FOOD AND DRUG ADMINISTRATION
TRANSMISSIBLE SPONGIFORM ENCEPHALOPATHIES
October 31, 2005
Holiday Inn Bethesda
CASET Associates, Ltd.
10201 Lee Highway, Suite 180
Fairfax, Virginia 22030
Freas, William, PhD
Priola, Suzette A, PhD
Bias, Val D
Creekmore, Lynn H. DVM
Hogan, R. Nick, MD, PhD
Telling, Glenn C, PhD
Kranitz, Florence J.
Johnson, Richard T, MD
Salman, Mo D. DVM, PhD
Sejvar, James J., MD
James R. Allen, MD, MPH
Paul Brown, MD
David C. Bolton, PhD
David Gaylor, PhD
Michael Geschwind, MD, PhD
Bernardino Ghetti, MD
Susan F. Leitman, MD
James W. Lillard, Jr, PhD, MBA
Arthur W. Bracey, MD
Administrative Remarks - William Freas, PhD, CBER,
Executive Secretary, TSEAC 1
Recognition of Committee Service - Jesse L. Goodman,
MD, MPH, Director, CBER 9
Opening Remarks - Suzette Priola, PhD, NIAID, NIH,
Chairperson, TSEAC 9
Update on US and worldwide BSE status - Lisa Ferguson,
DVM, APHIS, USDA 9
Scientific issues in evaluating products intended to
decontaminate surgical instruments exposed to TSE agents:
discussion of a recent FDA Device Panel 16
Topic 1: Progress Report on FDA's Risk Assessment for
Potential Exposure to Variant Creutzfeldt-Jakob Disease
in Human Plasma-Derived Antihemophilic Factor (FVIII)
Introduction and Questions to the Committee - Dorothy
Scott, MD, OBRR, CBER 26
Variant CJD risk associated with human plasma
derivatives: Introduction and overview of risk model
Steven Anderson, PhD, OBE. CBER 32
Update on vCJD in UK and other countries: estimates of
Richard Knight, MD UK Director, CJD Surveillance Unit
Azra C. Ghani, PhD, London School of Hygiene and
Tropical Medicine 83
Modeling risk of vCJD in US donors - residual risk
and efficiency of donor deferral - Alan Williams, PhD,
OBRR, CBER 104
VCJD infectivity of plasma - estimates from
experimental models - David Asher, MD, OBRR, CBER 119
Review of TSE clearance in FVIII product manufacturing
Dorothy Scott, MD, OBRR,CBER 128
FVIII product usage in clinical settings - Mark
Weinstein, PhD, OBRR, CBER 135
Open Public Hearing 146
Committee discussion and recommendations 164
Topic 2: Labeling Claims for Filters Intended to
Remove TSE Infectivity from Blood Components
Prospects for reduction or removal of TSE agent
infectivity from blood components by filtration and
criteria for allowing claims: Introduction
Jaroslav Vostal, MD, PhD, OBRR, CBER 219
Evaluation of prion reduction filters
Mark Turner, MB, ChB, PhD, FCRP(Lond) University of
Performance of Pall Corporation Leukoreduction filters
on TSE infectivity of blood components: experimental
studies and European experience - Dr. Sam Coker,
Pall Corporation 242
Selection and performance of resin-bound ligands for
removal of TSE infectivity from plasma - Robert
Rohwer, PhD, PRDT (with ProMetic and ARC) Rockville, MD 251
Other industry/academic filter chromatography developer
Dr. Ralph Zahn, CEO, Alicon AG, Schlieren, Switzerland 266
Open Public Hearing 275
Committee discursion and recommendations 275
P R O C E E D I N G S 8:06 AM
Administrative Remarks William Freas, Ph.D., CBER, Executive Secretary, TSEAC
DR. FREAS: Ms. Chairperson, members of the Committee, invited guests, consultants and members of the public, I would like to welcome all of you to this our 18th meeting of the Transmissible Spongiform Encephalopathies Advisory Committee.
I am Bill Freas. I will the Executive Secretary for today's session. The entire meeting today is open to the public.
At this time I would like to go around the head table and introduce to the public the members who are seated at the table.
Will the members please raise their hands as their name is called.
In the first chair on the right side of the room, that is the audience's right is Dr. David Bolton, head, Laboratory of Molecular Structure and Function, New York State Institute for Basic Research.
Next is Dr. Richard Johnson, professor of neurology, Johns Hopkins University.
Next is Dr. Glenn Telling, associate professor, Department of Microbiology, University of Kentucky.
Next is Dr. Lynn Creekmore, staff veterinarian, APHIS Veterinary Services, US Department of Agriculture.
Next is Dr. James Lillard, associate professor of microbiology, Morehouse School of Medicine.
Next is Dr. James Sejvar, medical epidemiologist, Division of Viral and Rickettsial Diseases, Center for Disease Control and Prevention.
Next is Dr. Nick Hogan, assistant professor of ophthalmology, University of Texas, Southwestern Medical School.
In front of the podium is Mr. Val Bias, Co-Chairman, Blood Safety Working Group, National Hemophilia Foundation, Oakland, California.
Next is Dr. James Allen. Dr. Allen is Chair of FDA's Blood Products Advisory Committee. He is, also, president and CEO of the American Social Health Association.
Next is the Chair of this Committee, Dr. Suzette Priola, investigator, Laboratory of Persistent and Viral Diseases, Rocky Mountain Laboratories.
Next is Dr. Arthur Bracey. Dr. Bracey will be serving as a non-voting consultant today. He is Associate Chief of Pathology, St. luke's Hospital, Houston, Texas.
Next is our consumer representative, Mrs. Florence Kranitz. She is President of the CJD Foundation, Akron, Ohio.
Next is Dr. Michael Geschwind, assistant professor of neurology, University of California, San Francisco Medical Center.
Next is Dr. Susan Leitman, Deputy Chief, Department of Transfusion Medicine, National Institutes of Health.
Next is Dr. David Gaylor, President, Gaylor and Associates, Eureka S[ring, Arkansas.
Next is Dr. Bernardino Ghetti, distinguished professor, Director, Indiana Alzheimer's Disease Center, Indiana University, School of Medicine.
Next is Dr. Mo Salman, professor and Director, Animal Population Health Institute, College of Veterinary Medicine and Biomedical Sciences, Colorado State University.
The chair at the end of the end of the table will soon be filled by Dr. Paul Brown. He is a consultant from Bethesda, Maryland.
Our non-voting industry representative could not attend today's meeting due to a medical emergency. Our efforts to recruit a replacement in time for this meeting were not successful.
I would like to thank the members for attending this morning. I, also, have one announcement to make. Dr. Alan Jenny passed away last Thursday night. Dr. Jenny served as a member of this Committee since September 2004 and was a consultant and speaker at many of our meetings prior to his service on the Committee. Dr. Jenny was a pathologist at the National Veterinary Services Laboratory for the US Department of Agriculture in Ames, Iowa. He was well known and respected for his research on investigative studies on numerous livestock diseases in the United States.
He was a kind and gentle man. He was a wonderful person. Alan Jenny will certainly be missed by his family, this Committee and many people the world over.
I would like to ask for a moment of silence in his honor.
I would now like to read into the record the conflict of interest statement for this meeting. Some of you might think you are in the wrong room. This has just recently been revised by our attorneys.
The Food and Drug Administration is convening today's meeting of the Transmissible Spongiform Encephalopathies Advisory Committee under the authority of the Federal Advisory Committee Act of 1972.
All members of the Committee are special government employees or regular federal employees from other agencies and are subject to federal conflict of interest laws and regulations.
The following information on the status of this Committee's compliance with the federal conflict of interest laws including but not limited to 18 US Code, Section 208 and 21 US Code, Section 355(n)(4) is being provided to the participants in today's meeting and to the public.
FDA has determined that members of this Committee are in compliance with federal ethics and conflict of interest laws including but not limited to 18 US Code, Section 208 and 21 US Code Section 355(n)(4). Under 18 US Code, Section 208 applicable to all government agencies and 21 US Code, Section 355(n)(4) applicable to certain FDA committees, Congress has authorized FDA to grant waivers to special government employees who have financial conflicts when it is determined that the agency's need for the particular individual's service outweighs his or her potential financial conflict of interest, Section 208, and when participation is necessary to afford essential expertise, Section 355.
Members of the Committee are special government employees including consultants appointed as temporary voting members. Committee members have been screened for potential conflicts of interest of their own as well as those imputed to them including those of their employer, spouse or minor child.
Related to the discussions of progress in the development of a risk assessment model for vCJD in human plasma-derived Factor 8 products and discussions of the reductions of TSE removal, TSE agent infectivity from blood components by filtration, these interests may include investments, consulting, expert witness testimony, contracts, grants, CRADAs, teaching, speaking, writing, patents and royalties and primary employment.
Today's agenda topics are considered general matters discussions. In accordance with 18 US Code, Section 208(b)(3) general waivers have been granted to all Advisory Committee members including Val Bias, Lynn Creekmore, Nick Hogan and Richard Johnson, Florence Kranitz, Susan Priola, Mo Salman, James Sejvar and Glenn Telling.
In addition, a general matters waiver has been granted to Paul Brown. A copy of the written waiver statements may be obtained by submitting a written request to the agency's Freedom of Information Foundation Office, Room 12A30 of the Parklawn Building.
With regard to FDA's guest speakers the agency has determined that information provided by these speakers is essential. The following information is made public to allow the audience to objectively evaluate any presentation and/or comment made by these speakers.
Dr. Lisa Ferguson is employed by the USDA in Hyattsville, Maryland. Dr. Azra Ghani is Director, UK, CJD Foundation, London School of Hygiene and Tropical Medicine, England. Dr. Richard Knight is Director, UK CJD Surveillance Unit, Western General Hospital, Edinburgh, Scotland. Dr. Marc Turner is Clinical Director, Scottish National Blood Transfusion Service, Edinburgh, Scotland.
As guest speakers they will not participate in Committee deliberations nor will they have a vote. They are more than welcome to answer questions though of the Committee and we hope they stick around to answer questions of the Committee.
In addition there may be regulated industry and other outside organizations making presentations. These speakers have financial interests associated with their employer and with other regulated firms.
The FDA asks in the interest of fairness that they address any current or previous financial involvement with any firm whose products they which to comment upon.
These individuals from industry were not screened by the FDA for conflicts of interest. The conflict of interest statement will be available for review at the registration table. We would like to remind members that if the discussions involve any products or firms not already on the agenda for which an FDA participant has a personal or imputed financial interest the participants need to exclude themselves from such an involvement and their exclusion will be noted on the record.
FDA encourages all participants to advise the Committee of any financial relationship you may have with firms that could be affected by the Committee's discussions.
That ends the conflict of interest statement.
Before I turn the microphone over to our Chair, I would like to request that everybody check their cell phones and please place either in the silent mode or turn them off so that it won't be so disruptive to the meeting.
Dr. Priola, I turn the meeting over to you.
DR. PRIOLA: Okay, everybody. There are lots and lots of questions and the first topic is basically to assess the risk assessment model for variant CJD exposure. They want us to refine the input parameters and help them to define input parameters. They just want advice. I am going to hold questions after each speaker to just technical questions and only one or two of those. We can ask after each group of speakers all the questions that we couldn't get to in that first part and also ask some questions in the discussion period.
Agenda Item: Recognition of Committee Service Jesse L. Goodman, MD, MPH, Director, CBER
DR. GOODMAN: Okay, I have the opportunity to recognize the services of three very distinguished individuals here and I would appreciate if they would each come up, Richard Johnson who is of course a distinguished neurologist. Should I do it one by one?
Okay, so, okay, it is the picture again. Dr. Johnson, thanks so much.
Then we have Dr. Arthur Bracey and I will try to do better with Arthur. Thank you, Dr. Bracey.
Okay, and Dr. Bracey, by the way is going on an HHS committee or has already done that. So, thank you for that, too, and then Sue Priola who obviously in addition to being a member has been the Chair of the Committee for the last 2 years.
Thanks, Dr. Priola.
Agenda Itedm: Opening Remarks Suzette Priola, PhD, NIAID, NIH, Chairperson TSEAC
DR. PRIOLA: Okay, we are ready for the presentations. The first is Dr. Ferguson from USDA.
Agenda Item: Informational Presentations Update on US and Worldwide BSE Status Lisa Ferguson, DVM, APHIS, USDA
DR. FERGUSON: Good morning. I just have a few brief slides. You probably get tired of hearing me every time you get together. Anyway I thought I would go over a little bit an update both on what we are doing for BSE here in the US, also, a brief summary of data from Europe as that is where the vast majority of BSE cases have occurred worldwide.
Total BSE cases worldwide at this point greater than 189,000 cases worldwide with a couple of important points to remember about that. First of all more than 96 percent of those cases have been in the UK but more importantly more than 89 percent of those have really occurred in 1996 and before.
If you are really interested in specific information on individual countries and cases reported by year I have listed the OIE web site and OIE is the world organization for animal health and as countries report cases OIE has a specific table for BSE status and a calculated apparent incidence rate.
Let us talk a bit about what the Europeans have done. They have had very intensive surveillance since 2001. So, we have very good numbers for comparison within the EU. One point though, as of mid to late 2004 there used to be 15 member states in the European Union and now it is 25. So, that has changed the numbers somewhat with the addition of 10 new member states to the EU, but in 2004 they did a total of more than 11 million tests.
Of those about 9.5 million were apparently healthy animals greater than 30 months of age at slaughter. One point five million are what they call risk animals. These would be the same as our targeted population in the US. These are animals with some type of clinical signs, falling stock, non-ambulatory animals.
Out of all of those tests in 2004 they had a total of 865 positive cases and this is a decrease in the number of cases by about 37 percent and overall apparent prevalence also was decreased by about 38 percent.
If you look at the numbers for the year before similarly in 2003 they had about a 35 percent decrease form 2002. So, you see things continuing to decline in Europe which is very encouraging. These reductions and also the increasing age of positive cases really indicates the success that they have had in Europe with their control measures. So, those measures that they have imposed increasingly stringently do seem to be working.
If you look at an analysis by year of birth the positive cases and we are assuming that most exposure occurs in that first year of an animal's life, at this point with the 2004 data and granted this can change; with long incubation periods you can still see increasing cases, but there do appear to be exposure peaks pretty well defined in certain member states, obviously the UK but outside of that. The data in 2004 show that France and Ireland have an exposure peak about 1995, Germany, Belgium, Italy and Netherlands about 1996. So, it would be interesting to see is that stays the same as they get more data over the next couple of years.
There, also, is one note. They have increased their TSE monitoring in small ruminants and are doing extensive analysis with that data from sheep and goats and they did find one goat positive for BSE. It is an animal that was actually slaughtered in 2002 with extensive analysis of how to define that case.
So, let us jump a bit to the US and I think as everybody knows we have been doing an enhanced surveillance program since June 2004. I would note we have done active surveillance for BSE in the US since 1990 but in response to the cases in Canada we ramped up our surveillance a drastic amount beginning in June 2004, and our goal of our enhanced program was to get as many samples as possible from the targeted population in a 12-to-18-month period. So, we were trying to go out there and get all the samples we could from that population where you are most likely to find disease present.
Our targeted population is those animals that have some type of a clinical sign that could be considered consistent with BSE. So, this is a pretty broad definition. It includes animals that have classic clinical signs of BSE, animals that have central nervous system signs, other types of clinical signs that could be considered consistent, animals that are non-ambulatory, animals that have died for unexplained reasons and then we are also sampling animals that are condemned on antemortem inspection at slaughter.
This talks a bit about our assumptions. We are looking in this targeted population as we have been since 1990 with the assumption that if we can't find disease there in that population where we are most likely to find it if it is present then we are even less likely to find it in a broader cattle population or in a non-targeted population, but we can use the data that we get from this targeted approach and extrapolate that as we try to estimate prevalence in the broader cattle population.
Then just a summary of the targeted populations where we are getting samples and we are working with diagnostic labs and public health labs around the country also as they get neuro cases or rabies cases. We ask them either to forward those samples on to us or to work with us to obtain those.
This actually is previous surveillance. I didn't put in all since 1990. So, you can see where we have been. Beginning in about 2002 we were looking at between 19,000 and 20,000 samples a year and then in our enhanced program since June 3, 2004, we have looked at more than 510,000 samples. So, we have had very good success in obtaining access to the populations that we need and getting these samples.
Out of all of this we had one positive case in June 2005, a case found in Texas. This was a 12-year-old animal, a Brahma-cross, so a Bos indicus type breed. We did an extensive investigation. Our FDA colleagues did an extensive feed investigation. Clearly with a 12-year-old animal it is sometimes a challenge both on the feed history and trace backs and trace forwards. We were a bit constrained by records on the premise of origin but we attempted to trace birth cohorts which we defined very broadly from this animal due to the lack of records, traced those animals, and we actually euthanized 68 animals total and tested those as birth cohorts.
Our FDA colleagues in their food investigation really didn't find something specific that they could pinpoint and say, "Here is a likely source of exposure," but then that is not entirely unexpected again especially if you are working with 12-year-old records.
As we go through my program we are continuing to evaluate where we are and to ensure that we are accessing appropriate representation around the country and we are getting pretty good geographic distribution.
The vast majority of our samples tend to be from dead stock and non-ambulatory animals. This is entirely what we expected and really what we wanted.
We are getting good representation from all the different collection sites but clearly the majority of these samples that we are getting are from animal disposal facilities, rendering facilities, what we call 3D, 4D plants or salvage slaughter plants, dead stock. This is where our targeted population shows up. So, that is where we are getting those samples.
We will be doing a very detailed analysis when we complete our program. We are now in the 18th month of this and we hope to do an analysis and have that out for public release very shortly thereafter after we are done completely with the program and I just realized I didn't actually put anything in about our Canadian colleagues and their situation. I know there is lots of interest in what the Canadians are doing and have done. They have a total of four cases and the last cases were those two in December 2004, January 2005.
They have ramped up their surveillance similar to what we have done, also, doing targeted surveillance and at this point in 2005, they are up to close to 40,000 samples.
So,they are, also, having very good success with obtaining the samples that they need and with no further positives since those four in an apparent cluster out there in Alberta, and in line with Sue's push to keep us on time if you are looking for more detail, more info we do try to post everything that we can on our web site including updated numbers and you can always look there to see the latest release from us.
DR. PRIOLA: You will be around for the next few hours?
DR. FERGUSON; Actually I need to leave at the end of the morning.
DR. PRIOLA: Our second international presentation is by Dr. Sheila Murphy and she is going to update on a new FDA Device Panel that discusses scientific issues in evaluating decontamination products.
Agenda Item: Scientific Issues in Evaluating Products Intended to Decontaminate Surgical Instruments Exposed to TSE Agents: Discussion of a Recent FDA Device Panel Sheila Murphey, mD, CDRH
DR. MURPHY: Good morning. Thank you for your interest in our committee. The Center for Devices and Radiologic Health has asked one of our advisory committees to address the scientific issues related to TSE screening of products intended to decontaminate surgical instruments. This was our general hospital and personal uses devices panel. The advisory committee is asked to address the issues surrounding the evaluation of products or processes intended to reduce the viability of CJD transmissible agents on contaminated surgical instruments.
We believed that we needed more guidance on these issues. There are a number of scientific issues addressing the removal of TSE from instrument proxies is increasing in the literature.
Public interest about CJD and variant CJD and its potential for causing disease in the United States is in fact increasing and DAGID believes that it should prepare for the possibility that products or processes intended to reduce TSE infectivity on surgical instruments will be submitted to FDA for premarket evaluation.
There were a number of presentations at the scientific panel. I am going to give you just some quick background on the four from FDA. First, Dr. Elaine Mayhall presented a general overview of transmissible spongiform encephalopathy concentrating primarily on iatrogenic transmission.
There are only six reported cases in the literature and four of those are considered possible, not probable. None have been reported since 1980. Small epidemiologic studies of the potential association between surgery and risk of CJD have not resulted in consistent associations, some of them positive, some negative. There are reports of patients exposed to instruments used in the care of patients with Jakob-Creutzfeldt disease where the recognition of that exposure did not occur until after the instruments had routinely processed and used on other patients.
There have been at the present time no reports of transmission of disease related to these exposures. TSEs of course in man are rare diseases. Iatrogenic transmission has been quite rare. We have a number of clinical procedures now in place to reduce this risk but the question is could we do better and how should we evaluate the possibility that products could in fact do this. I was asked to present the experimental design issues on this point. We looked at such things as the types of prions and animal models which could be used to possibly validate such studies, the problems with such validation which particularly relate to the small, simple instrument proxies on the one hand and the reality of complex used surgical instruments on the other hand with their hard-to-clean shapes and the realities that that poses.
This, of course, is a huge surface which is obviously dirty after initial cleaning. There are other issues such as the proper design of large animal studies which would be appropriate not so much to the making of a scientific point but rather to the validation of a commercial product with a degree of statistical significance that would be appropriate for that which is a little bit different from scientific investigation.
So, we asked the committee to consider the risk/benefit ratio in possibly approving such products. the benefit of course of having a product that could in fact reduce TSE transmission by contaminated surgical instruments would be to further reduce risk to patients.
Are there risks involved? These would be primarily behavior risks, the creation of a false sense of security about the possibility of transmitting TSE by contaminated instruments perhaps leading to the failure to adequately follow the practices currently recommended to reduce that risk for surgical instruments.
Is the potential benefit of approving processes or products which could reduce risk significant? Does this benefit outweigh the potential risks?
Dr. Estelle Russell-Cohen from the Division of Biostatistics in the Office of Biometrics and Surveillance gave a presentation on the statistical considerations of study design for product evaluation. The studies of course must support the intended use claim.
Again, this is a little different from scientific investigation. The labeling instructions for product use do need to be supported by the study design and there needs to be a reasonable degree of statistical confidence in those results.
Good study design of course would include removing systematic error and reducing bias, looking at the endpoint, the time to death or range of symptoms. Are there extraneous variables particularly in veterinary study design which could have an impact on study performance? What about calculation of uncertainty and/or statistical significance?
We paid particular attention to the most common endpoint in the prion literature which is the log reduction endpoint. In the device literature we pay particular attention to a 6 log reduction in infectivity in our evaluation of sterility processes. That is sterility not prion decontamination.
In conclusion Dr. Russek-Cohen pointed out that the actual study design will vary with the scientific model and exactly what is being examined. This will drive how the systematic sources of variation will be designed but it is very important that the study be sufficiently sized to produce an appropriate level of certainty in the results. Again, we are looking at the product evaluation not just the scientific studies.
To put our risk in a little perspective Dr. Ron Brown from the Office of Science and Engineering Laboratories presented a risk analysis on the likelihood of transmitting sporadic CJD in the United States by contaminated surgical instruments at the present time. Dr. Brown did not address the risk assessments that were developed in the United Kingdom but he has done it with numbers that are appropriate to the United States. He did not consider in this analysis the risk of transmitting variant CJD.
This is the formula that was used for a deterministic model and the default values are listed. He also looked at a probabilistic as well as a deterministic model and again used our default values appropriate to the United States in terms of number of neurosurgical procedures, estimated prevalence of sporadic CJD in the United States and when you solve these equations this is the range for the probabilistic model.
On the average you get a figure of a risk of less than one transmission of CJD by contaminated surgical instruments in the United States per year. It may be as low as .1. With a worst case scenario for the range of parameters it possibly could be as high as 3-1/2 to 4 cases over year, but we consider that rather unlikely.
So, we really feel that in the United States at the present time the risk of transmitting sporadic CJD by contaminated surgical instruments when the appropriate precautions are used is quite low. It is certainly not zero.
We were fortunate to have a guest speaker from the United Kingdom from our sister agency and that is the Medicines and Health Care Products Regulatory Agency. Mr. Hidderley is the representative at MRHA who is particularly in charge of products involving TSE contamination and I have shortened his speech to the high points which are his opinion that new decontamination products are being presented to the market but the models chosen may not be substantive enough to assure that the product or process is adequately validated.
I should point out that the product has been approved in the United Kingdom at the present time for reducing the risk of TSE transmission related to surgical instruments.
In the United Kingdom the processing of contaminated instruments has been centralized to a specialized center. They are not using this product at the present time. Mr. Hedderley also pointed out that there is a great deal yet to be understood about reducing prions on materials other than simple surgical instruments and at the present time there is not uniform agreement in the scientific community on the appropriate animal prion strain model which is most representative for variant CJD and studies appropriate to that.
I am going to present to you the questions that we asked of our panel and the answers which they gave us. The first question was assuming that a product sponsor seeks a claim for reducing TSE infectivity on stainless steel instruments is it in fact reasonable for such an indication to be validated using animal studies of TSE transmission and the committee pointed out that while there are other ways of studying the general issue of TSE biology that at the present time looking at such a claim animal studies would be the reasonable way to go. In fact, they are probably the only way to go today.
The committee was then asked to discuss the relevance of various design features for such validation studies. This was a rather wide-ranging discussion but the advisory committee agreed that the following points were particularly relevant to validation studies for a claim of TSE reduction and those were maximum study validation. One should try to go beyond 1 year in particular in observing the animals. The study population should be large enough for sufficient statistical validity.
Now, that was not further defined, but the committee made very clear that we want large-scale very statistically significant studies. The log reduction in infectivity shown by such studies should be as large as possible. The committee declined to put an exact numerical figure on that log reduction. The committee decided that human prion sources would be the most appropriate to these studies for both variant and sporadic CJD models and the committee did point out that they really felt that variant CJD strains as well as sporadic strains should be studied.
Transgenic mouse models were felt to be most appropriate for these studies and should a sponsor need a new not previously studied human CJD source the committee pointed out that it would be appropriate to characterize such a strain against known animal TSE models and/or the WHO reference strains.
There was a question as to whether or not the WHO reference strains should be suggested for actually doing these studies. It was pointed out by Dr. Aher that they are intended to be reference strains not primary study strains.
Another question for the panel of the three study endpoints cited in the literature log reduction in infectivity, mean incubation time and survival, either median survival or percent survival, which if any might be adequate for the validation of a reducing TSE infectivity indication. Should demonstration of a particular level of reduction of TSE infectivity in one or more endpoints be expected in order to support such a claim and how may clinical benefit be estimated from these endpoints?
The advisory panel agreed that the log reduction in infectivity is the appropriate study endpoint for the validation of a reducing TSE infectivity claim. The committee declined to specify a particular level of reduction and stated that linking the mean incubation time to a log reduction in infectivity could demonstrate clinical relevance.
They asked what additional issues should be considered by FDA when evaluating indications for use for devices other than simple stainless steel surgical instruments. How can devices constructed from or including materials other than stainless steel, devices with complex shapes or difficult-to-clean surfaces be in fact validated, and the panel suggested that we should look at modification of the test wire surface, consider testing different materials and test simulated or surrogate device shapes.
The committee was asked how closely should we consider the treatment conditions for a product or process that would be used for such a claim. Should we look at things like instrument cleaning, the risk of fixing proteins, interactions between various steps in the cleaning procedure and the committee said, "Yes," they wished to see the study conditions simulate the clinical conditions of instrument processing as closely as possible and specifically mentioned attention to a large bioburden dried on the instrument before reprocessing.
Finally, we asked the committee considering the current state of the science could an indication for use of complete elimination of TSE infectivity in fact be validated, and the committee said, "No."
This was the substance of our discussion.
Thank you very much for your time.
DR. PRIOLA: I think we will move on to Dr. Scott who will introduce the questions to the committee,
Agenda Item: Topic 1: Progress Report on FDAs Risk Assessment for Potential Exposure to Variant Creutzfeldt-Jakob Disease in Human Plasma-Derived Antihemophilic Factor (FVIII) Products
Introduction and Questions to the Committee Dorothy Scott, MD, OBRR, CBER
DR. SCOTT: I am going to start with satisfactory products. Now, as of the last meeting we presented to you a risk model looking at the potential risk of vCJD transmission by plasma-derived Factor 11 and at that time we also presented to you a model for US plasma-derived products. As we started down this way after that last meeting in February we recognized the complexity of the input for the risk assessment model for US products and so the purpose of this topic today is to have a public discussion and to ask you for your advice prior to selecting these input ranges.
Why a risk assessment? I am just going to go over this very briefly. Blood plasma may have risk. There has been transfusion transmission of vCJD reported in the UK, two cases and plasma has also been shown to be infectious in animal models of spongiform encephalopathies.
However, I would point out that there have not been any variant CJD infections diagnosed in derivatives of plasma recipients. Risk estimates provide a basis for examining the adequacy of current measures to protect blood in plasma-derived products and these assessments may trigger a threshold for actions including risk management communications, surveillance and these risk assessments contribute to public health decisions and I would just point out that with regard to actions that the UK and France and some other countries have taken actions based on their risk assessments.
Back in February we presented to you a risk assessment model and you were asked to comment with regard to the model per se. You will be seeing this again today. Dr. Anderson will be reviewing it for you and we also ask you what additional information is needed to improve risk estimates with the various plasma derivatives.
You approved in general of the risk assessment framework and you felt that additional refinements should be made as more input information is collected. Different products obviously may have different risk levels and the committee recognized this. The donor travel history is important to consider and you will find out today what some of the variables are in figuring this out. The committee was also concerned about the 1-month exposure in the UK by a Japanese traveler who later developed variant CJD because this suggested that there may be a residual variant CJD risk in donors even with very brief travel to the United Kingdom.
Very briefly these are the main elements of risk assessment, the prevalence of variant CJD in US plasma and of course this is linked to the exposure to BSE and basically to travel in the UK and other countries, the amount of variant CJD infectivity in plasma, TSE clearance by plasma derivative manufacturing processes and patient exposure, that is how much of a product the patient has used.
The main outcome of a risk assessment for the main parameter is exposure per patient per year to one infectious dose 50 or more and 1 ID50 is only defined in animal models but it represents a 50 percent risk of infection in these animal models.
Another outcome of risk assessments is to identify sources of uncertainty. That is what we will be talking about a lot today. I just want to point out to you that the uncertainty in a risk assessment model increases with its complexity and when data is lacking from multiple input parameters. We also get a sensitivity analysis out of the risk assessment and the sensitivity analysis identifies input parameters that have the most impact on the outcome and among other things this can focus data collection efforts to the most important parameters.
Some aspects of risk assessment are important to acknowledge. The uncertainties and ranges that are provided, it is really a probabilistic model. It doesn't give you an exact number. It gives you a range of numbers at the end. Data gaps need to be communicated with a risk assessment to provide context to the people who are affected by the risk assessment. Input parameters should be adjusted over time to reflect scientific findings and the outcome should be compared to the actual observed risk over time.
These are the folks you will be hearing from in this session. First Dr. Steven Anderson will give a review of the FDA risk assessment model and then we will hear from Dr. Ghani and Dr. Knight about an update on vCJD in the UK and other countries, estimates of prevalence.
Dr. Alan Williams will talk about the modeling risk of variant CJD or rather the modeling of variant CJD in US illness, the residual risk and the efficiency of donor deferrals, and we will hear from Dr. Asher about the variant CJD infectivity of plasma, the estimates that have been made from experimental models. Then we will be discussing a review of TSE clearance in factory product manufacturing and then Dr. Mark Weinstein will be talking about Factor 8 product usage in the clinical Setting. So, overall we are going to be covering all those main parameters of input variables that go into the risk assessment and we are going to ask you a lot of questions and don't worry about remembering these right now because you will be asked these before each individual talk and we will put these back up at the end but this is just to start you thinking.
What is the estimate that should be used to reflect the prevalence of variant CJD in the UK?
How effective are current donor deferrals for geographic risk of variant CJD. In other words what is the residual risk in US plasma after donor deferral?
What intravenous infectivity range in ID50 should be selected for plasma based on animal studies?
Is there sufficient evidence to estimate when during the incubation period human plasma is infectious?
Do you agree with our proposed approach for estimating clearance of variant CJD infectivity from Factor 8 by the manufacturing processes and what experiments might enable refinement of these clearance estimates and allow comparison of clearance offered by various steps and methods used to manufacture plasma-derived Factor 8?
What data should we use to estimate how much Factor 8 is used by typical patients?
What is the effect of plasma pool size, that is the number of donors that contribute to a final lot of product for Factor 8 recipients and can a cumulative effect from repeated exposures to low doses of the variant CJD agent be incorporated into the risk model? We are also providing you with proposals to help you discuss these questions, and finally at the end of all this we are also going to ask you given the present scientific uncertainties in the underlying assumptions of the Factor 8 risk assessment do you believe that the risk assessment model as applied to Factor 8 could provide a useful basis for communication to patients or families and health care providers?
DR. PRIOLA: Next is Dr. Steven Anderson.
Agenda Item: Variant CJD Risk Associated with Human Plasma Derivatives: Introduction and Overview of Risk Model Steven Anderson, PhD, OBE, CBER
DR. ANDERSON: Good morning. As Dr. Scott alluded to I am going to give an update and a progress report on the variant CJD risk model for US manufactured Factor 8 products.
So, I think most of you have seen this slide before in many of my presentations. It is sort of the grounding framework that we use for all of these risk assessments that we do. This is the elements of risk assessment that were developed by the National Academy of Sciences in 1983, and I am just going to point to some of the highlights of this particular slide as it sort of relates to what I am going to be talking about today.
Most of what I am going to be talking about today will deal with the exposure assessment component and what we are interested in this exposure assessment component is what is the frequency and level of exposure to a particular hazard.
In this case the hazard we are interested in is variant CJD agent in these Factor 8 products potentially and also then what we do is we relate this exposure which gives us the dose, and we add that along with the dose response information and the relationship which is a linear relationship based on the ID50s that are published in the literature for animal experiments to finally get the risk and then we characterize the risk in the risk characterization section of the risk assessment.
So, this just sort of provides some basic background of what risk assessment is. Risk assessment is basically conducted when information is limited and uncertainty is high. So, throughout my entire talk one underlying theme throughout the entire talk is going to be this issue about uncertainty because it is a considerable element in risk assessment.
I think one of the important things to do though is to highlight the value of risk assessments. What role do they play in our decision making? Well, they are a tool that provides an estimate of risk and that can be the magnitude of risk or more specific information about the level of risk.
It details the uncertainties around that risk. So, it gives us a confidence level or confidence bounds around our estimate of risk and also it allows us to determine the effectiveness of mitigations, compare mitigations, identify which of those mitigations are potentially most effective in reducing risk and then finally it helps us identify data gaps and then determine research priorities.
All right, so again just to talk a little bit more about uncertainty and how uncertainty arises in these models as Dr. Scott alluded to uncertainty arises in risk assessment when there is only limited information available or when data are lacking and if data are lacking where this information is very limited it forces us to use assumptions or expert opinion in the model which increases the uncertainty in the model. There are also errors in measurement or data collection that can be an issue.
So, if the experimental data that we are basing a particular parameter on are flawed then that potentially can add to our uncertainty in the model.
There is also incorrect specification of the model. So, there can be problems. The model actually is incorrect. It is not considering all the factors that it should and that is an important part of uncertainty as well. So, I think you will see in some of the models that are going to be presented on predictive modeling for the size of the epidemic in the UK that those models are, the earlier models at least sort of just focused on the homozygous individuals at position 129 so they were methionine homozygous at position 129 of the PRP protein.
Now, what we are finding out more about as time goes on of course is that there are other individuals that are becoming susceptible to this disease. Those haven't been included in some of these more recent models and some of the recent models under development actually are going to focus on those.
So, let me just go on?
Factor 8 risk assessment that we are developing, again I have to say it is under development. So, we don't have any results from the model specifically. So, we don't have any output. We haven't done any runs of the model. So, don't ask which I know people are usually are very interested in the results right away.
I think it is important just to tell you that the model for variant CJD risk that we are develop really specifically models the risk for these Factor 8 products made in 2002.
Again, 2002 we have a fair amount of data for that year. It is a recent year and also was the first year that the FDA deferral policies for blood donors and plasma donors was fully in place.
We can do assessments for additional years and I think you will see that we have proposed to do 1999 as a potential year and that would be prior to the implementation of the deferral policies here at FDA and through the blood centers and the plasma centers as well.
I think it is important just to emphasize that this is the beginning of a very long process to assess risk for plasma derivatives and in the future we may assess variant CJD risk not only for Factor 8 products but for additional product classes. So, for instance we can go down the line for Factor 9 to anti-thrombin or a variety of other products.
We can also change the type of risk assessments so that we can possibly assess risk by specific products and manufacturers that produce those products. We could do it per individual per specific patient population, per individual, etc.
So, there is a wide range of possible routes for populations and products that we can potentially go with and choose to conduct risk assessments for.
Again, what type of risk assessment are we talking about? We are talking about a process model that analyzes the probability and quantity that the variant CJD agent will be in plasma pools and then if it is potentially in these plasma pools manufactured in the United States what are the potential reduction levels during processing and manufacturing, in the amount and levels of variant CJD infectivity in these pools and then what is quantity of factor by used by patients and that gets really at the question of exposure to variant CJD ID50s.
So, moving on I am going to provide just a brief overview of this model in a sort of diagrammatic format.
So, let me just orient you. Going down the middle are the modules and the components of the model. We have a four-part model here and on the left hand side we have the input and this is very important to focus on because these are the actual input information and data that we are using in the model and then on the right hand side we have the actual outputs.
So, these are what is being predicted from the model; what is the model generating as output, and it is important to note that is what is generated here goes on and becomes input into the next stage of the model and so on.
So, what is generated here goes here, here, goes here and then what we are finally trying to do is to estimate the annual exposure of recipients of Factor 8 products for the variant CJD agent.
I can walk you through at least a few steps of this. Our first module we are predicting the variant CJD prevalence in the United Kingdom. I will talk more about that in a minute and I will talk more about each of these components in just a minute but also we go from this variant CJD prevalence in the United Kingdom and that is used as a basis to predict variant CJD prevalence in US donors.
I will talk and sort of detail that more. What we are interested in with donors is donor travel history, specifically those individual that traveled to the UK, France or Europe since 1980 and then what we do is we adjust that donor travel information that we have for each donor population, we adjust for duration that they traveled, the specific year they traveled and then donor age and then finally I think one of the most sort of effective mitigations that we put in place is we analyzed in this risk assessment the effectiveness of the screening questionnaire that reflected policies for donor deferrals.
So, that is a large impact on the model because that is the step where risk is reduced considerably. So, what we do then is we get the total number of variant CJD donors, the number of variant CJD donors post-screening and then the total number of variant CJD donations. That goes in and once we have calculated this those donations end up in plasma pools.
So, we go down to our next step which is the processing of those plasma pools into Factor 8.
We are interested in plasma pools because these donations are going to be going into those plasma pools. So, what size pools do they go into? What is the quantity of agent in the pools and then what is the reduction level through the manufacturing and process that the variant CJD agent if it is present undergoes during this manufacturing process?
So, just to sort of emphasize what we get after that, we get the percentage of plasma pools and vials that may contain variant CJD agent and that is if they are made from a pool that contains the variant CJD donation from an individual infected with variant CJD and then what is the quantity of variant CJD agent in those vials?
Then finally what we are interested in is if patients use these vials what level of exposure will they be exposed to of this variant CJD agent? So, what is the patient's annual dose of Factor 8 and then finally we use that not only to predict annual exposure to Factor 8, I am sorry to the variant CJD agent but then to calculate their risk to that potential agent and I will explain a little bit more about that toward the end of the talk.
What are our proposed modeling approaches for modeling prevalence of variant CJD in the United Kingdom? Our proposed modeling approach is we propose to use two sources of data to estimate UK variant CJD prevalence. First would be a predictive modeling approach based on variant CJD cases in the United Kingdom.
Our second approach would be to use surveillance data and that surveillance data specifically involves the examination of tonsil and appendix samples from UK patients I believe in the mid-1990s.
I should mention that there is a disparity right now of approximately 10 to 100 fold between these two approaches. This seems to give a higher level in prediction of prevalence than the models.
I think what we are going to find in the future is that these two sort of estimates and these data sources are coming closer together as far as their estimates and I think they are going to probably sort of meet perhaps somewhere in the middle of these two estimates but we propose to use actually data from both in our modeling. So, let me just go through what are the predictive mathematical models.
These are some of the data we may use again. There are probably hundreds of these types of models out there and hundreds of publications.
Dr. Azra Ghani is going to talk more about here work and I think I have got this right. Her publication in 1990, she estimated a median of 100 cases in confidence intervals of variant CJD in the United Kingdom. That worked out to a median of about 1 in 500,000.
There are some other recent estimates, too, by a French group and the author is Belleli et al. They estimated approximately 180 to 300 cases and then there is also Lewin who estimated a variant CJD infectious prevalence in the population perhaps of 1 in 15,000 to 1 in 30,000. That works out to about 1000 to 2000 infections. I think this one is probably quite different from these other two.
These other two basically just model the clinical cases that could potentially develop in individuals that are methionine homozygous at position 129 of the PRP gene while this considers the other backgrounds, not only those methionine homozygous individuals but also the methionine-valine heterozygous individuals and also the valine homozygous individuals and just to remind people that the methionine homozygous individuals represent about 40 percent of the population. The methionine-valine I believe represent about another 40 percent as well.
So, again, this sort of brings home the point about model specification maybe slightly incorrect for these earlier estimates but this is an evolving field as we get more information about the infection and the type of infections that can occur. These models are getting updated probably as we speak.
Let me go on to the surveillance data. I just wanted to remind people that the surveillance data is based on tissue samples in UK patients in the 1990s. It was a surveillance study, a very large study and what they found was 3 prion positive samples in appendices in 12,674 samples tested.
That works out to this mean positive of about 1 in 4200 individuals. That works out if you correct it to about 13,000 variant CJD infected individuals. This isn't an age corrected number at this point. It is just a rough estimate.
This comes out as a very high estimate you will notice and this is, if you will remember the estimate that w used in the Factor 11 model that we presented to the committee in February of this year.
At that time we thought that perhaps this type estimate was higher and we used it because we thought it might capture some of those other methionine, valine heterozygous individuals that are asymptomatic and then also the valine individuals that are homozygous at codon 129, that it might also catch those individuals and what you asked us to do with that data that we used in that model was to age adjust that.
So, if we used this data we would of course age adjust this because this was collected in individuals that were 20 to 30 years old. So, we would adjust it to reflect prevalence distribution across all age groups in the UK population.
I think it is important to talk about uncertainties in this data because basically all data and all the information that we used in these risk assessments have some sort of limitations or uncertainties associated with them and the uncertainties of the proposed modeling approach is that predictive modeling really is based on the known variant CJD cases in the UK for the most part.
Most of the estimates although some of the models now are estimating in other populations but most of the models are valid for clinical cases of again this methionine homozygous codon in position 125 for those individuals. It doesn't capture the other genetic backgrounds.
Again, they use assumptions in these models for incubation period, time of infection and other factors. So, again that adds another level of uncertainty to these types of approaches.
All right, the surveillance data are no better. Using those again these are examinations on appendices samples. One sort of critical drawback here is that you can't go back and determine what the disposition of that patient was from which that sample was harvested and so we don't have any idea whether those patients actually became symptomatic or not or whether they actually came down with variant CJD or another TSE disease per se and another sort of drawback is variant CJD agent maybe in the appendices at the time but they may not really represent a threat for the blood supply if the agent isn't in the blood.
So, we may be overestimating the risk if we use this particular type of approach. Some people may say that we may underestimate the prevalence of the disease again. In one cases, one of the infected cases that were identified the agent actually wasn't in the appendices at the time the infection was identified.
So, again, both types of data have particular drawbacks and uncertainties associated with them. Again, probably neither case sort of adequately addresses the clinical or asymptomatic cases for the methionine valines or the valine homozygous individuals. That potentially could exist in the population and then they probably don't also sort of well represent the variant CJD infections in all of these groups that don't progress to symptomatic disease.
Okay, just going on through model 1, I think it is important at this point as I sort of end on this module that this is a critical parameter in the model and it is used to estimate not only variant CJD. We are not only using this for the UK population but we are going to be using a relative risk approach to estimate variant CJD prevalence for France and for Europe, and then ultimately what we do is we add up all those prevalences for the donors in the United States that have traveled to those regions. So, we are going to be using that for this variant CJD estimate of prevalence to also estimate plasma donor risk in the United States.
So, this is a critical parameter really to get correct. So, we need the best information available and possible for this particular point in the model. Just going back to the next step which is modeling prevalence of variant CJD in plasma donors and there are several modeling approaches that were, not several modeling approach that we are using.
Our goal is to estimate the size of the US donor population with a history of travel to the United Kingdom, France or Europe since 1980. Again, if you will notice this model if we go to 2002 is going to span 23 years of information. So, it is a rather large model at this point.
If we model this portion correctly we plan to determine travel characteristics from a survey that we have and that data was from the American Red Cross and then what we would do is adjust the travel data for each individual that has traveled by their duration of stay, the year of travel and their age and then our plan is to estimate the probability of infection in those individual donors based on the amount of time that they stayed in these particular countries and then ultimately what we would do is we would hope that we will add up the potential number of variant CJD cases in US plasma donor groups and then get the number of donations that they would donate and those would feed into the next portion of the model.
So, our model output, what we hope to predict would be the potential number of variant CJD infected US plasma donors, the variant CJD infected donors that are actually deferred from donation. The real risk that lies here I should say is in the donors that aren't deferred.
Those donors that are deferred from donation even if they have variant CJD don't really pose a threat to the blood supply or the plasma supply. What we are interested in is those individuals that actually evade the screening process by some way either incidentally or accidentally.
Then finally what are the potential number of donations that potentially contain variant CJD agent that enter these plasma pools that are used in manufacture of Factor 8?
All right, so moving on about prevalence of variant CJD in the US plasma donors our major assumption in the model is that variant CJD in US donors basically derives mostly from dietary exposure to BSE agent during travel. Again this is mostly travel in the United Kingdom but then secondarily travel in France and Europe.
Now, our current deferral policy is listed below. It defers donors with a travel history for instance, travel to the UK for individuals that traveled 3 months or more from the years 1980 to 1996, 5 years or more from 1980 to present, either in France or Europe and I think the important thing for Europe is that it is travel for 5 years or more from 1980 to the present for blood donations only and not for plasma donors. Again, I think Alan Williams is going to talk a little bit more about this in the subsequent talks but we are estimating somewhere around a factor of 90 percent to 99 percent factor of effectiveness in eliminating these variant CJD donors and the risk they may pose in transmitting infection if indeed they have agent in their blood or plasma.
So, what is the actual residual risk then after this policy has been put into place and where may risk lie in the system for plasma and for blood products?
What we have identified are two specific groups of interest. Again, it is these with deferrable risk and it is those 1 to 10 percent that have the deferrable travel history but for some reason evade the screen and get through and are able to donate plasma or blood and then the second source potentially would also be those with short duration travel to these countries and that again would be just what is not covered by the policy. So, that would be UK less than 3 months, France less than 5 years and Europe less than 5 years again during these specific time periods.
So, again, we are not only planning to model the deferrable risk but we are also planning to model this short duration travel as well because we believe that may pose risk as well.
Again, a number of individuals, probably the bulk of individuals fall into this category and that is why we are modeling it whereas fewer individuals have traveled to France for 5 years or more or to Europe or the UK for these long periods of time.
So, then we move on to the concept of relative risk and how it is being used in the model. Again, we are doing this modeling for a period of about 23 years for all donors that potentially donate plasma in the United States and how we are doing that and what we are using for prevalence is we are using this concept of relative risk that was used when the policy initially was set up and it is used to estimate the variant CJD prevalence for France and Europe relative to the UK prevalence.
So, I will provide an example down below in a minute and this relative risk estimate is based on a number of factors such as the potential for BSE exposure in France or Europe, the number of variant CJD cases, imports of feed and those types of factors. So, for UK the relative risk assigned to the United Kingdom has been a value of one and that is equivalent to the UK variant CJD prevalence.
France is thought to have about 1/20th of the risk or .05 of the risk. In our calculations these factors just become multipliers against the UK variant CJD prevalence. Europe has a risk estimated at about 1.5 percent or 1/60. Again that becomes a model, a multiplier in our model against the variant CJD prevalence and then there are individuals that spent significant amount of time in the military. That is estimated at 3.5 percent. Again, that becomes a multiplier in our model as well and then a final group of interest to us that may have been exposed to the variant CJD agent is those that received Euroblood and Euroblood was blood that was collected in three regions in Europe and then was used in the New York City region in the United States and given to recipients from donors that lived in Europe.
So, that multiplier is .015 times the variant CJD prevalence and then we have again further adjusted for the age of the European donor in that case.
So, relative risk for UK plasma donors, for US plasma donors with travel history -- so just to remind people what we are actually doing is we are applying this relative risk concept to US plasma donors with this history of travel to UK France or Europe since 1980. Again, as I mentioned this is a 23-year period. I just wanted to reinforce that we are adjusting for the duration of travel during this time period. It is a very critical factor because a lot of individuals spend only a few days in the United Kingdom. So, this is actually the biggest adjustment to that relative risk factor.
The specific year of travel we are also interested in spanning this time period and what we are doing is we are linking this to the variation in the BSE epidemic essentially linking it to the BSE epidemic curve.
So, somebody that traveled for instance at the height of the epidemic in 1993, would have a higher risk than somebody that traveled for instance at the start of the epidemic in 1980 or at the end which is probably around 2000 or even currently, so, those individuals or 2002. So, we account for the specific year of travel. We are also accounting for the donor again to apply the age specific rates for variant CJD in the United Kingdom. Again, the median age is 28 years and this is very important since most of our blood and plasma donors fall into this around this age category in the age 20 to 40. It is important to do this age adjustment for the specific rate of variant CJD.
All right, so we adjust by those three factors. I, also, wanted to say that we plan to model two types of plasma donors specifically the plasma donors and that represents greater than 80 percent of the donations. That is the source plasma donors. Again, those are collected by processes such as plasmapheresis.
Again, the second population then would be recovered plasma donors. Those represent less than 20 percent of the donations and those are whole blood donations actually that are recovered. The plasma is recovered from those.
Again, we have age-specific donation rates for each of these groups and we are planning to include those in the model as well.
Then finally just to give you an idea of where this plasma donor travel information is coming from we estimated this from survey data that was conducted by the American Red Cross. The survey was conducted in December 1998 and January 1999, and was presented in front of this committee I believe in 2000 or 2001, and what that survey covers is it queried travel history and accumulated stay information for the UK and Europe and from that we can make a few assumptions and infer travel specifically for France during this period from 1980 to 1996.
So, that gives us the bulk of the risk that we are interested in and then we are having to extrapolate further to cover the additional years since 1996 for France as well.
So, we do have some survey data and just to remind people that this is survey data in blood donors; it is not for plasma donors. So, that adds a level of uncertainty to this information.
Again, modeling the effectiveness of geographic deferral is an important aspect to this model. Again, we have this deferral policy for UK, France and Europe. Dr. Alan Williams is going to discuss more about that in his presentation.
I think I would just lay out some of the basic values that we are interested in that could be used in the model. We haven't particularly modeled either of these yet but we could use a factor that reduces 90 to 95 percent of the risk for first-time donations and then add a second layer to that which is a level of reduction of 99 percent of the risk for those that are repeat donors, and I should mention that probably greater than 90 percent of the donors of plasma are repeat donors.
So, we are eliminating essentially two logs of risk or 99 percent of the risk if we use this factor. So, that is something the committee may want to think about as well when they are discussing this particular issue.
So, let me move on? Another key factor is when is variant CJD agent present in blood during the incubation period. There is going to be a detailed discussion of the data by Dr. David Asher.
I think that I should just touch on the two potential approaches that could be modeled. What we could do is model if the agent is present in the bloodstream or plasma during the entire incubation period and I wanted to remind people that this is an assumption that we used in the Factor 11 risk assessment that we presented earlier.
We could model it as being present in the last half of the incubation period or later in the incubation period and this is based on experiments or one experiment by Dr. Paul Brown in which blood was found to be present later in the incubation period for a specific TSE model. I think one thing to point out is that the modeling in this case would be complex. It would increase uncertainty perhaps in the model and then also we would have to make a few assumptions about the duration of incubation periods.
So, we would have to think carefully about whether this approach is really something that we want to do, but we would like, I think, some feedback on that from the committee.
Uncertainties in the model, I will sort of speed through some of this since I am running short on time. The survey conducted on whole blood, so we have survey information on whole blood donations and the travel history for those individuals but what we don't have is survey information on the travel characteristics of source plasma donations and anecdotally people believe, I think that source plasma donors may travel less. So, if that is true then our blood donor travel information that we are currently using may slightly overestimate the risk for the source plasma donations. So, that is a little bit of uncertainty in the model.
Estimation of the deferral effectiveness is a challenge because there is the issue of self-deferral. So, many people don't even come in to donate blood or plasma because they know about the policy that is in place and then they just don't show up to donate blood or plasma, and that is a significant problem.
So, we don't actually know the total number of individuals really being affected or deferred by this policy. We don't know the denominator information. So, that is a challenge and then estimation again of when variant CJD agent is present in the blood from the animal data whether this is accurate or not for humans we don't know. It may be present the entire time in humans and it may be present only toward the end of the incubation period. We just don't know.
Just going through quickly for getting onto module 3 for Factor 8 processing and manufacturing our proposed modeling approaches to estimate the probability that a plasma pool will contain a variant CJD donation estimating the quantity of variant CJD per ml of plasma and then the amount of agent per pool we would estimate the efficiency of exposure and incorporate this into the model for the IV route versus the IC route and we would also include log 10 reductions and the log reductions in the quantity of the infectivity during the processing and manufacture of these products.
So, once we have that type of information input into the model then we would use t he model to output to predict the percentage of pools and vials that contain variant CJD agent and then the quantity of agent per vial.
So, we are getting further and further down the chain to the point where we are at the point of estimating the percentage of vials and the quantity of agent that may be contained in those vials.
A proposed modeling approach at least for the quantity of infectivity, Dr. David Asher is going to discuss this more in a minute in his presentation and this is intracerebral ID50s of the variant CJD agent per ml of blood. We propose to use a triangular distribution with the minimum of .1, a most likely of 10 and a maximum of 310.
Again, these are just sort of ranges that we found in the literature and we are interested in what the committee's perspective is on these particular data.
Again, this is just proposed approaches for the estimation of the efficiency of the exposure route to variant CJD ID50s or infectivity. In the Factor 11 risk assessment we used a value with a range from 5-to-10-fold based on experiments by Kimberline and also by Paul Brown's lab.
Recent unpublished data suggest that this might be lower. It might be only 1-to-5-fold. So, that is a question for the committee. We would propose again to use the estimate of 1-to-5-fold now for the adjustment in efficiency from intracerebral to the intravenous route.
As far as the plasma pool size we would propose again, we have information that suggests the plasma pool size used in the manufacture of these products ranges from 20,000 up to 60,000 donations. I think it is clear that we need more accurate information on the size of these pools used in manufacturing. So, at this point what we would propose to do is we would propose to use a bimodal distribution that favors sort of this 20,000 and also favors the 60,000 as an estimate of pool size.
So, 20,000 to 60,000 is the estimate for pool size for these products at this point. Just getting towards the modeling of the reduction in the amount of infectivity during the processing Dr. Scott is going to talk about this, Dorothy Scott, some of the reduction levels again, we are expecting that at least some level of reduction will occur during processing and manufacture of Factor 8.
I think it is important just to emphasize the designations for the degree of Factor 8 purity whether it is intermediate or high purity products may have little relationship to the level of variant CJD ID50 clearance.
So, I think that is an important thing to keep in mind. Just because these are high purity we are not sure exactly if they are going to be a high level of clearance or not.
We propose to use three values for the log reductions around these ranges of two logs of reduction, five logs of reduction or eight logs of reduction.
Again, uncertainties in the data, there is only a limited amount of data available on TSE reductions for a small number of processing steps and few products. The levels of reduction have been achieved in these experiments with spiked infectivity. The question is will that reflect actual levels of reduction using the endogenous material that is used during the manufacturing process.
Experimental data obtained for TSE agents that are other than the variant CJD agent that we are interested in; so the question is is there going to be an exact correlation between variant CJD reduction and for instance scrapie reduction or other types of agents that were used in these experiments, and another bigger question is does addition of these orthogonal reduction steps reflect the actual reduction of ID50s during manufacturing, so a lot of uncertainties in this type of data.
Utilization of Factor 8 in module 4, again we are very near the end of the models. So, our proposed modeling approach is to estimate. Again we are getting this from the previous section, the percentage of vials with variant CJD agent, the quantity of agent per vial and then we want information on annual utilization and dose of factor 8 that each patient might use or patient groups may use and our goal would be to model the annual dose of variant CJD that a patient is exposed to either per patient, per year is a possibility and then a prediction of the risk ultimately of variant CJD infection based on animal dose response relationship.
Just for time I think I am going to skip over an important set of slides but moving on just to talk about utilization of Factor 8 some of the utilization factors we are considering in the model is the severity of hemophilia. So, we are going to be modeling risk for severe, moderate and mild individuals and under different treatment regimens, prophylaxis and sporadic types of treatment or episodic treatment.
The type of data we are interested in using for the model would be the data for instance that comes from the Centers for Disease Control's hemophilia treatment centers. They followed 3000 patients from 1993 to 1998, and have utilization information based on review of medical charts. So, this is very good data and we may end up using it in the model if nothing sort of more recent comes along, but if available we may use additional data sources for instance from medical databases such as Center for Medicare-Medicaid Services, HMOs or Medicaid organizations for particular states.
Okay, so again just to highlight some of the uncertainties I will sort of rush through these. The utilization data isn't the most current and may not accurately reflect some of the current prescribing practices.
Patients may be on multiple products and the data don't really sort that out. So, that is one challenge that we will have. Patients may move among categories and that may not be captured in the data specifically from for instance prophylaxis to episodic treatment.
So, we are seeking additional data sources in order to get the best information possible on how patients utilize these particular products for treatment of their disorders.
Should FDA model the apparent cumulative, non-linear effects of repeated dosing? So, one question that we have is if you get a dose one time of variant CJD agent how does that compare to somebody that is using this product multiple times; you know, how does that risk compare to one hit versus they take 10 injections of that product over a period of time? Is their risk higher?
So, we would like to at least try to incorporate that into the model in some way and we are sort of having a challenge as to figure out how to exactly do that. So, we are interested in the single dose episode but also repeated and cumulative doses.
Dr. Mark Weinstein is going to talk a little bit more about the details of this but again it is a big challenge for us to model this.
The limited data available suggest in some cases there may be an added non-linear increase in infection rates with repeated dosing. So, to account for this what we propose is to model the cumulative variant CJD exposure per annum or per year and we would assume a linear ID50 dose response with that.
So, that is our answer right now is to assume exposure for a year and then try to estimate risk based on the dose-response information that we have.
That is certainly something that we are seeking input on. Again, just to remind people this exposure assessment is going to give us information on the dose of variant CJD ID50. So, we get this estimated dose from the model coupled with dose response information that we have to finally estimate risk.
Another issue that has come up is this sort of reflects our relationship for the linear dose-response. We assume that an individual exposed to one ID50 has a 50 percent probability of infection. Somebody exposed to .1 ID50 has a 5 percent probability and we don't really know if this is true or not. What is the meaning of a fractional dose? And that is an important question I think that we need to answer in the future and definitely need more experimental data to resolve that issue, but again right now we are assuming a linear dose response to estimate the risk. So, we are taking dose times the dose-response relationship to estimate the risk or the probability that a person will be infected with the agent.
Finally, again, a lot of uncertainties here; we are using an animal dose response to estimate human risk, a lot of uncertainty there. It is limited data to get that dose response. Human data are not available just to remind people of that and the development of the human dose-response model therefore is not possible at this time.
Some of the conclusions are the estimate risk of infection based on the level of exposure can be predicted using the model. So, we can get a relative estimate of the risk and the level of exposure that a person using Factor 8 products may be exposed to this variant CJD agent.
The risk prediction is based on animal data and animal dose response. So, again, it is going to be highly uncertain but I think one of the important things is that it will highlight the data gaps and uncertainty and hopefully t hose will improve in subsequent iterations of the risk assessment.
It is important to just sort of emphasize that risk assessments provide information on relative magnitudes of risk and this is somewhat useful for risk management purposes depending on the types of predictions that are generated and that is the end.
DR. PRIOLA: Thank you.
Next is Dr. Richard Knight, Director of the CJD Surveillance Unit, Edinburgh.
Agenda Item: Update on vCJD in UK and Other Countries: Estimates of Prevalence Richard Knight, MD UK Director, CJD Surveillance Unit Edinburgh
DR. KNIGHT: Thank you very much. This is a two-part presentation and I am going to begin talking from a clinician's point of view which is what I am and then Azra Ghani is going to follow on with a more statistical point of view and I am going to after a very brief introduction talk about the illness and its diagnosis, deal with some important questions and then turn to human transmission before passing on to my colleague.
So, a brief introduction, I am not going to go into all this in detail because obviously everybody here is familiar with this, but I do want to stress that I am going to talk about CJD essentially and really variant CJD.
These are the underlying assumptions that I will take as given, that the key molecular event is the post-translational change of the PRP protein from the normal cellular form to the abnormal disease-related form, that t his is going to be deposited in tissue and that this is associated with disease and associated with infection but I am going to pass over the difficult and controversial issues as to the precise nature of that association and what is more and it is important of course, misconcepts that while the disease is limited to the central nervous system the deposition of the abnormal protein may not be, and as you have already heard about this I will pass over it quickly but the prion protein gene is of critical importance and in particular the common polymorphism at codon 120 and here you can see that people can codify the methionine or valine. Therefore all of those are either MM homozygotes, VV homozygotes or heterozygotes and the significance of this is that this to some extent may affect susceptibility to these diseases. It may affect the incubation period in he quiet forms and it can also affect the clinical pathological features of the resulting illness.
This is the normal UK population. Just over one-third of those are methionine homozygotes, about one-half MV heterozygotes and the rest valine homozygotes. This does vary from country to country with roughly an east to west drift so that in Japan over 90 percent of the population are methionine homozygotes.
This is variant CJD, and you can see straightaway that all tested cases to date, 154 in total have been methionine homozygotes.
So, I will turn now to the illness and this diagnosis and we have identified 158 cases to date in the UK. You can see it is a disease of the relatively young. The youngest age of onset so far is 12. The oldest age of onset is 74, median duration 14 months although it can be very short and we have one individual who is still alive at well over 40 months. Apparently more men are affected but this is not a statistically significant difference and at any stage we tend to have a few people alive with the illness in the UK and at present we have seven.
This is just the background and I am not going to discuss it in detail but these are the three elements in the theory of variant CJD and as you go down the line the color blue becomes lighter because the evidence becomes thinner.
Certainly the diseases do appear to be caused by identical agents although of course we haven't characterized the agent and therefore all the evidence is indirect. It does not appear to have passed from a third animal or somewhere else into cattle and man. It does appear to have gone from cattle to man and it does appear to have passed in diet. We don't have any other reasonable theories. Our observations in the UK do not suggest any other plausible route and indeed our case-controlled study has now started to produce evidence that modestly but does support the dietary theory.
This is a rough outline of what we think happens. Infectivity from cattle enters food. Food enters the human gastrointestinal tract and then enters obviously the human being in general. There appears to be an important lymphoreticular phase. These are the tonsils, the spleen and of course the appendix and they Peyer's patches in the intestine and after some period apparently by a neurological route the infection enters the brain.
It may be through nerves in the gut going into the spinal cord and descending northwards. It may be the vagal nerve going directly into the brain stem. It may be through the trigeminal or glossopharyngeal nerve into the brain stem, then into the brain where disease results, and this is a probably pattern of tissue infectivity in humans, infection, a rise infectivity in the lymphoreticular system which then plateaus and later rising of infectivity in the central nervous system and at some point the beginning of clinical disease with central nervous system infectivity being significantly higher than that found in the lymphoreticular system.
This is just to illustrate the points about preclinical and subclinical infection as I am going to use them. You may become infected. There may be no evidence of an infectivity and then there may be lymphoreticular colonization. Then there may be neurological disease and the incubation period is the time to neurological disease followed by clinical illness and death from variant CJD.
In subclinical infection you may get nothing to start with and you may get lymphoreticular colonization but you get no variant CJD and die from another cause, and why would that happen?
One thing is the incubation period may be so long that it exceeds the life span of that individual either because the incubation period is longer than the normal human life span or because this individual dies for another reason before they have a chance to develop disease.
It may be there is a genuine subclinical state whereby no matter how long somebody lived they would actually never develop infection.
The difference between these two situations is of course rather theoretical at present. I don't know any way of distinguishing between them in human beings.
So, that of course brings us on to the incubation period of this disease and I think Azra Ghani is going to approach this in a rather more rigorous form but just to outline it in general we think the minimum incubation period is likely to be around about 5 years, the mean somewhere around 10, but the maximum may indeed be very long indeed.
Now, when you look at prion diseases there are certain factors which have been called determinants of the clinical pathological features but they may in fact be associations. I will pass over that point, and they are first of all cause, secondly route where these are acquired, thirdly, agent strain, a rather complex and controversial issue but nonetheless these agents do appear to exist in strains with different biological behavior, the type of protein found in the tissue and the 129 genotype, and if you look down at variant CJD there is apparently one cause at present, one route at present, one agent, one protein type and one prion genotype and therefore it is hardly surprising perhaps that in the United Kingdom and elsewhere this disease has been very homogenous from a clinical and pathological point of view which is rather unlike what you see in some other forms of prion disease, but certainly from a surveillance system point of view the critical factors that you change any of these factors like have a route other than the oral or you have a different genotype, then it might be that the resulting picture would be different and we might have to look out for a new variant CJD.
The illness tends to present with psychiatric and behavioral symptoms rather than the typical neurological presentation of many of these diseases. You may get other symptoms but they are often non-specific and taken as part of the psychiatric picture, and neurological signs appear usually around about 6 months into the illness and this is an illness of median duration of 14 months.
So, these people present with multiples symptoms with a psychiatric flavor without neurological signs and early neurological diagnosis is really very difficult indeed. There is no simple clinical diagnostic test at present.
The diagnosis therefore requires neuropathology if you want to be certain and this is a critically important point if you really are wanting to know for sure about variant CJD in the population. You need good neuropathology with effective autopsy rates and that does not happen in many countries.
A clinician approaches this by suspecting the disease in the first place which requires a suggestive clinical picture and for the clinician to understand of course what the picture looks like. They have to exclude alternative diagnoses and that I should stress includes other forms of CJD and so you need a knowledge of clinical neurology and of CJD in general and there are supportive tests which while not absolutely diagnostic are helpful if you understand their role and one particular very useful test is the MRI and what you see in variant CJD is this high signal in the posterior thalamic region, the pulvinar sign which is present in over 90 percent of our cases if you use flare sequencing and of course if you use certain standard techniques you do not find the abnormal form of the prion protein in lymphoreticular tissues in other forms of illness but you do find it in variant CJD and so tonsilar biopsy is sometimes potentially diagnostically helpful.
Now, I mentioned the differential diagnosis and it may seem perhaps a bit odd to anyone who has seen some cases of sporadic CJD but they could be confused with variant, but some genetic forms of prion disease may look like variant CJD and there are atypical sporadic CJD cases that may be atypically young with unusual even psychiatric presenting features with an atypical clinical course and an unusually long duration and in the United kingdom the main differential diagnosis of variant CJD is sporadic CJD and throughout the world we get notified every now and again of cases that are thought to be variant CJD occurring for the first time in a new country which are in fact atypical sporadic CJD cases and I think that I would stress again the importance of autopsy and stress also that if you are really interested in surveillance of variant CJD you need to have an effective surveillance system for all forms of prion disease.
How would you tell variant from sporadic apart? The MRI appearances can be very useful but they are not completely reliable. Tonsilar biopsy is certainly a possibility but a negative tonsilar biopsy can't exclude the illness and it is a relatively invasive test.
The histological appearances of course are important and so is protein typing but they require of course brain material. In the end the final arbiter at present is experimental transmission characteristic, i.e., you take material put it into laboratory animals and you look at the incubation period and the neuropathological lesion profile. However, that is clearly difficult and expensive and is only done in particularly important and difficult cases and just to finish on this topic this is the US CJD collaboration showing the standardized mortality ratios which of course should be around one and the countries in yellow are those countries that have mortality ratios that are not significantly different from one. The ones in blue, Slovakia and the United Kingdom are countries which have statistically lower mortality rates and I have no idea exactly why that is and two countries, France and Switzerland that have statistically higher mortality rates and again I am not sure why that is and certain countries like for example, Switzerland have shown a significant increase in identified sporadic CJD cases recently.
The explanation for this is not clear but there is a lot of collaborative research going on within these countries and all I can say at present is that there is no evidence that these differences or changes are due to unexpected infections with BSE.
So, that leads to questions, the relative youth, the numbers of cases, other genotypes, other countries and then going back to the preclinical, subclinical issue.
This is the graph showing the age at death or the present age if they are still alive of cases in the UK and you can see that they mostly fall in this 10 to 20 age group and what is more striking is that over the whole period of the epidemic in the United Kingdom this age of onset has not changed which is a big curious for a disease which is supposed to have been due to a limited time exposure of infection and stands in need of some explanation, and the three explanations that I have heard put forward are one, different age-related exposures, two, age-related incubation period, and three, age-related susceptibility and all I can say is that from my UK data we do not think that age-related exposure is likely to be a major factor, and therefore it is likely to be one or both of these and indeed of course they may in some way be related.
These are the numbers of cases. These are onsets in the United Kingdom showing a rise and then a fall although the figures for 2003 and 2004 are incomplete and you can see that there almost appears to be a rise again in 2004.
Mick Andrews of the Health Protection Agency produces curves to fit these data and the best curve at present is a quadratic curve. These are deaths showing a peak in deaths in 2000 with a subsequent decline. However, these profiles relate to dietary and codon 129 MM cases and of course the numbers and predictions will be dealt with in more detail by my colleague afterwards.
So, will there be non-MM cases? The only thing that I can say is that it is virtually certain that there will be non-MM cases and why do I say that? Well, first of all if you look at other prion diseases, iatrogenic CJD or kuru, other genotypes are affected. Secondly, one of my colleagues in collaboration with other colleagues in Edinburgh is looking at the human transgenic mouse model and while I can't I am afraid disclose the results of this experiment in detail the experiment's preliminary data suggests very strongly that other genotypes will be affected by BSE infection and of course we do have a case report of lymphoreticular systemic involvement in an MV blood recipient. This of course was not variant CJD itself and presumed to be an infection related illness rather than a dietary one but nonetheless it does show that and MV individual can have BSE infection.
If all this is true what we expect is first of all that longer incubation periods will be apparent for these genotypes which is perhaps why we haven't seen them to date. There is also some reason for believing that you might get more subclinical infections in these non-MM types and certainly that is evidence that is coming again from these human transgenic mouse experiments and the clinical pathological phenotype might be different and again, the human transgenic mouse experiments are beginning to suggest that the clinical pathological phenotype in non-MM cases might actually show some significant differences which is an important point for surveillance.
Well, now, aside from the UK these are the figures and you can see that France is next in the lead with 15, Republic of Ireland 4 and the rest of the countries single cases only.
The dark blue countries here are those countries in whom the cases are thought on good grounds to be intrinsic to those countries. They are cases in those countries who were infected in those countries. These colors here are those countries in whom they have had cases but it is thought that they were infected during stays in the UK and you can see here for example the USA case is considered really to be a UK case.
Saudi Arabia, the status is uncertain but I think it is likely to be an intrinsic case to Saudi Arabia. The Japanese case you have heard already is attributed at present to the UK. It is very difficult. The decision was a problematic one since they stayed for less than a month in the UK but it was thought that it was very unlikely that they would have gotten the infection in Japan intrinsically. It was also felt very unlikely that they would have gotten the infection during the 1-month stay in the UK but the probability that it was in the UK was marginally higher than it was for due to Japanese intrinsic infection.
Now, I will move on to preclinical and subclinical and we do have definitive evidence that the lymphoreticular system can be involved preclinically. These are two cases, one with onset of disease in 1995, and one in 1998 both of whom had their appendix removed due to routine surgery 8 months and 2 years prior to onset of disease and in both of these cases the appendices were examined and found to be positive.
So, the appendix certainly can be positive at least 2 years prior to clinical onset of disease. We, also, of course, have two cases of probable transmission by blood donation from individuals who at the time they donated did not have variant CJD. They were preclinical cases and therefore infectivity implies infection. So, these individuals must have had some form of preclinical infectivity and there is also of course the case of the lymphoreticular positivity in a blood recipient who did not have variant CJD and therefore again was some kind of preclinical or subclinical case and we have the UK appendix study.
Now, this is something you have already heard referred to. This was undertaken by David Hilton and James Ironside and others and they looked at surgical specimens from across the whole of the UK population. We don't know anything about the genotype of these individuals who had their appendix removed and presumably unless there is something peculiar about the prion protein and appendicitis they should represent a cross sample of the normal human population.
They were tested for PrPsc and nearly 13,000 appendices were examined and three of them were found to be positive. Now, if you extrapolate this data across the UK population then it would suggest that 247 per million of the UK population have appendix positivity. Of course, the 95 percent confidence intervals are wide. Most appendectomies are done in the 10 to 30 age group in the UK. So, if you adjust for that that would suggest that in the 10 to 30 age group in the UK nearly 4000 people have positive appendices for this abnormal protein, again with wide confidence intervals.
Now, you have heard there is a discrepancy here and indeed there is. If you say in this age group here there are nearly 4000 people with lymphoreticular positivity bearing in mind the wide confidence intervals yet within this age group we have only noted 92 cases of variant CJD and variant CJD in the UK appears to be in decline; so, what is the explanation for this?
Well, people have suggested that there could be false positives in the appendix study. I am not a neuropathologist, but I am given to understand that this is extremely unlikely. Of course, we don't know if the codon 129 genotype of these three positive appendices. They are being studied but I cannot tell you the results at present.
Of course, all the cases so far have been MM and it may be that the appendix cases are non-MM and of course it also might imply as indeed most people are beginning to think that there may be substantial numbers of subclinical cases of BSE infection in the population which is very reassuring to a neurologist but not very reassuring to a public health doctor.
So, that is it really in the UK. BSE is controlled at least in the UK. Diet is controlled, at least in the UK. So, we are awaiting the outcome of this terrible accident to see what happens, but of course in the meantime there could be secondary iatrogenic spread particularly if there are preclinical or subclinical cases and that leads on to the concern of surgery and blood, and I am going to close by just talking a little bit about human-to-human transmission via blood.
So, first of all there are opportunities for this. It should be possible to avoid clinically ill people being donors. At least I hope so. Therefore the exposure will come either from the incubation period which of course may be very long or even more worrying from subclinical infectivity which may extend through the whole lifetime of the individual especially in the absence of a simple diagnostic test.
If we are to consider this issue we have experimental evidence and we have epidemiological evidence and I am not going to review the experimental evidence at all except to comment on the sheep blood experiment which had most relevance, at least by Nora Hunter and her colleagues BSE was given to sheep. The sheep then act as blood donors and other sheep are intravenous recipients and just to summarize the present status of the study there is transmission of BSE by whole blood or by buffy coat, if transmission by intravenous route of a unit of blood it so parallels the human situation. Transmission has been successful with clinical phase donations and preclinical phase donations. So, there is evidence of preclinical infectivity again and the whole blood transmission rate at present is around about 25 percent but because of the methodology of this particular experiment this equates probably to a successful transmission of about 40 percent.
So, this is concerning but people always say can you really go from animals to humans. So, we don't experiment with humans but we observe them and the transfusion medicine epidemiological review was set up in 1997 between our unit and the National Blood Services and in outline for this particular topic what we do is we identify cases of variant CJD. We give their names to the national blood authorities. They look to see if they act as blood donors. If so they identify the recipients. They give the names of the recipients to us and we check to see whether they have appeared or in the future appear on our register and there is of course a reverse study whereby we traced the donors of blood when variant CJD cases report being recipients and there is a parallel sporadic study.
The data are present. When there were 157 cases in the United Kingdom we had 23 that had got donor records and the numbers from which the components were actually issued was 18 and the recipients were 66, so, essentially 66 recipients potentially at risk.
In addition 9 variant CJD individuals donated to 23 plasma pools which were identified as going on to make plasma products. What has happened to these 66 people? The first thing you note is that most of them have died. Forty out of 66 have died already and I suppose the reason for that is simply that if you have to have a blood transfusion you are ill and if you are ill you may die from that illness and you can see the majority of individuals actually died within 1 or 2 years, i.e., within the time period of the incubation period as we expected. In other words they would never actually have had a chance to develop variant CJD.
The ones who are still alive, 26 of them, there are a number that are in a short period of time from transfusion and one would not have expected them yet to develop variant CJD if they were going to. There were a significant number in a kind of high-risk period. A few have lived beyond 10 years although of course we don't know what the maximum incubation period for this disease is.
This is just to illustrate that it is no longer a UK problem. These are variant CJD blood donors by year of disease onset and you can see that it was a UK problem until recently.
The Saudi Arabian case had been a blood donor. Some of the French cases have been blood donors. Spain and Ireland have had blood donors.
So, if variant CJD occurs in other countries clearly the risk exists there as well. We, also, identified two individual cases and I want to discuss them in a little detail.
The first instance occurred in 2004. An individual donated blood. Three point three years later they developed variant CJD which was neuropathologically proven. One unit of non-leuko-depleted red blood cells was given to a recipient who 6-1/2 years had variant CJD and died after a fairly typical illness neuropathologically confirmed.
There were 68 slightly older than the usual case of variant CJD but they were still codon 129 MM. Now, of course, this individual had lived in the UK throughout the risk period and therefore had been exposed to diet and the question was couldn't this just be an accident. Well, the figure you heard earlier on of 1 in 15,000 to 1 in 30,000 my understanding was that it was not a prediction for the whole population but an analysis to see whether of the people that we knew who had been recipients of variant CJD blood what would be the chance that they had developed variant CJD by diet simply by accident, by coincidence and the answer in this particular analysis for this patient it would be about a 1 in 15,000 to 1 in 30,000 chance that they would have actually developed variant CJD through diet rather than through blood. In other words, this was very unlikely to be a coincidence and therefore a probable case of transmission.
In the second case a donor gave blood and 1-1/2 years later developed variant CJD, again, proven. An individual received non-leuko-depleted red blood cells and 5 years later died of a non-neurological illness. They didn't have any symptoms of variant CJD and neuropathologically there was no evidence of variant CJD.
However, at autopsy they were found to have the appropriate form of abnormal prion protein in the spleen and in the cervical lymph node. So, they had evidence of BSE infection and interestingly they were an MV genotype.
What is quite interesting about this case is that although the spleen and cervical lymph node were positive the tonsil and appendix were negative and of course I don't know why that is. It could be that because they were MV they have a different tissue distribution. We don't know. It could be because this is a blood transmission case rather than oral transmission case. There might be another reason. We don't know, and just to finish on the reverse study we have six variant CJD cases who have records of having received blood in the past.
Two of them the timing was just wrong. So, it couldn't have been that they got infection from the blood. In the remaining four one was the case I have described to you, the probably transmission of variant CJD with the time interval indicated here.
The other three cases we don't know. As far as we know the donors did not have variant CJD and have not developed variant CJD yet.
In one instance this individual here received a total of 106 components during her transfusion for a very serious illness. The intervals as you can see at present are running about 5 to 6 years.
So, just to conclude the UK is showing a decline in variant CJD, but there are lots of concerns. First of all there may very well be MV and VV cases to come. Secondly, more countries are being affected and some of these countries are not entirely predictable countries. There is increasing experimental evidence of blood risk and there are now two instances of probable actual human blood transmission, one of them showing evidence of blood infectivity at at least 3.3 years preclinically.
The magnitude of the blood risk clearly must relate to the prevalence of infection and this is particularly important with increasing concern over the whole issue of preclinical and even subclinical cases and the UK TMR study continues to collect data and obviously various precautionary measures have been taken, but what I will do now is hand over to my colleague, Azra Ghani who will address the issue of prevalence in the population in a more rigorous and scientific manner than I am capable of.
Agenda Item: Azra C. Ghani, PhD, London School of Hygiene and Tropical Medicine
DR. GHANI: Okay, thanks very much. Hopefully I will follow on from Rich's talk and that has given you most of the background and I am going to very much focus on a more quantitative aspect which is the mathematical modeling work that I have been involved with over the past 8 years and that other groups are also developing which really focuses on the first part of Steve Anderson's presentation in terms of trying to estimate the prevalence of variant CJD in the UK, and all of the work I am going to present today is work that I have been involved in but which over a period of time would have come from Oxford University and the Imperial College and now the London School of Hygiene and Tropical Medicine.
There are other papers published by other different groups. The predictions that we are all getting now are very, very similar and the basis of the models is also very much alike.
Okay, all the models to date have really been sort of risk assessment based on primary infection and by primary infection I mean ingestion through consumption of BSE-infected material and this schematic just shows the general process one might want to go through in trying to understand the potential for variant CJD cases arising through consumption of BSE-infected material.
So, at bottom here you have some sort of profile of the BSE epidemic in cattle and this is very well estimated. Certainly in the UK we have very good records of the clinical cases of BSE and mathematical models have been used to translate those clinical cases into estimates of infected animals slaughtered for human consumption over time.
It is important to note that current estimates suggest that 3 to 4 million animals were infected and slaughtered for consumption in the UK over the course of the epidemic.
So, there was widespread exposure to the BSE infective agent. There are a number of steps then that BSE infected cattle will go through prior to being consumed by a human.
Very little is known about the production in particular types of tissue used for the food or indeed the effectiveness of certain precautionary measures that were put in place notably a specified bovine offal ban which removed the riskiest material from the human food chain in the middle of 1989.
Consumption patterns are also thought to vary and there have been studies looking at dietary data. Dietary data obviously is fairly difficult to analyze. Typically there will be some sort of recall bias, particularly when you are trying to ask individuals about what they consumed over a long period of time, but it is likely that there was some heterogeneity. These to date and these have been included in some other mathematical models have really focused on age and certainly for the UK the dietary data do not support an age-dependent exposure as being the main reason why we are seeing the majority of cases in young individuals.
There is then the infection process. Infectivity obviously varies by different tissues that were consumed and some being riskier than others, also, by the incubation stage of the cattle and it is thought that the riskiest cattle would be those up to 1 year prior to clinical onset or onset of clinical signs.
There has been some discussion earlier about the dose response. In all of the models so far we have assumed a linear dose response and this is very much the simplest type of response to include in these types of models but obviously it is possible that there are some other forms of the dose response curve.
Heterogeneity and susceptibility are very much focused to date on two factors, variation in age-dependent susceptibility and I will go on to show how estimates about age-dependent susceptibility arise from these models and also genetic susceptibility and all of the models to date as has been stressed earlier have really focused on trying to predict what is happening in the MM homozygous population. So, that is approximately 40 percent of the UK population.
The reason for making that assumption is not so much that it is an assumption but all clinical cases have arisen in MM homozygotes and so it is very difficult to predict anything in the other genetic subgroups because we haven't seen any clinical cases.
There is then of course the species barrier very much an unknown quantity and then we have an incubation period in humans. Once an individual has become infected there will be a long and variable incubation period. This again could be dose dependent. No models to date have actually included a dose-dependent incubation period really because it is mathematically quite difficult to do. It may be age dependent and genotype dependent and earlier models t hat we looked at with age-dependent incubation periods showed a shift in age profile over time. So, it would suggest that the average age of those coming down with disease would increase over time and that is inconsistent with the data that Richard has shown in his previous talk and so most models now only consider age dependent susceptibility and exposure.
It is also of course important to include survivorship for diseases with long incubation periods. People will die of other competing causes of mortality and so this is a schematic that was used to relate estimates of BSE-infected animals to what is happening in variant CJD cases.
So, the process that the models go through is to try to incorporate all the various uncertainties through this process to generate potential epidemics and then we look at those generated epidemics and say how well do they match what we have seen so far and we can exclude statistically those that are way out and those that match very well are our best estimates.
So, this is really the only equation which is the fundamental equation of the process that is being used for mathematical modeling in this area and it is an equation that arises from a technique known as back calculation which is used very much for HIV and was developed for HIV in the mid-1980s.
What you are basically trying to do is relate the number of people who are becoming infected at certain times to those cases that you will observe at a later date. So, you are trying to relate numbers infected to the cases through multiplying this by the potential for the incubation period. So, it is really a method very much developed for long incubation period diseases.
So, mathematics is a little bit more difficult to interpret for most people. So, this is the same method simply expressed in words. Suppose we have a case of disease that we have observed in 2005. Then under this very simple model we would say that the probability that we would observe this case is equal to the number infected in 2004 times the probability that their incubation period was 1 year because if they had been infected in 2004 and the onset of disease was in 2005 they would have an incubation period of a year.
We then sum that with the number infected in 2003 times probability that the incubation period is 2 years and so on and we would go all the way back and in these models infection risk starts in 1980, and so that would be the probability for having an incubation period of 25 years.
So, that is the basic process these models are going through.
All of the models to date have a number of parameters and these parameters are uncertain and the approach to this uncertainty is to try to generate lots and lots of different scenarios to different sets of parameters and see which sets of these parameters are consistent with the variant CJD cases we have seen so far.
The first important one is some sort of measure of the exposure which is always estimates of numbers of infected animals entering the food supply by time and by disease stage at slaughter. So, in all of our models we assumed that those animals that are close to onset of clinical signs are more infectious than those early in their incubation period.
They all have an incubation period for humans which is either the time from infection to onset of disease or death. The term "incubation period" is used a little bit more loosely in the models dependent on whether the model is being fitted to the onset or the deaths.
They have some sort of age dependent susceptibility or exposure function which allows younger individuals to be at higher risk and that is a function that can be varied and we can try to see what function best explains the current data.
They all include the effect of control measures and most notably the specified bovine offal ban in mid-1989 which would have removed the highest risk materials from the food supply but we very much suspect it wasn't 100 percent effective.
Finally they have a transmission probability. That is the probability that someone who consumes infected material develops infection and so that is really dependent on the species barrier and finally competing courses of survival and birth cohort size are taken from census data. So, they are fixed quantities.
So, predictions of future clinical cases or deaths, this is what most of the models really have been set up to produce and models are able to produce fairly statistically robust estimates of future cases once the epidemic has peaked and that is a feature really of the methods, that these methods are helpful once you have seen a peak in an epidemic. Prior to the peak as you will see from the sort of predictions that were made prior to 2000 you get very wide uncertainty in any predictions.
So, there was variation in the different predictions that were made prior to 2000 by the different groups and these were just really dependent on slightly different assumptions about the parameters or slightly different model structures but they now all give very much similar estimates.
One important thing to note is that the predictions really are only valid within the populations that are being studied. So, to date all the UK models include the UK cases but do not include any of those cases that were assigned to other countries but were probable UK acquired infections, for example, the US and Canadian cases.
They consider all the cases but one to be acquired from consumption of BSE-infected beef. So, we are assuming that all these cases that we have seen so far have been through consumption of BSE-infected materials and for example cases arising through blood transmission.
For reasons I have already explained we only considered the MM homozygous population to be at risk and related to that they all assume what we call a unimodal incubation period so that the probability of coming down with disease after a certain time period increases and then decreases. So, we couldn't potentially see two peaks and situations in which we might see more than a multimodal incubation period really would be if we were trying to model wider genetic susceptibilities and we could see peaks coming in other genotypes.
The majority as I said, assume no age dependency in the incubation period. So, the cases I am going to present in the next slide are based on the models that are fit both to the time and age specified variant CJD deaths and the most recent predictions are based on cases up to the end of 2004, and importantly they also fit to the results in this appendix survey, so, the prevalence estimate that you will have seen in the previous talk.
To do that we need to include some sort of carrier state, subclinical infection. The terminology effectively mean the same thing so that we have a portion of individuals who become infected and are detectable having detected PRPSC in their appendix but do not go on to develop clinical disease and I will show estimates of how that is obtained in the model later.
They exclude the one patient thought to have acquired infection via blood and again they were only in the MM homozygous population.
So, this shows you the long-term estimates and bounds or prediction intervals for variant CJD mortality that my group have published over time and we started fitting these actually back in 1998 and this shows just how the uncertainty has decreased and really critically. There was a big jump after 2000 from 2000 to 2001 when the epidemic started, the clinical cases started to decrease.
So, current projections are for fairly low numbers. So, the best estimate at the moment would be that there would be 37 future cases with confidence bounds that have of course decreased as well so showing decreased uncertainty.
So, these are just saying that the projections of future clinical cases in the MM homozygous population in the UK are fairly low. We are seeing a declining epidemic.
By doing those and producing those projections you can also get some estimates of how age-dependent susceptibility and exposure relates to the age distribution of the clinical cases and the best distribution looks something like this so that those aged really between about 5 and 25 appear to be at highest risk for acquiring infection. We can't distinguish in these types of models whether that would be due to exposure or due to some sort of biological susceptibility and very similar patterns are reported from other models.
We can also obtain our best estimates for the mean incubation period and this is the incubation period defined as the time from infection to death rather than to clinical onset and that gives the best estimate currently around 11 years and you can see a tail here. The tail actually cuts off quite quickly but of course this aspect of it is less well fit by the model. There is more uncertainty in the tail simply because we won't have seen those cases arising with long incubation periods. So, there is uncertainty in that tail.
Okay, moving on then to estimates of prevalence of infection I think the most important point to note is that these models cannot estimate the prevalence of infection. You are taking models looking at the risk from BSE and trying to relate them to variant CJD cases.
All your information in this model is coming from those clinical cases. So they are quite powerful in predicting, making short-term projections onwards for those clinical cases but they can't say anything about prevalence because we could simply scale up and down the prevalence and increase the tail on the incubation period and we would get very similar projections in terms of what is happening with clinical cases.
So, if they are fit to the clinical cases alone we are unable to make any estimate of prevalence of infection. One way we can try to understand the prevalence of infection is to also fit the models to this data, the appendix survey data because that then constrains the prevalence at a specific point in time. So, those appendix tissues were removed between 1995 and 2000 in a specific age group and we could therefore constrain our model to match that prevalence as well as the clinical cases.
The useful thing from the models is that we can then say that given the age distribution of the clinical cases and given that we have found this prevalence in our highest risk group, the 10 to 30 age group what would we expect the wide equivalence to be elsewhere in the population; so it is quite powerful in terms of extrapolating those survey results.
For the clinical case and survey to be consistent in any model that we fit we need to include the possibility of a carrier state or subclinical infection and just a technical point. The survey results only apply to the MM population when we enter them into the model because our model is only looking at the MM population.
So, including this carrier state this is our sort of estimates that we get. So, we get estimates that only 10 to 15 percent of infections would go on to develop clinical disease. So, we are saying that 85 to 90 percent of those that appear infected or have detectable infection in their appendix will actually be subclinical infections that would never go on to develop clinical disease.
Just to stress that this is just an estimate from a model, it is not saying that subclinical infection does exist. This is one plausible hypothesis for why we get this discrepancy between the appendix survey results and the clinical data. It doesn't say that this is the only reason. This is just the reason that we have come up with that best fits the data at the current time.
The model then can also be used to then extrapolate and give estimates of prevalence by age group and these sort of data that are now being used to try to focus testing of tonsil tissues that have been collected in the UK.
So, there is a storage of tonsil tissues that started 2 years ago in the UK and we are now looking at ways to best test those tissues to try to detect infection and age is one aspect of that.
So, you can see also how we might see past, current and future prevalence coming out of these models by age. We would see a cohort effect so that infection in those that were young in 1997 will gradually move through the cohort.
So, this is just saying that a specific cohort was at highest risk. Going on finally then to genotypes we have gone through a lot of this information already on the genotypes. So, one aspect that we wanted to look at was if we were to see cases in non-MM individual although we can't constrain epidemics in this group what can we say about the potential scale of an epidemic in the non-MM individuals based on what we have seen in the MM individuals and we know that they have either to have not seen any cases to date they would have either had to be less susceptible to infection and/or have longer incubation periods. So, we can do a sensitivity analysis say if we have an incubation period that is up to 5 years longer and if we have a reduced susceptibility what would this do to our projections?
So, I will just skip that. So, these are the sort of results that we would get in terms of cases in non-MM genotype. So, what this figure is showing is the scaling of the incubation period distribution in the non-MM genotypes. Because we don't have very much data we haven't distinguished between the VVs and the MVs here. For example, here we would be saying that the incubation period in the non-MMs is twice as long as that in the MMs and here we have the relative susceptibility. So up here we would be saying that they are as susceptible but they may potentially have much longer incubation periods and as we go down we are saying that the non-MMs are less susceptible.
So a value of .5 would mean they are only as half as susceptible to infection and the colors here are denoting the potential sizes of the epidemic. So, of course, our worst case scenario would be if they were almost as susceptible as the MM genotypes but have much longer incubation periods because that would suggest we have an epidemic that is going to occur later in time that we are waiting to see, but this sensitivity analysis did also suggest that future case estimates could really only be up to about five-fold higher because this depends on the relative frequency of the MM and non-MM genotypes in the population, also, that it is unlikely that we would have greater susceptibility. The biology doesn't really suggest that the non-MMs would be more susceptible but have much, much longer incubation periods.
Okay, finally I wanted to talk a little bit about blood transfusion. I will skip these couple of slides because they have been covered by Richard in his previous talk.
One question of interest at the moment is what the potential for an epidemic arising through blood transfusion and this is really the main concern because we have obviously seen cases arising through this route.
So, just to go through what actually determines the potential scale of a transfusion associated transmission mathematical models typically deal with this quantity called the reproductive number of R naught and this is basically saying how many on average, how many new infections are on average generated by one initial infection in an almost entirely susceptible population.
So, diagrammatically here we have the first person infected through the dietary route and that person donated blood and that went on to infect this second person and that second person could then possibly donate blood that was given to two more individuals but only infection occurred in one of those instances and so we could see this expanding tree of infections and the critical quantity is this factor R naught, the average number of new infections generated by each individual in this chain and if that is more than one so each individual is on average transmitting on to more than one other individual we will see that this chain is expanding and we will get an expanding epidemic. If it is less than one, so on average each new first infection generates less than one new infection then the epidemic will die out. It won't persist.
So, this is a quantity that we have been particularly interested in. This quantity R naught is really telling you whether you can get a self-sustaining epidemic and a self-sustaining epidemic is obviously something we want to avoid.
So, this just shows you diagrammatically some very simple models, the types of picture you could get. Here if you have an R naught on two so on average each initial infection gives rise to two new infections, then you will see an epidemic appearing in this sort of form. It will become endemic and you will reach a steady equilibrium state of prevalence and over this time you will be accumulating new cases and you could see that cases could rise very, very rapidly for a R naught of this value.
For an R naught of less than one we see that the prevalence will die down and decrease and eventually we won't have infection in the population and so there is a lot of focus on keeping R naught less than one for all epidemics but I think it is very important to note that this does not mean that there aren't substantial numbers of cases arising. You can still get a substantial number of cases arising even if on average the epidemic is not going to be self-sustaining.
So, these principles really hold not just for the blood transfusion associated epidemics but also for transmission occurring via surgical instruments. So, what are the factors determining R naught? Most important obviously is the infectious dose. So, what is the probability that an individual given say X units of infected blood would become infected, and that is something that we realize is highly uncertain. We would never be able to really put any bounds on that value but we could potentially look at scenarios in which we vary this infectious dose right up to the value of one so that anybody given an infected unit of blood becomes infected which is a plausible value but also not just the infectious dose but also the numbers of donations recipients of transfusion really determine this R naught value. So, it is the magnitude of that blood supply and this just gives some figures for the UK and for the similar figures available in the US from the national blood service figures for 1996-97. We had 1.9 million donors donating just 2.2 million units of blood. We also have some good data on red cell transfusions so where the blood is actually going to and that gives estimates of the units received per 100,000 population per year. It,also, gives good estimates of the age distributions and this is a final point I would like to stop on.
This is the age distribution of the recipients and the donors. So,you can see that blood donors are typically in the 30-to-50 age range but as we all know transfusion recipients are very much older individuals and this age distribution is also very, very important for this concept of a self-sustaining epidemic because if these were the primary infections and we do see infections in that age group they might go on to donate blood that is used and infects individuals in this age group but it is then fairly unlikely that we would get infections going back into this population simply because of the age distribution but those receiving red cell transfusions are unlike to then themselves donate blood. They are in a much smaller probability.
So, these age distributions are very important both for blood transfusion and also for surgical instruments in determining the potential scale of an epidemic by these routes and this is work that is ongoing.
So, in summary then the variant CJD clinical cases remain low and the current predictions are also low based on either fitting into clinical cases or including a subclinical or carrier state.
The prevalence estimated from the appendix survey is higher than would have been expected from the epidemic observed so far if we didn't include any sort of form of subclinical infection and then this inclusion of this subclinical infection or carrier state to explain this discrepancy between the two data sets gives rise to an estimate that approximately 20 percent of those infected will become clinical cases.
There is of course still uncertainty in the genotypes and Richard already covered some of this. Worst case scenario is where the non-MM genotypes have similar susceptibility but much, much longer incubation periods and this could result in a five-fold higher estimate of future epidemic and the big remaining uncertainties are blood transmission and potentials for secondary epidemics.
Okay, I would like to acknowledge my people who have helped with work and finally I have put a few of the key model references down for people to look at.
DR. PRIOLA: Thank you very much, Ghani. Will you and Dr. Knight be here for the rest of the day?
DR. GHANI: Yes.
DR. PRIOLA: Okay, so, you will be available for questions.
I think we will take a break now for about 15 minutes and reconvene at ten-fifty-five.
DR. PRIOLA: Once again I would like to remind the Committee because I know this is tough to sit through these talks without asking questions but to keep in mind the questions that we are supposed to address as we listen to these talks so that when we get to the discussion we can get through the 10 questions we have to discuss early this afternoon.
So, our first speaker for this next session is going to be Alan Williams and he is going to talk to us about modeling the risk of variant CJD in US donors.
Agenda Item: Modeling Risk of vCJD in US Donors Residual Risk and Efficiency of donor Deferral Alan Williams, PhD, OBRR, CBER
DR. WILLIAMS: Thank you. Combined with laboratory testing the screening of blood donors by medical history and other screening questions is a powerful adjunct to helping to maintain the safety of the blood supply but there are a handful of situations when safety of the blood supply is wholly dependent on the screening of donors because there simply isn't an available laboratory test and variant CJD is one of those instances.
Most of you have seen a pie chart similar to this in the past but this is a representation of the overall risk burden in the donor population prior to the initiation of any interventions, any donor screening for dietary exposure.
One can see that the major risk component is dietary exposure within the UK, components related to travel residence in Europe, a portion of the donor population that was imported from Europe at the New York Blood Center and something relating to the second set of deferrals was the exposure to UK beef on military bases.
The first set of recommendations was issued in November 1999. I am not going to go through the details because Steve covered those in this talk and you had those in great detail last October but basically the primary deferral was related to time spent in the United Kingdom for 6 months or greater and a major consideration during that policy determination was the impact on the blood supply.
If one deferred everyone who had a contribution to that original pie chart one would lose more than 35 percent of the blood supply. Obviously that would not be tolerable. So, with the first recommendation there was an estimated loss of 2 percent which was tolerated and left this as the fractional component of risk proportionally a larger component from the base exposure. So, it was a sizeable component 32 percent from UK travel and then the Euro blood and Europe exposure.
Then in January 2002 the second set of recommendations issued which tightened the deferral for UK travel added base exposure and this is the representation or the impact on that second pie chart. The total risk burden reduction was estimated to be about 91 percent from that original pie chart and one can see here the representation of the DOD exposure removed by the deferral, the DOD exposure left in because of time that was lower than the time period of the recommendation.
UK exposure cut not quite in half with a little bit of UK exposure left primarily due to individuals who traveled less than 3 months and similarly for European exposure.
Now, this gulf really I think shows the two elements that are incorporated into the model. The white areas were risk is removed what we are going to be looking at in a major component of my talk is what is the efficiency of those donor screenings; has everyone in fact with that risk been taken out of the blood supply or is there a false-negative rate for the donor screening questions? The segments of the pie that are left there represent the shorter term deferrals, those who have been to Europe for just a couple of days or a week and are not subject to the deferral but may well have had dietary exposure throughout this period and represents quite a large number of donors.
So, just to reiterate factors related to the current donor screening procedures are considered in the proposed FDA model that Dr. Anderson described. The first is periods of dietary BSE/vCJD risk among donors that are less than the current deferral criteria. These are estimated from the original donor travel survey data and estimated as about 9 percent of the total risk burden prior to any implementation of donor screening.
The second factor included in the model is the consideration of the sensitivity of current donor screening procedures to exclude those donors with deferrable risk when they appear for donation.
FDA is proposing that in the absence of empirical measures the proposed FDA risk model should incorporate an estimate range of 90 to 90 (sic) percent sensitivity to occur in donor screening procedures to exclude that deferrable risk.
The second point, due to the donor population size, the limited other risk-reduction measures and the fact that the donor population dietary risk really is the risk contribution to the model the impact of this factor of sensitivity on the model is quite large.
So, the balance of this presentation will concern the rationale behind the 90 to 90 percent sensitivity estimate for the questions.
I think one thing to understand is what are the different components of donor screening. A policy is developed which is a combination of regulatory policy and industry standards and local blood establishment policy and this is generally made known to potential donors through educational materials, web sites, donor calls with questions. Often they get a little prescreening over the phone.
So, the elements that contribute to donor screening are the reduction of entire population subsets such as occurred in the past. Paid donors are eligible for donation but in the whole blood segment those units need to be labeled as being from a paid donor. Prisoners for some time have been ineligible for blood donation.
Then through education there is a large component of self-deferral before a blood drive. This is the group that Dr. Anderson mentioned that are aware that they are not eligible and simply don't appear at the blood center.
A second self-deferral is folks who appear at the blood center. They may not have been aware of it beforehand that they are not eligible but they see the materials while they are sitting waiting to be interviewed and determined that they are not eligible and leave.
There are also those where it is very hard to get any sort of empirical measures of how many donors actually do that because they are not available for study. The highly visible area of donor screening is a deferral by staff during the medical history interview process and I will say more about that in a moment and then finally if the donor proceeds with donation and that risk factor is not appropriately elucidated sometimes additional information comes back to the blood center and we find out about that risk after donation and in instances when that unit has been issued for transfusion that would constitute a biologic deviation and be reportable to FDA.
So, there are limited data available but that tends to be somewhat a passive reporting system. So, again it is hard to get rigorous data.
I mentioned the deferral on site that I would say more about. The 1999 recommendations there was an estimate of 2.4 percent of deferral among the blood donor population and 2002 about 5 percent, a little bit over 7 percent overall.
You can see that the on-site deferral experience and these have been presented through the years by Red Cross and other blood establishments, the current deferral is 3 to 4 per 1000 overall, most of this representing not first- time donors because the definition of first-time donors gets very complex but Red Cross has considered this as donors who have no previous onset exposure to the variant CJ question. So, although they may have donated previously it was in a time before the deferral was put into place.
So, you can see that the bulk of those deferrals are those who had no previous administration of the question and about 7 per 10,000, those who had a previous experience of the question may have traveled since that time but may have been missed in the earlier question administration.
A small proportion but when looking at about 14 million donations per year it can constitute a sizeable number of individuals and then similar data among the source plasma population, about 4 per 1000 deferral among candidate donors. These are donors whose plasma has not become part of the plasma pool because they are not yet qualified donors and if they a positive question response their donation is removed and the ongoing rate about 4 per 10,000,not incompatible with the whole blood donor population but there are probably other factors such as level of education, materials that are encountered and demographics that in fact probably account for some of those numbers in the underlying rationale.
So, how does one assess the sensitivity of donor screening? The things that we would really like to get at but we can't for variant CJD are changes in donor seroprevalence, reduction in adverse events to recipients or comparisons with risk levels in the general population.
Now, we can't get at any of this for variant CJD. So, the approach that we took was looking at some non-agents and some extrapolations from analogous situations with other transfusion transmitted disease risk factors, and then I also wanted to be sure to mention that there has been a lot of recent work in the questionnaire process itself. The Blood Products Advisory Committee has worked closely with FDA and the blood collection community so that the questionnaires currently being used for donors now have been cognitively assessed for understanding by donors, a lot of this work by the National Center for Health Statistics. So, really for the first time we have some confidence that the questions have been through a rigorous qualification process.
So, to look at some of the comparisons, first looking at marker prevalence and although I am not showing data for incidence presumably data should be available to make some similar comparisons using incidence of infectious disease markers.
So, here we are looking at comparing the general population versus first-time donors. First-time donors have not donated. They have not been prescreened. It is the best population to try to correlate with general population data.
So, looking at HIV seroprevalence, a report by Jerry McQuillan which is the Dallas County Household Survey found a little less than 1/2 percent HIV-positive in donor age general population subjects in Dallas County. That can be compared with about the same time frame HIV seropositivity of .03 percent in first-time donors. So, the difference between those two should be the impact of the education, screening on the donors and results in a 93.6 percent reduction in HIV seroprevalence.
Similarly one can make that comparison by looking over time and this I would say is the classic graph developed by Dr. Mike Gershon and his colleagues in the early 1990s.
This was a combination of looking at Irwin(?) Memorial Blood Center seroprevalence data together with some of the prevalence data for targeted studies within the San Francisco area and the NHLBI transfusion safety study, and this graph was produced to show the risk of HIV transmission per unit of blood starting with the earliest days of the epidemic through 1985 which was the development of specific screening.
It shows that the epidemic was under way for several years before initially recognized, then recognized by clusters of Pneumocystis carinii and other clinical syndromes and then the first post-transfusion AIDS case report and the recognition of high risk donor subpopulations which were then subject to deferral and self-deferral, and a lot of self-deferral took place so that by the time the test actually became available the risk in the transfused blood had been reduced about 90 percent solely on the donor screening aspect.
Similar comparisons, we started looking at seroprevalence, and here we are looking at risk from accepted blood donors and there was a study published by the University of Chicago showing that in the general population of the country there is about a 4.1 percent prevalence of males who have had sexual contact with other males in the past 5 years.
Compare that with survey data and I think the publication was shared with you a little more than 1/2 percent of that risk, males who have had sex with males since 1977 in accepted male donors, so, again, general population versus the male donor population about an 86 percent reduction and then one last example comparing intravenous drug use risk. Again, the Dallas Household Survey documented a little under 4 percent IDU since 1978 in the general population versus 1/2 percent intravenous drug use ever among accepted donors, again about an 86, 87 percent reduction.
Now, those show the positive impact of screening but I think there are also data which show quite convincingly that there is a false-negative aspect of screening and this comes from several different types of observational studies but basically data collected from donors after they are accepted for donation may identify behavioral risks that should have prevented their donation and that is what is being called deferrable risks.
I think again some of the classic data come from interviews with HIV seropositive donors, many of these studies sponsored by the Centers for Disease Control. For the period of the late 1980s and then another study in 1997 for both males and females when a donor is found HIV seropositive and subject to one-on-one interview one clearly sees risks that should have resulted in the deferral of that donor including MSM risk in males, heterosexual contact in a small component of injecting drug use and then similarly for women a much larger group with no reported risk but again others with non-heterosexual contact and one can see those types of data whether it is interviews with donors who have been identified with hepatitis infection or HTLV. One commonly sees risks that should have resulted in deferral and then similarly the Red(?) study did a post-donation anonymous survey of risks among accepted donors. This was published in JAMA in 1997 and then repeated in 1998 and also published with very similar results but a wider range of risks assessed and this shows that really for all the risks that are being screened for there is some false negative that gets through into the accepted donor population.
Just one anecdote that I happened to look at and didn't make it on the slide but males who trade drugs or sex for sexual contact there is a general population figure for that, I think, again, the Dallas Household Survey reported 11.6 percent in the general population versus 1/2 percent in blood donors. So, here we are seeing about a 20-fold risk reduction there as well.
So, you can see we are looking at general ballpark of risk reduction ranges.
From a behavioral science perspective one can understand why there are difficulties both in screening individuals for histories as well as trying to get rigorous data to look at efficacy.
Information about personal behaviors is inherently difficult to collect. One big thing is the social acceptability of information and clearly for injecting drug use and MSM that would probably be a larger factor than for travel history.
Response rates on behavioral surveys tend to be low. There tend to be missing data when someone completes a questionnaire and inconsistencies are frequent.
Now, keep in mind that blood establishments are regulated and they simply need to have all the blanks filled in. So, that is less of an issue within a regulated blood collection environment but it is a background problem with history taking.
People tend to avoid careful reading. I think for the donor setting there are those who are convinced that a video presentation or a one-on-one interview is better than simply giving somebody two pages of educational material to read but still the community is very dependent on reading material.
It is known that there is an educational effect of donor screening that even if a donor has missed the first time they are in there is a likelihood that they will be caught the second, third or fourth time that they are in.
So, there is a reduction in deferrable risk with subsequent donations and the Red study has shown some of that information for some of the major screening factors.
What happens when a donor determines that they are in fact, the individual that is being talked about in that screening questionnaire? There can be an aspect of denial and there can be quite a proactive lack of respect for policy if you have an individual who is knowledgeable and feels that the donor screening is out of line that in fact out of lack of respect they simply will not identify themselves as the subject of that policy and fail to self-defer.
There can also be external factors such as secondary gain for donation be it peer pressure coming to the donation site with a group of friends and not wanting to appear to stick out by being deferred. Plasma donors in the country largely are paid for their donations. So, peer pressure, other secondary gains and the environment, what is the privacy of the donor screening procedure; can one in fact give a socially less acceptable answer and be assured that it will remain confidential? And then finally, the aspect that has been looked at recently with respect to the questionnaire is the issue of comprehension of the donors and question complexity. Both of these are an acknowledged problem for the travel histories where one has to think of the cumulative travel history and determine whether or not that exceeds a 3-month period for the UK. I think there are probably those, a few of those in this room who would have some difficulty doing that if they are not current blood donors.
So, in conclusion based upon limited data from analogous donor screening situations we believe that a 90 to 99 percent estimate of screening sensitivity is bound, is realistic. The 90 percent lower bound supported by screening experience with other transfusion transmitted infections, the 99 percent upper bound supported by behavioral factors, the lower likelihood of travel among plasma donors and the high proportion of repeat donors, I mentioned that qualified donor aspect for source plasma donors and in fact all donors now it is an industry standard, are prequalified so that de facto all donors for source plasma are repeat donors.
There may be other ways to get at the predictive value and sensitivity of the questions but the work hasn't been done. There could be targeted follow up of the reasons for the post-donation information reports. There could be on-site attempts at validation and there could be comparison of results from different donor screening modes such as post-donation interviews or surveys to help continue to study this aspect.
So, I remind you that the TSE Advisory Committee in fact gave a mandate for the initial collection of survey data which you have heard about now many times so that if there is a key area of data collection I urge you to identify that and make it known most likely to the National Heart, Lung and Blood Institute which funds these types of studies and would be capable of carrying that out.
DR. PRIOLA: Thank you, Dr. Williams.
We will move on to the talk by Dr. David Asher who is going to tell us about estimates of variant CJD infectivity in plasma.
Agenda Item: VCJD Infectivity of Plasma Estimates from Experimental Models David Asher, MD, OBRR, CBER
DR. ASHER: Thanks. First, I want to say that to comment on the loss of Al Jenny so soon after Beth Williams, they were two of the people that I have relied on for years for expert advice besides being two of the nicest people that I can think of and I, personally, feel a tremendous sense of loss at their deaths.
We are asking the Committee to comment on several assumptions used in FDA's provisional variant CJD risk assessment for plasma derivatives. We have attempted to use very conservative assumptions about infectivity in plasma. We have assumed the infectivity to be present in blood throughout incubation period although we certainly that that won't prove to be true.
The maximum and minimum incubation periods of variant CJD are not known. Depending on how you estimate the minimum incubation period somewhere between 5 and 12 years after oral exposure seems reasonable and the one case of transfusion transmitted vCJD 6 years.
About infectivity how much infectivity might be present in human blood during various stages of the incubation period, the shorter answer is we don't know. All we can say with any confidence is that two units of non-leuko reduced packed red blood cells each contained at least one human intravenous infecting dose which translates into something like .005 doses per milliliter minimum.
However, it seems likely that there is more infectivity than that and we have taken provisionally 0.1 dose per milliliter as a reasonable minimum.
The rest of the model is based on what we know from experimental studies in rodents and 10 infectious doses per milliliter has been taken as an approximation of a common result.
We took 310 as the maximum dose because that is the maximum reported anywhere in the literature for animals although as you will learn that number is not really well supported.
In the 8 minutes or so that remains in this talk I can't review much of what we know about TSE agents in blood but I do want to remind you that for sporadic CJD attempts to transmit disease by inoculation of patients' blood into non-human primates even whole units into chimpanzees have failed. Claims of transmission of CJD from patients into rodents have not been confirmed.
Those observations are consistent with the epidemiological studies that failed to show blood transfusion or exposure to blood products as a risk factor for sporadic CJD and especially the surveillance of more than 100 recipients of labile blood components from donors who later got sporadic CJD presented here found that no recipient has ever been recognized with sporadic CJD, quite a different situation from that that we see with variant CJD as you know.
Studies of animal TSEs revealed similar difficulties in detecting agent in blood but when those difficulties were overcome a troubling consistency in detection of small amounts of infectivity in blood of animals during the incubation periods of TSEs was seen.
For many years William Hadlow and Carl Eklund hypothesized that there might be a blood-borne phase of infection in scrapie. They tried and failed to transmit the disease from sheep and goat and mouse blood into mice.
Several years ago Hunter Houston and colleagues succeeded as reviewed by Richard Knight earlier today, succeeded in transmitting natural scrapie to sheep by transfusion demonstrating the previous failures had probably been limited both by the species barrier of the assay and by the small amount of blood that it was possible to assay in experimental animals by intracerebral inoculation.
Shortly before the scrapie transmission by transfusion was demonstrated by Hunter and colleagues the same investigator succeeded in transmitting sheep adapted BSE by transfusion to scrapie-free sheep.
Those experiments did not permit estimates of the amount of infectivity in blood but experiments with the BSE agents adapted to mice suggested that the amount of infectivity in mouse blood must be very small; a published incubation time assay and I will comment about that in a minute was about 5 intracerebral infectious units per ml of mouse blood.
Most work in estimating the amounts of infectivity that might be present in blood has been performed using other models of TSE adapted to rodents which was first demonstrated by Elias and Laura Manulites who demonstrated convincingly in 1978 in blood of guinea pigs infected with a strain derived for a CJD patient. Soon after that Koroda and Gibbs in 1983, reported similar findings in mice infected with a strain adapted from the brain of a patient with a familiar TSE later identified as Gershner-Streisler-Shankar(?) syndrome.
A particularly popular model has been the 263K strain of scrapie agent adapted to hamsters and I will present some of the work of Bob Rohwer who is here to participate in the next session but who kindly lent me several of his slides. That model was first reported by Heino Diringer shortly after that by Petricio Casachia and Mauricio Pochiari after hamsters were inoculated intraperitoneally with scrapie. The results of that study were reworked recently by Phil Colmer and his colleagues at Detnar Scarveratosh(?) to estimate the average of 310 50 percent infectious doses per ml of blood during the first part of the scrapie incubation period and I want to note here that those results have at least two serious problems. First the pattern of blood infectivity which in their studies was present only in the first part of the incubation period has not been confirmed in later work and does not appear to be predictive of what we have seen for variant CJD.
In this panel prepared by Bob Rohwer I will show you what I mean if the cursor comes on. Okay, the cursor didn't come on. Okay, and I don't know where the pointer is. Okay, Heino Diringer demonstrated infectivity. The red stars are successful isolation of transmission attempts and the open circles are negative attempts, failures. In the first part of the incubation period they stopped looking at 40 days. Casachios showed infectivity on through 90 days but failed to demonstrate it after that and found no infectivity at 120 days when the animals were sick with scrapie.
If you look at other studies this is Laura Manuilitis's study in guinea pigs and you can see that after a little infectivity during the first couple of weeks potentially coming from the initial inoculum a number of weeks went by with no infectivity and then periodic detection of infectivity and infectivity detected in some animals on into clinical disease and the work of Karoda and later Paul Brown and Delores Achevanukova no infectivity early. Infectivity appears later in the incubation period and on into clinical disease. So, the Casachio Prochiari work on which that number of 310 is based suffers from not being predictive for other models of TSE.
Another problem is that they used an incubation period assay that is highly inaccurate in estimating small amounts of infectivity.
The classical method for estimating infectivity is titration testing serial dilutions of samples looking for that dilution at which easy assay unit has a 50 percent chance of being infected. Another way of looking at it is that it estimates the volume of inoculum that has a 50 percent chance of containing infectious agent.
As the amounts of infectivity get smaller the incubation periods get longer and so during most of the range of infectivity incubation period can be used as a, or the time to death can be used as a rough estimate of how much infectivity is present.
The problem is that it at the very smallest amounts of infectivity the dose-response curve becomes non-linear so that in this panel also prepared by Bob Roy you will see that at the end point there is a very wide spread of infectivity and animals come down anywhere from oh, I don't know, 130 days or out to 450 days and if only a small number of animals are being used those numbers cannot be used to predict accurately what the end point dilution would have been.
Paul Brown and Lois Achovanukova performed more informative estimates of infectivity using mice infected with the Fu-1 strain of GSS, the same one that Kuroda and Gibbs have developed and they did something approaching a classical titration. They found that buffy coat contained 44 in one experiment intracerebral infectivity units and 106 in another. Since in their study an ml of blood contained about a 1/20 ml of buffy coat that means that anywhere from 4 to 10 infectious units per ml plus the infectivity in the plasma must have been present and that varied from 10 to 22 infectious doses per ml in the same experiment. So, adding together the buffy coat and plasma associated infectivity it might be reasonable to conclude that whole blood probably contains somewhere between 9 and 22 infectious units per ml. Both of them are here today and would be far better than I to answer any questions. An even more precise, if very expensive assay method for infectivity has been used by Bob Rohwer called the limiting dilution titration and it represents an expansion of the classical titration by increasing the number of assay units near the end point.
For blood that means it assaying an entire volume, large volume of material and dividing it into aliquots and injecting the whole thing intracerebrally into hamsters. The readout is therefore not an ID50 but a direct measurements of the number of lethal doses present in the whole volume corrected using the Poisson distribution for the possibility that some of the animals might have been exposed to more than a single dose of infectivity.
When they did that on multiple assays Bob and his colleagues have titrated the blood of a large number of hamsters inoculated both intracerebrally and orally at intervals after infection, the maximum amount of infectivity they found in terminally ill hamsters varying from about 2 infectious units per ml to as much as 25 infectious units per ml with both median and mean about 10 infectious units per ml.
The infectivity was first detected in this series of experiments just before about day 50 representing something less than 50 percent of the total incubation period rising gradually until the onset of illness which was of course quite different from the results that Patricio Casachia had obtained.
At least 30 percent of the infectivity was associated with plasma in these experiments, 45 percent with buffy coat and 25 percent with red blood cells although some of Bob's other work suggest that red cells and platelets may not be intrinsically infected. I am sure we will hear more about that in the next session.
So, to conclude the provisional risk assessment for plasma derivatives is proposing as an assumption using a triangular distribution for the amount of infectivity that might be present in blood of donors incubating variant CJD a minimum of .1 ID50 per milliliter of blood mostly likely 10 ID50 and a maximum of 310 ID50 acknowledging the uncertainty regarding the assumption.
As I noted the experimental basis for the 310 ID50 per ml is questionable. However, it does represent the highest published value for rodent blood. The value is only about a log and a half higher than the mean value estimated for hamsters and that difference could be considered as a safety margin.
Perhaps members of the Committee think that we should even use a higher margin of safety or perhaps you know about information that we are not aware of.
Until last week we had a preliminary sensitivity analysis that had suggested that differences of this magnitude were not likely to be major drivers of overall risk. However, when the values were recalculated changing some of the other assumptions it appears that the level of infectivity in blood may not be an inconsequential element in risk although it doesn't appear to be as important as the prevalence of infection in the donor population or the ability of the manufacturing process to reduce infectivity.
Thanks very much.
DR. PRIOLA: We will move on to Dr. Scott for her presentation.
Agenda Item: Review of TSE Clearance in FVIII Product Manufacturing Dorothy Scott, MD, OBRR, CBER
DR. SCOTT: Thanks. As Dr. Asher just mentioned the amount of clearance in steps used to manufacture any product really is important for the risk assessment, that is it is a variable that is found to have a lot of impact. So, right now I am going to review the publicly available information on TSE clearance by steps used to manufacture just Factor 8 products.
These are the questions to you. Do you agree with our proposed approach for estimating clearance of TSEs from Factor 8 products by manufacturing for use in the risk assessment model and what experimental data would enable refinement of these estimates and allow comparisons of clearance affected by various steps in Factor 8 manufacture?
Just to very briefly remind everybody this is how a typical clearance step is done. You take a TSE infected material. Usually this is brain because it has a very high titer, but this can also be plasma from an infected animal and you add that to the starting plasma which would be like a plasma pool or intermediate material. Obviously this is all done at the lab scale. Then you take a scale-downed process or a series of processes that you want to study to see if they result in any clearance and you measure at the end of this process the leftover infectivity and this measurement as has been mentioned is typically in logs of reduction.
There are two main kinds of studies that are done but the most common form manufacturing processes is the spiking studies.
The reason for that is because if you use infected brain you can demonstrated significant clearance levels because the spike might be 6 to 9 logs of infectivity. Therefore if you end up with three or two you can show a substantial amount of clearance if it exists.
However, this has been criticized because the physical similarity to the blood-borne form of TSE agents of brain material is uncertain. There has been similar behavior demonstrated in some situations, that is convergence of in general of fairly high clearance for alcohol precipitations with spiking experiments and endogenous infectivity experiments but this has particularly been questioned for some other steps such as filtrations that are size dependent because there may be some forms of TSE infectivity that are quite small.
At least when the characteristics of blood-borne infectivity are defined the relevance of spiking preparations can then really be studied and understood. Endogenous infectivity studies using starting plasma of an infected animal are certainly more relevant it can be argued but they cannot demonstrate high levels of clearance because the starting infectivity is usually on the order of 1 to 2 logs per ml.
Those are very difficult experiments to titrate. So, there are two ways that we thought of to estimate the TSE clearance in Factor 8 product manufacturing. One is to use the published literature to identify clearance values from similar steps for Factor 8 or other products and we have seen this done in other risk assessments but I would point out there is a probable major flaw in this which is the fact that different clearance levels are often demonstrated for what appear to be similar steps, but these differences are likely due to product and process specifics. You don't need to read all of this. This is just an example of TSE clearance studies. Here is your reduction factor and these all use depth filtration with some kind of upstream process that allows material to be filtered.
This is a starting material and you can see this comes from published information including a package insert but basically the reduction factors that you get with depth filtration are highly dependent obviously potentially on the type of depth filter that is used but even not entirely that and on the starting material and its characteristics which would include different protein concentrations, pHs, ionic strengths and levels of alcohol or other things that are used to precipitate.
So, we didn't choose to use this more general method. Rather we thought it would be better to use product specific studies to identify clearance values. These are obviously more relevant to the specific products but they are not available for all Factor 8 products in the US and even when they are peer reviewed and published many of them have not been evaluated in detail by CBER.
There are lots of variations in the study methods including different spiking preparations. Even the brain preparations have different methodologies and they might be variably clarified, solubilized, sonicated or filtered and then of course there are microsomal and fibril preparations.
Furthermore the assays for TSE at the beginning and at the end may be surrogates, that is PRP, scrapie measures or bioassays and these results can sometimes differ.
I would just point out that the labeling claim that we offer for TSE clearance is based on demonstration of infectivity reduction at this time.
Now, what about the Factor 8 products that we look at? I just wanted to mention overall the Factor 8 products that we have and the purity has typically been defined by their Factor 8 activity and these have been called intermediate purity and you can see that that has a lower amount of Factor 8 per milligram of protein than a higher purity preparation or what is called a very high purity preparation. However, there are lots of methods used even to produce these intermediate purity products and overall for all of these products we see cryoprecipitation as a very common feature but there are also PEG precipitations in some cases, size exclusion chromatography, ion exchange, monoclonal antibody affinity chromatography and heparin affinity chromatography.
So, it is more than just cryoprecipitation. All of these products have a number of other subsequent steps that further purify the Factor 8.
I would like to mention, also, that many of the intermediate purity products contain von Wiilebrand's factor, and this is very important for people with von Willebrand's disease obviously.
The potential clearance of variant CJD or any TSE may not correlate with the classic definitions of purity. It really is process dependent. So, if you think of studies of TSE clearance and Factor 8 products one of these is not published. What we did do though is we took six of these reports and looked at the amount of clearance that we got and basically this is a summary and actually it is very similar to what PPTA has presented to this Committee in the past that cryoprecipitations typically do not result in a lot of clearance, that PEG or glycine precipitation just each step by itself may result in some level of clearance, maybe slightly higher with ion exchange chromatography and maybe slightly higher with affinity purification.
Now, how are our current plasma-derived Factor 8 products made? There is not a lot of detail here and the reason is because this came primarily from package inserts and a lot of the processes that are used in the manufacture are proprietary.
So, I just listed some of these, the heparin affinity for alphanate, PEG precipitation for Koate and immunoaffinity for these three products as well as ion exchange for both of these.
So, in other words what I am telling you here is this is not complete and understandably not because of the proprietary nature of manufacturing.
Now, this is our proposal for TSE clearance values to be used in the risk assessment. There is a range of clearance values that is suggested by the available studies for different manufacturing steps and the ranges that we would like to select are consistent with additional data that is available to us which has not been published.
What we propose to do is to run the risk assessment three times with three different clearance ranges, a likely minimum of 2 to 3 logs and this would more or less reflect a single step with an intermediate level of clearance in that 2 to 3 range, a mid-range clearance level of 4 to 6 logs which would usually involved a single step with a higher clearance level or multiple additive steps, that is throughput experiments sometimes demonstrate that if you start and do a number of steps in sequence that you get a higher amount of removal than you would if you studied just one of those steps.
In other words, they can be additive, and then a likely maximum of 7 to 9 logs which typically would involve two higher clearance steps, that is if they are additive.
So, I just put the questions back up again. Do you agree with this approach for estimating clearance? I think that I hope I have helped you understand that we do not have a comprehensive data set for these products and what data would help us refine these estimates because this is an important risk parameter.
DR. PRIOLA: We will move on to the last talk by Dr. Weinstein.
Agenda Item: FVIII Product Usage In Clinical Settings Mark Weinstein, PhD, OBRR, CBER
DR. WEINSTEIN: As Dr. Anderson mentioned I will talk about the utilization of Factor 8 products in clinical settings. Utilization factors being considered for the model include the severity of the disease and treatment regimens.
I will also talk about our proposal to model cumulative variant CJD exposure per year assuming a linear ID50 dose response.
The first question that we are presenting to the Committee regarding this topic is what data should be used to estimate how much Factor 8 is used by typical patients. For this estimation we need to know what the definition of a typical patient is; how much product is used per treatment; what is the frequency of dosage and what should be the time period to be evaluated?
To answer these questions I will briefly review some of the characteristics of the patient populations we are considering and available data sources for estimates of dosage and dosage frequency.
I will also mention some of the limitations that we have in our data collection. Regarding the patient populations we are considering hemophilia A is an inherited X-linked recessive trait. It is caused by a deficiency of the coagulation Factor 8 related to mutations of the cloning of the clotting factor gene.
The age-adjusted prevalence of hemophilia in the United States is approximately 1 in 10,000 males. Factor 8 normally circulates in the plasma bound to a second very large protein, von Willebrand's factor. I will just point it out in the model. There is a Factor 8 protein bound to the von Willebrand's factor.
The von Willebrand's factor protects Factor 8 from proteolysis, enhances Factor 8 synthesis and concentrates Factor 8 at the site of active hemostasis.
Von Willebrand's factor is also needed for binding of platelets to wound sites. Von Willebrand's disease is caused by reduced levels of Factor 8 activity and there are several types of von Willebrand's and subtypes of von Willebrand's disease but we will be considering in our model type 3 which is the most severe and requires treatment with plasma-derived products.
Von Willebrand's factor protein occurs in higher concentrations than Factor 8. In hemophilia A one can have no circulating Factor 8 but normal levels of von Willebrand's factor but in severe forms of von Willebrand's disease with little or no von Willebrand's factor present there will be little circulating Factor 8 and I will discuss that in the next slide.
To treat hemophilia A one needs products that contain Factor 8 but may or may not contain von Willebrand's factor.
To treat type 3 von Willebrand's disease you need concentrates that contain both von Willebrand's factor and Factor 8.
This chart outlines some of the characteristics of the hemophilia A population and von Willebrand's disease population that we are considering. Each of the populations is treated on average with different total doses of product and at different frequencies.
Hemophilia A patients with severe disease have less than 1 percent Factor 8 and have spontaneous bleeding predominantly in joints and muscles. They constitute the largest proportion of the hemophilia A population and there were roughly 6200 patients in the United States in 2002, our base year for our calculations for our model that had severe hemophilia A.
Those with moderate disease with Factor 8 levels from 1 to 5 percent have occasional spontaneous bleeding with severe bleeding accompanying trauma or surgery. We estimate there are about 3600 individuals in this category or 25 percent of the total hemophilia A population.
Those with mild hemophilia characterized by severe bleeding with major trauma or surgery are about 30 percent of the total population or about 3600 individuals.
Type 3 von Willebrand's disease is quite rare. These folks have less than 1 percent von Willebrand's factor and 2 to 3 percent Factor 8. They often have mucosal bleeding and severe bleeding with trauma or surgery. We estimate that there are approximately 250 of these individuals in the United States.
Regarding product usage the picture is extremely complex. First most hemophilia A patients we estimate on the order of 70 to 80 percent use recombinant Factor 8 and this is particularly true in the case of children.
As I mentioned previously von Willebrand's disease patients have to use products containing von Willebrand's factor.
The frequency of product usage and dosage are highly variable and depend on the patient's weight, type of bleed and clinical severity of the disease.
Analyzing data about product usage from different data sources is complicated because of the variation in how investigators define terms such as prophylaxis, intermittent or secondary prophylaxis and episodic or on demand treatment.
For example in some studies prophylaxis has been defined as receiving a product every other day or twice per week for 45 weeks or more per year.
However, on the form used by CDC in their uniform data collection program prophylaxis is defined as receipt of treatment products to prevent bleeding or to prevent rebleeding. Prophylaxis is further divided into a category of continuous if products are to be administered indefinitely or on a regular schedule to prevent any and all bleeding.
Intermittent prophylaxis is defined as the patient receiving treatment products on a regular schedule for a period of at least 28 days and at least one occasion since the last annual visit but therapy was not expected to last indefinitely.
Episodic is defined as receiving product only in response to bleeding complications. You will notice that in this data collection form that there is no specified or amount of product to be used.
Now, the best quantitative data that we currently have access to is from a CDC study carried out from 1993 to 1998 on all hemophilia A patients in six states. These states included Massachusetts, New York, Colorado, Georgia, Louisiana and Oklahoma.
Data was obtained from patient reports obtained from physicians, hospitals, clinical labs and hemophilia treatment centers. The list of information obtained in this study is quite extensive and will be very helpful to us. It includes the severity of the disease based on the activity range of Factor 8, the total number of bleeding episodes per year, and very importantly an estimate of the amount of product used per year, the pattern of usage whether it is prophylaxis or episodic, the number of weeks scheduled for prophylaxis and the brand of product being used.
Possible limitations of this data include that it might not reflect current usage. Changes in the average weight and activity level of patients using plasma-derived products may have changed which will affect the average amount of product used. Extrapolations to other states in the United States might not be accurate although the diversity of states in the original study was planned to be representative of the entire country but unfortunately for our purposes no data was collected specifically on type 3 of von Willebrand's disease patients.
The second source of information comes from the current program that was initiated by the CDC in 1998 called the universal data collection program. Patients voluntarily give information to the survey and we can obtain information regarding the numbers of hemophilia treatment center patients from the start of the program, the disease type, whether it is hemophilia A, severe, moderate, mild and type 3 von Willebrand's disease, the treatment prescription, again, unfortunately not quantitative but whether it is episodic, continuous or intermittent and the product brand used by the patients.
No information has been collected about the amount of product used. We, therefore have to extrapolate the amount of product used from the survey study. About 85 percent of patients who go to hemophilia treatment centers are enrolled in the UDC program and from the 1993 to 1998 survey data about 70 percent of the total hemophilia A population visit hemophilia treatment centers.
The patient population that doesn't visit hemophilia treatment centers may be different from those who do. For example, they may have milder forms of the disease.
Also, we know that some patients may use more than one product brand or type of product. Nevertheless using a combination of data from the survey and from the UDC data we have calculated the following information.
Most of the patients who use plasma-derived products are in the episodic category. Most use monoclonal affinity purified Factor 8.
Of the total 6200 hemophilia patient population in the severe category about 29 percent use plasma-derived products. Twelve percent of the 3600 patients in the moderate category use plasma-derived Factor 8 and 6 percent of the 4600 in the mild category use plasma-derived Factor 8.
Of the 189 type 3 patients who receive plasma-derived clotting factors all were on episodic treatment. Now, as to the amount of product used we can make some educated guesses based on current practice. For example, regarding prophylactic use some clinicians have defined prophylaxis as using 25 to 50 units of Factor 8 per kilogram 2 to 3.5 doses per week and for a 70-kilogram man using 25 units per kilogram for 52 weeks per year we calculate that that would amount to about 182,000 units per year.
At 50 units per kilogram, 3-1/2 times per week he would use about 637,000 units.
Again, the term "episodic" is ill defined for our purposes but just to cite one study in the literature a product was termed episodic when it was used at doses from 12.5 to 53 units per kilogram and it was infused 4 to 72 times per year.
In this study if we use that frequency wit h our 70-kilogram individual this would amount to receiving anywhere from 3500 to 270,000 units per year and he would receive a median dose of about 45,500 units per year.
Now, we can get some sense from the literature that we are at least in the ballpark about the amounts of product used per year from a study by Linden. She used the information obtained in the 1993 to 1998 survey study and extracted only the information pertaining to New York State. Dr. Linden reported that patients under prophylaxis in the severe, moderate and mild category used about 200,000, 100,000 and 25,000 units per year on average respectively.
Those on episodic treatment who were in the severe category used 80,000 units, moderates 21,000 units and milds around 3600 units.
Now, unfortunately these data are confounded because the estimates were combined from both hemophilia A and B patients. However, a very important point is that we can go back to the original survey data from all six states and with the very generous help of Mike Souci at the CDC who is in charge of the database and does much of our statistical analysis we can separate our data out about the use of product in hemophilia A patients to get detailed information on product use for each patient.
So, in conclusion regarding clinical use the existing data are limited and have not been analyzed for estimates of clinical use of specific brands of Factor 8 products in patient groups. We plan to analyze data from the ongoing UDC survey to estimate the numbers of patients using specific brands and the distribution of disease types. We, also, plan to extrapolate data from the 1993 to 1998 survey in the six states to estimate the total number of US patients and product consumption per patient with stratification according to the clinical setting.
If there is inconsistent information from these two analyses it would be reconciled using patient-based medical records data.
Now, we also know that in the future we may be able to get more accurate data about product usage from a survey study of inhibitor formation in hemophilia A patients.
This study is part of a larger data collection effort called the National Bleeding Disorders Coalition. However, this project is still very much in the planning stage and it might be a number of years before we are actually able to get data from this source.
We are, of course, interested in whether the Committee has further ideas about how we should collect data.
I would like next to turn to the question of whether repeated exposures to low doses of variant CJD infectivity lead to clinical disease. The data about risk of cumulative exposure is limited to a few studies in animals and I will describe one such study by Diringer.
Hamsters were fed one dose of scrapie-infected hamster brain for 1 day, one dose each day for 10 days or one dose every 4 days for 10 exposures.
There were about 60 animals in each of these studies. Approximately 1 percent of the animals fed a single dose died while 8 percent receiving a dose every day for 10 days died and 4 percent died of those receiving a single dose once every 4 days for 10 exposures. The authors concluded that hamsters receiving a repeated standard infectious dose several times have a higher risk of developing scrapie than those receiving a single infectious dose.
In another part of the study the authors fed the hamsters one infectious dose once. They, also, divided the dose into 10 tenths and gave one tenth eery day for 10 days to one group of animals and one tenth every fourth day for 10 exposures.
The animals who received the full dose all at once had 11 percent risk of infection while 8 percent had a risk of infection when the dose was spread out over 10 days and 4 percent had a risk when treatment was spread out over 40 days.
The study suggested that there was a smaller risk of infection associated with longer intervals between feeding.
These findings suggested exposure to repeated low doses of variant CJD infectious material by the oral route increases the potential for infection but that increase in the time between doses may decrease the risk.
So, in conclusion the risk of variant CJD infection may not be linearly related to cumulative exposure. Nevertheless despite possible low prevalence of variant CJD in plasma donors and limitations to pool size repeated dosing substantially increases potential risk of variant CJD exposure in Factor 8 product recipients.
We propose in our model based on these considerations to estimate the risk per annum in plasma derived or von Willebrand's factor product users rather than the risk per dose and we are asking whether the Committee agrees with this proposal.
DR. PRIOLA: Thank you very much, Dr. Weinstein. We will move on to the open public hearing portion of the meeting.
Agenda Item: Open Public Hearing
DR. FREAS: As part of the FDA Advisory Committee procedure we hold open public hearings to give members of the public an opportunity to make a statement concerning matters pending before the Committee. As Chairperson at this time I have received two written submissions for the meeting record, an e-mail from Ms.Sachau and an e-mail from Terry Singeltary. These e-mails are currently in the Committee members' folders, the public viewing notebook and copies are available at the reception desk upon request.
We, also, have received three requests for oral presentations for this morning's open public hearing sessions. The first requester is Dr. Peter Ostrow, professor and former dean of the University of Buffalo Medical School on behalf of the CJD Foundation Medical Education Program.
I am sorry, the most important thing before you begin the Chair must read a statement regarding all open public hearing speakers.
DR. PRIOLA: Both the Food and Drug Administration, FDA and the public believe in a transparent process for information gathering and decision making. To ensure such transparency at the open public hearing session of the Advisory Committee meeting FDA believes that it is important to understand the context of an individual's presentation. For this reason FDA encourages you, the open public hearing speaker at the beginning of your written or oral statement to advise the Committee of any financial relationship that you may have with any company or any group that is likely to be impacted by the topic of this meeting. For example, the financial information may include the company's or a group's payment of your travel, lodging or other expenses in connection with your attendance at the meeting.
Likewise FDA encourages you at the beginning of your statement to advise the Committee if you do not have any such financial relationships.
If you choose not to address this issue of financial relationships at the beginning of your statement it will not preclude you from speaking.
Now, you can go ahead.
DR. OSTROW: Thank you. Thanks for the opportunity to present at this meeting, and there are no financial relationships to report. I am going to show you a brief preview of a continuing medical education program about CJD that addresses some issues that are not well understood by health care workers including physicians and even neurologists. The CJD Foundation became aware of this information gap through their many contacts with patients' families and the questions that these people had about the disease, questions that physicians were often unable to answer.
In some cases the families also complained that the diagnosis seemed to take longer than it should have, that they felt like they were in the dark during the workup and that once a diagnosis of CJD had been made health care workers at various levels were not well prepared to deal with it and to some extent this is quite understandable. CJD is a very rare disease and few physicians have had much experience with it and certainly it can be very hard to diagnose especially in the early stages, and there is also a lot of misinformation freely available in the popular press.
Our program is intended to provide information to physicians that will make them more familiar with these diseases, that will help them in the workup, that will facilitate more useful and supportive interaction with patients' families and that will help physicians counter some of the widespread misconceptions about CJD.
I would like to acknowledge three people who are here today who have starring roles in this production, Dr. Dick Johnson, Dr. Richard Knight and Florence Kranitz, President of the CJD Foundation.
Our producer, Cathy Nelson is also here to make sure that this thing works and that I don't speak too long and now I would like you to see our preview.
(A DVD was shown.)
DR. OSTROW: This DVD which is aimed at physicians and residents will be available at the end of this year through the CJD Foundation. We hope to distribute it widely. I should mention that it will also include a review of the Red Cross Look Back study that the CJD Foundation is collaborating in to look at blood transmissibility. It is our hope that we will obtain support to produce a series of these things aimed not just at physicians but also at other health care workers, at the funeral industry, at patients and their families and also at the general public and thanks again for the opportunity to present today.
DR. FREAS: Thank you, Dr. Ostrow. Based on that preview I will look forward to seeing the full feature film.
We have two additional presenters this morning. They are Mr. Mark Skinner, President of the World Federation of the Hemophilia Foundation from Montreal, Canada and Andy Wertzel from the National Hemophilia Foundation. They have agreed since we are behind schedule to pool their time and Mr. Mark Skinner, President of the World Federation of Hemophilia Foundation will be making the presentation.
MR.SKINNER: Good afternoon. Thank you for having us and I have no financial disclosures. In fact both the WFH and NHF adopt strict ethical guidelines and would preclude us from having an investment in any of the companies or products we are talking about here today.
The issue that we are talking about and the reason that I am speaking today on behalf of both organizations, the National Hemophilia Foundation located here in the US and the World Federation which represents 107 member nations around the world is in fact that the actions of the FDA on this issue are of vital importance to the global patient population and the actions of the leading regulatory bodies including the FDA will certainly be heard around the world and will resonate and have the potential for impact on the use and the treatment of patients globally.
The World Federation of Hemophilia in fact last October in October 2004 published our first guidance document or complete guidance document. We published several before that discussing the assessment of risk as it relates to TSEs and hemophilia products.
We do believe that it is feasible to counter this emerging threat in our patient population but we also recognize that it is important that we balance the issues of safety and supply and we think of risk in this total context which includes access to treatment and also includes the impact on the supply of products and certainly actions of the Committee, the TSE Committee and the FDA in considering this important issue are a clear sign that the regulators are not complacent.
As a patient population we certainly adopt; we adhere; and we encourage regulators to advance the precautionary principle when there is incomplete scientific information which is certainly what we are faced here with today.
We acknowledge as a patient population that we are dealing in an area of uncertainty and we don't expect precise and exact answers. We do appreciate the effort to communicate in an open and transparent manner. That doesn't always occur. It is in fact occurring here today but there are certainly steps beyond and before this that are important for us as a patient population to be involved in and perhaps most significantly the FDA is not the only regulatory agency around the world which in fact is talking about this subject and so as regulatory authorities in France, Spain, the UK, Australia have all opined on similar subjects we are certainly seeing a divergence among the risk assessments and so the confusion among the patient population globally or the risk for confusion is certainly growing every day.
As I mentioned before we published our risk assessment back in last October and the parameters which we identified as they would relate to TSE risk for patients using plasma derivatives are listed out here. These are certainly very similar to the ones that the Committee has been addressing.
What I would say as an aside is that this risk assessment from the World Federation has actually now been accepted by Vox Sanguine(?) and it will be published in an article authored by Albert Ferugia, James Ironside and Paul Giangrande and we expect that to be published electronically in December and in the January issue of Vox Sanguine.
I am going to start first with the sixth issue because it is the most important and in the interests of time I can certainly skip over some of the others but the area in question of risk communication with the patient population, the clear and unequivocal answer on the part of the patients is that risk communication is important and the actions of publishing and developing a risk assessment are very helpful with a couple of caveats, that there is in fact advance consultation, that the patients do have a right to be informed, that the patients do have a right to be consulted and that it is vitally important that the leading patient organizations have an opportunity to be consulted in the development of the risk assessments as well as our leading clinicians and this doesn't mean simply a consultation in the context such as this but throughout the development process we need to have an opportunity for the patients to interact, to understand the kinds of issues t hat are going forward because certainly as we move forward and a risk assessment is published it is vital that the patient population's knowledge and understanding of the issues of certainty and uncertainty are acknowledged and understood.
If a risk assessment is dropped into laps of the patient population without adequate information we will see significant harm done to the patient population and this notion is certainly reinforced from the experience of the eighties.
One of the conclusions and recommendations of the IOM study relating to the HIV decision-making process was that it is shared responsibility and groups such as the National Hemophilia Foundation and other patient organizations representing the hemophilia community have a right and should be involved and should share the responsibility of decision making as it relates to infections such as HIV.
So, we believe the guidance from the eighties is applicable today and we should continue to monitor and learn from the lessons of the eighties.
The consequences on the public health for the 5100 odd patients in the UK that have hemophilia were in fact quite severe. We saw rapid access denials for patients for dental procedures. We saw denials for basic procedures including endoscopies in major hospitals around the UK and in fact those procedures continue today, some out of fear and some out of just simply the cost on the public health system.
If you are left with little choice but to destroy the medical instruments afterwards who is going to bear that cost, and so the hemophilia patient population was rapidly stigmatized because of the notification process that was not optional but was mandatory in the UK in which letters were sent not only to their general practitioners but also to their dentists and it was put in all of their files that in fact they were at risk for variant CJD.
Now, granted there are several approaches and the approach that the UK took was one of a general health precaution. So, the patient population as a group was generalized and they did not adopt the approach of individual risk assessments for individual patients which of course for the many reasons we have heard here this morning would be extremely difficult but we think precisely because of the adverse outcomes that occurred in the UK and because the patient population for hemophilia and not those with immune deficiency or alpha 1 anti-trypsin deficiency or in fact for the general blood transfusion recipient where there is an absolute risk known for transmission of vCJD, they did not receive the same universal precautions and as a result they weren't stigmatized but in fact some of those patients may in fact have a greater risk.
So, we as a patient population were actually singled out and have suffered great harm to our health care as a result of that. So, we are hoping that the Federal Government here in the US and other regulatory authorities have learned from this lesson and understand the importance not only of advanced consultation which did occur to some degree in the UK but of thinking through this process and allowing the patient population and our health care providers, our physicians to work with you to develop the appropriate response.
So, in summary as it relates to risk communication we as a patient population do think that we have the right to be informed and consulted. We strongly support the precautionary approach as it relates to patient groups. We do think it is important that we need to balance over reaction and patient stigmatization with the under reaction and the lack of information and lack of transparency. It is an extremely difficult balance. It is sensitive and we think through working with the patient groups that we can achieve the appropriate balance.
Certainly in risk assessment or any guidance that is offered has to come with counseling, competent explanation and clear information and you have to acknowledge uncertainty. I mentioned before the lack of consensus and the confusion between the global risk assessments certainly is one that will not only confound here in the US because patients do read the Internet. They do hear about the information happening around the world but will certainly make it difficult to manage treatment and care around the world and perhaps have an impact on product supply ultimately around the world and certainly the patients locally and nationally all need to be treated the same, that we need to think of this as a global problem and certainly given that the treatment products on which we rely are in fact global.
I would use just by way of illustration to emphasize the point of the importance of risk communication just this past weekend we had the National Hemophilia Foundation Annual Meeting and I had an opportunity to present on risk communication as it relates to variant CJD.
Following the meeting a young mother with a son of the age of 12 came up to me and in tears. I mean she is truly worried. She doesn't know what to do in terms of treating her patient. He has a high titer inhibitor and he is looking at immune tolerance using the plasma derivative with von Willebrand's factor to treat his condition but the one thing that she did tell me clearly and I think this goes right to the answer to question No. 6 is she said, "I appreciate very much knowing what we do know and what we don't know. The absence of information is more fearful than knowing what we don't know and as a mother I appreciate the openness and having a chance to hear it early and to make decisions with my health care provider."
Just quickly I want to run through a couple of the other questions and I don't know that these require a great deal of response. I will point out just one or two instances where our position differs and most of our answers here in fact come from the risk guidance we published last October.
Similar to the FDA when there is a doubt and when there is a variety of estimates in terms of what prevalence should be we adopt a conservative approach as has the FDA and suggested that we would encourage the Committee to adopt that as well. Obviously most of the risk assessment will hinge from this and this is the critical issue which is why you have it No. 1 and we did as well.
The prevalence of vCJD in the US we agree very much with the donor deferral mechanisms and until there is a clear pathogen reduction or elimination process that has been validated we have to maintain the integrity of the donor process and we would encourage the Committee to do that of course keeping in mind that it needs to be updated as new scientific evidence is available going forward.
Perhaps this is the area of slight difference with the FDA's assessment and guidance to date is that we do believe that the larger animal models are more predictive and we have adopted a similar position of the UEO and the UK in a one-to-one equivalency between intracerebral and intravenous routes of transmission. The level of CJD infectivity in human plasma, again, the similar approach following the precautionary principle adopting between the two and the options presented of the conservative estimate.
Dorothy Scott talked about this and we agree completely that clearance values must be product and process specific and without analyzing and evaluating each product it is impossible to have a generalized approach to the Factor 8 derivatives and the other products on which our community depends and part and parcel to that and assuming that you do agree to go forward with the risk assessment and communicate it to the community we do believe that it is appropriate for the FDA to consider actually including a label warning in the plasma derivative product speaking to variant CJD and while we are not typical authors of this kind of warning that is done through the regulatory process there is some language that we have put together and some thoughts that we believe would be useful and certainly mirrors the experience of similar warnings for HIV and HCV that have been previously communicated.
Again, we think it is important that it be indicated that it is highly precautionary,that there are no known transmissions but communicating uncertainty is an important step in informing the patient population.
A comment on the CDC data. We certainly agree that the UDC data, I had been a participant in this study since its inception, is very useful and perhaps the only available source. I would remind you as Mark pointed out that there are in fact a number of deficiencies in this data. There was a reference earlier perhaps supplementing this data with insurer data and I would suggest that is not the best option to do given the experience of working with insurer data and reimbursement and trying to understand treatment patterns. I think that you would not find that of great use. You may in fact find that some other countries which have a more centralized health care system you might be able to find data that would be more instructive as to plasma use treatment and dosage and duration of treatment that could supplement the CDC data and I would encourage you to look outside the US before looking to the insurers and payers within the US.
That question was deleted in the final questions. We do believe that risk is communicated and certainly from our experience with other pathogens that the patient populations are dependent upon it and that is consistent with what the FDA has recommended.
So, thank you.
DR. FREAS: Thank you very much, Mr. Skinner.
Is there anyone else in the audience who at this time would like to make a short comment to the Committee?
MR. CAVANAUGH: Thank you. I am Dave Cavanaugh with the Committee of 10,000. I will be quite brief, no financial interests. We appreciate what the World Federation has already come to the conclusion of but some of what we heard this morning was quite disturbing and I just want to speak to it in the hopes that you can bookmark it in your discussions. The whole discussion of preclinical numbers in the UK was very upsetting and there was only a brief reference later on as to the infectivity of persons during that period.
We have a couple of ideas about latter half, former half of the incubation period. We know so little. It is very much a call to be cautious and advisory as much as possible.
There was a discussion of the fractionation and reduction of infectivity as well and every time I heard the figure 90 percent I thought of 10 percent. That is all it would take and it makes us very nervous. Again, we need to understand that only in a few spiking studies has anybody been able to claim an acceptable, putatively acceptable level of reduction. It is not established. It is transmissible through factor. We have to assume that until it is absolutely wiped out as a possibility and it certainly is through whole blood. There is no national association of blood transfusion recipients. The folks in the blood products family sometimes have to represent that but please take that into account as well.
On the usage the material that Mark was presenting about calculating amount consumed it is known that 70, 80 percent of persons with hemophilia utilize in some degree hemophilia treatment centers but it is often not those that are mild that don't but those that are severely affected and have complicating factors like HIV and hepatitis that have to seek out infectious disease specialists and hepatologists. So, I wouldn't discount that population. I would at least assume that it is at the same level of risk and finally I got here at eight-thirty this morning thinking we started at nine. So, I missed the first presentation from USDA and I haven't heard that agency mentioned elsewhere this morning but you are linked to USDA. You are our FEMA. I hate to say it that way but that means rising to the occasion.
We have a motto that we have started. I don't think it is ready to take it on Oprah yet but the motto is if you eat beef don't give blood. The item that I wanted to say that I had forgotten in a list of things that have already been discussed was the post donation information revelations about true MSN behavior, true IDU behavior. They are catching 1 in 200 donors. How many get away? I don't know what the estimate is and I would like to talk to Alan further about it but a lot of people come and they never have that "aha" 3 days later. So, we do have a risky blood supply.
Thank you very much.
DR. FREAS: Thank you, Mr. Cavanaugh. Is there anyone else in the audience at this time?
If not, we will close the open public hearing session for the morning. We will have one in the afternoon.
DR. PRIOLA: I think with that we will break for lunch and reconvene here at one-thirty.
(Thereupon, at 12:45 p.m., a recess was taken until 1:45 p.m., the same day.)
A F T E R N O O N S E S S I O N 1:45 PM
Agenda Item: Committee Discussion and Recommendations
DR. PRIOLA: Okay, most of the Committee members are back. If we could get started with the discussion we have 10 questions to discuss but only the very last one that relates to issue 6 is a voting question. So, the FDA is interested in our discussion and the recommendations.
So, Dr. Scott is going to present the first question to us.
DR. SCOTT: This is as you have in your sheet of questions and in the issue statement what estimate should be used to reflect the prevalence of variant CJD in the UK which of course is linked to the prevalence of vCJD potentially in donors in the US, and we had proposed using the surgical tissue surveillance data as the assumed prevalence of vCJD in the UK.
This would be the most conservative estimate that we think we can make at this time and then we also propose that we run the risk assessment model again using epidemiological predictions based on diagnosed clinical disease in the UK as an alternative assumption of prevalence and then there would be adjustments for possible latent infections during the incubation period.
So, wait for the Committee's comment on this especially about using the surgical tissue surveillance data.
DR. PRIOLA: So, just to remind everyone this was one of the critical parameters that was pointed out earlier in the presentations and I think it was Dr. Ghani who mentioned that you can't use the clinical data as a substitution for prevalence unless you include the carrier state which is what you sort of added on your slide, right,with the latent; that is the same as the latent, yes.
So, comments from the Committee?
DR. GESCHWIND: I guess one comment I had was my feeling is that the surgical data should in some ways be the lower estimate and the reason I say that is that it is under the assumption that the immunohistochemistry is not 100 percent sensitive and PNS papers so far myself and Stan Prusner and others in February of this year using the confirmation dependent immunoassay showed that many cases are missed by immunohistochemistry particularly many parts of the brain.
So, I think that we should be conservative with that.
DR. PRIOLA: Other comments from the Committee?
Does the Committee agree then in general with the FDA proposal that both sets of data should be used?
I see lots of nodding heads.
Any other discussion?
DR. BRACEY: Yes, I think it really is important to use the surgical data because again there are so many unknowns as far as the incubation period and clearly in terms of the precautionary principle the idea would be to avoid exposure rather than to avoid clinical disease.
DR. PRIOLA: Okay, if there is no further discussion we can move on to the second question.
DR.SCOTT: This is also an issue with a fair amount of impact on the ultimate outcome of the risk assessment because it has to do with how many donors in the pool might have vCJD that are not deferred. What is the residual risk? But the question reads how effective are current donor deferrals for geographic risk of vCJD and our proposal is that based on the currently available surveys of unreported risks for other conditions and allowing for margin of error that we estimate that the FDA recommended deferral policy has a 90 to 99 percent efficiency for deferring donors with a specified increased vCJD risk.
DR. PRIOLA: I forgot to bring up one point with the previous question and that is does anybody have any comments about the PRP genotype influencing that modeling because all the modeling or the model that was presented to us was met-met homozygous with the caveat that if you included the met-val it would sort of be included with the latent carrier, the sort of subclinical state. Was that correct? I just want to make sure that, is that true, Ghani? I just want to make sure that point doesn't get overlooked because of this heterozygous subclinical individual who is picked up.
DR. GHANI: At the moment the only model, the MM homozygous and if you include allowance for their being other genetically susceptible groups you increase your estimates of clinical cases but at the moment the models are assuming that they are behaving very much the same way so that all genetic groups have potentially a subclinical infection as well. I just think one important point to note is that the models are fitting to this data from this tissue survey. So, they are not producing anything different if you assume that you use the tissue survey data. They are assuming at the moment that the tests are only sensitive in the last 75 percent of the incubation period and that we have varied that for the last 50 percent and that scale is the actual magnitude and that is something that has to be scaled in terms of the interpretation of the tissue survey as well. The actual published estimates say, "One hundred percent sensitivity." More realistically figures are probably around 50 percent.
DR. PRIOLA: Okay, thank you. I meant to bring up that point, and I forgot.
All right, so, on to issue 2. So, how effective are the current donor deferrals for geographical vCJD risk and does the Committee agree, have anything to add to the FDA proposal to model that between 90 to 99 percent?
DR. TELLING: I have a question for Dr. Anderson in this regard and that is for the input factors for module two. It seems to be exclusively with respect to travel to the UK but to the extent that this has been analyzed in other countries and in particular France and I realize that France is not the same as the US that that aspect had a relatively low rate of calculating the relative risk of exposure to BSE and a much more significant determinant was importation of BSE infected animals and foodstuffs and feed and so on and so forth, but I notice that that isn't a factor in your calculation for the risk inputs for module --
PARTICIPANT: So, right now the relative risk rate that we are using for France is .05 or 5 percent of the prevalence or the risk in the UK. I agree actually. We were talking about that at lunch that it is probably necessary to actually go back and adjust that number a little but upward because France recently has I think gone up to about 12 cases.
PARTICIPANT: Do I hear 16? Now we are getting more towards you know, 8, 9, 10 percent rather than .05. It should be more like you know .08, .09, you know, .1. So, we may consider actually adjusting that particular relative risk rate. Does that answer your question?
But we are specifically addressing travel to the UK. We have also got the number of travelers set aside that traveled specifically to France and then those that traveled to Europe.
DR. TELLING: My question is what is the relative risk of actual exposure of the US, the North American population actually in general to BSE materials imported from the UK.
PARTICIPANT: Right, from domestic exposure to any BSE that may perhaps be here. We have gone ahead and done those calculations and we determined that risk to be pretty negligible. So, we will talk about that in the model but it is not absolutely integrated into the model because the number is so low for the estimate. So, we acknowledge that that risk is there, but it is just so low that it doesn't really --
DR. TELLING: I am sorry but that is where the vast majority of BSE cases, I mean it is three cases, right, but the majority of those three cases are a result of importation, we presume of feed or cattle or whatever to Canada and then via Canada to the US. So, it is an important factor.
PARTICIPANT: Right, but we slaughter. We have got three cases but we slaughter 35 millions a year for food production. So, in the background of everything it really ends up being very low risk.
DR. BOLTON; I just wanted to say that I thought that the data actually supported better a range of 85 to 95 percent. It seems to me that 90 to 99 is a little overly optimistic.
DR. PRIOLA: Why do you want it dropped in that way?
DR. BOLTON: In the presentation I think several of the estimates that actually gave hard numbers were in the range of 86, 87 percent. Plus you have the 2 percent overestimation of the efficiency due to the false reporting essentially or false self-deferring.
So, it is not clear to me how the 90 percent as the lower end of the efficiency of the deferral policy came about. We are talking about issue 2, right? How effective are the current donor deferrals? I think 90 to 99 percent is too optimistic.
DR. PRIOLA: Dr. Williams, do you want to comment on that at all?
DR. WILLIAMS: I think it is quite a reasonable observation. I think the rationale between the 85 and the 90 percent is those 85, 86 percent figures were referring to deferrals for which you know the social acceptability, the answer would produce a downward pressure on getting a response as part of the screening process. That said, you know, those are the numbers and I think it is a reasonable observation.
DR. PRIOLA: Dr. Leitman?
DR. LEITMAN: It has been pointed out that the majority of plasma donors for plasma products for fractionation are paid donors and that population for whatever reason doesn't travel as widely especially to Europe perhaps as do voluntary donors who donate whole blood. So, that makes me more comfortable with this geographic deferral.
Having said that paid donors are not always the most honest group because they are not altruistic. They are donating for self-interest. So, if a positive answer had financial implications they might not give that so readily. So, those two things tend to balance themselves.
DR. PRIOLA: So, you would agree with Dr. Bolton dropping it to 85 percent would be a reasonable thing?
DR. LEITMAN: No, I think 90 to 99 percent is okay.
MR. BIAS: I thought 90 to 99 percent was optimistic just because I know the industry has done a lot to clean up their system about donor deferral but they are paid donors. Those centers are still in places where people are trading those dollars for sustenance or other things that they might need and I just thought that 90 to 99 percent meant we were doing it almost perfect and I don't think that is what is going on out there.
DR. PRIOLA: Dr. Bracey?
DR. BRACEY: This is really a request for Dr. Williams. What we are talking about are in essence projected numbers and we don't have a hard number and since this is an area where the safety is absolutely dependent upon the screening question I think that it really would be important for the Committee to send a message to the funding agencies that we need to acquire more hard information because right now we can try to draw parallels from other risks that we are dealing with but they are not the same.
DR. PRIOLA: That was something Dr. Williams brought up. What sort of other studies would you want funded to give you the information?
DR. BRACEY: In terms of just a general design a study that would look at simply the effectiveness of the screening questionnaires and that is you know these are very complex sets of questions and there are a number of things that were discussed such as comprehension, things such as donor motivation. For example, you know, if you were in the work place there are many donors that we sometimes consider to be nearly pathologic because they must donate. If they aren't given the chance to donate they feel as though there is something wrong with them and that is not the majority of donors but you will find a very small number of donors that will sometimes not give you all the truth so that they can participate. They want to feel normal, and those are the sort of things that I think need to get teased out if you are using a question as your only means for screening.
DR. PRIOLA: Other comments from the Committee?
So, is there any major disagreement over dropping that range to the lower end to take into account the points that have been raised to go from like 85 and still keep the upper range 99; you just extend the range and they can do their model in that way? Is that acceptable to the Committee?
And recommendation for funding for further studies along the lines of what Dr. Bracey asked. Are there any other comments or can we all agree that, I mean it is always better to have more information. I don't think anyone will disagree with that.
Okay, last chance for comments.
MR. BIAS: I would add to that that I would have that study look specifically at certain economic, the placement of certain economic centers. I have been to a number of blood collection centers and one in particular comes to mind and it was very pristine in Green Bay with a child care center and everybody was a student and it was very pleasant to visit there and then I visited one in an inner city area and it was clear to me that folks needed the income and that is why they were in line and I would gear the study toward those centers where people are using it as income, needed income for themselves and not doing it for altruistic ways and the one I visited in Wisconsin certainly they used it for what they called it was football, beer money which was fine. They were all sort of you know middle income folks and stuff but I would be very interested in seeing a survey that looked particularly at the inner city sites where people are at risk.
DR. PRIOLA: Dr. Scott, you wanted to say something?
DR. SCOTT: I just wanted to point out that it is a general believe that plasmapheresis donors travel less perhaps because of their economic situation but to my knowledge we don't have any real data on that and it would be nice to have some.
DR. ALLEN: I think my suggestion would be that to the extent further studies are designed and developed in this area that they look broadly at the issues of donor deferral and not necessarily just focus on this one risk. I mean trying to look specifically at geographic risk when we don't have a good measure of risk from food intake; you know, we have made some presumptions here in terms of time factors but we don't have good data. Just to the extent that additional studies are designed and carried out I think my recommendation to the FDA would be to look at a very broad scale design that meets the needs of multiple points not just this one specific one.
DR. BOLTON: I hate to delay these proceedings but this has been said now several times and I am just curious since this is a science-based decision-making process I have heard several times now the idea that the plasmapheresis donors travel less, but Dr. Scott just said that there are no data. So, the question is where does that come from if there aren't any data in support of that then why are we bringing it up?
DR. PRIOLA: Dr.Scott?
Okay, Dr. Leitman?
DR. LEITMAN: There are deferral rates and we know how many persons in the volunteer collection centers are deferred for a positive answer to the various demographic deferrals. At my center I can give you exact numbers. Other centers can give you the number. So, I assume that the plasma collection centers, paid donors can also give you that data. That assumes an honest answer and you don't get quite the number of self-deferrals that don't make it into the center because they know they are not supposed to. I assume that that is much lower in the paid centers but we don't know about the honesty of the answers.
DR. BOLTON; So, it really is a question of weighing is the answer honest given the act that there may also be economic incentives to not tell the truth versus is there actually less travel.
DR. WILLIAMS: There is one other side to the answer that I think will hold up scientifically. In the blood donor studies there was a clear correlation with age and so you are covering a cumulative travel history. It is clear it is going to be an age-specific rise and it is known that the plasma donor population is younger. Now, that is corrected in part of the model but that would apply to the risk associations as well, I think.
DR. SCOTT: I would, also, point out that the amount of estimated and actual loss of donors in the plasma sector we understood was lower when we instituted the geographic donor deferrals. I guess mu point really was that we don't have numbers.
DR. PRIOLA: Okay, I think given that this was another one of those parameters that was a high-impact parameter the call for new studies to get those sorts of numbers is something the Committee is asking for.
So, let us move on then to question 3.
DR. SCOTT: Question 3 is what is the intravenous infectivity range in ID50? What level should be selected for plasma based on animal studies and we propose that this model uses statistical distribution of infectivity with a minimum value of 0.1 ICID50 and most likely a value of 10, a maximum value of 310 as Dr. Asher explained and we also propose because the agent in primates may more closely reflect the human situation than rodent models that we model the IC to IV ratio for infectivity over a range of 1 to 5 based on recent primate study.
DR. PRIOLA: Any comments from the Committee?
DR. SALMAN: I think using the maximum value of 310 is very high, that is not justifiable if we follow on the same assumptions and being even on the very pessimistic side I think it still is very high.
DR. PRIOLA: Dr. Brown?
DR.BROWN; I would second that. The notion of a maximum based on a study that you don't believe strikes me as inappropriate even though it is higher than anything, well, and also because it is higher than anything that you could possibly imagine. Every other study has established an absolute maximum of blood infectivity or blood component infectivity at about a log and one-half and blood consistently, whole blood consistently ranges between 10 and 25.
I would suggest that as a maximum if one must have a maximum that it be set at 100, that is 2 logs rather than 320.
DR. PRIOLA: I agree; 310 is somewhat arbitrary given you don't believe the experiments. You might as well pick 400, 500, whatever.
Any other comments?
So, the lower end of the range we can agree upon and then set that upper end at 2 logs, at 100.
DR. BOLTON: This is a range and given new information comes in it can be adjusted as the model develops.
DR. PRIOLA: Any other comments?
What about the IC to IV ratio? We heard one of the public hearing speakers say that they favored a one-to-one IC to IV ratio and this 1 to 5 is based on recent primate studies.
Does the Committee have any comments?
DR.BROWN; One to one is a little pessimistic. Our study in mice was about 1 to 7. Earlier studies have, an earlier study at least by Kimberlin was 1 to 10. The primate study is based on very few animals and I think it is possible that it is a perfectly valid observation but it doesn't have the impact of the numbers that the earlier studies do.
So, I would think that somewhere between 5 and 10 is probably the most appropriate range.
DR. PRIOLA: So, as a lower end 1 to 5 you would say that 1 to 5 is at the lower end of that.
DR. BOLTON: I will respectfully disagree. I think that it is important to maintain that most pessimistic 1-to-1 relationship and then the plan is to study the range of 1 to 1 to 1 to 5. Is that not correct? Right. Dorothy is shaking her head. So, why not maintain that? I mean if you wanted to go more optimistic perhaps to go include the 1 to 10 relationship but I would hesitate to make it 1 to 5 to 1 to 10.
You are leaving out that possibility that IV exposure is exactly as efficient as IC exposure and I don't think we have enough information to really know that that is true or that it isn't true.
DR. PRIOLA: Dr. Allen?
DR. ALLEN: The other variable there is obviously the volume. Generally intracerebral tends to be a very, you know when it is done in animal studies tends to be a very small volume. If you are getting a whole unit of blood or packed red cells even, the volume is much,much greater proportionally. So, I don't disagree with you on the pessimistic. I would think perhaps extend the range from 1 to 1 to even to 1 to 10 to take in the range of available data from animal studies.
DR. BROWN: I don't want to inhibit pessimism.
DR. BROWN: I was really just suggesting that the most likely answer is going to turn out not to be 1 to 1 but the volume point is one that we have been hammering away at now for several years. If you really wanted an accurate comparison you would be obliged to inoculate the same volume IC as IV. Under those circumstances I have no doubt whatsoever that you are not going to get a 1 to 1.
DR. PRIOLA: So, the Committee would basically like to recommend a range in keeping with the range previously from 1 to 1 to 1 to 10 which would cover the most likely range IC/IV. Is that fair enough?
If there are no other comments we will move on to question 4 or question 3B.
This is the is there sufficient evidence to estimate when variant CJD is present in the plasma.
DR. SCOTT: I think we can go to four. I am sorry about the numbering but that is actually the same thing and our proposal is that because of the uncertainties about the incubation periods of food-borne vCJD and the time during the incubation period at which infectivity appears in humans we propose to adopt a conservative approach and assume plasma to be potentially infectious throughout the incubation period.
DR. BROWN: That is really conservative with a capital "C." There have been actually several studies. I put them all together on the table for a book that Marc Turner is going to edit and it turns out that in about half a dozen studies infectivity turns up about half to two-thirds of the way through the incubation period in each one, very consistent. That is data from rodents and data from sheep. That is true, also in the sheep study. The only primate study is actually in humans and there we know that 5 years in advance at least 5 years in advance of the onset of symptomatic disease in one case the transmitted case that, or is it 3 years? I guess it is 3 years but at least we know that.
You know you can model and the more modeling you do, that is fine as long as we don't come to the conclusion before the story plays out. Again the likely answer is that the incubation period from about being conservative halfway through is where infectivity begins to appear in the blood in those models where it appears.
DR. PRIOLA: Other comments from the Committee?
So, one of the things I see about that data depending upon the study you get infectivity early; then it goes away; then it is late. It starts in the middle. Then it goes away.
DR.BROWN: it is not that inconsistent, Sue.
DR. PRIOLA: It is not?
DR. BROWN: The early infectivity is almost certainly due to residual infectivity from the inoculation. If you sort of take away that first week or two then consistently infectivity doesn't begin before halfway through the incubation period and usually about two-thirds of the way.
DR. PRIOLA: So, you would recommend basing the model on the appearance of infectivity one-third to one-half of the way through the incubation period?
DR. BROWN: Yes, I would recommend the last half of the incubation period as being pretty conservative.
DR. PRIOLA: Other comments from Committee members?
MR. BIAS: I just wonder if that is the most conservative we can be since we are talking about plasma and people who are taking it on a regular basis. Doesn't that make us want to go toward the lower end? I am not exactly sure of the science. I will leave that to others but I am just a little concerned about that.
DR. BOLTON: I can't imagine if I were the one running the model that since it is just a model that I wouldn't include the entire incubation period just to see what the answer was. It is sort of an exercise in trying to figure out what the parameters are and which parameters given the biggest risks or uncertainties and it seems to me that it would make sense to again cover that range, the entire incubation period, half, maybe even the last third and see what that does to the model.
DR. PRIOLA: Dr. Anderson?
DR. ANDERSON: We can do it both ways and we have done that with particular models I think with Creutzfeldt-Jakob disease and the predictions at least come out. It is not a huge influence on the model. It probably predicts about twice as many cases if you go through the entire period versus the last half.
So, I think what I said before was that we are introducing a fair amount of uncertainty doing just the last half just because we have got to make some assumptions about incubation period and some other factors.
The good thing about using the entire incubation period is that you can then say that it can't be any worse than this with regard to that factor.
If you say that we are only assuming the last half then there is always the suspicion well, you didn't include the first half. So, what does that do to the outcome?
In the other case it certainly can't be any worse than that.
DR. ANDERSON: Right. It is a very conservative approach.
DR. PRIOLA: But you said that it wouldn't affect the outcome that much actually if you --
DR. ANDERSON: It is not going to affect the outcome as much as other factors in the model.
DR. BROWN: One of the bad things about using a broad range which I don't disagree with, David at all is that they tend to get through a process of inertia factored into the final risk assessment no matter what the answer turns out to be. Somebody is always going to say, "But you know you didn't include the whole incubation period. You had better do that," and when the data as it exists simply doesn't jibe with that I think you are obliged to set the limits for which you have science.
DR. PRIOLA: Dr. Salman?
DR. SALMAN; I am in agreement with what Paul is saying. If we use the model only as a game we don't need it. We have to try to as much as possible to have the model to mimic the infection status of what we know with the science we know.
I will question like any of these diseases, any of the diseases we know that the infection starts on the day one of the incubation period which diseases we know about that, I mean. So, we need some credibility for the model. Otherwise it will be a game.
DR. BOLTON: Let me say this. I think one thing that we do know is that from the day of infection the agent is present. It is present in the body somewhere. It may very well be in the blood and the plasma. It may not be in amounts that are easily detected, but given some of the studies and especially the recent paper from Rocky Mountain Lab that is published in Nature about the multiple injections over time I think that we have to now suspend our belief in the linear relationship between the LD50 and outcomes, in other words multiple inoculations of very low titer in fact below 1 ID50 dose will in fact result in a very high proportion of infection.
So, science based, yes, but we do know that once an individual is infected the agent is there somewhere and one of the problems with some of the animal studies is that the route of inoculation is important. The type of agent, the strain of the agent is important and the host is important and I am not sure. I was going to ask Paul this question before but again I don't want to make these things go on too long, but in the table that you constructed, Paul how do those studies look in terms of the peripheral nature of the early part of the infection anyway? I mean are those studies that used agent strain combinations that have peripheral distribution that mimics variant CJD or are they like some of the other models that are primarily CNS based?
DR. BROWN: I suspect none of the studies were full autopsy studies in a time course manner but I can tell you that the studies that I am referring to were studies of hamster-adapted scrapie, GSS-adapted mice, vCJD, mouse adapted vCJD, sheep-adapted scrapie and sheep-adapted BSE. So, it is a fair range.
DR. PRIOLA: So, let me ask the Committee maybe just since we have two it seems like primary points of view, both of which I can understand, actually, does the Committee have any preference for this assuming infectivity present throughout the disease course or basing it on where we do have scientific data and that is saying from about halfway on?
Any preference for modeling based on infectivity, oh, hold on.
DR. EPSTEIN: I do have one suggestion. There is the question of what parameters to put in the model and there is the question of how to do sensitivity analysis and one pathway here would be that we use 50 percent of the incubation period to generate the estimate and then report out the results of the sensitivity analysis looking at infectivity throughout the incubation period and that I think has the virtue of getting at both sides of the argument without biasing if you will the model itself.
DR. PRIOLA: David, do you like that idea?
MR. BIAS: It works for me.
DR. PRIOLA: Okay, nice compromise. Okay, if we have no other comments let us move on to the next question.
DR. SCOTT: Does the Committee agree with our proposed approach for estimating clearance of vCJD infectivity from Factor 8 by manufacturing processes? We proposed to model three clearance ranges to represent a likely minimum, 2 to 3 logs, midrange 4 to 6, and maximum 7 to 9 logs of clearance of the vCJD agent from products manufactured in a variety of ways.
DR. PRIOLA: Comments from the Committee?
DR. BROWN: Is there any particular reason that you had for and again here we are playing games in terms of modeling but we already know that if you have a guaranteed 3 log per ml clearance as a concentration clearance you have done the job and 4 to 6 logs is nice; 7 to 9 logs is just insane. I am not sure why you are going to do that
DR. LEITMAN: For clarification just the 20,000 to 60,000 pool already gives you a log reduction, well, a 4-log reduction in concentration. So, this is over and above the effect of dilution in the pool. This is specific processing whatever that is, precipitation, filtration. So, is that beyond the several logs just by dilution?
DR. SCOTT: I guess that would be but I think the whole dilutional thing is another topic altogether. So, if you start out with 10 infectious doses per ml of plasma and you have 800 cc's of plasma of 8000 infectious doses that is going into a pool of 20,000 but that is still, if just one of those is enough to infect somebody you still have 8000 infections. Do you see what I mean?
So, you might want to have more than a few logs of clearance actually or that is one of the ways of thinking about it.
Now, in terms of how we selected these numbers we really used a combination of the data that is available to us to try to identify types of steps that are likely to result in greater or smaller amounts of clearance. This is a very sensitive parameter. At least it was in the Factor 11 risk assessment and we felt it was important to run this range to really see what happened and to see whether or not this made a difference in terms of the products.
DR. BROWN: We need I think to understand are you talking here about concentration or total mass? Are you talking about 9 logs of clearance in a unit or are you talking about a 9-log clearance per ml?
DR. SCOTT: Actually, Paul when I was talking to Dr. Leitman I was just making an example but usually we talk about per ml is my understanding.
DR. BROWN: Again, I don't understand the reason for the higher values in terms of are you modeling again or have I missed a beat here?
DR. PRIOLA: Dr. Epstein?
DR. EPSTEIN: Let me just clarify that the way the model works dilution doesn't affect the outcome of risk per dose or risk per vial because in the end you ask how many infective units are in the pool and then how many end up per vial or per dose.
So, the fact that you may have diluted it initially in the pool doesn't affect how many absolute infectious units would end up in a vial unless there is clearance. So, whether you describe the clearance as per ml or per total unit it doesn't matter. What you are computing is the residual number of ID50s per vial as mass, as total.
DR. BROWN: It does matter if you are talking processing reduction because that requires a knowledge of what you are aiming at. Are you aiming to get 9 logs reduction in a unit or 9 logs per ml? They are different.
DR. EPSTEIN: Okay, the goal in processing is per ml, but the point I am making is that the dilutional effect of pooling has no effect whatever.
DR. BROWN: I wasn't talking about pooling at all. I agree.
DR. EPSTEIN: But the earlier question and I forget who asked it was whether the clearance was above and beyond the effect of dilution itself because you have 3-log dilution and the answer is we are speaking about clearance above and beyond any dilutional effect, that the dilutional effect doesn't affect the risk estimate.
DR. BROWN: Right. As it is stated I agree. It is a different topic, but my question still stands particularly now if you are talking about per ml which is a concentration why would be the slightest bit interested in this 7-to-9-log clearance? I mean we love to have it but you are not going to because nobody in the history of the field has gotten a 9-log clearance per ml of anything by any method.
DR. SCOTT; I think it is pretty hard to estimate a 9-log clearance actually with the amount of infectivity that you can put into something but it seems possible. I think that the reason we have a 7 to 9 is really that reflects more of an 8 which really reflects a combination of two high-level clearance steps and in fact that apparently has been observed in studies.
DR. PRIOLA: But that has been observed in studies using spiked brain material which is much, much, much higher than you will ever get in blood.
DR. EPSTEIN: First of all Dr. Scott gave you our answer which is it comes from combining so-called "orthogonal" procedures but I think there is another point that needs to come across here which is whether when you get below 1 ID50 per dose what does that mean; are you now just talking about the probability that there is a residual infectivity or are you saying that no one gets an infectious dose, and we don't know really how to resolve that. So, just to be a little bit concrete let us say that a person, let us say that the clearance is such that infectivity per dose is .01 ID50; are we then saying that no on gets an infectious dose because 1/100 of an iD50 infects no one or are we saying that the risk of exposure to 1 ID50 is 1 percent and it is the latter model that we are putting forward and why that matters then is for patients who are repeatedly dosed.
Remember we are talking about patients who may have hundreds of exposures per annum and thousands per lifetime and if the probability of receiving an ID50 is low per single exposure it may on the other hand not be low for cumulative exposure. So, it matters whether it is a probabilistic model and for that reason very high levels of clearance if demonstrated may in fact matter ultimately to patient safety.
DR. PRIOLA: David?
DR. BOLTON: I agree completely but the problem is now not so much in the model. It is when you take this to reality because to demonstrate anything more than maybe a 3 log clearance with natural prions in the natural fluid is going to be impossible and when you do the brain spiking even microsomal fractions, I am not sure that anybody here that has done those studies can really believe that they represent what would really be happening in naturally infected blood. It is just that we don't know. I mean that is one of the things that if we would ever get that information what is the natural physical state of variant CJD agent in human blood that would be a tremendous piece of knowledge to have because when you take a brain homogenate or brain microsomal fractions and you spike them into a sample of blood to get a titer of 10 to the 7th or 10 to the 8th it is just we have no idea that that is in fact representing reality.
DR. BROWN: Well, you do if you know it is not. It is very simple, but this leads into what I think is going to be another question coming up. Both exogenous, that is spiked experiments and endogenous experiments are complementary and you get from one something and you get from the other something and they both should be done. In other words if you are going to be talking about optimizing validation studies what you want is validation using both because the spike study is going to give you maximum range of possibility and the endogenous study is going to give you absolute relevance so that even if you know you are only eliminating 2 logs in an endogenous study at least you have shown that what you have done applies to the endogenous infection.
So, there is contradiction and confusion. It would be terrible, for example, if you got a 7-log reduction in the filtration and then discovered you didn't get anything in endogenous. I mean I could imagine that happening if there were some size problem but doing both you have got all of the information that you can possibly have and you should do both. No one should ever be satisfied with a spike study for validation today.
DR. PRIOLA: So, in a way that argument applies to this question as well then as you alluded to because you can accept that higher range as being experimentally the most that you can clear using an artificial system.
So, in that regard this range of logs would be okay. The low end is the endogenous more realistic one and the high end is the experimental spiked one and a limit of what you can do, okay.
Does everybody on the Committee agree with that? Yes, so would the Committee agree based upon what, I think that is an excellent argument, based upon that argument that we keep the ranges as proposed by the FDA?
Okay, let us move on then.
DR. SCOTT: It is actually linked to the last and I think that Dr. Brown was alluding to it. What experiments might enable refinement of the clearance estimates and allow comparison of clearance offered by various steps in the methods used to manufacture plasma-derived Factor 8?
DR. PRIOLA: Okay, so we can I think use as our basis what Dr. Brown just said that the spiking experiments are fine as long as they are accompanied by some simulation of an endogenous spike which it is not exactly clear what sort of endogenous spike you would use except for blood from a naturally infected animal model.
DR. BROWN: It would be very nice when for example sheep because everyone is a little nervous about basing human therapy on mice with cause but when we get another species or two possibly even primates if we are very lucky if we can get some idea of the concentration of infectivity in the blood in non-rodent species we will be much better off in terms of knowing what really are the maximum levels of infectivity in the blood in endogenous infections.
Today there is no estimate, no fact, no observation of the amount of infectivity in any species other than the mouse and the hamster to the best of my knowledge and that should change I would hope in the next year or two.
DR. PRIOLA: That is because that requires titering experiments that are very long term and expensive.
Any other comments as to this question?
I think the issue of using multiple animal models comes up later as well. If there are no other comments from the Committee we will move on to whatever the next question is. I have lost count. I am sorry.
DR. PRIOLA: Question 7.
DR. SCOTT: Only the numbers have been changed, but actually for this question the proposal has also changed, that is it is really just I think more fully described and fleshed out. What data should be used to estimate how much Factor 8 is used by typical patients? This is from Dr. Weinstein's slide. We plan to analyze data from the ongoing UDC survey to estimate the numbers of patients using specific product brands and the distribution of disease types that is severe, moderate, mild hemophilia and type 3 von Willebrand's disease and we also plan to extrapolate data from the 1993 to 1998 survey in six states to estimate the total number of US patients and product consumption per patient with stratification by clinical setting and if there is inconsistent information from these two analyses it will need to be reconciled using patient-based medical record data.
DR. PRIOLA: So, basically you have three different routes to this question, right, three different proposals for this, that you can use all three of these data sets to address this?
DR.SCOTT: Yes, I think they are not exclusive. They really ought to be used --
DR. PRIOLA: Together. So, there is real data to be used which is very nice. Does the Committee have any comments? Do we all agree with the FDA proposal? It seems pretty reasonable.
There don't seem to be any more comments. So, let us move on to question 8.
Oh, hold on, sorry, Mr. Bias, excuse me.
MR. BIAS: I don't know how this might impact but if there are patients who are on immune tolerance therapy how do you account for them in the model?
DR. ASHER; That is a topic. Actually we didn't bring that into play here but some of the information is available that is as far as the number of patients on immune tolerance, we do have information from the CDC, UDC collection for numbers of patients immune tolerance. We would also again have to make an approximation from essentially literature review I think at this point to get an estimate of the amount of material that would be used there but those of course would be patients who are using very high levels of product. There are relatively few of them compared to some of these other population groups but that is another group that we consider and we would be able to I think get data probably more of a, the quantitative data is difficult to get hold of for those patients but probably with the review of current practices from physicians and looking at various protocols for treating immune tolerance we could extrapolate that information.
MR. BIAS: That would be good. I am glad you can get that and my other question would be occasionally a patient like myself has an accident in a particular year. Does the UDC data also account for those spikes in usage? For instance this year my monthly average tripled for the 6 months I was injured. So, that is a big spike.
DR. ASHER: There is information from the UDC data. There is a category of how many times people have been under say a prophylactic treatment or use but we don't again have quantitative data about the amount of product being used in those particular situations which is somewhat of a limitation and the UDC data as it is currently constructed.
DR. PRIOLA: Can you get that quantitative data?
DR.ASHER: It simply isn't being collected right now. I had inquired in fact whether an additional question might be put on the survey form of the UDC data and the opinion was that the information would be not very accurate, that it was felt that it just may not be sufficient for our purposes. We are going to get more quantitative data from the survey information.
DR. ALLEN: I certainly think that there are probably alternate sources of data that could be used on a survey basis to estimate the extraordinary needs that sometimes arise and I assume also and I forget from the presentation, but data sources do show recombinant factor usage versus plasma-derived?
DR. ASHER: Yes, we can get that but of course we are primarily interested in the plasma-derived materials.
DR. PRIOLA: Dr. Bracey?
DR. BRACEY; No, I was just going to comment on the possibility of using alternate sources as well. I think we have heard that there are probably some very good alternate sources available and if there are important data elements I think that we should go elsewhere.
MR. BIAS: I am wondering if the delivery industry, if you have ever tried to use them as a source but some of the larger providers. I mean there is not a unit of clotting factor that goes out that isn't tracked.
DR. ASHER: The question that I was uncertain about we had in fact discussed that element here but I was not certain about whether or not there was a per patient figure, in other words a delivery system may know in the aggregate, are you talking about a distributor or a home care company?
MR. BIAS: Yes, they will know per patient. They will know exactly per patient and they are very happy when it spikes. They produce a monthly report that indicates exactly how much each person uses and they will have very accurate figures on whether it spiked.
DR. ASHER: That is a good point.
DR. PRIOLA: Yes, excellent. Thank you. So, that is another point for data collection for you to put in the model.
Okay, let us move on to question 8.
DR. SCOTT: The next question is what is the effect of plasma pool size, that is the number of donors per final product for Factor 8 recipients and we proposed to estimate plasma pool size as a range between 20 and 60 thousand donations with a bimodal distribution to reflect expected source and recovered plasma pool numbers but we feel that we should or do need to seek additional data from plasma fractionators to get a better sense of the amount that we have in the lower and the upper range, and I think in terms of the way the question is worded, what is the effect of plasma pool size this goes back to Steve Anderson's presentation where he showed the information or the modeling that indicates more frequent treatment sort removal overall of the effect of plasm pool size.
DR. PRIOLA: So, you are proposing this as a starting range really for your model knowing that you will, assuming that you will be able to get better data from the plasm fractionators eventually?
This is just a beginning range for you, the 20 to 60 thousand?
DR. SCOTT: The 60,000 actually represents a voluntary ceiling for number of donors contributing to a pool and 20,000 is based on an estimate of what we think for the lower end of plasma pool sizes really based on ad hoc observations.
DR. PRIOLA: Dr.Leitman, do you want to comment on that at all? No? Other comments? Dr. Salman?
DR. SALMAN; This is for clarification and excuse my naiveness. How many donations do you need then of one single pooled unit?
DR.SCOTT: Each manufacturer really has their own unique batch size but I think the lower limit in the CFR is 1000 or something like that but typically, I am not sure I understand your question, how many do you need.
DR. SALMAN: When you have one pooled unit how many donors will contribute to that pooled unit?
DR. SCOTT: Right, so the manufacturing pool is comprised is plasma of a lot of donors and if you want to a 500-liter pool and you have recovered plasma and you have say 200, 250 cc's per recovered plasma you can just do the math and figure out how many donors that you would need to make up that pool size.
DR. SALMAN: But that will be a very important question for the model and maybe Dr. Anderson can answer that as we heard like the model will take that into consideration. So, the number of the donors contributes to the unit will be unimportant aspect to the model especially when you look on one of the three slides. You said they are very important and you skipped them and --
DR. ANDERSON: Do you want me to actually show those slides?
DR. SALMAN: One of them you said, "Well, there will be 60,000 donors pool and 10 percent of those may be contaminated.
DR. ANDERSON: Yes, if you had a disease prevalence of 1 in 500,000 and you had 60,000 donations per pool it would take about, in 10 pools one of those potentially would have an infected unit, contaminated unit.
DR. SALMAN: If I understand the model and you present it well I think what will contribute to the contamination is the number of the infectious donors that participate in the pool.
DR. ANDERSON: That is correct.
DR.SALMAN: So, the number of the donors for that pool will be an important aspect. Is that correct or not?
DR. ANDERSON: This is an important aspect of the model and we are specifically modeling it. So, let me just sort of emphasize that and we have one option that we put up here is we have a range of 20,000 to 60,000. So, we are going between that range based on the information that we have.
The other thing we can do is we have recovered plasma that we are interested in and source plasma. We could generate different distributions for each of those. The problem here is that we have very limited data by which to make these estimates and what we are doing is we are saying that we know it is between 20,000 and 60,000 and our anecdotal information is that most pools lie at for the most part at either end of these ranges. So, we are proposing a bimodal distribution. So, most of them will be either 60,000 or 20,000 in our model.
So, it is important and it is figured into the model.
DR. SALMAN: But is that reflecting reality like in any given pool unit? I am just asking the question for people who --
DR. ANDERSON: Manufacturers from our understanding different manufacturers have different pool sizes from which they make these products.
DR. BROWN: I think about 6 or 7 years ago when this whole thing with the FDA and precautions and so forth got going at that time manufacturers were using 100,000, 150,000 units for a plasma pool, sometimes. It depended on the product they were particularly interested in. After this story broke it was proposed that that was a too high limit and since then I believe in fact I think all manufacturers, there was a guidance. Is that not right?
DR. ASHER: No.
DR. BROWN: There was no guidance?
DR. ASHER: There is none, not a guidance.
DR. BROWN: There was advice. There was something in the air, don't go over 50 or 60. Is that right?
DR. ASHER: Sixty. That is again a completely voluntary --
DR. BROWN: Exactly but manufacturers pay attention to things that are in the air when they come from the FDA and so you can bet your bottom dollar that manufacturers paid attention to that. I know they did and virtually the maximum now that is used is 60 and I think this is probably a very realistic range, 20 to 60 at the moment.
DR. ALLEN: In actual fact from using source plasma donors however, you might have multiple donations from a single source plasma donor. So, if you had one infected donor there may be multiple donations from that one donor that go into a single pool.
DR.BROWN: That is actually an interesting questions. Would they in fact go into the same pool if they were donating say, I don't know once every 2 weeks? Sometimes they donate once every 2 weeks. So, they could wind up in the same pool.
DR. ASHER: And that is something we can model as well.
DR. SCOTT: I think part of minimizing the donor pool is actually trying to use some of the same donors in pooling from certain selected centers into one pool all the time. You see what I mean? So, it certainly could happen. It would be consistent with current practice.
DR. BROWN: I think the rationale was to limit the damage. If you had a pool that was contaminated with smaller pools it would be distributed to fewer people. I think that was the clear --
DR. EPSTEIN: That is correct for the infrequent product user. The problem that you get into is that the chronic product user will simply be exposed to more product lots made from a larger number of smaller pools and so those phenomena offset each other which is part of our motivation for trying to look at an annual patient risk but for the rare or infrequent product user yes, a smaller pool would have a lower probability of having an infectious donor.
DR. BOLTON: I just have a question. Is there a uniform relationship between the number of donated units and the units of Factor 8 that are manufactured from that? There is a critical relationship there in terms of translating donations to product vials.
DR. ANDERSON: Right, and we have that actually from the literature and manufacturer information. So, we have a range actually for that and I actually don't recall what that range is offhand, but we actually have put a range in for that estimate and it is using a yield calculation essentially.
DR. BOLTON: Is it pretty uniform across the industry or is there --
DR. ANDERSON: There is a little bit of variability from what we can see from the information that we have. So, we are incorporating that into the model.
DR. PRIOLA: Dr. Leitman?
DR. LEITMAN: This is getting off on a tangent but isn't there a voluntary hold practice in the industry so that a donor's plasma is held 6 months until their next visit to confirm they are not in a window period?
PARTICIPANT: Sixty days.
DR. LEITMAN: Sixty days. I knew there was a six in there. So, potentially a donor's unit could get into the same lot 60 days apart?
DR. PRIOLA: Okay, any other discussion from the Committee?
MR. BIAS: It is exactly because of that hold that you can get several infectious donations into a pool and because there is no rule, hard and fast rule we don't exactly know how many factors are handling that on site. We learned from the eighties that it was possible for them to in making batches leave a little bit of a batch from a previous batch at the bottom of the pool that would increase the infectivity if that previous pool was infected and previous batch was infected and therefore instead of 60,000 you had a pool that suddenly had donations from 120,000 people in there and because there is no hard and fast rule we have sort of a gentlemen's agreement with the manufacturers that they are going to lower the pool size. I would be very concerned that that still occasionally happens because it is a manufacturing process and a manufacturing process is one that produces a product that produces something they are going to sell and become income and they are certainly not going to pour it out.
DR. SCOTT: I can say that that practice is highly discouraged and I, personally, am not aware of any use of tailings anymore and it is an inspectional issue. Anybody that is found to be processing things this way will definitely get a problematic inspection.
MR. BIAS: I am sure they are all playing appropriately in the sand.
DR. ASHER; I want to just make certain it is on the record that this again is a voluntary system and the 60,000 is voluntary. It does not mean that there are no manufacturers that have gone above that.
DR. PRIOLA: Okay, so, are there any other comments from the Committee?
If not do we agree that it seems to me the range that the FDA proposes is okay?
All right, let us go on to question 9.
DR. SCOTT: Can a cumulative effect from repeated exposures to low doses of the vCJD agent be incorporated into the risk model and we propose to allow for the theoretical possibility of cumulative effects by having the model provide a cumulative risk for a 1-year period for these different types of patients.
DR. PRIOLA: Dr. Brown?
DR. BROWN: Yes, that is a good idea. In fact, the good news or shall I say the bad news first? The bad news is that in our model it can happen. It has been shown. Now, we were wondering about that for a long time. The answer is now on the table.
The good news is that despite that fact hemophiliacs are not dying and that is another way to look at it but absolutely it is almost more than a theoretical possibility now. It is something that really has to be included in any model.
DR. PRIOLA: Any other comments from the Committee?
So, we are in agreement that the 1-year cumulative is a good idea. I think it is.
All right, let us go on to the final question, question 10 which is the voting question.
DR. SCOTT: I don't think we will have a final solution today but we do want to understand the Committee's feelings about everything that you have heard and this is a question that we are asking you to consider now.
Given the present scientific uncertainties that you have heard about today in the underlying assumptions of the Factor 8 risk assessment do you believe that the risk assessment model could provide a useful basis for risk communication to patients, their families and health care providers?
DR. PRIOLA: Dr.Salman?
DR. SALMAN: I think the short answer is yes but I will put a condition like the sensitivity analysis should be done as part of the risk assessment model and that should become part of the communication with the public.
DR. PRIOLA: Dr. Brown?
DR. BROWN: It really depends on how well it is articulated to the patients' families and health care providers. We can't expect them to understand clearly what has been happening this morning with all the caveats and this, that and the other thing but the fact is that all the evidence to date indicates that sporadic CJD No. 1 is not associated with infectivity in the blood, the evidence to date.
Second, to date we have no cases of variant CJD in this country.
Third, the infectivity present in cone(?) fraction in the precipitate already has a low amount of infectivity. In fact, four is that processing currently in place for Factor 8 is more than adequate to take out any infectivity that might theoretically have been present.
I think the risk assessment will probably validate this overall scheme of safety.
DR. LEITMAN; I just want to second that. When you start to talk to patients and their families about risk assessment they assume you are talking about very real risk. This is theoretical risk or hypothetical risk because there has not been a case even in the highest-risk population which would have been UK hemophiliacs before the screening procedures were put in place.
DR. ALLEN: Let me concur with the previous comments and say that I think the FDA needs to move forward with this model development and to look very carefully at the results that come from it.
The subsequent steps as one tends to want to go public with the information, however, are to look carefully at how you approach this with the media because you can't keep it just to patients at risk, providers and the small community. It is going to and has got to be involved with the general media and I have got real concerns about that because they like to hype everything regardless of what the actual risks are.
The experience we had with HIV infection more than 20 years ago now clearly tells us that we can't just sit on this and wait. I think the risk is likely to be extremely low. Fortunately we have had enough experience. We have got much better surveillance systems than ever in the past but the answer is not going to come down to zero risk. We know that and we are going to have to look very carefully at how this is communicated so that it is useful and reassuring and educational rather than frightening.
DR. BRACEY: Again, I concur with the previous statements and I think the statement that was made earlier was very important and that is that there are groups already that have been discussing the risk with the members of that community and I think that rather than start anew it would be good to try to partner with those individuals to continue the counseling that has already begun.
DR. PRIOLA: Mr. Bias?
MR. BIAS: Although I want the risk assessment developed I have real concerns about how it gets communicated to patients and health care providers.
I am concerned about the possible stigmatism to patients as related to not their primary caretakers at per se hemophilia treatment centers or someone who is familiar with their background in hemophilia but their outliers; one of the weaker parts of comprehensive care is dental care. If their instruments are at risk they may choose not to treat patients with bleeding disorders. I think that other agencies within the government have to be alerted and have to, if we are going to publish this information there has to be some provision so that patients can continue to be treated and guaranteed that treatment on some level.
States are moving toward preferred product lists where they are limiting patients to one type of clotting factor and we are fearful that in some states they are going to select a plasma-based product because it is cheaper and if that is the case that is going to leave that family without any alternatives for care.
So, I am very concerned about how this risk assessment gets applied to the public and any slow news day, we are in a 24-hour-7 news cycle now. If the earth isn't cracking open this will be the major story of the day and it will run. If it is a holiday weekend it will run for 3 or 4 days and what you will have is a group of patients who are without care and without access to care and without alternatives because the Federal Government hasn't protected their right to have care or have access to other product choices, if their state says that this is the product that we have for you.
So, I am very concerned about how this is applied and I would strongly recommend that before this information is published in any way that there is major consultation with hemophilia organizations both in the world and nationally so you can get their perspective and guidance as you go forward and in addition to that that we work with HHS to ensure that there are going to be alternatives for patients to continue to receive care including saying to medical providers, "You don't treat these patients; you don't have access to Medicaid."
DR. PRIOLA: Dr. Hogan?
DR.HOGAN; Relative to that I am sure we are going to be hearing about the results of this model in this Committee. I think it would be possible and we would ask the FDA for this Committee to discuss those results relative to how they would be dispensed.
Obviously this is an open forum and the media is here but we can certainly stress that there are the uncertainties that are involved and hopefully have some sort of oversight as to what Mr. Bias is talking about.
DR. PRIOLA: Dr. Johnson?
DR. JOHNSON: I am concerned about the same thing that Val was talking about and that is the nature of the audience this is being released. There is a huge percentage of people who are well educated, well informed in this country who don't differentiate mad cow disease from anything else and think it is here and the level of confusion they have already undergone is enormous. Trying to explain something like this on top of it reasonably is going to be very hard.
I would like to hear Florence's comments on that. She deals with it every day.
MRS. KRANITZ: Thank you. i totally agree with Val. I could not agree more and as simple as you may make the explanation or as hard as you may try to show how this risk model made the assessment and realities of it, you still are going to have some panic, probably a lot of panic on the part of not only the patient but of the health care provider.
So, before you take on the project of informing publicly any part of the population you need to know that you might even have to do risk assessment on top of risk assessment as to what possible damages you are going to create by releasing this information.
DR. PRIOLA: So, the basic point that the risk model could be a basis for communication to family and health care providers, that is okay, and it is the way in which that information is communicated that is the biggest concern. Okay.
Any other comments?
DR. BRACEY: This may be a somewhat naive question. I think it may be a matter, well it is a matter of perhaps economics and a matter of supply, but you know the entire issue revolves around the current methods for preparing the product. If we were talking about recombinant products aside of course from the risk of the albumin that it is resuspended in I think we would be having a different discussion and so one of the things that I wondered about and I would like to hear from Mr. Bias or Val about this is if you indeed did have this risk model that you could address that looked at plasma, recovered Factor 8 wouldn't that potentially be useful in making an argument to bolster the use of the recombinant product?
MR. BIAS: It probably would be but my concern is the reality that we are facing on the ground and currently we are in a battle state to state to maintain the access to the care that we have and I am just concerned that without the force of the Federal Government behind it the publication of this information would not have the same impact on every patient in every state and there will be people who will fall through the cracks. There will be discrimination and so on and so forth. So, my guidance is that if we are going to release this information that other parts of the Federal Government that are responsible for health care and health care provision also be prepared to put laws into effect, put statutes into effect so that we can guarantee treatment for people with bleeding disorders. This is a disease that we have worked at high cost of lives for many years to make very livable for people. People are living very full healthy active lives now that we have gone to recombinant clotting factor and the plasma products are very clean. It is such a difference from when I was a child to today.
A child today plays on their school basketball team. I was not allowed in the gym and because of the switch of power from the Federal Government to the state government that understanding isn't there. That history isn't there and we see our access to care being rolled back in states all over the country.
So, I just want to make sure that if we are going to release this kind of information it can be used as an argument but we have got to have the Federal Government's power behind that argument.
DR. BROWN: The text says as I read it carefully, "Provide a useful basis," and let us vote.
DR. PRIOLA: Dr. Weiss, you had a comment you wanted to make?
DR. ASHER; Yes, I just wanted to clarify the element here that in fact there are no recombinant von Willebrand's containing factors and so there is a definite need for plasma-derived materials and secondly there are current studies going on there about potentially the advantage of using plasma-derived for immune tolerance. It is unclear whether or not this is really preferable to recombinant but there is some evidence that is being investigated now.
DR. PRIOLA: Let us go ahead and vote on the issue because we are voting on is it the basis for a reasonable risk communication.
DR. FREAS: For the record there are 17 voting members at the table. Dr. Bracey is a non-voting consultant at this meeting.
I will go around and call the roll.
DR. BOLTON: Yes.
DR. FREAS: Dr. Johnson?
DR. JOHNSON: Yes.
DR. FREAS: Dr. Telling?
DR. TELLING: Yes.
DR. FREAS: Dr. Creekmore?
DR. CREEKMORE: Yes.
DR. FREAS: Dr. Lillard?
DR. LILLARD: Yes.
DR. FREAS: Dr. Sejvar?
DR. SEJVAR: Yes.
DR. FREAS: Dr. Hogan?
DR. HOGAN: Yes.
DR. FREAS: Mr. Bias?
MR. BIAS: Yes.
DR. FREAS: Dr. Allen?
DR. ALLEN: Yes, with reservations noted during the discussion.
DR. FREAS: Dr. Priola?
DR. PRIOLA: Yes.
DR. FREAS: Mrs. Kranitz?
MRS. KRANITZ: Yes.
DR. FREAS: Dr. Geschwind?
DR. GESCHWIND: Yes.
DR. FREAS: Dr. Leitman?
DR. LEITMAN: Yes, with the reservations noted during the discussion.
DR. FREAS: Dr. Gaylor?
DR. GAYLOR: Yes.
DR. FREAS: Dr. Ghetti?
DR. GHETTI: Yes.
DR. FREAS: Dr. Salman?
DR. SALMAN: Yes.
DR. FREAS: Dr. Brown?
DR. BROWN: Yes.
DR. FREAS: The vote is unanimous.
Agenda Item: Topic 2: Labeling Claims for Filters Intended to Remove TSE Infectivity from Blood Components
DR. PRIOLA: Okay, let us move on to topic 2. The Committee will have noticed there is no break scheduled for the afternoon. So, I have asked the speakers to keep on time and if possible we will try to take a 10-minute break or something after the first couple of speakers if everybody is on time. We are about 20 minutes behind time now which isn't too bad.
So, our first speaker of the afternoon session will be Dr. Vostal.
Agenda Item: Prospects for Reduction or Removal of TSE Agent Infectivity from Blood Components by Filtration and Criteria for Allowing Claims: Introduction Jaroslav Vostal, MD, PhD, OBRR, CBER
DR. VOSTAL: Thank you for the opportunity to share with you some of our thinking in terms of validation of labeling claims for TSE reduction studies with blood processing filters.
Now, my talk actually starts off with several introductory slides and since these topics have been very well covered earlier today I am just going to go through them very briefly.
As you can see this is a brief schematic of the prion protein. It points out that there is different conformation of the protein. One of these is the pathologic form of the prion. It has protease resistance and less soluble.
The main point of this would be that TSE infectivity can be present even in the absence of PRPSC and so therefore PRPSC is only considered as a surrogate for TSE infectivity.
Now, this is a schematic of the organs of an animal and in terms of pathophysiology just very briefly if there is oral inoculation the infectivity goes through the LRS system and through the peripheral nervous system and migrates into the central nervous system.
As was discussed by Dr. Asher earlier today these models have demonstrated that there is infectivity in rodents during the asymptomatic phase of the disease and so it is pretty much agreed that at least in the rodent model there is infectivity in blood.
Those earlier experiments raised concerns that there is transmission by blood transfusion particularly for BSE. This was confirmed by the key experiments by Houston and Hunter where they used the sheep model and they had an asymptomatic BSE-infected sheep. From this animal they were able to collect a full unit of blood and transfuse that to a healthy sheep which then went on to develop or some of these animals went on to develop BSE.
This model is interesting in that respect that it uses a full-size animal that is capable of donating a full unit of blood that can then be processed by the devices that we are going to be talking about later on today.
Moving on from the animal experiments there is now currently epidemiological evidence for variant CJD transmission by blood transfusion in humans. This was reported by Dr. Llewelyn and Dr. Peden. As we already heard earlier today there is a national CJD surveillance system in the UK that identified 48 individuals. Actually these are numbers taken from these publications. So, those numbers are slightly different, current numbers are slightly different. Basically there were 17 individuals identified that are still alive who received products from 15 donors who later became diagnosed with variant CJD. Two of these living recipients were subsequently diagnosed with variant CJD. One died from symptoms of vCJD. One died of unrelated causes and was later identified as having PrPres in his spleen and both of these patients received non-leukoreduced red cells.
So, if there is infectivity in blood, TSE infectivity in blood we are very interested in looking at devices that are capable of reducing this and what we would like to do is to establish a system to validate the claims for these devices.
Some of the issues to consider for validation of the devices are the distribution of TSE infectivity in blood in humans or in the particular animal model, whether this infectivity is cell associated, whether it is intracellular or extracellular, whether the infectivity could be free floating in plasma and then we have to consider the physical attributes of infectivity in plasma, whether these are aggregates, fibrils or microvesicle.
We need to consider the interaction of the individual units of infectivity with the devices and finally to look into the distribution during and after processing to make sure that the devices actually eliminate infectivity and do not only distribute it such as would happen if you had a microvesicle formation from the infected cells.
So, the steps to validate the TSE claims of these devices would include in vitro spiking studies and this is as was already discussed before, these involve spiking of brain material usually homogenate into these products and detection of infectivity either by surrogate markers for PrPres, PrPsc or by bioassay again in an appropriate animal.
Complementary to this is a model that uses endogenous TSE infectivity and here because the levels of infectivity are so much lower usually the detection is done by the bioassay.
Now, when we discuss animal models there is always the question of whether these are predictive to the human situation and some of the things to consider in terms of deciding whether these animal models are predictive are the comparability of the animal blood to the human blood, for example, to look at the type, number, size of these cells and the physical properties of the blood cells. Also, we need to look at the interaction of the animal and human blood cells with the different materials that they are going to be exposed to and also there are questions about the transmissibility or infectivity of a TSE agent that can be influenced by the strain of the agent, the dose of the agent, distribution of infectivity in blood particular to the agent and also and sometimes we are concerned about the distribution of normal prion, whether that can influence a distribution of infectivity in blood cells.
Now, these are just a small table comparing the hematologic values for various species. The species that we have listed here are human, mouse, hamster and sheep.
You can see the human value here on the bottom line, the red cell volume and the platelet count. For the most part the platelet counts in these animals are equivalent and so are the white cell counts. The key differences between the human and the animal models comes in looking at the size of these cells. The human cells have about close to tens of liters of volume per red cell. You can see that in the mouse it is about half and particularly in the sheep these are very small cells and we always wonder whether these cells, whether the size influences the way those cells are processed by the devices and whether they can influence the way the infectivity can be distributed after it has been processed by the devices.
It appears that the small size of these cells is compensated by the increased number of these cells. So, the sheep and the mouse have a significant number more red cells than you find in human blood.
This slide talks about the normal protein expression on blood cells of different species. This is the difference. We are wondering whether the expression of prion protein on these cells would make a difference because there is a physical association between the pathologic conformation and the normal prion proteins in terms of propagating the infectivity.
So we looked at distribution in human cells by flow cytometry and for humans there is expression on platelets, erythrocytes, and particularly on lymphocytes and monocytes.
If you look at some of these animal models that have been used to show TSE infectivity in blood it is striking how different they are in terms of prion protein expression, particularly for the hamster. We were not able to detect any prion protein expression. The mouse has comparability at least in red cells and sheep again platelets and red cells are devoid of prion protein. So, we are not exactly sure how this can influence the distribution of infectivity but it is something to consider when deciding the appropriateness of the animal model for this purpose.
So, besides looking at validating the removal of TSE infectivity these devices will also have to be evaluated for their impact on the transfusion product quality. The evaluation of the red cells, platelets and plasma will have to be done after they are processed through these devices and the FDA follows a standard evaluation approach to each transfusion product based on previous experience with the devices that process transfusion products.
For example, we have a lot of experience with leukoreduction filters and the next couple of slides will just share with you the evaluation process that we go through to evaluate leukoreduction filters.
So, leukoreduction is the process of reducing the total number of leukocytes in the transfusion component. The methods that are employed are filtration or apheresis.
These leukoreduced products have been associated with the reduction of febrile non-hemolytic transfusion reactions, alloimmunization and reduction of CMV virus. Even though there is an association or beneficial aspect of these products none of the manufacturers actually claim these beyond meeting the criteria for leukoreduction and in the US the criteria for leukoreduction is less than 5 times 10 to the 6th leukocytes per transfusion product and in Europe this is reduced to less than 1 times 10 to the 6th leukocytes per transfusion product.
So, to validate leukocyte reduction filters for efficacy we look at the quantitation of leukocytes in a particular blood product before and after filtration. So, there is leukocyte count and the whole blood, red cells, platelets and plasma before and after.
We, also, look for definition of the timing of leukoreduction from the time of collection of the product. This is because leukocytes actually disintegrate within several days and in order to be able to remove the whole cells it is best to filter early on in the storage period.
We, also, explore the effects of temperature on filtration efficacy whether it is done at room temperature or cold temperature and we also look at validation of the efficacy for a particular anticoagulant which can influence the way those cells perform as they are being processed through the filter.
In terms of validating these products for safety we look at biocompatibility and integrity of the materials. We look at their effect on cellular products. In particular we have criteria for in vitro recovery which is 85 percent. This means that we expect to see 85 percent of the products, of the red cells that are passed through that product to be recovered after filtration. We also have criteria for hemolysis at the end of storage and it should be less than 1 percent and we do in vivo recovery of radiolabeled cells in normal human volunteers and this is done for platelets and red cells.
For plasma we look at levels and function of plasma proteins and also complement activation.
So, this slide here summarizes our proposal for validating a claim for reducing TSE infectivity in human blood products. We would like to see a demonstration of a reduction of endogenous TSE infectivity by bioassay in two animal models and we suggest that this would be a rodent model as well as a sheep model.
We like to see a full-scale blood unit and leukoreduction filter used. We would like to see the TSE infectivity to come from a BSE or variant CJD strain of TSE diseases.
The reduction of PrPsc in blood products will be considered supportive but not sufficient for a claim.
Study should be performed at two separate sites to minimize the issues of cross contamination and differences in laboratory practice and finally study size should be sufficient to support statistically valid conclusion from those experiments.
So, that is my introductory talk. Here are the questions that we are going to be asking you to comment on today. The question is are the FDA proposed minimal criteria for validation of TSE infectivity reduction by filtration adequate and appropriate? And we would also like you to comment on the following points. The rationale for use of specific animal models to study the properties of blood-borne TSE infectivity, are the experiments in rodents sufficient or should experiments also be done in sheep?
Is it necessary that each experiment should be done in two separate laboratory sites to ensure the reproducibility and accuracy of the clearance and are general descriptions of informative scaled-down processes for reducing TSE infectivity in blood?
There are several more comments that we would like your input on, for example, levels of clearance acceptable for claims of reduced TSE infectivity in blood components as used in clinical settings, the estimated logs of clearance of TSE infectivity required to conclude that blood filters have effectively removed infectivity from blood components and the methodology appropriate to use in evaluating TSER agent clearance, bioassays for infectivity, Western blots or other assay for prion proteins.
So, those are comments to the initial question. The follow-up question is does the FDA's proposed labeling for a filter meet the appropriate criteria for a claim of reduction of TSE infectivity in blood or blood components and here we have several options. We have an option A which is this filter or device has been shown to reduce TSE infectivity in blood from an infected animal model. Now, A would then be coupled with labeling C which is a disclaimer that states that due to lack of feasibility studies have not been performed to validate this claim in the human population and the other option for labeling would be using part B which is this filter has been shown to reduce transmission of TSE infectivity by transfusion in an animal model and this again would be coupled with the disclaimer C.
So, if you can help us out by commenting on some of these issues that we presented we would be very appreciative.
DR. PRIOLA: Thank you, Dr. Vostal.
The next presentation will be by Dr. Marc Turner who will discuss evaluation of prion reduction filters.
Agenda Item: Evaluation of Prion Reeduction Filters Mark Turner, MB, ChB, PhD, FCRP(Lond) University of Edinburgh
DR. TURNER: Thank you very much, Madame Chairman. I am going to speak to you about briefly the UK evaluation process for prion reduction filters. You would probably be aware we are really somewhat on the front line on this issue and that we probably have put quite a lot of thought into these issues over the past 6 to 12 months.
The first comment is that you are probably aware that there are four UK blood services, English National Blood Service. SMBTS, and the Irish Blood Service, the Non-irish Blood Service along with the Seven Irish Blood Service and all accountable to different jurisdictions and so we have elected to approach this problem from the collaborative power of a working group which really is aimed at reducing the points of contact for the companies themselves.
So, they have one joint service rather than five different services also reducing the amount of duplication of effort between the services.
So, the purpose of the group is to get a primary point of contact with any blood services or the manufacturers to foster a dialogue which I think has been very successful with those manufacturers providing expertise and advice from our end of the business on the clinical development requirements we require from these systems to liaise with regarding in-house operation evaluations, to liaise with other parts of the UK, Joint Professional Advisory Committees for Blood Services with regard to implementation of these devices and to ensure that the appropriate UK Department of Health decision making properties are also kept apprised and up to date and in the UK that formally means the UK Spongiform Encephalopathy Advisory Committee or SEAC and also the Committee on the Microbiological Safety of Blood, Tissue and Organs.
Now, this is a, I apologize for the complexity of this diagram but this is a diagram showing you the pathway that we have devised just to simplify it down for your comprehension. It really breaks down into three parts. The first part on the top left here is a process of establishing the technical specifications and that is being done through the JPAC process and there are three lines for that. One is the technical specification around efficacy and the second is the technical specifications around quality and third is an operation and technical specification and that work is being doing and has been brought together and approved by the JPAC approval process.
A very important aspect of this is a risk assessment which has been carried out on our behalf by colleagues in the Department of Health,Economics and Operational Research, Statistics and Operational Research and I will speak to that issue in a few minutes.
Really that is this point now in the process. On the right hand side are what I would describe as preclinical evaluations around independent evaluation of prion removal and also around the independent evaluation of component quality. Those will then lead into the bottom left hand corner which will be clinical safety evaluation studies which I will describe to you.
The key issue from the point of view of the risk assessment was the issue of how great a reduction in infectivity is going to be needed to be clinically useful for us and we have made a number of assumptions around this issue. First of all I think it is worth pointing out that the current generation of prion reduction filters are applicable to red cell products only and not to plasma and platelets at this stage. We assume a red cell concentration in optimum added solution with prior leukodepletion and 10 to 30 mls of residual plasma and I will just make an aside comment here that the two transmission events, e.g., prime ones occurred almost certainly using an earlier generation of red cell products and not only were they not depleted but they probably almost certainly weren't in optimum additive solutions or they would have been in an anticoagulant with probably about 100 to 150 mls of residual plasma.We are making the assumption that total residual infectivity of greater than 2 ID50s transfusing into one infectious dose will transmit for certain and we are also making the working assumption that prion removal is going to occur mainly from the plasma.
Now, in terms of background in risk assessment we have used a starting proposition of 10 IDs per ml of infectivity with the assumption that a 450 to 500 ml unit which gives you between 4-1/2 thousand and 5 thousand infectious doses per unit. We have made a pessimistic assumption of no differential between IV and IC and we have followed Bob Riller's work in assuming for the purposes of this assessment that about 60 percent of infectivity goes forward in the plasma and about 40 percent associated with the leukocytes. So, in general terms there are about 3000 infectious doses associated with a plasma component and about 2000 infectious doses associated with the lymphocytes.
During the component processing as you have heard the units are subject to leukodepletion and although it is not shown but it is true it is about 1 times 10 to the 6th residual leukocytes per unit. In point of fact in practice we are finding that we normally get down to around 2 times 10 to the 5th residual leukocytes, so about a 4 log reduction and that is where you get this residual leukocyte infectivity figure of 0.2 IDs and in addition to that the residual plasma that is probably about 225 mls in a unit before component processing and depending on processing technique that is reduced to somewhere between 10 and 30 mls. So, traditional top-top component processing will give you in the order of magnitude of 25 to 30 mls plasma and a more modern bottom to top processing will leave you in the order of magnitude of 6 to 10 mls of plasma and in some scenarios that differential is actually of some importance. I have to say that those various assumptions that we have bought into with risk assessment are going back to the UK SEAC probably in this month for evaluation. So, they will be addressing some of the very similar questions that you are addressing at this table, ladies and gentlemen and I am vaguely hopeful that they will come up with similar kind of answers but we shall see.
In this illustration you can see that where you are only getting say for example 1 or 2 logs of reduction of infectivity and in the plasma you have still got enough infectious doses to infect the patient for sure given the relatively large volume of plasma or large volume of the product and it is really only if you start getting to around 3 logs of reduction across the prion reduction filter itself that is in addition to any other effects of plasma reduction or leukocyte reduction you start to make a significant impact on the risk of infectivity. As you don't get beyond that in this model you find that reducing the plasma infectivity even further doesn't assist you that much because you have still got the residue of infectivity associated with the residual leukocytes.
So, the conclusions from that assessment are that 1 to 2 logs are likely to give value as clearly highly dependent on their route of infectivity and highly sensitive to that in the plasma. I haven't shown you this but the estimates are that if we were to achieve 3 logs that might amount to 75 percent reduction in the incidence of secondary transmissions in the UK and I think important further reduction in residual plasma could augment reduction in infectivity over the prion reduction filter and therefore the incidence of secondary transmission and so there is an issue as to whether if and when we implement these technologies we should really be looking also to from top to bottom in processing with maximal plasma reduction and in addition any further effect on residual leukocyte count could be of additional significant benefit. So, if your second filter serves to drive down the residual leukocytes that might be of overall benefit as well.
So, that is obvious. The analysis has driven the prion reduction specifications and we are looking for a 3 log reduction to be shown and we then are sponsoring these immunoassays and bioassays. We would also like to see data from endogenous infectivity studies up to a limit or the model and we are asking that the model be capable of demonstrating at least 1 log reduction by immunoblot if possible but certainly by bioassay and asking the manufacturers to look at the issues of process, their levels including we have processed some blood at plus 4 degree C and some at ambient temperature. We need to know the filters work in both those environments, also, issues such as anticoagulant and use of the head height and things like that are found to be important for example in leukocyte reduction and we have asked the companies to propose surrogate markers or process monitoring which is perhaps one of the most challenging issues, but clearly it is not going to be possible on a day-by-day basis to measure infectivity in the blood that we are prion reducing. So, we are looking for surrogate markers. We need to be able to demonstrate parallelism of infectivity in a surrogate marker if at all possible.
Then where possible surrogates present themselves PrPc is an obvious choice but the residual plasma in the red cell concentrate is minimal and therefore the gross levels of prion protein,normal prion protein are very small and it looks as if this might not be possible and so the discussion we are having at the moment is around a variety of different coagulation factors as potential surrogate markers.
Just to comment on the component policy specification we are advising all the various issues that of course were mentioned 5 minutes ago as to the Payee(?) Redbook guidelines which will be very similar to the guidelines here. We are looking at volunteer red cell survival studies. The only point I would draw out is that we have also patterned more detail than we normally would at changes in red cell membrane physiology and red cells do appear to express normal prion protein albeit low levels and probably in a truncated form and that is just a little matter of concern for those who operate by removing prion protein itself and so we are looking at the expression of protein on the surface of red cells before and after filtration and at the other factors which might lead to a reduced red cell survival as well as remember that these red cells have to be cross matched in hospital blood banks up and down the country. So, any alteration in antigenicity would be of concern. I have to say for those that we have evaluated so far we have not really come up with any problems using any of this preclinical evaluation and I guess our main concern at the moment would be that this does take 14, 15 mls of blood out of the pack in addition to the leukodepletion filter which is kind of unavoidable and probably doesn't matter for most patients but for some patients who are receiving large volumes of blood or frequent blood transfusions may have a consequence of them receiving more units at the end of the day.
Now, the UK, SEAC and MSBTO and UK Blood Service chiefs have asked us to carry out an independent evaluation study and the purpose is really twofold. One is to provide some independent replication of some of the key data that is being provided to use by the manufacturers and where possible to extend that data into a more clinically relevant or clinically informative model and certainly I think we would be keen to look at more than one model. Data generated on a particular strain and animal species is generalizable. What we proposed in the first instance is that we will start very close to some of the data that has been provided by the manufacturers. We will introduce it in 3K in hamster homogenate for the exclusion homogenate for microsomal fraction and sonicated fraction assessed by immunoblot and bioassay and in parallel with 301B BSE stedic(?) spike assessed by Western blot if possible and by bioassay and those two studies have now gone out. Because of the size of the studies they have to go out through a formal procurement process. So, the process is being instigated and we expect that to be completed by the end half of the new year. We are expecting data to be available within about 6 months of the initiation of the study. Bioassays are going to take up to 2 years. So, we are hoping that we will have initial data available by the third quarter perhaps of 2006. We obviously also very mindful that we would like to see endogenous infectivity studies probably in a rodent model and in a sheep study. We haven't planned these at the moment. We decided to focus on these initial studies in the first instance as a kind of gatekeeper and we will be putting further thought into these in due course. These will take an even longer period of time to create a sheep study which takes many years and I will come back to this theme if I may at the end. There is clearly a trade-off here certainly from our point of view between the comprehensiveness of the evaluation and the kind of time lines needed to implement these technologies in a timely way if they are effective.
Finally we are running a series of clinical studies and the primary aim here obviously is to look for the incidence of adverse events and adverse immune responses. Study 0 has been commenced or is imminently going to commence in Ireland where we will be putting in the very first instance a model there that is full use of blood from the patients through the prion reduction filter. So, 20 single units will go into patients that are being transfused and 20 double units as it were in trying to keep the patients that are receiving 3 units which may take perhaps to the end of this year, the end of 2005.
Early in 2006 we will be initiating two much larger studies one in patients undergoing complex cardiac surgery, 300 patients in the UK all of whom are to receive the prion reduction filter treated with concentrates. Study 2 is a randomized blinded study in chronic transfusion dependent patients probably MDS patients rather than thalassemic patients because we want to note particularly these kind of patients who could be vulnerable to allo-antibody development and we want to look at that issue specifically and the statistical estimates are that these two studies have around 90 percent power of picking up one additional adverse event in these study populations.
So, that is really all I wanted to say ladies and gentlemen. The baseline assumptions and these proposals will go to the UK SEAC probably at the end of November for their consideration and commentary and the only other comment I wanted to make is that I think that clearly the situation in the UK is an order of magnitude more grave than it is perhaps here in the US and I think that we are going to be faced with very difficult decisions and as to this balance between wanting to ensure a comprehensive evaluation on the other hand and wanting to move forward on the precautionary principle and implement these kind of technologies as soon as is reasonably possible.
Thank you very much.
DR. PRIOLA: Okay, thank you, Dr. Turner. I am going to put a question to the Committee. We have another hour's worth of talks on removal of prions using various filters. We can take a 10-minute break or we can plow through. If we take a 10-minute break we are going to be later. We are about one-half hour behind.
So, do the Committee members want to break or do you want to plow through?
We will take a short break for 10 minutes to sort of re-energize people and then we will come back at ten to four.
DR. PRIOLA: Take your seats so we can get going. Our next speaker will be Dr. Sam Coker from Pall Corporation.
Agenda Item: Performance of Pall Corporation Leukoreduction Filters on TSE Infectivity of Blood Components: Experimental Studies and European Experience Dr. Sam Coker, Pall Corporation
DR. COKER: Thank you very much. What I am going to share with you today are some of the results of the validation work that we have done on this particular filter which is currently being licensed for use in Europe.
Some of the topics that I will be going over today include how we validated the particular product. I will talk a little bit about the process that we use and then I will give you some of the results that we obtained using this particular Western blot.
I will go to part three very quickly just to give you a brief update in terms of the ongoing validation work that we are currently doing.
In part 4 of my discussion I will also give you some of the work that is also ongoing at Pall to identify a particular surrogate that can be used as a QC for looking at the efficient removal of infectivity using this particular product and I will give you a brief rundown of some of the red cell quality that we have already done and finally I will just give you an understanding of some of the work that we are currently doing in Europe.
Some of the tests that we had done to validate this particular product including Western blot which is mainly an exogenous spiking study to kind of give us an idea of how efficiently this particular product is working and then we also did a bioassay to give us a much more realistic indication of the log removal and finally which we think is also the most important aspect of this is to demonstrate the removal of the infectivity from blood that had been infected endogenously using hamster model. So, these are the three main approaches that we use to validate this particular product.
The type of materials that we used in the validation included brain homogenates in this particular case using the hamster model. We also have data using mouse adapted variant CJD which is the closest we can get to variant CJD itself and we also did additional work with sporadic CJD but the main validation data that we used for CE map of this particular product was based on the hamster model.
The Western blot assay that we used is based on the publication from Wadsworth and his group in Edinburgh and we also validated assay using an outside contractor BioReliance which is an FDA licensed contract lab. I will skip some of this but these are just the dilutions that we used to validate the Western blot assay. So, you have an indication of how to measure the infectivity level or the level of the residual amount of PrPres that is present in the blood.
So, the spiking study we did especially we used brain homogenates from hamster. We homogenized it. We added it to the red cells and we simply just filtered. We measured the level of infectivity before and after filtration using the Western blot and we evaluated several different processing conditions, the effects of different anticoagulant filtration heights, effects of leukocytes, the effects of different filter lots.
This is mainly to kind of study the filter that we developed. This is just an example of what a typical Western blot looks like. This is before filtration. This is the proteinase K resistant form which is believed to be the infectious form of the prion. This is before filtration and this is what we obtain after filtration and looking at this over a whole lot of processing conditions, anticoagulants, we saw no significant difference especially between the CPD and the SAGM. However, there is a slight increase or improvement in removal efficiency using CPDA-1 but a critical aspect of this is that the filter is effective in removing infectivity using different anticoagulants.
We also looked at the leukocyte reduced and non-leukocyte reduced blood that had been spiked with infectious prions.
Again, we didn't see any significant difference between that. We looked at different filter lots and again t he level of removal is consistent which again indicates that the manufacturing process that is being used to produce the filter is very consistent and we looked at the effect of filtration heights. We identified that the lower the filtrational height the more effective the removal process. So, that allows that to kind of identify the particulate filtrational height that would be used by this particular product.
We, also, identified filtrational temperature as well as the contribution to improve it in prion removal. Just to summarize this initial aspect of this particular filter it has already been CE Mark in Europe. So, it is available in Europe for use and the CE Mark is based on the hamster model. We have about 2.9 plus or minus .7 logs and we have demonstrated that this filter is effective across all processing conditions and most importantly we saw no significant difference between different lots of filters which indicates consistency in the production of this particular filter.
So, what I am going to share with you now are some of the additional tests that we did to kind of confirm that the filter is able to remove not only the hamster prion but also prions from sporadic CJD and most importantly from mouse adapted variant CJD which I think is very critical and relevant to discussion here.
Again, when you think of the same experiments again, but here using brain homogenates with red cell measured in levels before and after using the Western blot assay and typically for most of the sporadic CJD you can see very clearly the three bands or the three different forms of the PrPres,the diglycosylate and monoglycosylate and this is very important because this shows that the sporadic CJD the amount of these different bands can be used as a way of identifying different strains of sporadic CJD.
Over here in type 4 which is believed to be very typical of variant CJD the main impact of this particular one is to demonstrate the effectiveness of the filter in removing different strains of prion.
So, we collected brain homogenates from different groups of patients that have various forms of neurological disease. Some of them have sporadic CJD and some of them have Alzheimer's disease and this was done in a double-blind study with the New York Medical School and the results show very clearly that the fourth filtration will identify those three bands that are present and identify what particular group has type 1 or type 2. The fourth filtration will identify the presence of those bands after filtration, The filter completely removed the level of infectivity that is present.
The next part of the study was to now go forward and repeat this same experiment using mouse adapted form of infectious prion.
Again, this is before filtration using mouse adapted prion and after filtration again there is removal below the limit of detection of the current assay that we have.
This is just to summarize what we have done to date with this particular product. This is the result that we used for the CEMAC(?) using the scrapie infected hamster and you can see the fourth filtration. So, we are removing about 2.87 plus or minus .7 logs and this is using mouse-adapted variant CJD and there is about 2.2 plus or minus .32 logs and again we will find significant removal with sporadic CJD.
The next aspect of our work is to now demonstrate that the endogenous infectivity can be moved on to similarly. We can see additional confirmation of the exogenous work using a bioassay to determine the log reduction.
We, also, have an experiment going on using endogenous infectivity study and essentially this is just a simple endpoint titration assay to kind of give us an indication of whatever we had with the Western blot correlates with the bioassay. These experiments are currently ongoing and the results should be available early next year and this is the endogenous infectivity again. This we believe is probably one of the most critical aspects of the validation program because as we have discussed earlier today the use of the brain homogenates has its limitations and the best way to avoid the controversy regarding brain homogeneity is to actually use endogenously infected blood samples. So, we have blood collected from about 100 to 200 hamsters and these are then processed as you normally process with the red cell and plasma and the red cell that is endogenously infected is then filtered with the filter. The filtered blood is transfused or intracerebrally injected into about 400 hamsters. Two hundred of them receive the pre-filtration sample and these particular experiments are finally ongoing and by the middle of next year to early part of next year we should be able to get some indication as to what the results are.
So, in summary some of the validation work that we used to study to see clearly were based on the Western blot. We are currently doing the endogenous infectivity study to obtain the actual log reduction that will complement what we have with Western blot.
We, also, have the endogenous infectivity study that is also currently ongoing. So, we will be able to get an indication of how effective the filter is in preventing the transmission of prion disease.
The next part of my talk actually relates to what Marc Turner mentioned earlier about identifying a particular surrogate that can be used to monitor the effectiveness of removal of infectious prions from blood.
We have done quite a lot of work in looking at several different plasma proteins that are present and we have currently identified a couple of proteins instead of the PrPc because of this limitation that we can use to monitor how effective the filter is in removing prions from blood. So, this can be easily incorporated into any blood bank and we feel that this should be a very good way of performing a QC.
The next aspect from my work, I mean this we have already gone through is to give an indication of the red cell quality. We talked about how we validated the product. We have also talked about the effectiveness of the filter being able to remove different strains. The next aspect is to see what is the quality of the red cell after going through the filter.
We did a whole series of studies including hemolysis study looking at the membrane integrity and the neurologic properties in the survivor and by all accounts of what we have done to date this is just an indication of the results. We didn't see any particular change at all and when we look at the hemolysis at the end of a 42-day study especially for sagendazen(?) it is still well below the Council of Europe guidelines as well as the standard from the FDA.
In addition to looking at the quality we also look at the safety of the product itself and all of this is really according to the regulations from the FDA that had been established for leukocyte reduction filters and so today all of this has been passed and we have not seen any particular concern.
So, in summary the red cell quality is very well maintained and we did not see any concerns about the safety of the product or of any of the parameters that we measured to kind of look at the safety.
Overall the filter that I have just described to you has been able to demonstrate that we can remove at least a significant level of infectious prions from blood using brain homogenate. We have demonstrated it can remove different strains of prions, sporadic as well as mouse adapted and overall the cell quality was very well maintained throughout the whole process and most importantly we have also identified a series of proteins or a couple of proteins that can be used as you see of the prion removal efficiency of the filter.
The European experience is that currently we have a series of studies that are currently going on in Europe just to try to validate some of the work that we have done. So, we expect some of these results to come back sometime in 2006.
Thank you very much.
DR. PRIOLA: Thank you, Dr. Coker.
Our next speaker will be Dr. Bob Rohwer who will talk about selection and performance of resin-bound ligands for removal of TSE infectivity.
Agenda Item: Selection and Performance of Resin-Bound Ligands for Removal of TSE Infectivity From Plasma Robert Rohwer, PhD, PRDT (with Prometic and ARC)
DR. ROHWER: Thank you very much. I am wearing a little different hat than I usually do here because I am representing Pathogen Removal and Diagnostic Technologies who is the company which I helped found and which is developing this removal technology.
I am going to concentrate mostly on the infectivity studies because that has been my contribution to this effort and I will summarize the work of the other partners in this.
As we heard earlier in the day we do have now a confirmation that there is a transfusion risk associated with variant CJD. I won't go over this anymore and Dave Asher very nicely summarized this data for me earlier, that at least in the hamster model that we have characterized in our laboratory we get a median value of about 10 infectious doses per ml in blood and that doesn't seem like very much of a risk unless you consider it in terms of a unit. For example, at the same time in this disease when we have 10 infectious doses per ml in blood we have got 10 billion infectious doses in the brain per gram of brain in the hamster model.
So, it really is a very, very small effect compared to what is going on in this animal at that stage of the disease.
On the other hand if we consider the way in which we actually use blood it is not on a per ml basis but a per unit basis. In a 500 ml unit we might have as much as 3-1/2 logs of infectivity.
The other important piece of data is our studies on the appearance of infectivity in blood. Dave Asher showed this earlier and the main point here is that we first saw the infectivity in this part of the infection but because we took points along the way if we extrapolate this back to here it is about one-third of the way through the infection that we first start seeing infectivity.
This would be nothing on the basis of infectious doses per ml, but if we put it in terms of infectious doses per unit even at these very early times we have significant amounts of infectivity probably plenty to cause an infection if a whole unit was given.
How do we deal with this? This is the usual triad of, triad because usually these are lumped, of approaches to controlling TSE pathogens and these particular diseases this group up here are all problematical in various ways and therefore we decided to or I have had been advocating this approach to controlling the risk from these particular pathogens for some time and there are some other real advantages to this in my opinion.
First, it removes infectivity that can't be detected with diagnostics. Every diagnostic has a limit of detection and a perfectly working which we may never get to, but in theory at least a removal device would be able to remove infectivity that could not be detected, i.e., infectivity that was below the window period limit of detection for any pathogen, not just these pathogens.
In the case of TSE diseases this would be even clinical disease for blood because we still don't have a convincing assay for detection of the infectivity in blood or the POP signal in blood from the infection, but it also applies to preclinical disease from brain or other tissues where if you go back early enough in the infection you still may have a risk for example for tissue donation or something like that and still be up against this same limit of detection.
The other advantage of this method is a very big advantage. There is no reason necessarily to need to discriminate between the abnormal from the normal form of PrP.
In the case of blood we can remove both and there is actually even some advantage to doing so because there is measurable PrPc in blood and so far there has been no demonstration of PrPres in blood or we don't have methods sensitive enough to detect it.
So, we can use the removal of PrPc as a assay for the removal of both as long as we have a device that will use both and in fact we selected on purpose for resins that do both and because it can actually access this area of the infection that is below the limit of detection of diagnostics it may be more comprehensive than a diagnostic and in the end it could even be less costly to deploy than diagnostics.
PRDT, Pathogen Removal and Diagnostics Technologies was a company that was put together by Dave Hammond and Rubin Carbonell, a couple of combinatorial chemists. Dave Hammond has, also, had a lot of experience in the plasma industry. Rubin is an engineer at North Carolina State and myself providing the prion expertise and we interested the Red Cross Informetic(?) in investing in this and it has become a joint venture of these two corporations and we now have Macropharma(?) as a major blood bank manufacturer in Europe as the partner for marketing and production of devices.
The way these things came into being is we screened several libraries of various compounds. They represented over 64 million combinations in total. We looked at 8 million beads in the course of doing this, doing a selection assay based on protein methods, a blotting method and the Western blot. Once we picked out the first 200 candidates we decoded the beads and then made larger quantities of these materials so we could go through a secondary characterization based on protein and then out of that we got another group from which we selected seven for infectivity characterization, first by spiked TSE experiments and then an endogenous experiment and I am going to spend the rest of the talk talking about these experiments here.
These are the kinds of things you get out of these screenings. What you are looking at here is the binding in duplicate. These are duplicate samples which is why you see pairs here without protein PHK and with protein PHK in plasma as a couple of resins. So, this particular resin binds well in buffer and in plasma. Here is one that binds well in buffer but not in plasma. Obviously we are interested in this type of resin for further development. That doesn't mean that this particular compound, this particular resin wouldn't be useful but it won't be useful for this particular application and we have in our pocket about 200 of these.
The infectivity experiments, this has been discussed before but I want to go over it once more just to emphasize the difference of what we can get out of the various modalities.
If we spike brain-derived infectivity into red blood cells we have the advantage of high titer, high levels of removal that can be demonstrated but will have uncertain relevance because we don't know how well this spike regardless of how we may manipulate it before spiking represents the infectivity in blood.
This is a somewhat earlier readout. It is a fairly crude measurement and it is less costly. If we do an endogenous experiment relevance is not an issue. It is relevant. It is blood-borne infectivity but the titer is very low about 10 infectious doses per ml in whole blood and even lower for our red blood cells where the plasma concentration is lower and the white blood cells have been removed and the most we can hope to demonstrate inoculating 5 mls of this product into hamsters, into 100 hamsters is about 1-1/2 logs of removal.
It is a long experiment. We have to take the animals to the end of their life essentially but the measurement is very precise by the limiting dilution method which Dave Asher referred to earlier and I really don't have time to explain right here and it is quite a bit more costly for those reasons.
Typically in this type of experiment to measure the infectivity in these models we do endpoint dilution titration. We do serial 10-fold dilutions, inoculate into groups of animals. They get sick and at some point you run out of infectivity and you can calculate a titer from that.
There is also a dose-response associated with this. These animals come down quicker than these and the dose response as Dave showed earlier disappears in this group right here. We are going to make one set of measurements using the dose response in this part of the curve and the endogenous measurements using the infection at the limiting dilution using the Poisson distribution of infectivity into animals at the end.
Now, the dose-response measurements I have always had a problem with this. I have always resisted it but we had a lot of samples we wanted to screen this way and as a consequence we needed something that we could afford to do basically and the main problem I have had with it, one of the problems I have had with it is that the endpoint is hard to define because there is a progression of symptoms especially in the hamster model but saying exactly when one stage of the disease ends and another starts there is a lot of interobserver variation in that.
We developed this method of just weighing the animals. They gain weight throughout their life and as they get sick their weight falls and taking this cutoff at 80 percent of maximum weight as an endpoint.
So, from this we get an empirical determination. It is observer independent. We developed our dose-response curves from duplicate measurements, two completely independent measurements. They are indicated here. Each animal is indicated by a circle here and the means by these triangles and they are displaced around these values. So, you can see the data actually and it is actually much more tightly clustered than I ever would have thought.
Here we are getting at limiting dilution where some of the animals do not get sick and so how does this assay work? We took a large pool of human red blood cell concentrate and then divided it into one unit quantities after spiking it and mixing it. So, all of these challenges were identical and then passed it through our device. This was a prototype device at this time in the development and then collected the unit here, looked at the PrP scrapie signal that is retained on the device here and measured the incubation time of the infectivity that remains in this bag in a test group here.
So here is a case where we removed the infectivity. We can see it here. There is not enough infectivity left. There is not enough PrPres signal left in this bag to measure it by Western blot. So we have to go through the infectivity measurement. Here is our standard curve again. Here is our test group. Here are the incubation times for the test group. They are at this concentration but they are displaced off the curve to this level.
We carry this down to where it belongs on the dose-response curve and we see that we have got 4.33 log of dilution between here and here and we presume that we have removed around 4 logs of infectivity.
We did that for a number of samples. These are all different resins here. These are controls here. They are all at 10 to the minus 3. They all belong on this line but I displaced then so you can see them. This is the data I just showed you. Here is the next best one and here is the worse one right here.
So, the clustered in this fashion. Here is the data summarized in a bar graph, log removal on this side and log reduction on this side. These are our controls again and these are all probably pretty much equivalent within the error associated with this type of measurement.
Just to remind you we challenge with a million infectious doses per ml. The actual blood will contain about 10 infectious doses per ml, but if you will remember we did get infections and that worked out to about 20 infectious doses per ml. Not all of the infectivity was removed by this filter and this infectivity right here when passed through subsequent resins of the same type was not removable. This is in some form that is not recognized by the resin.
So, what we have is we have a spike at this level, a residua at this level and in blood if the proportion is the same we will start with 10 infectious per ml with a residual way out here at .0002 infectious doses per ml. This would not be significant and we would still have quite a significant margin of safety.
Nevertheless we can't be sure of this. We don't know that this distribution is the same because of this question that has been discussed throughout the day. We don't know what the form actually is in blood. Therefore it behooves us to measure this to the best that we can that we can actually remove the endogenous infectivity from blood. What if it is all in this form, for example?
There is a form that would not be removed by our device and that is cell associated infectivity. Everything that we will have to do looking at endogenous infectivity will have to be done with leukoreduced blood because we don't claim that the resin would remove cell-associated infectivity and we know that there is a significant amount of cell-associated infectivity from this leukoreduction experiment that we did a couple of years ago in our laboratory where we took a unit of whole blood prepared from hamsters, that was 500 mls of blood, passed it through a leukoreduction filter and titered the infectivity before and after leukoreduction and got this type of data. This is the leukoreduced whole blood. These are the incubation times down here. These are the animals that did not come down and there is about 40 percent removal here of the starting infectivity by leukoreduction.
This, also, gave us a way to do the experiment because we had this as a precedent and we knew what to expect in terms of the amount of infectivity we could in the leukoreduced blood in order to challenge the device with endogenous infectivity. So, we expected about six infectious doses per ml in whole blood. We knew that if we had made red blood cell concentrate from this and ended up with 20 or 30 mls of plasma our expectation was that we would be down to less than a single infectious dose per ml in that material and this would not be, we didn't consider this to be even though this is the target of the device and the target we were going for we didn't think we could do a meaningful experiment with this material.
So, we actually have done this with whole blood even though we consider it to be a, leukofiltered whole blood even though we consider it to be a worst case. It gives sufficient titer from measurable effect and we have the preceding experiment to inform us.
We are currently at about 420 days, well, at 420 days when I made this slide which was about 30 days ago and this is where we are in this experiment. What I have plotted here is this is incubation time on this axis and the number of animals on this axis. This is all of the limiting dilution titrations we have done to date in our laboratory that are summarized in that first slide that I showed you showing you 10 infectious doses per ml where this is a distribution of about 500 animals from blood infections that have come down over the course of those experiments and this is the distribution of those infections and the point I want to make here and this is, if we add these all up and say, "What proportion of the infections have occurred by 215 days for example?" it is about 50 percent of them, and that is indicated on this red curve here. This is the cumulative number of infections that have occurred at any given time. You can see that they go all the way out to 550 days. So, you can get infections out at this level but at 420 days when I summarized the data that I am going to show you next we are about two-thirds of the way through the infection but we are about 94 percent, we have seen 94 percent of the infections we are going to see. That is the main point. We are very close to seeing everything we are going to see in this experiment. So, here is the data. In this case we took our non-leukoreduced whole blood before leukoreduction and put it into 50 animals instead of 100 and so you need to multiply these numbers by two to get a direct comparison with these values over here.
Each dot here is an animal. Each S means an animal that has come down with scrapie. Here we are in the challenge. This is the titer in the leukoreduced blood. One of our disappointments is we are seeing a lot less infectivity in the leukoreduced blood than we saw in our first experiment, the experiment that was published in the Lancet last year.
Here is the final flow through the device and thank God we haven't seen any infections yet though every time I get this update on this data I get a little heart flutter because we are getting so close to the end of this.
We are 94 percent complete. This is the data in terms of infected over total number of animals inoculated and this is what if we project what it should look like at 100 percent completion we will get another infection here, another infection here, another infection here and hopefully we will see no more infections here.
This titer is coming out to exactly what we always get, about 10 infectious doses per ml or what we usually get but on the other hand we are seeing a lot less infectivity in the challenge. We were expecting about 6.2 infectious doses. Here we are only seeing 2.6, about half that and what has happened here is that the leukoreduction was much more efficient this time than it was the first time around where you lose 75 percent of the infectivity in the leukoreduction instead of 40 percent.
Nevertheless if this relationship remains because we have inoculated 5 mls of this we will have 13 infectious doses in that 5 mls of blood and we will be able to demonstrate a log. We have already demonstrated a log of removal.
Now, just two more comments on what is going on in this leukoreduction. We have done another experiment during the last year and that is we have spent a lot of time over the last 8 years or so trying to figure out what white blood cell type actually harbors the infectivity and every time we purify the white cells we seem to lose the infectivity. So, we just did a simple experiment. We collected the white blood cells from a buffy coat and measured the infectivity before and after a simple centrifugal wash in PBS and that centrifugal wash removed 80 percent of the infectivity.
So, in a typical leukoreduction we have been thinking about this in terms of 50 percent of the infectivity in plasma, 50 percent in the white blood cell fraction because we have shown in other experiments that it is not in the red blood cells intrinsically associated with red blood cells or platelets at least in this model and 80 percent of this plus this leads us to believe that we really should be thinking about the infectivity as plasma associated and it is not that tightly associated white blood cells at that and as a consequence there may be some variability here in the leukofiltration results just based on things like flow rate or pretty mild parameters that we don't have a understanding of yet.
So, that is just summarizing that in words. So, where are we now with this? We have this resin which we have now characterized in an endogenous experiment. It has a very high affinity for the prion protein, 10 to the minus 9 KD. This is mysterious to me and I think it is indicating that there is some cooperativity in this binding. It removes PrP from rodent brain and human brain. We have looked at scrapie, sporadic CJD, familial CJD and variant CJD using the WHO standards and it binds to all of these. It works in red blood cell concentrates, whole blood and plasma. The plasma work has only been done in vitro so far. We get 4 logs of removal of brain derived infectivity greater than 1 log from endogenous infectivity to date and the human compatibility studies have all been done by the Red Cross and of course they have done them very well and so far we have seen no impact on red blood cells, plasma proteins or platelet activation.
By this I mean plasma proteins that are important for therapeutic development. We have a partnership now with Macropharma and they will manufacture and supply this and it is in the latter stages of development.
I wanted to acknowledge Louisa Grigoria in my laboratory and her staff who have spearheaded this effort on its day-to-day basis and I will conclude there.
DR. PRIOLA: Thank you, Dr. Rohwer. We will move on to our final presentation and that is Dr. Ralph Zahn from Alicon.
Agenda Item: Other Industry/Academic Filter Chromatography Develper Dr. Ralph Zahn, CEO, Alicon AG
DR. ZAHN: Good afternoon and thank you very much to the Committee for inviting me here to talk which is a big honor for a small Swiss company actually. So, when we started with our company at the beginning of last year we decided to work on BSE diagnostics and also on biochemicals which are somehow related to BSE but then we found that our technologies can actually also be applied for other diagnostics like scrapie or CJD and it also works with prion filtration and most probably also for Alzheimer's diagnostics and so we currently have 35 products to supply which are 33 different recombinant prion proteins and two monoclonal antibodies and so two of these prion proteins are probably important for diagnostics and also for prion filtration, PrP pure and also PrP beta.
So, why are they important? I think they are important for TSE diagnostics and prion filtration because they can be used for research and development. They can be used as positive and negative controls and they can also be used for checking the quality control for assays and filters and so Alicon PrP pure corresponds to the natural prion protein found in healthy humans and animals or in other words PrPc and this is available for different species including bovine, deer, hamster, human, mouse and sheep.
So, we not only have different species available but all the different lengths of different constructs in particular for the human protein. We have seven different constructs. Also for sheep we have three different polymorphisms and then on the other hand we have PrP beta which corresponds to the natural form of the prion protein found in infected humans and animals or in other words the PrPsc and again we have different species available for this protein.
So, PrP beta is produced starting from PrP pure, the same principle in a three-step procedure which should somehow mimic the production of PrPsc in nature. So, we start with PrP pure and then in the first step we have a conversion from PrP pure to PrP beta star where PrP beta star is the better suited protein. It is oligomeric and it is completely water soluble, and in the second step we have a conformational transition in 2 PrP beta fibrils. So, we think that these three forms of recombinant prion proteins resemble very much the three forms in vivo, so, PrPc, oligomeric PrPsc and also PrPsc fibrils and I also should mention that this procedure has been worked out at the ETH in Zurich in the lab of Professor Retrich and the main work was done by Atoss and Vias and this is some of the biophysical data just to show the conformational transition from PrP pure into PrP beta. You see PrP pure before conversion there is an alpha helical secondary structure and after conversion there was a better secondary structure as indicated by the single minimolar to 15 approximately and this also works for a different construct as is shown here. Then this is the PrP beta fibril which is typically formed off of filaments which are elegantly wound as shown here and where the single protofilaments(?) show this beta substructure and exactly the same morphology has been described for PrPsc, for natural PrPsc and similar to PrPsc the PrP beta also binds Congo red and shows this typical Congo red bifringence(?) and it is also, PrP beta is also more proteinase K resistant compared to the normal prion protein PrP pure. There is at least a factor of 10 difference and there is also an accumulation of this typical 16 kieregard(?) fragment which has been ultimately described for PrPsc and PrP beta binds also to PrPsc conformation specific antibody which is the 15 P3 antibody from another Swiss company. So, this binds PrP beta not PrP pure.
Now, we also have done the biosafety checks because we wanted to know whether this protein, PrP beta is also infectious. So, this was actually what we wanted to show at the beginning but we are trying this since a very long time but we never could really show infectivity. So, we did infectivity checks in TG20 mice but also in wild-type mice. We inoculated about 18 micrograms of mouse PrP beta into these mice. We, also did a serial transmission experiment but we never got clinical signs or pathological signs or proteinase K resistance in the brain homogenate.
Of course, this also has some advantages because if you want to work under less stringent biosafety conditions then the PrP beta has some advantages over natural PrPsc.
So, the main project we are working on or we have worked on is the matrix which has a very high affinity to all kind of prion proteins and so why is this so? Because this matrix does not have only one binding site for PrP but it has three different binding sites indicated here by different colors, so making this contact between PrP and the matrix very efficient and does really very tight binding and another interesting feature of this matrix is that the specificity for the PrPsc conformation can be modulated. In the absence of aligning with X, so zero concentration the matrix binds to PrP pure. So, corresponding to PrPc and to PrP beta corresponding to PrPsc and it also binds to dimeric forms of those proteins and there is only a very low amount of unspecific binding of these averages here in 1000-fold excess over the prion proteins, but if you increase the concentration of this ligand X let us say for example to four then there is only PrPsc bound but not PrPc anymore. So, the specificity of this matrix can be modulated.
So, the applications of this matrix technology, the name of this we call Octapetform(?) for more for historical reasons so that there are two main applications. One is PrP enrichments which can be used for prion detection and the other application is PrP removal which is of course necessary for prion filtration.
So, I would like to show you some examples for PrP enrichment. So, this is an enrichment experiment where we started with 4 mls of plasma and we did an 800-fold enrichment of PrPc in a lot of healthy cows which are not infected with BSE and as you can see here we have a nice signal of PrPc which is 800-fold more sensitive compared to the normal Western blot assay and so we mostly observed the diglycosylated form of PrPc in the one blood in this case in plasma and if we add some proteinase K then of course the protein is degraded.
A similar picture we see also in interface cells or in white blood cells. Again we have a strong signal here after enrichment corresponding to diglycosylated PrPc. This is our standard protein and this is a dimer of the standard and again if you add proteinase K then we get an intermediate fragment first and then at 5 micrograms per ml the PrPc is completely degraded and the same also works for platelets, again this strong signal here for the platelets and this also works for PrPsc from scrapie brain homogenate which was added to bovine plasma in this case here. So, we started with 1.5 mls of plasma. So, we have a 300-fold enrichment and we can show how we can measure very low amounts of proteins for example here 800, 200 and 40 picograms of PrPsc using this enrichment procedure.
Now, so to summarize using this technology we are able to at least 5000-fold increase the signal for example in Western blot assays using this matrix and I am sure we can go even to a higher concentration if we would try. We didn't try so far and in terms of protein concentration we even have a more than 50,000-fold enrichment process going on here.
Now, so we have applied this technology for a BSE live test for cattle and so I would like to show you just one result here. So, on the left side you see a cow which has been experimentally infected with BSE prions about two twenty months ago and this blood was sent us from Germany and on the right hand side you see a control. There is no proteinase K resistant protein seen and this pattern of four bands we observed for, as well as we observed experimentally infected cows as well as natural BSE cows.
This is also an important figure which shows different variants of our matrix which has been used for treatment with BSE infected cows again but which will also spike with PrPsc and as you can see here only one matrix bound to this typical four band pattern which indicates endogenous PrP from an infected cow but the other matrices only recognize PRSE after spiking. So, this means that this is quite important in my opinion because this shows if you have a matrix which binds to PrPsc from brain homogenate this doesn't mean that it also hinds at the same time endogenous PrP.
Now, I am now changing to prion filtration. As we heard this morning there are some important applications like plasma fractionation, plasma banks and also for pharmaceutical industry. This is an example where we have completely removed PrPc from bovine plasma of a healthy cow. So, we did this experiment similarly like before. So, we started with 20 mls of plasma, treated this plasma with our matrix and then we diluted the bound protein and loaded this on a Western blot here for two different cows, A and B. You can see one cow has a little bit more PrPc than the other cow and this is the recombinant protein again and this is done after recombinant protein and if you do this experiment a second time then after the second time there is no PrP left anymore in the plasma. So, this means that we have completely removed the PrPc from the plasma using our matrix and with a detection limit of less than 1 picogram per 20 mls which is about 50 femtograms per ml which corresponds to approximately 0.5 infectious units per ml of blood plasma.
Now, this is a similar experiment with human plasma which was spiked with PrPsc. Again, after the first treatment we see a lot of PrPc and PrPsc before and after proteinase K digest on the Western blot and after the second treatment we only see recombinant prion protein which we used as a marker to show that the matrix works actually but after proteinase K there is no protein left anymore. So, again we have completely removed PrPsc in this case at the concentration of lower than 1 picogram per 20 mls and of course we wanted to know whether our matrix has some effect on blood coagulation. So, we did some different, we did various tests on the different variants of matrices and the results are summarized here. So, there is some effect for some matrices for example here in this global factor but there is a small difference compared to the control. There is also some effect of matrix on the two. There is a slight increase in Factor 7 and there is also a small decrease of von Willebrand's factor of measures 1 and 3 but all the other parameters for example, fibrinogen Factor 5, Factor 8, Factor 9, the fibrin dimers, the three inhibitors of the protein concentration are not changed at all and most importantly there is one matrix where we didn't observe any change in these parameters. So, this matrix is probably quite useful for that transfusion.
So, to summarize the advantage of our technology in our opinion at a really high affinity matrix first and second we can completely remove all prion proteins including PrPc and PrPsc and also endogenous PrPsc as we have seen from this in cows and this matrix also has a high compatibility to blood plasma.
These are the people who did the work finally and thank you very much.
Agenda Item: Open Public Hearing
DR. PRIOLA: Thank you, Dr. Zahn.
Okay, I think we will move on to the open public hearing portion of the afternoon.
DR. FREAS: Dr. Priola, at this time we have not received any request to speak in the open public hearing in the afternoon. Is there anyone in the audience at this time who would like to address the Committee on this topic?
I see none. So, we will move on.
Agenda Item: Committee Discussion and Recommendations
DR. PRIOLA: Okay, so I think we should address the two questions that the FDA has posed to us based upon topic 2.
So, the first question is are the FDA's proposed minimal criteria for validation of TSE infectivity reduction by filtration adequate and appropriate and I think if you will put the slide up there this is a voting question but before we vote they would like us to go through and comment on some of the points behind this question.
So, if you look in the handout from Dr. Vostal on the last page you will see those comments, the points they would like us to comment on.
The first is the rationale for the use of specific animal models to study the properties of blood-borne TSE infectivity. Specifically are experiments in rodents sufficient or should experiments also be done in sheep or any other sort of TSE model?
So, I would like to open that for comments from the Committee.
One model, two models?
DR. TELLING: I had a comment talking about using sheep but another large animal model that springs to mind is infected cervines because obviously there are large amounts of blood available from such models, and it would appear that the lymphoreticular distribution of infectivity to the extent it has been looked at may mimic variant CJD.
DR. BROWN: Just as a practical matter I don't think the FDA can require companies to use small ruminants if they are going to require bioassays. I think bioassays and small ruminants or large ruminants or any ruminants simply won't get done in time to be of any use. We won't have a problem anymore when the titers are finally in.
If they were to use Western blot as a marker for infectivity then I think it is possible to require larger animals. My inclination would be to go with two strains in an appropriate rodent model and I think one of those strains would have to be 301V mouse adapted variant and the other model could be sort of anything you want, 263K in a hamster but I think maybe two species, two models, that is a strain in a mouse and a strain in a hamster and the mouse strain clearly should be 301V as the closest thing that is in a rodent to variant CJD.
DR. PRIOLA: One of the speakers brought up earlier that the size of the blood cells for example differs in blood from different species. What about trying to address that sort of issue? That might be one of the things the FDA is thinking of with using different animal models and using rodents, either mice or hamsters wouldn't necessarily address that unless you just want to stick with Western blot for ruminant models.
DR. BOLTON: What if the infected blood came from naturally infected larger ruminants but the bioassays were done in transgenic mice? Then you have a chance of getting the data back in some sort of reasonable time but you are actually studying the natural product which is closer to what you are looking for.
DR. BRACEY: Are transgenic mice of this sort readily available? I am not a transgenic person. If they are not that kind of throws a wrench into practicality, but if they are --
DR. BOLTON: I don't have them either but I understand that they are in development. Glenn, ovinized transgenic mice, are they --
DR. TELLING: Yes, the ovinized mice have certainly been published on by the French group in particular. As to whether or not they are available, you know, there are cervinized mice.
DR.BOLTON: We produce cervinized mice as you guys have produced them and the group at Case Western has and Stan's group has and we are certainly committed to making all of our transgenic mouse models available and those include not only cervinized but also ovinized and bovinized. So, yes, as far as we are concerned they are available.
DR.ALLEN: I am certainly not expert in this area but I will make the general comment that we have heard an update today and certainly been given enough background information to suggest that there is a lot of variation here both in terms of the host animals as well as the prions themselves and I would suggest as the former speakers have said that looking at multiple models makes a lot of sense and not going with just a single model.
In addition assuming that this is to be used on human blood and plasma certainly one wants to make certain that there is no damage to any of the cellular components or to the end product from the use of the filtration.
So, one looks at it, needs to look at it from both the safety as well as the efficacy points.
DR. PRIOLA: So, from a practical point of view where would this blood come from if it were, I mean would manufacturers be required to have scrapie-positive sheep and CWD positive deer and elk or are there other sources for that available?
DR. BOLTON: They can just go to Colorado and Wyoming, can't they?
DR. PRIOLA: It is possible I guess. That is a practical point but if that blood were available then it would provide a basis for that sort of test. It would also sort of negate the second comment there, is it necessary that each experiment should be done at two separate laboratory sites and that has to do with the contamination issue from people not being careful enough when injecting their animals and if you had common source blood and the same sort of transgenic model systems it would be easier for independent labs to do that.
Comments on any of those first two points? I think it is a really good idea, actually taking blood and using the transgenic mice to test.
DR. ROHWER; I could say a lot about this but I will try to keep my remarks fairly brief. Obviously the best of all possible worlds would be to use human blood inoculated into a transgenic animal that was sensitive enough to assay it and I think that has always been the dream behind the transgenic work was to make that possible but as far as I know it is not possible or not possible yet and certainly there have been a lot of people trying.
The next best thing would be to have a large animal model like the sheep assayable in a transgenic and I think there are a number of us who are trying to do this and we have been talking to each other and hopefully that will come to fruition but we have no idea whether it works now. You know we know you can infect transgenics with brain-derived sheep infectivity but whether that will work for blood and at what efficiency who knows, and the final thing I want to say is just you should consider the following aspects. There is a reason that we use the hamster and we stumbled on it but it turned out to be a pretty ideal model. It produces enough blood that we can actually obtain the blood in quantity. Using 120 hamsters we can make a unit of blood and that is a doable thing. It takes a morning, etc., and we can quantitate it because the animal is small enough we can put it into a large number of animals to do that.
Doing the same types of experiments in mice and we have done this a couple of times now in the 301V mouse it takes eight times as many animals to get the same volume of blood and it takes twice as many animals to do the titration because you can only put half as much inoculum in the brain of a mouse compared to a hamster. So, the cost and logistics go way up, plus it takes a lot longer for the infections to develop in mice.
So, they are also much more sensitive to blood. Blood is toxic when it is inoculated IC. The hamster can tolerate it if you do it right. In the mouse it is much harder to do and so there are lots of things working against the mouse and for that reason we have recently passaged through OMV into the hamster just so we will have access to it but of course by putting it in the hamster that doesn't give us access to transgenics because so far no one has made a transgenic hamster though we have been watching that very carefully and then the other thing I think you should consider is in the sheep model besides these, there are some serious differences in the behavior of sheep blood and plasma compared to humans and hamsters. Actually we find the hamster to be a better model for human blood than sheep even though we use the sheep a lot.
Nevertheless it is going to be hard to quantitate it unless we have a transgenic and if try to do the Fiona Houston type of experiment back into sheep when would we ever know whether the experiment is over? What we are looking for is a negative and a negative result. We don't want the animals to get infected. Well, do you wait until they die? Is that 10, 12, 15 years, something like that? They have a fairly long life span in captivity anyway and so anyway these are some other things that I think have to go into the planning of this and I think it would be important for what I would like to see the FDA do which is voice their concerns in a more general way and give people as much flexibility as possible in meeting the requirements that you want to see them meet so that as animal models develop they may converge on this need or not but basically what we want to do is the best possible experiments we can whatever they happen to be at a given time.
In terms of two sites I think you are going to have trouble doing that with sheep. There aren't that many places you can do this. We have a sheep flock that we couldn't use for this particular application because they are all infected.
DR. BROWN: That recapitulates the notion that several of us have already expressed which is the need for two strains, two hosts. Yes, give people enough flexibility to make their own decisions. As information comes in one host strain combination may turn out to be the ideal. Mice I think are certainly as Bob said in some ways, many ways less practical than hamsters. On the other hand you can get around most of that by doing a spiking experiment with 301V in the mouse and do an endogenous infectivity experiment in the hamster and you don't have to wait 2 years when you use high input infectivity in the mouse and you don't have to collect a unit of blood to necessarily do the spiking experiment.
So, that would be a sort of reversal that might be practical.
DR. EPSTEIN: I just wanted to ask the question how important is it to do an actual transfusion experiment because the advantage of the large animal model is you can actually transfuse an intact unit into a whole animal with volume relationships comparable to human transfusion and I know you certainly can study infectivity with IC inoculations in readout animals but I think part of the idea which embedded suggesting a model in sheep is the actual transfusion experiment and also you know looking forward to question 2 we are sort of suggesting that one might stratify efficacy labeling according to whether an actual transfusion experiment had or had not been done. So, you know the idea of waiting forever for the result you might be able to approve products with more limited labels pending a more definitive experiment.
So, I would like to hear opinion from the Committee specifically on the question of whether an experiment needs to be done with actual transfusion in a large animal model in order to mimic human transfusion.
DR. BROWN: The goal is to detect infectivity in blood before and after a process and so you want to use the technique which is optimum for detecting infectivity. If that turns out to be a transfusion then yes you would want to use a transfusion experiment but I think there probably are more optimal ways, more sensitive, right, sensitivity and one is the transgenic mouse and it may well be that intracerebral inoculation of smaller volumes in the proper host strain model will be more sensitive than transfusion where you know the transfusion is a very sensible method to detect infectivity in sheep. It is if I am not mistaken, Bob, much less sensitive in hamsters, that is when you inoculate it in blood intracerebrally you got a far greater number of takes than when you transfused blood even in larger volume.
So, in that sense transfusion would be less sensitive than intracerebral even if the volume was smaller.
DR.ROHWER: That is true except that we don't know what the actual titer was in the sheep blood. That same effect may be present there. It is just that by giving 500 mls instead of 2 mls you overwhelm it and you don't see it. So, I don't think that question can be resolved by that comparison.
DR. BROWN: At least not yet. We just don't know yet, but the point, the principle is what you want is the most sensitive method using more than a single strain.
DR. PRIOLA: Any other comments from the Committee on the comment?
So, if we go on to B is it necessary that each experiment should be done at two separate laboratory sites I think we sort of addressed that to ensure reproducibility and accuracy of clearance. Any comments on that? I know that this is always an issue in any scrapie lab where you are looking at low levels of infectivity. Is it a practical issue in this instance?
DR. BROWN: The difficulty is actually in the wrong direction that is to say if you do get cross contamination you are out of the business. So, it behooves anybody who does such an experiment in a single laboratory to be extremely careful and so there is an enormous motivation to avoid cross contamination if you do it in a single place.
DR. PRIOLA: And with appropriate and rigorous controls that is completely doable. I am not so concerned about the two laboratory sites anyway as long as the experiments are appropriately controlled.
DR. GESCHWIND: Particularly with that issue of large animals making it really just impractical.
DR.VOSTAL; I would just like to point out the fact that when we evaluate things like leukoreduction filters we always ask for two laboratories to minimize laboratory differences and practices and such.
DR. PRIOLA: It is a bit more difficult situation in the TSE field because of the specialized nature of the infectivity and there are very few labs that can do it. So, it might not be as practical to do that but I see your point.
Any other comments?
Now, C, is general description of informative scaled down processes for reducing TSE infectivity in blood and I have got to admit I am not exactly sure what that means.
DR. VOSTAL; I think we are trying to ask whether scaled down experiments are acceptable or whether it would be better to do a full-scale transfusion like in the sheep.
DR. PRIOLA: Anybody want to hazard a comment?
DR.BOLTON: Paul actually beat me to it. I guess at some point if you were trying to certify a particular filter you are going to have to have that particular device and geometry to run the full unit of blood through it or some configuration. I don't know how you would do 500 microliters of infected mouse blood through this thing and get any kind of meaningful answer but as Paul said there is a lot of experiments that you can do sort of in the preliminary stage to get a foundation to say that yes, we should go on, but I think eventually you would have to run the particular geometry that is going to be approved.
DR. ALLEN: I think that there are two components to that. As Dr. Bolton said you are going to want to make sure that the process runs on a whole unit of blood.
On the other hand how you analyze it doesn't mean that you have to then infuse that whole unit of blood in a large animal and follow it for X amount of time. If you have got another more sensitive or equally sensitive method of detection and a residual infectivity that should be perfectly adequate. You do it to assess the process itself in full volume.
DR.BOLTON: And the demonstration of the filter, the product coming through the filter still has all the appropriate biological specifications, wouldn't even need to be done on an infected unit. At least in my opinion you could run a parallel unit on normal uninfected blood that we would then define the parameters in terms of its blood qualities and separately test in infected units for the removal of infectivity. Does that make sense?
DR. PRIOLA: Any other comments?
Let us move on to D which is what are the levels of clearance acceptable for claims of reduced TSE infectivity in blood components as used in clinical settings. So, this harks back a bit to what we talked about this morning. I think Dr. Rohwer showed with his filter he can remove so far at least as far as he can tell all the massive infectivity he has in his blood model using an endogenous sample as well as a good chunk of something in a spiked sample but what would be the clearance that would be acceptable?
DR. BOLTON: Paul, I thought you would jump in here with the two species, two strains. I am reading your mind but I think you would say that it depends on the, if you are doing a spiking study you are going to get one potential level of clearance. If you are doing an endogenous study you are not going to be able to achieve that clearance. So, Bob, you have got what 1 log. You could demonstrate 1 log endogenous. You can't do more than that at least right now.
DR.ROHWER: It is a matter of how many animals you inoculate but there is a point of diminishing return because the returns go down as a factor of two actually, but to inoculate a whole unit would need 10,000 hamsters and I don't think anyone is prepared to do an experiment like that. So, we do 5 mls. It is something we can do and occasionally we have gone to 10 mls but generally you can get an idea of what you are going to get from that kind of data. For example, you don't get another log for that. You get another fraction of a log.
DR. BROWN; The other interesting thing from one of the presentations is it is conceivable that a methodology such as we saw from Alicon could concentrate infectivity in assay experiments so that you could in fact by inoculating, by using a concentrating device you could get the equivalent of a whole unit of blood in a couple of milliliters. That is something that might be considered.
DR. PRIOLA: Yes, especially in combination with the two-mouse model or two-rodent model.
DR. ROHWER: All these devices by definition are concentrating infectivity in the device and in our particular case we haven't been able to figure out how to get this stuff back off without killing it because it sticks so tight, but we are still working on that and if we can figure that out that is definitely a way you could go.
DR. ALLEN: As I read the question it is what are the levels of clearance that should be acceptable for clearance of induced TSE infectivity in blood components as used in clinical settings. Obviously in clinical settings you are not going to have anything above endogenous infectivity. So, you know if using spiked samples you can show a level of clearance is well above what would normally be found in endogenous infectivity and in the endogenous experiments that are done you don't get any evidence of transmissibility it would seem to me that you have satisfied the claim and I understand that this is all hypothetical. It is a statistical process. Nonetheless given what we know now I will be reluctant to accept any evidence that suggested that there could be a breakthrough with or likely be a breakthrough with endogenous level of infectivity. I would like to see it well below that.
DR. TELLING: So, the answer is a log?
DR. PRIOLA: You mean for the endogenous.
DR. BROWN: I don't think anybody would be happy with that and I think your question, well, I said, anybody would be happy; you know, you are looking for 5 and historically 6 logs of with HIV and a few other things. Certainly you have to sterilize endogenous infectivity whatever it is, if it is a log, 1 log, 1-1/2, 2 logs. We can't have a single particle left. There can be no transmission. Any transmission from an experiment on endogenous infectivity is a failure. That is one criterion. I think you can set your own criteria for spiking. We learned from Marc that according to his model it only requires about 3 logs. I think that is more than you need but that is what he got. So, he is the expert.
DR. PRIOLA: Also, this morning we came up with an upper limit of 2 logs. So, from a limit of 2 up to 100. So, perhaps for a validation study 2 logs would be the upper limit in this case as well because that should clear everything in the blood based upon what work has been presented.
DR. BROWN: And that would be my feeling but I know that that makes other people uncomfortable because they like more margin for error and there is always this issue of transferring the exquisite care that goes into the laboratory experiments to the manufacturing scene and so it is not possible just to say, "Okay, in the laboratory here we get 2 logs and say that that is good enough for the manufacturer," So, partly for that reason I think that people want a margin of error. So, I think 3 logs in the spiking experiment might be appropriate.
DR. ALLEN: Yes, one would like to see a reasonable significant margin of error. I think it is easier to get rid of that first 99 percent than it is to get rid of that last 1 percent or the last 1/10 percent or 1/100 percent. So, to get to the point that you have sterilized, you know one of my mentors when I was much younger always, "Sterility is actually a theoretical concept. You can never guarantee it," and you try to set up a process that goes well beyond whatever would be detectable in a clinical situation and I think that is what we would like to do here.
DR.BROWN: And the other thing that I would emphasize is that almost simply as important and possibly more important than defining a minimum with a margin of error is requiring absolute guaranteed reproducibility in test after test after test and only in that way can you get a feel for whether or not your margin of error is satisfactory.
DR. LEITMAN: So, this whole process tends to put a lot of responsibility on the manufacturer. For leukoreduction filters the blood center validates and does quality control very easily because the readout is so simple. It is a flow white cell count or something like that but the blood center here, that is where this will be used and the surface service won't have the tools to do the correct readout. So, they can't validate their process really. So, I had a little difficulty with that because everything you perform as a manufacturing step we can do a quality control on but not this.
DR. PRIOLA: Perhaps that is another reason for having it done at two independent laboratories, the manufacturer and somebody else to basically back the manufacturer up or not by doing the same studies. Would that be better? You would never be able to do it at the local level. There is no way unless someone comes up with a surrogate easily detectable marker and even then it might be questionable.
Dr. Weinstein? Oh, I am sorry, Dr. Vostal?
DR.VOSTAL: I just want to ask when you do an endogenous infectivity experiment do you do a leukoreduction on that product first and then process your product through your device or does leukoreduction become part of that, the accounting of the infectivity from start to finish?
DR. COKER: From what we are doing at Pall there was no leukoreduction prior to doing the filtration. So, the whole blood is actually not leukoreduced at all.
DR. ROHWER: In terms of PRDT device it is not a leukofilter. It is a light and it binds PrP specifically and it will only remove from plasma. There is no claim that it would remove cell-associated infectivity. So, this device would be docked below the leukofilter in a collection scheme and will be an add on and in terms of evaluating what these things do it is important I think to get an idea of whether it removes from plasma, whether these devices remove from plasma or not because basically I think that is the residual risk we are trying to get rid of. We already know the leukofilter will remove white blood cells and we can get rid of that risk with the leukofilter and so in terms of figuring out what these things actually do and whether they actually work I think it is important to test it against the residual plasma component.
DR. PRIOLA: So, D and E sort of go together in a way. It is all about levels of clearance, the first in blood components in clinical settings and the second to conclude that blood filters have effects with related infectivity from those blood components. So, does the Committee have kind of a consensus as to what that range of clearance is then? Are we talking as we did earlier this morning about spike being a really good way to show high clearance and then doing it with endogenous infectivity as well?
DR. BROWN: Yes, I think there is a consensus. I think everybody agrees that sterility of an endogenous infectious sample is mandatory and I don't know we might leave it up to the FDA folks in attendance to make their own decision about what kind of level of concentration of infectivity represents something that they are comfortable with. We certainly had a lot of discussion about it this morning and whether it is 3 logs or 2 logs or 4 logs maybe it is best that they decide. I doubt that we are going to be able to make the decision here.
DR. PRIOLA: I think that is exactly right. We are not going to come up with a bottom line here for this.
So, I think the final point that they would like us to comment on is the methodology appropriate to use in evaluating TSE agent clearance and we heard this afternoon that they start with Western blot and then move to bioassay and I know my opinion is always the bioassay has to be in there before anything is approved because that is the most sensitive technique so far.
DR.BROWN: In terms of reproducibility you can design a spiking experiment and do 100 spiking experiments and use a Western blot. That is okay. You can't do 100 bioassays. That is unreasonable but you could do one or two bioassays to complement and you could use Westerns for reproducibility and the bioassay as the most appropriate test for what you are looking for which is transmissibility.
DR. PRIOLA: As long as your Western blots are reproducible, right.
DR. TELLING: Along those lines we heard about almost 3 log reductions based on Western blot and what is the dynamic range of that assay?
DR. COKER: It is between about 1 log and 3 logs. So, 3 logs is about the maximum.
DR. TELLING: So, how do you know that you are going to get 3 log reduction if you are at the limit of your --
DR.COKER: I don't understand. Do you mean for the endogenous?
DR. TELLING: Three logs is your limit. You are saying that that is what you are achieving. How do you know that you are not --
DR. BROWN: The best Westerns that I know of were done by Bayer and on a good day they could detect close to 2 logs of infectivity. So the 3 log minimum threshold is sort of your everyday best. So, you certainly have to if you want to demonstrate 3 log reduction of infectivity using a Western blot as a marker you have to start with 6 logs and then if you get nothing in the filtrate you know you have got at least 3 logs. You might have more but you know you have got that.
DR.ROHWER: Considering this is another complication and that is that doing a Western blot out of plasma is almost the most difficult challenge you can present a Western blot with, so you have to do some sort of preprocessing in order to get a signal in the first place.
DR. PRIOLA: Yes, maybe the Western blot would only really work well with the spiking experiments.
DR. GESCHWIND: I just had maybe a point of clarification. Are we leaving open the option of tests other than a Western blot just checking that there are tests that are currently out there and tests that have just been presented in Dusseldorf that are certainly more sensitive than the standard Western blot; so, I want to make sure that we are not restricting it to the Western blot.
DR. PRIOLA: You mean like CDI or PMCA or something? Yes, I am sure we are not restricting it.
DR. CREEKMORE: It actually says, "Or other assay for serum proteins."
DR. GESCHWIND: And we are also in the bioassay. Is there another question about biomarkers or is that a separate question?
DR. PRIOLA: I don't think that, but go ahead and comment on the biomarkers.
DR. GESCHWIND: Just from the clinical side and seeing the biomarkers that have been touted for the diagnosis I am very worried about the use of biomarkers when we actually have the actual protein itself or the disease, the bioassay. So, I just would say that I am against the idea right now of biomarkers particularly because of the difficulty with reproducibility between lab to lab. I think we should really go down that path with great caution.
DR. PRIOLA: I think it was Dr. Turner who presented some data that they might have a couple of proteins in plasma, but I agree that really has to be validated and studied very hard.
Dr. Allen, did you have anything you wanted to say?
DR. ROHWER: Marc, please correct me but my understanding is that the idea of using the biomarker is simply to get a way of routinely testing whether the thing is working at all, you know, are there holes in it, that kind of thing and if you have a biomarker protein that you know also binds to the device you can at least assay for whether it is being removed. If it is easier to assay for it then the PrP protein and the idea is to do that on a routine basis to make sure that the device is working.
DR. PRIOLA: Just for reproducibility, yes, although I still think it would be nice to someday have another biomarker for TOC here but I would like a lot of things.
Any other comments from the Committee?
Should we vote on this issue? Are the FDA's proposed minimal criteria -- we are not voting?
DR. FREAS: We are not voting on this issue.
DR. PRIOLA: We are not voting on this issue. So, we have had a discussion. Would the FDA like us to have any other thing they would like to mention or point out or have us discuss?
DR. EPSTEIN: I guess we would like a vote overall whether we have the right set of criteria.
DR. PRIOLA: Okay, so we will vote on do they have the right set of criteria given what has been discussed this morning and this afternoon. Are the FDA's proposed minimal criteria for validation of TSE infectivity reduction by filtration adequate and appropriate?
Do we have a slide of that?
So, again, these are the minimal criteria.
I don't know if they have to delete the sheep but I think the point is just to have two animal models, or one that might be applicable to transfusion experiments and that would be sheep if they can do it.
DR. BROWN: Susan, yes, I think rodent and sheep should probably be excluded.
DR. PRIOLA: Just two animal models in general.
DR. BROWN: Two animal models and I would in that first one demonstrate elimination not reduction. We were just talking about that. If you can only reduce a log and one-half you are not in business, okay? So, it is the elimination of endogenous TSE infectivity and then reduction of spiked infectivity.
DR. CREEKMORE: I agree with that and then also in our discussion about the two separate sites how does the Committee feel about that? It seemed like if there were adequate controls that the two separate sites weren't that critical.
DR. PRIOLA: That is what I would think. I mean I agree with that but apparently with other infectious organisms the FDA requires that. So, it might be nice and then Dr. Leitman brought up that it would be another way to validate the manufacturer's claims to have it done independently but I agree. I don't think it is essential, but Glenn, did you have a comment?
DR. TELLING: Basically only to underscore your earlier comments that if these are adequately controlled and from reputable laboratories then I wouldn't have any problems from one location.
DR. BOLTON; I recall going through this once before some years ago but it seems to me that we are in fact rewriting the criteria that we are asked to give. The question was whether or not we thought that they were appropriate and adequate and it seems to me that we have decided that they are not appropriate and adequate and we are not rewriting them. Which would you prefer? Do you want us to rewrite them or do you want us to just tell you what we thought about the original ones?
DR. VOSTAL: I think we already know what you thought about the original ones. I think any comments are very helpful and if you care to rewrite them for us that would be helpful in itself.
DR. BROWN: We are not doing a whole lot of rewriting. We changed a word and eliminated a parenthesis. That is not bad and if you want to eliminate necessity to perform the study at two different sites it seems to me there is a kind of a sense around the table that that would be okay as well but since we are talking about minimums it could also be left in.
DR. VOSTAL: The concern we have is that if we get data from a manufacturer who has had 5 years of perfecting their procedure we are always wondering whether that will work in somebody else's hands and I think that is what Dr. Leitman was trying to point out.
DR. PRIOLA: And that is a very valid point because we have had that experience in the research field where you can't replicate what somebody does.
Yes, Dr. Turner?
DR. TURNER: That is exactly the concern we had in the UK which is exactly the reason we are going down the road of commissioning our own independent validation studies.
DR. ALLEN: The third bullet point that requires TSE infectivity from BSE or vCJD, I assume that would be spiked experiment?
DR. PRIOLA: That would presumably be along the lines of what Dr. Brown said which is 301V. It is the rodent version of BSE, yes, and that could be spiking.
DR. BROWN: And the study performed at two separate sites, what you might want to do is just add since apparently the real reason is indicate that this is a test which is doable in more than one place which is neither an issue of cross contamination or maybe even differences but we might want to reword differences in laboratory practice and just indicate reproducibility in different labs.
DR. COKER: I just have one comment on that. If the manufacturer decides to use to do the tests well then will they be required to use another contract lab because most of the manufacturers or some of them do not do the studies in house? They actually contract it out. So, if you are requesting two sites that means they will have to have two contractors doing the tests.
DR. PRIOLA: So,with the modifications that have been made to these bullet points does the Committee feel comfortable on voting?
DR. GESCHWIND: There is only one thing. The infusion aspect doesn't seem to be put in there that there has to be a model.
DR. PRIOLA: You mean the transfusion?
DR. GESCHWIND: The transfusion or is that No. 2 where there has to be a model in which you are infusing a unit.
DR. PRIOLA: I think that is why sheep was originally up there.
DR. GESCHWIND: Right, but I thought we had taken that out.
DR. PRIOLA: Okay, rodent, sheep or other model.
DR. BRACEY: I thought we changed it to say that the key point was demonstrating lack of infectivity.
DR. PRIOLA: Yes, elimination.
DR. BRACEY: Elimination of infectivity so that we weren't really, we were requiring the process of an entire unit but not infusion of an entire unit.
DR. PRIOLA: Any other comments?
Shall we vote?
Oh, sorry, Bob.
DR. ROHWER: I have a question for Paul. Paul, I am not clear on what you mean by demonstrating that there is sterility in a unit because right now we don't have, endogenous sterility in a unit because right now we don't have as far as I know the capability of doing that unless we do a transfusion in sheep in which case we could but it would take years to know whether the sheep is actually transmitted or not especially if we are looking at very low individual titers and so, I see a practical issue here that needs to be considered.
DR. BROWN: You used a full unit in the hamster.
DR. ROHWER: Yes, but we only measured 5 mls of blood for that unit.
DR. BROWN: But that represented a unit for a hamster, more than a unit.
DR. ROHWER: Oh, that is more than a unit for a hamster. Is that what you mean? That is what I want to get at. Is that what you mean?
DR. BROWN: Yes, right.
DR. PRIOLA: Dr. Epstein?
DR.EPSTEIN: This is the very issue I was trying to raise before in other words if a whole unit in a human might contain 5000 infectious units and if the processed leukoreduced unit still might contain 2 or 3 thousand infectious units and then you only do an assay on a few milliliters then you do not have the ability to assure that you have sterilized the unit. If the residual infectivity were for argument's sake 20 or 30 infectious units then at any feasible volume studied in a rodent model you could not exclude a residual infectivity of a whole unit and in fact that is --
DR. BROWN: Of a whole human unit, but that is what we are saying.
DR. EPSTEIN: But how else will you model filtration of a volume which may contain endogenous infectivity of 5000 or 3000 IU? This is precisely the dilemma and that is why we have stratified two potential labels. One potential label says that the filter has been shown to reduce the infectivity. The other label essentially says that it has been shown to remove the infectivity of the transfused unit and we understand fully that you can't get to that second endpoint without either waiting a long time or some advancement in the experimental models but we are not really comforted that showing logs reduction in a scaled down model shows that you have eliminated endogenous infectivity and that is precisely the point.
DR. BROWN: I couldn't agree more with you, Jay but we can't figure out how to do the experiment or we would do it.
DR. BOLTON: You know how to do it, 10,000 hamsters.
DR. ROHWER: I agree that I do know how to do the experiment but I despair of talking anybody into doing it.
DR. EPSTEIN: But the endpoint may be that we get to labeling A and not labeling B because the end point isn't that we can't review claims for filters. The issue then becomes what exactly is the claim that has been supported by that experiment.
DR. BOLTON: You could achieve the first bench mark by a spiking and clearance study or the hamster endogenous study maybe, but the second bench mark you would have to meet by say for example doing a sheep transfusion study or
a 10,000 hamster study or something that would end up being equivalent to examining an entire unit of blood.
DR. TELLING: But even in the sheep transfusion study a negative result wouldn't necessarily give you the confidence that there was no infectivity in that blood.
DR.BOLTON: There would have to be more than one sheep. What would be the statistical number? I don't know that anybody is going to do that study to get that second label.
DR. PRIOLA: Because you have to determine the level of sensitivity in sheep.
DR. BROWN: Of course, you could also use a filter with a T-median level column, I mean do a real scale down which is actually a serious problem in terms of people being comfortable with X degree of scale down or Y degree of scale down as opposed to the real thing and you can never do the real thing. You can't do an experiment on 10,000 pints of plasma. So, it is a question of degree and I have to say that personally I am comfortable in seeing a reduction of a concentration of whatever is in endogenous infected blood whether it be 10 or 30 or 50 infectious doses per ml eliminated in a reasonable number of mls to say that I have sterilized, I mean just as a practical matter as we have just been talking about.
DR. PRIOLA: Any other comments?
Okay, let us move to the vote, keeping in mind what has just been said and we have proposed slight modifications to those bullet points.
Let us vote yes or no to are the FDA's proposed minimum criteria for validation of TSE infectivity reduction by filtration adequate and appropriate.
MR. BIAS: I think we can just state what has been modified.
DR. PRIOLA: Okay, good point. So, the modifications as I remember them were to demonstrate elimination and to take out rodent and sheep. Do we still agree for that first bullet point? Do we still agree with that?
MR. BIAS: Are there two animal models that are okay if we are taking out rodent and sheep?
DR. PRIOLA: I think we just leave it to the FDA. I mean we can leave that open. You don't have to constrain the animal models. I think that was the idea.
DR. BOLTON: I think Paul's suggestion was two different species, two different strains.
DR. PRIOLA: Yes, but that is getting, I mean stili that is two animal models or more than one animal model. So, I don't think we have to get that specific.
DR.BOLTON: Would you accept mouse scrapie and hamster scrapie? I am trying to figure out are you going to select --
MR. BIAS: I am sure they will but are we sure they are going to select an animal model that applies?
DR. BROWN: I think it is probably important to specify two strains, two hosts rather than two animals because you could. You know, you could study two different strains of scrapie and you could study two different strains of scrapie in mice and that wouldn't be adequate in my opinion.
DR. PRIOLA: So, two strains, two hosts in those parentheses instead of sheep and rodents, two strains, two hosts.
Okay, was there another? Now, I can't remember if there was another modification. The two sites? Do we still want the two separate sites? Oh, to minimize issues of cross contamination and reproducibility, so to take out differences in laboratory practice.
DR. BROWN: You could even depending on how the Committee felt include the word "preferably" after the word "study."
DR. PRIOLA: Preferably performed to give them some leeway.
DR. TELLING: So, a clarification, implicit in what one of those combinations of animal, host and strain will be is bullet point No. 3, question mark. So, one of them should be BSE or vCJD.
DR. PRIOLA: Right.
DR. BROWN; The other point here is that I see nothing on this slide about a spike experiment.
DR. PRIOLA: There is a recommendation, that is right, for the first one. So, you had recommended demonstrate elimination of endogenous infectivity and reduction of spiked infectivity going back to the two approaches by 3 logs.
DR. TELLING: Again, that is implicit in bullet point No. 4 because the only way you would be able to detect the scrapie is by spiking, right?
DR. PRIOLA: Right.
Lynn, did you have something you wanted to add?
DR. CREEKMORE: Just that with the point that Glenn was making about the TSE infectivity from BSE or vCJD strain it doesn't say endogenous versus spiked. So, it doesn't connect back to that demonstrate reduction of endogenous because they could choose to do a spiked experiment for that.
DR. BROWN: Susan, I think you have to have a bullet in here. Either put it in as a first sentence on bullet point 4 or add another bullet but you have to mention explicitly a spike experiment. You can't just leave it implicit in the PrP bullet.
DR. PRIOLA: We can add it to the first one, reduction in spiked infectivity and that will take care of that.
So, it is starting to get a little bit confusing. So, I just want to make sure. You know, I don't want to get stuck on this for the next 20 minutes. So, the modification of the first one is elimination of endogenous TSE infectivity or reduction of spiked infectivity, and, excuse me, and reduction of spiked infectivity by 3 logs in animal models, two strains, two hosts. We can't give this detail. The 301B is implied. It is definitely implied in TSE infectivity from BSE or vCJD strain, I am sure. So, I think that is okay. Then the other change was bullet point 5 and that is cross contamination and reproducibility. Is that correct?
Okay, so, let us go ahead and vote on this issue.
DR. FREAS: We will go around the table.
DR. BOLTON: Yes.
DR. FREAS: Dr. Johnson?
DR. JOHNSON: Yes.
DR. FREAS: Dr. Telling?
DR. FREAS: Dr. Creekmore?
DR. CREEKMORE: Yes, amended.
DR. FREAS: Dr. Hogan?
DR.HOGAN: Yes, as amended.
DR. FREAS: Mr. Bias?
MR. BIAS: Yes, as amended.
DR. FREAS: Dr. Allen?
DR. ALLEN: Yes.
DR. FREAS: Dr.Priola?
DR. PRIOLA: Yes.
DR. FREAS: Dr. Geschwind?
DR. GESCHWIND: Yes.
DR. FREAS: Dr. Brown?
DR. BROWN: Yes,with the modifications.
DR. FREAS: That is unanimous, yes, with modifications.
DR. PRIOLA: All right, the final thing that we have been asked to address is -- is this a voting question as well? I am sorry I don't know. Is question 2 on top of 2 a voting question? It seems like it is. It is, okay.
So, does the FDA's proposed labeling for a filter meet the appropriate criteria for a claim of reduction of TSE infectivity in blood or blood components and we have already gone through all this I think with the first question.
So, this filter has been shown to reduce TSE infectivity in blood from an infected animal model plus C, right?
DR. FREAS: Right, that is the disclaimer that goes along with it.
DR. PRIOLA: And that disclaimer goes with both or if you get a transfusion model the label would be B.
DR. JOHNSON: Shouldn't it say it has been shown to eliminate? We already said that in the other.
DR.BOLTON: No, because the spiking experiment won't necessarily --
DR.JOHNSON; Or reduce by 3 logs.
DR. BOLTON: That is going to be too confusing.
DR. PRIOLA: I think maybe David is right that since we have reduced for the spiked reduced would be better there. Does anybody have any major objections to the way these labels are phrased?
DR. BOLTON: No major objections. I just would point out that phrases like infected animal model I am not sure what that means tot he general public. So, some thought maybe could be given to what phrases of those type mean the most to people.
DR.JOHNSON: The general public won't be buying filters.
DR. PRIOLA: Let us go ahead and I don't sense any major problems with this. So, let us go ahead and vote on this final question.
DR. FREAS: Dr.Bolton?
DR. BOLTON: Yes.
DR. FREAS: Dr. Johnson?
DR. FREAS: Dr. Telling?
DR. TELLING: Yes.
DR. FREAS: Dr. Creekmore?
DR. CREEKMORE: Yes.
DR. FREAS: Mr. Bias?
DR. BIAS: Yes.
DR. FREAS: Dr. Allen?
DR. ALLEN: Yes.
DR. FREAS: Dr. Priola?
DR. PRIOLA: Yes.
DR. FREAS: Dr. Geschwind?
DR. FREAS: Dr.Brown?
DR. BROWN: Yes.
DR. FREAS: Again, a unanimous yes with nine people voting, I believe.
DR. PRIOLA: Okay, that I think, thank you all very much for your patience. I know it has been a very intense day and this meeting is adjourned.
DR. BROWN: Sue, congratulations. In 5 years as Chairman I never had a clean slate of unanimous votes at a single meeting.
DR. PRIOLA: That is because of the participants.
DR. FREAS: I do have one more announcement regarding Dr. Alan Jenny. I received this message. The services will be held Tuesday, November 1, at 1:30 p.m., at the Starkwellin Funeral Home, 609 7th Street, Boone, Iowa, and memorial contributions will be accepted in Dr. Jenny's name for the local Ike's(?) Club for Conservation, and if you need more information, please see me.
(Thereupon, at 5:50 p.m., the meeting was adjourned.)