Does the storage duration of blood products affect outcomes in critically ill patients?

When oxygen delivery fails to support aerobic metabolism, a complex state of physiologic dissonance emerges, broadly designated as shock. As the oxygen debt in shock mounts, cascading organ failure will lead to death unless physiologic control is restored. Two principal blood functions are impaired in this setting: oxygen transport by red blood cells (RBCs) and the balance between coagulation and fibrinolysis.^1-3 This deteriorating state is compounded by the additional dysregulation of immune and endothelial function in critically ill patients who have minimal physiologic reserve to compensate for these changes. Paradoxically, mounting evidence suggests that blood component therapies (to correct anemia or coagulopathy) may exacerbate, rather than resolve, this condition—possibly due to unintended consequences of blood product processing and storage.^4-9 This may be particularly true for those with shock and coagulopathy since exacerbation of either condition will adversely affect the other.^2,3 For critically ill patients with marginally compensated organ failure or those with shock and coagulopathy, it is essential that resuscitation is composed with such issues in mind. These high-risk patients may benefit from blood products of decreased storage duration that have increased function immediately upon transfusion as well as an improved safety profile.^10-16

Conversely, non–critically ill and elective surgical patients who are not in a state of dysregulation may not be as tenuous due to increased physiological reserve. Therefore, the adverse impact of blood products with sub-optimal function is associated with less risk in this population. Moreover, certain RBC functionalities may be restored in vivo within 24 to 48 hours after transfusion,¹⁷ and thus benefit may eventually accrue to hemodynamically stable patients with minimal transfusion volume requirements.

The two-hit hypothesis has been well described for many clinical conditions and may explain the different relationship between resuscitative agents and outcomes in these two patient populations.⁷ Based on this concept, the risk-benefit ratio for blood product administration is dependent on the baseline physiologic state and differs between critically ill and non–critically ill patients. Specifically, patients with marginally compensated organ failure are less tolerant of additional insult, including adverse effects attributable to transfusion. As such, evaluation of blood product efficacy and safety is further informed by distinctly separating analysis of outcome in critically ill and non–critically ill populations.

The impact of storage duration on blood product efficacy and safety is currently under intense scrutiny.¹⁸ Rigorous evaluation of outcomes attributable to specific therapies in the intensive care setting presents challenges; however, it is essential we carefully evaluate blood products where efficacy and safety most influence outcome. Two general areas of current research include the efficacy and safety of RBCs to treat patients in shock and the efficacy and safety of fresh whole blood (FWB) for the patient with shock and coagulopathy.

For the majority of the 20th century, transfusion was administered as whole blood.¹⁹ With development of blood component therapy, extended RBC unit storage duration, low blood donation rates, and unavoidable market forces, near exclusive use of blood components replaced transfusion with whole blood by the late 1980s. Since the majority of patients requiring blood products are hemodynamically stable with single-component deficiencies, this transition from whole blood to components was logical and prudent. Alternatively, for the minority of transfusion recipients who are critically ill in a state of life-threatening coagulopathy or shock, concern has been raised that the transition from briefly stored whole blood to components stored for longer durations may have adversely influenced outcomes.^4-9,20,21

Notably, the transition from whole blood to components occurred without substantial investigation to determine if efficacy differed in high-risk critically ill patient populations, or if the extension of RBC storage duration was safe for this patient group. In fact, the Food and Drug Administration (FDA) has “grandfathered” RBC and whole blood storage solutions with a definition of efficacy or safety for licensing that (in light of new data on the RBC “storage lesion”) may now be insufficient, in that criteria for improved oxygen delivery to tissue (the purpose of transfusion) are not required. As such, licensing by the FDA for RBC and whole blood storage solutions only requires that hemolysis be less than 1% at the end of storage and that RBC recovery rates are greater than 75% 24 hours after transfusion.²² Recent literature suggests that limited hemolysis during storage and adequate posttransfusion survival does not necessarily ensure transfusion will augment oxygen delivery for a patient in shock. In addition, determination of quality by these parameters does not appear to obviate more recently characterized adverse effects of storage (e.g., immunomodulation, inflammation, hypercoagulation, impaired vasoregulation, and perfusion).^1,4-9,23-29 Such preclinical data suggest that current licensing criteria may be inadequate. In summary, the current body of literature evaluating blood product storage duration appears insufficient to either support or change current regulations or allocation policies.

Similarly for platelet (PLT) storage solutions, FDA criteria do not require functional measures of PLTs nor address potential adverse effects of storage. A recent review of the PLT storage lesion highlights the lack of efficacy and safety data for stored PLTs and indicates the difficulties of generalizing in vitro data with in vivo function.³⁰ Two pediatric studies have indicated decreased PLT storage duration was associated with less clinical bleeding with improved PLT function measured in one of these reports.^10,11

How can efficacy and safety of blood products be measured in critically ill patients? It would be optimal to determine if a stored RBC can deliver oxygen and improve cellular respiration in prospective randomized controlled trials (RCTs). Direct measurement of microvascular perfusion or oxygen consumption with calorimetry is difficult to perform in the intensive care unit (ICU) setting. However, two current trials will compare surrogate measures of oxygen delivery in critically ill patients.^31,32 Noninvasive tissue oxygenation measures using vascular occlusion methods to assess oxygen consumption and microvascular reactivity are promising but require validation.^33,34 Of note, transfusion of human RBCs and PLTs in rodent models has been used to measure storage lesion effects;^35,36 such studies indicate that RBC storage duration is inversely related to perfusion and oxygen delivery.²⁹ Given the limitations of such data, the efficacy of transfused human RBCs to improve oxygen utilization should be directly evaluated in humans. To date, clinical trials examining this issue have not directly measured O₂ delivery to tissue and analysis of outcomes is challenged by multiple confounding design elements. Notably, there are at least five active RCTs examining the efficacy of RBCs according to storage duration.^18,31,37-39 These trials will provide a high level of evidence regarding the clinical efficacy of RBCs attributable to storage age.

While the risk of transfusion-transmitted diseases has dramatically decreased over the past few decades,⁴⁰ noninfectious safety risks associated with RBC storage duration remain a concern and are the focus of significant investigation and funding.¹⁸ The biologic plausibility of adverse clinical outcomes in critically ill patients secondary to the transfusion of RBCs of increased storage age has been established.⁶ In vitro and animal laboratory data indicate that the risk of inflammation and oxidative injury, hypercoagulation, microparticle production, altered vasoregulation, and perfusion correlates with RBC storage duration.^{5,8,23-26,28,29,41} Comprehensive ancillary studies of the Age of Blood Evaluation (ABLE) and Red Cell Storage Study (RECESS) trials are underway and will examine each of these variables.⁶ Definitive determination of clinical relevance for results from these ancillary studies will likely require very large thoughtfully designed trials. The same degree of investigation of storage duration is needed for PLT efficacy and safety in patients with severe life-threatening bleeding.

In this issue of TRANSFUSION, the topics of blood product efficacy and safety are explored. van de Watering and the BEST collaborative⁴² review common pitfalls of retrospective research on the influence of RBC storage duration on outcomes. Ho and Leonard⁴³ examine whether FWB use is associated with improved outcomes.

The review by van de Watering of common pitfalls associated with study of RBC storage duration in transfusion recipients outlines several categories of bias in the hope that readers, policy makers, physicians, blood bankers, reviewers, and editors will avoid them in the future.⁴² This excellent and thorough review both highlights pitfalls and, uniquely, suggests measures to minimize the bias inherent to retrospective research. As such, van de Watering calls for future hypothesis-generating research to 1) avoid comparison of nontransfused to transfused cohorts or employ propensity analysis to minimize selection bias arising from this approach; 2) adjust for the volume of RBCs transfused and other potential confounders such as ABO type and year of transfusion; 3) perform multiple analyses of RBC age “bands,” analyze RBC age as a continuous variable, and stratify RBC storage duration across groups large enough to allow for meaningful comparisons; 4) match cohorts by RBC volume transfused; 5) optimize sample size to avoid Type II error, but also be mindful of large studies reporting significant outcomes lacking clinical relevance.

This otherwise thorough review might also have addressed the influence of severity of illness on (what van de Watering describes as) “conflicting results.” Specifically, most efforts failing to identify an influence of storage duration on outcome either 1) study non–critically ill populations (patients with a low event rates who may be resilient to the physiologic impact of the “storage lesion”) or 2) lack power to determine if a difference exists. Of note, severity of illness is difficult to define across pathologic states, although it is most simply indicated by the incidence of morbidity and mortality in a population. An incidence of more than 5% to 10% of adverse outcomes in a population is a reasonable threshold to define significant critical illness.

Recently published meta-analyses and reviews indicate that there are many studies that do not report an association between RBC storage age and adverse clinical outcomes.^17,22,44 Examples of these “negative” studies include those performed in healthy volunteers,⁴⁵ colorectal cancer patients,⁴⁵ or cohorts with low (<5%-10%) risk of morbidity or mortality that were underpowered for outcomes measured (sample size ranging from 17 to 902).²² Prospective studies, all of which are “negative,” examine populations of 17 to 66 patients.²² The largest study completed with mortality as the primary outcome was a retrospective study in a postoperative cardiac surgery population.⁴⁶ This project evaluated 2715 patients in a mixed RBC age analysis and 1895 patients in an exclusive RBC age analysis, with an overall mortality incidence of 3.5%. With standard alpha (0.05) and beta (0.2) values and testing for a clinical meaningful reduction in mortality of 20% (reducing 3.5% mortality to 2.8%), a study would require 19,000 patients to be adequately powered. Even a study powered to measure an absolute mortality reduction from 3.5% to 2.0% (42% reduction) would require 3700 patients. It should be noted that some reviews of RBC age and outcomes include studies of FWB.⁴⁴ The efficacy of FWB is influenced by many factors other than RBC age and should not be grouped with single component therapy in analysis of storage duration.

Since the publication of these reviews,^17,22,44 a few reports have avoided many of the pitfalls mentioned by van de Watering.⁴² Each examines only critically ill patients or those with significant morbidity and mortality (incidence >10%). In a study of 600 trauma patients transfused 3 or more RBC units, Weinberg¹² reported an independent association between RBC storage duration and increased mortality. This analysis adjusted for the amount of RBC units transfused among other potential confounders, and patient groups compared received either only fresher RBCs (all <14 days) or only older RBCs (all ≥ 14 days). Eikelboom and colleagues¹⁵ studied more than 2000 transfusion recipients with cardiopulmonary disease and reported an independent association between RBC storage duration and mortality, after evaluating both dichotomous and continuous RBC age values and adjusting for transfusion timing and volume. Ranucci and coworkers¹⁶ performed a retrospective study over a 2-year period of 192 children who received transfusions during cardiopulmonary bypass. After adjustment for potential confounding variables, maximum RBC storage duration analyzed both as a dichotomous variable (≤ or >4 days) and as a continuous variable was independently associated with major morbidities (defined as cardiac, pulmonary, hepatic, and renal failure). Pettila and colleagues¹⁴ reported a prospective multicenter observational study of 757 critically ill adults from 47 Australian and New Zealand ICUs; in an adjusted analysis, for patients exposed to one or more RBCs stored for more than 11 days, the odds ratio for hospital mortality was 2.01 (95% confidence interval, 1.07-3.77). Importantly, in each of these studies, the incidence of the adverse outcome in the older RBC group was between 10 and 27%.^12,14-16 The only pitfall identified by van de Watering remaining in these studies was failure to adjust for ABO type. Since ABO type has not been demonstrated to be independently associated with mortality it is difficult to interpret the importance of this omission.

As noted, there is renewed interest in the value of FWB for patients with severe or life-threatening coagulopathy and shock, although the literature on this subject is also of varying quality and difficult to interpret. The US military defines FWB as blood that is stored at 20 to 24°C for less than 24 hours (warm FWB).⁴⁷ Civilian institutions have defined FWB as blood that has been stored at 2 to 6°C for less than 48 hours (cold FWB)¹⁰ or for less than 5 days by others.⁴⁸ While there are theoretical advantages of FWB for patients with (or, at high risk of) coagulopathy and shock, few prospective RCTs compare FWB to component therapy. Potential advantages of FWB are: 1) increased concentration of hemoglobin, coagulation factors, and PLTs relative to reconstitution of individual components;⁴⁹ 2) limited impact of processing on function (with regard to both O₂ transport and coagulation);^28,29,50-52 3) avoidance of risk associated with extended storage duration;^4-9,53,54 and 4) reduced donor exposure.¹¹ It is currently unknown if these potential advantages of FWB would improve outcomes for patients with life-threatening hemorrhagic shock.

Certainly, there are also risks associated with FWB use in the critically ill. FWB is not typically leukoreduced, due to the associated removal of PLTs. As a result, transfusion of viable white blood cells (WBCs) increases the risk of febrile reactions and cytomegalovirus transmission.⁵⁵ It also increases the rare fatal risk of transfusion-associated graft-versus-host disease and of transfusion-associated microchimerism in trauma patients that may or may not have clinical consequences.^56,57 When FWB is used in combat or other austere settings, standard transfusion-transmitted disease testing is not feasible, with resulting risk of transfusion-transmitted disease transmission.²¹ Increased risk of acute respiratory distress syndrome has also been reported with FWB use in combat settings, but these results were unadjusted and may be affected by survivorship bias.^54,58 Methods that may mitigate the potential risk of FWB include the use of whole blood leukoreduction filters that spare PLTs⁵⁹ or ultraviolet light techniques that inactivate both pathogens and WBCs.⁶⁰ Improved safety of whole blood is very important to the developing world where it is used often and the risks of infectious complications are high.⁶¹ Optimal methods to improve safety according to location still need to be determined.

There are no RCTs of FWB in adults, although two recently published retrospective studies from the US military report different results in combat casualties.^54,58 While both reports indicate improved adjusted 24-hour survival, adjusted survival at 30 days was conflicting, which may be due to the inability to adjust for all confounding variables as well as differences between studies in inclusion criteria, logistic regression methods, severity of illness of patients compared, and subject attrition.

There have been two prospective RCTs of FWB in pediatric cardiac surgery patients, also with different methods and conflicting findings. Manno and colleagues¹⁰ performed a prospective double-blind study in which 161 children were randomly assigned to warm FWB (<6 hr at 20°C), cold FWB (stored for 24-48 hr at 4-6°C), or reconstituted blood (RBCs ≤ 5 days of storage, fresh-frozen plasma [FFP], and PLTs). Their data indicated that both warm and cold FWB resulted in less postoperative blood loss and improved PLT function by aggregometry, with findings most pronounced in children less than 2 years of age. Multivariate linear regression analysis affirmed these findings. This RCT was limited by randomization that was dependent on FWB availability. Mou and coworkers⁶² reported a RCT comparing cold FWB to reconstituted blood (RBCs and FFP) in the pump prime for 200 pediatric cardiac surgery patients less than 1 year of age. Notably, this study (unlike Manno) did not provide FWB postoperatively. The primary outcome of this study was a composite score for survival and ICU length of stay (LOS), with no difference found between groups. There was an increase in secondary outcomes of ICU LOS and total fluid requirement for patients transfused cold FWB. These outcomes were not adjusted for potential confounders.

The analysis by Ho and Leonard⁴³ in this issue of TRANSFUSION adds to the literature comparing FWB to components, although there are challenges to interpreting these data. Analyzing both elective surgery and traumatic injury patients simultaneously does not allow for comparing severity of injury or preoperative risk for bleeding. The inability to determine if groups studied were of similar severity of illness makes it difficult to determine if the results are valid; for example, the APACHE score is not validated in trauma patients. Moreover, it is critical to be certain that exposure to FWB occurred before the observation measures that are used to compare outcomes and adjustment for variables such as RBC volume should be included in either the propensity analysis or the regression analysis. It is not possible to conclude that there are no differences between groups if insufficient numbers of patients were included in the analysis. Most importantly, a study comparing FWB to components should have a significant amount of FWB transfused; it appears in the report by Ho and Leonard that approximately 10% to 15% of all blood transfused to the FWB group was actually FWB. Of note, the study that indicated an association with survival had roughly 30% of the blood products transfused as FWB,⁵⁴ whereas the negative study had approximately 21% of the blood volume transfused as FWB.⁵⁸ Regardless, each of these studies highlight the need for prospective RCTs to address the efficacy of FWB in select populations.

Research is under way to determine optimal storage temperature and duration for FWB as they relate to efficacy and safety. A recent in vitro study by Jobes and colleagues⁶³ suggests that FWB can be stored at 4 to 6°C for up to 2 to 3 weeks without diminished hemostatic activity. Hughes and coworkers⁶⁴ reported that whole blood maintains its hemostatic properties for at least 3 days when stored at room temperature (20°C). These results must be validated in vivo, employing bleeding models. The Hemostasis and Oxygenation Research (THOR) network led by Hervig and Strandenes have developed a series of experiments comparing storage lesion effects to include immune, coagulation, endothelial, and microparticle variables between FWB and reconstituted blood components stored for increasing durations. An adult RCT has started in Houston, Texas, which compares components to cold stored FWB and PLTs in 400 total trauma patients.⁴⁸ The primary outcome measure in this trial will be a reduction in transfusion requirements.

In the 1970s and 1980s, the storage duration of RBCs increased and there was a shift from the use of whole blood to components; recent, deeper understanding of storage effects has renewed interest in comprehensive study to determine if these changes were efficacious and safe, especially in critically ill patients at risk of shock and/or coagulopathy. Thirty to 40 years later it is encouraging that there is significant research aimed at determining the efficacy and safety of blood products transfused to critically ill patients. In such work, it is imperative we take into account the degree of illness since the risks are higher and there is less physiologic reserve in the critically ill. Moreover, quality improvement and outcome research would be enhanced if blood bank clinical databases provided data on storage age of every blood component. While these data exist on every unit transfused, current AABB reporting requirements do not require this information to be recorded in standard blood bank databases.

Interest in studying FWB in patients with hemorrhagic shock is renewed and prospective trials will provide needed data. Current initiatives evaluating efficacy and safety of stored RBCs are also of major importance. Based on biologic plausibility, laboratory and animal data indicating 1) increased O₂ delivery efficacy by younger RBCs and 2) increased adverse effects with older RBCs in the critically ill, it is imperative to definitively test the hypothesis that the transfusion of RBCs of decreased storage duration will improve outcomes in this vulnerable group. Results of current prospective RCTs are eagerly awaited and, hopefully, will help us guide transfusion policies in the future.

These are challenging times for transfusion medicine experts and the physicians who care for the critically ill. The blood banking community is charged with providing a scarce resource that is in high demand, supporting physicians whose approach to transfusion varies widely and is infrequently evidence based. Moreover, blood banks are expected to provide a beneficial product to a wide variety of potential recipients with storage criteria that many now believe may be insufficient. In the absence of regulatory requirement for direct efficacy and safety data evaluating oxygen delivery, tissue respiration, and PLT function we lack incentive for industry to invest the resources necessary to acquire this information. In an era of quality improvement, evidence-based medicine, and significant cost constraints, determining the optimal transfusion approach for critically ill patients with shock and/or coagulopathy is difficult, but worthy of pursuit.

CONFLICT OF INTEREST

The authors report the following conflicts of interest: PCS: CaridianBCT, Citra; NB: Pall Biomedical, Fenwal, and CaridianBCT; AD and JBH: none.