Advances in the understanding of delayed cerebral ischaemia after aneurysmal subarachnoid haemorrhage

Introduction

Aneurysmal subarachnoid haemorrhage (aSAH) has an incidence of 6–11 per 100,000 people per year and accounts for only 5% of all strokes^1–4. Despite this, aSAH is the cause of one third of all stroke-related years of potential life lost before the age of 65⁵. Approximately 70% of all people with aSAH will either die or require help with activities of daily living at six months after the initial injury⁵. The mean age of onset of aSAH is 55 years and, when combined with its poor morbidity and mortality, it causes an enormous socioeconomic burden^6,7. The significant morbidity attached to aSAH can be attributed to rebleeding, delayed cerebral ischaemia (DCI), hydrocephalus, and other medical complications, despite successful treatment of the ruptured aneurysm. Of these complications, DCI is the most important cause of morbidity and mortality in patients who survive the ruptured aneurysm^5,8,9. Between days 3 and 10 after the initial aSAH, 30–40% of patients will suffer DCI and half of these will have a poor outcome^5,10,11.

Our understanding of DCI is meagre at best. Conventionally, DCI was thought of as a neurological deficit observed at least three days after aSAH with radiological confirmation of large vessel narrowing and was often termed “vasospasm”. However, more contemporary articles question whether the relationship between angiographic cerebral vessel narrowing and neurological outcome is associative rather than causative and have highlighted the possibility of a multifactorial aetiology^12–15. One of the problems with the disease and research surrounding DCI is the terminology applied. Terms include DCI, delayed ischaemic neurological deficit (DIND), delayed neurological deficit, secondary cerebral ischaemia, and vasospasm. In 2010, a consensus statement was issued defining DCI as a focal neurological impairment or decrease of ≥2 points on the Glasgow Coma Scale which lasts for ≥1 hour, is not apparent immediately after aneurysm occlusion, and cannot be attributed to other causes by means of clinical assessment, blood tests, or imaging¹⁶. The Neurocritical Care Society’s consensus definition was similar for DCI and also defined vasospasm as radiological evidence of cerebral vessel narrowing with corresponding neurology¹⁷.

Cerebral artery narrowing

Over six decades ago, cerebral vessel narrowing was demonstrated by angiography after aSAH¹⁸. A decade later, a link was found between cerebral vessel narrowing and focal neurology¹⁹. Then in the late 1970s, it appeared that vessel narrowing was not only localised to the vascular territory of the aneurysmal bleed but also proportional to blood load and occurred between days 3 and 12 after the aSAH^20,21. More contemporary authors found the onset of vessel narrowing started on day 3, was maximal by days 6–10, and lasted for up to two weeks^22–24. The density, duration and volume of subarachnoid blood are key predictors of vessel narrowing^21,25. Narrowing of cerebral arteries may cause a reduction in cerebral blood flow distal to the constricted vessel and contribute to secondary ischaemia²⁶. The cause of vessel narrowing after aSAH is unclear but is thought to involve oxyhaemoglobin release, an inflammatory-mediated response, decreased nitric oxide levels, and an increased concentration of endothelin-1 (ET-1)¹⁴.

Oxyhaemoglobin

Oxyhaemaglobin induces cerebral artery vasoconstriction in vitro and in vivo in primates, which is not seen with methaemoglobin or bilirubin^27–29. It is thought that oxyhaemoglobin decreases the production of prostacyclin and increases prostaglandin E2 in vessel walls, thereby causing vasoconstriction. It can also inhibit endothelial-dependent relaxation. The oxidation of oxyhaemoglobin to methaemoglobin, which occurs spontaneously, causes lipid peroxidation and vasoconstriction³⁰. It is plausible that oxyhaemoglobin causes vasoconstriction by some or all of these mechanisms but attempts at modulating them have not completely reversed vessel narrowing or, importantly, improved outcomes.

Nitric oxide

Nitric oxide, which is responsible for the relaxation of vascular smooth muscle cells, appears to be depleted after aSAH. This may be due to a number of reasons, one of which is that nitric oxide is scavenged by haemoglobin, released during the breakdown of subarachnoid blood, due to nitric oxide’s high affinity for haemoglobin^31,32. In addition to this, the production of nitric oxide may also be decreased due to the down-regulation of endothelial and neuronal nitric oxide synthase, which occurs in spastic arteries after aSAH^33–35. Both of these mechanisms will lead to a decrease in the bioavailability of nitric oxide, which is then unable to counteract the effects of the vasoconstrictor ET-1³⁶. Furthermore, exogenous donors of nitric oxide, such as sodium nitroprusside and nitroglycerin, although associated with systemic side effects, have been shown to ameliorate cerebral artery narrowing^37,38. In addition to the hypotension seen with these exogenous donors, there is also a concern that exposing nitric oxide to oxyhaemoglobin and deoxyhaemoglobin will lead to the formation of methaemoglobin, S-nitrosohaemoglobin and ferrous-nitrosyl-haemoglobin³³. Interestingly, Kida et al. note in their comprehensive review that inhaled nitric oxide acts as a selective pulmonary vasodilator and avoids the hypotension seen with intravenous administration. Animal studies have demonstrated a reduction in ischaemia-reperfusion injuries after nitric oxide inhalation in extrapulmonary organs after cardiac injury. These have also been supported by proof-of-concept human trials³⁹. The research discussed is used to support post-cardiac arrest ischaemia but Garry et al. also encourage further investigation of nitric oxide as a treatment of secondary brain injury in their review with reference to aSAH⁴⁰.

Endothelin

Endothelin is key to maintaining the vascular tone of blood vessels, with ET-1 being the most potent endogenous activator of vasoconstriction. The amount of ET-1 in serum and plasma increases within minutes after the aSAH and peaks around days 3–4, the time at which DCI starts to occur. There also appears to be an excessive release of ET-1 by astrocytes around the time of onset of ischaemic symptoms^41,42. ET-1 concentrations appear consistently elevated in patients with DCI. However, there are conflicting reports of ET-1 concentrations within the normal range in patients with radiological evidence of cerebral artery narrowing who do not have DCI^43–45. Authors have questioned whether increased ET-1 marks ischaemic damage rather than arterial vessel narrowing in DCI¹⁴. Therefore, there are a number of different mechanisms that could be contributing to the arterial narrowing commonly seen after aSAH.

Alpha calcitonin gene-related peptide

Alpha calcitonin gene-related peptide (CGRP) is an endogenous neuropeptide and a potent vasodilator. CGRP exhibits its vasodilating properties by two mechanisms: one is nitric oxide and endothelium-dependent and the other is cyclic adenosine monophosphate mediated and is endothelium-independent⁴⁶. Endogenous CGRP appears to be released, and is subsequently depleted, after aSAH to combat cerebral vasoconstriction which has led to the theory that exogenous CGRP may be beneficial in managing DCI^47–49. Because CGRP can act independently of endothelial cells, which are morphologically damaged after aSAH, it may be successful in treating DCI. A number of animal studies and three human trials have investigated the effect of CGRP on cerebral arteries after aSAH. All animal studies appear to show either a reversal or improvement in cerebral artery narrowing⁴⁶. The largest human trial, the European CGRP in aSAH study, demonstrated little improvement in morbidity or mortality from intravenous administration but noted that systemic side effects, such as hypotension, were limiting and suggested that intrathecal administration may be more beneficial, as endogenous CGRP acts on the abluminal side of vessel walls⁵⁰. A trial investigating the effect of CGRP after intrathecal administration is still awaited.

Radiological evidence

An often-cited argument against cerebral vasoconstriction being a causative factor of DCI is that, whilst up to 70% of patients demonstrate cerebral vessel narrowing on angiography, only 40% of these will manifest neurological deficits and only 30% develop DCI^51–54. However, it must be acknowledged that even the consensus definition of DCI provided in the introduction has its limitations¹⁶. Patients with poor grade aSAH (World Federation of Neurosurgical Societies Grades IV and V), the group of patients most likely to develop DCI, are often sedated and mechanically ventilated and are particularly difficult to assess clinically⁵⁵. Therefore, it is likely that we are under-diagnosing and under-treating DCI in this group of patients. Furthermore, it may be that the degree of large cerebral vessel narrowing does not correlate well with symptom severity²⁶.

Following a review of current tests available for the diagnosis of delayed cerebral ischaemia, Rodriguez et al. advise clinical examination and transcranial Doppler (TCD) in the screening and diagnosis of “vasospasm”. The authors reserve multi-modal magnetic resonance imaging (MRI) and computed tomography (CT) for specific situations, and acknowledge digital subtraction angiography (DSA) as the gold standard for diagnosis (Figure 1)⁵⁶. Rabinstein et al. found that TCD and angiogram demonstrating cerebral vessel narrowing (termed vasospasm) only had a positive predictive value of 67% for cerebral infarction on CT⁸. We would expect this to be higher if cerebral vessel narrowing was the primary cause of DCI. Rates of cerebral infarction in patients with evidence of cerebral vessel narrowing range between 24 and 35% using CT^57,58, but have been found to be as high as 81% in some studies using MRI⁵⁹. In addition to this poor correlation between cerebral vessel narrowing and infarction, there is clinical evidence that up to 25% of delayed infarcts on CT are not in the same territory as the vessel narrowing, or are found in patients that did not demonstrate vessel narrowing at all^60–62. Rabinstein et al. note that TCD and angiogram only agreed on the diagnosis of “vasospasm” in 73% of cases and so it could be that vessel narrowing simply wasn’t identified in patients who were later found to have evidence of infarcts on CT⁸. Despite these conflicting messages, clinical studies do report that those patients with radiological evidence of cerebral vessel narrowing are at greater risk of DCI^62,63.

Figure 1. CT image of subarachnoid haemorrhage.

Non-contrast CT scan of brain showing subarachnoid haemorrhage in classical “star sign” distribution with blood distributed along basal vessels.

Herz et al. directly visualised pial artery constriction after application of blood or microtrauma to pial arteries in animal studies⁶⁴. Further in vitro research has suggested that constriction of intraparenchymal arterioles occurs after aSAH and may contribute to DCI⁶⁵. Maximal luminal narrowing has been seen between days 3 and 7 and repeated in vivo in mouse studies. The correlation between decreased regional cerebral blood flow and microvascular constriction appears stronger than that seen with large vessel narrowing^65–67. Uhl et al. identified constriction of small vessels in surgical patients within the first 72 hours after aSAH by spectral imaging, and Pennings et al. later directly observed cerebral arterioles contracting after aSAH^68,69. Therefore, it may be that vessel narrowing is consistently occurring with DCI but that we are not visualising it because it is microvascular and not readily visible on catheter angiography or TCD⁵⁶.

CT perfusion scanning (CTP) may provide haemodynamic evidence to support the diagnosis of DCI. Dankbaar et al. evaluated the diagnostic value of CTP for DCI and reported 84% sensitivity, 79% specificity, and 88% positive predictive values⁷⁰. Sanelli et al. found that more CTP deficits occurred in patients with DCI than in those without⁷¹. Dankbaar et al. later suggested that patients with DCI exhibit worse cerebral perfusion (measured on CTP) than patients without DCI even before focal signs occurred. Encouragingly, they demonstrated partial recovery in areas of poor perfusion, suggesting that DCI could be partly reversible⁷². However, Killeen et al. concluded from their retrospective comparative study that CTP and DSA had similar test characteristics for identifying DCI in aSAH patients⁷³.

Endothelin-antagonists

A shift in theory from cerebral vessel narrowing to a multifactorial aetiology occurred after the CONSCIOUS trials and a recent meta-analysis of pharmacological treatments for delayed cerebral ischaemia^74–76. The meta-analysis demonstrated that, despite a reduction of cerebral vessel narrowing, no statistically significant effect on poor outcome was observed⁷⁴. However, the authors note that the dissociation between a reduction in cerebral vessel narrowing but not poor outcomes could result from methodological problems, sample size, and insensitivity of outcome measures, in addition to a multifactorial aetiology of DCI. The CONSCIOUS trials were multicentre randomised controlled trials (RCT) investigating the effect of clazosentan, an endothelin-A (ET-A) antagonist, on “vasospasm” after aSAH. The first of these trials, CONSCIOUS-1, demonstrated that, despite a significant reduction in angiographic cerebral vessel narrowing, there was little evidence to support its use to improve morbidity and mortality and it was associated with increased rates of pulmonary complications, hypotension and anaemia⁷⁶. CONSCIOUS-2 demonstrated no benefit from clazosentan in patients treated with surgical clipping, which led to the early termination of the trial⁷⁵. Laban et al. recently published a review of animal studies investigating endothelin receptor antagonists after experimental aSAH and found no improvement in functional outcomes⁷⁷. Perhaps more importantly, the review described insufficient animal data supporting endothelin receptor antagonists to warrant progression to a human trial. The authors also suggest that cerebral artery diameter, or “vasospasm”, is not a clinically relevant outcome measure in experimental aSAH studies⁷⁷.

The example of clazosentan appears to provide evidence that cerebral artery narrowing is not the sole cause of DCI. However, there is conflicting evidence as more invasive methods of reducing vessel narrowing can improve outcomes (Figure 2). Kimball et al. reviewed 49 articles relating to interventional techniques to treat “vasospasm”. A total of 24 of the 27 publications (1,028 patients) reporting the use of transluminal balloon angioplasty noted an improvement in vessel diameter and neurological deficits. Twelve case series reported good angiographic and clinical results for patients who received papaverine (a vasodilator) administered approximate to the site of vessel narrowing⁷⁸. Both techniques were associated with significant side effects and the quality of the studies was reported as very low to moderate (based upon the GRADE classification system)⁷⁹. Nevertheless, the review does provide evidence that cerebral artery narrowing is likely to be strongly involved in the pathology of DCI.

Figure 2. Angiograms demonstrating cerebral vessel narrowing after subarachnoid haemorrhage.

A and B: Anteroposterior (A) and lateral (B) angiograms of the left internal carotid artery demonstrate vessel narrowing at the level of the carotid siphon, the terminal internal carotid artery, the A1 segment of the anterior cerebral artery and the middle cerebral artery. C and D: Anteroposterior (C) and lateral (D) angiograms obtained after intra-arterial injection of nimodipine.

Nimodopine

The calcium channel antagonist nimodipine is the only proven intervention to reduce the incidence of DCI and improve outcomes after aSAH. Nimodopine was initially investigated as a vasodilator in the hope that it would aid post-ischaemic reperfusion, as it was thought that an increase in calcium in vascular smooth muscle cells led to “vasospasm”^80,81. In 1989, the British Aneurysm Nimodipine Trial subsequently demonstrated a significant reduction in cerebral infarction rates and improved neurological outcomes at three months after aSAH⁸². A Cochrane review in 2007 concurred with these findings but noted that the supporting evidence was based mainly on one large study. This led to oral nimodipine becoming standard care for patients after aSAH⁸³. Interestingly, the review found no statistically significant results to support the use of other calcium antagonists, magnesium sulphate, or intravenous administration of nimodipine.

Magnesium sulphate is a non-competitive inhibitor of calcium channels and has vasodilatory and neuroprotective properties, similar to nimodipine. Hypomagnesaemia is common in patients after aSAH, appears to be proportional to the severity of the bleed, and is predictive of DCI⁸⁴. Magnesium sulphate has also been shown to reduce cerebral artery narrowing and the size of ischaemic lesions after aSAH in animal models⁸⁵. However, the Neurocritical Care Society guidelines advise against the routine administration of magnesium in patients with aSAH¹⁷. This is supported by data from the intravenous magnesium sulphate for aneurysmal subarachnoid haemorrhage (IMASH) and MASH-2 trials and a recent meta-analysis demonstrating no beneficial effect of magnesium in this group of patients^86–88. A post hoc analysis of the IMASH trial reported an association between high plasma levels of magnesium and worse clinical outcomes⁸⁹.

In summary, one calcium channel antagonist, nimodipine, has been shown to be effective in the prevention and treatment of DCI after aSAH whilst other calcium channel antagonists and a non-competitive inhibitor of calcium channels appear to have little effect on, or worsen outcomes.

It remains unclear how nimodipine exerts its neuroprotective effects but its action seems independent of any effect on large vessel narrowing^90,91. It was thought that nimodipine may exert its effect by stopping calcium influx at a neuronal level, but no beneficial effect has been seen from administration in patients after ischaemic stroke or traumatic brain injury^92–94. In addition to this, a recent systematic review found no benefit from nimodipine after traumatic SAH, suggesting that the mechanism of action of nimodipine is unique to aSAH⁹⁵. Nimodipine has two properties that it does not share with other calcium channel antagonists. Firstly, it increases endogenous fibrinolytic activity, which may reduce the incidence of microthrombosis⁹⁶. Secondly, it antagonises cortical spreading ischaemia in rats, which may be one of the culprits in DCI and is discussed in further detail below⁹⁷.

Contemporary hypotheses

Therapies

Conclusion

In summary, cerebral vessel narrowing is consistently seen after aSAH, but its location and severity is not predictably linked to DCI. There is no conclusive evidence to support the treatment of vessel narrowing in the management of DCI, despite some studies reporting improved outcomes, specifically after more invasive techniques. Nimodipine is the only effective treatment for DCI but we still do not understand how nimodopine exerts its neuroprotective effect, although it does not seem to work by reversing cerebral artery narrowing, at least not in large vessels. It is possible that we are not detecting microvascular vasoconstriction or ischaemia on CT and TCD and so our understanding of the pathology is limited. Furthermore, improved recognition of DCI clinically, from imaging and/or biochemical markers would not only allow for quicker treatment but also better assessment of interventions. DCI almost certainly has a multifactorial aetiology and it may be that only by combining interventions will we see improved outcomes, but first we must understand the aetiology. Understanding how nimodipine, the only drug with proven efficacy, exerts its effect may be the key to creating new interventions with improved efficacy. There remains a large amount of work to be done in understanding DCI and investigating future potential treatments.