The critical care management of poor-grade subarachnoid haemorrhage

A PubMed search for articles published until May 2015 was performed by using the terms “Subarachnoid Hemorrhage” [Mesh] AND (“poor-grade” [Title/Abstract] OR “high-grade” [Title/Abstract]), which returned 236 articles. Additionally, the reference lists of the most recent guidelines on the management of SAH were searched [8‐10]. Lastly, the authors’ personal databases were used as an additional source for this review.

During aneurysmal SAH, extravasation of high-pressure arterial blood in the subarachnoid space (and often into the brain parenchyma and ventricles) is associated with a sudden ICP increase that, if severe and sustained, may compromise cerebral perfusion, causing global cerebral ischaemia and EBI (Fig. 1). If the haemorrhage does not stop, acute cardiopulmonary instability associated with intracranial hypertension or compromised cerebral blood flow (CBF) leads to patient death before hospital admission. In patients who survive the initial haemorrhage, re-bleeding is the most severe early complication; the reported incidence is up to 15 % in the first 24 hours, and the fatality rate is approximately 70 % [11‐13]. Patients with poor-grade SAH are at higher risk of re-bleeding [14]. Initial management therefore should focus on strategies aimed to prevent re-bleeding and to control ICP.

Early aneurysm repair is generally considered the most important strategy to reduce the risk of aneurysm re-rupture [15]. However, evidence for optimum timing of treatment is limited, and it is unclear whether ultra-early treatment (less than 24 hours) is superior to early aneurysm repair (within 72 hours). A recently published retrospective data analysis comparing ultra-early treatment with repair performed within 24–72 hours after haemorrhage suggests that aneurysm occlusion can be performed safely within 72 hours after aneurysm rupture [16]. The American Heart Association/American Stroke Association [9] recommend as a Class IB Recommendation that “surgical clipping or endovascular coiling of the ruptured aneurysm should be performed as early as feasible in the majority of patients to reduce the rate of re-bleeding after SAH”. This recommendation for timing of aneurysm intervention is corroborated by the European Stroke Organization Guidelines for the Management of Intracranial Aneurysms and Subarachnoid Haemorrhage [10], which stated that “aneurysm should be treated as early as logistically and technically possible to reduce the risk of re-bleeding; if possible it should be aimed to intervene at least within 72 hours after onset of first symptoms”.

The results from an ongoing trial only enrolling patients with poor-grade SAH may help answer the question of whether early treatment (within 3 days) is associated with improved outcome compared with intermediate (days 4–7) or late (after day 7) treatment [17].

The choice of treatment modality between surgical clipping and endovascular coiling is a complex endeavour, which requires the expertise of an interdisciplinary team, including neurointensivists, interventional neuroradiologists and neurovascular surgeons. For aneurysms considered to be equally treatable by both modalities, the endovascular approach is superior, being associated with better long-term outcomes [18‐20]. Randomised trials of clipping versus coiling included mostly good-grade patients, leading to controversy as to whether their results apply also to poor-grade patients. Retrospective data on clipping and coiling in poor-grade patients seem to suggest that surgical clipping and endovascular are equally effective [21]. An early and short course of an anti-fibrinolytic drug such as tranexamic acid, initiated as soon as the radiological diagnosis of SAH is established and stopped within 24–72 hours, has been associated with decreased rate of ultra-early re-bleeding and a non-significant improvement in long-term functional outcome [22]. This approach remains controversial [23], and short-term administration of tranexamic acid to prevent re-bleeding is being further studied in a multicentre randomised trial (Dutch Trial Registry number NTR3272) [24]. Another medical intervention applied to prevent aneurysm re-rupture is the avoidance of extremes of blood pressure. The American Heart Association/American Stroke Association [9] and the Neurocritical Care [8] guidelines suggest keeping the mean arterial blood pressure below 110 mm Hg or systolic blood pressure below 160 mm Hg (or both) in the presence of ruptured unsecured aneurysm. The European guidelines are less aggressive and suggest keeping the systolic blood pressure below 180 mm Hg [10]. These parameters should not be used after aneurysm treatment, when spontaneously high blood pressure may be beneficial [25].

Intracranial hypertension (ICP of at least 20 mm Hg) is a relatively common complication of SAH, especially in patients presenting with poor neurological condition [26‐28]. Multiple factors such as cerebral oedema, intraparenchymal haematoma, acute communicating hydrocephalus, intraventricular haemorrhage, aneurysm re-rupture, complications related to aneurysm treatment, EBI, and DCI can contribute to the development of intracranial hypertension [29]. High ICP is associated with severe derangements of cerebral metabolism [30], increased risk of neurological deterioration [25], and poor outcome, especially if refractory to medical treatment [29, 31]. ICP of greater than 20 mm Hg is an independent predictor of severe disability and death in aneurysmal SAH [30].

Principles of management of intracranial hypertension after SAH have been traditionally adopted from traumatic brain injury (TBI) literature [32] and are not specifically designed for the SAH population. However, these two entities are different from a pathophysiological perspective, and the use of therapies tested in patients with TBI in the SAH population is controversial. Currently, the role of therapies such as hyperosmolar agents, hypothermia, barbiturates, and decompressive craniectomy is not well established in SAH patients with intracranial hypertension refractory to first-line treatments.

The initial approach to raised ICP includes head of bed elevation (between 30° and 45°) to optimise cerebral venous drainage, normoventilation (arterial partial pressure of carbon dioxide (PaCO₂): 35–40 mm Hg) [33], use of sedation and analgesia to achieve a calm and quiet state (Richmond Agitation Sedation Scale score of −5 or Sedation-Agitation Scale score of 1), and surgical intervention in the presence of mass-occupying lesions [34]. The use of neuromuscular blocking agents is sometimes applied to prevent ICP surges during tracheal suctioning and physiotherapy; however, the role of these drugs for ICP management is not well established, and some authors suggest that they may be more deleterious than beneficial [35]. If ICP remains elevated despite these interventions, a short course (less than 2 hours) of hyperventilation (PaCO₂ of 30–35 mm Hg) might be considered while new brain imaging is obtained and other interventions are planned and initiated [36‐38].

Cerebrospinal fluid (CSF) drainage is a mainstay in ICP management of patients with SAH, especially when hydrocephalus is present [39]. Acute hydrocephalus is common in SAH, and approximately 50 % of patients are affected on admission [40]. When hydrocephalus is associated with a decreased level of consciousness, an external ventricular drain (EVD) should be inserted to allow CSF drainage and ICP monitoring. EVD insertion before aneurysm treatment has been shown to be safe and not associated with increased risk of aneurysm re-rupture [40, 41], if accompanied by early aneurysm repair. Additionally, when EVD insertion is performed before aneurysm repair, CSF drainage should be practiced with caution because rapid and aggressive CFS drainage can increase transmural pressure, increasing the risk of aneurysm re-rupture [41, 42]. Interestingly, approximately 30 % of patients with poor-grade SAH improve neurologically after EVD insertion and CSF drainage. These responders have a functional outcome similar to that of good-grade (WFNS I–III) patients [39].

Hyperosmolar agents, such as mannitol and hypertonic saline, are usually considered when the above strategies fail to control ICP, although their role on clinical outcome in the SAH population is not well established. We could not identify any study addressing the role of mannitol in the management of raised ICP in the SAH population; for hypertonic saline, we found only case series [43‐46] and a small placebo-controlled trial in patients with raised but stable ICP [47]. In these studies, hypertonic saline was effective to control ICP and improved CBF [43‐47] and may improve outcome in the poor-grade population [43].

The last line of treatment includes the use of barbiturates, induced hypothermia, and decompressive craniectomy [38, 48]. Therapeutic hypothermia has been shown to be effective to control ICP in SAH but has not been associated with improved functional outcome and reduced mortality rates in patients with poor-grade SAH [49]. The association of barbiturate coma and mild hypothermia (33–34 °C, median treatment of 7 days) was studied in 100 SAH (64 poor-grade) patients with intracranial hypertension refractory to other medical interventions [50]. Approximately 70 % of patients were severely disabled or dead at 1 year, and more than 90 % of patients developed medical complications associated with the hypothermia/barbiturate treatment (i.e., electrolyte disorders, ventilator associated pneumonia, thrombocytopenia, and septic shock). Decompressive craniectomy is another possible strategy for refractory ICP management in patients with SAH. Poor-grade patients are more commonly exposed to this rescue therapy than patients with good-grade SAH [51, 52]. Decompressive craniectomy has been associated with decreased mortality [53], significant reduction of ICP [34], improved cerebral oxygenation [54, 55], and improved cerebral metabolism [56]. However, most patients undergoing decompressive craniectomy due to refractory ICP have poor outcome, with severe disability or death [56]. Many authors suggest that, if any benefit can be achieved with decompressive craniectomy, this may be best obtained when the procedure is performed early (within 48 hours from the bleeding) [52] and in the absence of radiological signs of cerebral infarction [51]. Finally, in poor-grade patients with large intraparenchymal or Sylvian fissure haematomas usually from middle cerebral artery aneurysms, prophylactic decompressive craniectomy should be considered [34].

It is important to mention that long-term outcome after acute brain injury is markedly improved when patients are managed in a dedicated neurologic/neurosurgical intensive care unit (ICU) [57, 58]. Especially after SAH, outcome is affected by hospital caseload, and better outcomes happen in high-volume centres (centres treating more than 60 patients per year) [59]. Six-month mortality is inversely associated with hospital annual caseload; there is a 24 % reduction in mortality for each 100 patients admitted per year [60]. Regardless of initial grade, early transfer to a high-volume centre is safe and cost-effective and should be pursued [61‐63].

Delayed neurological deterioration occurs frequently in the first 2 weeks after SAH. Common causes of this deterioration include neurological events such as progression of EBI, hydrocephalus, seizures, ischaemia, and systemic conditions, such as fever and infections, respiratory failure, and electrolyte abnormalities. Any delayed neurological deterioration presumed to be related to ischaemia that persists for more than 1 hour and cannot be explained differently has been defined as DCI [7] (Table 1). DCI occurs in up to 30 % of SAH patients surviving the initial haemorrhage. It can present as an acute or insidious change in the level of consciousness or as a focal neurological symptom, such as aphasia or hemiparesis, or as both. These symptoms can be reversible if treated promptly and aggressively; otherwise, DCI tends to progress to cerebral infarction, which is associated with higher rates of disability and mortality. Traditionally, DCI has been considered to be related to a cerebral vasoconstriction (angiographic vasospasm) that begins approximately 3 days and peaks 1 week after the haemorrhage and starts resolving after 2 weeks [64]. However, recent evidence suggests that DCI is a complex, multifactorial syndrome, which can include additional pathophysiologic processes beyond angiographic or sonographic vasospasm (Fig. 2 and Table 2) [65]. DCI may also occur in cerebral territories without evidence of angiographic vasospasm [66]. EBI (defined as brain injury developing in the first 72 hours after haemorrhage) has significant impact on likelihood and severity of subsequent ischaemic changes [67, 68]. For example, poor-grade patients, who have worse EBI, as well as patients who lose consciousness at the time of SAH (and therefore have at least a short episode of transient global cerebral ischaemia) have increased risk of DCI [67, 68].

Cortical spreading ischaemia (CSI) is a wave of depolarisation in the grey matter that propagates across the brain at 2–5 mm/minute [69, 70], leading to depression in evoked potentials and spontaneous electroencephalogram activity. The use of invasive subdural electrocorticographic monitoring combined with regional CBF measurements has shown that CSI can occur isolated or in clusters, and the depolarisation waves are associated with profound cortex hypoperfusion secondary to vasoconstriction [71]. The vast majority of cortical spreading depolarisation waves usually happen in the first 2 weeks after aneurysm rupture, and 75 % of all CSIs recorded occur between the fifth and seventh day post-bleeding [72]. In a prospective multicentre study, Dreier et al. [73] assessed the incidence and timing of spreading depolarisations and DCI after SAH. Eighteen SAH patients requiring craniotomy for aneurysm treatment were monitored for up to 10 days with subdural electrodes. Cortical spreading depolarisations were detected in 13 patients (72 %). DCI was detected in seven patients and was time-locked to a sequence of recurrent spreading depolarisations in all seven cases. Additionally, delayed ischaemic strokes verified by serial computed tomography (CT) scans or magnetic resonance imaging occurred in the recording area in four patients. In another prospective study, using a novel subdural opto-electrode technology for simultaneous laser-Doppler flowmetry and direct current-electrocorticography, combined with measurements of tissue partial pressure of oxygen, Dreier et al. [71] studied 13 patients with SAH. Isolated spreading depolarisations were detected in 12 of those. These waves of depolarisations were associated with physiological, absent, or inverse regional CBF responses. Normal haemodynamic response was associated with tissue hyperoxia, whereas inverse response led to tissue hypoxia. Five patients presented clusters of prolonged spreading depolarisations with persistent depressions. These clusters of spreading depolarisations were closely associated with structural brain damage as observed by neuroimaging. Similarly, Bosche et al. [72] have reported low cerebral measurements of tissue partial pressure of oxygen occurring during clusters of spreading depolarisations.

Microthrombosis is common after SAH [74]. Subarachnoid blood and blood products activate inflammatory pathways, along with tissue factor in the microcirculation of cerebral vessel wall, leading to endothelial cell activation and damage, which in turn cause mural thrombus formation and release of microemboli [75]. Markers of increased activity of the coagulation cascade have been associated with DCI, cerebral infarction, and poor outcome [76]. For example, in a group of 90 patients with SAH, early (within 3 days of SAH onset) elevated concentrations of von Willebrand factor were associated with poor outcome (crude odds ratio (OR) = 4.6, 95 % confidence interval (CI) 2.0–10.9; adjusted OR = 3.3, 95 % CI 1.1–9.8), ischaemic events (crude hazard ratio (HR) = 2.3, 95 % CI 1.1–4.9; adjusted HR = 1.8, 95 % CI 0.8–3.9), and occurrence of spontaneous DCI (crude HR = 3.5, 95 % CI 0.9–13.1; adjusted HR = 2.2, 95 % CI 0.5–9.8). The hypothesis is that this early elevation in von Willebrand factor levels probably reflects the formation of microthrombi in the cerebral circulation [77, 78]. Autopsy studies have shown that patients who developed DCI-related cerebral infarction had significantly more microthrombi compared with SAH patients who died because of re-bleeding or acute hydrocephalus [75, 79].

Haptoglobin is a complex tetramer glycoprotein, consisting of two α and two β chains, synthesised mainly by the liver [80]. It is an acute-phase protein that increases in plasma during major stress situations, such as sepsis, burns, and major trauma. Some recent studies have suggested that the haptoglobin α1-α1 isoform could be protective after SAH [81‐83]. Haptoglobin binds free extracellular haemoglobin, which reduces free haemoglobin ability to generate oxygen-free radicals and therefore interferes in one of the possible pathophysiological pathways of angiographic vasospasm (i.e., haemoglobin-mediated oxidative stress) [82].

Kantor et al. [82] found, in a cohort of 193 patients with SAH, that the haptoglobin α2-α2 isoform was associated with worse functional outcome at 3 months when compared with the α1-α1 genotype. The haptoglobin α2-α2 isoform has a lower affinity for binding haemoglobin and possibly inhibits haptoglobin-haemoglobin clearance because of its larger size [84]. The α2-α2 genotype remained significantly associated with worse functional outcome (OR 4.138; P = 0.0463) after adjustment for age, sex, Fisher grade, and Hunt and Hess grade. A previous study had already shown that haptoglobin α2-α2 genotype was associated with higher rates of angiographic vasospasm by transcranial Doppler (TCD) and conventional angiography performed between days 3 and 14 after SAH [81]. A recent study by Leclerc et al. [83] showed, in a cohort of 74 patients with SAH, that haptoglobin α2-α2 genotype was an independent risk factor for the development of focal and global angiographic vasospasm and also predictive of unfavourable functional outcomes and mortality.

The hypothesis is that patients with haptoglobin α2-α2 genotype do worse because of reduced CSF clearance of haemoglobin, increased reactive oxygen species, and therefore development of more inflammation. This hypothesis is corroborated by an experimental model of SAH, which showed that mice expressing human α2-α2 haptogobin developed more severe angiographic vasospasm and increased macrophage/neutrophil counts in the CSF after SAH, when compared with wild-type α1-α1 haptogobin-expressing mice [85]. Although there is no clinical intervention directly designed to address this important recent finding on the pathophysiology of SAH, the genetic effect on outcome after SAH may increase our knowledge of the disease.

It is well described that medical complications after SAH have a negative impact on survival and functional outcome. Up to 80 % of patients will develop a serious medical complication during phase 2, increasing the risk for secondary brain injury [129].

Cardiac complications following SAH can range from benign electrocardiogram changes to overt cardiogenic shock requiring intra-aortic balloon pump [130, 131]. Positive troponin is a good marker of left ventricular dysfunction after SAH [132], which increases the risk of hypotension, pulmonary oedema, and cerebral infarction [133]. The treatment is mainly supportive, and most of the cases will recover spontaneously within 2 weeks [134]. However, aggressive ICU management may be required in the setting of severely impaired left ventricular function and DCI. Thus, the use of inotropic agents such as dobutamine [135], levosimendan [136], milrinone [137], and even intra-aortic balloon pump counterpulsation [138] has been described and can be considered to optimise the cardiac function in order to improve CBF.

Patients with poor-grade SAH are at higher risk of cardiac and pulmonary complications [139]. Additionally, hypovolemia and pulmonary oedema are common phenomena in this population, increasing the risk for delayed cerebral ischaemia [140, 141]. Therefore, the poor-grade SAH population may benefit from advanced haemodynamic monitoring. Yoneda et al. [139], in a multicentre prospective cohort study of haemodynamic monitoring using a transpulmonary thermodilution system (PiCCO Plus), which included a group of 138 patients with poor-grade SAH, showed that extravascular lung water index (P = 0.049), pulmonary vascular permeability index (P = 0.039), and systemic vascular resistance index (P = 0.038) were significantly higher in the poor-grade group when compared with the good-grade population. Additionally, poor-grade patients displayed significantly lower cardiac index on days 1 and 2 (P = 0.027 and P = 0.011, respectively) and developed heart failure-like afterload mismatch at an early stage, and those who developed DCI had haemodynamic measures of hypovolemia, as shown by a decreased global end-diastolic volume index [139]. The same group described the mean global end-diastolic volume index (normal range, 680–800 ml/m²) as an independent factor for the development of DCI (HR 0.74, 95 % CI 0.60–0.93; P = 0.008). Patients who developed DCI had significantly lower global end-diastolic volume index compared with patients who did not (783 ± 25 ml/m² versus 870 ± 14 ml/m²; P = 0.007). A threshold of less than 822 ml/m² was correlated with DCI development, whereas a global end-diastolic volume index above 921 ml/m² was associated with the development of severe pulmonary oedema. These finding suggest that maintaining global end-diastolic volume index slightly above the normal range may be effective to prevent hypovolemia and severe pulmonary oedema, which may decrease the risk of DCI.

Pulmonary complication, such as hospital-acquired pneumonia, cardiogenic or neurogenic pulmonary oedema, aspiration pneumonitis, and pulmonary embolism, occur in approximately 30 % of patients after SAH [142]. Acute respiratory distress syndrome can affect 27 % of cases and is independently associated with worse outcomes [143]. In this clinical scenario, extra caution should be taken to avoid fluid overload; however, diuretics might be dangerous because of the risk of hypovolemia-induced cerebral ischaemia.

Hyponatremia (serum sodium of less than 135 mEq/dl) is the most common electrolyte derangement after SAH, occurring in up to 50 % of patients. There are two possible mechanisms responsible for the development of hyponatremia after SAH: (1) cerebral salt wasting (CSW) and (2) the syndrome of inappropriate secretion of antidiuretic hormone (SIADH) [144]. These entities are fundamentally different in their pathogenesis; however, they are difficult to distinguish in clinical practice and may concur in the same patient [145]. Importantly, CSW courses with intravascular volume contraction, which increases the risk of DCI and poor outcome [145]. Likewise, the treatment of SIADH on the basis of fluid restriction is not indicated, because of increased risk of hypovolemia-associated cerebral infarction [146, 147]. Therefore, in clinical practice, the management of hyponatremia in the setting of SAH is based on the avoidance of hypovolemia and the judicious repletion of volume and sodium losses [144].

In a retrospective study in a single academic centre, Wartenberg et al. found that a single occurrence of hyperglycaemia, fever, or anaemia after aneurismal SAH was independently predictive of poor outcome, even after adjustment for traditional prognostic variables, such as age, clinical grade, aneurysm size, re-bleeding, and cerebral infarction [129].

Fever is the most common medical complication after SAH and is associated with longer ICU and hospital length of stays, worse functional outcomes, and higher mortality [148, 149]. Although non-infectious fever is common, especially in the presence of intraventricular haemorrhage and poor-grade patients [150], it is strongly recommended that frequent temperature checks and careful assessment for possible infectious cause are made. During the time window of vasospasm, it is desirable to maintain normothermia with antipyretic drugs, followed by advanced fever control with surface cooling or intravascular devices [151, 152]. In this situation, especial attention should be paid to detect and treat shivering. The protocol for diagnosis and treatment of shivering has been published elsewhere [153].

Ideally, blood sugar should be kept less than 200 mg/dl and hypoglycaemia (less than 80 mg/dl) should be strictly avoided. Both have been shown in microdialysis studies to be associated with metabolic crisis and worse neurological outcome [154, 155].

Anaemia can be easily corrected, but blood transfusion has been implicated with worse outcome after SAH [156, 157], including higher mortality, after adjustment for the most common clinical indications of transfusion [158].

Although there is no clear threshold for transfusion in patients with SAH, general ICU thresholds are not applicable for this population [7, 101, 159]. Dhar et al. [160], in an elegant study using positron emission tomography scan, demonstrated that transfusion in patients with haemoglobin levels of less than 9 g/dl was the only intervention capable of increasing global CBF and oxygen delivery, when compared with crystalloid bolus and induced hypertension. The clinical applicability of these findings needs to be addressed in a large trial because the study enrolled a small number of patients (38 in total) and had only physiological endpoints.

Patients with poor-grade SAH are at high risk of venous thromboembolism [161]. Guidelines on management of SAH suggest starting mechanical prophylaxis with intermittent compression devices before aneurysm treatment [8‐10]. Pharmacologic thromboprophylaxis seems to be safe if started within 12 to 24 hours after aneurysm treatment [162].

Aneurysmal SAH is a complex neurovascular disease associated with multiple neurological and systemic complications and requires multidisciplinary specialised care, best provided in high-volume centres. Patients who survive the initial bleed can deteriorate within 2 weeks, especially because of DCI. DCI is a syndrome with a complex multifactorial pathophysiology that extends beyond the historic explanation of angiographic vasospasm.

Although the risk of death and moderate and severe disability has decreased significantly over the last three decades, the extent to which the prevalence of cognitive deficits, poor quality of life, and day­to­day functioning have changed over the same period is completely undetermined. Further studies should simultaneously tackle multiple pathophysiological pathways, and long-term functional outcomes shall be assessed by means of outcome measures able to recognise subtle cognitive changes.