Cerebral Blood Flow

In 1948, Kety and Schmidt quantified normal cerebral blood flow (CBF) in “healthy normal men” to be 54 mL/100 g per minute.¹ Sundt and others noted that a minimal CBF of 18 mL/100 g per minute is needed to maintain normal electroencephalographic (EEG) parameters during carotid endarterectomy (CEA).^2,3 A further decrease in CBF causes neuronal electrical silence and decreased synaptic activity to preserve energy stores. Irreversible cellular damage occurs when CBF is below 10 mL/100 g per minute.^4–8 In response to ischemia, the cerebral vessel autoregulatory mechanism induces vasodilation to increase collateral blood flow and thereby increase oxygen and glucose extraction for preservation of viable neurons (Fig. 345-1).

FIGURE 345-1 Cerebral blood flow thresholds for critical functions.

(Adapted from Astrup J, Symon L, Branston NM, et al. Cortical evoked potential and extracellular K⁺ and H⁺ at critical levels of brain ischemia. Stroke. 1977;8:51-57.)

Ischemic Penumbra

Laboratory and clinical evidence demonstrates a spectrum of injured neurons when CBF is interrupted as a result of vessel occlusion. In the affected territory, there is a core of irreversibly damaged neurons surrounded by an area of electrically silent but viable neurons known as the ischemic penumbra.^4,9–11 CBF in the penumbra has been determined to be between 10 and 20 mL/100 g per minute.^7,12–16 These neurons may survive for hours, although the exact duration is unknown. Many factors may reduce the time for survivability of these marginal neurons. Hyperthermia of 1°C to 2°C and serum hyperglycemia have been shown to accelerate cell death.^17,18 Restoration of CBF to the penumbra within the “critical hours” may salvage viable neurons and minimize the neurological deficits (Fig. 345-2). Without intervention, the core of the infarcted territory will subsume the penumbra as demonstrated by positron emission tomography (PET) and magnetic resonance imaging (MRI).^19–21 The temporal and spatial evolution of the infarcted core is highly variable.^19,21 Baron demonstrated by PET that the onset of the penumbra occurs as late as 16 hours after infarction.²¹ Susceptibility to ischemic injury also varies in different areas of the brain, with hippocampal CA1 and striatal neurons being more vulnerable than cortical neurons in ischemic models.^22,23

FIGURE 345-2 Relationship between time of ischemia and reversibility of neurological deficit. CBF, cerebral blood flow.

(Adapted from Jones TH, Morawetz RB, Crowell RM, et al. Thresholds of focal cerebral ischemia in awake monkeys. J Neurosurg. 1981;54:773-782.)

The traditional concept of an infarcted core surrounding by a penumbra may be too simplistic. Many patients who improve with thrombolytic agents do not demonstrate a core surrounded by a larger penumbra on imaging. The core may be surrounded by heterogeneous regions of penumbra, with possibly separate areas of minimal flow or core within the penumbra as a result of variable degrees of collateral circulation.²⁴

Energy Failure

The most immediate biochemical change in neurons affected by ischemia is diversion of the cellular machinery from aerobic to anaerobic metabolism. Insufficient oxygen inhibits the aerobic catabolism of pyruvate and promotes anaerobic glycolysis, thereby leading to accumulation of lactic acid. Diminished CBF limits lactate removal and consequently results in additional accumulation of this by-product. As pH decreases from lactic acidosis, various destructive cascades are activated and lead to cellular death.

The decrease in energy production from anaerobic metabolism (2 adenosine triphosphate [ATP] molecules) versus oxidative phosphorylation (32 ATP molecules) causes failure of the ionic electrochemical gradient established by the sodium-potassium adenosine triphosphatase (Na⁺,K⁺-ATPase) pump. Peri-infarct depolarization can be observed in the cortical penumbra with the use of direct current potential. These phenomena are associated with efflux of K⁺ and influx of Na⁺ and Ca²⁺.^25–28 Biochemical analysis of PO₂ and ATP in the penumbra confirms the depletion during depolarization,^29,30 thus placing an additional burden on compromised tissue and increasing the size of the ischemic lesion by 30% to 100%.^31–34 Ionic influx also results in oncosis,³⁵ a condition characterized by swelling of cells and organelles, increased membrane permeability, and nonspecific DNA damage. Swelling of astrocytes causes large amounts of excitatory amino acids (EAAs) to be released through volume-activated anion channels.³⁵ Although the death pathway of astrocytes is unclear, oncosis may contribute to the mechanism.

Altered Calcium Homeostasis and Excitotoxicity

Ischemia causes an increase in the intracellular calcium concentration from depolarization of voltage-gated channels, activation of ligand-gated channels, release of intracytoplasmic stores, and loss of ATP-dependent calcium extrusion pathways. In addition, there is an influx of calcium down an electrochemical gradient (Fig. 345-3).³¹ The calcium-mediated release of glutamate and EAAs from depolarized neurons plays a major role in ischemic cell death.^36–38 In the presence of energy failure, depolarization of somatodendritic and presynaptic voltage-dependent calcium channels leads to excessive extracellular glutamate and EAAs (Fig. 345-4).³⁹ When combined with failure of glutamate reuptake, toxic accumulation of extracellular EAAs causes excessive activation of the postsynaptic glutamate receptors N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoazolepropionic acid (AMPA). Overactivation of NMDA and AMPA receptors facilitates a massive influx of calcium and sodium ions, respectively.³⁹ Animal models and microdialysis during neurosurgical procedures have shown glutamate and other EAAs to be markedly elevated over baseline levels.^40–46 Accumulation of ions leads to cellular edema and peri-infarct depolarization.^47,48 Such depolarization has been recorded for 6 to 8 hours after an initial ischemic insult.³¹ The depletion of energy stores needed for restoration of normal depolarization is thought to contribute to enlargement of an infarct. In addition, excessive intracellular calcium directly and indirectly initiates a series of detrimental cytoplasmic and nuclear cascades, including activation of Ca²⁺-dependent proteases and phospholipases, alterations in plasma membrane permeability, depolarization of mitochondria and release of mitochondrial proapoptotic proteins, production of reactive oxygen species (ROSs) and reactive nitrogen species (RNSs), and activation of poly(adenosine diphosphate [ADP]-ribose) polymerase (PARP).^49–56

FIGURE 345-3 Simplified overview of pathophysiologic mechanisms in a focally ischemic brain. Energy failure leads to depolarization of neurons. Activation of specific glutamate receptors dramatically increases intracellular levels of Ca²⁺, Na⁺, and Cl⁻, whereas K⁺ is released into the extracellular space. Diffusion of glutamate (Glu) and K⁺ in the extracellular space can propagate a series of spreading waves of depolarization (peri-infarct depolarizations). Water shifts to the intracellular space via osmotic gradients, and cells swell (edema). The universal intracellular messenger Ca²⁺ overactivates numerous enzyme systems (e.g., proteases, lipases, endonucleases). Free radicals, which damage membranes (lipolysis), mitochondria, and DNA, are generated and in turn trigger caspase-mediated cell death (apoptosis). Free radicals also induce the formation of inflammatory mediators, which activate microglia and lead to the invasion of blood-borne inflammatory cells.³¹

(Used with permission from Barrow Neurological Institute.)

FIGURE 345-4 Calcium handling by cells. Calcium enters the cell through voltage- and agonist-operated channels and through nonspecific channels, such as those opened by reactive oxygen species (ROS), whereas Ca²⁺ is extruded from the cell by an adenosine triphosphate (ATP)-driven pump or by Ca²⁺-Na⁺ exchange. Cytosolic calcium (Ca²⁺) is sequestered by mitochondria and endoplasmic reticulum (ER) at the expense of respiratory energy or ATP, and it is bound to calcium-binding molecules. Ca²⁺ is released from the ER in response to agonist activation of receptors coupled to phospholipase C (PLC) and the formation of inositol 1,4,5-triphosphate (IP₃) from phosphatidylinositol 4,5-bisphosphate (PIP₂). A rise in Ca²⁺ can cause translocation of protein kinase C (PKC) to membranes where it may be activated by diglycerides (DG) formed during hydrolysis of PIP₂. Accumulation of Ca²⁺ in mitochondria can lead to the activation of a mitochondrial permeability transition (MPT) pore and result in the release of Ca²⁺ and other molecules with molecular weights of up to 1.5 kD from intramitochondrial compartments.³⁹ ADP, adenosine diphosphate; APCD, 1-aminocyclopentane-1S,3R-dicarboxylic acid; ICDH, isocitrate dehydrogenase; OGDH, oxoglutarate dehydrogenase PDH, pyruvate dehydrogenase; Quis, quisqualate.

(Used with permission from Barrow Neurological Institute.)

Free Radicals

With its high oxygen consumption, the brain is especially susceptible to free radical damage. It also has high levels of iron and ascorbate,^57–59 which can act as pro-oxidants under pathologic conditions, as well as low levels of antioxidant enzymes such as catalase, glutathione peroxidase, and superoxide dismutase.⁶⁰ The cellular membranes are rich in polyunsaturated fatty acids, which are vulnerable to radical-induced peroxidation. The acidic environment from lactic acidosis during ischemia promotes dissociation of iron from protein and enhances the conversion of superoxide anion to more reactive radicals.^58,61

ROSs and RNSs play significant roles in the pathogenesis and exacerbation of ischemic injury. It has been postulated that low levels of ROSs and RNSs trigger apoptosis and higher levels cause necrosis.^62,63 During ischemia, nitric oxide (NO), a vasodilator, is generated by endothelial nitrous oxide synthase (NOS) to increase local CBF.⁶⁴ Although this may be beneficial locally in the vasculature, a higher concentration of NO is produced by neuronal NOS and inducible NOS from inflammatory cells in the ischemic territory and may play a role in neuronal injury.^64–68 NO and its derivative peroxynitrite, in addition to oxidizing effects on cellular structures, contribute to the oxidative stress by activating PARP, thus further depleting energy stores.⁶⁹ PARP catalyzes the transfer of ADP-ribose units from nicotinamide adenine dinucleotide (NAD) to nuclear proteins. For every mole of ADP-ribose transferred, 1 mol of NAD is consumed. Four free energy equivalents of ATP are necessary to regenerate NAD.

Although ROSs are generated during ischemia, reperfusion after ischemia greatly increases ROS generation.^57,70,71 Reoxygenation provides an excess supply of substrate for oxidation, thus producing more ROSs and RNSs. This further depletes endogenous antioxidants, including radical scavengers such as tocopherol (vitamin E) and carotene and enzymes such as superoxide dismutase and catalase. ROSs and RNSs are also produced in mitochondria during ischemia and reperfusion and are involved in the release of cytochrome c.^58,72 Studies have shown that depletion of glutathione, a reductant, can cause a 100-fold increase in mitochondrial production of ROSs.^60,73 Free radicals have been implicated in signaling pathways involving dephosphorylation and translocation of the proapoptotic Bad protein from the cytosol to the mitochondrial membrane.⁶⁰ By affecting the redox state of neurons and glial cells, free radicals can alter their genetic expression, activate destructive proteins, and inhibit protective enzymes, which ultimately leads to cellular death and breakdown of the blood-brain barrier.

Free Fatty Acids

During ischemia, stimulation of glutamate receptors and an increase in intracellular calcium activate various subtypes of phospholipases, which causes a large amount of fatty acids, particularly arachidonic acid, to be released from the cellular membrane.^74–76 With free radicals, arachidonic acid is further metabolized by cyclooxygenase/lipoxygenase to produce leukotriene C₄, prostaglandin E₂, and additional free radicals. These metabolites have been implicated in neuronal demise.⁷⁷ Recent laboratory evidence suggests that arachidonic acid induces cytosolic and mitochondrial Na⁺ and Ca²⁺ overload via nonselective cation conductance, thereby leading to the formation of mitochondrial permeability transition pores, release of cytochrome c, and activation of caspase-3–dependent apoptosis.⁷⁸

Inflammatory Response

Reperfusion facilitates the arrival of inflammatory cells at the ischemic zone, and postischemic inflammation contributes to ischemic injury in several ways.⁷⁹ Inflammatory cells can aggravate ischemic injury by compromising blood flow in the microcirculation.^80–82 These cells also stimulate platelet aggregation and the formation of microthrombi, which further propagates the ischemic injury. After adhesion to the endothelial walls, granulocytes migrate (diapedesis) into the parenchyma and can be seen in peri-ischemic areas within hours of arterial occlusion.⁸³ Inflammatory cells are also responsible for the production of proteolytic enzymes and the generation of free radicals.⁸⁴ Activation of metalloproteinases and proteases destroys surrounding tissues and the extracellular matrix, independent of the energy status of neurons.^24,31 This inflammatory response is enhanced by proinflammatory chemokines such as interleukin-1β (IL-1β), IL-8, and tumor necrosis factor-α (TNF-α) and mediated by upregulation of adhesion molecules, including selectins, integrins (CD11/CD18), and the immunoglobulin supergene family.^82,85–90

Cell Death

The two processes by which injured neurons die are necrosis and apoptosis, and these processes occur in both the infarcted core and the penumbra.^53,91–95 When necrosis develops,⁹¹ cells swell in the first few hours with the formation of eosinophilic neurons. The cells then shrink and undergo pyknosis (condensation of nuclear chromatin). The number of necrotic neurons increases most markedly during the initial 6 to 12 hours. Phagocytosis of necrotic tissue takes place over a period of days. Pan-necrosis can occur in the infarcted core and give rise to a cystic cavity.²⁵ In the penumbra, infarcted neurons are removed and reactive astrocytes proliferate, thereby creating an incomplete infarct.

Ischemic apoptosis may be triggered by oxidative stress and overactivation of glutamate receptors, although how ischemia triggers the apoptotic mechanism is unclear. The process begins with shrinkage of the nucleus and cytoplasm, followed by condensation of chromatin and nuclear fragmentation. The hallmark of apoptosis is a “laddering” pattern of DNA on gel electrophoresis caused by endonuclease cleavage of DNA into segments of nonrandom length.^96,97 Eventually, the cells separate into small, pyknotic bodies that undergo phagocytosis with minimal associated inflammatory response.

Evidence has shown that the Bcl-2 family of proteins plays a significant role in determining whether ischemic neurons resist or succumb to apoptosis.^98,99 Proapoptotic Bcl-2 proteins, such as Bad and Bax, are translocated into the mitochondrial membrane, where they induce the formation of pores and release proapoptotic enzymes that lead to the activation of proteases and cell death.^100,101 The antiapoptotic proteins Bcl-2 and Bcl-xL may prevent apoptosis by inhibiting the proapoptotic proteins from inducing mitochondrial pore formation (Fig. 345-5).¹⁰² Other proteins involved in the mitochondrial pore complex include pancortin-2 and the Wiskott-Aldrich syndrome protein verprolin homologous-1 (WAVE1).^103,104 Further evidence in ischemic animal models has demonstrated that pancortin-2 sequesters Bcl-xL in association with WAVE1 to promote the translocation of Bax to mitochondria, which causes release of cytochrome c and apoptosis.¹⁰³

FIGURE 345-5 Several steps occur between the initial hypoxic-ischemic insult and frank neuronal death. Within this cell death cascade, several molecules and proteins facilitate neuronal survival and compete with factors that contribute to the cell death cascade. Ultimately, the balance between survival and death factors determines the fate of the cell. Pharmacologic approaches to stroke can be directed at facilitating the survival factors or inhibiting the cell death machinery, with the goal of preserving neurons until reperfusion and reoxygenation of the tissue can be established. DNA-PK, DNA–protein kinase; IAP, inhibitor of apoptosis; ICE, interleukin-1β–converting enzyme; NAIP, neuronal apoptosis inhibitory protein; NO, nitric oxide; PARP, poly(adenosine diphosphate ribose) polymerase; PCNA, proliferating cell nuclear antigen; PKC, protein kinase C; ROS, reactive oxygen species.

(From MacManus JP, Linnik MD. Gene expression induced by cerebral ischemia: An apoptotic perspective. J Cereb Blood Flow Metab. 1997;17:815-832. With permission from Lippincott Williams & Wilkins.)

Other degradative enzymes activated during apoptosis include endonucleases, proteases, and caspases such as IL-1β–converting enzyme. Analysis of the DNA of apoptotic cells shows the laddering of approximately 200 base pairs of DNA fragments, which is specific for apoptosis rather than necrosis. These DNA fragments, indicative of neuronal death through apoptotic mechanisms, have been observed after focal ischemia by many investigators.^93,105,106

Integration of Cellular Injury Mechanisms

Ischemic injury causes dynamic and highly interrelated biochemical changes that have variable outcomes. In the absence of collateral circulation, neuronal tissue undergoes necrosis. When the collateral circulation is able to maintain cellular integrity, these neurons can survive for a finite period but are susceptible to injury leading to various end points, including necrosis, apoptosis, and reperfusion injury (Fig. 345-6).

FIGURE 345-6 Putative cascade of damaging events in focal cerebral ischemia. Early after onset of the focal perfusion deficit, excitotoxic mechanisms can damage neurons and glia lethally. Excitotoxicity triggers many events that can contribute further to demise of the tissue. Such events include peri-infarct depolarizations and the more delayed mechanisms of inflammation and programmed cell death. The x-axis reflects evolution of the cascade over time, whereas the y-axis illustrates the impact of each element of the cascade on final outcome.

(From Dirnagl U, ladecola C, Moskowitz MA. Pathobiology of ischaemic stroke: An integrated view. Trends Neurosci. 1999;22:391-397.)

Limiting the Duration of Ischemia

The use of temporary arterial occlusion for dissection of aneurysms and permanent clipping is now a common standard during aneurysm surgery, but it is crucial to establish a safe occlusion time with emphasis on the anatomic level of occlusion and the vascular territory occluded. The duration of focal ischemia that can be tolerated safely without clinically evident sequelae varies between individuals and vascular territories.^107–111 The current consensus for temporary vessel occlusion is brief repetitive clipping periods, which provides increased safety and less risk for postoperative neurological deficit than a single episode of occlusion does, but definitive data have yet to confirm this consensus.

Laboratory studies of multiple ischemic events in focal models have demonstrated varying results.^{107,112–115} Sakaki and coworkers examined the response of three 20-minutes episodes of middle cerebral artery (MCA) occlusion in cats versus a single 1-hour period.¹¹² They noted that intermittent occlusion produced less damage than single, longer occlusion did. This report did not specify the length of reperfusion between episodes of ischemia, so it is difficult to assess whether reperfusion injury could occur. Steinberg and colleagues used a rabbit model of multiple intracranial vessel occlusion and demonstrated a 59% decrease in the area of cortical ischemic neuronal damage but no difference in the extent of striatal ischemic damage with intermittent occlusion versus uninterrupted occlusion.¹¹⁵

Goldman and associates studied repetitive MCA occlusion in rats in a protocol designed to simulate intraoperative occlusion techniques.¹⁰⁷ The total infarcted areas after 60, 90, and 120 minutes of uninterrupted occlusion were significantly greater than those after identical cumulative ischemic periods but with 5 minutes of reperfusion after each 10-minute ischemic period. In an investigation of normotensive rats undergoing MCA occlusion by Selman and colleagues, biochemical and pathologic evidence demonstrated that a single, prolonged episode of reversible ischemia was more deleterious than multiple occlusions of a similar total duration.¹¹³ Statistical differences were seen only when total single occlusion time reached 2 hours. To provide an infarct size that can be statistically evaluated in different treatments, the experimental models used to date have required total occlusion times longer than those generally needed in the clinical setting. The nature of these experimental paradigms and species differences must be kept in mind when attempting to generalize the results and apply them to clinical use.

Most reports on the use of temporary arterial occlusion in humans have been retrospective analyses of case series in which the use or nonuse of temporary occlusion was based on the experience and judgment of the surgeon.^{109,116–128} No randomized, prospective studies have been designed to compare the outcome of cerebrovascular procedures with temporary occlusion and those without local circulatory arrest. In 1961, Pool stated that bilateral anterior cerebral artery occlusion was safe for up to 20 minutes with the protective effects of hypothermia.¹²⁹ Both Suzuki’s group¹²⁷ and Ljunggren’s group¹¹⁹ estimated the maximum safe occlusion time of the MCA to be 20 minutes under normothermia. Other authors have recommended maintaining occlusion times at less than 15 minutes when possible,^{116,123,125,130} although some have reported occlusion lasting longer than 90 minutes without deficit.¹¹⁸

Samson and coworkers reviewed 121 patients from a group of 234 consecutive aneurysm patients.¹²⁵ One hundred patients who did not experience intraoperative complications were analyzed. These patients underwent elective temporary occlusion under a standard neuroanesthetic regimen, including etomidate-induced burst suppression, normotension, and normothermia. Infarctions were noted in specific arterial territories as follows: basilar, 41%; middle cerebral, 26%; internal carotid, 7%; and anterior communicating, 16%. The authors identified the following factors as significant predictors of postoperative radiographic evidence of infarction: age (≥61 years), poor preoperative grade (Hunt and Hess grades III to IV), protracted duration of temporary occlusion, and the use of incomplete circulatory arrest.¹²⁵ This series documents that temporary occlusion is not without risk; however, factors other than the duration of temporary occlusion, such as aneurysm configuration and the location of perforators, may be critical in determining this risk.¹³¹

In a nonconcurrent, prospective analysis, Ogilvy and colleagues studied the results of 132 consecutive aneurysm clipping procedures performed with temporary vascular occlusion in combination with mild hypothermia (33°C to 34°C), induced hypertension (systolic pressure of 150 mm Hg), and mannitol administration (100 g).¹²⁶ Multivariate analysis demonstrated that intraoperative rupture and duration of clipping longer than 20 minutes were independently associated with stroke outcome. The average clip application time in patients with radiographic evidence of stroke was approximately 42 minutes as compared with 29 minutes in patients without radiographic evidence of stroke, whereas in patients with a clinically significant stroke, the average time was 50 minutes.¹²⁶ The overall stroke rate in patients with an occlusion time of less than 20 minutes (1/67, 1.5%) was significantly less than that in patients with occlusion times longer than 20 minutes (12/65, 18.5%).¹²⁶

Ferch and coauthors reported on 106 aneurysm patients retrospectively analyzed with regard to age, neurological status, aneurysm characteristics, intraoperative rupture, and duration and number of occlusive episodes.¹³² The overall symptomatic stroke rate attributed to temporary clip placement in this series was 17%. The incidence of stroke was 12% in patients with occlusion times of less than 10 minutes and 35% in patients with occlusion times longer than 10 minutes. In the group of patients who did not suffer stroke, the mean temporary occlusion time was 5.4 ± 5.2 minutes, whereas in those with stroke, it was 7.4 ± 8.1 minutes. In other series, the mean temporary occlusion time of the internal carotid artery without causing a stroke was 7.1 ± 3.8 minutes, and that for the MCA was 5.3 ± 3.9 minutes.¹³²

Juvela and associates prospectively investigated temporary vessel occlusion times in 101 of 156 patients with ischemic lesions seen on computed tomography (CT) 3 months after subarachnoid hemorrhage.¹³³ Thirty-five patients who underwent temporary vessel occlusion and had an ischemic lesion on follow-up CT had a total occlusion time of 8.5 ± 8.2 minutes versus 2.5 ± 2 minutes and a longer single occlusive episode of 6.1 ± 6.2 minutes versus 2.5 ± 2 minutes, with a trend toward significance in patients with a greater number of occlusive episodes lasting 1.78 ± 0.95 minute versus 1.14 ± 0.38 minute.¹³³ Additional occlusion times of 25 to 42 minutes (mean, 33 ± 7 minutes) have been described in superficial temporal artery–MCA bypass patients without permanent neurological deficits.¹³⁴

As alluded to in earlier studies, temporary clip placement and the duration of occlusion are not the only factors influencing postoperative ischemic injury. Other factors must also be considered, including patient age (>50 years), grade and time of the initial hemorrhage, poor clinical condition (Hunt and Hess grade III to IV), occlusion of perforators, intraoperative rupture, hypertension, the specific vascular territory, multiple aneurysms, and timing of surgery.^132,133

Augmentation of Cerebral Blood Flow

Systemic arterial pressure can readily be manipulated in the setting of planned temporary ischemia. Symon and colleagues showed that MCA occlusion in primates results in loss of autoregulation in the ischemic region.¹³⁵ Elevation of blood pressure should increase cerebral perfusion because of the passive nature of the vessels that have lost autoregulation in the ischemic territory. Other investigators have shown that the decrease in CBF with focal ischemia is less in animals with phenylephrine-induced hypertension than in controls.^136–138 The beneficial effect of induced hypertension may, however, be limited to relatively brief episodes of ischemia. Smrcka and coworkers reported that hypertension reduced infarct size by 97% in rabbits subjected to 1 hour of arterial occlusion but achieved only a 45% reduction in animals with 2 hours of ischemia.¹³⁹ In the clinical setting, use of hypertension is limited by the patient’s cardiac tolerance. Close monitoring of cardiac function with limitation of the elevation in blood pressure to approximately 10% above baseline is advisable.

Prevention of Iatrogenic Ischemia (Intraoperative Cerebral Blood Flow Monitoring)

Preoperative measurement of flow rates through cerebral vessels can be achieved with noninvasive studies such as quantitative magnetic resonance angiography (a noninvasive tool also used for bypass surveillance) and xenon-enhanced CT.¹⁴⁰ These studies help the surgeon identify patients who have low cerebrovascular reserve and may be at higher risk for iatrogenic ischemia before entering the operating room.

The neurological examination provides a sensitive, qualitative assessment of CBF during cerebrovascular procedures but remains limited to CEA and by patient tolerance of an awake operation.¹⁴¹ Vessel patency after aneurysmal clip placement can be confirmed by a number of modalities, including direct microvascular Doppler (MVD) or transcranial Doppler (TCD) ultrasound, ultrasonic flow probe, intraoperative angiography, EEG studies, electrocorticography, multimodality evoked potential (MEP) and somatosensory evoked potential (SEP) monitoring, brain tissue oxygenation, and more recently, fluorescent angiography with fluorescein sodium or indocyanine green (ICG).

TCD can provide only a relative estimation of CBF based on normal ranges, and results vary according to vessel diameter and operator technique. TCD has been used extensively to monitor MCA CBF during CEA^142–144 and postoperatively for the detection of vasospasm. MVD is a relatively fast and easy method of establishing vessel patency and aneurysm filling after clipping and can be used in patients undergoing multiple aneurysm clip applications. Use of the probe is limited by vessel depth and confounded by adjacent vascular tributaries. Unfortunately, these techniques do not determine whether flow is sufficient to prevent ischemia.

Direct intraoperative flow measurements can be made with the use of a microvascular ultrasonic flow probe. The device consists of an electronic flow detection unit and a flow-sensing perivascular probe. The probe uses the principle of ultrasonic transit time to assess intravascular flow without close vessel contact.¹⁴⁵ The use of intraoperative angiography (IA) remains controversial; some reports recommend IA for complex aneurysms, such as giant aneurysms or those arising from specific locations, including the superior hypophysial, superior cerebellar, and posterior communicating arteries and the internal carotid artery bifurcation.^146–149 The argument against IA includes increased operative time, poor image quality, exposure of personnel to radiation, and a high cost-benefit ratio. Lopez and coauthors reported on a prospective cohort of 191 patients with various cerebrovascular pathologies in whom 204 angiograms were performed.¹⁵⁰ Intraoperative findings were positive in 23% of the patients (residual lesions in 12%, parent or vessel occlusion in 6%, vasospasm in 5%) and resulted in clip repositioning or additional clip placement in 8% of patients. Other studies have found that IA altered surgical treatment in 11% or 12% of cases and was associated with less than a 0.5% morbidity rate.^151,152 Whether IA should be performed is ultimately left to the cerebrovascular surgeon’s discretion. Klopfenstein and colleagues prospectively assessed the cerebrovascular surgeon’s accuracy of predicting whether IA was necessary. In this series, 200 patients with 235 aneurysms underwent IA. The results led to changes in surgical treatment in 7% of cases. In 4.4% of these patients, the surgeon thought that IA was unnecessary.¹⁵³

IA is safe and useful when treating patients with a wide variety of vascular pathologies and may be beneficial in patients for whom surgery has been prospectively deemed unnecessary by the surgeon. However, IA is limited in its evaluation of small perforating vessels, which cause 8% of iatrogenic clip infarcts.¹⁵⁴ By no means should IA be a substitute for sound surgical technique and judgment.

EEG, electrocorticographic, SEP, and MEP studies are also used to evaluate CBF. EEG monitoring is commonplace for CEA, but its usefulness is limited with barbiturate use, hypothermia, and global ischemia and in the evaluation of subcortical structures. Intraoperative neurophysiologic SEP monitoring is sensitive in detecting cortical ischemia, but MEP monitoring provides more information on subcortical structures and the motor cortex and pathways.^155,156 MEP has been found to be more sensitive than SEP monitoring in detecting insufficient blood flow in the MCA and anterior choroidal artery.^157,158 MEP monitoring is limited by the fact that the patient is fixed in a head holder and any movement can be disruptive to the surgeon.

Angiography/fluorescence imaging with ICG, a near-infrared fluorescent tricarbocyanine dye used for several decades by cardiologists, is relatively new in neurosurgical application. An intravenous bolus of 0.2 to 0.5 mg/kg (maximum daily dose, 5 mg/kg) of ICG allows the surgeon to perform real-time evaluation of blood flow within vessels in the operative field. de Oliveira and associates performed a prospective study of 60 patients with 64 aneurysms; ICG angiography demonstrated vessel patency in 62 aneurysms.¹⁵⁹ ICG angiography is safe and reliable for evaluation of perforating vessels in the surgical field but is limited in patients with deep-seated aneurysms, blood in the surgical field, complete occlusion, and clip obstruction. It is currently undergoing active evaluation in many neurovascular centers.

Complex and giant aneurysms of the skull base and distal vessels present a unique challenge. These lesions may not lend themselves to open surgical clipping or endovascular coil embolization. Even in light of recent advances in endovascular therapy (i.e., stent-assisted coiling), parent vessel ligation and aneurysm trapping, in addition to bypass, may be the only alternative. Surgical success with selective bypass begins with adequate preoperative planning and implementation in appropriate patients. Flow-assisted surgery allows direct intraoperative flow measurement and may help ensure success of the bypass.¹⁶⁰ This measurement can be quantified as a cerebral flow index (CFI): CFI = bypass flow/cut flow. With a CFI of 0.5 as a threshold value, the bypass patency rate was 92% in patients with CFIs greater than 0.5 versus 50% in patients with a CFI of less than 0.5.^145,160,161

Reduction of Metabolic Activity