Cerebral Perfusion Pressure and Delayed Cerebral Ischemia After Aneurysmal Subarachnoid Hemorrhage

The sample population consisted of patients with aSAH admitted to a regional medical center between May 1999 and October 2011 and enrolled in an ongoing National Institutes of Health study (R01NR004339). SAH was diagnosed by using computed tomography (CT), and presence of an aneurysm was diagnosed by using digital subtraction angiography. Eligible patients were 21 to 75 years old and had spontaneous rupture of an aneurysm and an original Fisher grade of 2 or greater. In addition to the eligibility criteria of the parent study, the current study included patients who had external ventricular drains (EVDs) for measurement of intracranial pressure (ICP). Exclusion criteria included preexisting chronic neurological deficit or traumatic or mycotic aneurysm. Patients were recruited within the first 5 days after aSAH and followed up for 14 days after rupture of the aneurysm or until discharge.

CPP was defined as the difference between mean arterial pressure (MAP) and ICP and was measured in millimeters of mercury. MAP was measured via an arterial catheter or a blood pressure cuff. ICP measurements were obtained by using the EVD. The transducer was leveled at the external acoustic meatus. MAP and ICP were measured every 2 hours or more frequently if a patient’s neurological or systemic status was unstable. Data were available for 14 days unless patients were admitted more than 1 day after bleeding or discharged before 14 days. Minimum CPP values were obtained within 12 hours before occurrence of DCI.

DCI was defined as neurological deterioration not attributable to rebleeding, seizures, hydrocephalus, or cerebral edema and evidence of impaired cerebral blood flow.³  DCI was diagnosed when a patient had neurological deterioration and 1 or more indications of impaired cerebral blood flow (Table 1). Indications of neurological deterioration included decrease in level of consciousness, new focal neurological deficit, deterioration in pupillary reaction, and worsening condition as indicated by the score on the National Institutes of Health Stroke Scale. Cerebral blood flow assessments within 12 hours of the observed neurological deterioration were evaluated.

Cerebral blood flow was assessed by using head CT and CT perfusion scans, cerebral angiography, and/or transcranial Doppler imaging for measurement of blood flow velocities. When patients had daily transcranial Doppler imaging, impaired cerebral blood flow was defined as a systolic velocity greater than 200 mL/s in the middle cerebral artery or a Lindegaard ratio greater than 3.0. Head CT and CT perfusion scans within the 12-hour temporal window of the neurological deterioration were reviewed for the presence of cerebral ischemia, infarction, and abnormal blood flow. Finally, cerebral angiographic findings obtained within the same temporal window were independently reviewed and evaluated for evidence of vascular narrowing, with narrowing greater than 25% indicative of clinically significant vasospasm.

Patients were excluded from analysis if their DCI status could not be determined (eg, they were comatose and sedated, and thus a decline in neurological status could not be evaluated). The variable time to DCI was defined as the number of hours from rupture of the aneurysm to the time of DCI diagnosis.

The Hunt and Hess grade was determined by using a 5-point scale to quantify the severity of non-traumatic SAH on the basis of signs and symptoms at the time of admission to the hospital.¹³  For the analysis reported here, the grade was dichotomized into poor (grade 3–5) and good (grade 1–2).

A comprehensive set of measurements that included mortality and neuropsychological, functional, and physical outcomes were assessed as part of the ongoing parent study. Neuropsychological outcomes assessed at 3 and 12 months included 7 domains: attention, learning and memory, psychomotor speed, mental flexibility, executive function, visuospatial ability, and language (Table 2). Because of concerns about sample size in each domain, neuropsychological function was dichotomized as impaired and not impaired. All test scores in all domains were converted to z scores. A z score less than or equal to −1.5 in at least 2 tests or a z score less than or equal to −2.0 in at least 1 test indicated neuropsychological impairment.¹⁴  Functional outcomes were assessed by using the Modified Rankin Scale (MRS): good, 0–2; poor, 3–6. The physical component score of the Medical Outcomes Study 36-Item Short- Form Health Survey II was used to assess physical function.15 Validity and reliability of the preceding outcome measures have been well established in patients with neurological injury.¹⁶ 

Because higher measures of depression and anxiety have been associated with poorer scores on neuropsychological tests,^17,18  controls for these variables were used in the analysis of neuropsychological outcomes. Depressive symptoms were assessed by using the Beck Depression Inventory II.¹⁹  Anxiety was measured by using the State component of the State-Trait Anxiety Inventory.²⁰  Validity and reliability have been well established for both the Beck Depression Inventory II and the State-Trait Anxiety Inventory in patients with neurological injuries.^19,21  Age, years of education, Hunt and Hess grade, and aneurysm treatment method (surgical clipping or endovascular coiling) are often associated with DCI and outcomes after aSAH^22,23  and so were included as control variables.

Data were analyzed by using IBM SPSS 19 (IBM SPSS), Mplus 6 (Muthén & Muthén), and SAS 9.2 (SAS Institute Inc) software. Descriptive statistics including means, standard deviations, and percentages were used to describe the sample. Logistic regression was used to test the relationship between CPP and DCI. The accelerated failure time model was used to test whether CPP was associated with the onset of DCI. Finally, bias-corrected bootstrapping was used to determine if DCI mediates the relationship between Hunt and Hess grade and outcomes. All comparisons were performed a priori; thus no correction for a level was made.

The 211 patients in the study had a mean age of 53 (SD, 11) years and were predominantly female (66%) and white (88%). Mean years of education was 13 (SD, 2). Approximately 62% had aneurysm coiling, 67% had a poor Hunt and Hess grade (grade 3–5), 70% had EVDs, and 42% had DCI. DCI could not be determined for 13 patients (6%) because sedation or coma precluded assessment of deterioration in neurological status. The mean value for CPP was 53 (SD, 17) mm Hg. Mean scores on the Beck Depression Inventory II were 10 (SD, 8) at 3 months and 12 (SD, 10) at 12 months. Mean scores on the State component of the State-Trait Anxiety Inventory were 47 (SD, 6) at 3 months and 45 (SD, 6) at 12 months.

Logistic regression was performed on DCI as a function of age, sex, Hunt and Hess grade, aneurysm treatment option, and CPP (Table 3). The overall model was significantly predictive of DCI: χ² (df = 5; n = 196) = 71.4; P < .001; Nagelkerke R² = 0.41. Analysis revealed a significant positive relationship between CPP and DCI; for every increase of 10 mm Hg in CPP, the odds for DCI increased by 2.78 (95% CI, 2.00–3.87). Furthermore, the mean value for CPP was significantly greater (P < .001) for patients with DCI (mean [SD], 64.6 [17] mm Hg) than for those without DCI (mean [SD], 46 [11] mm Hg).

The mean time for DCI diagnosis was 6 days (SD, 2.3 days) after bleeding. The analysis of the association between CPP and DCI was controlled for age, sex, method of repair of the aneurysm, and Hunt and Hess grade. Table 4 shows that the estimate of CPP was negative (−0.283), indicating that high CPP values are associated with shorter time to DCI (P < .001).

Data on neuropsychological outcomes were available for 60 to 77 patients. Missing data were due to time of recruitment (neuropsychological assessments were not initiated until 2003), loss to follow-up, death, and refusal to participate in the study. Approximately one-third of the patients had a poor score on the MRS at 3 months (32%) and at 12 months (30%). Impaired neuropsychological function was observed in 33% of patients at 3 months and 17% at 12 months. Mean scores on the physical component score of the Medical Outcomes Study 36-Item Short-Form Health Survey were 0 (SD, 5) at 3 months and 22 (SD, 5) at 12 months. A total of 26% of patients were dead at 3 months and 29% at 12 months. The direct effects between Hunt and Hess grade and outcomes were assessed before the mediation effect was analyzed. The mediation effect was analyzed only when the direct effect was significant. Hunt and Hess grade was significantly related to mortality and functional outcomes at 3 and 12 months but not to physical or neuropsychological function (adjustments were made for age, method of repair of the aneurysm, education, and depression and anxiety). However, DCI did not significantly mediate the relationship between Hunt and Hess grade and functional outcome or death at either 3 months or 12 months (Table 5).

The primary goal in this study was to determine whether a measure of global cerebral perfusion could explain, in part, the pathogenesis of DCI. Our findings indicate that CPP was positively associated with DCI and time to DCI. Further, the relationship between Hunt and Hess grade and outcomes was independent of DCI. To our knowledge, this study was the first to attempt to explain the relationship between Hunt and Hess grade and outcomes by using DCI as a mediating variable.

Our findings indicate that CPP was related to DCI, but the direction of the relationship was not as expected. High CPP values were associated with DCI. Most likely, high CPP values were, in part, the result of medical interventions (eg, drainage of cerebral spinal fluid and induced hypertension and hypervolemia) that most patients received prophylactically to prevent vasospasm. Because MAP and ICP (and thus CPP) are highly manipulated at the bedside, the CPP trends we observed were either a marker or a complication of treatment. Patients with higher bleeding grades (who have higher risk for complications such as vasospasm and DCI) most likely received more aggressive therapy. However, if the observed relationship between CPP and DCI were due to complications of treatment, this finding would raise several questions about induced hypertension as a prophylactic measure. Current guidelines recommend use of induced hypertension to treat or prevent DCI. However, no guidelines indicate when hypertension can be therapeutic or too aggressive or whether different treatment strategies should be used with normotensive patients compared with hypertensive patients. Future studies are needed to clarify these issues.

Bijlenga et al²⁴  reported that aSAH patients had higher CPP values during vasospasm than before vasospasm as a result of triple-H therapy (medically induced hypertension, hypervolemia, and hemodilution). Other investigators²⁵  found that CPP less than 70 mm Hg was associated with increased risk for brain tissue hypoxia and metabolic crisis. However, in those studies triple-H therapy was used to treat DCI, rather than as a prophylaxis, and thus did not affect CPP values before DCI. In our patients, induced hypertension and hypervolemia were used as prophylactic therapy to prevent vasospasm after the aneurysm was treated.

Our patients had impairment in 4 of the 7 neuropsychological domains (Table 2). The prevalence of impairment was highest in the learning and memory domain and lowest in the executive function domain. Other researchers²⁶  have reported a similar trend. Impairment (per domain) occurred in 3% to 23% of patients. A range of 14% to 61% for impairment was previously reported.^27,28  In our sample, the overall prevalence of neuropsychological impairment ranged from 33% at 3 months to 17% at 12 months. Haug et al²³  and Mayer et al²⁶  reported a prevalence of 27% to 46% at similar time points. Our patients’ mean score on the Beck Depression Inventory II indicated normal variability at 3 months and mild mood disturbances at 12 months. However, the mean scores on the State component of the State-Trait Anxiety Inventory suggested clinically significant anxiety (score > 39 indicates significant anxiety).²⁹ 

DCI did not mediate the relationship between Hunt and Hess grade and outcomes. This finding might be biased because of the missing data on neuropsychological function. However, the result suggests that the relationship between Hunt and Hess grade and poor outcomes is either direct or influenced by mediators other than DCI. Other potential mediators may include delayed neurological deficit, generalized cerebral edema, infarction, severity of initial bleeding, and early brain injury. These variables can have strong associations with poor outcomes after aSAH^28,30,31  and thus can be considered possible mediators. Unlike DCI, delayed neurological deficit incorporates many causes for neurological deterioration, such as hydrocephalus, fever, seizure, edema, and electrolyte abnormalities, with substantial influence on morbidity and mortality after aSAH.³²  Therefore, delayed neurological deficit might be an important mediator.

Our study had several limitations. All patients in the sample had EVDs, often used in patients with high-grade bleeding. Our results may therefore not be applicable to patients who do no have EVDs or who have low-grade bleeding. In addition, the incidence of DCI may have been overestimated in this sample of patients who required EVDs. We did not collect data on use of induced hypertension and hypervolemia or the intensity of these treatments. Such information might have provided further insight into our findings and might have helped us determine the effect of induced hypertension and hypervolemia on CPP. We did not use any formal assessment of cerebral blood flow, rather we used surrogates such as transcranial Doppler imaging. Last, patients’ neuropsychological function was classified as impaired or not impaired because of concerns about the size of the sample in each domain. This classification may have resulted in loss of information. Future studies with more patients will be needed to address the effect of Hunt and Hess grade on specific domains or subdomains of neuropsychological function.

Patients with DCI had higher CPP values than did patients without DCI. For every increase of 10 mm Hg in CPP, the odds for DCI increased by 2.78. Future studies will need to determine whether this relationship is causal and whether increased CPP is bad or is simply a marker for interventions to treat or prevent DCI. High CPP was associated with earlier onset of DCI. The relationship between Hunt and Hess grade and poor outcomes after aSAH was not mediated by DCI, suggesting that this relationship might be direct or due to other factors not identified in our study. Findings raise concerns about safety of induced hypertension and the need for studies that define limits for induced hypertension that are lacking in current guidelines.

This research was done at the University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania.