Relationship between Angiographic Vasospasm and Regional Hypoperfusion In Aneurysmal Subarachnoid Hemorrhage

Abstract

Background

Angiographic vasospasm frequently complicates subarachnoid hemorrhage (SAH) and has been implicated in the development of delayed cerebral ischemia. Whether large-vessel narrowing adequately accounts for the critical reductions in regional cerebral blood flow (CBF) underlying ischemia is unclear. We sought to clarify the relationship between angiographic vasospasm and regional hypoperfusion.

Methods

25 patients with aneurysmal SAH underwent cerebral catheter angiography and ¹⁵O-PET imaging within 1 day of each other (median of 7 days after SAH). Severity of vasospasm was assessed in each intracranial artery while cerebral blood flow (CBF) and oxygen extraction fraction (OEF) were measured in 28 brain regions distributed across these vascular territories. We analyzed the association between vasospasm and perfusion and compared frequency of hypoperfusion (CBF < 25 ml/100g/min) and oligemia (low oxygen delivery with OEF ≥ 0.5) in territories with vs. without significant vasospasm.

Results

24% of 652 brain regions were supplied by vessels with significant vasospasm. CBF was lower in such regions (38.6±12 vs. 48.7±16 ml/100g/min) while OEF was higher (0.48±0.19 vs. 0.37±0.14, both p<0.001). Hypoperfusion was seen in 46 regions (7%) but 66% of these were supplied by vessels with no significant vasospasm; 24% occurred in patients without angiographic vasospasm. Similarly, oligemia occurred more frequently outside territories with vasospasm.

Conclusions

Angiographic vasospasm is associated with reductions in cerebral perfusion. However, regional hypoperfusion and oligemia frequently occurred in territories and patients without vasospasm. Other factors in addition to large-vessel narrowing must contribute to critical reductions in perfusion.

Introduction

Patients with subarachnoid hemorrhage (SAH) may develop focal and/or global neurological deficits in a delayed fashion, which have been attributed to underlying cerebral ischemia.¹ When cerebral blood flow (CBF) is insufficient to provide adequate oxygen delivery (DO₂) to brain tissues, cellular metabolism fails and, if substrate delivery is not promptly restored, ischemic cell death ensues. One of the most powerful predictors of poor neurological outcome after SAH is the development of cerebral infarction, the end-result of such critical regional ischemia.²

Classically delayed cerebral ischemia (DCI) has been attributed to anatomic narrowing of proximal cerebral arteries, a process which frequently complicates the course of patients with SAH.³ Such vasospasm is most accurately assessed by catheter angiography and may be severe enough to delay intracranial circulation.⁴ This proposed relationship between vasospasm and ischemia forms the rationale for monitoring and treating vasospasm in order to prevent ischemic deficits and infarction.⁵ However, the causal link between this common vascular abnormality and tissue hypoperfusion has not been clearly established. Infarcts may be seen in the absence of significant proximal vasospasm and treatments that significantly reduce angiographic vasospasm have not consistently reduced ischemic brain injury.^6,7 Other factors including disturbed autoregulation, microvascular thrombi or spasm, and cortical spreading depression may alternatively explain reductions in CBF and contribute to DCI.^8,9

We employed positron emission tomography (PET) to assess the relationship between regional hypoperfusion and angiographic vasospasm. Specifically, we wanted to determine how frequently cerebral hypoperfusion and oligemia occur independently of proximal vasospasm.

Methods

Patient selection

We selected patients from our PET research database who met the following criteria: 1) aneurysmal SAH; 2) PET performed during period of risk for DCI and vasospasm (days 4–14 after SAH); 3) cerebral catheter angiography within 1 day of PET. We included patients both with and without vasospasm and/or ischemic deficits. The Human Research Protection Office and Radioactive Drugs Research Committee of Washington University School of Medicine approved all PET protocols. Informed consent was obtained from each patient or their legally authorized surrogate.

Clinical Management

All patients underwent baseline catheter angiography and ruptured aneurysms were treated within 24 hours of admission. All received enteral nimodipine and were maintained in a euvolemic state by adjustment of intravenous fluids to keep ins and outs balanced; prophylactic hypervolemia/ hypertensive therapy was not employed. New or worsening neurological deficits were promptly evaluated, and if no alternative cause was identified, patients underwent cerebral angiography. In the absence of intervening symptoms, all patients underwent cerebral angiography on or around day 7 after SAH. Hemodynamic augmentation (primarily induced hypertension) was initiated in cases of presumed ischemic neurological deficits pending angiographic confirmation of vasospasm (although PET was performed prior to institution of induced hypertension in a number of cases, as part of a research protocol).

Data Collection and Angiographic Assessment

Clinical data was recorded at the time of admission and on the day of PET, including Glasgow Coma Scale (GCS) and World Federation of Neurosurgical Societies’ (WFNS) scores.¹⁰ Admission CTs were graded for amount of subarachnoid blood using the Fisher scale.¹¹ Cerebral angiograms performed within 24 hours of PET were retrospectively analyzed by a single trained investigator blind to PET data.

The diameter of each large intracranial artery (i.e. distal ICA, vertebral, basilar, and proximal segments of MCA [M₁], ACA [A₁], and PCA [P₁]) was measured and quantitative percent stenosis was calculated in comparison to its diameter on baseline angiograms. Distal segments of the ACA, MCA, and PCA were qualitatively assigned a vasospasm severity of none, mild, moderate, or severe based on visual inspection (see Supplemental Figure 1). Test-retest reliability for this grading scheme was assessed using kappa statistic. We noted the presence of fetal PCA circulation (i.e. hypoplastic P₁ segment with dominant posterior communicating artery) and cases with significant cross-flow via the anterior communicating artery (AComm). For three patients whose angiograms were no longer available, coding was carried out by reviewing the neuroradiologist’s interpretation of vasospasm severity in each vessel. For those patients who underwent endovascular treatment of vasospasm (e.g. angioplasty) and had PET after angiography (n=4), the post-intervention (not baseline) vessel measurements were used to correlate with PET findings.

Severity of vasospasm was categorized in each vessel based on established cutoffs for both proximal and/or distal involvement (see Table 1).^12,13 Significant vasospasm was considered present if proximal stenosis was at least 50% or if severity of distal vasospasm was moderate or severe. Patients with at least one vessel with significant vasospasm were considered to be affected by angiographic vasospasm for patient-level analyses.

Table 1.

Classification of vasospasm severity based on degree of arterial narrowing along with frequency of regions affected

Severity of Vasospasm	Degree of Narrowing	Frequency
None	≤ 5%	246 (38%)
Mild	6–33%	226 (34%)
Moderate	34–66%	108 (17%)
Severe	> 66%	72 (11%)
Significant Vasospasm	Degree of Narrowing	Frequency
No	< 50%	495 (76%)
Yes	≥ 50% †	157 (24%)

^†or if distal vasospasm was moderate-severe

PET Methods

All PET studies were performed on either the Siemens/CTI ECAT EXACT HR 47 or HR+ scanners located in the NNICU using 15-oxygen labeled radiotracers.^14,15 Patients were studied while on maintenance fluids and any ongoing therapies including vasopressors (although a number of studies were performed in patients with suspected ischemic deficits prior to initiation of vasopressors). Image acquisition was performed as detailed previously to measure CBF and, in the more recent studies, oxygen extraction fraction (OEF) and cerebral metabolic rate for oxygen (CMRO₂).16 Physiologic data were recorded at the time of each scan including central venous pressure (CVP) and intracranial pressure (ICP), when available. Arterial blood was analyzed for hemoglobin and arterial oxygen content (CaO₂).

All PET scans for each patient were co-registered and aligned to the initial baseline CBF study using Automated Image Registration software.¹⁷ They were calibrated for conversion of PET counts to quantitative radiotracer concentrations, as previously described.¹⁸ Radioactivity in arterial blood was measured using an automated blood sampler. The arterial time-radioactivity curve recorded by the sampler was corrected for delay and dispersion using previously determined parameters. Images were then co-registered to a reference brain image and resampled into Talairach atlas space. Global values for CBF, OEF and CMRO₂ were obtained using a standard image mask covering the brain from below the superior sagittal sinus down to the level of the pineal gland. DO₂ was calculated as the product of CBF and CaO₂.

Regional Analysis

Spherical regions of 10-mm diameter were placed in 14 predefined locations distributed across the major supratentorial vascular territories bilaterally (28 total per patient, 3 in each ACA, 8 in each MCA, and 3 in each PCA territory, outlined in Supplemental Figure 2).¹⁹ Regions containing hematoma, infarcted tissue, or within the ventricular system on concurrent CT images were excluded. Regional values for CBF, OEF, and CMRO₂ were then determined within each of the remaining spheres.

Regions were classified as being hypoperfused when CBF was below 25 ml/100g/min. A threshold for low DO₂ was set at 4.5 ml/100g/min (equivalent to a CBF of 25 at low-normal CaO₂ of 18 ml/dl).²⁰ For patients in whom OEF was measured, we also determined which regions had OEF greater than or equal to 0.5 (as evidence of increased extraction compensating for insufficient oxygen delivery). Oligemia, a state of particularly vulnerable hemodynamic impairment, was considered present when DO₂ was low and OEF was elevated. Thresholds utilized are conservative estimates guided by data from normal controls,^21,22 and previous PET studies of subjects with SAH, both with and without vasospasm.^23,24

Association of Vasospasm with Brain Regions

Each brain region was assigned a grade of vasospasm based on the maximal severity of angiographic narrowing in any of the arteries supplying the vascular territory in which it was located. Pattern of vasospasm was further categorized into proximal only (e.g. ICA and/or proximal ACA or MCA for ACA or MCA regions), distal only (when vasospasm was moderate-severe in distal segments supplying that territory, without proximal narrowing), or both proximal and distal vasospasm. In the presence of a fetal PCA circulation, PCA regions were classified not on basilar or proximal PCA vasospasm, but on ICA involvement. We determined whether cross-flow across the AComm could attenuate effect of proximal ACA stenosis (i.e. if there was no distal ACA vasospasm).

Statistical Analysis

Physiologic and PET measurements were compared between patients and brain regions with vs. without significant vasospasm, using student’s t-test. Chi-squared tests were used to compare the proportion of regions with and without vasospasm that exhibited PET evidence of hemodynamic impairment (low CBF and oligemia). Perfusion to regions in each vascular territory was compared in those with proximal, distal, combined, or no vasospasm using ANOVA. We additionally examined the impact that vasospasm had on perfusion depending on whether adjacent vascular territories were also affected, an estimation of impaired pial collateral circulation. We also performed these analyses limiting our sample only to patients with vasospasm (i.e. those who had both affected and unaffected regions) to further examine the within-patient effects of vasospasm on perfusion. We further excluded patients being treated with vasopressors at the time of PET to avoid hemodynamic augmentation confounding the assessment of hypoperfusion.

Results

Of 38 SAH patients studied with PET, 25 had angiography within 24 hours of PET and were included in this analysis (Table 2). PET was performed a median of 7 days after SAH and a median of 5 hours from angiography. 14 patients had significant vasospasm in at least one vessel on catheter angiography; of the 12 patients with presumed ischemic neurological deficits, only 3 were being treated with hemodynamic augmentation at the time of PET. The other nine subjects were symptomatic but studied before institution of induced hypertension. Baseline physiologic data at the time of PET is shown in Supplemental Table 1. MAP was not significantly higher in those with vasospasm (113±17 vs. 106±11, p=0.23), and even this slight difference disappeared after excluding patients on vasopressors. Hemoglobin was higher in those with vasospasm (10.6±1.9 vs. 9.3±1.7, p=0.09) resulting in a higher CaO₂ (14.2±2.4 vs. 12.2±2.2, p=0.05). Median ICP was 8 mm Hg (range 6–20); no patient had elevated ICP at the time of PET. Global CBF tended to be lower in patients with vasospasm (Table 3) but this difference was not statistically significant, even after excluding patients on vasopressors. Global DO₂ was comparable as a result of the higher hemoglobin in patients with vasospasm.

Table 2.

Demographics and clinical characteristics

	Frequency (n/%)
Number of subjects	25
Age, years (mean±SD)	57 ± 11
Sex, female	18 (72%)
Race, white	20 (80%)
GCS (admit), median/range	13 (3–15)
Poor Grade (WFNS IV–V)	10 (40%)
Fisher grade 3	23 (92%)
Clip / Coil	15 / 10
PET study on day (mean±SD)	7 ± 3
Angiographic vasospasm	14 (56%)
Ischemic neurologic deficits	12 (48%)
On pressors at time of PET GCS (day of study), median/range	3 (12%) 13 (7–15)
Length of Stay, days (mean±SD)	23.6 ± 8.7
Discharge Disposition (Home/Rehab/LTC/Death)	5/17/3/0

LTC = long-term care

Table 3.

Global PET measurements in patients with and without angiographic vasospasm

	Overall mean±SD	Angiographic Vasospasm (n=14)	No Vasospasm (n=11)
CBF	44.3±14.5	38.4±9.7 †	46.9±12.4
DO2	5.7±1.7	5.4±1.4	5.7±1.7
OEF (n=12)	0.36 ± 0.13	0.40±0.17	0.33±0.11
CMRO2 (n=12)	2.2±0.7	2.1±0.7	2.0±0.4

^†p=0.07

CBF, DO₂, and CMRO₂ are expressed in ml/100g/min

48 of 700 total regions were excluded for being within infarcts, hematoma, or an enlarged ventricular system, leaving 652 for analysis. The distribution of vasospasm severity is outlined in Table 1. 157 regions (24%) were affected by significant (proximal and/or distal) vasospasm across 34 vascular territories. Distal vasospasm of at least moderate severity, without proximal involvement, was present in 14 arterial territories supplying 75 brain regions. This compared with 82 regions with proximal involvement (including 51 regions across 14 territories with both proximal and distal vasospasm). Kappa statistic for grading of distal vasospasm severity (as moderate/severe vs. none/mild) was 0.91. There were 4 cases of fetal supply of PCA regions from the ICA circulation; in none of these cases was there significant ICA vasospasm that would affect coding of PCA regions. Four patients had significant AComm cross-flow, all associated with significant A₁ vasospasm. Only one of these did not have distal ACA vasospasm as well, meaning that for all except the 2 distal ACA regions in that single patient, AComm flow did not alter coding of regional ACA vasospasm.

Mean CBF was similar in regions with moderate as compared to severe angiographic vasospasm, but significantly lower in such regions than those with none or only mild vasospasm. Regions with significant (≥ 50% proximal or moderate/severe distal) vasospasm had mean CBF of 38.6±11.7 ml/100g/min compared to 48.7±16.5 in regions without significant vasospasm (p<0.001). DO₂ was similarly reduced (5.3±1.6 vs. 6.3±2.3, p<0.001) and OEF was elevated (0.48±0.19 vs. 0.37±0.14, p<0.001) in affected regions. Perfusion was reduced in regions with only proximal vasospasm (39.3±12.8 ml/100g/min), only distal vasospasm (39.3±11.4) or both proximal and distal narrowing (37.3±11.6, p<0.05 for post-hoc comparison of each category to regions with no vasospasm).

Regions with Hypoperfusion

Regions with low CBF were found in 10 patients including 7 of 14 with and 3 of 11 without significant vasospasm. Nine of the 10 patients with hypoperfusion had neurological symptoms prior to or at the time of PET (none were on vasopressors), including 2 of the 3 with deficits but no vasospasm on angiography. A total of 46 regions (7%) had low CBF whereas low DO₂ was evident in 159 regions (24%) across 20 patients (including 8 of those without any vasospasm and 8 of whom were asymptomatic).

Analysing perfusion in the 14 patients who had angiographic vasospasm, 157 (42%) of 374 brain regions were located in territories with significant vasospasm. Such regions had lower CBF (38.6±11.7 vs. 45.4±16.1, p<0.001) and higher OEF (0.48±0.19 vs. 0.38±0.16, p=0.002) than those without vasospasm. Despite lower CBF on average across regions with vasospasm, Figure 1 provides a per-patient representation of regional CBF values in regions with and without vasospasm, highlighting the predominant overlap in perfusion between affected and unaffected territories within a given patient. There was a trend to lower perfusion in regions with vasospasm when the adjacent vascular territory was also affected by significant vasospasm (37.5±10.7 vs. 40.8±13.3, p=0.12).

Figure 1.

Boxplot of CBF in brain regions supplied by vessels with and without vasospasm in 14 affected subjects.

Hypoperfusion (CBF < 25) did not occur more frequently in territories with vasospasm (15/157 regions, 10% vs. 20/217, 9% of unaffected regions) in these patients, meaning that a majority (57%) of vulnerable regions were located outside territories with vasospasm. In fact, 54% of hypoperfused regions occurred in territories with no vasospasm anywhere in the ipsilateral carotid (or, if appropriate, vertebrobasilar) circulation. Distribution of hypoperfusion across each of the vascular territories in relationship to presence and location of vasospasm is shown in Table 4. Hypoperfusion was as frequently found in the absence of any vasospasm as in territories with proximal and/or distal vasospasm, regardless of vascular territory. Involvement of adjacent vascular territories did not predict a higher risk of hypoperfusion in those territories with vasospasm. Hypoperfusion additionally occurred in 11 (4%) brain regions in 3 patients without any vasospasm, meaning that overall 31 of 46 (67%) hypoperfused regions were not in the territory of any affected vessels (Supplemental Figure 3).

Table 4.

Distribution of hypoperfusion across vascular territories in relation to presence and location of angiographic vasospasm

Territory	Incidence of Hypoperfusion (overall)	No Vasospasm	Proximal Vasospasm	Distal Vasospasm	Both Proximal & Distal
ACA regions	12/68 (18%)	6/28 (21%)	1/7 (14%)	2/15 (13%)	3/18 (17%)
# of Territories		12	3	5	8
MCA regions	16/222 (7%)	8/120 (7%)	2/24 (8%)	4/54 (7%)	2/24 (8%)
# of Territories		15	3	7	3
PCA regions	7/84 (8%)	6/69 (9%)	0/0	0/6	1/9 (11%)
# of Territories		23	0	2	3
All Territories	35/374 (9%)	20/217 (9%)	3/31 (10%)	6/75 (8%)	6/51 (12%)

Note: this only includes the 14 patients with vasospasm in at least one vessel

Oligemia was present in 6 of 12 patients with OEF data, equally divided between patients with and without vasospasm, for a total of 24 (of 313 regions, 8%). Oligemia was non-significantly more likely to occur in territories with vasospasm (10/60, 17%) compared to 6/75 (8%) in normal vascular territories, (p=0.12), in patients with vasospasm. However, like hypoperfusion, a majority of oligemic regions occurred in the absence of vasospasm when looking across all patients.

If the 3 patients on vasopressors were excluded from this analysis, the majority of hypoperfused regions were still seen outside territories with vasospasm in the remaining untreated patients. In fact, the gradation of CBF was more obvious in separating degrees of vasospasm (Figure 2), with a significant difference in CBF not only between severe (30.8±8.6) and mild vasospasm (46.7±17, p<0.001) but also a trend to lower CBF with severe as compared to moderate (38.1±12.6, p=0.07) vasospasm.

Figure 2.

Comparison of mean regional CBF between territories with no, mild, moderate, and severe vasospasm in untreated patients (error bars represent ± 2 standard errors of the mean)Table 1: Classification of vasospasm severity based on degree of arterial narrowing along with frequency of regions affected.

Discussion

Vasospasm has been associated with reductions in CBF in a number of previous studies as early as the 1970’s.²⁵ Most have suggested that moderately severe narrowing is necessary to reduce distal perfusion.²⁶ Yet this association is not clear or necessarily causative. Even severe angiographic vasospasm is not always associated with ischemic deficits or infarction.²⁷ Elevated TCD velocities consistent with vasospasm were frequently seen without evidence of hypoperfusion in a previous PET study.²⁸ Infarcts may occur in the absence of vasospasm or in territories where no proximal narrowing is seen.^29,30 A study using CT perfusion found that 15% of those without vasospasm (on CT angiography) still had areas of hypoperfusion.³¹ We found that 3 of 11 patients without any significant vasospasm (on catheter angiography) still had regions with low CBF while 8 of 11 had regions with low levels of oxygen delivery. No previous studies have examined the correlation between angiographic abnormalities and matched regional tissue hypoperfusion. Most have investigated the relationship between vasospasm and global perfusion or CBF averaged over an entire affected vascular territory.^28,31,32 However, ischemia is a focal/multifocal process with infarcts often being patchy and scattered.^33,34 By examining multiple regions throughout the various vascular territories (both affected and unaffected by vasospasm, in symptomatic and asymptomatic patients), we were able to sample the spectrum of tissue at risk and determine the relationship between vasospasm and regional hypoperfusion in this SAH cohort. Furthermore, we used conventional catheter angiography, the gold-standard for evaluation of vasospasm, rather than TCD or CT angiography as utilized in most previous studies.^27,28,31,35 We restricted our analysis to those patients who had PET and angiography within a day of one another (median of 6 hours apart), allowing meaningful contemporaneous comparisons to be drawn without excessive interference from intervening confounders. Our use of PET allowed us to not only measure CBF in multiple stereotactic brain locations but also assess level of oxygen extraction and metabolism (including oligemia), which may better reflect risk of ischemia.

We found that although flow was lower in patients and within territories affected by moderate-severe vasospasm (whether proximal or distal), there was a poor relationship between arterial narrowing and matched regional hypoperfusion. A minority of regions within territories with significant vasospasm had low CBF or DO₂, and both hypoperfusion and oligemia were found as commonly in territories (and patients) unaffected by vasospasm. Even when limiting our analysis only to patients with vasospasm, we still found no consistent association between territories with vasospasm and hypoperfusion. This held even when excluding the 3 patients on vasopressors at the time of PET, whose inclusion could cloud this relationship.

CBF and DO₂ reflect tissue perfusion and delivery of oxygen to the microcirculation. In contrast, vasospasm is an anatomic phenomenon that does not directly translate into cerebral physiology. Large-vessel narrowing may not have a major impact on circulation to brain tissue in the presence of autoregulatory vasodilatation of distal arterioles or in the presence of collateral flow. Our study confirms this dissociation and reminds us to not assume vasospasm equates with hypoperfusion or, equally importantly, that the absence of angiographic abnormalities precludes the presence of tissue ischemia. Basing management decisions on angiographic findings alone may misestimate the true risk of ischemia.

Treatment strategies that are clearly effective in reducing vasospasm have not translated into a reduction in cerebral infarction or improved outcome.^7,13 Our findings suggest that other factors must be responsible, at least in part, for impairing tissue perfusion. These may include microvascular spasm, thrombosis, or the increasingly appreciated phenomenon of cortical spreading depression.^9,36 Focusing on radiographic vasospasm alone risks missing areas of hypoperfusion in territories and patients without this radiographic abnormality who are still at risk for ischemia and cerebral infarction. Measurements of tissue perfusion may better evaluate DCI to improve outcome for patients with SAH.

Limitations

Such a correlative study requires precise comparisons between measures of vasospasm and perfusion. We employed modalities that comprise the gold standard for both such measurements. We used PET to provide quantitative regional CBF data on a number of regions throughout the cerebral vascular territories. However, it is possible that we could have missed small foci of hypoperfusion in an affected territory. We evaluated each angiogram retrospectively (blinded to PET data) and quantitatively assigned a degree of narrowing to each vessel. This rigorous approach formed the basis for categorization of vasospasm severity using standard cutoffs. However, the two studies were not performed concurrently but could be as much as 24 hours apart (although over half were done within 6 hours of each other). As perfusion and vasospasm are potentially dynamic processes, using measurements separated in time could introduce some misestimation into our analysis.

Hypoperfusion was also found in only a minority of brain regions, with low CBF occurring in only 7% of regions. This is likely related to our proactive management of SAH patients with fluids and permissive hypertension, minimizing risk of active ischemia. Even if the few patients on vasopressors might mask the effect of severe vasospasm, after excluding these patients, the relationships remained unaltered (i.e. hypoperfusion still existed as commonly outside of regions with vasospasm). We also analyzed a number of patients who had neurological deficits and underwent PET before hypertensive therapy was initiated, allowing an unfiltered perspective on the relationship between untreated vasospasm and perfusion abnormalities. We did not have data on the precise localization of neurological deficits exhibited in these patients, but we did document hypoperfusion in 2 patients who were symptomatic but did not have any angiographic vasospasm. Further, the poor association between territories with vasospasm and regions with hypoperfusion did not seem to be explained by presence/absence of collaterals, as estimated in this study by involvement of adjacent vascular territories which would compromise pial collateral circulation, or flow across AComm in cases of proximal ACA vasospasm. However, a more refined quantitative measurement of collateral circulation (not possible in this retrospective study) might better correct for this important factor.

Conclusions

While more severe cerebral vasospasm is associated with reductions in regional CBF, regional hypoperfusion and oligemia are found as often in absence of significant arterial narrowing. Large-vessel narrowing does not likely explain the spectrum of ischemia seen after SAH; other factors must contribute to critical reductions in oxygen delivery underlying DCI and cerebral infarction. Measurement of CBF and/or adequacy of oxygen delivery may be more useful than evaluation of vasospasm alone in determining risk of ischemia and managing patients with SAH.