Widening our gaze of red blood storage haze: a role for metabolomics

There can be little doubt that developing the technical capability to store red blood cells (RBCs) and to create blood banks represented a major medical advance of the early 20th century.1 Indeed, in the United States alone, approximately 1 of every 70 individuals receives a RBC transfusion annually. Like many clinical situations, the indications are obvious at the extremes of the curves; transfusion for anemic patients with symptomatic hypoxia seems a clear indication, whereas a normal hematocrit is typically a contraindication for RBC transfusion. In contrast, consensus on the exact clinical indications for RBC transfusion in the context of less extreme anemia has been difficult to achieve, with a variety of transfusion triggers proposed over the years. There is a tremendous volume of clinical information indicating that transfusion of stored RBCs saves lives. Trauma patients who have had their entire blood volume replaced with stored RBCs can survive and make a full recovery. Likewise, RBC exchanges of stored RBCs for patients with sickle cell disease (SCD) can lead to drastic improvements. However, the fact that stored RBCs have efficacy does not tell us what fraction of optimal efficacy they have, nor how much efficacy might be increased with improved storage. Moreover, the observation that many patients who receive large volumes of transfused RBCs do well clinically does not address the possibility that there may be coincident sequelae of transfusion. There have been a large number of observational studies that have associated more liberal RBC transfusion strategies with worse medical outcomes (e.g., infections and recurrence of neoplasia, posited to occur due to immunomodulatory effects of transfusion).2 Such concerns have promoted the study of how more restrictive transfusion practices affect medical outcomes. The results of such studies reported that restrictive transfusion practices were not associated with worse outcomes in patients in the intensive care unit, but were on the contrary associated with a benefit (albeit not of significance).3 Subsequent studies have been performed in other settings with similar results.4 Recently, no difference was reported in clinical outcomes between liberal versus restrictive RBC transfusion strategies, over longer-term periods of time (3 years,) in geriatric hip fracture surgery patients.5 Additional data also exist, and a meta-analysis of 18 randomized trials with more than 7500 patients reported a decreased risk of adverse outcomes with restrictive transfusion practices.6 A separate, but related, issue is the hypothesis that transfusion of RBCs that have been stored for longer periods of time are associated with worse medical outcomes. Although there are myriad retrospective studies that report worse outcomes with older RBCs compared to fresher RBCs, there are likewise many reports showing no difference at all and several showing a benefit to transfusing older units. In this context, several prospective randomized clinical trials have been conducted, focusing study on premature infants, critically ill adults, and complex cardiac surgery patients. The Age of Red Blood Cells in Premature Infants (ARIPI trial) reported no difference in medical outcomes between groups receiving fresher (5.1 ± 2.0 days) versus older RBCs (14.6 ± 8.3 days).7 Likewise, both the RECESS8 and ABLE9 trials have recently reported no difference in groups receiving fresher versus older blood (presented in abstract form and at meetings); the field awaits formal publication of these studies. It remains unclear to what extent one can infer a general conclusion that older RBCs result in the same medical outcomes as fresher RBCs outside of the context of the patient populations studied. There are also concerns that small effects cannot be ruled out, which may nevertheless result in a significant difference in patient outcomes, due to the very large number of patients transfused each year. Nevertheless, the outcomes of these recent randomized controlled trials (RCTs) provide substantial evidence, and give reassuring data, to support the notion that current standards of storage do not result in a worse outcome of a magnitude that could be detected by the trials that have been performed. While members of the field may have very different views on how efficacious stored RBCs are or the extent to which stored RBCs may cause significant sequelae, it is likely that all can agree that we should always focus on refining storage to achieve a safer and more effective product. However, the best strategy to accomplish such a goal is not clear. While RCTs may ultimately be the gold standard to determine the best RBC storage conditions, their cost and complexity make them unavailable for the discovery phase of innovative and novel storage approaches; indeed, RCTs have yet to be used even to compare efficacy of most of the different existing storage systems. As a more achievable metric, although still of considerable cost and complexity, current food and drug administration (FDA) licensing criteria necessitate a 24-hour posttransfusion recovery of 75% of stored RBCs (with a standard deviation of less than 9% and with a one-sided 95% lower confidence limit for the population proportion of successes of greater than 70%). The reasoning behind this requirement is that RBCs, which do not circulate, cannot function biologically. Born of similar reasoning, FDA criteria limit average in-the-bag hemolysis to 1%. Although the above metrics do not measure efficacy or toxicity of the product, they do exclude conditions where RBCs are sufficiently “nonviable” to exclude sufficient product efficacy. While 24-hour posttransfusion recoveries are an FDA-mandated licensing requirement and are thus carried out on all approved storage systems, such studies are nevertheless expensive, time-consuming, and involve the ethical conduct of research on healthy human volunteers. Thus, innovative development of RBC storage does not typically use in vivo recoveries in the discovery phase. Rather, as is the case for most biologic therapeutics, innovative development is guided by in vitro assays and laboratory-based determinations that serve as surrogates to guide development. Despite lacking nuclei, mitochondria, or ribosomes, mature RBCs nevertheless maintain a complex regulatory biology, with ongoing glycolysis, metabolism of lipids and antioxidant pathways, and regulation of ion gradients. RBCs also face a particular challenge in handling the autooxidation of hemoglobin and the resultant cellular damage that ensues. The storage of RBCs adds a level of complexity, due to nonphysiologic conditions (e.g., increased glucose, decreased calcium, accumulation of lactate). In addition, damaged RBCs persist in the bag during storage, due to the absence of a reticuloendothelial system that would normally remove them in vivo. It is thus of little surprise that a large number of changes take place in stored RBCs (known in aggregate as “the storage lesion”). Such changes include alterations in cellular metabolism, changes in glycosylation of proteins and lipids, alterations in rheologic properties, morphologic changes, alterations in membrane lipid composition and symmetry, generation of microparticles, oxidative damage to proteins and lipids, and additional changes (recently reviewed).10, 11 A number of analytic technologies have been developed to measure the changes that occur during the storage lesion (e.g., biochemical tests [ATP, 2,3-DPG, etc.], pH, microscopy, rheology, flow cytometry to measure changes in surface molecules, and others). These measures reflect clear and reproducible changes that take place during RBC storage. However, each of these measures is essentially serving only as a surrogate for the actual biology of interest, in particular 24-hour posttransfusion recoveries. Thus, if the laboratory metrics that guide development do not correlate with 24-hour recoveries, it becomes difficult to see the rational basis for predicting progress using such measures as guidance. The fact that 24-hour recoveries may not themselves correlate to an efficacious RBC product brings even an added layer of tenuousness. Regrettably, it is the observed case that none of the in vitro metrics has a strong correlation to high 24-hour recoveries. Indeed, extreme values (e.g., very low pH, extreme morphologic changes) can predict poor 24-hour recoveries; however, good laboratory values in no way guarantee acceptable 24-hour recovery. This lack of correlation may be the result of previous progress and innovation. Because such measures were utilized in the previous advances in RBC storage, the current conditions may have already taken advantage of the biologic differences elucidated by such measures. In any case, the fact remains, and it is a serious problem to the field of engineering improved RBC storage systems, that the in vivo laboratory metrics used to evaluate novel storage solutions have only weak (or in some cases no) correlation to the in vivo measure of 24-hour recoveries, which itself has unknown correlation to efficacy and/or sequelae of transfusion in any given clinical setting. Thus, the field of RBC storage biology is very much in need of new measures of RBC storage quality that predict how the RBC units will perform in vivo. Some progress may have been made in the elegant engineering of ex vivo human systems. For example, a system has been described in which RBCs are passed through whole living human spleens (ex vivo) to look at clearance;12 however, such systems are difficult to set up, would be challenging to standardize, and depend on a resource that may be complicated to control regarding quality or quantity (e.g., human spleens surgically removed for benefit of the patient). It is in the light of the above challenge that attempts have been made to harness the “omics” revolution. Whereas traditional systems of analysis required measuring one (or at best several) analytes at a time, and required substantial sample volume to do so, the linkage of chromatography to tandem mass spectrometers has provided the capability to simultaneously elucidate thousands of analytes on small specimens. Depending on the technical particulars, one can analyze peptide fragments of proteins, species of membrane lipids, or the metabolic chemicals of the cells. Sophisticated informatics analysis is then required to translate raw data into a processed format that has quantitative meaning and that can be juxtaposed between samples and conditions. The general application of these technologies to the RBC storage lesion has resulted in a previously unimaginable panoply of new data for a wide array of new chemicals and has captured the attention of blood storage development.13 In many ways, it is unclear if the application of omics technology to the RBC storage lesion will make things better or worse. It is currently unknown which, if any, of the thousands of peptides, lipids, and metabolites that have now been reported to change during RBC storage correlate with 24-hour recoveries or RBC efficacy. Thus, this approach has thus far widened our gaze, but done little to clarify the haze, since the reported omics data have not been paired with in vivo recovery studies on the units being analyzed. An added problem can be found in the application of traditional statistical measures of significance. If one is comparing 2000 analytes, and using a p value of 0.05 as significant, then 100 analytes would be predicted to be increased, just by chance alone. This latter problem can be addressed, in part, by grouping analytes together into common biologic pathways, allowing a measure of aggregate probability of all of the members of a pathway moving in a common way, as opposed to isolated components. Metabolomics (the analysis of metabolites in stored RBCs) is perhaps most amenable to this type of analysis, due to the fact that many analytes are either precursors or downstream products of each other. Thus, by analyzing multiple members of a metabolic pathway, one can gain degrees of confidence regarding changes between storage conditions. In this issue, D'Alesandro and colleagues14 apply a sophisticated metabolomics analysis to the question of changes in RBCs stored in AS-3. The particular approach used not only provides the ability to generate a relative quantitation between specimens, but elegantly places the individual analytes into the relevant biologic pathways. Although this analysis is not the first application of metabolomics to RBC storage, a comprehensive pathway analysis is put forward, in addition to consideration of individual analytes of interest, so as to allow analysis from global to granular. The fashion of information processing employed by D'Alesandro and colleagues is precisely the type of data analysis and presentation that make metabolomics approaches elucidating and powerful, potentially leading to meaningful observations. As the field continues to develop and embrace omics technology, increasing sophistication of informatics will be required to maximize meaningful observation in the context of what can be overwhelming volumes of data. The ongoing generation of high-quality omics data, with appropriate and proper analysis, seems almost certain to teach us a great deal about the biology of RBC storage and, in all likelihood, the biology of RBCs in general. Indeed, the compilation of such detailed information is the first necessary step to the generation of future understanding. However, it also seems that unless and until there are careful studies performed to evaluate the extent to which analytes or pathways correlate to the primary biology of interest (e.g., 24-hour recoveries), the lack of a suitable surrogate of the primary biology will persist, and thus barriers will remain to rapid and efficient innovation and refinement of novel storage technologies. Moreover, should such correlations be observed (and even causality ultimately established), then the new understanding would provide mechanistic insight to guide rational modification of RBC storage systems. Thus, from a basic research standpoint, the expansive new body of data can only be a good thing. However, the pragmatic challenge to the field of RBC storage remains one of sorting out the meaningful metrics from the ever widening array of variables known to change during RBC storage which differ between storage systems. James C. Zimring, MD, PhD e-mail: [email protected] Puget Sound Blood Center Research Institute and Department of Laboratory Medicine and Department of Internal Medicine Division of Hematology University of Washington Seattle, WA The author has a sponsored research agreement with Immucor, Inc., unrelated to the current work.