Does red blood cell storage affect clinical outcome? When in doubt, do the experiment

Questions about the efficacy, safety, and availability of blood products are major concerns facing the FDA and the National Heart, Lung, and Blood Institute (NHLBI) as well as health care providers and administrators at hospitals. One of the basic questions that has been posed episodically but repeatedly for decades has been whether or not the changes that occur in red blood cells (RBCs) during storage affect clinical outcomes. Although the urgency of the question has waxed and waned over the years, it remains unanswered. This issue of TRANSFUSION offers two articles that address this issue. Lelubre and coauthors1 review our current level of understanding of the impact of RBC storage on patient outcomes and point out some of the difficulties of interpreting the data from these clinical studies. The clinical design feasibility study of Bennett-Guerrero and coauthors2 highlights some of the challenges of conducting definitive clinical trials in this area. Together, they help us focus on the direction we need to move, to address the questions that persist about the clinical impact of RBC storage. Questioning the clinical implications of RBC storage is again timely, because RBC transfusion practice is currently a widely debated topic in administrative, financial, and legal venues as well as the academic forum. Minimizing blood component inventory, establishing transfusion guidelines, policing utilization, and implementing systems to eliminate adverse events related to product administration are but a few discussion points. Unfortunately, rigorous clinical trial data are lacking to support many of the policies being developed. Even where clinical evidence to change practice does exist, such as the safe reduction of RBC transfusion threshold demonstrated in the TRICC trial,3 attempts to change practice have had variable success.4, 5 In addition, generating the science to establish sound clinical practice becomes increasingly difficult in the face of public opinion influenced by the media, in particular, those reports sensationalizing the detrimental effects of “old blood.”6 Although important, the issue of how RBC storage affects patients is only one of many fundamental questions in transfusion medicine that remain to be answered. The delivery of oxygen and its utilization by tissues is only partially understood. In theory, RBCs are transfused to improve tissue oxygen delivery, although it is not always clear that increased delivery translates into increased local availability of oxygen at the tissue level or increased intracellular consumption,7-9 nor for that matter, that all microcirculatory beds respond similarly.10, 11 Basic clinical questions that warrant study include: How do we determine if an RBC transfusion is effective? Which variables should we assess to identify patients who might benefit from transfusion? How do we decide which patients need RBC transfusion? What are the outcomes of patients who receive one, some, or many RBC transfusions? “Red cell transfusion has never undergone prospective randomized testing in the fashion of that of a new drug,”12 but clearly the need persists to examine its clinical effects more closely. Certainly, a growing body of clinical work demonstrates that clinically stable adult and pediatric intensive care patients tolerate lower hemoglobin (Hb) levels than had been thought over the previous decades.3, 13, 14 However, much more remains to be done to determine the extent to which a lower RBC transfusion threshold is safe and efficacious in other patient populations and what information other than the Hb level could be used to make individualized, physiology-driven decisions about RBC transfusion. With such gaps in our understanding of the basics of oxygen delivery and consumption, and how they affect patients, it is little wonder that we are still trying to come to grips with the effect of RBC storage on these processes. The fact that many changes occur in stored RBCs is very well established and has attracted considerable attention. RBC storage produces deformability changes such that the membrane flexibility needed to traverse capillaries is reduced; cells need to synthesize ATP to maintain flexibility but ATP levels decrease with storage. RBC storage is also accompanied by reduced levels of intracellular 2,3-diphosphoglycerate, which increases Hb affinity for oxygen; levels are almost undetectable after 2 weeks of storage, although recovery begins with a few hours of transfusion and is restored within 72 hours. RBC storage also reduces nitric oxide release, a normal response of the RBC under hypoxic conditions; research is ongoing to determine if nitric oxide release mechanisms are restored after transfusion.15-19 In addition to these changes, to the RBC itself, are the variables of RBC processing and storage, such as leukoreduction methods, preservative solutions, and timing of irradiation. It is possible to construct physiologically plausible hypotheses as to how these changes might affect RBC function in the transfused patient. However, it is one thing to hypothesize how storage- or processing-induced changes may affect a patient, but quite another to demonstrate that it translates into a clinically meaningful outcome. For this, well-designed and executed clinical trials must be performed. There is a modest body of data attempting to address the clinical impact of RBC storage. In this issue of TRANSFUSION, Lelubre and coauthors1 update earlier reviews20, 21 with a compilation of clinical studies identified through a systematic literature search and then subjected to a review of the evidence. Twenty-four adult studies were reviewed after grouping them into primary patient population categories of cardiac surgery,8 colorectal surgery,3 and so forth. Twenty-one studies were single center, 12 were retrospective, population sizes varied from 15 to 6002 subjects, and the interval over which the studies were conducted ranged from 1 to almost 8 years and occurred between 1995 and 2008. The authors felt that the studies were too heterogeneous to perform a true meta-analysis and thus compared results from within specific patient diagnostic groups only to each other. This strategy clearly highlights one of the difficulties in conducting clinical trials—how generalizable are the results of a given study to other populations? The analysis by Lelubre and colleagues suggests that experience with one diagnostic group may not necessarily be applicable to another. For example, two of the eight cardiac surgery studies reported an increased risk of mortality, but had statistical limitations. Two retrospective intensive care unit studies suggested longer storage of RBCs was correlated with impaired microcirculatory changes or mortality, but could not be replicated in prospective, double-blinded studies. Supporting this suggestion are the observations that unique end-organ microcirculatory beds react to anemia, hypovolemia, and/or hypoxia differently, and therefore the pathophysiologic response of the microcirculation to RBCs stored for varying periods of time may be different in these states.11 The five trauma surgery studies cited correlated longer RBC storage with infection, multiple organ dysfunction, or mortality. However, subpopulations of trauma patients may be more likely to receive massive RBC transfusion and thus may be unique in having increased mortality solely on this basis. Lelubre and colleagues1 point out some of the difficulties in interpreting these studies which result from both weaknesses in study design as well as heterogeneity among the studies. Among the limitations that they cite for different studies were retrospective design, with its inherent vulnerability to the effects of confounding variables; inconsistent results even among similar studies; small sample size; single-center design; and long study duration with its susceptibility to the effects of temporal changes in practice. In addition, there is considerable heterogeneity in the patient populations, the methods for reporting RBC storage time, the study end points, and the details of the processing and storage of the RBCs. It is quite clear from the review of Lelubre and coauthors1 that the question of the effect of RBC storage on patient outcomes has still not been answered rigorously. We have a hypothesis based on the well-described changes that occur in RBCs stored under conventional blood bank conditions, as well as some data from animal models which demonstrate deleterious effects when stored RBCs are transfused. The hypothesis is that storage of RBCs impairs their ability to function and produces worse patient outcomes. However, the clinical data adduced so far to substantiate or discredit this hypothesis have not produced a clear result. Some studies have shown that there is a difference, but others have not. We therefore remain in a state of clinical equipoise. We simply do not know the answer to the question. We've been here before. Several years ago, a very large number of retrospective, observational studies suggested an association between the transfusion of nonleukoreduced, allogeneic blood components with poorer outcomes. However, when it came to the outcomes of randomized, clinical trials, the purported deleterious effects of allogeneic white blood cells proved to be elusive.22, 23 This recent history should teach us that the results of nonrandomized studies must be interpreted with circumspection and that the smoke of a significant association does not always indicate the existence of a fire of causality. We trample through equipoise to draw premature conclusions at our peril. The need for adequately powered, prospective, randomized controlled trials to address the effect of RBC storage on clinical outcomes is therefore well recognized. To this end, two pilot/feasibility studies have been undertaken, one of which is reported in this issue of TRANSFUSION. In the previously published pilot study by Hébert and colleagues24 for the current ABLE trial, cardiac surgery or intensive care unit patients were randomly assigned to receive transfusions of RBCs that had been either stored for fewer than 8 days or stored and issued according to “standard therapy,” that being RBCs “issued from the hospital blood bank with the longest storage times first in accordance with standard blood bank procedure.” To maximize the difference in storage time between groups, patients were only randomly assigned when the “average age of RBC in the blood bank exceeded 15 days.” Compliance with the age assignment 90% of the time occurred in 73% of the patients overall, with 91% compliance in the less than 8-day arm, but only 59% in the standard therapy group, and the median storage ages were 4 days versus 19 days. The authors concluded that although it is possible to maintain a supply of RBCs stored for less than 8 days while achieving a significant difference in storage time between the two arms, considerable preplanning was required to adhere to the randomization assignments.24 In the feasibility study reported in this issue of TRANSFUSION, Bennett-Guerrero and coauthors2 randomized cardiac surgery patients in the first part of the study to receive RBCs stored for 21 days or less or to their “standard operating procedure being ‘oldest out first.’ ” In the second part of their study, patients were randomly assigned to receive RBCs that had been stored for 7 ± 4 or 21 ± 4 days. All patients received ABO group–specific RBCs; use of out-of-group ABO-compatible RBCs was not permitted, nor were RBCs intentionally held in inventory to enhance availability of units stored for 21 ± 4 days. In the first part of the study, there was considerable overlap in RBC storage times between the two arms. In the second part of the study, transfusions were compliant with the storage time assignment for 90% of the RBCs ordered 100% of the time for the patients in the 7-day arm but only 50% for those in the 21-day arm, due to both failure to dispense units in storage for the appropriate time and lack of available inventory meeting randomization assignment; however, a significant difference in the mean storage times of the RBCs transfused to the two arms was achieved. The authors concluded that if a nonoverlapping RBC storage interval difference between study arms is desired, assigning one group to the standard of care would not be likely to produce enough of a storage time divergence to elicit any disparity in clinical effects. Additionally, using their institution's current procedures for inventory management, none of the product arm assignments resulted in the transfusion of a significant number of RBCs, which approached the currently approved 42-day storage limit. This study points out that one of the challenges in executing such a clinical trial is the management of the RBC inventory so that enough of a difference exists between the storage times to adequately test the hypothesis that storage time affects the outcome measure of the study. As Bennett-Guerrero and coauthors point out, randomizing patients to the extremes of RBCs stored for 3 days or less versus more than 35 days would maximize the likelihood of uncovering any clinical effects of RBC storage. However, this randomization scheme would be extremely difficult to execute with respect to inventory management and perhaps also from the standpoint of the likely reaction of the public and possibly the clinicians. The choice of the appropriate storage times to study was vigorously debated by transfusion medicine physicians, hematologists, and cardiac surgery care teams who designed the Red Cell Storage Duration Study (RECESS), which is being organized by the NHLBI Transfusion Medicine and Hemostasis Clinical Trials Network (TMHCTN). RBCs stored for 10 days or less versus 21 days or more were chosen for the two arms of this study of the effects of RBC storage on outcomes in cardiac surgery patients. A storage time of 10 days or less is readily achievable with respect to inventory management given that most hospital blood banks receive, for at least some portion of their RBCs, units that have been redistributed within their region at varying times in their shelf life. “Standard practice” was considered for the longer storage arm, but the concern was that the difference in storage times between the two arms would be relatively small and thus lessen the likelihood of eliciting a clinical outcome difference. A storage time of 28 days or more was also considered but there were concerns that it might not be uniformly acceptable to human subjects committees after the recent publicity surrounding this storage issue. A survey of several of the TMHCTN sites revealed that between 28% and 50% of RBCs administered to cardiac surgery patients at those institutions had been stored for 28 days or more at the time of transfusion, which is similar to what was found in the pilot by Hébert and colleagues, although it is in contrast to the observations at Bennett-Guerrero's institution. Thus, a storage period of 21 days or more is not widely divergent from current practice but still allows for a clear separation from the arm receiving units stored 10 days or less. Given the current interest in the potential impact of storage of transfused RBCs on patient outcomes and the concern about the implications for blood inventory management, rigorous, prospective, controlled studies are long overdue. Fortunately, in addition to the RECESS trial, which is still being developed, two other trials are already enrolling patients. The Canadian Clinical Trials Groups has begun enrolling patients in a prospective, multicenter, blinded, randomized clinical trial, the ABLE study, the pilot for which was mentioned above. Patients 16 years old or more being admitted to the intensive care unit are being randomly assigned to receive RBCs that have been stored for 2 to 8 days or standard of care (2-42 days) with 30-day mortality as the primary endpoint. The Age of Red Blood Cells in Premature Infants (ARIPI) study, a multicenter, blinded, randomized clinical trial that is also being conducted in Canada, is randomly assigning premature infants weighing 1250 g or less to receive transfusions of RBCs stored for less than 7 days or to the standard of practice, which is to use multiple aliquots from a single RBC unit dedicated for a specific infant until its expiration date.25 The primary outcome measure is a composite of mortality, necrotizing enterocolitis, retinopathy of prematurity, bronchopulmonary dysplasia, and intraventricular hemorrhage, which is being followed out to 90 days after randomization. RECESS, the trial mentioned above, is designed as a multicenter, blinded, randomized study that will randomly assign patients more than 12 years old who are undergoing complex cardiac surgery to receive RBCs stored for 10 or fewer or 21 or more days. The primary endpoint will be the change in the Multiple Organ Dysfunction Score (MODS) by Postoperative Day 7.26 Secondary endpoints include changes in the MODS by Day 28, mortality by Day 28, and several composite and individual organ markers of morbidity. The in vitro measurements of the changes that occur during RBC storage and data from animal experiments have suggested the plausible hypothesis that stored RBCs may have affects that, if not deleterious, then at least are less beneficial, compared to RBCs stored for shorter periods. The results from the largely observational clinical studies have been inconsistent and their interpretation has been fraught with the limitations inherent in their retrospective design, heterogeneous populations, small size, or other factors. The only randomized studies performed so far have been too small to draw any conclusions. The question of whether or not the effects of storage on RBCs contribute to poorer clinical outcomes in patients remains in equipoise. Fortunately, several well-designed clinical trials are being mounted to address the effect of RBC storage on clinical outcomes and to provide the data which will eventually tip the scales of equipoise in the direction of scientific truth. Until we have those studies to evaluate, it would not be prudent to alter transfusion algorithms based on length of storage. The authors have no conflict of interest.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.