Old blood bad? Either the biggest issue in transfusion medicine or a nonevent

Blood transfusion in 2012 is probably safer than at any time in the past. Recipient screening for red blood cell (RBC) alloantibodies, use of male donor plasma to avoid exposure to (pregnancy-acquired) white blood cell alloantibodies, and leukoreduction have dramatically reduced the risk of immunologic and other acute reactions to blood, and exhaustive screening of donor blood has lowered the risks of human immunodeficiency virus, viral hepatitis, and other transmissible infections to almost undetectable levels. Attention is now shifting to more fundamental questions concerning the efficacy and safety of blood transfusion: Is blood transfusion of benefit? What threshold should trigger transfusion? Is new blood better than old blood? In this issue of TRANSFUSION, Wang and colleagues1 from the National Institutes of Health report the results of a meta-analysis of 21 studies involving 409,966 patients demonstrating that transfusion of “older” versus “newer” blood is associated with a significantly increased risk of death (odds ratio [OR], 1.16; 95% confidence interval [CI], 1.07-1.24). Despite including a heterogeneous mixture of study designs (prospective observational, retrospective observational, randomized controlled trials [RCTs]) that involved different patient types (adult and pediatric; cardiac, critical care, and trauma), various cutoffs to define “old” (mean, 9 to ≥28 days) and “new” (mean, 1.6 to ≤21 days) blood, RBC storage solutions (AS-3, CPD-2, CPDA-1, SAGM), durations of follow-up (in-hospital, 30 days, and “long term”), and analytic methods, a consistent pattern of increased mortality with older blood was evident. Further, the authors found no evidence for publication bias. Based on these findings, they concluded that “use of older stored blood is associated with a significantly increased risk of death.” How should we interpret the authors' provocative conclusion? In theory, the results of any study can potentially be explained by random error, systematic error, or truth. The estimates of increased mortality provided by Wang and colleagues are associated with narrow CIs and the p value is extreme (<0.0001), making it very unlikely that the results are due to chance. Furthermore, it is sobering that the published risk ratios from 17 of the 21 studies suggest harm and none of the four remaining studies indicates a significant benefit of older blood. If the findings of excess harm were due to chance alone, one might have expected a protective effect in at least some of the included studies. Systematic error cannot be excluded as a possible explanation for the findings of this study. The analyses are based primarily on the results of observational studies (six prospective, 12 retrospective) that account for most of the data. The three small RCTs included in the meta-analysis contributed only 126 patients or less than 0.05% of the total patient numbers. A key potential source of confounding is that older and sicker patients are more likely to be exposed to older blood because they receive more blood. Furthermore, other as yet unknown confounders might also have accounted for their findings. The medical literature is replete with associations reported in observational studies that are not confirmed when subjected to randomized evaluation (e.g., antioxidants2 and homocysteine-lowering vitamins3 for cardiovascular prevention). Even if confounding could not be demonstrated by Wang and colleagues, the massive potential implications of harm of older compared with newer blood mandate rigorous evaluation in randomized studies that minimize the potential for systematic error. Is there a biologic basis for harm of older compared with newer blood? Pooled data from the meta-analysis by Wang and coworkers suggest that older blood is also associated with an excess of “multiorgan dysfunction syndrome” (OR, 2.26 [95% CI, 1.56-3.25]; n = 3 studies) and “pneumonia” (OR, 1.17 [95% CI, 1.08-1.27]; n = 3 studies) as well as “renal dysfunction” (significantly increased in two of three studies) and “sepsis” (significantly increased in one of two studies). Both organ dysfunction and infection may explain increased mortality with older blood. Another as yet unproven mechanism of adverse outcomes after blood transfusion is tissue hypoxia resulting from progressive RBC depletion of adenosine triphosphate, 2,3-diphosphoglycerate, and nitric oxide during storage that reduces the capacity of transfused blood to deliver oxygen to tissues.4 If, in truth, older compared with newer blood is harmful, and if the estimate of harm provided by this meta-analysis (a 16% relative increase in death) is correct, this result would translate into an excess of approximately one death for every 100 transfused patients that received older versus newer blood.1 This predicted increase in death represents a massive burden of harm that would account for thousands of potentially preventable deaths in the United States alone, each year, and tens of thousands worldwide. With this possible scenario in mind, what impact should the findings by Wang and colleagues have on current clinical practice and future research? First, we agree with the authors that no immediate changes in transfusion practice are required. Even if older blood were proven to be less beneficial than newer blood, transfusion of older blood is not necessarily contraindicated. On the contrary, transfusion of older blood is very likely to be life-saving in patients with major bleeding and in those undergoing life-saving surgery. In the absence of a placebo or no-transfusion control, it is not possible to determine benefits and risks of transfusing any blood, whether old or new. Second, if older blood were proven to increase mortality compared with newer blood, this observation would raise a challenging problem in attribution of causation. Unlike previously described harmful effects of blood transfusion (e.g., ABO incompatibility, delayed hemolytic transfusion reaction, viral transmission, transfusion-related acute lung injury), in which a causal relationship between a particular transfused unit of blood and harm (including death) can be demonstrated (i.e., unit “x” harmed patient “y” through mechanism “z”), in the case of older versus newer blood, one could never reliably attribute causality between exposure to older blood and the death of an individual patient. This problem with causality is because the magnitude of the adverse effect is relatively “modest” (a 16% relative increase in risk related to exposure to older versus newer blood per this meta-analysis), and because the older blood evinces harm in “nonspecific” ways, for example, greater risk of life-threatening organ failure through poorly defined mechanisms. Even if somewhat newer blood could be provided through operational changes to the transfusion service, the demonstration of any resulting benefit would require large-scale tracking studies that relate patient mortality to age of transfused blood (and even then it would be impossible to know whether other concomitant changes in transfusion practice were actually responsible). We urgently require well-designed and adequately powered RCTs to evaluate the effects of older compared with newer blood. Two pilot RCTs5,6 did not suggest harm of older compared with newer blood transfusion, but several large RCTs have been initiated—in cardiac surgery patients,7,8 critically ill patients,9 and neonates10—and a 24,000-patient “pragmatic” RCT is under way in unselected hospitalized patients receiving a blood transfusion.6,11 The demonstration of an association (if indeed there is one) between age of transfused blood and outcome would be useless if we could not do anything about it. Thus, at the same time as performing these studies, we must seek to clarify the nature of the association (is it continuous or is there a threshold effect?) and the possible mechanisms. Only armed with this additional information can we begin a data-driven approach to better manage blood storage and inventories and thereby potentially improve outcomes for patients who require a blood transfusion. None.

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.