FDA Home Page | CBER A-Z Index | CBER Search | Contact CBER | CBER Home Page

DEPARTMENT OF HEALTH AND HUMAN SERVICES FOOD AND DRUG ADMINISTRATION CENTER FOR BIOLOGICS EVALUATION AND RESEARCH AND PLASMA PROTEIN THERAPEUTICS ASSOCIATION COMPARABILITY STUDIES FOR HUMAN PLASMA-DERIVED THERAPEUTICS Thursday, May 30, 2002 8:00 a.m. Doubletree Hotel Rockville, Maryland C O N T E N T S Opening Remarks Opening Remarks, Christopher Healey, J.D., PPTA Agenda Overview, Mark Weinstein, Ph.D., FDA Industry Perspective: Comparability,       Michael Gross, Ph.D., Aventis Behring CBER Perspective: Comparability,       Christopher Joneckis, Ph.D., FDA Comparability Studies for Human Plasma-Derived Therapeutics Product Characterization FDA Perspective, Andrew Chang, Ph.D., FDA Industry Perspective,       Ghiorghis Ghenbot, Ph.D., Aventis Q&A Preclinical Studies Industry Perspective,       Michael Saunders, M.D., Baxter Relevance of Animal Modeling in Predicting       Immunogenicity, Basil Golding, M.D., FDA Comparability Studies for Human Therapeutics, Preclinical and Pharmacokinetic Aspects,       Martin D. Green, Ph.D., FDA FDA Perspective on the Design of Clinical Studies to Evaluate Comparability,       Charles Maplethorpe, M.D., FDA Q&A Reporting Manufacturing Changes Changes to be Effected/Changes to be Effected in 30 Days and Pre-Approval Supplements FDA Perspective,       Timothy Lee, FDA Industry Perspective,       Frank Rauschen, Ph.D., Bayer Q&A Comparability Protocols FDA Perspective, John Finkbohner, Ph.D., FDA Industry Perspective, Paul Gil, Bayer Q&A Case Studies Case Study Presentation from Industry Angela Blackshere, Baxter BioScience Ghiorghis Ghenbot, Ph.D., Aventis Behring Case Study Presentation from FDA,       Basil Golding, M.D., FDA Q&A P R O C E E D I N G S Opening Remarks DR. HEALEY: Good morning, ladies and gentlemen. I would like to start today's program. My name is Chris Healey, and I am the executive director for PPTA North America. On behalf of PPTA, I want to thank you all very much for coming. It is wonderful to see so many people here. I think some of you know that the program is so popular that we actually had to stop registration. So, consider yourselves fortunate, I guess, you were able to sneak in. But thanks very much. Just a couple of things, I did want to, on behalf of PPTA, acknowledge how privileged we feel to have the opportunity to co-sponsor the program with the Food and Drug Administration. It has been a great process. I can tell you it has been one that has been a long time in the making. For over a year we have been working with the folks at FDA and within industry, putting together the program and the agenda to make sure that it is a really meaningful and informative series of presentations over the next day and a half. In particular, I want to acknowledge Andrew Chang from FDA, who has spent a lot of hours and a lot of his personal time and resources helping design the program, and put it together, and making sure that it is appropriate and beneficial for all of us. Also, PPTA Craig Mendelsohn, who is our director of regulatory affairs, who has worked really hard on this over the past year or so, making sure also that the industry folks kind of had their act together, that we were working well with FDA and making sure the agenda came together. So, particular thanks to the two of them. Also, in terms of administrative matters, Helms Briscoe is our meeting planner and I think they have done an outstanding job, and our AV resources, Capital AV have been a big help in making sure everything is set up. So, if you have complaints, feel free to go to them. If you have compliments, feel free to come to Craig, myself or Andrew. We will be welcoming those throughout the day. With respect to questions, we anticipate that there are going to be a lot of questions. This is obviously a very important topic and, given the level of interest that has been expressed, we hope to have a really dynamic program. We anticipate that we may be running tight in time with questions in particular, but we have index cards that are available so, please, throughout the day jot down any questions you have and we will have folks walking around the room, collecting those index cards. If we don't get to your particular question during the Q&A, there will be panel sessions where we can address additional questions. So, don't worry, we will get to all the questions that we possibly can. With that, thank you very much once again and I am going to turn it over to Dr. Weinstein. Agenda Overview DR. WEINSTEIN: Thank you, Chris. Well, it is a pleasure to be here and I anticipate this meeting will be very productive. The issue that we will be discussing primarily over the next day and a half will be how best to implement the provisions of our guidance concerning demonstration of comparability of human biological products, including therapeutic biotechnology-derived products. Henceforth we will refer to this as the guidance. I think that as we go along here, we realize that understanding the provisions of this guidance on part of the FDA and on industry's part will help us to better formulate submissions; will understand each other's concerns and expectations; and eventually it should lead to reaching our overall goal, to provide consumers with safe, pure and potent products in the most expeditious manner. It should also be remembered that a guidance is just that. A guidance contains recommendations; it does not contain requirements. So, I will emphasize over and over again that there is no substitute for good judgment, and this good judgment is based on good data and on experience and on knowledge. This will be, again, a continuing theme throughout this meeting. CBER has a number of concerns that are rather routine. We learn at CBER school that there are a number of issues here that are rather common and that we can anticipate might affect the plasma derivatives if there is a manufacturing change. I should point out that this document was issued in 1996 with the intention of reducing the need for clinical trials if it could be demonstrated by other means that product, before a manufacturing change, are comparable to the product produced after the change. Our problem is in defining exactly what "demonstrate product" and "comparability" actually means. These will be topics that we will be discussing throughout this entire meeting. I should mention also that a copy of my slides I guess is being handed out now so you might not have to write everything down immediately. Anyway, you know, one can consider this as sort of a rogues' gallery of issues that we have concern about. This includes the production of neoantigenicity, new aggregation, polymerization and degradation after a manufacturing change, oxidation, deamination, altered glycosylation. We will be discussing to some extent during the course of this meeting introduction of vasoactive substances, particularly the pre-kallikrein activator; change in reactivity towards substrates or receptors; introduction or removal of so-called impurities that affect the product safety and efficacy; a change in molecular species distribution, for example, a change in the distribution of immune globulins; and the introduction or proteolytic enzymes. The important point here is that the more information that a manufacturing can supply us about these particular issues up front in their submission, the more likely is it that we will be able to review the submission quickly and that we will be able to have reduction in the review cycles. I point these particular ones out because these are pretty much the hot topics. We will hear some other issues as we go along, but these are ones that are quite uppermost in our minds when we do a review. The questions that we are confronted with following a manufacturing change: what is the clinical significance of an observed change in an analytical result? Now, often it occurs that one might see a slight shift in a peak on a chromatography readout, or you might see a slight change in the distribution of product species. The questions that a reviewer has and that you should have in anticipating what the reviewer question is, is what is the significance of that analytical result? How are you to evaluate it? And, it is very important that you assess your data; that you give us up front your assessment of what that change might be rather than just giving us raw data where we pick out the anomaly and try to decide what it might mean. The other issue is how confident can you be of product comparability if you observe no change in analytical results. The question there is whether your methodology is sensitive enough, and how can you assure us that, in fact, there isn't something that has crept into the new product? We think, of course, that there needs to be an approved reciprocal communication about the expanding body of experience in paradigm situations that influence our regulatory decisions. What does that mean? The notion is that we have an increasing body of knowledge as we go along and try to manufacture. The PKA incident is one that, again, we will discuss in some detail. These are experiences that a particular company might have in the production of a product here, but it becomes the paradigm for our thinking about many other products. The point is that it is very important that we have an opportunity to share what these paradigm experiences are, and communicate them throughout the industry so that we will have a common reference point to go to, to be able to have better submissions; that we are all working essentially off the same page. So, what I presented in the previous slide was actually a number of issues that are our chief concern here. You will get a sense of our highest levels of concern here are, but again the idea of sharing is very important and, as one outcome of this meeting, I think we should work toward establishing some way of being able to share our experiences both at the FDA and from industry. Actually, this meeting is an example of that kind of sharing of information. Now, to quickly go through some of the issues that will be presented during the rest of the one and a half days, we will have a discussion from industry and from CBER representatives on perspectives in general and comparability. We will discuss the review guidance in much greater detail. We will be talking about the characteristics of plasma-derived therapeutics as opposed to specified biotech products, can we use the same kinds of instrumentation of analysis on plasma proteins as we use on biotech products? We will talk about approaches to establishing comparability and we will have a rather extensive review of latest analytical biotechniques that can detect changes in protein structure and product content. We will also discuss the preclinical testing for product characterization and compatibility. The issue here is what can animal studies tell us? What are the weaknesses and strengths? We will also have a discussion of potentially a new system of detecting neoantigenicity that in the future may be used to give us a sense of a potential of a product to have neoantigen formation during the product change. We will also talk about the design of clinical studies to demonstrate comparability. The issue there perhaps is should we always strive to establish comparability, or are there some instances where it would be actually a better strategy to consider a product as a new product. You may need less patients in the clinical trial under some circumstances than if you tried to establish that a product is comparable before and after a manufacturing change. We will also discuss the classification of submissions and/or reports, CBEs, CBE-30s or prior approval supplements. What should be the criteria for characterization? What are the advantages and disadvantages of a given classification? And, we will give examples of successful and unsuccessful applications of this regulation. Here is a little quote, "though success is more pleasing, failure is often no less instructive." This is Francis Bacon, in 1590. I actually got this quote from John Finlayson. Regarding case studies, we will be talking about successful use of a comparability protocol and a comparability study that reduced clinical trial requirements. We will also then have a very extensive discussion of two situations where we had the pre-kallikrein activator as an issue that affected product quality and, again, the notion here that small changes in manufacturing can have a large effect on product quality. Finally, on the second day we will be talking about comparing fractionation intermediates. We will be discussing the need for fractionation intermediates in the marketplaces, the issue of reduced supplies of plasma, and we will be discussing the criteria for comparing the intermediates. We will also review the draft guidance entitled, Cooperative Manufacturing Arrangements for Licensed Biologics, and have a case study of the acceptance of a fractionation intermediate. This is a very ambitious agenda. Chris already alluded to the idea that we do want participation. I think that part of the success of this meeting will be the idea that we can exchange ideas freely. Please, do fill out these cards here and we will try to discuss as many of your questions as possible during our session later in the afternoon. Thank you for your attention. The first speaker will be Michael Gross, who will give an industry perspective on comparability. Industry Perspective: Comparability DR. GROSS: Good morning, everyone and thank you for coming to this meeting that has taken us almost two years to put together. My name is Michael Gross, and I am responsible and proud to present the industry viewpoint on comparability as it relates to plasma derivatives. Change can occur throughout the product life cycle of a biological product to, amongst other things, improve product quality, enhance compliance with regulation, improve facilities and equipment, extend production capacity and improve efficiency. A plasma derivative change can occur in basic fractionation, purification, formulation, packaging, storage, testing. Biologics are said to be defined by their manufacturing process, so making change in the process means making change in the product. This raises issues about adverse effects of change, and it also raises issues about the relevance of previously established clinical data to the product made by a variant process. The comparability process is a tool with great potential for expanded application in the management of change in our industry. New chemical and biological bioprocess analytical methodologies provide tools to address difficulties encountered in the characterization of biologics, including plasma derivatives. Successful application of modern analytical methods encourages the management of change through comparability approaches. Over the next day and a half we will discuss challenges, difficulties, success, failure and the future of managing change in the product of plasma derivatives and plasma protein products. We hope to come away from this meeting with an improved understanding of how to best apply the comparability concept and what changes might need to be made in the future. I will begin with an attempt to define a few key terms, raise a few questions and make some suggestions for areas that need improvement. The comparability concept concerns make scientifically sound judgments based on data, based on data comparisons; manufacturing and manufacturing history of change, perhaps as much as forty years of it; clinical experience; and it compares two pharmaceutical entities, perhaps a drub substance, a drug product, a key intermediate, one derived from a variant manufacturing process that was used to produce the other, to determine if they are sufficiently the same or similar enough that they can be considered--and I use the next word with some trepidation--equivalent in their effects. In particular, the chemical, biological and clinical effects. Controlled change is good and the comparability concept was developed to provide a vehicle to accommodate the manufacturer's need to make changes and FDA's need to control change. Comparability was intended to provide post FDAMA regulatory relief, an approach to controlling change that does not place the regulatory barrier so high as to discourage it. The comparability protocol is codified in 21 CFR 601.12(e) and had to apply comparability as described in the guidance, 1996 CBER guidance entitled, FDA Guidance Concerning Demonstration of Comparability of Human Biological Products, Including Therapeutic Biotechnology-Derived Products. A comparability study is an experiment, a side by side comparison of properties of a drug substance, a drug product, a process intermediate made by an established process and a variant process. A comparability program is a collection of comparability studies that are well reasoned and well designed and are intended, in part, to eliminate the need for clinical data that provides evidence that a manufacturing change has not adversely impacted identity, purity, safety and potency of a biological product. It is not a testing hierarchy. Rather, it is a program of studies designed with an understanding of the performance of a validated manufacturing process and the plasma protein product produced. The design of a comparability program is driven by the nature of the change and its location in the process; the stage of development; the product characteristics; the potential for them to change and affect product purity; the type of product and its intended use; the physicochemical and biological properties of the product and their potential to produce related substances; the suitability and availability of analytical methods to asses the impact of change on these characteristics, and the relationship between product quality and biological activity and product safety and potency. Comparability comparisons or chemical, physical and biological data for drug substance, drug product or intermediates will usually include routine release test and stability data. This may be supplemented with in-process tests at the manufacturing step most likely to be impacted by the process change. They may also include non-routine tests, including methods at early stage used to characterize the consistency of production, and they also require more advanced tests, newly developed tests if established methods are not sensitive enough. Besides physical/chemical characterization studies, comparability studies will often include in vitro or in vivo bioassays in models and may include animal pharmacokinetic and pharmacodynamic studies and toxicity studies and, in some cases, even clinical data is required such as immunogenicity data, pharmacology data or maybe even safety data. After data from a comparability program provides evidence to reasonably conclude that the manufacturing change has not adversely affected product characteristics, the change may be approved by FDA. When comparability, however, cannot be established because it appears that the product has been adversely impacted or perhaps the methods applied are not sensitive enough to establish that no change has occurred, then clinical studies may be required to show that the change is not associated with an adverse effect. For older, long-established products routine release tests and methodologies may be based on classical bioanalytical methods, but for comparability studies more advanced methods are required and if they are not available clinical studies may be requested. A comparability protocol is a prior approval supplement describing the comparability program. It is intended to facilitate the review and approval of a discrete change in facilities, equipment or manufacturing process by establishing agreement between FDA and the sponsor over the content of a comparability program and the acceptance criteria. It is intended to provide a route to reduced reporting category. The biological licensing/reporting categories that are codified in 21 CFR 601.12 are the annual report, changes being affected supplement, changes being affected 30 supplement and the prior approval supplement. Typically, a comparability protocol would be used to downgrade reporting category for a prior approval supplement to a CBE-30 since that is where the biggest bang for the buck occurs. Once FDA approves a prior approval supplement containing in part the comparability protocol and acceptance criteria, and when the protocol is exercised and preestablished criteria are met, the change may be approved on the bases of a CBE-30. In a workshop concerning well characterized biologics held about seven years ago, FDA specified four types of biological products considered at the time to be well characterized, namely, therapeutic DNA plasmids, therapeutic synthetic peptides of less than 40 residues, monoclonal antibodies for in vivo use in proteins derived from recombinant DNA. While the term "well characterized" is not rigorously defined in regulation or guidance, it was stated in the workshop that a well characterized biologic is one whose identity, purity, impurities, potency and quantity can be measured and controlled. The term also suggests to me having detailed knowledge of the mechanism of action, product process and clinical performance such that consistent and predictable manufacturing can be controlled. Well characterized seems to be related to the concept of comparability, although it seems that this relationship is not essential. Plasma derivatives are not specified by FDA as being well characterized and comparability approaches are still allowed. The importance of well characterized designation in the regulation of change through comparability approaches is something that will be clarified in this meeting. What is the downside if plasma proteins are not considered to be well characterized? Can we still manage process change using comparability approaches? Plasma protein products are typically not as highly purified as products derived from biotechnology. They are frequently highly enriched concentrates of endogenous proteins. These may be considered to be less characterizable than a product derived from biotechnology. Is this a problem? Is this characteristic of proteins isolated from natural sources a stumbling block to considering plasma derivatives to be well characterized? If the answer is yes, then what is the effect of being considered to be less characterized? Are we waning in our level of understanding of a relationship between structure function of plasma-derived proteins and the impacts of change on structure function? Is it a problem that in fractionation multiple products are stripped from source plasma pools but in biotechnology the manufacturing process is directed at detection of a single product? If plasma derivatives are not considered to be well characterized and if that matters, then perhaps we can suggest that they be considered to be substantially or approximately characterized. The product characteristics of plasma derivatives are routinely measured and controlled using both classical and modern tools of protein chemistry, and some examples are shown here. Safe, potent and pure products have been made, for the most part, consistently over long periods of time. Today using modern methods, biologics can be characterized to a level not previously achievable providing improved opportunities and a stimulus to apply comparability approaches in the regulation of plasma derivatives. You will hear in one of our next talks how modern methods of bioanalytical characterization have been successfully applied to plasma derivatives. We hope to dialogue over the next day and a half to better understand FDA's expectation about the characterizability of natural biologics, and the relationship of comparability, and what differences there might be between a well characterized biologic and one that is highly characterized but still may not qualify for this distinction. Three important areas where comparability concepts are being applied are likely to be applied with greater frequency are in the development of new product presentations, such as new strengths, packaging presentations, the upgrading of facilities, manufacturing processes, equipment and in the exchange of fractionation intermediates. The nature of our industry today requires finding efficient ways for manufacturers to exchange fractionation starting materials, intermediates, under an appropriate level of control. Using a company's A cryoprecipitate to manufacture company's B Factor VIII, or company's C manufacture of immune globulins from company's D Fraction II plus III paste is not uncommon, and the practice is likely to increase over time. Another new and important area for increased application of comparability concepts in the plasma derivatives area is in process, equipment, facilities changes that enable the rapid and economical deployment of new technologies aimed at achieving approved improved process, assuring pathogen safety, reducing product shortages and providing access to new processes, facilities and equipment that better conform to ideal models of GMP. For comparability to be a useful tool, plasma derivative manufacturers need to develop comparability protocols with confidence that FDA's requirements are understood, and protracted negotiations over the content of a comparability protocol will not be routinely encountered. Predictability is very important. We recognize that a lot of our comparability depends on the specifics of a particular process change. Historically, everything has been case by case; maybe it is time to challenge this. The comparability concept was developed to reduce regulatory burden, and the need is for industry and FDA to establish together, as best we can, rules, paradigms, guidelines, expectations, etc. to improve our ability to anticipate and plan to meet requirements. The comparability concept has the potential to be an important regulatory tool but for it to be useful, it must be predictable and provide realizable benefits. The Plasma Protein Therapeutics Association membership hopes that the utility of comparability protocols can be improved through clarification and specification of requirements for plasma derivatives, and we hope that this meeting will catalyze this effort. Thank you. DR. WEINSTEIN: The next speaker will be Chris Joneckis, who will give the CBER perspective on comparability. CBER Perspective: Comparability DR. JONECKIS: Good morning. My name Chris Joneckis. I am the senior advisor for chemistry manufacturing and work for Dr. Zoon, which means I do a lot of interesting things and one of those things recently, for a period of time, has been comparability. Let me point out that this conference is very timely in that we are taking an internal look across CBER at what we have done regarding comparability over the past six years since the guidance, as it has been termed, has been issued. So, this conference is very useful and important to, I think, let us sit back and help us in our internal thinking as well. Secondly, it is important I think also because there is an ICH initiative to development a concept or, rather, a guidance on comparability and, although the proposed scope will focus on biotechnology products, sometimes those guidances seem to have applicability in other regions towards a broader variety of products. So, that is also very useful. What I am going to talk about today is a little bit more of the broader perspective of comparability across CBER. Everyone is well familiar with the CBER mission statement, but the part I would like to focus on is the fact that the regulation of these products is founded on science and law to ensure that purity, potency, safety efficacy and especially availability of those products. Before that term "availability" was formally added to the mission statement, CBER had a long history of partnering and working with industry to facilitate the delivery of products and approved product changes. As you have heard, that in part facilitated the development of this guidance back in April of 1996. It was clearly made possible by a lot of advancements in time of manufacturing methods, process control methods such as validation and other tests, and analytical tests to assess products, the drug substance, the intermediates and such. The guidance was devised for all biological products regulated by CBER and, as you have heard, the key product is to demonstrate product comparability of a pre-change and a post-change product, and the whole issue was at this point that, depending upon several factors which some speakers have spoken about, one may allow a change without the necessity for preclinical and/or clinical testing. Well, what is comparability? The closest we come to a CBER definition is taken from that guidance, and it is that FDA may determine that two products are comparable if the results of comparability testing demonstrate that the manufacturing change does not affect safety, identity, purity or potency, essentially a very broad and operational term. Others have proposed different definitions that I have seen in various forums. I think the other way to define comparability is what it is and what it is not. The tests that are used for a comparability program do not really allow us to determine that the pre- and post-change products are identical in fact, although, for example, the analytical methods may be able to say that the pre- and post-change products are indistinguishable is the key point, I think, to make. The second point is that you can also say what it is not. Well, clearly you can move into the range of it being different. That also depends upon how one defines "different" and what the operational types of terms are used, the criteria that are applied, although with the caveat that you may have certain differences for example in analytical assessments as long as they don't translate into significant clinical safety and efficacy effects. So, comparability falls somewhere, in a sense, in that area. The concept of comparability, again, is across the life cycle of the product and that is how it is applied in the guidance and how it is applied at CBER. Of course, there is deference to the fact that there has to be some flexibility in where one is in this whole life cycle to allow for flexibility for process and product development, especially early on, but, again, in those situations you are always doing some type of clinical and perhaps preclinical testing. The real issues that come into play are more during the Phase III or post Phase III changes and post-approval changes where, again, the need may not be to do preclinical and/or clinical testing, and I think that is largely the focus of the issues today. The other point is, as people have said, that the comparability protocol, which is basically an operational mechanism which may allow one to get a reduced reporting burden on this post-approval change and may in certain cases allow for expedited product release, has been used and has been used quite successfully by a whole variety of products across CBER. The point to make though is that the concept of comparability clearly falls into this comparability protocol. The elements, as other speakers have said, of the comparability concept, as well as various considerations in developing this comparability assessment program, can be sort of grouped into several areas: process, product, analytical, predictability and manufacturing changes. This is the way I have characterized them. Just a few quick comments on that, the old dogma or the dogma that the process is the product may or may not still be applicable; may or may not be applicable depending on the product class that one is speaking about. Whether that is a debatable topic, I think it is still clear to say that the process clearly defines what that product is going to be regardless of the starting source or starting material that may be used. Again, the products across CBER are still heterogeneous. There is an inherent heterogeneity across those various products. There are also considerations for certain molecular complexity and the influence of impurities which subsequent speakers will discuss. Also, the analytical ability, where we are currently in the analytical capabilities--what are the capabilities to detect differences? What are the inherent limitations of all methods that we face not only in looking at the comparability issue but in approving new products? What is the ability to detect small differences in large molecules that may have profound consequences? Again, the predictability, as other speakers have said, is based largely on the knowledge, history and experience of your product and your process come from developmental studies. Lastly, the type and extent, the complexity, if you will, and where you are in the life cycle concept of the manufacturing change. So, all these are looked at in considering how to determine the comparability assessment program. In many ways they are interrelated--again, what kind of process and the purity of that resulting product and the heterogeneity and complexity of that product to some extent determine what analytical methods you may be able to apply, and the results that come from those analytical methods. The type and extent of this change will also influence what types of additional studies may be needed. Again, as has been identified in the guidance, the key component is what you can predict. The next three slides is material I have already covered on this slide, so I will just skip them. This is the life cycle of CBER, as it has been called the world of products. When you look at comparability, comparability really has been applied to this upper quadrant of products, mostly because these products for the most part are still in the development cycle and there are very few licensed types of products. Plasma derivatives will be the subject of this conference so I won't discuss those but just a few comments on these other types of products. Vaccines has had very limited use of post-approval types of comparabilities. In many ways, they are heterogeneous, naturally traditionally derived material and they have inherent concerns about being able to define the heterogeneity and the complexity of those constituents in the drug substance, including the active ingredient. So, there has been rather limited use of the comparability post-approval, at least in terms of being able to not require some type of preclinical or clinical testing. The largest group of experience we have had with comparability has been for the specified products, in particular these two highlighted here, the therapeutic DNA-derived products and the monoclonal antibodies from recombinant or naturally derived sources. As other speakers have alluded to, what is the relationship among these various products? If one looks at a synthetic product which may include things such as chemical identity or a synthetic peptide versus the specified products as defined by the FDA versus some of the more traditional products, in general I think people have indicated that as one moves from left to right across the screen there is increasing impurity, increasing ability to be characterized and perhaps to some extent decreasing heterogeneity, although I should point out that I think there are, as I have tried to show here schematically, overlaps within these various types of product classes. So, in fact, there has been a naturally derived product that does have properties similar to that of the synthetic. So, it may be easier to talk about individual examples or products within a class as opposed to the entire class altogether. I think that is reflected in how we approach comparability and is part of the overall assessment that one has to make in developing a comparability program. I would like to spend the next few slides on examples that we have learned from specified products. Time does not allow me to go into substantial amount of detail, but I think it is important that we take a look at what we have learned from a well characterized group of products. The reasons for manufacturing changes in these products is summarized here. I think it is generally applicable to say that these are also the types of changes that we see for various products across CBER. Again, there are the standard types of pre-approval process development. This is a big one, increasing product supply for many of our products and it has been accomplished through process optimization such as increasing yield, scale-ups or duplication of existing processes or additional manufacturing sites, either from a contract nature or from additional sites put on by the initial manufacturer. Again, they may be driven by just process optimization, updating the process for economic concerns or newer technologies coming on line. Again, there are always compliance-driven elements, for example, the need to eliminate human and animal derived components, or inspectional issues, or general facilities improvement. So, I think those are sort of the types of changes that we have seen. So, what can we say about specified products generally? Well, overall the scale-ups are generally less problematic in causing problems. It sort of makes sense when one stays with the same process and the same principles of those processes, they seem to be less problematic in types of concerns and issues that we have or have seen. Changes in an early manufacturing stage, the cell bank or the cell culture fermentation are of great concern and do have great potential to impact the drug substance. There are several examples where we have some additional problems in terms of looking at comparability. There have been examples of inadequate characterization of raw materials or components that can affect products, and those can be naturally derived materials, biologically derived materials or actually even chemical entities. There are examples where if chemical entities have not been completely characterized or characterized to the right quality issues, they can profoundly change the whole molecule and cause products to not be comparable. There have been examples of that. Formulation changes can affect products, especially for sub cu. and im. products. There are several examples where that has happened. And, there are several examples where there have been changes at site of manufacture that can affect the drug substance and drug product. Picking up the product and moving it to a new location with no or extremely minimal changes that one would not predict to have an impact have had changes. Examples have occurred during late product development, such as Phase III type of situations or even post-approval. Overall, meeting drug product release specifications alone is not sufficient. Characterization studies must be performed, I think, for all the changes that we have had for specified products, as well as other products--the complete analytical characterization, the complete package has to be performed, as well as perhaps relevant in-process control tests on appropriate intermediates. Changes for within specifications may be important. There have been examples where even though specifications have been established for products and all of the resulting post-product changes have met specifications, there have been situations where one product or one product lot has been out of trend, if you will, and that has been shown not to be particularly comparable. What does that say? Does that say that perhaps specifications were not appropriately established to begin with? Appropriate characterization or process development studies were not performed? That is possible, but it is important to highlight that, again, within specifications changes can be important and they don't tell you everything. What can be very sensitive measures is trending the acceptance criteria for multiple pre- and post-change lots, in addition to doing side by side characterization of, say, your typical three lots that people like to do. Looking at those trends has been very enlightening in terms of seeing what types of changes have occurred and, again, combining that with what type of changes may be seen within specifications has been particularly illustrative. It is difficult, I realize, to do this especially as it requires a substantial post-change material. Of course, drug product and drug substance can change upon storage and you need to consider that in development of the post-approval comparability protocol. Analytical testing has been the basis for establishing product comparability in specified products in some cases, actually in quite a few cases. The PK data may or may not be needed. Again, it is part of that overall algorithm and considerations that have to be looked at. For more substantial changes, depending on how one looks at that, PK studies are again particularly needed. PD studies really aren't done for specified products for many reasons. There are very few PD studies that we have had. Hence, it is really not utilized per se. Clinical efficacy and safety data generally is less likely to be needed, although it has been requested in some cases. Again, assessing immunogenicity is of increasing importance. To illustrate that point, I will talk about one specific example which I am sure is familiar to many people in the room, and that is with the product erythropoietin Eprex, which is made in a mammalian system. It is not marketed in the United States but is marketed and distributed in Europe, Canada and Australia. There was somewhat recently a manufacturing change that was made, including a protein free formulation that included removal of human serum albumin. As of April 30th, there have been over 116 cases of suspected red blood cell aplasia that have been reported to the FDA. Approximately 85 percent of those cases have been confirmed through bone marrow biopsy to, in fact, be red cell aplasia, and approximately 50 percent of the patients evaluated have had high antibody titers against erythropoietin. In other words, this is an example where making a change in a process that would not be predicted to cause this effect has resulted in the development of an antibody against the Eprex molecule and has resulted in removal of the endogenous epogen in the individuals in these cases. Those cases need red blood cell transfusions in order to continue to survive in many cases. There are theoretical hypotheses as to why this has happened, but nothing has been conclusively determined at this point. There are reference articles that describe some of these earlier cases, and also a response from our epidemiology group and other groups within the agency that you can look at if you would like. I bring this point sort of towards the end to make the point that these changes in comparability can have profound enduring consequences, and our ability to predict these kinds of changes from our knowledge and history is still limited. So, there is no established formula for determining comparability testing requirements for specified products but I think that is the universal theme across all of our products. Despite best efforts to detect product differences and predict the impact of manufacturing changes, these surprises do continue to occur, again, echoing the theme made by the earlier speakers about where are we with our knowledge and experience database, if you will? The algorithm that is described, if you will, in the comparability guidance has in many cases caught changes for specified products that have not been seen through analytical testing. So, we have been fortunate in not being able to have any of those severe consequences from comparable products, such as the Eprex example, happening for products that are regulated by CBER. In one of Henry James' novels, he writes experience is never limited and is never complete. We realize that we only see what you all show us; we don't make these products. In speaking with several colleagues, I know there are other experiences out there from changes that have been seen in untoward effects. So, I think that that be shared and discussed is very important because, again, it goes to expanding the general knowledge base that we all have. The road ahead? Well, the changes that allow us to implement the 1996 guidance, those manufacturing methods, analytical methods and expanding our knowledge base continue to advance and change and I think they may be useful in future to affect what we can determine from analytical, for example, assessments. There are certainly new analytical methods coming along that provide more direct measures of structure and, of course, in relationship to activity which is very important. We, at CBER, also spend a lot of time doing this. We have NMR facilities to look at the structural techniques. We are spending a lot of time doing proteonomics and other laboratory CHIP types of techniques to try to see what additional types of information can be provided from these methods, and the confidence that can be gained from those particular methods. The bottom line, however, is that approval of any product or product made from that manufacturing change must always ensure quality, safety and efficacy of that product. Lastly, I would like to thank the individuals in the Office of Therapeutic Research and Review for some of the information provided regarding specified products, and the Office of Vaccines regarding the information they provided on the comparability experiences that they have had. Thank you. DR. HAYES: Good morning. I am Tim Hayes, with the American Red Cross, and I am going to be co-moderating in the next session with Mark Weinstein. The next session is the comparability studies for human plasma-derived therapeutics, and we are going to be taking a closer look at the individual components of that, product characterization preclinical studies, as well as clinical studies and how those work together. It is my pleasure to introduce our first speaker, Dr. Andrew Chang. He is a special assistant to the director for the Division of Hematology, Office of Blood Research and Review. He is going to give us the FDA perspective on product characterization. I would just like to add while the slides are coming up that Andrew and some of the other people in the audience have actually worked together with another FDA co-sponsored event, for well-characterized biological products conference, and have actually been working and gotten a very good start on this issue of appropriate application of characterization to these types of studies. Andrew? Product Characterization FDA Perspective DR. CHANG: Thank you, Dr. Hayes. Thank you for the introduction. Before I start my presentation, I would like to use this opportunity to thank the working committee for this workshop. Those include industry people and the people in the FDA. For the industry people, we have Craig Mendelsohn, who is a co-chair for this workshop, and Michael Gross and Chris Healey and Jean Huxsoll. I apologize if I have missed anybody from industry side of the working committee. From the CBER side we have Chris Joneckis, Mark Weinstein, John Finkbohner, Timothy Lee and also Joe Wilczek and myself. I would like to thank the committee members for their hard work and effort to make this meeting a success. Also, I have two housekeeping items. For people asking questions, please use the microphone. We are recording this workshop so we want to capture the questions. We may not have enough time for the questions that you are going to ask. You also can use index cards and, as Dr. Chris Healey mentioned, we will have a panel discussion later and use your index cards for your questions. Another thing is, speakers, please, keep within your time frame. We want to finish this workshop on time. This also applies to our moderators. I would also like to thank all the speakers for volunteering to give their speech in this workshop. My talk will focus on the product characterization for plasma derivatives. I have the following topics and I will very briefly go through the comparability policy that we have. The previous speaker has already covered that area extensively so I will keep that very brief. I want to focus on the experience that we have had in the past five years since the publication of the comparability guidance in 1996, associated with the plasma derivatives and some of the recombinant products that are found in the hematological product criteria. Lastly, I will go through some of the FDA's perspectives. The FDA 1996 guidance on comparability actually is a policy that we published in terms of comparability. The policy has resulted from the desire to make improvements in the test methods and product production process within a single manufacturer--I emphasize single manufacturer, the same sponsor to make manufacturing changes for licensed products. FDA may determine that two products are comparable if the results of the comparability testing demonstrate that a manufacturing change does not affect safety, identity, purity or potency. This policy allows for changes in product characteristics if they have no adverse effect. As Dr. Joneckis pointed out in one of his slides, the comparability is really between the identical and also differences. We have room to work with in terms of comparability as long as the differences will not affect the safety, identity, purity and potency. In the guidance, as spelled out, there is a three steps approach. The first step is analytical functional approach, then preclinical and clinical. The one thing I want to emphasize here is that this is not simply a hierarchical system but, rather, a complementary one. For example, if you find some differences in your preclinical study that may trigger some additional in vitro analytical functional study. So, it is not that you have an individual study and then you leave that stage, it is, rather, complementary. Now I am going to change my topic to the experience we have had in the plasma derivatives area. Some people may ask if the comparability concept been used for plasma derivatives, and the answer is yes. We have seen the comparability approach applied to the manufacturing changes for plasma derivatives. Then, some people may ask when does that complication approach start? Did it start after the 1996 comparability guidance? Dr. John Finlayson has said, no, we have had that concept a long time ago, before the first biotech recombinant procedure was licensed in this country. So, the concept is there and has been used for the plasma derivatives. Do we have any concerns? Yes, we do have many concerns, as Dr. Weinstein pointed out in his presentation. He listed some major concerns that we have in dealing with plasma derivatives, as well as all the biological products. Later I will point out some significant concerns for plasma derivatives. I have tried to categorize the plasma derivatives that we have licensed in this country. These include coagulation factors, such as anti-hemophilia factors and von Willebrand factors. I have a list here and I am not going to go through the whole list. Then, we have another type of plasma derivatives such as albumin and TPF. We have a couple of NT protease inhibitors, such as alpha-1 antitrypsin and NT thrombin 3. We have a family of immunoglobulins that I have listed. That may not be the complete list for that type of product but I have listed these over here. We also have five recombinant coagulation factors that have been licensed in this country. Those include the BeneFIX, ReFacto, Kogenate FS, Recombinate and the Novo Seven. We have one recombinant antihemophilic factor concentrate for further manufacturing. What is our experience? How often are we seeing major manufacturing changes? With the help of the staff in the Division of Hematology, I have gathered some information just for your information. In the past five years we have received from 70 to 100 supplements for major changes. The previous speaker already gave you some indication of different categories of changes. Also, we have another session in the afternoon to further discuss the reporting category and requirements. I am going to focus on the pre-approval supplements which are normally used for the major changes to a licensed product. The number for major supplements is actually quite steady since 1996 and also this is true for plasma derivatives and also for recombinant hematological biologics. At the top I list the number of the major supplements that require clinical data to support the change. You see that there are actually very few numbers of supplements that require clinical data. We have six of the prior to approval supplements that require clinical data to support the change. Actually, three of them are efficacy supplements for a new indication. So, in terms of manufacturing changes we only have three major supplements for manufacturing change, and those changes include formulation changes and also alternate manufacturing processes. The percentage for those major manufacturing changes requiring clinical data is about 1.31 percent. As I said earlier, three of them are efficacy supplements for a new indication. So, the number is cut in half so it is, let's say, 0.7 percent. If you look at the data in 2001 and then separate it to the different categories in terms of the changes, we have found that 13 of the major supplements involved a change for the process; 13 of them with assay, which could be a new assay or introducing a new standard for the analytical assays; 24 of them are one-time exceptions. I don't know whether you know this term. One-time exception supplement is that the manufacturer sends a supplement to the agency requesting to release one or a few lots that had some manufacturing differences from what they have normally have as permitted by the license. We have other 35 supplements involving other categories, such as stability issues and labeling changes. Again, this data was collected for the plasma derivatives and also the hematological recombinant products. What is our approach, regulatory approach for major manufacturing changes? It has been handled on a case by case situation. The factors that influence that approach really depend on the following three areas, one is product. We have different products and we have different knowledge on the different products, and we know the risk involved with the product are different. The second element is type of manufacturing changes and, lastly, risk analysis and assessment that play a role in the regulatory approaches. Instead of giving you some specific examples, I decided to give you two types of examples that will cover each end of the major manufacturing changes. The first series of examples includes type of changes such as a new facility with no change in in-process control, no change in specification, demonstration of individual comparability. Another type of change is a new assay standard for quality control, lot release. The third one is a one-time exception. What is the review mechanism for this type of change? We review this type of change as a prior approval supplement for which, under PDUFA 2, we have four-month review time for that. This normally includes data for individual biochemical, biophysical characterization and for some of supplements, such as new facility, we also conduct a pre-approval inspection. Another type of example, which is at the other end of the major manufacturing changes, is the most difficult change, involved with a lot of changes. This type of change is like a new facility with an automated process, and also changes in specification for the drug substance and the drug product for which a company will demonstrate comparability. Now, very often--actually, it is always the case that the sponsor voluntarily phases out the old process after agency approval for the new process. The review mechanism under PDUFA 2 is ten months review time because for this type of manufacturing change clinical data is required. This will include in vitro biochemical, biophysical characterization, preclinical studies and some of the preclinical studies such as human pharmacokinetic data and some of the safety and efficacy clinical data. Pre-approval inspections are always required for this type of change. Very often the sponsor proposes a new proprietary name for the product manufactured with that alternate process. Conclusions, comparability approaches apply to both plasma and the biotech-derived biologics. I said earlier that that has been used a long time ago. In our experience, clinical data has seldom been required to support manufacturing changes, however, major concerns remain. What are those concerns? In addition to the concerns that Dr. Weinstein mentioned earlier, we have the following concerns, such as poorly defined study material, source plasma versus recovered plasma, different pool sizes can be used for manufacturing, and demographical and racial differences in the source material and, in addition, some of the manufacturing steps with different intermediates. That is another topic that we are going to cover tomorrow. Lack of robustness of the manufacturing process, minor changes with major impact, and we will have some example case studies in this workshop to demonstrate minor changes with major impact on the product safety and efficacy. We have a type of product with very low purity, and impurities may affect activity, immunogenicity or absorption. Often highly complex and heterogeneous proteins for plasma derivatives and also it is quite important that we have a history of viral transmission for this type of product. So, what is the FDA perspective? Since my presentation is under the individual characterization I am going to focus on the individual characterization part. Analytical and functional testing, physicochemical, functional, biological, immunological studies have been used and performed to support the comparability. Sensitive tests measure all criteria functions of the product. We would expect that because of the complexity of the product, some of the proteins have more than one active site. The example for that is, for example, von Willebrand factor. Considerations for product related and process related impurities and contaminants; qualitative and quantitative assessment, validated and qualified methods should be used for your study. Another thing has been touched upon by Dr. Gross, and that is that we are not satisfied with your routine testing to support comparability studies. That routine testing includes in-process control testing, the final release testing to meet specifications and we ask for additional testing to support the comparability. What type of additional testing will be required for that particular manufacturing change? Well, it depends on the change involved and the product involved. So, there is no easy way to give you a formula to do that. You know best because you know the product and the process more than anybody else. So, you should be able to come up with a program to assess the comparability. The approaches to establish comparability started out in the 1996 guidance. What I am going to do here is just point to some language in the guidance. Side by side comparison--manufacturers should provide to FDA extensive chemical, physical and bioactivity comparisons with side by side analysis of the old product and the qualification lots of the new product. This is language in the guidance for side by side comparison. Also, reference standards should be used in the comparability studies. This is also spelled out in the guidance. When available, fully characterized reference standards for a drug substance in the final container material should also be used. Lastly, comparison with historical data. That is especially important when the reviewer looks at some in-process control parameters and we are expecting the company to do some statistical analyses to demonstrate that a particular step has not been changed or is comparable. I want to emphasize "know they process and they product." This term has been used in one of the previous workshops for process validation so I quoted that. Establish validated manufacturing experience and product history. That is a very important element. Conduct thorough drug product, drug substance characterization. Establish specifications, statistical trending, sensitive discrimination assays. As you all know, specifications should rely on your historical data as well as supportive clinical data. In order to have meaningful specifications, your process should be well controlled. Those in-process control parameters and the range for those control should have good correlation with your final specifications. It is difficult to generalize impact regarding a specific change to all classes of products. Again, we have three different categories that we are looking at when we deal with manufacturing changes. Those are product and type of manufacturing change, and also risk assessment. Issues for consideration--these are issues that Dr. Hayes mentioned. We had a workshop, a 2002 symposium. In that symposium Dr. Hayes and I co-chaired a workshop to look at some manufacturing gaps that we have between the specified product and the plasma-derived product. We actually found that there are some gaps for the plasma derivatives. These include the source material, can or should study source material, in this case for plasma derivatives plasma be better characterized? Can or should the manufacturing process for human plasma biologic be validated to the same extent as that for a specified product? If a plasma protein is heterogeneous, to what extent should heterogeneity be characterized? How should impurities be assessed? My last slide is with some recommendations. Control source study materials and characterize them if feasible. Establish robust, reproducible and validated processes. Establish sensitive and discriminating characterization and release testing. Document manufacturing history and experience. Establish normal variation of a product supported by statistical analysis and clinical experience. Conduct and submit developmental pilot studies for manufacturing changes. This is especially important for a comparability protocol. Consider applying new analytical technologies. Qualify impurities. Toxic impurities should be identified and controlled. That will be all for my presentation. Thank you for your attention. DR. HAYES: Thank you very much, Andrew. Our next speaker--we need to be moving along here because, again, we have a very ambitious schedule--is Ghiorghis Ghenbot, who was also a participant in that workshop at the WCBP, and he will be giving us the industry's perspective as far as product characterization is concerned as involved with comparability protocols. Industry Perspective DR. GHENBOT: Good morning. I will present on product characterization industry perspective. When I was preparing for this seminar or workshop, I had several specific questions to answer for myself. Somehow, they will be the same questions that you kind of ask yourself when you consider characterization of plasma derivatives or, for that matter, biotechnology derived products. Here are the questions. Number one, why should we characterize a product? Two, what should be a characterization strategy be for a plasma derivative protein package include? Three, is there a procedure or a precedence for this activity? Do we have some experience from the past? If so, can this experience be used to characterize plasma-derived proteins? Finally, how is this characterization activity related to a comparability protocol and comparability activities? After all, that is exactly why we are here. We may vary in terms of answers to these questions but I hope we have some sort of agreement somewhere during this presentation. I will try to give you my answers. Here they are. Why should we characterize a product? I think we should characterize a product because particular characterization is an integral component of setting specifications, product specifications. This is very clearly stated in Q6B. As a matter of fact, the ICH guidelines for test procedures and acceptance criteria for biotechnology/biological products under Q6B state that specifications are one part of a product strategy designed to ensure product quality and consistency. Secondly, we should characterize a product because proper characterization ensures the safety, purity and potency of the biological product. Third, characterization helps in identifying the criteria parameters for defining product quality. Characterization data places comparability programs on a secure foundation. Again, back to a CBER publication of 1996, it states the following, manufacturers should provide to FDA extensive chemical, physical and bioactivity comparison with side by side analyses of the old product and the qualification lots of the new product. So, the directions are already stated in the documentation. Also, proper characterization of plasma-derived proteins ensures the development of well-characterized plasma therapeutics. The bottom line is we need to characterize it to make sure that things are fully under control. Finally, I think product characterization is a sound business decision. In this case, the FDA has accepted manufacturing or controlled changes for well-characterized products again case by case, without clinical proof of product safety and efficacy in terms of money and dollar investments, I think this is a huge saving for any company which is involved in this type of activities. Now, when do we characterize the product? This is not going to be an easy one. The bottom line is I cannot say that I can characterize a product at one time and that is it. No way. I think you should cover the lifetime of product. Therefore, I submit to you that we should start characterization activities during early stage of product development. Why is that? Because characterization efforts should proceed hand-in-hand with normal procedures of assay development, with particular emphasis on biological activity, so that one assures that the assay is a surrogate of the proposed physiological activity. Second, we should also continue characterization of the product in Phase II when clinical data are evaluated. In this case, dose and efficacy targets are being devised. Process or formulation changes may be made, and physical chemical tests that ensure lot-to-lot consistency, safety issues surrounding impurity profiles, product heterogeneity and stability need to be further defined. Also, we should continue characterization in Phase III, and that is because this is the stage where emphasis is placed on validation to show that the product meets specifications. In this case, product characterization is needed at this stage to assure and justify specifications. We have a chance to get rid of some of the assays that we have developed throughout Phase I and Phase II. Also, we should continue characterization because at some point there is a need for definitive data. Finally, and very important, times are changing and times do change. And, almost every speaker before has alluded to this one, here. Now, what should a characterization strategy for a plasma-derived protein therapeutic include? The elements that I think this characterization strategy package should include are the following: identity, quantity, purity, impurities, potency and safety. There is nothing new here. The difference is how should we form this package in such a way as to take advantage of the past in terms of biotechnology-derived products and now for our plasma-derived products? It was also very clearly stated that actually the experience is based not on biotechnology-drug products but actually on plasma-derived products. So, we are kind of coming back. In terms of identity, I submit to you that one has to look for a highly specific test reflecting the unique aspects of the product structure. In other words, focus on what you need for that particular product and not go by a certain specific definition. In this case, one or more tests based on physicochemical, biological and/or immunochemical methods should be appropriate. In terms of identity, I think we can sum up the activity in terms of structural characterization as well as physicochemical characterization. For the structural characterization aspect, there is basic information such as amino acid composition and N-terminal sequence and peptide mapping to look for some specific type of activities that we care about. Most importantly, these two, I submit to you, should be very carefully monitored for plasma-derived protein studies, post-translational modification in the form of carbohydrate composition, if there are any issues of sulphation and other things. In terms of carbohydrate composition, as I said before, we need to look at the structure as well as composition. Bear in mind that I am not focusing on sequence here; I am actually looking at the package of that profile and now it changes throughout the process and if there is any issue regarding sort of the total amounts. To go back a little bit, I would submit to you I would be concerned about this issue if there is any concern about heterogeneity of that product. Furthermore, in physicochemical characterization there are techniques available, old techniques as well as new techniques. There will be cases where we need to revisit the molecular size of the product that we have in hand. There is always that issue here. We have done it twenty years ago, we have done it thirty years ago, but technology has changed and now we have newer elements or newer instruments so that we can fine-tune our studies in this case. It is very easy and it is possible to do it either by instruments such as MALDI-TDF as well as the ES-MS, right here. In terms of the electrophoretic profile, we also have newer techniques, SEC. In terms of chromatographic patterns, we can also look at the various procedures such as reverse phase HPLC as well as affinity chromatography. As you know, in my presentation I am really focusing on an in vitro characterization approach because the other speakers are going to cover the in vivo approach. I kind of like the bottom part of this slide, and that is because there is a possibility, in addition to the physicochemical characterization outlined above, to correlate activity with structure and perhaps, with some data in the future, with immunogenicity. I strongly believe that if there is some correlation between structure and function, the product should be well under control. In this case, you can look at the tertiary structure of the molecule of interest using spectroscopy, calorimetry, analytical ultracentrifugation, as well as SPR technology. Now, the following are few examples of a plasma-derived product. In the process of characterization some of the experience that we have here is very simple. This is HPLC. There is no big deal about it. But I think we also need to look at the reverse phase HPLC simply because if there is a possibility for breakdown products or related impurities, one can look at extended programs of this sort of approach. The bottom part of this slide shows the electrophoretic pattern of certain plasma-derived products of the same product, by the way, and this one was also controlled by capillary zone electrophoresis. Another example on identity is based on evaluation of the product, in this case using CD. You can look at far UV CD which is not that important in my experience, but I kind of believe that near UV CD has pretty good correlation in terms of structure and function. More importantly, you could also use the same CD to look for the protein stability, and this issue becomes very important when you deal with formulation changes, stability issues and shelf life. That is because we would like to know at what temperatures we can start to store material. One can monitor or simply make a certain baseline based on the CD profile of the transition temperature measurement for the protein, or you could do the same thing using differential scanning calorimetry. In this case, it is interesting to note that there is no difference whatsoever in terms of the temperature observed. Now, if that is the result we get every time, I think that would be very easy. Moving on to the other part of the elements of product characterization, there are certain things to consider when we talk about quantitation of a biological product. There are a number of methods that every one of us is really kind of familiar with. Methods such as shown here have been around for a while, but in more recent days people have also shown that analytical ultracentrifugation can also be used to quantitate proteins. Why do I point to this particular one? Simply because with the same system you can also look at binding characteristics; you can look at fragmentation; you can look at product heterogeneity, although the system is extremely expensive and issues of validation and things like that are of question. That is, it is purely investigational at this stage, at least as far as I know. The other element of product characterization is the purity/impurity profile of the product. In this case, I would like to look at two issues. I would like to look at product-related impurities as well as process-related impurities. In terms of product-related impurities, we can talk about truncated forms of the product. There is a possibility that some sorts of fragments of that product can be formed. I kind of think that it is very unlikely, especially with plasma-derived products, although there are very well documented cases where there can be some sort of heterogeneity in terms of the N-terminal sequence. I haven't seen that much in terms of C-terminal sequence. This perhaps is important when one has to consider the purity of the protein as well as the biological activity. Also, and very often, chemically modified forms do occur. That is perhaps due to oxidation, most of the time, or perhaps due to deamination and maybe sometimes due to isomerization. I kind of believe that post-translational modification for plasma-derived proteins in the dorm of modification of carbohydrate groups is very unlikely to take place. The other part of this purity/impurity profile is looking into process-related impurities. Now, the previous speakers have mentioned the safety issues with plasma source material, and that is a huge chapter in itself and I have no interest in really focusing on that particular topic now but it is very important and it is hard to cover all the aspects of safety issues at this particular time because I am focusing on in vitro methods. On the other hand, there are methods for looking for those agents that we suspect to be there. The only point or the only place that I would like to talk about is the bottom part of this slide. We are talking about downstream process and you can use LC methods to monitor these types of impurities. Again, it is not generally in plasma-derived proteins. Other methods for procedures to look for structurally related impurities in the product that we have, what are the analytical studies? An easy way to look at this is probably is actually to break down the type of changes that you see in the problem that you have at hand. You can very easily classify the type of structure in terms of those types of changes that are derived because of the chemical instability of the product of interest, or it is simply because of the physical instability of the product. In terms of chemical instability, you can have some sort of breakdown or degradation, as well as oxidation, deamidation, disulfide exchange and glycosylation issues. The strategy that one can follow is a combination of seven different procedures, and there is no one particular method that one can say this is the method of choice to look for these problems but, in general, a combination of reverse phase HPLC and mass spectrometry, as well as capillary zone electrophoresis will do the job. The final part of this part though is looking at the physical stability of the protein, which is expressed in the form of aggregation or denaturation. In terms of aggregation, as I alluded to before, you can look at systems analytical ultracentrifugation and simple procedures like SDS-PAGE. On this part, here, this is also a problem that can be monitored by this approach. This is an example of looking at the purity of a plasma-derived product. In this case, there was a possibility that this protein could actually lose activity because of certain residues which can be oxidized, in particular, if you know the chemistry of the protein and if you know what residues are really related to the activity of your protein, you can kind of specifically look at those. In our case, for this particular plasma protein, we knew that there was some sort of possibility that it could oxidize and lose its activity. If you look at this protein, we generated a peptide map of the protein, and this is before oxidation and here it is after oxidation. What you see here is only a snapshot of the entire peptide map. It is evident. Here you have an area which is basically two main peaks, and the bottom line here shows you there is something creeping up. In this case, what you had up here is actually starting to disappear. Actually, in this particular procedure we could monitor it in terms of activity as well as in terms of CD spectroscopy and reverse phase HPLC. The other example in terms of looking at purity, as I said before, we have technologies such as FACE where you can actually look at the carbohydrate components of your product, in which case you label them; you glycosylate it and then you look at the labeled parts. One very easy and cheap approach of looking at your is actually a combination of chromatography and light scattering detector, using the light scattering detector. In this case, we decided to look what would happen if you really heat your protein with an aggregating agent. Obviously, the mono peak, which is sitting here, completely goes away to an aggregate peak and you could actually devise this procedure. Why do I care about this? Simply because I would like to know whether the components that we normally call dimers, trimers or aggregates adverse event discrete components of the original material and whether they can be monitored very carefully. Finally, the other issue in terms of the package, I would like to consider the product potency. Again, this is nothing new for all of us but we do have our own different options that we develop. The only thing that I would like to really focus on is this ligand and receptor type of binding assays, where we should try to kind of look at the in vitroactivity and then try to correlate to the physiological activity. In terms again of potency, we normally develop this potency or activity measurement under very, very controlled buffer conditions, salt concentration, temperature and excipients. That really takes most of the time when you start development. But what happens as time goes on? We keep on using the same assay, assuming that it should be good throughout the development process. Oftentimes that is not the case. So, we need to go back and check this approach here. We can also control this by including reference standards, reference standards that we can have either in-house or reference standards from somewhere else. These activities should also be correlated with some additional information in terms of stress testing, stability-indicating assays and shelf life. That would actually enlarge or narrow our specifications in terms of product potency. It is very hard if we stick to the first assay, the first time we develop it, and we want this assay to tell us exactly what we want in Phase II, in Phase III and even after clinical material. It is going to be very hard. So, it is really important to go back and say my experience shows so much variation in the activity. In shelf life there is so much variation in activity. So, go back and modify the activity and then come up with new values. Now conclusions, I would like to submit to you that product characterization is not an end in itself but, rather, a means of identifying the critical parameters required for defining product quality. Secondly, and very important and every one of the speakers has really hit this point here, recent advances in analytical biotechnology allow one to characterize biologicals to the levels that were previously unattainable. One needs to take advantage of these developments. I think the issue of biotechnology-derived products as well as plasma-derived products should really flourish. I really focus on protein therapeutics rather than classifying them this way or that way as far as this particular approach is concerned. Third, these advances apply equally to methods of purification and process control and, as such, the information in product characterization package should reflect product development history, clinical and licensure experience. Also, such data, and that is why we are here today, is invaluable in properly managing post-licensure comparability activities. Finally, the concept of "well-characterized" biotechnology products provides a golden frame of reference. If you look at the elements of the characterization strategy that I put forth for plasma-derived products, it is exactly the same topic that has been covered and well characterized. There is a debate as to whether we have to call this process well characterized, substantially characterized or well understood. In the end, there will be no issue whatsoever if we have to focus on the product itself and the approach that we use. In this particular case, there is one more slide. If we have to look at the elements that we put together before and look at the agency's publication for well-characterized proteins, and then we put against this plasma-derived proteins, what would be my conclusion? I think my conclusion would be the definition that was given for monoclonal antibodies in what we call well-characterized therapeutic biotechnology products, the monoclonal antibodies were defined at that time as those proteins, the identity of which would be determined by reverse physicochemical, immunochemical characterization without fully knowing its chemical structure I think the biotechnology products, the concept of well-characterized products and the plasma-derived products meet in this particular definition of well-characterized proteins. In addition, for these monoclonal antibodies one needs to know the purity and impurities. We have covered that aspect, that the purity be identified and the impurities also be quantified. Thank you for your attention. If you have any questions, I will be very happy to answer them. Q & A DR. HAYES: So, we have the opportunity for some questions now. Again, as was mentioned earlier, we need to have everybody that is orally asking some questions come to one of the microphones. We have a portable microphone that will be going around. Additionally, we are going to try to keep this to ten minutes and cut our coffee break short. So, I would like to have the different speakers that have been here this morning be available for the questions. We have some questions from the audience that we can begin with. Do we have anybody who wants to voice one at the moment? DR. VAN GEODEREN: I am Hans Ven Geoderen, ZLB Bioplasma. We actually still have pending a submission with CBER where we requested a change--and this is about IGIV--where we tried to change our release test for anti-measles antibodies. We tried to replace the current release test for that by an enzyme immunoassay, ELISA. While doing so, we got into several discussions with CBER and the last one we had was one where we were requested to set up a comparability exercise between the two antigen mixtures that are in the two kits, so the existing test and the ELISA, a comparison comparability testing of the two antigen mixtures. So, this is not about changing the method of manufacture; it is about changing the release test. Given the fact that this information about kits and which antigens are used is proprietary information which is held by the manufacturer of the kits, to me it is really a sort of stretching it a little bit. Can you comment on this? Have other companies had the same type of experiences with trying to change release tests? And, do you think this is appropriate? I will ask the question to Dr. Chang. DR. CHANG: I am not exactly sure whether I got your question, but Dr. Basil Golding is probably the proper person to answer that. Let me just give you what I think. For release testing, the agency believes that any change associated with release testing is very, very critical and you have to demonstrate the assay sensitivity and it should be very close, if it is not better. Dr. Timothy Lee will have one example this afternoon to demonstrate how critical small changes in release testing could influence a major impact. Again, I am not sure what exactly--can you rephrase your question? DR. VAN GEODEREN: Well, we have a set of release tests and specifications. We were trying to get rid of one of them. The tests were anti measles. I believe this test makes use of monkey erythrocytes. So, we set up a side by side comparison of tens of lots of IGIV, testing with the old test and with the new test. We saw that the new test actually measured more anti-measles antibodies than the old test would do. Therefore, we proposed to raise the specification in order to compensate for that. But, still, we weren't allowed to do it because the concern was that perhaps the new test would measure antibodies that were not clinically relevant, that wouldn't work clinically. To me, that is amazing because, still, I think the clinical efficacy of a product is tested and is demonstrated in clinical studies, and the release testing that you do for these lots in essence should be consistency testing. DR. GOLDING: I am Basil Golding and, unfortunately, I don't know the details of this case because I wasn't directly involved, but I don't think we can resolve your question here without having the reviewers who were directly involved and have them be able to express their concerns. But maybe we can use this to just go over a few general principles. For looking at antibodies in a release test, one way is to do binding assays; another way is to do a neutralization assay. Now, the information that you get from a neutralization assay is probably closer to in vivo efficacy than the binding assay. So, there are different assays and they mean different things. The actual measles assay is one of the very few assays that we ask for as a final release test for IGIV. Originally the test was one of the tests because measles was a much more problem and it was very important to have anti-measles activity in the product. But now the test is more used as a marker of the product and its consistency over time and its comparability to previous similar products. So, that is a critical test, and the changing of the test, as Dr. Chang indicated, is a critical issue. Now to go into the details of your question I think is not reasonable because I don't have all the details at my fingertips to deal with that and, you know, we can have a conference and set up a time to discuss this in more detail with the reviewers. DR. HAYES: We need to move on to another question. DR. JONECKIS: Good morning again. This is a question for Dr. Chang. Thank you very much for your presentation. It was very pertinent and I think raised a lot of important issues. I just wanted to tell the audience we will make sure to try and get copies of the revised presentation out to them before the end of the day. I am sure many of you were curious about that. More specifically though, I saw that one of the items you listed was better characterization of starting material, particularly plasma and you listed a number of bullets. I am wondering if you could elaborate a little bit on that, particularly with respect to demographic and racial differences that you mentioned among donors and starting material, and if you could relate that to how that plays in the product characterization or finished product comparability, that would be helpful. Thank you. DR. CHANG: First, Chris, thank you for the comment, that nice comment you made. The concept that we are talking about, comparability, if the study material started differently, then the downstream process when you compare them with the same process when you use a different study material, it is very likely that you will see some differences in the process downstream. Now, can we identify some critical elements that we can characterize to provide better means later when you use a comparability study? When you compare a plasma-derived product to a recombinant product you know that the recombinant product, the study material, is a cell substrate which is well controlled. You know what you started with. But that is not the case for plasma-derivatives. I mentioned several possibilities that contribute to variations, such as the source plasma was recovered plasma; pool size. I am not very clear whether or not for the same product, the same company, do they always start with the same pool size. My understanding is that they may not. See, the different pool size, with so many donors, has variations. In normal production, if we control that it may not be a problem in terms of safety and efficacy, but when you talk about comparability that becomes problematic because you have a different study material. You have the same process. You are expecting some differences. Then the question is whether that difference is significant. Does that difference contribute to the safety and efficacy? So, those are the concerns. DR. WEINSTEIN: I would also like to make a comment here. I guess, you know, there are studies in the literature that at least have purported to show that starting plasma that has perhaps undergone some degree of proteolysis in the manufacture of Factor VIII may end up with a product that, after say solvent detergent treatment and heating, resulted in new antigenicity of the Factor VIII product. It has been alleged that it was the quality of the starting material that resulted in a product that had epitopes on it that were seen by the patients as being new entities resulting in a new antigenic site. You know, the issue has been raised whether or not one should monitor starting plasma for degrees of degradation. The issue is that potentially if you start out with a product that is more degraded in some way, the process which would normally give you a product that would not have a new antigen may result in a final product that does have some abnormality. What I am trying to say is the starting material certainly can have an effect on the end product here, and one should have specifications for that starting material that will lead to a final product that you can have confident will have specified properties. DR. HAYES: I hate to do it but we need to cut off the questioning at this point. We have used our ten minutes. I invite everybody to talk with the speakers during the coffee break, which we are going to reduce. Well, we can't be back in ten minutes; let's have a compromise and we will start the session, say, five minutes late and try to be back here at 10:15. [Brief recess] DR. WEINSTEIN: We would like to resume now. Our next speaker will be Michael Saunders, from Baxter, who will talk about the industry perspective on preclinical studies. Preclinical Studies Industry Perspective DR. SAUNDERS: Thank you, Dr. Weinstein and Dr. Hayes. I would also like to say thank you and congratulations to the meeting organizers and to the earlier speakers. I think the state has been well set for further discussion of the comparability issues. This morning I have the opportunity to share with you some views on preclinical studies for comparability. I also submit the following as a subtitle to my presentation because it truly is a clinician's perspective on the use and reliance on preclinical testing for product characterization and comparability. My background is in clinical research and for the past several years my primary responsibility has been leading the clinical development of the second generation recombinant hemoglobin program for Baxter. This is added on to many more years of previous experience in drug development. So, you will see interspersed throughout my presentation references to recombinant product development to illustrate certain points. As an outline for my presentation and for purposes of definition and establishing some common ground, I will talk a little bit about background and standards; suggest an approach for comparability preclinical studies; and share some examples to illustrate the points that I am trying to make. I will make a few comments on preclinical testing as it relates to clinical trials, and then make a few concluding remarks. If you haven't already guessed, I should tell you that I am fairly methodical in nature so, to being with some basics, we know that as demands grow so do manufacturing facilities and capabilities. It is also recognized that process development, enhancements and changes are inevitable as you gain experience with a process, and as you have additional desires for improvements in economic and efficiency in the process, those changes will occur. The third point that I am trying to make here is that we are constantly faced with the confounder of biologic variability in our evaluations and their results in our testing and their interpretation. It is also acknowledged that resource and time is required as testing increases as we go up the hierarchy of studies, and I do believe that it is a hierarchy that might also be the complexity of the studies. For standardization purposes and to try and submit our own idea on what maybe the definition of comparability should be, our group in Boulder, Colorado, the Hemoglobin Therapeutics Group, came up with this definition where we say that a newly manufactured biological product does not have any detrimental differences in identity, strength, quality, purity, potency, safety of effectiveness relative to the pre-change product. The product characteristics can differ between the two products as long as there is no adverse effect on the above parameters. I recognize that might be a bit provocative, but so it goes. There are also several important statements I would like to reiterate and emphasize in the 1996 guidelines. I think you have seen this already, but I think it is important to point it out again and paraphrase. There may be manufacturing changes which occur in the development of new biological products but which do not necessarily require the need to conduct clinical trials as long as comparability test data demonstrate to FDA that the product, after manufacturing change, is safe, pure, potent and effective. A couple of other points, FDA recognizes that it is an important consideration for product comparability is whether or not it is anticipated that any of these manufacturing changes will translate into significant changes in clinical safety or efficacy. FDA has also stated that depending on the type of in vitroassays and animal studies and the quality of the data, extensive clinical data demonstrating equivalence may not be necessary. I am hammering home a point that you will see develop as my presentation goes that by prospectively and carefully creating the preclinical development, you can actually avoid the need for comparability clinical trials. The types of process changes are multiple and their significance varies greatly. We have categorized these into the major, moderate and minor categories. Without going through each one of these in detail, I would just point out that changes in process, purification, manufacturing and formulation would be considered major. Duplicate processes changes in equipment and testing site would be moderate process changes. Minor changes would include things like analytical modifications, tightening specifications, etc. The most important point here is that the assessment of this categorization impacts the extent of testing necessary to demonstrate product characteristics and comparability. To begin with, the approach to comparability testing with preclinical studies involves also several categories. Broadly, these are analytical testing, bioassays and then, importantly, preclinical and animal studies. Analytical testing, primarily chemical and physical assays to characterize the structure, identity, consistency, purity, stability and solution properties, and may also have some bearing on safety parameters. The bioassays are functional tests to evaluate the activity, potency, integrity, mechanism of action and may also have some predictability for human biological effects. Finally, the animal studies looking at pharmacokinetics, pharmacodynamics and toxicity, it is important to recognize, and I think everyone agrees, that we may not need to repeat all of the toxicity studies with a process change. The characterization of these studies involves tests which provide an accurate description of the product, as you have heard from the speakers earlier this morning, and that they are consistent, reliable and predictive of the effects in humans. The important point of emphasis is that there needs to be investigation and evaluation of models to select those which provide the highest degrees of selectivity, sensitivity, specificity, precision and predictability. Focusing on the preclinical animal studies, over the last several years we have seen a lot of technology advances, with enhancements in analytical procedures, non-invasive hemodynamic monitoring such as impedance cardiography, as well as in the immunogenicity area where models using transgenic animals, knockout models, human gene replacement models, genetically manipulated models have significantly changed the landscape, and you will be hearing more about that from later speakers. With respect to the pharmacokinetic studies, we primarily want to look at the time to maximum concentration, the maximum concentration, the integrated area under the curve, the half-life, also all of this done with an understanding of the appropriateness of the use of parallel or crossover design trials. In pharmacodynamic studies we want to evaluate the pharmacologic and physiologic effects, as well as the overall effectiveness, potency and mechanism of action. In toxicity studies for comparability, this is built upon the safety profile of the existing product, and it may be significantly influenced by the presence of a narrow therapeutic range or specific safety concerns. Experience with the product and the process change should also lead to an anticipated toxicity profile, as well as recognition of potential toxic impurities, and special consideration for immunogenicity issues. So, the overall approach can be summarized by saying it is based upon product characterization and anticipated effects of process change which can predict the expected identity, purity, safety and effectiveness as they impact on the post-change product. As I have emphasized, it is important to select the optimal and most appropriate models for the preclinical comparability testing, with the goal to do enough to establish comparability but not necessarily to create any obstacles which unnecessarily delay the development of process changes of important new products. So, with reference to the previously given definition, the ultimate goal is to demonstrate comparability as no clinically significant adverse experiences between the pre- and post-change product. Also, we recommend prospective determination of a clinically significant change which generally involves development of a decision tree. Possible examples might be threshold of greater than one grade severity of an adverse effect that is observed; a greater than two standard deviation change of a laboratory result. It is also recognized that often this needs to be done on a case by case basis. Let me go to a few examples to try to illustrate some of the points I am raising. Here is one from the literature of a platelet antagonist, XV459, fibrinogen receptor antagonist for glycoprotein 2B3A antagonist, which is in development. It is a potent selective product. The LC-MS assay has been used for quantitation of levels in guinea pig plasma. They have also noticed that there has been similar binding in guinea pigs and humans, as well similar dissociation rate constants, and 50 percent inhibitory concentrations for platelet aggregation assays, therefore, one might conclude that the guinea pig is an appropriate model of PK and distribution studies of this product and, therefore, could be a reasonable model for comparability studies. The next major area of examples that I would like to explore with you has to do with the area I am most familiar with, and that is the group I have been working with for the past several years in hemoglobin therapeutics. While I will be talking a lot about the recombinant product that we currently have in development, I should also point out that much of this experience is really built upon a human-derived product, DCLHb that you may be familiar with. It was really the major point of emphasis of the Baxter Hemoglobin Therapeutics group over the past decade. So, in principle and in action, our group has advocated a strong relationship between preclinical and clinical study development. We have done thorough and exhaustive explorations and evaluations of preclinical models to select those which are optimal. With that in mind, we designed the preclinical studies with clinical input and with an intent to mimic the planned clinical trial settings. The clinical trials are then built upon those results coming out of those preclinical trials, and it helps to direct subsequent study designs and the endpoint selection, with the hope that we can develop correlations between the outcomes of the preclinical studies and the clinical trials to try to make a match. We believe that this groundwork and collaboration allows for predictive comparability assessments. In order to be able to make some sense of these examples, I need to give you a little bit of background with respect to the recombinant hemoglobin development and some of the issues that we have encountered. Recombinant-based hemoglobin is to be administered intravenously, and should effectively transport oxygen to tissues. It has the potential of allowing reduction or avoidance of transfusions in the surgical setting, and may also have the potential of being a unique resuscitation fluid in trauma settings. One of the key issues we encountered with respect to the hemoglobin development was a finding of an effect on nitric oxide. Nitric oxide is an important clinical messenger in the body that, among other things, causes smooth muscle relaxation. We know that the heme moiety of the hemoglobin binds nitric oxide, and this binding leads to unopposed smooth muscle contraction which produces several clinical effects. Among those are arterial smooth muscle contraction resulting in hypertension. Now, let me step back a moment and characterize the two groups that we are talking about, first and second generation hemoglobin. The first generation hemoglobin we define as those hemoglobins which have a reactivity rate with nitric oxide comparable to native human hemoglobin, as opposed to the second generation hemoglobin where we have modified the molecule and modified and significantly reduced nitric oxide reactivity. To bring this to some clinical bearing, we did encounter clinical safety concerns with the first generation hemoglobins that Baxter had in development. As a consequence, we stopped the development of those first generation hemoglobins three and a half years ago--four years ago now--and focused the development on second generation hemoglobin. We identified nine features of the first generation hemoglobin which we desired to change, and we conducted a plethora of biology studies to fully characterize the second generation hemoglobin for an investigational drug status, as well as to distinguish it clearly from the first generation hemoglobin. We utilized cumulative experience, literature and consultations to identify the most appropriate models which were optimal for the biologic evaluations. With respect to the background that I have just presented, and referring back to the blood pressure effects, we noted that there was increased blood pressure seen with the first generation hemoglobin in man, as well as in animal studies. It represented a safety concern in a number of patients, and the second hemoglobin was targeted to overcome the blood pressure among other effects by reducing the nitric oxide reactivity. We put together a ran hemodynamics model where we found that the blood pressure response corresponded to nitric oxide kinetics. I will demonstrate that to you in a moment. We anticipate that it will correspond to the human blood pressure response as well. We believe this is a very predictive model for the blood pressure effect, correlating with the nitric oxide kinetics, therefore, any process changes which should occur with the second generation hemoglobin development that might affect nitric oxide reactivity could be assessed by this model, and we anticipate being able to predict what the blood pressure effects might be. Here is the demonstration with a series of hemoglobin molecules which were synthesized having a variety of nitric oxide binding kinetics. Here is the binding constant for nitric oxide on the lower axis. On the upper axis is the blood pressure response. You see a linear correlation, that the reduction in nitric oxide reactivity corresponded with a reduction in the blood pressure response with the increase in mean arterial blood pressure. The second example that I would like to share with you with respect to the hemoglobin is related to the pharmacokinetics and half-life. Second generation recombinant hemoglobin is a polymerized product to increase the size with the intent to increase half-life. We found that pharmacokinetic determinations in rat model did make a correlation with molecular size of the product. So, subsequent potential process changes which might affect molecular size distribution could be effectively assessed by this rate PK model, and we are looking forward to demonstrating that these correlate with the human results. I promised that I would give you a few relevant comments toward clinical trials, and with apologies to Dylan Thomas, I want to make the point that a decision to go into a clinical trial should not be taken capriciously. There are bioethical considerations for either doing or not doing clinical trials. Subjecting patients or volunteers to procedures in an unnecessary or avoidable clinical trial can represent ethical issues. There are certainly resource and time considerations that go into performing a clinical trial and, therefore, we would advocate reserving clinical trials for those instances where the preclinical investigations fail. When a clinical trial is truly deemed to be necessary, we would utilize those preclinical results as a guide to focus the clinical trial design and endpoints. So, the answer for the decision to conduct a comparability clinical trial may not be to do a clinical trial. It may, in fact, be to do a better preclinical profile ahead of time. We would urge that we enhance the preclinical testing for selection of the best predictive models, and that should be done in concert with FDA input and collaboration to exhaust all of the preclinical study alternatives. Finally, I would leave you with the following thought, summarized in this statement: Preclinical testing should be performed to identify the most appropriate sensitive and predictive models for product characterization and for evaluating the effects of process and manufacturing changes in order to avoid the conduct of unnecessary clinical comparability trials. Thank you. DR. HAYES: I have one housekeeping announcement to make. Because this meeting is being transcribed, you can be looking for that on the CBER web site in the next few weeks. Our next speaker is Basil Golding, M.D. Dr. Golding will be discussing the relevance of animal modeling in predicting immunogenicity. Relevance of Animal Modeling in Predicting Immunogenicity DR. GOLDING: I am going to be talking about the relevance of immunogenicity in animal models and testing for immunogenicity for products. Just an outline of my talk, I call it a road map--defining the problem. Animal models, are they useful? Differences between the animal and human immune system, and new approaches for testing immunogenicity of protein. When a product is given, such as a plasma-derived product, and antibodies develop--the main problem we are talking about in terms of immune response is antigenicity; in terms of outcome are antibodies. So, antibodies can interfere with the product's safety and efficacy. So, you may get inhibition of product function and this could be either due to binding of the antibody to the functional site of the protein, or you can get binding of the antibody to a non-functional site but, because there is increased clearance of the protein and alteration of pharmacokinetics, that will, in effect, reduce the function. Antibodies can be formed which cross-react with self causing autoimmune reactions. The example given by Christian Eckers probably falls in this category. Then, immune complexes, regarding the EPO effect and the effect of antibodies with EPO, are associated with aplasia. So, immune complexes, in other words, antibodies binding to the product forming a complex, can cause adverse reactions, and these are the type of reactions that can be seen as arthritis or even kidney disease and are similar to serum sickness. In rare instances, but of important clinical effect, is the fact that some proteins that we use can induce IgE antibodies and this can cause allergic reactions and even anaphylaxis. The example that comes to mind is selective IgA deficient individuals who are given immune globulins who can develop anti-IgA and this can result in anaphylaxis. Animal models have been used in the past. I don't want to give the impression that I don't think there is a place for animal models but they do have limited usefulness because immune responses may be different from humans. In particular, I am going to go into some detail later of this, there are differences at the level of the MHC Class II genes, the TCR repertoire and the antibody repertoire. This is for the future, and already strides have been made in this direction. It has been possible to take mice and introduce human genes into these mice, and humanize these mice so that they now express some of these human genes and, therefore, their immune systems are beginning to look like human immune systems. So, mice have been made which express three of the four IgG subclasses, and it has been shown that you can generate antibody responses in these mice. So, this may be an important model to use for testing immunogenicity of proteins that are going to be given to humans. In addition, mice have been engineered that express human MHC Class II genes, and I will explain the importance of this in a second. So, the classical testing of a protein use to see if it had gained some immunogenicity during production was to use rabbits or mice, and depending on your testing system and the product, you may use these animals that have targeted gene deficiencies. For example, if you wanted to test a new Factor VIII, you may want to use mice that are Factor VIII deficient. The basic protocol was to immunize with a native protein, to immunize with the innovator protein and to ask the question whether the innovator protein generated antibodies that were different from the native protein as judged by absorption profiles to the native protein, and also to ask whether the innovator protein generated antibodies that inhibited action of the protein. Now, I think there still is a place for this and, at a minimum, even though the immune system of these animals is very different from humans and can in no way predict what type of responses you are going to get in human, the fact that you do get a positive result when you are using an innovator protein would suggest that there is something different in structure in the new protein compared to the old protein. Just to go through some immunological concepts because what I am going to do in the next few slides is introduce you to some possible testing that could be done in the future, making use of human cells and human responses rather than relying on animal responses, and I must warn you that when I presented it to my colleagues I saw body language that told me that, hey, you're going to be way over, but I think I made a particular effort to change the slide and to try and explain this in a way that would also be digestible by non-immunologists. We will see if I succeeded. The first point to make is that antibody responses to proteins require T cell help, and that this help is related to MHC Class II expression and T cell receptor repertoire. I will explain this in a moment. Because these genes are different in different animal species and are, in fact, different from one human to another and any outbred species, you cannot predict from a response in one species that there is going to be a response to the same protein in another species. The same goes for humans. One human may respond and another human may not respond. This is a slide that is very critical to the whole presentation of the next few slides. I indicated to you that there is a critical cell, called the T-helper cell, that is required to respond in order to provide--it is called a T-helper cell because it helps, among other things, antibody production. So, this T-helper cell has on its surface a T cell receptor. This T cell receptor recognizes antigen in the context of MHC antigens. So, there are two cells here that are interacting. One is a T-helper cell and the other cell is an antigen presenting cell. The antigen presenting cell has taken up the antigen from the outside, has processed it, and has expressed a small peptide, usually eight or nine amino acids, in a groove of the MHC Class II and the T cell receptor is seeing this complex. Now, there is a tremendous diversity in the system to allow us to respond to all the multiple types of antigens we see in the environment. So, the T cell receptor consists of an alpha chain and a beta chain, and each chain is along a gene, the T cell receptor gene, which is very analogous to the immune globulin gene and has many genes. So, it has variable regions, V regions and J regions, and these can combine in various combinations in the alpha chain and the beta chain to give an alpha/beta chain receptor. The diversity here is mind boggling. In the human it is approximately three million different T cell receptors that could be formed by this molecular rearrangement. These genes are different in different humans and in different species. So, this T cell receptor repertoire is very different depending on whether you are a human or a rabbit or a mouse. In addition to that, the MHC Class II genes represent in the human at least 100 alleles per gene. Again, you have tremendous diversity over here. Because of this tremendous diversity, it is very difficult to predict whether a human response is going to occur if you see a response in a mouse or a rabbit. But the question is how could we use our knowledge of this system to devise a better way of screening proteins to see if they are going to be immunogenic or not? I should just point out that in immune responses the T-helper cell plays a central role, and when it is stimulated by the antigen presenting cell that I showed on a previous slide you get various things going on between the T-helper cells and the B cells, and then you get antibody production. So for proteins the T-helper cell is critical. Without first activating the T-helper cell you are not going to get antibody. This is the paper that I am going to be talking about, and I have only seen one paper to this effect but it is logical and I think is something that I would like to draw to your attention because I think more people should try and replicate these results, and maybe this is a mechanism for testing proteins that are going to be given to humans. The basic idea is that you could use naive human T cell responses to predict T cell epitopes in an antigen, in other words, to predict those peptides within an antigen that are going to stimulate the immune system, particularly T cell. If you want to look up the reference--I left out the data, this was probably in the year 2000. So, this looks like a very complicated slide, and I actually lifted it from Dr. Estell's presentation to the FDA and it is on the FDA web site. Although it looks very complicated, it is actually very simple. You are taking peripheral blood from a patient, you take that blood and you purify dendritic cells, which are the professional antigen presenting cells, and you also purify the helper cells. You then add the antigen. So, these dendritic cells will process the antigen, present it on the surface in association with MHC Class II and stimulate the CD4 T cells to divide provided, of course, that the T cells contain within their repertoire the T cell receptor that will recognize the peptide that is being presented by the dendritic cell. You can do this in microtiter plates and you can measure T cell stimulation by looking at thymidine uptake, which is a standard assay. What you can also do once you have seen a response to the total protein, you can make overlapping peptides from the protein and you can actually determine which are the epitopes which stimulated the T cells. You can do this from a large number of human individuals so that you get a good representation of the human population in terms of the MHC Class II that is used and the T cell repertoire. What they did, they used human peripheral blood mononuclear cells. They used an inactivated enzyme as the protein just for proof of concept, and they looked for proliferation of the T cells, and they called a response a stimulation index greater than twice background, and a weak response was slightly above background, and no response was same as background. This is just an example of the type of experiments that they did. I haven't got time to go over the whole paper, but when they took the native protein and used the system they found that over 50 percent of the T cells from normal individuals responded to the protein. So, obviously, this would be a problem if you wanted to give this protein as a treatment. You know, there was a weak response and no response in about 20 percent of the normal individuals. What they did, they identified, using peptides, which part of the protein was actually inducing the T cell response. Then they genetically engineered the protein so that those peptides, based on all kinds of algorithms that relate to the binding of the peptides to the MHC and so on, they could change them by just doing point mutations, and then test the protein again in the in vitrosystem to see if now the protein had lost its immunogenicity. What you see, for example, in this variant is that now most of the individuals are no longer responding to the protein by making a single mutation in the protein. So, this approach not only allows you to detect whether a protein is going to induce an immune response, but it allows you to engineer the protein in subtle ways so that it will no longer be immunogenic. Obviously, this has to be don e very carefully and even single point mutations can have a drastic effect on protein folding and glycosylation, and all kinds of things. So, all those other concerns have to be looked at before making the changes. What they were also able to do, and this is getting back to the situation where you could actually take a mouse and humanize the mouse and use that to test your protein, so if you had a protein that you wanted to use in patients and you did those previous studies in human T cells, and you knew from the human responses in vitro which DR alleles were involved in the response--the high responders were people that expressed these alleles--you could then make a mouse that expresses the human HMC Class II, and some people have already done this using different human alleles, and test in vivo in the mouse whether that mouse responds at the T cell level, and whether that mouse makes antibodies to the protein. In summary, this epitope mapping assay determines relevant priming epitopes. The epitopes can be modified to redu