Article Text

other Versions

Original research
Delayed infarction following aneurysmal subarachnoid hemorrhage: Can the role of severe angiographic vasospasm really be dismissed?
  1. Alex M Mortimer,
  2. Brendan Steinfort,
  3. Ken Faulder,
  4. Timothy Harrington
  1. Department of Radiology, Royal North Shore Hospital, Sydney, New South Wales, Australia
  1. Correspondence to Dr Alex Mark Mortimer, Department of Radiology, Royal North Shore Hospital, Reserve Road, St Leonards, Sydney, NSW 2065, Australia; alex_mortimer{at}


Background The recent literature pertaining to delayed cerebral ischemia following aneurysmal subarachnoid hemorrhage has downplayed the role of angiographic vasospasm. However, it is our hypothesis that angiographic vasospasm has a significant pathophysiological role in this disease. We undertook an observational radiographic study of patients who presented in a delayed manner (>72 h from ictus) with evidence of severe angiographic vasospasm on initial angiography in order to describe an apparent association between vasospasm and infarct location.

Methods This was a retrospective study of consecutive patients treated at our unit. Initial, subsequent, and follow-up cross-sectional imaging with CT or MRI was analyzed in conjunction with initial angiography. Sites of angiographic narrowing, angiographic hypoperfusion, and subsequent sites of infarction were assessed.

Results Thirteen patients (6 women, 7 men) of mean age 49 years were assessed. Mean time to presentation was 6 days. All had severe angiographic vasospasm. Nine of the 13 patients suffered infarction; the infarcts in seven of the nine patients were large. There was correlation between sites of angiographic narrowing and infarction in all cases and eight of the nine cases showed angiographic hypoperfusion in a location corresponding to eventual infarct location.

Conclusions Severe angiographic vasospasm may be linked to infarction in patients who present late. These infarcts are mostly large despite maximal treatment. We question the notion that proximal vasospasm has a minor role in delayed ischemia.

  • Aneurysm
  • Angiography
  • Subarachnoid
  • Stroke
  • Angioplasty

Statistics from


Cerebral infarction is strongly associated with poor outcome following subarachnoid hemorrhage (SAH).1 ,2 There are multiple potential mechanisms for infarction in this setting that broadly include the acute injury,3 complications of aneurysm treatment, or the delayed ischemic process. There is considerable controversy regarding mechanisms of delayed infarction following aneurysmal SAH and the role that angiographic vasospasm (aVSP) plays in this process.4–7 Although for decades aVSP has been assumed by many to represent the primary mechanism for delayed infarction,8 several recent reviews have suggested that proximal vasospasm has a relatively minor role in delayed ischemia7 or doubt its contribution.4 The principal reasons for this notion are that (1) delayed infarction may occur in patients without proximal vasospasm;9 ,10 (2) pharmacological agents that reduce rates of moderate to severe vasospasm have not been shown to improve outcomes;11 ,12 and (3) one of the few treatments with proven efficacy in this setting (eg, oral nimodipine) has not been shown to improve aVSP.13 ,14 An alternative mechanism of ischemia—cortical spreading depression—has also been shown to result in delayed infarction in the absence of proximal vasospasm,15 and this has reinforced the trend in the literature to downplay the role of aVSP.

Contrary to the above findings, there is a strong correlation between the severity of aVSP and the incidence of cerebral infarction, supporting its role in the pathogenesis of delayed infarction. Radiological studies suggest that the majority of delayed infarcts occur in association with aVSP,16 ,17 and that the majority of aVSP-related infarcts are associated with severe aVSP,16 ,18 ,19 which results in the most severe perfusion deficits.20–22 We hypothesize, as have many, that severe aVSP has a significant role in delayed infarction. We felt it would be important to describe our experience in those patients who present late with evidence of aVSP on initial angiography who may be at greater risk of delayed ischemia23–25—a cohort in whom the natural history of the disease could most closely be reproduced. We aimed to identify the rate and characteristics of infarction and correlation between the location of vasospasm, hypoperfused regions seen on digital subtraction angiography (DSA), and eventual regions of infarction demonstrated on follow-up CT or MRI.


This was a retrospective observational study of consecutive patients treated between November 2009 and March 2014. All patients treated during this period were identified from a departmental database. Each patient's angiographic imaging was reviewed for the presence of vasospasm on initial imaging. After subsequent case note review, patients were then included in the study if they presented more than 72 h after the initial hemorrhage and had severe arterial narrowing on the initial DSA, as defined below. Patients were excluded from the study if the SAH was non-aneurysmal in etiology or they were aged <18 years. The aVSP was managed using an intensive treatment paradigm involving transluminal balloon angiolasty (TBA) and/or repeated intra-arterial verapamil, as previously documented.26

The presenting CT scan, initial and subsequent angiography, CT scans performed during the course of the patient's treatment, and CT or MRI scans performed 2–6 weeks following vasospasm treatment were assessed by a neuroradiologist blinded to the clinical outcome. The initial CT was interpreted for distribution of hemorrhage. Fisher grade and maximal clot thickness were recorded in line with previously published protocols.27 ,28 Presenting CT scans were also interpreted for evidence of any intraventricular hemorrhage and intraparenchymal extension of hemorrhage and for the presence of hydrocephalus. The distribution of vasospasm was recorded from the initial DSA. Severe vasospasm was defined on DSA as internal carotid artery (ICA), basilar artery, M1 and M2 middle cerebral artery (MCA), A1 and A2 anterior cerebral artery (ACA) and P1 posterior cerebral artery (PCA) vessel diameters of ≤2, ≤1.5, ≤1.5, ≤1, ≤1, ≤0.75 and ≤1 mm, respectively with the interpretation of more distal vasospasm based on a subjective assessment of focal arterial narrowing (percentage narrowing as traditionally used could not be applied as baseline imaging was not available). Angiographic hypoperfusion was defined as a region that failed to opacify or showed marked delay in opacification during the parenchymal phase (figure 1).

Figure 1

Example case demonstrating mode of angiographic interpretation. Lateral arterial phase right internal carotid artery digital subtraction angiography (DSA) demonstrating severe M2 middle cerebral artery vasospasm (left, arrow). The parenchymal phase DSA shows a region of right parietal hypoperfusion (center left, arrowheads) with delayed filling of pial collateral vessels. This correlates with the infarct location seen on the follow-up CT imaging (right).

CT or MRI scans performed approximately 24 h after initial angiography were interpreted in conjunction with the angiographic and subsequent cross-sectional imaging to record the rate and characteristics of aVSP-related infarction. This was defined as hypodensity in a vascular territory or of a watershed distribution that correlated with the distribution of angiographic spasm and angiographic parenchymal phase hypoperfusion (see figure 1). Infarcts were defined as large if they occupied at least 50% of an M2 MCA territory, M1 perforator infarction involving the entire striatum, the A1 ACA or P1 PCA territories. Hypodensities relating to external ventricular drainage or hydrocephalus, those lying adjacent to hematomas or relating to aneurysm treatment were excluded. Clinical outcome was graded prospectively in the clinic by means of a modified Rankin Scale (mRS) at mean post-treatment follow-up of 6 months.


Thirteen patients fulfilled the inclusion criteria. Descriptive statistics are shown in table 1. The mean time from ictus to presentation was 6±1 days. Six of the 13 patients (46.15%) were poor grade (WFNS 4–5). Eleven of the 13 patients had Fisher grade 3 blood distribution on the presenting CT scan. One patient had early ischemic change on CT at presentation; 10/13 (76.92%) had evidence of hydrocephalus. Seven of the 13 responsible aneurysms (53.85%) were large (≥7 mm) and 12 (92.31%) were anterior circulation. The majority of patients were treated with coiling (10/13; 76.92%).

Table 1

Baseline characteristics of the study cohort

The distribution of severe aVSP is shown in table 2. Nine of the 13 patients suffered delayed infarction. Seven of the nine patients had large infarcts. Co-analysis of angiograms and sequential cross-sectional imaging showed that vasospasm distribution, angiographic hypoperfusion, and eventual infarct location correlated in eight of the nine cases (88.89%) with subsequent infarction. Pictorial demonstrations of these are shown in figures 28. In eight of the nine cases, evidence of ischemia was demonstrated on the 24 h follow-up CT scan. One patient demonstrated watershed ischemia on follow-up MRI at 3 months. The patient who did not show regional hypoperfusion on parenchymal phase imaging suffered a perforator infarct involving the lentiform nucleus but the associated MCA was severely spastic (figure 7). Each patient had intensive endovascular treatment in conjunction with maximal medical therapy in order to limit the hemodynamic impact of the aVSP and prevent infarct growth. Ten of the 13 patients underwent proximal vessel TBA and each territory had vasodilator therapy for a mean of 5.6 days. The mean±SD dose of verapamil per territory at each treatment was 14.5±2.34 mg. Six of the 13 patients (46.15%) were mRS 0–2 on clinical follow-up. Two patients had moderate functional follow-up scores (mRS 3) and two patients died; death was secondary to perioperative hemorrhage in one patient and uncontrollable intracranial hypertension in another.

Table 2

Aneurysm location, vasospasm distribution, and infarct location

Figure 2

Case 2. Frontal and lateral early arterial right internal carotid artery (ICA) digital subtraction angiography (DSA) demonstrating severe right ICA, M1, M2 middle cerebral artery and anterior cerebral artery vasospasm (left and center left, arrows). Parenchymal phase lateral DSA demonstrating right parietal hypoperfusion (center right, arrowheads). Parasagittal follow-up CT imaging (right) demonstrating right parietal infarction.

Figure 3

Case 3. Frontal and lateral early arterial left internal carotid artery digital subtraction angiography (DSA) demonstrating severe left A1 anterior cerebral artery and M1 and M2 middle cerebral artery vasospasm (left and center left, arrows). Parenchymal phase lateral DSA demonstrating left frontoparietal hypoperfusion (center, arrowheads). Parasagittal follow-up CT imaging (center right, right) demonstrating patchy left frontoparietal infarction on the background of diffuse oedema.

Figure 4

Case 6. Frontal and lateral right internal carotid artery (ICA) digital subtraction angiography (DSA) demonstrating severe right ICA, M1, and M2 middle cerebral artery vasospasm (left and center left, arrows). Parenchymal phase lateral DSA demonstrating right temporoparietal and frontal hypoperfusion (center right, arowheads). Parasagittal follow-up CT imaging (right) demonstrating right temporoparietal and frontal infarction.

Figure 5

Case 7. Oblique frontal right internal carotid artery (ICA) digital subtraction angiography (DSA) and frontal left ICA DSA demonstrating severe bilateral anterior cerebral artery vasospasm (left and center left, arrows). Parenchymal phase frontal left ICA DSA demonstrating left medial frontal hypoperfusion (center right, arrowheads). Axial follow-up CT imaging (right) demonstrating bilateral frontal infarction.

Figure 6

Case 8. Frontal left internal carotid artery (ICA) early arterial digital subtraction angiography (DSA) (left) and frontal left ICA late arterial DSA (center left) demonstrating severe ICA, A1 and middle cerebral artery (MCA) vasospasm (arrows) and sluggish flow in the distal anterior cerebral artery and MCA territories. Parenchymal phase frontal left ICA DSA demonstrating left parasagittal hypoperfusion (center right, arrowheads). Axial follow-up T2-weighted MRI (right) demonstrating left parasagittal deep white matter infarction.

Figure 7

Case 9. Frontal left internal carotid artery (ICA) digital subtraction angiography (left) demonstrating severe ICA, A1 anterior cerebral artery and M1 middle cerebral artery vasospasm. Follow-up CT imaging (right) demonstrating large left caudate nucleus and lentiform nucleus infarct.

Figure 8

Case 13. Frontal and lateral early arterial left internal carotid artery digital subtraction angiography (DSA) (left and center left, arrows) demonstrating severe left M2 middle cerebral artery vasospasm. Parenchymal phase lateral DSA demonstrating left paracentral hypoperfusion (center, arrowheads). Axial and parasagittal follow-up CT imaging (center right and right) demonstrating left paracentral infarction.


This analysis highlights the apparent role of cerebral vasospasm in the delayed ischemic process, particularly in this subgroup of patients who present in a delayed manner. We recognized correlation between the site of aVSP, angiographic regional hypoperfusion, and eventual infarction in eight of nine patients who suffered infarction after presenting in a delayed manner (mean 6 days) with severe aVSP. The infarcts were large in seven of the nine patients who suffered infarction.

The role of proximal vasospasm in delayed ischemia is controversial.4 ,7 It has been suggested that aVSP may not have a dominant role in delayed ischemia for a number of reasons. First, the incidence of aVSP is 70%, but clinically detected delayed ischemia is approximately half of this.4 ,9 It is true that some patients with evidence of vasospasm are clinically asymptomatic or do not show overt clinical deterioration. This may be because moderate degrees of vasospasm that are not necessarily hemodynamically significant are included in analyses. In many studies, non-invasive measures of vasospasm with limited accuracy, sensitivity and specificity have also been used. Furthermore, current measures of vasospasm severity often do not incorporate collateral flow or perfusion information or the eloquence of ischemic parenchyma, perhaps best assessed through combining angiographic and perfusion imaging. Moreover, patients have variable hemodynamic reserve and respond differently to flow limitation; clues to this can be gleaned from the ischemic stroke literature. Hakimelahi et al29 have recently reported that patients with ICA or proximal MCA occlusions demonstrated a wide range in diffusion-weighted infarct volumes within 30 h of occlusive stroke onset; there was no correlation between infarct volume and time from stroke onset. The observations suggest that highly variable cerebral perfusion via the collateral circulation may primarily determine infarct growth dynamics, and this has implications for patients with significant vasospasm.

It is also commonly suggested that vasospasm has a limited role in delayed ischemia as treatments that reduce the rate of moderate to severe vasospasm (eg, clazosentan) fail to result in an improvement in outcome,11 ,12 whereas nimodipine has been shown to improve outcome without an impact on aVSP.13 ,14 There are many potential reasons for a failure in correlation between reductions in the incidence of vasospasm but not in rates of poor outcome in the clazosentan trials. First, poor grade patients who suffer a more significant acute neurological injury are those who are more likely to suffer the most severe vasospasm and therefore detecting differences with current outcome measures would probably require very large trials or alternative outcome measures. Second, only severe vasospasm may result in significant reductions in cerebral blood flow so as to result in infarction and the inclusion of those with moderate vasospasm may dilute the treatment effect. Third, vasospasm is likely to exert its clinical impact through infarction: factors such as age30 govern the impact of infarct size on functional outcome. Furthermore, the eloquence of the infarct area will result in variable functional impact. Fourth, the systemic side effects of the treatment may have resulted in poor outcome through alternative mechanisms (medical complications) in the treatment arm. Lastly, rescue treatments such as endovascular intervention and systemic hypertension may have been effective in preventing a poor outcome in patients in the placebo arm. In this study, for example, despite the high rate of infarction, 46% of patients had a favorable outcome.

An additional intriguing possibility for the lack of correlation between occurrence of vasospasm and delayed ischemia is that an alternative mechanism of delayed ischemia exists that may occur alone or in combination with proximal aVSP. It has also been demonstrated that infarction can occur independently of severe cerebral vasospasm,9 and spreading depolarization with cortical ischemia may represent one mechanism for this. Woitzik et al15 have demonstrated in a series of 13 patients that the number of spreading depolarizations in patients suffering delayed ischemic neurological deficits was significantly higher than in those without, when proximal vasospasm was minimized through treatment with nicardipine beads. However, just as vasospasm does not necessarily result in infarction, this process occurred without subsequent infarction in 6 of 10 patients on follow-up CT or MRI and the infarcts that were demonstrated in this series were small. Spreading depolarization with vasoconstriction also occurs in the ischemic penumbra of stroke patients,31 raising the possibility that this process could be triggered or compounded by the sustained ischemia of proximal vasospasm with the two mechanisms acting as a ‘dual hit’.32 Of note, the incidence of spreading depolarizations is similar in patients with and without proximal vasospasm,15 ,33 suggesting that the spreading depolarization is not triggered by proximal vasospasm. However, when spreading ischemia was demonstrated in the absence of severe proximal vasospasm, infarcts were small.15 In contrast, we have shown (and it is well established) that patients can develop large infarcts in the presence of proximal vasospasm. Is it possible that perfusion deficits due to proximal vasospasm could increase the severity of spreading ischemia and expand infarcts? Experimental evidence suggests that spreading ischemia is intensified by a decline in cerebral perfusion pressure, while increasing perfusion pressure can reverse the spreading ischemic process to an almost normal spreading hyperemic response to depolarizations.34 ,35

An association between aVSP in the development of delayed infarction is supported by the results of radiological studies that suggest that the majority of delayed infarcts occur in association with aVSP,16 ,17 and that there is a correlation between aVSP severity and the incidence of infarction. The majority of aVSP-related infarcts are associated with severe aVSP,16 ,18 ,19 which results in the most severe perfusion deficits.20–22 Of patients with severe aVSP, 50–100% develop cerebral infarction compared with 3–5% of patients without significant aVSP.16 ,18 ,19 Severe aVSP is associated with poor cognition, worse patient-relevant outcomes, and greater inpatient healthcare resource use.36 It is also an independent predictor of poor outcome.9 However, if there are at least two processes—both macrovascular and microvascular—it is possible that more moderate degrees of proximal aVSP may become more relevant in the context of an additional microvascular process.

Nogueira et al37 have shown that moderate angiographic cerebral vasospasm can be associated with CT perfusion hypoperfusion. Furthermore, in some cases in their series, suboptimal angiographic response to intra-arterial nicardipine still resulted in CT perfusion normalization, suggesting a microvascular component to CT perfusion hypoperfusion. A recent meta-analysis demonstrated a 23-fold higher incidence of delayed cerebral ischemia in patients with CT perfusion abnormalities.38 This analysis did not correlate CT perfusion with cerebral angiography, but it could be that CT perfusion hypoperfusion is an estimate of both macrovascular and microvascular hypoperfusion. In one study,39 if there was macrovascular vasospasm with luminal narrowing ≥50%, there was a high likelihood (83%) of perfusion abnormality in the territory of the vasospastic vessel. Perfusion abnormalities without macrovascular vasospasm were also identified in the watershed areas or in the vicinity of sulcal clots. It is therefore possible that CT perfusion may identify those patients who might benefit most from endovascular treatment. CT perfusion could also be used as a surrogate for clinical deterioration since these patients are often poor grade, comatose or sedated, and difficult to assess. They may suffer silent infarction that itself has a negative impact on outcome.40 ,41 The main drawback is radiation exposure, but a CT angiography and perfusion approach appears to be cost effective compared with transcranial Doppler.42

If it is assumed that both spreading ischemia and proximal aVSP act in concert, perhaps with the latter augmenting the former, the rationale for treatments that increase proximal luminal diameter including endovascular treatments may therefore be to improve cerebral perfusion and limit the impact of a microvascular process such as spreading ischemia. This may be most relevant in instances where medical measures employed to improve flow (ie, therapeutic hypertension) cannot overcome severe proximal stenoses and where collateral flow is limited. TBA is effective in sustainably improving vessel diameter, but it is limited to use in the proximal vessels and suffers from procedural risks of vessel rupture and thromboembolism.43 Vasodilators are lower risk and can treat more distal vasospasm but suffer from a short duration of action that often necessitates multiple treatments. In a recent survey of US centers,44 the calcium channel antagonist verapamil was the most commonly used intra-arterial (IA) vasodilator. More recently, a number of other agents have been employed including IA milronone,45 a potent selective phosphodiesterase III inhibitor with inotropic, vasodilatory and anti-inflammatory effects, or IA fasudil hydrochloride,46 an inhibitor of myosin light chain kinase and vascular smooth muscle contraction. The effect of these agents on both macrovascular and microvascular processes certainly warrants further investigation.

This case series is limited in terms of its size and retrospective nature and is descriptive only. We relied on a subjective assessment of perfusion based on angiographic imaging which is not equivalent to CT perfusion-identified hypoperfusion, and it is uncertain how this corresponds to CT perfusion hypoperfusion. It was not possible to quantify or obtain a statistical analysis of the impact of severe vasospasm as this was a single-arm study. We did not investigate the rate of infarction in the absence of severe vasospasm but, in line with other studies,16 ,18 ,19 we believe that the incidence of this is low as we employ a screening program for vasospasm and monitor and treat delayed ischemia aggressively. The deleterious effect of proximal vasospasm in terms of ischemia and infarction may be greater in this population of late presenters,23–25 ,47 owing to the fact that most patients presenting in a timely manner will have had their aneurysm secured earlier and closer blood pressure monitoring and support in order to maintain cerebral perfusion pressure and negate the early hemodynamic effect of vasospasm and/or spreading ischemia.


Severe aVSP appears to be linked to regional perfusion deficits that correspond to eventual infarction by location. This indicates that proximal vasospasm is a significant factor in the etiology of delayed cerebral infarction. We suggest that the role of proximal vasospasm in delayed cerebral ischemia should not be underestimated, and the interaction of proximal vasospasm with other mechanisms of delayed ischemia is an area that warrants continued research.


View Abstract


  • Contributors All authors contributed to the manuscript and study.

  • Competing interests None declared.

  • Ethics approval Ethics approval was obtained from the Northern Sydney and Central Coast ethics committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All parties interested in data sharing may contact AMM.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.