|Home | About | Journals | Submit | Contact Us | Français|
The advent of preservative solutions permitted refrigerated storage of red blood cells. However, the convenience of having red blood cell inventories was accompanied by a disadvantage. Red cells undergo numerous physical and metabolic changes during cold storage, the “storage lesion(s)”. Whereas controlled clinical trials have not confirmed the clinical importance of such changes, ethical and operational issues have prevented careful study of the oldest stored red blood cells. Suggestions of toxicity from meta-analyses motivated us to develop pre-clinical canine models to compare the freshest vs the oldest red blood cells. Our model of canine pneumonia with red blood cell transfusion indicated that the oldest red blood cells increased mortality, that the severity of pneumonia is important, but that the dose of transfused red blood cells is not. Washing the oldest red blood cells reduces mortality by removing senescent cells and remnants, whereas washing fresher cells increases mortality by damaging the red blood cell membrane. An opposite effect was found in a model of haemorrhagic shock with reperfusion injury. Physiological studies indicate that release of iron from old cells is a primary mechanism of toxicity during infection, whereas scavenging of cell-free haemoglobin may be beneficial during reperfusion injury. Intravenous iron appears to have toxicity equivalent to old red blood cells in the pneumonia model, suggesting that intravenous iron and old red blood cells should be administered with caution to infected patients.
The earliest modern transfusions of human blood were either performed at the bedside immediately after removal of blood from the donor or even simultaneously via arterio-venous anastomoses1. The advent of solutions containing citrate anticoagulant and dextrose permitted blood collected for transfusion to be separated from the blood donor in both space and time. Refrigerated storage of whole blood or red cells (RBCs) permitted the establishment of blood depots to support the troops in World War I and eventually the modern blood bank2,3. However, the convenience of having inventories of stored refrigerated blood was accompanied by a drawback. Red cells clearly changed during cold storage as documented by a number of in vitro measurements of size, shape, enzyme content, rheology, and filterability4. Furthermore, modern tools can now detect thousands of changes in red cell metabolomics that occur within a few weeks of refrigerated storage5. The critical question, however, concerns the clinical importance of these changes.
For much of the 20th century, a major challenge in RBC transfusion involved efforts to extend component shelf life without sacrificing quality and efficacy. Early investigators settled on RBC recovery and survival in vivo at the end of storage as the key measure of RBC quality. The US Food and Drug Administration, which licenses RBCs, containers and storage solutions, relies primarily on two surrogate measures of efficacy and safety: 24-hour recovery and survival of >75% of radiochromium-labelled RBCs, and haemolysis of <1% at the end of storage. No clinical studies have ever been required. These were clearly minimal standards and although a variety of other analytical measurements were routinely provided (ATP, 2,3, DPG, lactate, whole blood oxygen binding curves, potassium concentration), surrogate measures had not been correlated with patient outcomes. Nevertheless, these surrogate measures seem to have served us well.
My research interest in this area was stimulated in 2003 when I was asked to write an editorial about a retrospective study that had concluded that stored RBCs posed a risk of toxicity to patients6. It occurred to me that whereas small compromises in erythrocyte quality and quantity, commonly referred to as the “storage lesion”, are tolerated as the price of increased blood availability, clinicians do not expect this price to include harm to their patients. A few years later, an influential and controversial paper published in the New England Journal of Medicine provided a retrospective analysis of several thousand cardiac surgery procedures and concluded that transfusion of older red cells was accompanied by increased mortality7. This was not the first such study, but it was the largest and most widely publicised. In response, a collaborative group from the Department of Transfusion Medicine and the Critical Care Medicine Department in the intramural program of the National Institutes of Health undertook a series of studies to determine whether “old blood” is associated with more clinical adverse events than “fresher” units. We determined that studying the oldest and freshest RBCs transfused to patients in specific clinical states would pose logistical and ethical problems, so we determined to adapt a well-validated canine model of severe pneumonia with a mortality end point to the transfusion setting. We believed such a model might detect a risk of transfusing stored RBCs if such a risk existed, and if so, could explore the mechanism(s), dose responses, critical length of storage, and potential interventions.
Before developing the pre-clinical study model, we undertook a literature review and meta-analysis of all studies published in English that reported the age of the RBCs transfused and included mortality as an outcome. Twenty-one such studies met the pre-determined criteria and our analysis revealed a highly statistically significant difference in mortality [odds ratio 1.16, 95% confidence interval (CI) 1.07–1.24, p=0.0001] favoring transfusion of fresher cells8 (Figure 1). Despite the obvious limitations of such meta-analyses, a series of sensitivity analyses of these data [adult intensive care unit (ICU) studies, paediatric studies, surgery studies, non-surgery studies, studies in different decades] all yielded consistent findings, further supporting the hypothesis of the storage lesion.
The pre-clinical model we developed examined the extremes of fresh and old RBC storage as well as massive amounts of RBC transfusion in critically ill animals. We reasoned that if such a model showed no differences in mortality, then it would seem fruitless to pursue further studies of RBC storage. We infused a known concentration of bacteria into the lungs of purpose-bred beagles, and between hours 4–16 after bacterial challenge, performed four 20 mL/kg exchange transfusions with commercially prepared, leucoreduced RBCs, either 5 days old (fresh) or 42 days old for a 70% RBC exchange. Animals were supported with antibiotics, mechanical ventilation and titrated pressor therapy similar to management of patients in an ICU. At 96 hours, any surviving animals were sacrificed for necropsy. Older blood transfusion increased mortality (p=0.0005), the arterial-alveolar oxygen gradient (p<0.01), and histological lung damage (p=0.03). Older blood resulted in increased in vivo haemolysis, releasing free iron in the form of non-transferrin bound iron (NTBI) and cell-free haemoglobin (CFH) (p<0.03) and decreasing plasma haptoglobin levels9. Consistent with the vasoconstrictive effect of CFH, older blood increased both systemic and pulmonary pressures (all p<0.02). Mortality was associated with extensive pulmonary necrosis; other organ toxicity was not observed. This was the first randomised blinded animal trial showing blood transfused at end of storage period can increase mortality during infection (Figure 2).
Based on our studies in canines, we have proposed two main mechanisms as underlying the reported adverse effects of older stored blood. Briefly, we found that older RBCs are more fragile and prone to in vivo haemolysis when transfused, resulting in increased release of plasma haemoglobin and iron. This is also the case with human red cells. Both cell-free haemoglobin (CFH) and free iron have recognised toxicities and can potentially worsen outcomes in transfusion settings10,11. CFH is well known to scavenge nitric oxide (NO), an endogenous potent vasodilator, resulting in vasoconstriction. The vasoconstrictive effects of CFH can induce ischaemia and vascular endothelial injury. Haptoglobin, the plasma protein known to bind CFH, may become saturated during haemolysis and unable to promote its clearance by the reticuloendothelial system (RES). In addition, haemolysis releases free and protein-bound iron. Iron is a critical nutrient for bacterial growth. Iron metabolism is ordinarily carefully controlled to prevent direct oxidative toxicity and to limit access by pathological microorganisms. Human physiology has evolved multiple mechanisms to remove free iron available for micro-organisms from the circulation. In conditions of iron overload (haemochromatosis, transfusional haemosiderosis) the binding capacities of such proteins as transferrin and ferritin are saturated resulting in increases in plasma non-transferrin bound iron (NTBI). However, bacteria have evolved sophisticated strategies to scavenge iron directly from binding proteins with the help of high-affinity siderophores.
Next, we investigated the effect of increasing bacterial doses and severity of infection on the risks associated with age of blood transfused in this canine model12. Forty-eight animals were challenged intrabronchially with either 0 (n=8), 1.0×109 (n=8), 1.25×109 (n=24), or 1.5×109 (n=8) S. aureus colony-forming units/kg and then exchange-transfused with either 7- or 42-day-old canine universal donor blood (80 mL/kg in four divided doses). Without bacterial challenge, levels of CFH and NTBI were significantly higher with older vs fresher RBC transfusion but there were no significant differences in measurable injury. With higher-dose bacterial challenge, the elevated NTBI levels declined more rapidly and to a greater extent after transfusion with older blood, and older blood was associated with significantly worse shock, lung injury, and mortality. The CFH levels were markedly elevated over days regardless of severity of infection. The augmented in vivo haemolysis of transfused older RBCs, resulting in excess plasma CFH and iron release, appears to require the presence of established infection to worsen outcomes. These data suggest that transfused older RBC increase the risks from infection in septic subjects and define an infection dose-response.
During canine bacterial pneumonia with septic shock, but not in controls, older stored RBC were associated with significantly increased lung injury and mortality. We wondered whether transfusion of older RBCs would cause similar adverse effects during shock and inflammatory injury without infection. Therefore, 2-year-old purpose-bred beagles (n=12) were transfused similar quantities of either older (42-day) or fresher (7-day) stored universal donor canine RBC 2.5 hours after undergoing controlled haemorrhage producing shock13. With older transfused RBCs, CFH (p<0.0001) and NTBI (p=0.004) levels increased, but lung injury (p=0.01) declined and there was a trend toward lower mortality (18% vs 50%). The increased levels of CFH with older RBC transfusion were associated with an improved haemodynamic response to haemorrhage-reperfusion, with lowered exogenous norepinephrine requirements (p<0.05) and cardiac outputs (p<0.05). This haemodynamic effect is consistent with the ability of CFH to scavenge NO causing vasoconstriction. Thus, in haemorrhagic shock, older RBCs altered resuscitation physiology but did not worsen clinical outcomes. Elevated CFH lowers norepinephrine requirements and cardiac output, ameliorating reperfusion injury. In our infection model, we had previously shown that older blood increases NTBI levels transiently during transfusion and the rapid clearance of iron is associated with increased lung injury and mortality. In contrast, during haemorrhagic shock, NTBI levels persist longer after transfusion, and increased levels of iron are not associated with worsened outcomes. These pre-clinical data suggest that whereas iron derived from older RBCs promotes bacterial growth, worsening septic shock mortality during infection, release of CFH and NTBI during haemorrhagic shock is not necessarily harmful.
We next conducted a blinded randomised controlled study (RCT) of RBC washing in this canine infection model of transfusion injury14. We hypothesised that washing older units of blood before transfusion would improve clinical outcomes by removing older fragile RBC and prevent increases of CFH and iron, whereas washing fresher units would have no effect on outcome. Twenty-four 2-year-old purpose-bred beagles (n=24) with S. aureus pneumonia were exchange-transfused with either 7- or 42-day-old washed (commercially available Haemonetics blood cell processor with standard washing procedure, Haemonetics Corp., Braintree, MA, USA) or unwashed canine universal donor blood (80 mL/kg in 4 divided doses). Washing RBCs before transfusion had a significantly different effect on canine survival, multiple organ injury, plasma iron, and CFH levels depending on the age of stored blood (all p<0.05 for interactions). Washing older units of blood improved survival rates, shock score, lung injury, cardiac performance and liver function, and reduced levels of NTBI, possibly by lysing and washing away old cells and supernatant. In contrast, washing fresh blood worsened these clinical parameters and increased CFH levels. Our data suggest that fresh blood should not be washed routinely because washing induces sublethal membrane damage to the RBC, and in a setting of established infection, washed RBC are prone to lyse, release CFH and iron, and result in worsened clinical outcomes. These findings, along with our previous studies, indicate that transfusion of fresh blood in infected subjects results in less haemolysis, CFH and iron release, and is less toxic than transfusion of older blood in critically ill infected subjects. However, if older blood must be used during established infection, washing prevents elevations in plasma circulating iron and improves survival and lessens multiple organ injury.
We next examined the results of altering volume, washing, and age of RBCs. Two-year-old purpose-bred infected beagles were transfused with increasing volumes (5–10, 20–40, or 60–80 mL/kg) of either 42- or 7-day-old RBCs (n=36) or 80 mL/kg of either unwashed or washed RBCs with increasing storage age (14, 21, 28, or 35 days) (n=40)15. All volumes transfused (5–80 mL/kg) of 42-day-old RBCs resulted in like increases in iron, CFH, lung injury, and mortality rates after transfusion. Transfusion of 80 mL/kg of RBCs stored for 14, 21, 28 and 35 days resulted in increased CFH and NTBI in between levels found at 7 and 42 days of storage. However, washing RBCs of intermediate ages (14–35 days) does not alter NTBI and CFH levels or mortality rates. Thus, our pre-clinical data suggest that any volume of 42-day-old blood potentially increases risks during established infection. In contrast, even massive volumes of 7-day-old blood result in minimal CFH and NTBI levels and risks. In contrast to the extremes of storage, washing blood stored for intermediate ages does not alter risks of transfusion or NTBI and CFH clearance.
Our most recent as yet unpublished studies examined the effect of intravenous iron infusion on mortality in the canine pneumonia model and compared it with the effects of RBC transfusion. Intravenous iron has been aggressively marketed as a safer alternative to RBC transfusion and is used widely in intensive care unit (ICUs)16,17. We reasoned that if iron released from old, stored RBC enhanced bacterial growth, a similar effect might result from administration of intravenous iron. We compared two forms of commercially available intravenous iron, iron sucrose, a “2nd-generation” preparation, and ferumoxytol, a novel iron oxide nanoparticle with a polyglucose sorbitol carboxymethylester coating designed to minimise immunological sensitivity, with fresh RBC transfusion. In the canine model of mild anaemia and pneumonia, both iron preparations resulted in similar mortality and pulmonary toxicity, both statistically increased compared to transfusion of 7-day-old RBC. These findings suggest that both old stored RBC and intravenous iron should be used with caution in the setting of established infection.
Because the number of clinical studies investigating the effects of age of blood on mortality has recently surged, and several large, RCTs have been completed, we up-dated our meta-analysis18. Among 31 observational studies, we found an overall significantly increased risk of death with older blood, confirming our original findings. In our up-dated meta-analysis, 6 RCTs now account for more than 4,000 patients, sufficient to perform a separate analysis from the observational studies. A significantly different overall effect on outcome was found in RCTs (p=0.02), with no differences in survival between older and fresher blood (OR: 0.91, 95% CI: 0.77–1.07). We attribute the different results found in observational studies and RCTs to methodological differences. While observational studies divide current transfusion practice into two groups of older and fresher blood transfusion, and “old” is up to 42 days, RCTs compare current practice (with an average storage age of 2–3 weeks) to transfusion of the freshest available RBCs (1–10 days old). Consequently, the median age for both the fresher and older blood arms was significantly higher in observational studies compared to RCTs (p=0.01). Our analysis of RCTs confirms that the freshest RBCs are not superior to the 2–3-week-old stored RBCs transfused during current practice. However, the RCTs cannot exclude the possibility, which observational studies suggest, that older stored units available for transfusion, namely 4–6 week old RBCs, increase mortality risk. Since publication of our analysis, a pragmatic, controlled trial conducted at six hospitals in four countries randomised 31,497 patients to receive either fresher or older RBC; the findings are consistent with our conclusions19.
Randomised controlled trials remain the “gold standard” for determining the risk of stored red cells, but studies in man have both ethical and practical limitations. One cannot ethically “store RBCs” in order to give the oldest cells to patients in a toxicity study. In addition, all clinical conditions cannot be studied. Pre-clinical studies can provide important clues to risk, pathophysiology, and therapy. Our studies suggest that the oldest RBC should not be administered to patients with established infection. We have, therefore, limited the age of stored RBC at NIH to 35 days. Our studies further suggest a potential benefit to washing the oldest RBC, but an increased risk to washing fresh cells. Finally, in a canine model, IV iron increases the mortality risk in the setting of established infection. We believe that until appropriate studies are conducted in patients, IV iron should be used with caution in infected patients.
These studies were performed in collaboration with Drs. Charles Natanson, Steven Solomon, Irene Cortés-Puch, Dong Wang, Kenneth Remy, and Jungfen Sun. The studies were supported by intramural research funding, the Intramural Research Program, and the National Institutes of Health.
This summary was presented in part at the 2016 Presidential Symposium of the International Society of Blood Transfusion, Dubai, United Arab Emirates. The opinions expressed are those of the Author and do not represent the positions of the NIH or the US Department of Health and Human Services.
The Author declares no conflicts of interest.