Blood Products Advisory Committee

91st Meeting, May 1-2, 2008

Rockville, MD




Topic I:  BEST (Biomedical Excellence for Safer Transfusion) Committee Report on Red Blood Cell Recovery Standards


Issue:   FDA seeks the advice of the Committee on an industry proposal to change the current acceptance standard for evaluation of red blood cell (RBC) studies based on in vivo RBC radiolabeling recovery studies.




Red blood cells (RBCs) are a life-giving therapeutic transfusion component for which, currently, there is no substitute.  Approximately fifteen million RBCs units are collected and transfused in the US per year with the intent of providing increased capacity to deliver oxygen to tissues of patients who are anemic. 


FDA evaluates novel red cell products to assure that they are at least equivalent to other red cell transfusion products already on the market.  Novel red cell products include red cells collected into bags containing novel anticoagulants; stored in new additive solutions; stored in new storage bags; processed by new devices - such as apheresis instruments; prepared with new leukocyte reduction filters; made with bedside component separators; new blood warmers; pathogen reduction treatments; or any other factor that could effect the quality of the red cell during its shelf life. 


The FDA evaluation process relies on in vitro tests and on red cell recovery based on in vivo radiolabeling clinical studies.  The in vitro tests characterize RBC morphology and biochemical status that include red cell shape, size, hemoglobin concentration, hemolysis, cell surface markers, glucose utilization rates, lactate production rates, and ATP levels and utilization rates.  Some of these tests have been in use for decades while others have been recently developed and adapted.  The tests do not include evaluation of oxygen carrying capacity because, historically, it has been assumed that minor modifications to storage conditions do not alter the RBC ability to deliver oxygen.  Extrapolation of in vitro tests results to predict in vivo function of the RBC has had, over the years, only limited success.  Highly abnormal tests are usually predictive of poor in vivo performance but normal or slightly abnormal in vitro tests may also be associated with decreased in vivo performance. 


Mollison (1) indicated that “The viability of a sample of stored red cells cannot be accurately predicted from any measurement made in vitro, so that measurements of post-transfusion survival remain essential in developing improved methods of preservation”.  Because of the difficulty of extrapolating in vitro results to clinical performance the FDA has relied on in vivo studies as the gold standard for evaluation of RBC products.  The in vivo test involves radiolabeling of autologous red cells donated by a healthy volunteer and re-infusing these cells into the same volunteer.  The amount of re-infused cells remaining in circulation is followed by serial blood sampling over time.  It has been shown that damaged cells are usually removed from circulation during the first 24 hours and thus a single 24 hour time point has been used to evaluate the quality of RBCs (1).  The percentage of red blood cells recovered at a 24 hour time point after transfusion is the recognized surrogate endpoint or threshold value of red blood cell efficacy.  Historically, based on the expert opinion in transfusion medicine, FDA has used a threshold value of ≥ 75% radiolabeled RBC recovery at 24 hours to evaluate a RBC product (2,3).  Studies showed the longer the storage of RBC, the poorer the RBC recovery (1).  In recent years there have been numerous studies that retrospectively analyzed outcomes of red cell transfusions and have suggested that transfusion of older blood may be associated with a higher rate of morbidity (4-6).   


In a public workshop in 1985, the FDA and the transfusion community agreed on a minimum criterion for recovery of radiolabeled red cells at 24 hours post infusion.  The criterion was set as a mean recovery of >75% in a group of 20-24 volunteers.  In 1998 the FDA modified the standard based on requests from manufacturers to provide additional statistical criteria related to the mean recovery of 75% value, namely a standard deviation requirement.  The FDA adopted the following criteria based on examination of previously submitted data for approval or clearance of red cell products:


Radiolabeling studies should be performed in at least two separate centers (laboratories) with a total of 20-24 healthy donors.  The mean recovery at 24 hours should be > 75% with SD < 9%.


The FDA used this criterion to evaluate RBCs until 2004.  At that point the FDA reviewed several applications for new RBC products where this criterion was satisfied.  In one study, one third of the RBC collections tested (8/24) did not have recovery > 75% even though the overall acceptance criteria were met.  The FDA was concerned that approving an RBC product with such a high proportion of failed RBC collections was not in the best interest of public health.  To address this problem, the FDA proposed a change to the acceptance criteria as follows (new criteria known as the “95-70 rule”):


Radiolabeling studies should be performed in at least two separate centers (laboratories) with a total of 20-24 healthy donors.  The mean recovery at 24 hours should be >75% with SD < 9% and with the one sided 95% lower confidence limit for the population proportion of successes >70% (Successes mean individual units with recovery ≥75%).  To meet the criteria, if sample size is 24, no more than three individual units should have the 24-hour recovery less than 75%.  The three failures are allowed to accommodate volunteers who may have a naturally reduced recovery of red cells.  If a 24-hour RBC recovery study passes the criteria, one can conclude with 95% confidence that more than 70% of the units will have a 24-hour RBC recovery ≥ 75%.


Under these new criteria (known as the “95-70 rule”) a successful study in 24 volunteers will mean that RBC from at least 21 volunteers will have recoveries above 75% in 24 volunteers.  This approach was discussed at the July 2004 BPAC meeting (7).  Consistent with the advice of BPAC in 2004, since that time FDA has been evaluating new RBC products by these revised criteria. 




In the last year the FDA was approached by manufacturers of RBC processing devices in regards to the new criteria.  The manufacturers were concerned that some new RBC products would not able to meet the new criteria and speculated that some RBC products already on the market would also not meet the new criteria.  The manufacturers volunteered to provide archived data from radiolabeling studies that had been used to support previous approval/clearance of products currently on the market for re-analysis by the new criteria.  Drs. Larry Dumont and James P. AuBuchon collected and analyzed the available dataset with the current criteria (8).  The objective of the BEST study was to define the ability of currently available RBC collection and storage systems to satisfy the revised RBC recovery criteria proposed by the FDA for approval of RBC systems.  There were 641 radiolabeling studies performed on liquid cold stored red cells between the years 1990 and 2006.  Other storage conditions included in this data set were frozen red cells and gamma irradiated red cells, which are special circumstances and will be considered separately in future.  After analysis of the liquid cold stored red cell data, the authors concluded that many of the approved RBCs would fail the current criteria.  They suggested that current red cell products would meet acceptance criteria 99.9% of the time (Table 1 last column) if these acceptance criteria, i.e., the success threshold or cut-off for individual RBC recovery were lowered from 75% to 67%.


Table 1. BEST study results from Drs. Dumont and AuBuchon.  Probability (power of the study) to meet the one sided 95% lower confidence limit for the population proportion of successes >70% if different cut-off values or threshold values are used to define a successful in vivo recovery (i.e. 75% vs. 70% vs. 67%).  Results are based on analysis of 641 liquid stored RBC recovery studies done between the years 1990-2006.





with mean

 > 75%


with SD

 ≤ 9%

Probability to meet current FDA criteria (95-70 rule) (i.e., Number of successes > 21/24)







> 75%




> 70%




> 67%










The FDA has been given access to the dataset used in the BEST study (FDA appreciates Drs. Dumont and AuBuchon sharing this information).  The BEST analysis included 4 studies performed after the revised criteria were put in place in 2004.  The FDA has been able to add additional 4 studies (n = 94) that were approved/cleared by the FDA that met the revised criteria after the new criteria were put in place.  To date, there have been a total of 8 submissions approved/cleared by the FDA since 2004.  --------------------------- ---- ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------  FDA has analyzed the dataset provided by Drs. Dumont and AuBuchon in combination with the 8 recent successful submissions and notes that the more recent studies performed between 1998 to 2007 had a higher rate of success in meeting the current criteria (see Table 2):


Table 2.  Comparison of the success rate (fraction of recoveries that meet the threshold value in a study of 24 RBC collections) and power (probability of success for a study to meet the threshold of > 21/24 recoveries) for studies done in different time periods for liquid stored RBC recovery studies.  Results are based on combined BEST (n = 595) and FDA (n = 94) data sets analyzed by FDA between the years 1990-2007.  Two studies (n = 46) were included in the BEST, but not in the FDA analysis because the dates when they were performed were uncertain.  However, both studies have passed based on current criteria.




Threshold value


(n = 140)


(n = 376)


(n = 173)


(n = 549)

 success  rate

power         n = 24

 success  rate

power         n = 24

Success  rate

power        n = 24

success  rate

Power        n = 24

RBC% ≥ 75%









RBC% ≥  74%









RBC% ≥  73%









# of studies meeting current criteria

4 out of 8


9 out of 11


8 out of 8


17 out of 19



The observations in Table 2 suggest that the overall quality of red blood cell products cleared or approved by the FDA is improving with time.  Consequently, relaxation of the standard based on the quality obtainable in earlier time periods would mean lowering the standard even though most recent products submitted to the FDA passed the higher standard and, therefore appears unnecessary.  Based on the data in Table 2 the power for a 24 volunteer RBC recovery study from 2004-2007 to meet the current criteria (after “95-70 rule” was introduced) with the threshold value of > 75% is 0.92.  In contrast, for products reviewed from 1998-2003 (before the “95-70 rule” was introduced) the power for a study of 24 volunteers to show conformance with the new FDA standard was only 0.69.  Most clinical studies performed to satisfy FDA criteria for drugs are done at 0.80 power.  The most recent studies, 2004-2007, exceed 0.80 power based on an observed success rate of 93.1%.  If we consider an expanded recent time period, i.e. 1998-2007, a study of 24 volunteers will have 0.82 power (> 0.80) if the threshold value for RBC recovery is decreased to > 74% (see Table 2).  


A threshold value of > 75% has provided the standard for red blood cell quality over the last 24 years.  FDA recently modified (2004) the statistical approach to these studies in an effort to assure that >70% of the RBC products will have RBC in vivo recovery greater than 75% as has been recommended by the experts in the field.  A retrospective analysis of recovery studies demonstrates that most products submitted to the FDA in the last 10 years (17 out of 19) can meet threshold value of 75% recovery indicating an improvement in red cell products in comparison to the previous decade (4 out of 8). 


Based on these considerations, and the premise that such enhanced RBC recovery is achievable and is clinically desirable, FDA proposes to continue applying the criteria adopted in 2004 (“95-70 rule”) to the quality evaluation of red blood cell products using in vivo radiolabeling studies. 


Questions to the committee:


1. 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 (laboratories) with a total of 20-24 healthy donors.  The mean recovery at 24 hours for each unit should be > 75% with SD < 9%; and the one sided 95% lower confidence limit for the population proportion of successes > 70% .)


2) Alternatively, does the Committee recommend that a change in the criteria is needed based on the data?


3) If the answer to question 2 is yes, what changes does the Committee recommend for the threshold value of individual subject red blood cell in vivo recovery, with a          sample size of 24?


Examples to consider are based on Table 1 and 2:


a)  Based on combined data from 1998-2007,  > 74% as the threshold value, power 0.82


b)  Based on combined data from 1998-2007,  > 73% as the threshold value, power 0.93


c)  Based on BEST data from 1990-2006, BEST recommends 67% as the threshold value, power 0.999


(Note that most clinical studies performed to satisfy FDA criteria for licensure of Drugs and Biologics have 0.80 power.)




1. Mollison’s Blood Transfusion in Clinical Medicine, 11th Edition.  (2005) Klein, H.G. and Anstee, D.J. editors.  Chapter 9.  Transfusion of red cells. Pages 362-373. Blackwell Publishing Ltd.


2. Ross JF, Finch CA, Peacock WC, Sammons ME.  (1947) The in vitro preservation and post-transfusion survival of stored blood. J Clin Invest 26:687-703.


3. Grindon AJ. (2001) Blood collection.  In: Clinical Laboratory Medicine.  McClatchey KD (eds).  2nd ed., Williams and Wilkins, pp. 1526-1528.


4.  Tinmouth A, Fergusson D, Yee IC, Hebert PC.  (2006) Clinical consequences of red cell storage in the critically ill.  Transfusion 40:2014-2027. 


5.  Basran S, Frumento RJ, Cohen A, Lee S, Du Y, Nishanian E, Kaplan HS, Stafford-Smith M, Bennett-Guerrero E.  (2006) The association between duration of storage of transfused red blood cells and morbidity and mortality after reoperative cardiac surgery.  Anesth Analg 103:15-20.


6.  Koch CG, Li L, Sessler DI, Figueroa P, Hoeltge GA, Mihaljevic T, Blackstone EH.  (2008) Duration of red-cell storage and complications after cardiac surgery.  N Engl J Med:358:1229-39.


7.  July 2004 BPAC transcripts:  Irradiated RBC acceptance criteria. Pg 60-204


8. Dumont LJ, AuBuchon JP.  (2008) Evaluation of proposed FDA criteria for the evaluation of radiolabeled red cell recovery trials. Biomedical excellence for safer transfusion (BEST) collaborative.  Transfusion Feb 21 [Epub]