mobility, though, on an NMR time scale? A lot of mobility or?

DR. GAWRISCH: In the coating, you have some impact on the hydrocarbon chains. You see a lot of mobility. These hydrocarbon chains in liquid, crystalline formulations -- these are the ones where the chains are shorter. They behave almost liquid-like, so maintaining this bilayer structure, but otherwise these molecules have tremendous degrees of freedom, and that is important for the uptake.

But you could have situations where this is not a critical property from what I understand. When you go, for example, to something that has 18 carbons in the hydrocarbon chain, you get really crystalline packed hydrocarbon chains and that may be beneficial for a certain application. That is a parameter that is pretty straightforward to measure.

DR. BYRN: Now, is mobility in these products a critical parameter if you make a change, do we know?

DR. GAWRISCH: I presume.

DR. BYRN: Or maybe we can't talk about it completely, but in a general way.

DR. MARTIN: It's a delicate balancing act because in one instance you want to keep the drug in your carrier to take it to tissues, for example, in one situation. But you don't want the drug to stay in there forever. So, you don't want covalent polymerized lipids, for example. You want the lipids to slowly hydrolyze, which would be their natural fate. So, it's this delicate balance between getting the drug there and then getting it to be bioavailable at some point later or provoking it to become bioavailable by putting a ligand on it, for example. It speeds things up.

DR. GAWRISCH: I would just like to give one example that demonstrates how important quality control is even for an ongoing process. When we study liposomes in research, we became aware that just at different times of the year our results differ. The simple truth was that the chicken got different food. That is something that's easy to control if you really know the background, know what has to be stable, and you just feed them the same stuff all the time. You keep them so that they don't sense the changes of the time of year, and everything becomes reproducible.

For natural products, that could be a very, very critical issue as you may get very different results. This should be very well documented, and there are well-known procedures to determine, for example, fatty acid content, content of other trace amounts of substances that could be critical for this. I think all these techniques -- this is good news -- have been worked out for the past 25 years. All that is very well known. And it should be part of the characterization of the product, and there should be ongoing process control to make sure that these critical parameters are not changing.

DR. BYRN: Yes, Bill.

DR. BARR: Just a comment and then maybe a question. First of all, I think one of the questions that was asked is do we ever want to use clinical endpoints or PD endpoints. I think probably the answer is almost never. I can't imagine using a clinical endpoint for the kinds of diseases that these products are used for. So, that I think we can rule out of the picture, except in maybe very extreme cases, as being any kind of a practical approach. PD is almost as difficult.

So, it really means that we almost have to start from the other side, it seems to me, in the way that we do some other complex products. So, we start using the physicochemical characterizations to the degree of identifying the critical manufacturing variables, knowing then how they may impact upon some kind of in vitro release characteristics -- and that's what I wanted to come back and ask about -- and then how that relates to what we know about the PK and the bioequivalence.

We haven't discussed very much about how well we can characterize the release in vitro and how that relates to the critical manufacturing variables. Could you amplify on that?

DR. MARTIN: Because of the precedence before it with other products, we were asked to develop an in vitro release for our product, which turned out to be very difficult to do because the drug is trapped so well inside the liposome that if you incubate it in any biological fluid, including plasma, the drug does not come out. It just stays in there. So, we developed a provoked release assay. It's held in there by an ionic gradient. It's a gradient of ammonium ions. If we break that gradient, the drug comes out by adding 200 millimolar ammonium sulfate to the outside of the liposome. Well, of course, that would not happen in vivo. So, it's a provoked assay that gives some information on the stability of the liposome probably, but I don't think it has any in vivo correlation.

So, in terms of developing in vivo/in vitro correlations that has been done for other dosage forms, we are in our infancy. I think there's room to hope there, but depending on the formulation, it's a difficult challenge because these things are so stable in the biological system.

DR. BARR: Of course, 10 years ago we said the same thing of most other dosage forms too. There weren't very many correlations until SUPAC came around and it was worthwhile doing it. It seems to me that probably is a worthwhile approach to some of these problems.

DR. BYRN: We probably should go on to BA/BE, but are there any other questions or comments by the committee on PE or any questions from the agency?

(No response.)

DR. BYRN: Let's go onto BA/BE questions. We were delving into that. One of the key questions is on page 5 of Dr. Kumi's discussion under "The Key Issues," and maybe we could begin by discussing the whole issue. Maybe Dr. Barr could just repeat again his question or ask a follow-on.

This question relates to blood, plasma, and serum levels. I'm in Dr. Kumi's presentation on page 5. The question is: Is blood/plasma/serum concentration of drug adequate to determine BA and BE? I don't know but I think it's pretty clear that the answer to that is no.

So, we need to delve into this question I guess a little bit more, this question of could you have surrogate tests, challenge tests. How do we proceed on this issue?

DR. SHARGEL: The question in my mind is how do we know that for sure that it's no if you've got some adequate in vivo tests such as release tests that Dr. Barr brought up. So, if you had adequate in vitro releasing that would mimic what possibly would be happening in vivo, then would the answer to a blood level study be adequate?

DR. BYRN: So, that's the question. I guess we can go on. If you had some kind of test, would that be adequate to show equivalence in the face of that change?

DR. BARR: It seems to me that when we say the plasma assay -- and I'm not sure because I don't know enough about the chemistry and how this is done, but it seems to me that the problem is that you have various degrees of aggregation. I wonder if you measure the plasma. The question is how well you can separate the absolutely free drug from the clearly encapsulated versus some aggregates in between. Maybe it's that middle part that we really need to focus in on. Is that a correct assessment?

DR. MARTIN: No. I think the actual separation of the liposome from unencapsulated drug that may have been released is fairly straightforward actually. Liposomes are like a little formed element. You can remove them by a small column, for example, a little prep column. The problem is with the drugs we're dealing with, for example, doxorubicin itself is cleared so rapidly that once it's out of the liposome, it's in the tissues within a few seconds. So, you have that uptake competing with trying to detect what is there.

Not only that, you're dealing in plasma in a fluid where you can get rid of the formed elements. The liposomes remain in the plasma so you can do a separation, but in tissues, getting back to the tissue problem, this is not possible because to quantitatively extract drugs like doxorubicin you require acidified solvents and things like that. So, you would break open any liposome. So, in a solid tissue, we have been unable to repeat the kind of simple separation method that we are able to perform in blood. That's where the challenge remains.

DR. BARR: This is more of a specific problem that would relate to that specific compound.

In terms of the tissue uptake, though, are there surrogate tissues that can be used like white blood cells, red blood cells, other tissues that also take up the drug rapidly, or is it only in --

DR. MARTIN: They would take up the free drug.

DR. BARR: The free drug, right.


DR. BARR: So, you could measure those as a measure of the free drug that's been released. So, there may be ways.

It would seem to me, again, that this has to be directed back to what we can do in plasma and how well we can separate those and maybe other surrogate tissues, those that are freely accessible like formed elements or things like that, or maybe even consider fat tissue or something. I don't know. And then getting back and trying to correlate that with the in vitro at some point in time to show that linkage in some way. Does that seem reasonable as an approach?

DR. MARTIN: It could be. I think the biggest problem in our case is that no matter what fluid we soak these liposomes in, the drug does not come out unless it's provoked to do so or unless we cook it or subject it to some totally irrelevant conditions.

DR. BARR: But presumably the body does that fairly efficiently, and all we need to do is to mimic that in some way, either by the enzymes or some process.

DR. MARTIN: But when you consider what's happening to the liposomes in different tissues, for example, in macrophages, sure, they're being digested by lipases. But in the case of tissues, such as skin or tumors, for that matter, the liposomes find themselves in the interstitial spaces, and in some tissues that's poorly regulated. In tumors, for example, the interstitial fluid is poorly regulated. The pH is not regulated well and so on. But in skin, interstitial fluid is an ultrafiltrate of blood, which is highly regulated. Again, you have differential treatment of these extravasated liposomes in different tissues. So, it complicates the matter even more.

DR. BARR: But again, using basic pharmacokinetic principles, we assume that there are some distribution equilibriums and what's measured in one site is reasonably predictive of what will happen in other sites. Otherwise, it appears that we're setting up such a system that it would be almost impossible to ever show bioequivalence in all tissues and therefore almost impossible to be able to approve new product changes as they may come along. So, it seems to me we have to pare that down into something that's reasonable and ignore perhaps all of the tissues but maybe try to focus on those that are most important for a specific drug.

DR. SHAW: I think we should keep in mind that demonstrating in vivo stability is not always going to be straightforward. It may be variable from product to product. So, we can't always assume that that will be easy to demonstrate.

DR. MEYER: Steve, I've always felt that the official CFR definition of bioavailability that speaks about availability at the site of action was kind of silly. I think it appears as though this may a case where that's essential and not silly at all.

DR. BYRN: Also, I think regarding this point about targeted liposomes, obviously we're going to have to think collectively about a way, just to talk about site, to assess, when you have targeted products, that they go to the target without doing clinical experiments. And that's a whole other range. Now, we not only have the liposome, but we have it targeted. So, we have to determine all the parameters of stability plus how well it goes to the target and reproducibly it goes to the target.

DR. GAWRISCH: I have a question. Would it be possible to design an animal model for every particular drug that would help to figure out if the formulation meets quality standards? At the moment, we certainly could ask to characterize liposomes by all means. That's something which I would not consider very difficult, and the critical parameters could be established and then could be verified. But then there is still this little part that is unknown and you would like to have some tests on a real-life model. Perhaps there would be certain animal tumors that could be, for example, used as an example to demonstrate efficiency of the drug.

DR. CHIU: Actually that topic was discussed internally. We were thinking before IV/IVC, maybe we have something in the middle, develop an animal model to show targeting, to show the site of absorption and the rate of absorption in lieu of clinical trials.

DR. LEE: Steve, I just want to say one comment and I'm not sure whether it is appropriate. I think this might be an ideal case for noninvasive monitoring, the use of spectroscopic methods. The more I listen to it, the less uncomfortable I am of relying on the blood data to make judgments.

DR. BYRN: Interesting. You could do that with doxorubicin also.

Other comments on BA/BE? Yes, Mei-Ling.

DR. CHEN: Just for my clarification, just now we discussed about the adequacy of using blood/plasma/serum concentrations for determination of bioavailability and bioequivalence. According to what I heard, the panel thinks that it will be highly difficult to use this method for demonstration of BA or BE. Am I correct?

DR. BYRN: That's what I heard. Are there any panelists that disagree with that characterization? Highly difficult. I think that's a good choice of term.

DR. CHEN: And I want to ask whether you have different degrees of comfort if we could classify liposomes according to the MPS uptake or avoidance. Would you give me different answers?

DR. BYRN: So, the question is now if we could classify them into the MPS uptake categories, could we then maybe be more confident in normal BA/BE measurements under some circumstances.

DR. CHEN: Because I realize that Dr. Martin's example was focused on Doxil, and he's saying that it's closer to the category of MPS avoidance. The drug gets stuck in the liposome, so you don't see the drug released in the bloodstream.

DR. MEYER: I think I would be more comfortable in looking at that.

DR. BYRN: Yes. I think from what I heard this classification is a good approach. At first we classify and then we decide the rigor or the complication for the BA/BE. I think that's what the committee felt.

Yes, Bill?

DR. JUSKO: I would say, however, though you should ask for criteria by which they establish that classification, what evidence does the company have for setting their liposome as being one type or another.

DR. CHEN: Yes.

DR. JUSKO: It sounds like it's a lot of after-the-fact --

DR. CHEN: Empirical.


DR. CHEN: I have another question. There is actually a slight difference between bioavailability and bioequivalence. So, I don't know whether we could really address the questions in a similar way.

DR. BYRN: Does anybody want to comment on their feeling as to whether these ideas apply both to BA and BE or just one of those?

DR. MEYER: Generally the BA is done for the NDA, so you have this wealth of clinical data as well as the kinetics of whatever you can measure adequately. Bioequivalence is much more difficult than bioavailability. It's kind of whatever you get, you get, you report it. Whereas, bioequivalence, you have to match something.

DR. CHEN: Right. Well, my question was actually in the context of this core issue that we have, whether we could use blood/plasma/serum concentrations for demonstration of, say, bioavailability.

DR. MEYER: That looks to me the best available currently. Certainly if someone comes up with a better way, that ought to be applied, but I wouldn't not do blood just because there are some doubts about its application to the clinic because you have the clinic also.

DR. CHEN: Yes, I would agree. In the NDA areas, we do have a body of information, including clinical trials. So, a bioavailability demonstration is not so much a concern there.

Thank you.

DR. BARR: Just a further comment. While BA and BE are somewhat different in the fact that whoever is doing the test may have more information, the end result is the same. Ultimately the innovator at some point is going to have to compare probably some lot or some new formulation with their original formulation and will have to find a way to justify that that's an acceptable way to do it. So, hopefully that may have to happen before we have to decide the other because they have most information and can probably help solve the BE problem as well.

DR. BYRN: Any other comments? Is the agency happy with our input? Do you have any other questions from the agency?

DR. CHIU: I just want to answer yes. We're very pleased with input. It helps a lot.

DR. BYRN: I think we can adjourn the meeting. Thanks to all the speakers today for keeping on time, and thanks to the committee members. And I wish everybody a safe trip.

(Whereupon, at 12:46 p.m., the committee was adjourned.)