Introduction

Anemia is ubiquitous in the critically ill population, affecting close to 100% of patients with admissions longer than 7 days [1]. Although it may be intuitive that moderate anemia (hemoglobin [Hb] level < 10 g/dL) is well tolerated in healthy individuals, there is now substantive evidence that it is also tolerated in most critical care populations as well [2, 3]. Further, transfusion to maintain higher Hb levels has not proven to be beneficial and, in fact, may be associated with harm in specific subgroups of critically ill patients, specifically those who are younger and less sick [2]. Unfortunately, the neurocritically ill population has been largely underrepresented in these trials, and specific data regarding both anemia and red blood cell (RBC) transfusion are lacking in this population. A systematic review of comparative studies in neurocritically ill patients from 2012 concluded that because of a general paucity of evidence for “clinically significant outcomes” and the overall risk of bias of the included studies, no specific recommendations could be made regarding whether a liberal or restrictive RBC transfusion strategy should be used [4]. In the case of aneurysmal subarachnoid hemorrhage (aSAH), the unique disease course and pathophysiology may make patients particularly vulnerable to the negative effects of anemia [5]. These unique aspects of aSAH should give us pause in applying what we know about managing anemia from other critically ill populations while we seek evidence specifically in this patient population. In this review, we provide an overview of anemia in aSAH, the physiologic considerations for RBC transfusion, and a review of the evidence on which these considerations are based.

Normal Cerebral Oxygenation

The brain is highly metabolically active, with the third highest metabolic rate of all human organs and tissues [6]. The brain relies almost exclusively on oxygen-coupled glucose metabolism to maintain normal function, so it requires a steady flow (delivery) of oxygenated arterial blood [7]. Despite its small contribution to total body mass (2%), the brain receives a large proportion of the resting cardiac output (10–20%) and accounts for 20% of the total basal oxygen consumption at rest (approximately 50 mL O2/min or 3.5 mL/100 g/min) [8].

Delivery of oxygen to the brain (CDO2) is determined by cerebral blood flow (CBF) and the arterial oxygen content (CaO2) of the blood, as follows: CDO2 = CBF × CaO2 [9]. CBF in normal adults is tightly maintained at 50 mL/100 g/min [10]. It can be conceptualized as laminar flow through a tube and therefore be described using both Ohm’s and Hagen-Poiseuille’s equations [11]. As such, the determinants of CBF are arterial blood pressure, venous pressure or intracranial pressure (ICP) (whichever is greater), the radius of the cerebral blood vessels, and blood viscosity [11]. The CaO2 is almost completely determined by the concentration of Hb and the oxygen saturation (SaO2), with a very small contribution from dissolved oxygen: CaO2 = Hb × 1.36 × (%SaO2/100) + (0.003 × PaO2) [9]. Under normal conditions in the resting brain, the ratio of oxygen consumption to delivery (known as the oxygen extraction ratio or fraction) is around 40–44% [12]. This provides a degree of buffer capacity against tissue hypoxia in the event of reduced CDO2.

Anemia and the “Normal Brain”

Multiple human studies have explored the impact of a reduced Hb concentration (and therefore a reduced CaO2) on cerebral oxygen physiology using healthy volunteers who underwent isovolemic hemodilution as a model for anemia. CaO2 has an inverse linear relationship with CBF: as CaO2 is decreased, there is a corresponding increase in CBF [13,14,15,16]. Animal studies suggest that in response to hemodilution, the regulatory mechanism leading to increased CBF is both complex and multifactorial but predominantly occurs via vasodilation and is primarily mediated by nitric oxide metabolites [9, 17, 18]. This increase in CBF attempts to maintain CDO2 and ultimately preserve tissue oxygenation. However, studies in both animals and humans show that the increase in CBF is not sufficient to completely maintain CDO2, especially at lower Hb concentrations [14, 15, 19,20,21]. Importantly, despite this drop in CDO2, the cerebral metabolic rate of oxygen (CMRO2) remains unchanged because of an increase in cerebral O2 extraction [14, 15, 19]. Therefore, to maintain the CMRO2 in the setting of isovolemic hemodilution (anemia), cerebral O2 extraction increases to compensate for the reduced CDO2.

Although the compensatory increase in CBF may not always be sufficient to maintain CDO2 in the setting of severe anemia (as previously illustrated), it is nevertheless a key component of the response that attempts to preserve tissue oxygenation. Both animal and human studies show that vasodilation is an important contributor to increased CBF in the setting of isovolemic hemodilution [22, 23]. Although hemodilution-induced vasodilation increased CBF, it also caused a reduction in the vasodilator reserve, which could limit the autoregulatory response to additional insults [23]. This vasodilation is part of the cerebrovascular autoregulation response to variation in pressure [24].

Pressure, in this case cerebral perfusion pressure (CPP), is another potentially important factor in the determination of CBF. Any substantial reduction in mean arterial pressure or increase in ICP could reduce CBF. However, because of autoregulation of the cerebral vasculature, the radius of the vessels changes in response to changes in CPP to maintain a constant CBF across a wide range of CPPs or mean arterial pressures (50–150 mmHg) [25]. In the normal brain, autoregulation is preserved, but in the injured brain, autoregulatory capacity can be altered or abolished entirely, leading to impaired CBF and delivery of oxygen [26, 27].

Conversely, although blood viscosity is a factor in CBF, its clinical significance is questioned. Both animal and human studies have found that viscosity likely has a limited to nonexistent role in the control of CBF [19, 28].

Anemia and the Injured Brain

In the injured brain, the negative effects of anemia can be exacerbated by both the features of the injury itself (e.g., elevated ICP or vasospasm) and the compensatory mechanisms that attempt to normalize tissue perfusion and oxygenation (vasodilation via autoregulation). Indeed, animal models of brain injury suggest that the injured brain is highly susceptible to anemia. For example, a canine model of global cerebral ischemia, using isovolemic hemodilution to simulate anemia, demonstrated that at a hematocrit level of 26% there was uncoupling of CaO2, CDO2, and the CMRO2 [29]. Although CBF increased, both CDO2 and CMRO2 fell, and the oxygen extraction fraction increased above the critical level for irreversible ischemia [29]. This may reflect a decrease in the cerebral perfusion reserve and the occurrence of irreversible ischemia.

In a rat model of traumatic brain injury, injured regions of the brain suffered significant secondary injury in the setting of anemia from isovolemic hemodilution [30]. This is compared with noninjured regions, which maintained normal cerebral oxygen tension and did not suffer injury [30]. Finally, In a mathematical modeling study of the effect of anemia on focal ischemic stroke in rabbits, when Hb concentrations were < 10 g/dL, the oxygen uptake in the ischemic penumbra “decreased progressively,” converting potentially salvageable tissue into a permanent infarct [31].

These findings appear to translate to the human population with aSAH. A study of eight patients with transcranial doppler ultrasound-confirmed vasospasm in aSAH assessed cerebral oxygen physiology before and after isovolemic hemodilution, from a hematocrit level of 36–28% (Hb 11.9–9.2 g/dL) [21]. By measuring global and regional CBF with radiolabelled tracers (133Xenon and 99mTc-HMPAO (technetium-99m-labeled hexamethylpropyleneamine oxime)), they demonstrated that anemia leads to disordered cerebral oxygen physiology and increased ischemic brain volume, which could ultimately lead to worse clinical outcomes [21].

Anemia in aSAH is Common and Problematic

Anemia is a common occurrence in aSAH, with approximately 50% of patients developing moderate anemia during their hospital stay [32]. Despite the fact that few patients (~ 5%) have anemia on presentation to the hospital, moderate anemia tends to occur early (day 4 post ictus) in the majority who develop it [32].

Multiple independent risk factors have been associated with the development of anemia in aSAH, including female sex, aSAH clinical severity (modified Fisher grade of 3 or 4 and/or a poor clinical grade), surgical aneurysm treatment, a baseline hematocrit level of < 36%, a history of hypertension or oral anticoagulant use, and the presence of ≥ 3 Systemic Inflammatory Response Syndrome criteria on admission [32, 33]. Inflammation is thought to be a key factor leading to increased consumption and decreased RBC production through myelosuppression as well as abnormalities with erythropoietin and iron metabolism that occur in critically ill patients [34]. Additionally, anemia can be exacerbated by a variety of factors associated with treatment, including hemodilution from fluid resuscitation, intraoperative loss during aneurysm repair, or blood sampling for laboratory testing [34].

Anemia and Adverse Clinical Outcomes in aSAH

Multiple studies have examined the association of anemia with clinical outcomes in aSAH (Table 1). Anemia appears to be associated with unfavorable outcome, though there are conflicting findings across studies [32, 35,36,37,38,39,40,41]. It is difficult to draw firm conclusions because these studies are generally single-center retrospective observational cohort studies with relatively small sample sizes. Further, there is significant heterogeneity across the studies with respect to a number of features, including the definition of anemia used, the time course of anemia studied, the number and type of adjusted confounders, and the specific outcome measures assessed. It thus remains unclear whether anemia is simply a marker of comorbidity and severity of illness or truly an independent prognostic factor.

Table 1 Studies published in the last 15 years showing the association of anemia and mortality or unfavorable outcome in aneurysmal subarachnoid hemorrhage
Full size table

Physiologic Effects of Anemia in aSAH

Physiologic data strongly support the association of anemia with harm in aSAH. A prospective observational cohort study of 20 patients with poor-grade (Hunt and Hess grade 4 or 5) aSAH used multimodal monitoring to evaluate the association of anemia with brain tissue hypoxia (using brain tissue oxygen tension [PbtO2]) and cellular energy dysfunction (using cerebral microdialysis lactate/pyruvate ratio) [42]. An Hb level < 9 g/dL was a strong independent predictor of both brain tissue hypoxia and cellular energy dysfunction, adjusted for CPP, central venous pressure, the ratio of partial pressure of oxygen in arterial blood to the fraction of inspiratory oxygen (P/F ratio), and presence of symptomatic vasospasm [42].

A retrospective cohort study of 34 patients with subarachnoid hemorrhage used multimodal monitoring data (PbtO2 and cerebral microdialysis) from a total of 259 h of monitoring to evaluate the association of low Hb levels with brain tissue hypoxia and cellular energy dysfunction [43]. The magnitude of anemia impacted brain hypoxia; for every 1-g/dL decrease in the Hb level, there was a 70% increase in brain hypoxia [43]. Compared to an Hb level of 10.1–11 g/dL, an Hb level of ≤ 9 g/dL was associated with a 280% increase in cellular energy dysfunction.

Although aSAH shares the basic physiological principles of cerebral oxygenation and metabolism with other forms of brain injury, there are unique features that may contribute to a particular vulnerability to anemia of this population (Fig. 1). The postbleed period of aSAH is thought to be complicated by two distinct phases: a period of early brain injury (typically during the first 72 h) and a period of delayed cerebral ischemia (DCI)/vasospasm (beyond 72 h until 21 days) [44].

Fig. 1
figure 1

Adapted from Okazaki T, Kuroda Y. Aneurysmal subarachnoid hemorrhage: intensive care for improving neurological outcome. J Intensive Care. 2018;6:28. https://doi.org/10.1186/s40560-018-0297-5 and from Dodd WS, Laurent D, Dumont AS, et al. Pathophysiology of delayed cerebral ischemia after subarachnoid hemorrhage: a review. J Am Heart Assoc. 2021;10:e021845.

Impact of anemia and RBC transfusion in aSAH: a conceptual model. aSAH aneurysmal subarachnoid hemorrhage, CPP cerebral perfusion pressure, DO2 xxx, ICP intracranial pressure, PbtO2 brain tissue oxygen tension.

Full size image

Anemia During the Early aSAH Phase (0–72 h): Early Brain Injury

Early brain injury is a relatively new concept in aSAH, and as a result, the pathophysiologic processes are not well understood [44]. There is a variety of proposed mechanisms believed to contribute to injury during this phase, including the following [45]:

  • Sudden increase in ICP leading to decreased CPP, impaired autoregulation, and possibly transient or persistent ischemia

  • Neuronal cell death and endothelial damage leading to cytotoxic edema and breakdown of the blood–brain barrier potentiating vasogenic edema

  • Further cell death via microcirculatory failure, microthrombosis, altered ionic hemostasis, excitotoxicity, oxidative stress, and neuronal swelling

  • Energy dysfunction due to cortical spreading depolarizations or mitochondrial dysfunction

  • Subarachnoid blood or intraventricular blood may cause microglial activation and inflammation

During this early phase, CDO2 is at particular risk of being reduced because of elevated ICP (from edema or hydrocephalus), intentionally lowered blood pressure (to prevent aneurysmal rebleeding), and possible cardiac dysfunction (neurogenic stress cardiomyopathy). Meanwhile, in the setting of impaired autoregulatory mechanisms, liberalized blood pressure targets post aneurysm securement could lead to hyperemia, edema, and elevated ICP, further worsening CDO2, particularly to regional areas of injury and ischemia.

In general, most of the studies of anemia in aSAH do not discriminate between the early and the delayed phases of aSAH. However, a secondary analysis on 413 patients from the CONSCIOUS-1 (Clazosentan to overcome neurological ischemia and infarction occurring after subarachnoid hemorrhage) study was able to achieve some distinction between these phases. The CONSCIOUS-1 study was a phase 2 randomized, double blind, placebo-controlled dose-finding trial in patients with aSAH that compared clozosentan (an endothelin receptor antagonist) to a placebo, with the primary end point of angiographic vasospasm [36]. The secondary analysis found that anemia (Hb level < 10 g/dL) was present in 29% of patients during days 1–3, and it was associated with poor neurologic outcome and increased risk of death [36]. Interestingly, anemia that occurred during the “peak DCI risk period” (days 5–9) was only associated with poor neurologic outcome but not mortality [36].

Anemia During the Delayed Phase (72 h to 21 Days): DCI/Vasospasm

The delayed phase of aSAH is classically the period from 72 h up to 21 days post rupture and is typified by both vasospasm and DCI. Previously, it was thought that macrovascular vasospasm led directly to DCI; however, contemporary evidence shows that this is not the case: approximately 70% of patients develop angiographic vasospasm, whereas only 30% develop DCI [44]. Other mechanisms are clearly contributing, of which CDO2 and Hb concentration may be one important element.

Based on the earlier discussion of the physiology of cerebral oxygen delivery, there is a significant possibility that anemia will reduce the CDO2, leading to increased risk of ischemia. In a study of eight patients with aSAH and vasospasm, isovolemic hemodilution to a hematocrit level of 28% produced a 12% decrease in global cerebral oxygen delivery and a 7% increase in the volume of ischemic brain tissue, despite an overall increase in CBF [21].

Additional evidence suggesting the importance of CaO2 and CDO2 in the time period of vasospasm and DCI comes from a retrospective cohort study of 245 patients with aSAH [38]. Patients with unfavorable outcomes had a lower Hb level between days 6 and 11 [38]. However, the notion that a higher Hb level is associated with better outcomes is challenged by conflicting data which suggests that an elevated Hb level on admission is associated with a higher incidence of DCI and deep vein thrombosis [46].

Transfusion in aSAH: Available Evidence

Transfusion and Physiologic Outcomes

Studies evaluating physiologic outcome measures support the intuitive notion that transfusion addresses the physiologic and clinical harms of anemia. A small prospective study of patients with aSAH with anemia found that RBC transfusion led to an 18% increase in CDO2, with global CBF remaining stable [47]. Similarly, in a study using oxygen-15 positron emission tomography, RBC transfusion in patients with aSAH who were at risk for DCI led to an increase in CBF and CDO2. Because the findings were similar to those of other interventions, including a fluid bolus (15 mL/kg) and induced hypertension (25% increase), the authors concluded that transfusion may be a meaningful alternative or adjunct to hemodynamic augmentation [48]. Another small cohort study of 15 patients with aSAH found that transfusion led to a significant increase in PbtO2 (every 1-g/dL increase in the Hb level was associated with a 1.39-mmHg increase in PbtO2 [p = 0.036]) [49]. Finally, a prospective cohort study evaluated 56 anemic (Hb level 7–13 g/dL) patients with aSAH who had their aneurysm secured and were deemed at risk for DCI [50]. Using oxygen-15 positron emission tomography scans to measure CDO2 pretransfusion and post transfusion of a single unit of RBCs, they found a 10% increase in global CDO2 and a 16% increase in CDO2 to vulnerable brain regions [50].

Transfusion and Clinical Outcomes: Observational Evidence

Despite this apparent physiologic rationale that transfusion improves CDO2, studies evaluating clinical outcomes have yielded conflicting results (Table 2). However, these results are affected by several limitations. First, the majority of these studies included data prior to the publication of the International Subarachnoid Aneurysm Trial and widespread acceptance of endovascular aneurysm treatment [51]. Patients requiring transfusion may have been more likely to have had open surgical treatment of their aneurysm. Second, given the observational nature of these studies, confounding may have played a significant role. Other factors that are associated with both transfusion and poor outcome (confounding by indication) could have biased the results. Third, there is significant variation in the transfusion thresholds used in these studies (from 7 to 11.5 g/dL), creating significant heterogeneity and limiting the comparison of results across studies. In actual practice, it appears that most clinicians are using transfusions thresholds in the 8–9-g/dL range. A North American survey from 2011 showed that for good-grade aSAH, the mean transfusion threshold was 7.85 g/dL, and for patients with DCI, it was 8.58 g/dL [52]. Fourth, the majority of studies on RBC transfusion in aSAH are retrospective in nature. These do not account for the timing of RBC transfusion (before, during, or after DCI), and as a result, DCI may be the cause of the unfavorable outcomes as opposed to the RBC transfusion [53]. Interestingly, a recent systematic review from 2021 describes a number of the same concerns but concludes that the overall benefits of transfusion likely outweigh the risks [54].

Table 2 Studies published in the last 15 years showing the association of RBC transfusion and mortality or unfavorable outcome in aneurysmal subarachnoid hemorrhage
Full size table

Transfusion and Clinical Outcomes: Randomized Controlled Trial Evidence

Clinical trials evaluating anemia and transfusion thresholds specific to aSAH are lacking. There is a single pilot trial evaluating the safety and feasibility of RBC transfusion in patients with aSAH at high risk of vasospasm [55]. The trial enrolled patients within 3 days of ictus who had a World Federation of Neurological Surgeons grade of 2–4 (or grade 1 with thick subarachnoid clot) and randomized them to an Hb transfusion trigger of 10 vs. 11.5 g/dL. The primary hypothesis was that an Hb level of 11.5 vs. 10 g/dL would not increase the number of days with a core temperature (> 100.4 °F) or decrease the number of ventilator-free days (during the first 13 days). The primary feasibility end point was that a higher Hb goal level would lead to a significantly higher Hb level during the study period. Radiographic (cerebral infarction) and clinical outcomes (impairment of functional status) were assessed. Cerebral infarction was evaluated at 14 days using magnetic resonance imaging, and clinical outcomes were measured at multiple time points over 3 months (using the National Institutes of Health Stroke Scale and/or the modified Rankin Scale). The results showed that the Hb level was different between the groups starting at day 4, but there was no difference in clinical outcomes. As a small pilot study, it was not powered to detect meaningful differences in outcomes, leading the study authors to conclude that a phase III trial is needed. Additionally, although the results confirmed both the safety and feasibility, the transfusion targets were considerably higher (10 vs. 11.5 g/dL) than those used by most clinicians, limiting the generalizability of the study protocol and, consequently, the external validity of the study results.

Guidelines

Not surprising given the existing physiologic and clinical data, the recommendations for the management of anemia are either inconsistent between guidelines or absent entirely (Table 3) [56,57,58,59,60]. The British Committee for Standards in Haematology guidelines recommend a target Hb level of 8–10 g/dL in patients with aSAH, while recognizing that the optimal Hb level has not be defined and that there is uncertainty surrounding the effects of transfusion on outcomes in this population [56]. The Neurocritical Care Society previously recommended maintaining a higher Hb level (8–10 g/dL) in patients with or at risk for DCI [57]. However, the new Neurocritical Care Society guidelines identify that there is insufficient evidence to recommend a transfusion threshold of greater than 7 g/dL in patients with aSAH [61]. The American Heart Association/American Stroke Association guidelines state that RBC transfusion “might be reasonable” for patients at risk of DCI, but the ideal Hb level is unclear [58]. The Chinese Stroke Association guidelines mention that a higher baseline Hb level is associated with better outcomes, but they recommend against RBC transfusion [62]. And finally, both the European Stroke Organization and the Japanese Society on Surgery for Cerebral Stroke guidelines do not address anemia or transfusion in aSAH [59, 60].

Table 3 Major society guideline recommendations for anemia/transfusion in aSAH
Full size table

Upcoming Evidence/Future Considerations

Clinical trials evaluating the effect of RBC transfusion on meaningful clinical outcomes are needed. As afore reviewed, there is little existing evidence to guide clinicians on, specifically, when to transfuse, in which patients, and at what threshold. The effect of transfusion on the aSAH disease course, as well as patient-important clinical outcomes, needs to be better understood.

There are two relevant ongoing studies in the field:

  1. 1.

    The SAHaRA (Aneurysmal SubArachnoid Hemorrhage - Red blood cell transfusion And outcome) trial (NCT03309579) addresses anemia and transfusion thresholds specifically in aSAH. It is a Canadian-led international prospective open-label, blinded end-point randomized controlled trial that plans to enroll 740 adults with aSAH and moderate anemia (Hb level ≤ 10 g/dL) [63]. The study compares a liberal versus a restrictive RBC transfusion trigger (≤ 10 vs. ≤ 8 g/dL) during the first 21 days. The primary outcome is functional neurologic status measured at 12 months using the modified Rankin Scale. The estimated study completion date is in the fall of 2024.

  2. 2.

    The TRAIN (TRansfusion strategies in Acute brain INjured patients) trial (NCT02968654) is a European prospective randomized controlled trial that plans to enroll 1,000 patients within 10 days of acute brain injury, including aSAH (as well as other causes, such as traumatic brain injury) and randomize them to either a restrictive RBC transfusion trigger (7 g/dL) or a liberal RBC transfusion trigger (9 g/dL) for 28 days or until hospital discharge. The primary outcome is good neurological outcome at 180 days, defined as a Glasgow Outcome Score Extended score of 6–8 [64]. The estimated study completion date is the summer of 2023.

The results of these studies should help inform the timing and Hb trigger for transfusion in the acute phases of management of this particularly vulnerable patient population.

Suggested Practice Guidance

Evidence supporting a specific transfusion threshold in the general critical care population should not necessarily be extrapolated to aSAH because of the unique disease processes that can affect CDO2, including early brain injury and DCI. The majority of the current evidence, both physiologic and clinical, supports an association between anemia and unfavorable outcomes in patients with aSAH.

Evidence supporting a particular transfusion strategy in aSAH is extremely limited [65]. Physiologically, RBC transfusion in anemia leads to improvements in CDO2 and reduction in ischemic burden. However, studies evaluating clinical outcomes have yielded conflicting results. The studies reviewed in this article have transfusion triggers ranging from 7 to 11.5 g/dL. Most studies on anemia and transfusion seem to support a more liberal transfusion strategy, with a target Hb level ≥ 10 g/dL [54]. However, this must be balanced with the known risks of transfusion and the limitations of this evidence (Table 4).

Table 4 Key summary points of proposed practice guidance
Full size table

In light of the limited evidence, a practical approach to anemia in aSAH includes avoidance of significant anemia with a transfusion threshold of 7–8 g/dL during the acute postrupture period (up to 21 days). Aiming for a higher Hb level in higher risk patients, such as those with clinical and/or physiologic risk factors, including poor clinical grade, presence of intracranial hypertension, active vasospasm or DCI, physiologic markers of hypoxic distress, or systemic factors that may negatively affect cerebral CDO2 (cardiac dysfunction, vasodilatory shock, etc.), is intuitively appealing but not supported by current evidence. With these points in mind and depending on local protocols, it is reasonable to transfuse patients with aSAH with the objective to aim for Hb levels above 7–10 g/dL. Randomized controlled trial evidence is needed to better inform RBC transfusion thresholds in anemic patients with aSAH, and such evidence is forthcoming.

Conclusions

In summary, evidence surrounding RBC transfusion in aSAH is bolstered by a strong physiologic rationale but limited to small largely single-center retrospective observational studies with conflicting results. The timing of transfusion in the aSAH disease course is poorly understood, as is its effect on outcome. Further trial evidence is needed.