The red blood cell storage lesion: a method to the madness

Abstract

For it is not enough to possess a good mind; the most important thing is to apply it correctly. Whether transfusion of “older” blood is as beneficial as transfusion of “fresher” blood is a topic of active debate. Considering the sheer volume of recent publications including some excellent reviews2, 3 and a meta-analysis4 on the issue, one may wonder why there is a need for yet another editorial5, 6 addressing one more observational study on the subject.7 Distinctively, this editorial strives to 1) examine how a well-conducted observational study can intelligently inform a scientific debate, 2) demonstrate the benefit of evaluating an issue from a variety of angles using different study methods, and 3) illustrate how research in transfusion medicine can impact and promote scientific knowledge in other fields. Several studies have reported an association between transfusion of older blood and highly clinically significant outcomes, including an increase in length of hospital stay, postoperative infections, prolonged mechanical ventilation, multiple organ failure, and mortality (see Lelubre et al.3 and Zimrin and Hess2 for in-depth reviews of the literature). Two mechanisms could be conjectured to explain these potential detrimental outcomes if a “cause-and-effect” relationship between duration of red blood cell (RBC) storage and morbidity/mortality is present; note that this is purely speculative since such a causal link has not been demonstrated. One hypothesis proposes that the RBC storage lesion defects that occur in stored blood components cause immunomodulatory and inflammatory complications in transfusion recipients including changes in vasoregulation. The second hypothesis, which could coexist with the first, proposes that susceptibility factors predispose certain patient populations to the clinical-pathologic side effects of older RBC transfusion. This problem if present (meaning if a causal relationship exists) would represent a major public health impediment considering the 5 million US patients who, every year, rely on RBC transfusions for their care. A shortened RBC storage time would have a major impact on the availability of blood. Blood banks would need to adapt recruitment practices to ensure a steady supply of fresher cells and/or develop storage procedures that prevented the development of lesions which contribute to significant adverse clinical outcomes. Faced with this specter, it behooves us to evaluate this problem in a scientifically rigorous and rational manner. Scientific objectivity is crucial because the consequences of misinterpreting the data are significant. As discussed, incorrectly concluding that blood storage duration is causally related to increased morbidity and mortality would severely impact blood availability and put patient care at risk. On the other hand, erroneously concluding that storage duration does not impact morbidity and mortality, if indeed it does, would have appalling consequences. Just under 400 years ago, Descartes put forward four methods in Part II of the Discourse1 which, to this day, can guide our approach to research and scientific discoveries. In his words, “the first [method] was never to accept anything as true that I did not incontrovertibly know to be so; that is to say, carefully to avoid both prejudice and premature conclusions; . . . ”; “the second [method] was to divide all the difficulties under examination into as many parts as possible, and as many as were required to solve them in the best way”; “the third [method] was to conduct my thoughts in a given order, beginning with the simplest and most easily understood objects, and gradually ascending, . . . to the knowledge of the most complex”; and “the last [method] was to undertake such complete enumerations and such general surveys that I would be sure to have left nothing out.” In other terms, we should refrain from jumping to conclusions when there is insufficient evidence (sounds familiar?); evaluate the issue from different perspectives using several study methods to collect a comprehensive set of data points; resolve simplest issues first; be comprehensive and systematic in our approach; check for consistency; double check our results; and use our “good sense,” that is, “the power of judging correctly and of distinguishing the true from the false”1 when conducting data analysis and interpreting results (as a side note, some philosophers argue that truth is never attainable). Armed with these principles and the desire to fund research that will most rapidly advance our understanding of this issue, the National Heart, Lung, and Blood Institute (NHLBI) undertook a review of the studies conducted in this area starting with the clinical literature. We concluded that it is extremely difficult to interpret existing data. Many of the studies are observational in nature and suffer from lack of adjustment, a small sample size, and/or the inability to generalize results (only a few small randomized clinical trials [RCTs] have been conducted so far). In particular, the frequently observed imbalance in baseline characteristics of the comparison groups renders adjustment for confounders (characteristics that are associated with both the outcome and the main independent variable) a critical requirement for interpretation. For example, the “number of transfusions,” which is linked to both severity of disease and increased RBC storage duration, is a major confounder in all RBC storage analyses.8 It is therefore crucial to carefully scrutinize these reports to evaluate how data were analyzed and interpreted and note that adjustment can still be problematic because not all confounders may be known and the assumptions underlying the statistical method that was used may not hold. Some studies suggest that transfusion of older blood is associated with increased morbidity and/or mortality while others suggest a lack of association or even the converse.2-4 A cause-and-effect relationship was not demonstrated and equipoise was present—meaning that there is genuine uncertainty as to whether transfusing fresher blood is more, less, or equally beneficial as transfusion of older blood. These observations begged for the conduct of studies evaluating comparable patient groups (e.g., RCTs) and, possibly, observational studies that were sufficiently powered and adequately analyzed to better inform the debate. NHLBI decided to pursue support of a RCT using the Transfusion Medicine and Hemostasis Clinical Trials (TMH) Network infrastructure (see below) and an observational study which is the subject of the report by Edgren and colleagues7 in this issue of TRANSFUSION. Using the Scandinavian Donations and Transfusions (SCANDAT) database, a powerful tool that merges data from population and migration, death and inpatient care, and blood donations and transfusion registries in Sweden and Denmark, Edgren and colleagues conducted a retrospective cohort study aimed at evaluating mortality among transfused patients receiving RBC units stored in a saline-adenosine-glucose-mannitol solution for 0 to 9 days (25.5% of transfusion episodes), 10 to 19 days (34.7%), 20 to 29 days (15.7%), or 30 to 42 days (8.1%); 15.9% of the transfusion episodes involved RBC units of mixed ages. They included data from more than 350,000 patients, 15 to 90 years old, who were transfused with RBCs between 1995 and 2002. Data on a little more than 400,000 transfusion episodes were analyzed using a variety of sophisticated methods and adjusted, at a minimum, for number of transfusions, age, sex, blood group, calendar period, season, weekday, hospital, and transfusion indication. Several other subanalyses were conducted to evaluate cause-specific risks of death as well as the effect of leukoreduction. This study evaluated a robust outcome (mortality) and included a large number of patient transfusion exposures allowing for adjustment of major known confounders. There was unbiased ascertainment of both exposure and outcome because the data were compiled from national registers. The investigators evaluated the data through a variety of approaches and still reached similar conclusions: that the 1-week risk of death after transfusion did not significantly differ between recipients of RBCs stored for different periods. Conversely, there was a significant increase in the 2-year mortality risk in recipients of 30- to 42-day-stored RBCs (5% increase) and mixed-age (3% increase) RBCs compared to the risk of recipients receiving RBCs stored for 10 to 19 days. The hazard ratios for patients receiving blood stored for 0 to 9 days and 20 to 29 days were 1.01 and 0.99, respectively (neither value was significant). Observing a significant increase in risk of death at 2 years but not at 1 week for recipients of 30- to 42-day-old or mixed-age blood is contrary to the expectation that the effects of an acute harmful event usually decrease over time. This observation, along with the absence of a clear dose-response effect and not finding that any particular cause(s) of death explained the excess risk, led the authors to attribute these findings to residual confounding by severity of disease, for which they were not able to directly adjust. One may also wonder if these two significant findings could just have occurred by chance considering the number of comparisons made. However, adjustment for multiple comparisons can be problematic and the authors would probably have been criticized for using such an adjustment to “get rid of” significant associations. Subset analyses did not reveal significant associations. In particular, looking at patients undergoing coronary artery bypass graft (CABG) surgery, the hazard ratios were similar, 1.09 and 1.10 (both nonsignificant) among patients receiving either blood stored for less than 10 days or blood stored for 30 to 42 days, respectively. The results of this well-conducted observational study are reassuring in the sense that 1) these investigators did not detect associations of the magnitude observed in other studies such as the study by Koch and colleagues9 and 2) the two significant findings can be reasonably explained by residual confounding and/or a multiple comparison effect. However, uncertainty still remains. What if these findings really reflect a cause-and-effect relationship (although it would be difficult to explain biologically why recipients of mixed-age blood would be at higher risk)? What if the subset analyses were insufficiently powered and patients undergoing CABG are at greater risk of complications if they receive fresh (<10 days) or old (>30 days) RBC units rather than blood that has been stored for 10 to 29 days? Clearly, RCTs are needed in a setting of equipoise because the ramifications for patient care and blood banking practices are profound. While RCTs can provide the strongest clinical evidence of whether RBC storage duration does or does not affect morbidity and mortality (the random allocation of patients to treatment arms provides for balance of baseline characteristics between comparison groups), RCTs are logistically difficult, lengthy, and expensive to conduct. Four large RCTs are now planned or under way in three different patient populations, highlighting the importance of this issue. The Age of Blood Evaluation (ABLE) study10 supported by the Canadian Institutes of Health Research (CIHR) is randomizing about 2500 intensive care unit patients at 25 or more Canadian sites to receive, if transfused, either less than 8-day RBCs or standard-issue RBCs (2-42 days); their primary outcome is 90-day all-cause mortality. The Age of Red Blood Cells in Premature Infants (ARIPI),11 also funded by CIHR, plans to randomize 450 premature infants (≤1250 g) to receive either less than 8-day RBCs or 2- to 42-day RBC aliquots (the standard practice is to take aliquots from the same unit of RBCs until its expiration); the primary endpoint is a 90-day composite measure of all-cause mortality and organ dysfunction. The Red Cell Storage Duration and Outcomes in Cardiac Surgery, which is being conducted at one center, the Cleveland Clinic (Ohio), is randomizing cardiac surgery patients who are 18 years and older to receive, if transfused, either less than 14- or more than 20-day RBC transfusions; their target is 2800 patients and their primary outcome is postoperative morbidity.12 Finally, the NHLBI TMH Network Red Cell Storage Duration Study (RECESS) currently plans to randomize approximately 1800 cardiac surgery patients at an estimated 15 centers across the United States to receive either 10-day-or-less or 21-day-or-more RBC units if transfused. Patients randomized to receive RBCs stored for 21 days or more will receive units from the hospital's existing blood bank inventory using the standard inventory management practice of releasing the oldest RBC units first. The primary outcome for RECESS is the change in the composite multiple organ dysfunction score (MODS) from baseline.13 Although results will not be known for several years, these clinical trials will be key to our understanding of whether the RBC storage lesion produces a clinically important phenomenon. Let us now turn our attention to the RBC storage lesion itself.14, 15 What biologic and/or pathologic mechanisms could underlie some of the adverse clinical findings observed in transfused recipients? Can RBC products be optimized to increase oxygen delivery? The latter question should be considered independently of the first. Indeed, since about 1 in 60 individuals receive RBC transfusions in the United States every year, transfusion therapies should not only be safe but, importantly, as effective as possible. While some elements of the RBC storage lesion are well known (e.g., increase in free hemoglobin [Hb] and potassium; decrease in pH, adenosine triphosphate, and 2,3-diphosphoglycerate; and increased RBC morphologic changes leading to loss of deformability),14, 15 less is known about other components such as the shed microparticles (MPs), RBC-dependent vasoregulatory compounds, and immune and inflammatory mediators that may be accumulating during storage. Further, surprisingly little current research has been performed to evaluate the effects of these elements on host cells, the vessel wall, and tissue microoxygenation. To begin rectifying this gap in knowledge, NHLBI recently established a targeted research program titled “Immunomodulatory, Inflammatory, and Vasoregulatory Properties of Transfused Red Blood Cell Units as a Function of Preparation and Storage.” Eight teams of investigators around the country are conducting basic and translational research including animal models and/or basic human physiologic studies aimed at further characterizing the storage lesion and understanding the interaction between the storage lesion components and recipient/host cells such as endothelial and hematopoietic cells. As shown in Table 1, several of these teams (Gladwin and Kim-Shapiro; Patel, Barnum, and Weinberg; Roback; and Stamler) propose that transfusion damage from stored blood can be traced to suboptimal tissue oxygenation engendered by a loss of proper control of blood flow. These investigators will evaluate the role of nitric oxide (NO), nitrite, ATP, and/or S-nitrosohemoglobin (SNO-Hb) in this context. For example, consideration will be given to whether loss of vasoregulation could be secondary to increased levels of Hb and MPs in the plasma which scavenge NO and inhibit beneficial NO signaling and/or due to a loss of RBC-mediated hypoxic vasodilation (see Table 1 for each group-specific aim). Blumberg and Phipps will investigate the RBC units' supernatant including Hb, MPs, and fats and its effect (as well as that of RBCs) on patients' platelets (PLTs). Jy's group will explore the role of MPs in transfused products by studying their potential procoagulant, proinflammatory, immunosuppressive, and endothelial activities. Norris' team will focus on the potential inflammatory and immune effects of stored RBCs by characterizing immune and inflammatory mediators in fresher and older RBC units, evaluating RBC-endothelial cell adhesion as a function of RBC storage, determining if novel storage solutions or pathogen reduction can ameliorate the “storage lesion,” and determining how RBC unit storage affects immunomodulation and inflammation in transfusion recipients enrolled in ABLE. Finally, the team led by Spitalnik will evaluate whether transfusion of older stored RBCs can produce a proinflammatory cytokine response due to delivery of substantial amounts of Hb iron to the monocyte-macrophage (reticuloendothelial) system and its consequent pro-oxidant effects (see Table 1 for additional descriptions of each of these eight projects). This research is expected to significantly increase our understanding of the RBC storage lesion and its potential clinical effects, as well as optimize methods and further development of assays to detect biomarkers of clinical relevance. New storage solutions could be developed. For example, identification of components in RBC units that may be potentially harmful to a specific patient population could lead to either development of improved RBC therapies that do not contain the harmful component or personalization of RBC therapy with transfusion of biomarker-negative units only in the patient population(s) at risk. Further studies may identify elements that are missing or in suboptimal concentrations because of preparation, manipulations, and storage time; RBC units may benefit from addition of such elements during component manufacturing or storage or before infusion. This multifaceted research could also have important ramifications for other fields. There are many examples of how research in transfusion medicine has positively impacted other scientific domains. For example, research conducted on infectious (e.g., human immunodeficiency virus, hepatitis B and C, West Nile virus) and immunologic complications of transfusions (e.g., transfusion-related acute lung injury) has not only helped advance the field of transfusion medicine by improving the safety of the blood supply, but has also been hypothesis generating. Thus, this work has fostered additional research to understand the pathogenesis of a wide variety of infectious and noninfectious diseases, including underlying host genetic and immunologic mechanisms involved in responding to and controlling infections and allogeneic immunologic challenges. The latter studies are critical to advancing the development of vaccines, for example. Similarly, we can anticipate that research on NO bioavailability, blood flow, tissue microoxygenation, interactions with the endothelium, PLT biology, immune and inflammatory modulators, and iron metabolism will be highly informative to the fields of vascular biology, hematology, immunology, and cellular therapies. It is hoped that this systematic and rational approach will lead to a better understanding of the RBC storage lesion and its clinical effects as well as to the development of enhanced transfusion therapies and strategies. These discoveries will help us achieve our mission, which is to improve and optimize patient care. The author declares no conflict of interest.