Members have access to national West Nile reactive donation data by state, and by donor residential zip code, via the AABB Biovigilence Network, and by way of communication plans established by blood organizations.  West Nile activity, as observed from West Nile RNA-confirmed- positive blood donors, occurs from the beginning of May through November.  Available data indicates that the use of ID-NAT during these times of West Nile activity is implemented by either single facilities or multiple facilities representing overlapping and adjacent areas, will greatly reduce the risk of transfusion-transmitted West Nile virus.

            During 2007, no cases of transfusion-transmitted West Nile virus were reported.

            Recommendations put forward in AABB Association bulletins regarding West Nile are developed based upon available data that are reviewed by the AABB's West Nile Task Force, which includes representation from the national blood organizations and obtains input from liaison representatives from the FDA and CDC.

            Such recommendations are also reviewed by AABB's Transfusion-Transmitted Diseases Committee, which has broad representation from national and international experts.

            Lastly, all AABB Association bulletins are approved, prior to release, by the AABB board of directors.  One of the objectives of the association bulletin developed for 2007 was to validate the recommended minimum trigger, which was based on the absolute number of West Nile reactive donations in a given geographic area as well as the rate of West Nile reactive donations in that same area over a seven day rolling period.

            The validation data, which was collected by a prospective study conducted by the American Red Cross, in six regions having experienced recurrent West Nile activity, revealed that 25 of 30, or 83 percent of confirmed positive donations were not detected using the recommended trigger.

            Increased sensitivity could be achieved with the elimination of the rate criterion and the use of only the absolute number of West Nile reactive donations.

            Prior to the release of revised recommendations regarding the use of a new trigger--that's for this year--agreement was reached that the use of presumed viremic donations or PVDs--those are donations with elevated signal-to-cutoff values, or those that are reactive--should be used in place of initially reactive donations for triggering and detriggering determinations.

            This is the practice that has been used successfully by blood centers for the past several years.

            In contrast to initially reactive results, PVDs have a sensitivity and positive predictive value of greater than 95 percent.  So that as decisions to convert from minipool to ID-NAT are made, resources, including both labor and reagents, are used wisely and focused in the areas of greatest need.  The false-positive rate of the NAT assays, relative to initially reactive results of one to two per thousand tests--that's the Red Cross experience, which is lot-dependent--would result in unnecessary triggering events, but, more importantly, would make it difficult or impossible to ever transition back to minipool NAT.

            The validation data revealed that 56 of 57, or 98 percent of donors testing falsely positive, occurred during West Nile ID-NAT, and therefore, more specific criteria than the use of initial reactivity for determining conversion from minipool to ID-NAT, and back to minipool NAT are needed.

            The West Nile Task Force reviewed three options for triggering based on the validation data and elimination of the rate criterion.  The three options reviewed prior to the release of the association bulletin included one PVD, not an initially-reactive donation with a rate determination, two PVDs, again without a rate determination, and thirdly, a hybrid approach which included a blending of one and two PVDs, again without a rate determination.

            One PVD would be used in predetermined areas of West Nile activity and two PVDs would be used in areas that have never experienced previous West Nile activity.

            The intent, again, was to carefully evaluate where the need was greatest and not falsely trigger in a given area such that labor reagents, donors and their donations would be conserved.

            Following consideration of the above three options, the decision was made to recommend two PVDs within a seven day roll-in period without a rate requirement.  Facilities are encouraged to review the validation data provided in the association bulletin and make local triggering and detriggering decisions as appropriate.

            However, it is recommended that facilities consider the benefits of establishing uniform criteria.  Facilities are also encouraged to review carefully location conditions, and that triggering on one PVD is appropriate.

            In the event that facilities in overlapping or adjacent areas have PVDs or local conditions indicate ongoing West Nile activity, such as reported clinical cases or reports of positive birds or mosquito pools, detriggering is based on seven days without a PVD, ID-NAT should continue beyond seven days in areas with ongoing West Nile activity, including positive blood donors from facilities that collect in overlapping areas or if other conditions such as prior history, ongoing clinical avian or mosquito activity exist, or at the discretion of the medical director.

            In these situations, ID-NAT for 14 days should be considered.

            The new triggering and detriggering criteria are believed to represent an approach that is believed to reduce the already low risk of West Nile transfusion transmission.

            Lastly, the AABB and other professional organizations, and blood collecting organizations, will be responding with comments to the FDA draft guidance.  Thank you.

            DR. SIEGAL:  Thank you very much.  All right.  At this point we'll adjourn for lunch, but I ask the committee to come forward to talk a little bit about the new voting procedures.  We'll resume at 1:30.

            (Whereupon, a luncheon recess was taken, the committee to resume at 1:30 p.m., the same day.)

            DR. SIEGAL:  Okay.  Let's come to order, and we'll start right away with topic one, relating to the BEST study, and the committee report on red blood cell recovery standards.

            We're first going to hear from Ping He, M.D., the medical officer at FDA, on FDA's approaches to evaluation of red cell products.

            Dr. He.  Have I pronounced that correctly?

            DR. HE:  Good afternoon.  My name is Ping He.  I am from Division of Hematology at Office of Blood Research and Review, and this afternoon, I'm going to talk about the FDA's criteria for evaluation of red blood cell products.

            Here's the issue summary for what we are going to discuss this afternoon.  FDA seeks the advice of the committee on an industry proposal to change the current acceptance criteria for evaluation of red blood cell studies based on in vivo radiolabeling recovery trials.

            Here's the key issue that we are going to talk about this afternoon.  Should the RBC in vivo recovery threshold value be changed?  The threshold value of the greater than equals 75 percent for RBC recovery study, serve as a cutoff line for determining RBC viability for unit during the evaluation of RBC in vivo recovery studies.

            The threshold value of greater than equals 75 percent has been--oh.  Okay.             The threshold value for--the threshold value of greater than equals 75 percent has been used by FDA in the past 24 years for evaluation of RBC in vivo recovery studies, based on the expert opinion and historic data.

            And this diagram shows that if a unit of RBC has in vivo recovery greater than 75 percent, that it was considered as a successful unit, and that means that the minimum amount of RBC, viable RBC in this unit would be greater than 75 percent.

            However, recently, the industry did a study based on the historic data, and indicated that the overall RBC recovery studies from 1990 to 2006 would not meet the FDA's current acceptance criteria--I'm going to talk about later--unless the threshold value, it changed from 75 percent to 67 percent, meaning that the viable RBCs in each bag should be dropped to 67 percent, while in contrast to the industry analysis, based on the more recent data, FDA believes that RBC products are improving with time.

            During the period 1998 to 2007, 17 out of 19 RBC recovery studies met the current acceptance criteria.

            Therefore, FDA maintains that the threshold value of greater than equals 75 percent should not be changed.

            We all know that RBC products play an important role in transfusion medicine.  RBCs are life-saving products that deliver oxygen to tissue.

            Each year, about 14 million units of full blood are collected, and on a given day, more than 38,000 units of RBC products are needed for patients with anemia, trauma, cancer or surgical procedures.  Up to now, there is no available substitute for RBC products, and the demand for RBC products are continuously growing.

            However, the collection or processing or manufacturing of RBC products may cause storage lesion.  A unit of RBC products can be collected either through the whole blood collection procedure which goes through steps of centrifugation and separation, or can be collected through the complex procedure.

            And then RBC products can be stored at different anticoagulants or additive solutions at 1 to 6 degree for a maximum shelf life of 42 days.  Studies have been showing that any preservation or manipulation can induce RBC membrane damage, therefore producing changes in the biochemical properties of RBCs and shortening their in vivo survival.

            Here, the potential harmful effects of the RBC storage lesion from the practical view, the most important change of RBC storage is the loss of RBC viability, and therefore shortens time in circulation after transfusion into the recipient.

            It therefore will decrease the oxygen delivery to tissue and the transfusion of damaged RBCs to recipient may saturate macrophage clearance mechanisms, and therefore reduce the bacterial clearance.

            The storage lesion can also decrease the levels of ATP in RBC, which correlates with the reduced RBC viability, and decrease the tissue perfusion.  Decrease 2,3-DPG, reduce oxygen delivery to tissue.  Loss cell membrane can make RBCs rigid, and therefore increase the spontaneous lysis, decrease microvascular flow.

            Storage lesion can also cause increase in hemolysis which has a harmful effect on organ function, and increased plasma potassium can cause increased potential hazard to neonates.

            Therefore, in order to ensure a new device for presence of RBC products, safe and effective, here, the FDA recommended tests on approval/clearance of stored processed RBC products.  Basically, a unit of the whole blood will be collected from a healthy volunteer into this new bag, and then the blood will be stored for 42 days, and we know that at end of the storage some of the RBC becomes nonviable.  Now in order to determine the levels of viable cells in a stored RBC bag, a portion of RBC will be collected from the storage bag and radiolabeled, and reinfused back to the same donor, and the 24 hour RBC in vivo recovery study will determine the viability of the RBCs.

            Meanwhile, a panel of the individual RBC test will also be determined from the stored RBC product, such as ph, hemolysis, levels for ATP, 230 pg, hemoglobin, hematocrit, cell morphology counts and glucose, and so on.

            However, the RBC individual test are used as screening tests, and the individual test are not predictive of in vivo RBC performance.  Therefore, the 24 hour RBC in vivo recovery remains critical for evaluation of effectiveness of a new bag.

            So the in vivo RBC recovery at 24 hours provides a surrogate end point for RBC product to evaluation.  It is to demonstrate that the safety and effectiveness of novel RBC collection processes --

            This diagram shows the comparison of RBC in vivo recoveries with different RBC storage period.  It was published by Mollison in 1951.

            These three lines shows the post-transfusion red blood cell survival from fresh blood, of red blood cell stored for 14 days or stored for 28 days in ACD anticoagulant at 4 degrees, and here the axis shows the days of the transfusion, the Y axis shows the percent of the RBC survival.

            I would like to draw your attention to the 24 hour post-transfusion time point, showing that the 24 hour post-transfusion recovery for the fresh blood is great, about a 100 percent.  It drops to 95 percent when the cell is stored at 14 days, and it drops to 75 percent when the cell is stored for 28 days.

            This means that the longer the storage, the poorer the RBC survival and recovery.  As I had mentioned earlier, that the greater than, or equals 75 percent of RBC recovery at 24 hours, is a threshold value for individual unit recovery, that it has been used for evaluation for in vivo recoveries.

            Now in order for FDA to determine if a clinical study is successful, certain criteria has used in the past, and this is a historical review of the FDA acceptance criteria for evaluation for in vivo RBC studies.  Basically, we recommend the study should be done in more than two different centers, with a minimum of twenty healthy volunteers.

            In 1985, at the FDA workshop on red cells stored in additive solution, based on the historic data and expert opinion, the acceptance criteria for RBC in vivo recovery started out with a mean recovery of  multiple individual units greater than equals 75 percent, meaning that if you have a sample size of twenty, regardless, the recovery ranged from 30, 40 percent to 80, 90 percent.  Or regardless of individual units that has recovery less than 75 percent, three out of twenty or five out of twenty, as long as the average or mean recovery of 20 units, greater than equals 75 percent, would be acceptable.

            Well, a decade later, in 1998, based on industry request, in addition to the mean of greater than equals 75 percent, FDA added a division for less than--or equal 9 percent, to eliminate some of the outliers.

            In 2004, when FDA reviewed some of the new submissions, we noticed that some of the studies, they can meet mean and the standard division.  However, the number of individual units that has recovery less than 75 percent is high.

            For example, five out of 20 units, or six out of 20 units has individual recoveries less than 75 percent, and this raised the concern about the quality of the RBC products, and to ensure the proportional successes, another parameter which is proposal for units with recovery greater than equals 75 percent was a one-sided 95 percent lower limit, greater than 70 percent, was added to the mean and standard division.

            So this was also called the current exemption criteria, and this criteria was discussed at 2004 BPAC meeting, and it was also communicated with the regulated industries through pre-meetings, and it was also presented at a couple of different workshops.

            So since 2004, majority of the submissions to FDA in this time period passed the current acceptance criteria.  Those that do not meet the current acceptance criteria also failed the previous mean and standard division criteria.

            So we believe that maintenance of quality for new RBC products is important.  Reduction in the approval of clearance criteria would allow RBCs to be the more severe damage, storage lesion to the market, which may correlate with a poorer clinical outcome.

            In next couple slides, I'm going to show you example for some retrospective analysis for transfusion of aged red blood cells associated with the adverse clinical outcome.

            This paper, published in 2006, titled as the Association  Between Duration of Storage of Transfused Red Blood cells and Morbidity/Mortality After Reoperative Cardiac Surgery.

            This shows that when you transfuse, the older the blood cells, it is associated with increased in-hospital mortality and associated with increased acute renal dysfunction.

            So the results indicate that there's an association between prolonged RBC storage and adverse clinical outcomes such as mortality and organ failure.

            This paper, published a few weeks ago, talks about the duration of red cell storage and the complications after cardiac surgery.

            The X axis shows the year the patient received the red cell transfusion, and the Y axis shows the survival.

            The orange line shows the patient received the new red blood cells which is less than 14 day old, and the blue line shows the patient received older red blood cells, older than 14 day old.  And the study shows that the patient has a better and higher survival when they receive the newer red blood cells.

            So here are the reasons to revise the in vivo RBC recovery and acceptance criteria.  For example, we have three studies, studies A, B, and C, and the sample size, 24, 21, 21, and all three studies met the acceptable criteria for mean recovery greater than 75 percent, and all three studies met the standard division of less than or equal 9 percent.

            However, in study A, there were eight individual units, has RBC recovery less than 75 percent, that is, eight out of twenty-four, and that makes the one study, 95 percent lower cost B- a limit for proportion of units having recovery greater than equals 75 percent, was only 47.9 percent.

            And in study B, we have five out of 21 units has RBC recovery less than 75 percent, and that makes the proportion of units having recovery greater than 75 percent, 56.3 percent.

            Study C, two out of 21 failures, and that makes the proportion of units having recovery greater than 75 percent was 72.9 percent.

            So the products from the study C has a much higher proportion of units having recoveries greater than 75 percent.

            So the high rate of individual units is that it's A and B, raised a concern about the quality of RBC products, and led FDA to consider revising the criteria.

            This slide shows, further explains the three studies that we mentioned in the previous slide.  The red line shows the RBC 24-hour recovery at 75 percent, and was again that all three studies met, the mean met the greater than 75 percent, and all three studies had the standard deviation, less than 9 percent.

            However, the number of individual units that has recovery less than 75 percent was that eight out of 24 from study A, and five out of 21 in study B, and only two out of 21 in study A.  Therefor, the products from study B offers much higher proportion of the units has individual recovery greater than 75 percent.

            So the revised current acceptance criteria emphasize the population proportion of successes, also called the 95-70 rule.  It is to ensure that most products, meaning greater than 70 percent of the products, have recovery greater than equals 75 percent.

            To meet this 95-70 rule, a specific number of maximum failures are allowed in a study, depending on the sample size of the study.

            For example, if the sample size of the study is twenty, then the number of units with recovery less than 75 percent would be two, and three out of 24, four out of 28, five out of 33.  So if the clinical study met this kind of outcome, one can conclude that with one study, 95 percent lower count limit, greater than 70 percent of the units will have RBC recoveries greater than 75 percent.

            So here comes the key issue we are going to discuss today.  Should the RBC in vivo recovery threshold value of greater than equals 75 percent be changed?

            The reason we ask this question was that since 2004, after FDA institute the revised criterion, some of the manufacturers raised the concern that some of the new RBC products may not be able to meet the current acceptable criteria, and the products already on the market would not meet the criteria either.

            Therefore, the manufacturers volunteered to provide the RBC in vivo recovery study data used for the support and approval of RBC products already on the market, to reassess the current criterion.

            And Drs. Dumont and AuBuchon collected the data and analyzed the data.  The objective of the BEST study was to determine if the products already on the market would be able to meet FDA's current acceptance criteria, and the BEST study concluded that the overall data analysis from 1990 to 2006 will not meet FDA's current acceptance criteria unless the threshold value changed from 75 percent to 67 percent. 

            Well, FDA also analyzed the data after Dr. Dumont kindly shared the BEST data with FDA and here's the FDA analysis of a combined BEST and FDA dataset in different time period, from 1990 to 2007.

            Well, this is a very complex table that Dr. Kim is going to go through this table in great detail in her talk later.  I'm only going to point out a few key points from this table.

            First of all, the BEST data collected three subsets of the studies.  One is the conventional 42 days liquid stored with the blood cells.  Another set, the gamma-irradiated red blood cells.  Another set is the frozen blood cells.  Today, we're only going to focus on the 42 day conventional liquid stored red blood cells to assess the current acceptance criteria.

            The gamma-irradiated red blood cells and the frozen red blood cells are considered as the special circumstances and they will be discussed at other times.

            Second is that here's the--595 are from the BEST datapoint.  So in addition to the 595 BEST datapoint, FDA added an additional 94 datapoint.  That was from the full approvals from the 2004 to 2007, which were not included in the BEST data analysis.

            And so that makes the final N of the 689.  Well, also to point out that studies 40 and 41, that were included in the BEST analysis, were not included here because the information of the year study was not available and we were not able to put into any of this time period.

            Although studies 40 and 41 are excellent studies, both studies met the FDA's current acceptance criteria.  And so we analyzed the--so in contrast to the BEST analysis, which lumps all the datasets from 1990 to 2006, to see if that meets the current acceptance criteria, FDA actually analyzed the data in three time period based on the year of the acceptance criteria use.

            For example, 1990 to 1997, the acceptance criteria was to meet the mean of greater than 75 percent only.  From 1998 to 2003, we added standard division of less than nine.  In 2004 to 2007 added a proportion greater than 70 percent.

            So if you look at the third value of greater than equals 75 percent, the success rate to meet this value increased with time, for example, from .83 to .93, the current time period.

            And the success rate--the power to meet the current acceptance criteria also increased from .43 to .92, and the number of the studies to meet the current criteria increased from the first time period of four out of eight to nine out of eleven, and to eight out of eight.

            I would also like to point out that the success rate, or the datapoints to meet greater than 75 percent is more than 90 percent, if you combine the datapoint from 1998 to 2003, to 2004, 2007, more than 90 percent of the datapoints can meet the threshold value of greater than 75 percent, meaning only 10 percent of the datapoints cannot have the recovery greater than 75 percent.

            So here are the observations of the combined data analysis.  Overall, the quality of RBC products approved or cleared by FDA is improving with time.  Most recent RBC products, meaning from 2004 to 2007, submitted to FDA passed the higher standard with a power of .92, with a threshold value of greater than equals 75 percent.

            This actually answered one of the concerns, that the manufacturer concern that some of the new products are unable to meet the current acceptance criteria, and it is also known that the most clinical studies performed to satisfy FDA criteria for drugs are powered at .80.

            So the threshold value of greater than equals 75 percent has provided a standard for RBC quality evaluation over the last 24 years.  The current criteria show that most of the RBC products, meaning greater than 70 percent, have a recovery greater than equals 75 percent.

            Therefore, based on these considerations, FDA proposes to continue applying the criteria adopted in 2004 to quality evaluation of RBC products using in vivo radiolabeled studies.

            Here are the questions I'm going to ask for us.  I think that we can come back and ask questions again during the discussion.

            So here are the questions to the committee.  Question number one.  Does the committee agree with FDA's proposal to maintain the current criteria? 

            The current criteria are: Radiolabeling studies should be performed in at least two separate centers with a total of 20-24 healthy donors.

            The mean recovery at 24 hours for each unit should be greater than equal to 75 percent with a standard division of less than equal 9 percent; and the one-sided 95 percent lower confidence limit for the population proportion of successes greater than 70 percent.  Here, the success means that each individual of RBC recovery greater than equals 75 percent.

            Question number two.  Alternatively, does the committee recommend that a change in the criteria is needed based on the data presented today?

            And question number three.  If the answer to question two is yes, what changes does the committee recommend for the threshold value of individual subject RBC in vivo recovery with a sample size of 24?

            Examples to consider.  We give you three examples here.

            A.  based on the combined data from '98 to 2007 with greater than equals 74 percent as the threshold value.  Power equals .82.

            B.  Based on combined data from '98 to 2007 with greater than equals 73 percent as the threshold value.  Power equals .93.

            C.  Based on the BEST data from 1990 to 2006, BEST recommends 67 percent as the threshold value.  Power equals .999.

            And I would like to take this opportunity to thank all the FDA staff for their expertise and support, and I also like to thank Dr. Dumont and Dr. Jim AuBuchon for sharing the BEST data with us.  And thank you for listening.

            DR. SIEGAL:  Thank you, Dr. He.


            DR. SZYMANSKI:  Now for clarification, when you say that the requirement are, now, 75 percent mean recovery--correct?

            DR. HE:  Right.  Yes.

            DR. SZYMANSKI:  Now 9 percent standard deviation?

            DR. HE:  Yes.

            DR. SZYMANSKI:  And then you say if you have 20 to 24 subjects done in at least two different institutions, only three should be failures.  Is that what you say?

            DR. HE:  Yeah.  Three out of 24 meaning--

            DR. SZYMANSKI:  Three out of 24 should be less than 75 percent?

            DR. HE:  Yes.

            DR. SZYMANSKI:  How can you say that you have a mean of 75 percent?  Twenty-four sample studied, and only three is under 75 percent.  Usually, when you have a mean, sort a half is up and the other half is down.  That is like an impossibility.

            DR. FLEMING:  Just to explain, there are two things happening here.  One thing is what is the criterion for success or failure and--

            DR. SZYMANSKI:  She said 75 percent.

            DR. FLEMING:  Right.  So on the individual basis, if your recovery is at least 75 percent, you're called a success, and, in fact, what's being questioned here is should the definition of what, on an individual basis, would be called a success, should it be dropped from 75 to 74, 73, etcetera?  That's the definition of success.

            Now take 20 to 24 people and compute the average success rate, and that average success rate has to be high enough to rule out that it could be 70 percent or lower, and that's achieved by having in 24 people and no more than three failures. 

            So what they're saying is in fact logically consistent.

            DR. SZYMANSKI:  I don't believe so, because if you have 24--no--now if you have 24 people, and only three can be under 75, your mean can't be 75.  Your mean must be over 75.

            DR. FLEMING:  So let me change the scenario, because this type of argument could happen in any disease setting.

            Suppose you have an HIV/AIDS patient and you're trying to reduce viral load.  What do we define success to be?  Maybe we define it to be getting to undetectable levels below 50 cells.  Okay.  Then we define success/failure based on whether a patient has rendered less than 50 cells.

            Now take 24 such patients, and can we ensure the success rate in those 24 patients is at least above 70 percent?  That's the same situation here.

            So what's confusing, somewhat, is what's defined to be success is a 75 percent recovery.  That's the individual basis for defining success.  Now take 24 people and conclude that the success rate is at least 70 percent, which occurs when you see at least 21 of 24 successes.

            DR. HE:  Right.

            DR. DI BISCEGLIE:  Can I offer a clarification?  I'm not sure how this, how these criteria are used.  I mean, what is being served by the--the blood bank center product?  I don't know how these things are used.  Please clarify.  Well, you said these are acceptance criteria.

            DR. HE:  Right.

            DR. DI BISCEGLIE:  Acceptance of what?

            DR. HE:  Accept the in vivo recovery studies.

            DR. DI BISCEGLIE:  I'm sorry.  This is probably obvious to the hematologists in the room but I have not--

            DR. VOSTAL:  -- new products for isolating, collecting, processing red cells.

            DR. DI BISCEGLIE:  And device?

            DR. VOSTAL:  For a device; yes.  Let's say a storage bank.  You get a new storage bag for evaluation, we want to make sure that the quality of those red cells stored in that bag are assured for 42 days.  So we do a study at the end of 42 days, this radiolabeling study, and we set the criteria, that the recovery should be greater than 75 percent with these additional statistical criteria.

            DR. HE:  Okay; thank you.

            DR. SIEGAL:  Any other questions or discussion?

            DR. DI BISCEGLIE:  If I may.

            DR. HE:  Sure.  Please.

            DR. DI BISCEGLIE:  You show the data on survival versus--survival, patient survival versus age of the blood.  Are there any data that correlate patient survival or clinical trials to red cell viability?

            DR. HE:  Well, no, we don't have data on that, but we do know that when the red blood cells store longer, and then the survival drops.  And we know that, many of the late studies showing that the aged red blood cells also associated with adverse clinical outcomes.

            So here we give the example, wants to try to say that before a device, or before anticoagulant, for storage of the red blood cells to be on the market, we want to assure that the survival is stored in those--the red cells survive, stored in those bags, will be higher than 75 percent.

            DR. DI BISCEGLIE: Is that certain?

            DR. HE:  It is not but, however, I can tell you that the 75 percent recovery has been debated in the past, many, many years.  In fact, since 1947, that was 60 years ago, Dr. Rose already published his paper saying that when you're evaluating a new RBC collection bag, or RBC storage, storage in solution, and you want to ensure that more than 70 percent of the red blood cells stored in that device are viable.  Okay?

            And that was 60 years ago.  At that time the red blood cells were stored in the glass bottles, in a much less advanced anticoagulant such as ACP.  And this is why, 20 years ago, in 1985, the experts in the field, they feel that, well, 40 years ago, in 1947, the experts were ready to propose and to suggest that the recovery should be greater than 75 percent.

            Now, in 1985, we have much more advanced additive solutions, we have plastic bags, and the survival should be increased a little bit higher.  That's why, at that time, they increased recovery, 24 hour recovery to 75 percent as the threshold value.

            And also I'd like to point out that the most recent combined study from FDA and BEST was 689 datapoints.  You can see that.  More than 90 percent of the datapoints from that study, showing that more than 90 percent of the datapoints can meet the 75 percent recovery, meaning only 10 percent of the datapoints will not be able to meet 75 percent recovery.

            DR. ZIMRIN:  You showed two studies that suggested that with older red cells, associated with the adverse clinical outcome--just a point of clarification.  There are certainly studies that do not suggest that.  You didn't mention that in your presentation; right?

            DR. HE:  That is true.  That is definitely true.  But on other hand, the study I showed, the second one, that's probably one of the largest study, and from a single center.  They'd actually be able to separate, to actually differentiate that some people only received the new red blood cells and some people only received the old red blood cells, and the people who received the new red blood cells has a better survival.

            Of course understand, we all understand that there are some variations, and there are some other points, should be further studied and further analyzed, and also this is a retrospective study--

            DR. ZIMRIN:  It's a retrospective study.  The graph you showed was an unadjusted comparison; right?

            DR. HE:  We all understand that debate but--

            DR. ZIMRIN:  Well, no, you did make that point.

            DR. HE:  Right.  We understand that; yes.

            DR. DI BISCEGLIE: I don=t know if this is the point you're trying to make.  I think, you know, we need to have a clear understanding if this is or is not going to be significant, but I'm not sure you've really answered the question.  I think you're saying the question can't be answered.

            DR. GOLDING:  Yes.  You know, I think you are pointing out relevant points, and we agree with what you're saying.  What we're saying is the current state of knowledge is not perfect, but what we do know is that we would--that it's likely, that if you have more viable cells, and you transfuse them, that there's less chance, we think, of getting adverse events.

            You know, it might take another five or ten years before the studies that are being alluded to are done in a more satisfactory way.  Do we want to reduce our standard in the meantime, before those studies are completed, or do we want to maintain a standard that we think allows most of the products that we see to be approved at that level?

            And the tests that we have are not perfect.  So the 75 percent survival, how does that relate to a clinically meaningful outcome? which is your question.  We don't have a definitive answer to that, but our approach is that this test has stood the--has been used as a criterion and can assure a certain viability of red cells, and that we'd like to maintain the criterion unless we have evidence to say that it's not a good approach in terms of approving products, and trying to assure that these products are safe in effect.

            DR. FINNEGAN:  I'm even less sophisticated than the hematologist.  Are you saying that we are discarding good blood?  Or are you saying that you're using this to test new bags and new fluid?

            DR. HE:  This is a test of new bags and new fluid.

            DR. FINNEGAN:  So we're not throwing blood away?

            DR. HE:  No; no.  Not for already approved on the market.  Only for the new bags.  Yes.

            DR. SIEGAL:  Just to clarify for me, these are blood samples that have sat for 42 days.

            DR. HE:  Right.

            DR. SIEGAL:  So they're way at the end of the shelf life.

            DR. HE:  Storage period; yes.

            DR. SIEGAL:  And they have no bearing, really, on the New England Journal article that just came out, really, which tested 15-day-old as the cutoff, where you have a clinical end point, blood.

            DR. GLYNN:  And again, these studies, the two studies you refer to, these are retrospective studies?

            DR. HE:  Right.

            DR. GLYNN:  I don't think there are any--they haven't been--well, there have been one pilot clinical trial done so far on just like 57, 59 patients, but otherwise there are no prospective data that I'm aware of.

            DR. HE:  That's true.  Yes.  Thank you.

            DR. SIEGAL:  Dr. Epstein.

            DR. EPSTEIN:  Yes.  Back to Dr. Zimrin's point which is well-taken.  Two things.  First, the Advisory Committee for Blood Safety and Availability will be addressing the question of what do we really know about age of storage in relation to clinical outcome, and I think we would readily concede that we don't know the answer now because the available studies go both directions, some showing no effect, some showing an effect, and the vast majority are retrospective and they're full of confounders.  So it's a little bit of a distraction.

            The reason for having mentioned that is that there's at least some body of data suggesting that there is a storage lesion.  We didn't have to cite those data.  You can see that, just from the reduced recovery, or the slide that was shown based on the studies done by Mollison, shows that age of storage correlates with reduced viability of the cells you infuse, or you recover a smaller percent the longer you store the blood.

            So we know that there's damage to cells.  So the point here really is we also know that the survival of the infused red cells relates directly to the proportion that are nonviable and are rapidly cleared.

            So what we're basically saying is if you're going to transfuse red cells, what you want is that they should be functional in vivo.  We think that the length of survival correlates with functionality.  In other words, the body's getting rid of bad cells.

            So the test at 24 hours for recovery is actually a surrogate, it's a predictor for the survival of the red cells in vivo, which we think is a surrogate or predictor for their functionality in vivo.

            So what we're basically saying is that it's a quality characteristic, that if the recovery is better, it means you've damaged the cells less, and we think that that has to be good.  So it's not that we're linking it directly to knowledge about the clinical outcome of aged blood versus fresh blood.  It's that we believe that all stored blood has a storage lesion, and that is shown by the fact that recovery goes down and survival goes down the more you store, and all we're saying is, well, if you look at the end of the storage period, we want to set a lower limit on the amount of nonviable cells that are in the population, cause we think that that's a quality factor for the product.

            So, again, the clinical outcomes with age versus fresh blood is a little bit of a distraction.  It was only there to illustrate that we know there's a storage lesion.  But we know that anyway.

            DR. SIEGAL:  Dr. Szymanski.

            DR. SZYMANSKI:  I agree with that but what I have difficulty with is to set these requirements, as presented, without the data that now exist, because it feels to me that here statistics are used to "strangle" the biology, and we know that there are, when you do the survival studies, there is individual variation, and there are certain donors and recipients who are their own recipients, own donors, that do not have a 24 hour recovery as is decided in this statistical model.

            And I went back to some of my own data, that was done in 1999, it was part of an FDA approval, and these data were accepted, or the methodology was accepted on the basis of these data.  And so I had 14 studies, and the mean was 78.7 percent, and four of these studies failed to be 75 percent.  One of them was rather lousy, was 57 percent, and when I looked, in detail, at this case, this particular donor had sort of removed that number of red cells at 24 hours, a 100 minus 76--57--but then he released those cells later on.

            And if you look at the long-term survival data, and project it to the Y axis, you have more than 80 percent recovery of these cells.

            So this presents a biologic variation of the donor-recipient, who kept the cells, probably in the spleen, and then released them, having actually more viable cells than would appear on the basis of 24- hour survival.

            And so therefore, I just feel that there are so many different biologic variables, that if you say that a certain number only can be under 75 percent, it is very difficult to come up with a study that meets the statistical requirements.

            Whereas the whole average is indicating that the product itself, the process of producing these type of red cells was good.

            DR. SIEGAL:  Any further commentary?  Or questions?

            DR. MANNO:  From a pediatric point of view, there are well-documented hazards to transfusion of old blood, even at 36 or 42 days, to low birthweight infants, and it's a practice that we don't accept.  We use fresher blood for infants.

            And then there was some data from years ago in chronically transfused thalassemia patients, demonstrating a benefit, although it was not anything that can be put into practice, of transfusing what they call neocytes, that is cells that were fresh, and much more buoyant, separated off from older stored red cells.  Better survival and less dumping of iron into the RE system for those chronically-transfused thalassemia patients.

            So if we're talking about testing devices, I think that it's fine to consider this, but if we're talking about what goes into at least infants and young children, we certainly don't want stored blood.

            DR. SIEGAL:  Comment?

            DR. FLEMING:  As I'm listening to your example, I understand the frustrations, that there's imperfections here in what we're calling success/failure.  I would call that that's inherent to the use of a surrogate, in most instances where we're using surrogates, and the example you gave, even under the refined definition, that person you referred to would be called failure, because that's still an estimate of less than 67 percent.  The two articles that were referred to for us to be looking at, the Ross article in 1947 laid this out as an arbitrary value of 70 percent, with a statement that we don't wish to imply that blood providing cells of this viability is as satisfactory as blood with a greater percentage, and then in the Grendan article, the 75 percent in 2001 was stated at 24 hour survival, to ensure that at the end of the allowed storage period, most units of blood will have adequate recovery and survival despite individual variation.

            So my "read" on this is people are struggling to try to formulate a surrogate that will allow this assessment in 20 to 24 people, rather than to have to have thousands of people that it would take to truly validate whether or not one product is in fact truly, from a clinical outcome perspective, noninferior.

            There are always, then, arbitrariness--there are always uncertainties with surrogates, this applies in any setting, and because, if you weaken a surrogate, you make it more likely that a result will be positive, is not a justification to weaken a surrogate.

            And I already mentioned one example in HIV/AIDS.  If you're treating somebody and you're trying to get to undetectable viral loads, or low viral loads, if you define success to be getting below 50 cells, clearly, more people will be successful if you make that getting below 400 cells, and more people further will be called successful if you make it below a thousand cells.

            Or in oncology, where you're trying to delay progression as a surrogate for prolonging survival.  If you declare success being progression-free at four months, more people will be successes if you say progress at three months.  More still at two months.

            Well, yes, you get more successes, but the fundamental issue is what defines the surrogate is adequately reliable evidence to establish clinical benefit.

            And so to change the viral load threshold, I have to have clinical data that would say, okay, if it's success just to get to a thousand cells, there's clinical data to say that's just as good as another therapy that gets you to fifty cells.  We don't have that data here, and in the essence here, the burden is not to assume the surrogate is valid until it's established to be invalid.  No; it's the other way around.

            We have to validate a surrogate. By congressional mandate, therapies need to have substantial evidence of efficacy.  So from my perspective, to weaken the surrogate, if a sponsor is advocating weakening a surrogate, show me the scientific data that says this weaker definition of success still protects efficacy.  So the question isn't: FDA, could you defend why you need a more rigorous standard?  The question is: Industry--or anybody who's advocating otherwise--can you defend why a weaker standard is still protecting efficacy?

            DR. ZIMRIN:  But when you look back at the transcript in the 2004 meeting, the fact, when this rule was established, there was no clinical data that said that it made more sense, and, in fact, looking back at that meeting, I'm amazed, actually, that the FDA concluded from that discussion that the BPAC was advising this at all.

            In fact, it seems to me that there was no consensus and the decision was to go to lunch.  So I agree with you that we need a real clinical end point, but I'm not sure--I mean, I think that this process has been involved and somewhat circuitous.  But--

            DR. FLEMING:  So I'm just asking: What would be the data, the--clearly, the weaker you make the criterion for success, the more people will be called success.  That's not how we justify weakening surrogates.

            We justify weakening a surrogate by having scientific evidence that shows when you do so, you still maintain clinical efficacy.  We haven't been shown any data, at least yet, to see that that's true.

            DR. SZYMANSKI:  I don't think that there is a purpose to weaken the surrogate.  If you mean with surrogate, the 75 percent level, that on the average the survivals would be in the healthy volunteers.

            That is not the case.  But when you do studies, like survival studies, those people who have done them, you see that there is individual variation, variability of donor recipients, who do go below that level, and that's a biological fact.

            And now, when you tell that even though everybody else has a mean, which is good, and you have a few people who are below that, it doesn't mean that the process is bad.  And if you go to the 42-hour red cells, AS red cells, that have been approved years and years ago, and I have here some data on them, individual data--it will not meet the current criteria.

            So this means that your products now available have different level of acceptance than the products that you are going to accept from now on.

            So you can't really--like apples and oranges.  The 42-day-old AS-103 red cell is not as good as some of the current new protocols provide.  And still we accept that we can use them.  Are we going to rethink the shelf life of AS-1 and AS-3?  I think we probably are not.

            DR. HE:  Can I say a few words about that.  Well, basically, that for the threshold value of the 35 percent confidence interval that greater than 70 percent of the products should have a recovery greater than 75 percent, we actually--when you're talking about it, that some people, they naturally have perfect recoveries, yes, we actually considered that fact.  That's why we actually have two out of 20 failures, three out of 24 failures, and if you are willing to do studies, 100 percent -- you are allowed to have 22 out of 100 failures.

            However, we all know that the blood cell studies are very expensive studies, and also the healthy volunteers will have to be exposed to radioactive material.  We try to limit the exposure as minimum as possible but to get the--as much useful information we can.  That's why the study sample size ranged from 20 to 24 instead of for hundreds.

            But we do give enough room for those people who has naturally poor recoveries.  And also for those products, or any approved, 20, 30 years ago on the market, many of them are now leukoreduced.  I think that most recently we have many of the red blood cells, particularly since 1996, '97, when the universal--I mean, the leukoreduction field test put in the market.  Most of the RBC products currently used in the market are leukoreduced products, and they're all from 1998, and we have plenty of products available in different hospitals for this more recent RBC product.

            And from the data you can see that.  If you count the studies from 1998 to 2007, 17 out of 19 studies current criteria and more than 90 percent data points can have recoveries greater than 75 percent.

            DR. SIEGAL:  Adrian, you had a comment.

            DR. DI BISCEGLIE:  Just a different tack, if I might.  You may have presented and I missed it.  Dr. He, can you tell us about the coefficient and variability.  Let's say you take 24 subjects and you measure it, and with a particular process, six of them fail.  If you take another 24 subjects with the same process, six of them had failed at the end, there might be one or two the next time round.  So the coefficient of --

            DR. HE:  I don't think we have data for that kind of studies, but we do think that if you want to do a study, you can do a two-arm study.  The same volunteer. You can take a unit in the approved product, which called controlled product, store up to 42 days, and you take a portion of the studies and ask that same donor come back at 42 days and donate another unit in the test product.

            DR. DI BISCEGLIE: It's the fact that we don't know the coefficient of variability, does make it a little harder to insist on a certain proportion, and that proportion may be negated.

            DR. SIEGAL:  Maybe we should go on and hear a little bit more about this problem before we--cause we're running a little late.  So if everybody agrees to that, let's go on to Dr. Richard Davey, professor of Pathology at The Methodist Hospital in Houston, talking about red blood cell recovery and survival, in vivo studies.

            DR. FLEMING:  Just as we go on--we will be able to answer your question, but we can wait till the end.

            DR. SIEGAL:  Sure.  We're going to have to have a lot of discussion about this anyway.

            DR. DAVEY:  Thank you, Dr. Siegal, and I want to thank the FDA for inviting me to present some information today.

            I think that some of my slides will relate to some of the questions that I think the committee's already asked Dr. He.

            So I'm at The Methodist Hospital in Houston.  Most of this work that I'm going to at least present in some overview fashion today was done when I was at the NIH in the '80s and '90s.

            Again, and I think what I'd like to do is talk a little bit about the evolution of radiolabeling, and then get into some of the evaluation criteria that you've already touched upon.

            First, we'll talk about labeling, a little bit of history of radiolabeling.  It was in 1950 when Gray and Sterling first described 51 chromium as a red cell radiolabel, and they used, Mallison and others used this technique to confirm the fact that this was an excellent label, and it also confirmed some studies by Ashby and others, back in the forties, of the red cell life span.

            So this has been a very useful label since 1950.  In 1967, Fisher described the use of 99m technetium as a red cell label.  Again, this has been very widely used in nuclear medicine and is useful for some adjunctive studies in terms of red cell recovery and survival.

            Then in 1989, Jim AuBuchon refined the use of 111 Indium as a red cell label and that has also proven to be useful.  We'll talk a little bit more about that in a minute.  111 Indium is especially useful, however, for labeling other blood elements like platelets and not so much for red cells.  So in terms of an ideal red cell radiolabel, we'd like to see a certain number of characteristics that would define what we might call an ideal red cell radiolabel.

            First, you want to have minimal preparation or manipulation of the red cells.  You don't want technical procedures to be too complex or to expose the technician to too much radioactive material in preparation of the red cells.  You want the label to be specific for red cells, nontoxic to the red cell, certainly nontoxic to the recipient, and there we get into radiation exposure.

            You want the label to not be metabolized by red cells, or to elute from the red cell.  And you like to have the radioactive half-life appropriate for the duration of this study and no relabeling of other cells in vivo.

            We'd like to have all these things, and no radiolabel meets all of these criteria, but one does better than the other, and that is 51 chromium.

            This is a label that now has been most useful for red cell radiolabeling and you can see perhaps why.  The half-life is ideal, in many ways, for the type of studies that we're looking at, where we're looking at red cell recovery, not only a day but sometimes it carries out even longer, several days, depending on the questions that we're asking.  The elution is really better than the other radionuclides.  Only about one percent a day, and that after some early elution, it's a little bit faster, is really a very steady state of elution that we can calculate in our determinations of recovery and survival.

            The major gamma photon emission is a bit high, 320 kilovolts, and that means it's really not suitable for imaging.  If you look at technetium, the problem we have there is the half-life is very, very short and the elution rate is very high.

            So this is really useful only for measuring red cell mass or red cell volume, and not for any longer survival studies.  And we can say the same about Indium, where we have a high and variable elution rate and a fairly short half-life.

            The advantages that both technetium and indium do have over chromium is that their gamma photon emission is really quite ideal for the sodium iodide gamma counters that we use.  So they're a bit easier to count and they can be used for imaging purposes.

            But for all practical purposes, what we use is 51 chromium.

            Just a very brief overview of the technical steps, which really have not changed in many years.  Basically, a suitable sample is of red cells that's been stored under the experimental condition, is chosen, these are almost always autologous studies.  We don't want to do allogeneic studies for ethical reasons.  So we use a reasonable size sample of 15, 20 milliliters of the materials.  The sample is taken from the experimental material, the experimental blood bag, etcetera.

            An amount of 51 chromium is used to ensure that you're going to have minimal adequate counts throughout the study.  That's about 4000 counts per sample, five to 20 microcuries or grays.  You incubate the test red cells and then you wash away, usually just by saline washing with the N-bound chromium, you make an appropriate standard for blood volume determination, the labeled red cells are infused into a peripheral vein, the exact volume of the infused material is determined, and samples are drawn from the opposite arm.

            You draw several early samples, and we'll see why in a minute, to determine the 100 percent survival rate.  Then the samples are counted in a gamma counter. Several different manufacturers make adequate counters for this particular technique.

            Just a couple points I think about how we learn to evaluate these survivals.  One I think key point is you have to--it's very important to get several early samples after the material is infused.

            We have to wait about three minutes for adequate mixing to occur, so we really can't get a zero time, can't get a zero time because mixing hasn't occurred properly.

            So the first sample can really get, we found that where mixing has occurred, is about three minutes.  We're going to get three, five, seven and a half, ten minute samples, because what we found is there is a rapid early loss of senescent or damaged cells, and if you don't get these early samples, you're going to underestimate the counts and therefore overestimate the survival at 24 hours.

            So you get these early counts, you extrapolate back to zero, you get your 100 percent survival point, compare that to your point at an hour or 24 hours down the line.

            The studies that Larry Dumont and Jim AuBuchon reviewed for the best analysis did use a single label technique that the previous slide illustrated.  There is another technique that's called a double label technique in which the actual red cell mass is determined by using a different radionuclide.  You can use 99m technetium along with the 51 chromium, labeling autologous cells at the same time with both radionuclides, and injecting them both, actually in a mixed sample.

            And with using the technetium to determine your red cell mass, and then the chromium to look at your long-term survival of the bag or the plastic, or the anticoagulant understudy.

            This is a tad more accurate.  Voitler has shown in some studies, a number of years ago, however, that the error that's introduced, the small error that's introduced by only using the single label technique is really deminimis, and in general now, most people just use a single label back extrapolation technique for these studies.  And just again to address a couple of the questions from the earlier speaker.  There have been a number of red cell products and red cell derivatives have been evaluated, using these red cell recovery studies.  I guess my pointer, I think, is not working.

            Certainly new plastics and plasticizers, new anticoagulants and preservatives.  New filters, for high-efficiency filters that we've all used.  Leukoreduced products.  Irradiated red cells.  Red cells that have been collected and--please; if you would.  Thank you very much.   Okay.

            Irradiated red cells.  Undercollected red cells.  We've looked at red cell units that don't quite meet the criteria when a donor vein collapses, or whatever, and you don't get quite enough in the bag.  That's been evaluated, to see if those red cells survive adequately.

            More recently, these technologies have been used to look at pathogen inactivation.  For instance, the material that I think Dr. Korash was presenting earlier, the pathogen inactivation technologies that he's perfected are evaluated with these technologies.

            There's been some interest in enzymatic removal of A&B antigens from INB red cells, and those studies have also been studied with these radiolabel survival studies.

            So they've been very useful in helping the FDA find ways to approve and evaluate some of these important products that we're using today.

            Just one set of examples here, if you can see this.  One study that was done a number of years ago, where we looked at the effect of prestorage irradiation on 24-hour post-transfusion red cell recovery.  We were beginning to learn about irradiation, there was damage to red cells by irradiating red cells.  So the question was, and the FDA asked this question of us: Does irradiation damage red cells to the extent that we might have to adjust the storage time for irradiated red cells?

            Now these studies were, again, I think illustrating some of the issues that we faced a decade or two ago, and that these studies were done at two laboratories under slightly different conditions.

            You can see the radiation dose differed between laboratory one and laboratory two.  Laboratory one studied irradiated red cells 21 days after irradiation and 28 days after irradiation.  My lab studied them 42 days after irradiation and you can see here's control, here's the irradiation, control irradiation, control --

            That over time, the irradiated red cells, even from 21 days, seemed to be in a bit worse shape in terms of storage than their control counterparts.  With potassium always higher, with ATP always lower, and with a 24 hour recovery--I guess you can't quite see it here--but anyway, you can see the recovery with the irradiated red cells went from 90, 85, to 78, and I think actually I'm afraid these got out of--you've got the line shift off-line.  I'm sorry.

            So that actually, this 68.5 should be under the irradiated 42 day column.  So this line shift should be shifted one notch over to the side.  I'm sorry.

            So the irradiated cells compared to control, with the first study was 90 versus 82, second study, 85 versus 80, and the third study, 78 versus 68.

            So here's 68.5 mean recovery after 42 days, after irradiation, and these data I think were important for the FDA to revise their storage time for irradiated red cells to 28 days after irradiation, or 42 days, whatever came first.

            So that's a bit about radiolabeling.  Again, the technology has not changed and the laboratories that are doing studies of red cell recovery and survival these days, like Dr. AuBuchon's and others, are really expert at doing this technique.

            So if we look at the evaluation criteria, and again, Dr. Hay and others have already gone over this, in '47, Ross did propose a 70 percent red cell survival, and it was somewhat arbitrary, as you've heard, that that percentage was determined.

            In 1985, and again you have heard that the minimum survival was raised to 75 percent following an FDA workshop.  In 1998, the minimum survival was still 75 percent with a standard deviation of less than 9 percent, based on historical data, and the current data, the current criteria are now based on historical data and input from BPAC a few years ago.

            And again you have heard these FDA requirements now for the approval of red cell products.  I don't think I need to go over them in any more detail.  I think again the key point that I think the industry has questioned is, as you've heard, now the requirement for one-sided 95 percent lower confidence limit, for a proportion of success, for successes must be greater than 70 percent, with a success threshold for individual recovery of greater or equal to 75 percent.

            And what this means is--and again we've heard that a successful trial can have no more than three of 24 red cell recoveries below 75 percent.  That's a pretty high bar.  There is individual variation and you do find recoveries in many studies, just because of ill-defined idiosyncracies in the way certain individuals manage their red cell recovery and removal from the circulation below that particular percentage.

            Other in vitro measures that we use are hemolysis.  You compare the control in the test for ATP, 2,3-DPG, ph, glucose, lactate, and a number of other in vitro comparisons.

            All of these, as I think it was brought out in the earlier discussion, are really somewhat surrogate markers for how well these red cells are doing in storage and how well we assume, without a lot of definitive data, they're going to do after transfusion in vivo.

            So let's talk a little bit about old blood because it is an issue today.

            We want to make sure that the red cells that are collected from our donors are collected in the most optimal manner and the most optimal solutions and plastics that we can.  These are accumulating data, and they're data on both sides of the coin.  But the data are accumulating, that the storage lesion we see with old blood is indeed significant.  There are changes.

            If you look at blood, over time, and these are exaggerated a bit, but you're going to see after prolonged storage red cells changing to spherocytic forms, and then ecanthocytes, and after really prolonged storage, beyond normal storage times, you're going to have really a very deteriorated-looking red cell smear.

            And there is a storage lesion that we've been aware of, we've been aware of for years, but I think the data are accumulating that it may be more extensive and more clinically significant than we may have thought in the past.

            On storage, you're going to see 23, decreased 2,3-DPG which affects oxygen, transfer from red cells to the periphery.  The red cells get more rigid.  There's a high PA-1 level.  There's a high CD-40 level in the bag.  Nitric oxide is depleted.  There's an increase in IL-10.  There's a decrease in tissue necrosis factor.

            All of these, and other factors, we have been finding are a result of, in prolonged storage with poor 02 transport, and inflammatory problems, thrombotic problems, and there is increasing evidence that old blood is, indeed, not as clinically efficacious as fresher, newer blood.  This is causing some issues for us in blood banking, because we do have shortages, we can't give people fresher blood when they request it, we have to use the first in/ first out technology, like we use the milk at the grocery store.  First in/first out.  Otherwise, we're not going to be able to maintain our inventory.

            So again just some slides illustrating some of these factors.  This is looking at 2,3-DPG.  During storage, you can see by a week, 2,3-DPG is almost a nondetectable levels.

            It does recover after transfusion but it takes a day or two to get back to clinically important levels.

            The red cells become more rigid.  There's a measure of red cell deformity, over time, and by several weeks, you can see the deformability index, indeed, indicating a more rigid, less pliable red cell after storage for extended times.  And this is some data just looking at nitric oxide.  We heard a lot about nitric oxide the last couple days when we were looking at ATRX, another study at NIH.  But it certainly seems to affect the quality of stored red cells also.

            Here's just a cartoon that was taken from one of the hematology papers, recently, just showing that red cells that are rich in nitric oxide are able to maintain or increase vasodilation, as we would expect, in vivo.  However, stored red cells get rapidly depleted in nitric oxide, and therefore may indeed not only not maintain vasodilation but may actually contribute to vasoconstriction.  This is hypothetical.

            There is some research now under way, where there's, actually, technology's now looking to enrich stored red cells with nitric oxide again, hopefully resulting in a more beneficial, in most patients, vasodilatory effect than nitric oxide conveys.

            This is a study, I hope you can see it, it isn't too dim--an important study, again helping us understand that less blood is better.  There's some other studies, that fresher blood may be, quote, better.  This is a very landmark study, ten years ago, by Hebert.  I'm sure many of you are familiar with this, a TRIC study done up in Canada.  A very nice study, looking at patients who received--ICU patients very carefully matched, about 400 and some in each ARM.  Half the patients got a restricted transfusion policy, meaning they were only transfused up to hemoglobins of 7 to 9, versus the liberal transfusion policy, transfusing these patients to a hemoglobin of ten to twelve.

            And you can see, that if you look at the p values of death, 30 day ICU hospital death, nothing reached exact significance, nor were there any significant changes in length of stay.  But the trend was clearly toward restrictive blood transfusion seemed to be better, across the board than liberal transfusion, certainly in terms of the death in 30 day ICU, and hospital death statistics.

            So what this showed was not so much that restrictive is the best way to go, but it showed that it is at least equal to the more liberal transfusion policy, and may indeed be better. 

            Here's a study, and I hope I'm not overlapping the next speaker's presentation, from Koch, et al, and from 2006, looking at factors associated with postoperative morbidity, post-CABG.  and you can look at all the factors that were associated with postoperative mobility, all these preoperative factors, higher body mass index, and normal ejection fraction, blah, blah, blah, all significant.

            But the most significant factor was transfusion, and they concluded, if I can read this--that "Perioperative red cell transfusion is the single factor most reliably associated with postoperative morbid events."  A pretty important conclusion, I think from that particular study.

            And we've heard about this study already, the study in the New England Journal, six weeks or so, by Koch.  I'll go over it very briefly.  The hypothesis with this group was that serious complications of morbidity after cardiac surgery may increase when red cell units are transfused, after they've been stored for more than two weeks.

            In this case patients underwent coronary bypass grafting, valve surgery or combination, a big study, over 6000 patients enrolled from 1998 to 2006.  Pretty much all serious morbid events were considered.  When blood was issued, it was issued, like I say, first in, first out, and the median time of storage in their blood bank was 15 days.

            And you've seen the slide from the previous presentation.  They did indicate, and with a number of other parameters, that the death rate--although these are small percentages, if you look at the percentages--the death rate for those receiving older blood was indeed higher than those receiving fresher blood.  And they concluded--although this study had problems.  Some of the confounding variables were not really appropriately addressed.  But they did conclude that patients given older blood had a greater in-hospital mortality, were more likely to need prolonged ventilatory support--these are all significant findings--more likely to have renal failure, more likely to have increased sepsis, and have increased multisystem organ failure.  Not so good for the old blood story.

            So just in conclusion, then, putting this altogether, we don't have definitive data that the way we evaluate red cells in storage directly affects clinical outcomes.  But I think we have a lot of surrogate data and ancillary data that can help us decide, especially some of the questions that the committee is considering today.

            We've seen, I think, the technical labeling procedures are well-established and are performed by skilled laboratories, and I think we heard in the previous presentation that the FDA's experiencing, recently, that many, if not most of the submissions, are meeting the fairly strict criteria that are now in effect.

            But I think again, looking at these increasing data that storage lesion is real, that older blood may be a bit more problematic than fresher blood, it might be reasonable to be prudent, and to be cautious in liberalizing threshold value at this point.

            However, we don't want the bar to be so high, that we lead to unnecessary trial failures.  And I'm just jumping ahead to some data that Larry and Jim kindly shared with me, that I think we'll hear from Larry.

            If you look at his data, there was a 67 percent historical success rate for liquid products, if the 75 threshold value currently in effect were applied to the historical studies.  Only two-thirds of those studies, looking at liquid products, would have been approved.  That's a high bar.  However, I don't think we want to set the bar so low, you have almost a 100 percent success.  And again if we look at--if we set the bar at the 67 percent threshold that's been proposed by BEST, you'd have a 99.9 percent historical success rate for liquid products, if I understand the data correctly.  That's a pretty low bar.

            I suggest that we might have alternatives to consider that have already been presented, to some extent, with the prior presentation.  A 70 percent threshold value would give a 95.5 historical success rate for liquid storage.  That, to me, seems like a reasonable percentage of success versus failure.  Now others may differ.

            And you might want to consider somewhere in the medium, we've seen some data that is 72, 73 percent threshold value, may be a reasonable place to come down.

            So with that, I'd like to conclude my comments, and any questions I'd be happy to answer.

            DR. SIEGAL:  Thank you very much.        Are there any questions?

            DR. CRYER:  I've got a methodological question.  That the process that you showed for measuring this, it looked like you said something about 11 cc's coming out, and then incubating, and then going in.  But what you're measuring here is  a process of storage, so that doesn't make sense.  So I'm going to have to take a unit out and store it the way that it would be stored--

            DR. DAVEY:  Right.  What you do--hypothetically, it's a new anticoagulant.  So yes, we would collect those units and the new anticoagulant, and also some control units in the standard approved anticoagulant, and then after the proper storage period has elapsed, you would take a sample of that material for radiolabeling and injecting.  We don't inject the whole unit but we'll take a sample from that unit, radiolabel it and inject it, and measure recovery.

            DR. CRYER:  Then you said you have a control, so that implies a gold standard?

            DR. DAVEY:  It implies--in most cases, you measure it against an approved product.  It depends on what you're looking at.  For instance, in the irradiated study, I looked at nonirradiated versus irradiated, and if you're looking at a new anticoagulant, you might look at the approved anticoagulant versus the new anticoagulant.

            DR. CRYER:  That ought to overcome the variability issue that Dr. Szymanski was worried about.  No?

            DR. DAVEY:  You're still going to have individual variation, but if you're controlling it with the same subjects, which you do, you often use the same subjects for the control and the experimental, you often can--but a person who has a poor survival in the control will often have a poor survival in the experimental.

            So while it does cancel it out, you still have the poor survival problem that we've run into in terms of the survival, the successful threshold problem.

            DR. SIEGAL:  Dr. Szymanski.

            DR. SZYMANSKI:  You talk about threshold.  Is that threshold the same as mean survival required?

            DR. DAVEY:  No.

            DR. SZYMANSKI:  Mean percentage in vivo survival.  Or what do you mean with the "threshold."

            DR. DAVEY:  I'll let the FDA define how they're defining threshold.  I think we can go back to the second or third slide; if we can.  I can read it directly.

            The threshold is one-sided 95 percent lower confidence limit, with the proportion of successes must be greater than 70 percent, with a success threshold for individual recovery of equal or greater than 75 percent.  Okay?  This was reading what their current criteria are.

            DR. SIEGAL:  Dr. Epstein.

            DR. EPSTEIN:  Yes, thank you, and thank you, Rick.  Just on your last slide.  I'm hoping you can comment on one point.  Dr. He pointed out that part of what conditions FDA's thinking, is that as technologies have progressed since 1990, an ever-higher proportion of technologies are meeting the higher standard, and the argument that you're making under alternatives really comes down to lumping the entire time period again.

            And so what I'd like you to comment upon is do you not think that FDA ought to hold better products to higher standards in the interest of putting better and better products on the market?  I think that lies at the core of what we're going to ask the committee.

            DR. DAVEY:  Right.  I think, Jay, you're obviously getting to the core of the question.  Are we setting the bar appropriate for current products that are being presented to the FDA?  Is the bar appropriate for the present products, and the laboratories that are now very skilled in looking at these products, and the technical advances in storage and plastics--is the bar appropriate for what's being submitted for the FDA today?

            Is that your question, more or less?

            DR. EPSTEIN:  Well, the data that you're showing aren't really for the 2004 to 2007 period.  In other words, it's a little bit confounding the issue cause you're lumping the data back to 1990.

            DR. DAVEY:  These data are actually from AuBuchon-Dumont paper, Jay.

            DR. EPSTEIN:   Right.  and the AuBuchon-Dumont paper aggregates all the studies for which data were publicly available, going back to 1990, and it overlooks the argument that Dr. He made, which is that if you look at how the approved products would have performed, over time, there's a trend toward improvement in products, and that that's consistent with the notion of holding newer products to a higher standard.  If we don't ever do that, then products don't ever improve, and again, I think that that lies at the core of what we're going to ask the committee.

            In essence, if the committee, in the end, concurs with FDA, to adhere to the 2004 standard, the committee is saying yes, in ongoing product approvals we want products to meet the standard that products were capable of meeting from 2004 on, but were not necessarily capable, or at least not at the same rate, of meeting before that time.

            So we're basically saying isn't there product improvement to which we should hold all future products.

            DR. DAVEY:  Right.  I think that was--

            DR. EPSTEIN:   And I think that's a core point that gets overlooked here.  Again, the reason is that you've used the statistics from the Dumont-AuBuchon paper which aggregate all the data from the year 1990.

            DR. SIEGAL:  Dr. Fleming.

            DR. FLEMING:  Actually, I'd like to amplify your question, because technically, the FDA is not even asking as much as what you just said.  I think what you just said is products are improving and shouldn't we hold you to a standard of being maybe even above where we currently are.  You're not asking that.  You're basically saying--

            DR. EPSTEIN:   We're not raising the bar.  We're saying that--

            DR. FLEMING:  You're not raising the bar.  In fact, essentially what you're saying, if you use the data from the table, and there was an 83.6 percent success rate, individually, in 1997, up to 88 in '98 to 2003, and up to 93 in 2004 to 2007, you're saying if you're just average relative to the last five years, you'll have a very high chance of success.

            Even if you're just average relative to the preceding half decade, you'll still have a very good chance, and you're asking is that in fact an acceptable standard to hold, which is not saying you have to be even better than the products now.  It's saying if products have improved, over time, is it in fact important to be at least, on average, similar to those products?

            DR. DAVEY:  I think we could think, if we look at what has been approved in the past--I mean, there are other correlates in the pharmaceutical world about post-licensure surveillance data, like with Vioxx or aspirin, if you will.  That if we hold currently-licensed products to the criterion of today, many of them would really not be on the market, and I think we have to consider the fact that the 75 percent threshold--to take Jay's point--we still have to look back at what's on the market toady.

            One-third of them would not have been approved, and I think we'd be missing out on some very key red cell products and derivatives and manipulations, if we had put the current standards into play today, and I think on the other side of the bar.  We shouldn't send it as low as I think has been proposed by BEST by far.  We don't want a 100 percent approval.  I think we can find a middle ground.

            DR. ZIMRIN:  I think that's right.  I mean if these products are so bad, then why are you happy with them being on the market?

            DR. EPSTEIN:   They're not so bad.  It's just we're making better ones, and, you know, this gets to a core issue in regulation, which is that as technologies improve we don't delicense the products that were previously approved, unless we have specific safety concerns.

            And so you could say, well, that's not even-handed, but it is a fact of life, and what we rely upon is that the practitioners, if informed about the characteristics of products, over time, will make medical decisions on which products to use.

            So this is a common feature with product approvals of all sorts.  The old products, unless they are found overtly harmful, don't come off the market because of regulatory standards.  In other words, we have a high threshold for taking a product off the market, but it doesn't mean that we shouldn't seek product improvements over time.

            So it's unfortunate, but I think we have to dissociate those two questions.  In other words, to argue that just because we had a lower standard in 1990, we shouldn't have had a higher standard in 2004, means that we should never raise standards, and the market effect of raising standards is less direct, but it's achieved through labeling and it's achieved through, you know, aggregated clinical experience.

            I mean, doctors have to ultimately figure out, as Dr. Di Bisceglie was saying, it doesn't really matter, and we don't know right now, but I don't think it's an argument against wanting better standards for newer products.

            And again when I say better standards, it's the standard of 2004.  It's the thing already in place.

            DR. SIEGAL:  Tom again.

            DR. FLEMING:  Just very briefly.  I understand your point.  We should seek better products, over time.  I just want to again reiterate--maintaining your current standard isn't forcing you to better.  Maintaining your current standard that you have here gives you a very high chance for success if you're just average relative to the current day products, and even if you're just average to the products a decade ago, which is less than the average now, you still have a very high chance of success.

            So we should be striving for better products but maintaining this current standard isn't requiring you to better.  You can be just average for what's out there, or even be somewhat less than average for what's out there, and still have a high chance of being approved.

            DR. DAVEY:  I just think that having heard what we've heard is that what we're doing now is potentially hurting people, at least if you're getting blood that's been stored longer than 14, 28 days, depending on which paper you're looking at, and that we ought to be considering a higher standard.

            DR. ZIMRIN:  I really don't think that's been established.  I think the conclusion of every study that's been published on this is that prospective studies need to be done to establish whether this is true or not.  That's the conclusion, uniformly.  There's so many confounders in looking at this data, they're all retrospective, that to take from these presentations, that we know that old blood is bad, I think is wrong.

            So I think that we need to move towards studies that look at this question, I think it's a tremendously important question, but I think we clearly don't know the answer right now.

            DR. SIEGAL:  Yes?

            DR. CRYER:  I would agree with that, and I would even say even if the data are convincing that old blood is bad, I'm still struggling for a link between red cell survival and old blood being bad.  I mean, maybe a few residual white cells being found, that's bad. 

            But I have a question, Dr. Epstein, if you could help me understand, because this threshold is used to regulate a number of different products, and some of them may be "me too" products, in which case we definitely want better, we want to improve.

            However, is it not possible that we might be trying to regulate a new product that is unique, that is worthy, where there's a great need for it, and then you might--if it's a "me too" you'd say of course, you know, we want to stick with it, want to keep the standard as high as we can.  That some people which there=s a great need -- are we still going to regulate it with the same --

            DR. EPSTEIN:  Stringency maybe.

            DR. CRYER:  Stringency.

            DR. EPSTEIN:   Well, the answer's no.  I mean, we do recognize alternative situations, different needs, and I think one example here is the irradiated red cell, which is important to prevent graft-versus-host disease.  Another example is the frozen and thawed red cell, which enables keeping rare units, and we're not actually applying the same standard.  And you'll see data, if we ever get to the other presentations, that those products would not meet the current general FDA standard.

            So we do recognize that there are different products meeting different needs.

            DR. SIEGAL:  With that in mind, maybe we should get on to the next presentation.  This is a presentation by Dr. Mark Stafford-Smith, professor of Anesthesiology at Duke.  He's going to talk about age of transfused blood, short-term mortality and long-term survival after cardiac surgery.

            Dr. Stafford-Smith.

            DR. STAFFORD-SMITH:  Okay.  Thank you very much for inviting me to present today.

            To some extent, I'll be covering topics that have been discussed, but hopefully shed some more light on the topic.  The topic is age of transfused blood and its association with mortality and long-term survival after cardiac surgery.  I have some disclosures but none really pertinent to the presentation topic.

            Just to review in terms of cardiac surgery, I think it's a particularly good model to examine, as much as there are limitations in retrospective evaluations, it's one of the very large datasets we have.

            Quality assurance datasets are being accumulated to very sizeable numbers of patients at many institutions for quality assurance, and are very accurately merged with blood bank datasets.

            It's also a large part of our transfusion burden.  Over 2 million red blood cells a year, about 20 percent of all red cell transfusions, and approximately 400-, 500,000 procedures a year.  So we're looking at a very large dataset we can look at.

            It's also a relatively homogenous result in many other ways, and as well as we're going to be able to look at things retrospectively, it has those elements that make it possible.  Other datasets that are also looked at in evaluating these issues are septic patients, trauma patients, and critically-ill datasets.

            So these are really the people who are receiving a lot of blood.

            One of the issues we've talked about here is, you know, what is the actual end point we're looking for?  Is it cell survival, and so on?  And I think in at least the clinical world that I live in, a lot of what we've been coming up with recently is drugs that we know were efficacious, and it's post-marketing involvement that demonstrates subsequently.

            In fact, even in retrospective datasets, for example, in the case of  aportinin, were accepted essentially as evidence to take the drugs off the market, and so we're really talking about efficacy versus safety, and my potential interpretation of this is that efficacy is the cell survival whereas these other elements are almost a different question, the safety evaluation.

            Given this, we have started to look, obviously, in blood products at the same issue in the U.K., looking at retrospective data, decided to change their management of