Red Blood Cell Transfusion in the Neurological ICU

Abstract

Red blood cell transfusion (RBCT) is a common therapy used in the intensive care unit to treat anemia. However, due to deleterious side effects and questionable efficacy, the clinical benefit of RBCT in patients who are not actively bleeding is unclear. The results of randomized controlled trials suggest there is no benefit to a liberal transfusion practice in general critical care populations. Whether the results of these trials are applicable to brain injured patients is unknown, as patients with primary neurological injury were excluded. This article reviews the efficacy and complications of RBCT, as well as the relationship between RBCT and its outcome in both the general intensive care unit and neurologically critically ill populations.

Electronic supplementary material

The online version of this article (doi:10.1007/s13311-011-0094-5) contains supplementary material, which is available to authorized users.

Efficacy of Red Blood Cell Transfusions

Red blood cell transfusions (RBCTs) are intended to improve tissue oxygenation; however, whether they do so remains unclear. The notion that RBCT augments tissue perfusion presumes that transfused blood efficiently stores and offloads oxygen, and that compromised tissues utilize the additional oxygen. However, studies have shown that although transfusion increases oxygen carrying capacity by augmenting hemoglobin concentration, it often fails to increase oxygen utilization [1–5]. The inability of transfused blood to increase tissue perfusion has been attributed to a series of biochemical and biomechanical changes that occur during red blood cell (RBC) storage collectively termed the “storage lesion.”

The storage lesion results in decreased oxygen delivery to tissues. After 7 days, stored blood is depleted of 2,3-diphosphoglycerate, a compound that is responsible for enhancing oxygen release from hemoglobin to tissues [6, 7]. A loss of 2,3-diphosphoglycerate in stored blood shifts the oxygen dissociation curve to the left and reduces the amount of oxygen available for tissue consumption. Furthermore, time-dependent changes in stored blood lead to acidemia and hyperkalemia, which result in RBC lysis and release of free hemoglobin [8]. Free hemoglobin is a nitric oxide scavenger and therefore may result in vasoconstriction and exacerbation of organ dysfunction [9].

Structural changes, induced by RBC storage, have been shown to compromise microvascular circulation [10]. RBCs undergo a predictable change from biconcave disks to sphero-echinocytes, resulting in loss of deformability. The loss of the biconcave structure of the 8-μm erythrocyte impairs its ability to navigate capillaries with smaller diameters (e.g., 3–8 μm) and may result in vessel occlusion. Sphero-echinocytes result from microvesiculation and loss of surface-to-volume ratio. The formation of microvesicles is associated with increased osmotic fragility and decreased RBC survival [11, 12]. Depletion of adenosine tri-phosphate may also contribute to corpuscular changes [13]. Storage duration of greater than 42 days may cause vasoconstriction due to lysophosphatidyl choline species released from the cellular membrane of senescent RBCs [14]. The storage lesion also promotes increased RBC aggregability and RBC endothelial cell adhesion, which may compromise or obstruct microvascular circulation [15, 16]. Part of this effect may be mediated by microparticles, anucleoid membrane vesicles, which increase in concentration with storage duration [17, 18]. Microparticles have been implicated in post-transfusion thrombosis likely due to the expression of phosphatidyl serine, which promotes plasma-mediated thrombin generation [17–19].

In summary, RBCT may not accomplish the intended effect of improved tissue oxygenation. The lack of efficacy is mediated in part by the storage lesion.

Complications of Transfusions

Blood transfusion therapy is associated with adverse side effects, including transmission of infections and immune activation or immunosuppression. Both viral illness and prion diseases may be transmitted by blood transfusion. Risks of transmission of human immunodeficiency virus, hepatitis C virus, and hepatitis B virus are 1 in 1,900,000; 1 in 600,000; and 1 in 220,000, respectively [20]. Cytomegalovirus is present in 4% of transfusions from healthy donors due to the reactivation of latent cytomegalovirus in leukocytes [21, 22].

Transfusion modulates the immune system in 2 opposite ways: 1) it may heighten the immune response (“alloimunization”), as in transfusion reactions, or it may quell the immune response (“tolerance induction”), which predisposes to nosocomial infections. Human leukocyte antigens (HLA), specifically HLA-DR antigens, on donor leukocytes partly determine which response ensues following RBCT; shared HLA-DR antigens between donor and recipient induce tolerance, whereas antigenic mismatch results in immunization [23].

Alloimmunization and consequent induction of HLA antibodies and T-cell activation results in a number of clinical syndromes, including transfusion reactions, transfusion associated graft-versus-host disease, transfusion-related acute lung injury, and potentially the development of various autoimmune diseases [23]. Transfusion-related acute lung injury is the number 1 cause of transfusion-related mortality [24, 25]. Transfusion-related acute lung injury is defined as a new episode of respiratory distress, which is not explained by an alternate etiology that occurs during or within 6 h of a completed transfusion [26]. The most common clinical features include: bilateral pulmonary edema, hypoxemia, fever, dyspnea, and hypotension in the presence of normal cardiac function [27]. Plasma-rich blood components (fresh frozen plasma and platelets) and high-volume transfusion may predispose to transfusion-related acute lung injury, although it has been associated with all blood product components including intravenous immunoglobulin G and cryoprecipitate [28–30].

Tolerance induction after RBCT is associated with a decrease in natural killer cell function, defective antigen presentation, and a reduction in helper/suppressor T-lymphocyte ratio, and has been linked to increased predisposition to nosocomial and postoperative infections and even to cancer recurrence [31–39]. Transfusion has been associated with increased risk of infection in cardiac surgery patients [40–42], trauma patients [32–35], and critically ill patients [43, 44]. Of 45 studies included in a meta-analysis of transfusion in the intensive care unit (ICU), 22 studies examined the effect of transfusion on infection and all demonstrated an independent association [45]. The pooled odds ratio for developing an infectious complication was 1.8 (95% CI, 1.5-2.2). Transfusion was also associated with an increased risk of multi-organ dysfunction and acute respiratory distress syndrome.

Clinical side effects due to the storage lesion result not only from erythrocyte changes, but also from leukocyte changes. Transfusion-transmitted infections are thought to be due to contaminated leukocytes [46]. Donor leukocytes secrete cytokines, which have been associated with both febrile nonhemolytic transfusion reactions and hemolytic transfusion reactions, in a time-dependent manner after storage [47]. Accumulation of lipid mediators, capable of priming recipient neutrophils and exacerbating multi-organ dysfunction, may result from leukocyte activity on RBC membranes during storage [48, 49]. Leukocyte degradation during storage results in release of oxygen free radicals and proteases, which may also incite inflammation in the transfusion recipient [50].

Leukoreduction of stored blood might mitigate the immunomodulatory effects of transfused RBCs, but its clinical impact and cost-effectiveness have not been fully studied. Although leukoreduction has been shown to reduce the rate of febrile nonhemolytic transfusion reactions [51], it has not consistently prevented such reactions [52]. A Canadian study, in which 7000 patients treated after institution of universal leukoreduction were compared to 7000 historical controls, suggested that universal leukoreduction was associated with lower mortality, but not a significant decrease in infection rate [53]. Smaller randomized controlled clinical trials of LR have failed to demonstrate any beneficial effect of leukoreduction on clinical outcome, including in-hospital mortality, ICU length of stay (LOS), and antibiotic usage [54–56]. Although the utility and cost-effectiveness of universal leukoreduction remains controversial, the majority of blood banking centers in the United States have followed the lead of Canada by adopting this practice, given the putative benefit and minimal risk.

Transfusion Practice in General ICUs

Transfusion is commonly used to improve cellular oxygen metabolism in sepsis and other shock states; however, physiological studies of its effects on tissue perfusion have not consistently demonstrated benefit [57–59]. Although RBCT, as part of a bundled approach to early sepsis management, did confer a survival advantage in a landmark study of patients with septic shock [60], subsequent studies have not confirmed this finding [61]. It should be noted that RBCT remains a recommendation of the 2008 Surviving Sepsis Campaign guideline as part of an algorithm for the treatment of sepsis.

Transfusion administration has been shown to exacerbate sepsis-related microcirculatory dysfunction and impair tissue perfusion. In a study of patients with sepsis, splanchnic perfusion, as measured by gastric tonometry, did not improve after transfusion [62]. In another study, transfusion of 3 units of RBCs did not measurably increase systemic oxygen uptake; however, it did result in splanchnic ischemia in those who had received blood stored for >15 days [63].

Prospective, multicenter observational studies suggest an association between RBCT and increased mortality and morbidity in general ICUs [64, 65]. The Anemia, Blood Transfusion in Critically Ill Patients (ABC) trial demonstrated that transfused patients had more severe organ failure, increased LOS, and higher mortality rates than nontransfused patients [64]. Higher mortality rates were observed regardless of admission hemoglobin level, thereby accounting for any effect of pre-morbid anemia. A dose–response relationship was identified between number of units transfused and mortality; the mortality rate for patients receiving 1 unit of RBCs was 15.9%, whereas those receiving >4 units it was 44.8%. Overall, RBCT increased the odds of death by a factor of 1.4. Propensity score analysis, performed on a subset of patients, demonstrated a higher mortality rate for those transfused, and this coupled with the fact that mortality rates were higher for transfused patients at nearly all levels of organ dysfunction, suggested that transfusion-related mortality was not simply a surrogate for increased severity of illness. A meta-analysis of transfusion therapy in the ICU concluded that transfusion was an independent predictor of mortality in critically-ill adult, trauma and surgical patients [45].

Concern about the safety and efficacy of red blood cell transfusion, including immunosuppressive and microcirculatory complications, has led to a reappraisal of transfusion practices. The Transfusion Requirements in Critical Care (TRICC) trial was designed to determine whether restrictive and liberal transfusion strategies in the ICU produces equivalent all-cause mortality at 30 days and equivalent organ dysfunction [66]. In this randomized controlled study, a hemoglobin threshold of either 7 g/dL or 10 g/dL was targeted as a trigger for transfusion to maintain hemoglobin concentrations in the goal ranges of 7 to 9 g/dL and 10 to 12 g/dL, respectively. The primary endpoint of the 30-day mortality was not significantly different between the 2 groups; however, in patients with an Acute Physiology and Chronic Health Evaluation II (APACHE II) score of <20, or ˂55 years of age, a restrictive approach was favorable (p = 0.03 and p = 0.02, respectively). Thirty-day mortality rates between treatment groups were not significantly different in the subgroups of patients with cardiac disease, severe infections or septic shock, or trauma. Although no significant difference in mortality rates was observed for the subset of patients with cardiovascular disease, there were significantly more cardiac complications in the liberal transfusion arm than in the restrictive arm (21.0% vs 13.2%; p < 0.01). The authors recommended that critically ill patients receive RBCT when the hemoglobin concentration falls below 7.0 g/dl, and it is also recommend that hemoglobin levels be maintained between 7.0 and 9.0 g/dl.

A similar study was performed in a larger cohort of patients admitted for hip surgery with a history of risk factors for cardiovascular disease [67]. Approximately 2000 patients were randomized to transfusion, either when clinically symptomatic or at an arbitrary threshold of 10 g/dL. The primary outcome was death or inability to walk unassisted for a pre-specified distance at 60 days. There was no significant difference in outcome between the high threshold and symptomatic groups. The authors concluded that a symptomatic transfusion practice conserved blood and did not negatively impact outcomes in elderly patients with underlying cardiovascular disease or cardiovascular risk factors.

In the general critical care population, RBCT has been associated with increased mortality and morbidity. Although the TRICC trial provides the best evidence for transfusion practice in the management of the critically ill, it should be recognized that patients with primary neurological injury were excluded.

Impact of Transfusion in the Neuro-ICU

Ischemic Stroke

Although trials studying RBCT in ischemic stroke are lacking, many trials exist studying the relationship between hematocrit and ischemic stroke. Although it seems intuitive that augmentation of oxygen carrying capacity (CaO₂) and oxygen delivery (DO₂) might ameliorate symptoms of an ischemic disease, concerns about viscosity-related reductions in cerebral blood flow (CBF) have limited consideration of RBCT in stroke. Conversely, hemodilution has been studied extensively as a putative treatment option for ischemic stroke because of the inverse relationship between hematocrit, the main determinant of viscosity, and CBF, although it has not been shown to improve outcome [68].

It has been postulated that increased viscosity exacerbates stasis, compromising microcirculatory blood flow to the ischemic penumbra [69]. In support of this theory, Allport et al. [70] demonstrated that higher baseline hematocrit (>50%) was associated with expansion of infarction and less reperfusion [70]. This effect may be more pronounced in women with hematocrit >50% [71]. Moreover, elevated hematocrit levels have been associated with carotid atherosclerosis [72, 73], atrial fibrillation [74], unilateral cerebral infarction [75], greater infarct size [75], early mortality [71, 74], and major disability after stroke [76].

Other studies have noted a “U-” or “reverse J”-shaped relationship between hematocrit and increased risk of stroke [74, 77, 78], suggesting that not only high hematocrit concentrations (>50%), but also low hematocrit concentrations (<30%) are also associated with an increased risk of ischemic stroke [79]. Reduced admission hemoglobin concentrations have been associated with larger infarct volume and infarct expansion on magnetic resonance imaging, and hemoglobin and hematocrit nadir have been associated with poor outcome [80]. Failure of compensatory autoregulation and oxygen extraction due to active ischemia was proposed as the mechanism of reduced oxygen use and infarction. Low baseline and intraoperative hemoglobin and hematocrit have also been associated with postoperative strokes in cardiac surgery patients [81, 82], although this finding has been inconsistent [83, 84]. A study of more than 10,000 cardiac surgery patients found that each percent decrease in hematocrit from baseline resulted in a 10% increase in stroke risk [81]. Independent of hemoglobin concentration, intraoperative transfusion has been associated with increased postoperative stroke in patients undergoing cardiopulmonary bypass [82]. Further studies are needed to determine whether RBCT or other interventions to augment oxygen delivery might reduce cerebral ischemia in both acute ischemic stroke and cardiac surgery patients.

Intracerebral Hemorrhage

A dose-dependent relationship between anemia and intracerebral hemorrhage (ICH) volume has been demonstrated in the literature [85], and a single study exists regarding the effect of RBCT on outcome after ICH [86]. Sheth et al. [86] found that RBC transfusion was associated with improved survival at 30 days (odds ratio, 2.76; 95% CI 1.45-5.26; p = 0.002) and decreased mortality at 30 days (odds ratio, 0.40; 95% CI 0.19-0.69; p = 0.02). However, despite transfusion, there was no significant increase in hemoglobin concentration among patients who were transfused; therefore, the mechanism by which transfusion appeared to be protective remained elusive. The authors tried to reconcile their findings with the fact that restrictive strategies appear to be beneficial in general ICU patients. They postulated that predictors of mortality in general ICU populations may not be analogous to those in patients with ICH, and thus the results may lead to different conclusions. They surmised that the benefit of transfusion might have been derived from augmented intravascular and/or cerebral blood volume given the fact that hemoglobin did not increase with time. Whether increased oxygen carrying capacity is required in the post-ICH period during times of cerebral edema, hydrocephalus, intracranial hypertension, or seizures remains unknown. Secondary injury after ICH, characterized by iron-mediated neurotoxicity, macrophage activation, and matrix metalloproteinase upregulation, may further increase the demand for oxygen.

Anemia may be associated with increased hematoma volume, the primary predictor of mortality in patients with ICH. Whether transfusion can limit hematoma volume and improve outcome remains to be determined.

Subarachnoid Hemorrhage

The impact of transfusion on subarachnoid hemorrhage (SAH) outcome is largely unknown, and the relevant data are mainly derived from retrospective clinical studies and from physiologic studies. There exists relatively more data on the relationship between anemia and outcome in patients with SAH. Anemia consistently predicts, in a dose-dependent fashion, infarction, death, and dependency [87, 88], and may also predict cognitive impairment [89]. Whether transfusion mitigates these risks is unknown.

Transfusion in patients after SAH is associated with poor outcome in some studies. A small prospective study of fluid therapy in SAH management found that RBCT was associated with poor outcome at discharge, but not at 6 months [90]. A single center, retrospective study found that discharge outcome in transfused patients was worse only in patients with arterial vasospasm [91]. Intraoperative transfusion also may be associated with poor 6-month outcome [92].

Red blood cell transfusion may be associated with increased risk of cerebral vasospasm [90]. In a retrospective study of 441 SAH patients, postoperative RBCT was associated with an increased risk of both angiographic and symptomatic vasospasm, independent of smoking, Hunt-Hess grade, Fisher group, and intraoperative rupture [92]. It was postulated that transfused blood, depleted of nitric oxide, might result in a blunted vasodilatory response to vasospasm.

Other retrospective studies have demonstrated an association between transfusion and increased frequency of cerebral and extra-cerebral complications in patients with SAH [93, 94]. Transfusion significantly increases the risk of major (cardiac, pulmonary, renal, or hepatic) and minor (rash, deep vein thrombosis) medical complications [93]. In a study of 620 SAH patients, RBCT was identified as a risk factor associated with the development of acute lung injury, independent of severity of illness, clinical grade, and severe sepsis [95]. RBCT may be associated with an increased risk of infection and inflammation; this could prove particularly deleterious to the patient with SAH, as infection has been shown to exacerbate delayed cerebral ischemia (DCI) and potentially worsen outcome in SAH [45, 96]. Erythrocyte transfusion may also be associated with an increased risk of thrombotic events in SAH patients, independent of injury severity, baseline demographics, comorbid conditions, other blood component transfusion, and Transcranial Doppler Ultrasound (TCD) vasospasm [97]. This may be explained by rheological and storage-induced changes in RBC structure, coupled with alterations in coagulation and fibrinolysis in SAH patients that may result in an increased risk of thrombosis.

Studies using physiological endpoints, such as cellular metabolism and brain hypoxia, have examined the relationship between hemoglobin concentration and cerebral metabolic dysfunction. Studies that used brain tissue oxygen tension (P_btO₂) monitoring and lactate-to-pyruvate (LPR) measurements via cerebral microdialysis demonstrated increased brain hypoxia and cell energy dysfunction when hemoglobin levels were less than 9 g/dl [98]. This association was independent of other factors such as cerebral perfusion pressure and vasospasm. Other studies have shown that reduced hemoglobin concentration was associated with hypoxia, impaired autoregulation, and increased cellular injury [99, 100]. One mathematical model suggested that hemoglobin concentrations <10 g/dL were associated with increased brain damage in animals with focal ischemia [79].

Physiological studies have also been used to examine the relationship between RBCT and outcome. In 1 study ¹⁵-Oxygen labeled positron emission tomography (¹⁵O-PET) was performed in 8 SAH patients before and after transfusion of 1 unit of RBCs [101]. Transfusion resulted in a 15% rise in both hemoglobin concentration and CaO₂, but did not reduce global CBF, demonstrating that CBF may not be negatively impacted by changes in viscosity induced by transfusion of 1 unit RBCs. Although oligemic regions demonstrated improved DO₂ and CBF after transfusion, CBF fell in areas of vasospasm. It was suggested that transfusion might improve perfusion-related reductions in DO₂ in areas with preserved autoregulation, but not in territories affected by vasospasm. The authors concluded that RBC transfusion could provide benefits similar to CBF augmentation (with intravenous crystalloid infusions and vasopressors), the current mainstay of DCI prevention.

At present, a multidisciplinary consensus panel of the Neurocritical Care Society strongly recommends that patients should receive RBCT to maintain a hemoglobin concentration >8 to 10 g/dl, based on moderate quality data [102]. A pilot study has shown that it is feasible to target hemoglobin thresholds and to prospectively assess outcome after transfusion; a prospective, randomized controlled trial to definitively determine transfusion thresholds in SAH patients is required [103].

Traumatic Brain Injury

There are no prospective randomized studies of transfusion in patients with traumatic brain injury (TBI), and data are largely derived from subgroup analyses of prospective studies in other populations. Trials in general trauma patients suggest that transfusion may increase the risk of both infection and mortality. Dunne et al. [104] demonstrated that transfusion of more than 4 units of blood increases the risk of death by a factor of 3 and increases the risk of perioperative infection by a factor of 9 [104]. In another study, blood transfusion was found to be an independent predictor of mortality, systemic inflammatory response syndrome, ICU admission, and increased ICU length of stay [38]. Trauma patients who receive blood transfusions have a twofold to sixfold increase in systemic inflammatory response syndrome and a more than fourfold increase in ICU admissions [38]. However, some studies of general trauma patients have failed to demonstrate a relationship between RBCT and adverse outcome [105]. In a substudy of the TRICC trial, 203 trauma patients, 25% of whom were identified as having brain injury, were randomized to a liberal versus restrictive strategy transfusion [106]. No significant differences in mortality, multi-organ dysfunction, ICU, or hospital LOS were identified. In a separate analysis, study results confirmed that a liberal transfusion strategy demonstrated no mortality benefit in a smaller cohort of patients with moderate-severe TBI [106].

Transfusion has been shown to improve P_btO₂ in 4 studies of severe TBI [107–110]; however, the magnitude of augmentation was small, the significance was questionable, and the effect was inconsistent. Interestingly, in 1 of these studies, RBCT resulted in a statistically significant increase in brain oxygenation; however, 13 of 30 patients (43%) experienced a decline in P_btO₂ after RBCT [107]. In a study of 49 TBI patients with 564 episodes of compromised P_btO₂, blood transfusion improved compromised P_btO₂ only 50% of the time [111]. In contrast, interventions, such as fraction of inspired oxygen (FiO₂) augmentation, vasospressor utilization, hyperosmolar therapy, and benzodiazepine administration resulted in a >70% response rate.

The variable response to RBCT observed in patients with TBI may be due to the fact that secondary ischemia is less common than previously thought. Physiologic studies suggest that what was thought to represent ischemia may actually signify mitochondrial failure [6]. Cerebral microdialysis has traditionally been used to identify regional ischemia by sampling the interstitium for metabolites, such as glucose, lactate, and pyruvate. The principal marker of cerebral ischemia is the LPR. An LPR >40 has been correlated with PET evidence of regional brain ischemia, particularly in patients with SAH [112–114]. However, an elevated LPR has also been identified in TBI without evidence of ischemia; PET data suggest that an elevated LPR corresponds to nonischemic reductions in cerebral metabolic rate for oxygen, a measure of mitochondrial function [114–116]. Therefore, RBCT intended to improve compromised blood flow may not be warranted.

Optimal use of transfusion in TBI remains unclear. Data from subgroup analysis of the TRICC suggests that there may be no benefit to transfusion in TBI. However, this analysis lacks the necessary power to make definitive conclusions. A survey of physicians at trauma centers in the United States found that transfusion practice in TBI widely varied, and many clinicians held the belief that a restrictive transfusion threshold may be inappropriate for patients with intracranial hypertension [117].

Conclusions

It remains unclear whether RBCT to correct anemia in brain injury patients who are not actively bleeding is warranted. Although RBCT may augment brain oxygenation in compromised tissue, it may also result in a paradoxical decrease in perfusion. This variable effect may be the result of prolonged blood storage. It is possible that patients with impaired baseline perfusion benefit from RBCT more than those without. Transfusion may be more effective in patients with SAH, given the associated risk of DCI, whereas the ischemia previously identified in TBI patients may more accurately represent mitochondrial dysfunction less responsive to transfusion. Individualized transfusion therapy may be preferable, with use of physiological endpoints instead of arbitrary hemoglobin levels. Large, prospective, randomized controlled trials are needed to better define the role of anemia, optimal hemoglobin threshold, and utility of RBCT and are of paramount importance to improve the management of patients in the neurological ICU.