Brain Protection

The literature on cerebral ischemia and brain protection is vast, and a detailed discourse on this topic is beyond the scope of the present discussion. A number of excellent recent reviews on the subject are available.^{[461] [463] [467] [468] [469] [470] [471]} The numerous brain protection strategies that have been studied in both the laboratory and the clinic can be classified according to the phase of this sequence that each addresses. They are listed in Table 21-5 . The focus of the present discussion will be anesthetic-mediated neuroprotection.

Considerations Relevant to Complete Global Ischemia (Cardiac Arrest) (also see Chapter 78 )

Maintenance of adequate perfusion pressure after cardiac arrest is of considerable importance. Hypotension developing after resuscitation from cardiac arrest may aggravate

838

Both barbiturates and calcium channel blockers have been administered after cardiac arrest. The former

839

Induced mild hypothermia has recently been shown to be effective in reducing mortality and morbidity in patients who sustain a cardiac arrest.^[482] Induction of mild hypothermia in the range of 32°C to 34°C for a period of about 24 hours improved neurologic outcome and survival 6 months after cardiac arrest when compared with the normothermic group. Mild hypothermia was induced without difficulty. Passive rewarming of patients was accomplished slowly over a period of 8 hours. The incidence of complications was similar to that in the control normothermic group. This important study is one of the first to demonstrate the feasibility and efficacy of induced hypothermia as a treatment to prevent injury from global ischemia. Undoubtedly, induced hypothermia will now be added to the armamentarium for the treatment of cerebral complications of cardiac arrest.

Considerations Relevant to Focal (Incomplete) Ischemia

Protection by Anesthetics

Before discussing individual drugs, it should be noted that a general theme that can be extracted from the protection-by-anesthetics literature, in particular for volatile anesthetics, is that anesthesia per se is protective. It appears that for undefined reasons, reducing the level of systemic stress associated with a standardized experimental insult results in an improved outcome.^{[483] [484]} In reviewing the protection-by-anesthetics literature, readers should be conscious of the possibility that the protective benefit ascribed to an intervention with an anesthetic may in fact be the product of exaggeration of the injury in a high-stress control state (e.g., N₂ O "sedation").

BARBITURATES.

The protective efficacy of barbiturates in focal cerebral ischemia has been reported numerous times in animals,^{[485] [486] [487] [488] [489] [490] [491]} with only a single demonstration of effectiveness in humans.^[492] The effect has been attributed principally to suppression of CMR. However, CBF redistribution effects and free radical scavenging^[493] have been suggested to contribute, and there is evidence that CMR suppression is not the sole mechanism.^{[494] [495] [496]} Suppression of CMR might logically be expected to be of benefit to brain regions in which oxygen delivery was inadequate to meet normal demands but was sufficient to allow energy consumption by some ongoing electrophysiologic activity (i.e., in which the EEG was abnormal but not flat). Such regions are likely to be limited in size in the setting of focal ischemia, yet several of the animal investigations suggest a very substantial protective effect.^{[485] [487] [488] [489]} Review of these experiments reveals that the methods used to monitor and maintain temperature, though accepted at the time, were below the standards that have evolved from more recent understanding of the effects of both deliberate^{[497] [498]} and inadvertent^[499] hypothermia. Unrecognized cerebral hypothermia may well have been a factor in some of the cited investigations, and it is therefore possible that the protective efficacy of barbiturates may have been overestimated. Although more recent publications involving suitable temperature control methods do in fact indicate a protective effect by barbiturates, ^{[496] [500] [501]} the magnitude of that effect was modest in comparison to the results of earlier studies. Barbiturate-induced EEG suppression in an already anesthetized patient may still be logical therapy when it can be applied before or early in the course of a period of temporary focal ischemia, such as temporary occlusion during aneurysm surgery. However, the judgment to institute such therapy should be made after consideration of the risk of the occlusive event, the patient's cardiovascular status, and the physician's willingness to accept the possible prolongation of arousal, as well as after objectively determining the probable magnitude of the protective effect.

Numerous investigations in animals and humans have failed to demonstrate any protective effect of barbiturates in the setting of global cerebral ischemia (e.g., cardiac arrest).^{[472] [502]}

Because CMR suppression has been the presumed mechanism of effect, barbiturates have traditionally been administered to produce a maximal reduction in CMR (which is nearly complete when EEG burst suppression has been achieved). However, data presented by Warner and colleagues^[496] demonstrated that the same protective benefit (expressed as reduction in infarct volume) could be achieved with a third of the burst suppression dose.^[496] This finding raises a clinically important issue. The various barbiturates (thiopental, thiamylal, methohexital, pentobarbital) have similar effects on CMR and have generally been assumed to have equal protective efficacy. However, if the mechanism of protection is a pharmacologic effect other than CMR reduction, is it reasonable to assume equivalence among the barbiturates? Recent data suggest that the neuroprotective efficacy of barbiturates is not similar. In a direct comparison of three clinically used barbiturates, methohexital and thiopental, but not pentobarbital, reduced injury in an animal model of focal ischemia.^[503] These data suggest that mechanisms other than, or at least in addition to, metabolic suppression may contribute to the protective effect of barbiturates.

ISOFLURANE/VOLATILE ANESTHETICS.

Isoflurane is also a potent suppressant of CMR in the cerebral cortex, and EEG evidence suggestive of a protective effect in humans^[440] has been reported. In comparison to the awake or N₂ O-fentanyl-anesthetized state, isoflurane has consistently been shown to be neuroprotective in models

840

PROPOFOL.

EEG suppression can also be achieved with clinically feasible doses of propofol. Anecdotal information suggests that it is being used to provide "protection" during both aneurysm surgery^[518] and carotid endarterectomy. In experimental models of cerebral ischemia, the extent of neurologic injury in propofol-anesthetized animals was similar to that in halothane-anesthetized animals.^[519] Given the previous demonstration that halothane can reduce injury, these data provide indirect evidence of propofol's neuroprotective efficacy. In a more recent investigation, cerebral infarction was significantly reduced in propofol-anesthetized animals than in awake animals.^[520] A direct comparison of propofol to pentobarbital has also demonstrated that cerebral injury after focal ischemia is similar in animals anesthetized with the two agents. ^[521] Collectively, these data are consistent with the premise that propofol can reduce ischemic cerebral injury, at least in the short term. Sustained neuroprotection with propofol has not been evaluated.

ETOMIDATE.

Etomidate was proposed as early as 1988 as a potential protective drug in the setting of aneurysm surgery.^[522] It too produces CMR suppression to an extent equivalent to the barbiturates, and like the barbiturates, etomidate is an agonist at the (inhibitory) type A γ-aminobutyric acid (GABA_A ) receptor. At that time, no investigations of etomidate using histologic end points had been conducted. Subsequently, in a model of temporary forebrain ischemia, a very small protective effect, relative to anesthesia with 1.1 MAC halothane or isoflurane or anesthesia with thiopental, was demonstrated in the hippocampus, but no protective effects were seen in other brain regions, including the cortex, reticular nucleus of the thalamus, and the striatum.^{[523] [524]} In a later investigation using a model of temporary MCA occlusion,^[501] the volume of injury was not reduced by etomidate relative to a 1.2 MAC halothane-anesthetized control group. In fact, the volume of injury with etomidate was significantly larger than in the control group. In patients subjected to temporary intracranial vessel occlusion, administration of etomidate results in greater tissue hypoxia and acidosis than equivalent desflurane anesthesia does. The aggravation of injury produced by etomidate (an imidazole) may be related to direct binding of NO^[525] as a consequence of etomidate-induced hemolysis,^[526] combined with direct inhibition of the NO synthase enzyme by etomidate.^[527] Therefore, no scientific support exists for the current use^[528] of etomidate for "cerebral protection."

Calcium Channel Blockers

It is now an established clinical practice to administer nimodipine orally (the intravenous preparation is not approved for clinical use in North America) for 21 days beginning as soon as possible after subarachnoid hemorrhage.^[529] However, it has not yet become standard practice to administer nimodipine or any other calcium channel blocker routinely after neurologic stroke occurring in the operating room or any other environment. In spite of favorable results in small trials,^{[530] [531]} not all investigations in stroke victims have confirmed the benefits of nimodipine. ^[532] Those that did demonstrate beneficial effects saw them only in limited post hoc patient subsets, that is, those in whom therapy was initiated quickly.^{[530] [531]}

Other Drugs

A remarkable number of drugs have shown neuroprotective efficacy in animal studies. However, to date, large-scale randomized trials of a variety of drugs in patients with stroke have not demonstrated neuroprotection with any drug (see Table 21-5 ). With the exception of tissue plasminogen activator for thrombolysis and the calcium channel blockers nimodipine and nicardipine for the management of subarachnoid hemorrhage, pharmacologic neuroprotective drugs are not available for the treatment of patients with cerebral ischemia.

Cerebral Ischemia: Influence of Physiologic Variables

CEREBRAL PERFUSION PRESSURE.

Measures designed to augment CBF (an important determinant of energy supply) are also important. In the "ischemic penumbra" (described earlier), small improvements in CBF have the potential to prolong neuronal survival substantially. Maintenance of a high normal CPP can augment collateral CBF^[101] and has been shown to result in improvement in various neurophysiologic parameters, including neurologic function.^{[533] [534] [535]} By contrast, hypotension can reduce CBF and exacerbate injury. Therefore, in patients with cerebral ischemia, hypotension should be treated promptly and normotension should be restored. The target MAP should be based on knowledge of a patient's preexisting blood pressure. In most patients, maintenance of MAP in the 70- to 80-mm Hg range should be adequate. Note, however, that augmentation of CPP in the high-normal range carries the inadequately explored risks of increased edema and hemorrhagic infarction if used as support during more than brief periods of ischemia, particularly when several hours has elapsed since the onset of ischemia.

CARBON DIOXIDE TENSION.

Normocapnia should be maintained. Hypercapnia has the potential to cause an intracerebral steal phenomenon and may worsen intracellular pH. In spite of some support for the occurrence of a favorable so-called Robin Hood or inverse steal,^{[252] [536]}

841

TEMPERATURE (also see Chapter 40 ).

Hypothermia is firmly and justifiably established as the principal cerebral protective technique for circulatory arrest procedures. It unequivocally enhances cerebral tolerance for episodes of ischemia. For deep hypothermia, this effect has been presumed to be largely a function of reduction of CMR. Whereas pharmacologic agents, such as the barbiturates, reduce only that component of CMR associated with electrophysiologic "work" (about 60% of CMRO₂ in the awake state), hypothermia causes a reduction in both electrophysiologic energy consumption and energy utilization related to maintenance of cellular integrity, and mild hypothermia may preferentially suppress the latter.^{[35] [36]} Recently, there has been a surge of interest in mild hypothermia as a cerebral protective technique. A substantial number of laboratory studies have demonstrated that mild degrees of hypothermia (2°C to 4°C) during an episode of ischemia can confer substantial protection as measured histologically.^{[33] [497] [498] [542] [543] [544] [545]} In addition, evidence from animal studies indicates that hypothermia initiated in the immediate postischemic period confers protective benefits.^{[499] [546] [547] [548] [549]}

These recent observations regarding mild hypothermia have cast doubt on the notion that CMR suppression is the sole mediator of the cerebral protective effects of hypothermia. This doubt has arisen because the amount of CMR suppression associated with protective degrees of hypothermia is in some instances less than that associated with pharmacologic interventions that have been less clearly protective. Microdialysis investigations in the setting of cerebral ischemia have revealed that hypothermia has a very striking inhibitory effect on the release of various neurotransmitters ^[542] (recall the pathophysiologic "cascade" discussion earlier), and it may well be that the protective effects of hypothermia are far more dependent on influences on several steps in the biochemistry of ischemia ( Table 21-6 ) than on suppression of CMR.^[550] On

TABLE 21-6 -- Possible mechanisms of cerebral protection by mild hypothermia *

Reduction of the cerebral metabolic rate

Decreased release of excitatory neurotransmitters

Inhibition of activation of protein kinase C (early)

Preservation of late postischemic enzyme function

Protein kinase C

Calcium-calmodulin kinase II

Ubiquitin

Reduction of free radical formation

Inhibition of DNA transcription

Inhibition of apoptosis

Inhibition of proinflammatory cytokines

*For details, see Maher J, Hachinski V: Hypothermia as a potential treatment for cerebral ischemia. Cerebrovasc Brain Metab Rev 5:277–300, 1993; and Colbourne F, Sutherland G, Corbett D: Postischemic hypothermia: A critical appraisal with implications for clinical treatment. Mol Neurobiol 14:171–201, 1997.)

The development of cerebral ischemia contributes to the pathophysiology of head injury.^{[557] [558]} Based on this evidence, induction of mild hypothermia has also been evaluated in head-injured patients. Although the initial small single-institution trials of mild hypothermia in head injury^{[559] [560] [561] [562]} were promising, the results of a large randomized clinical trial suggest that hypothermia may not necessarily be neuroprotective in the setting of head injury.^[563] Methodologic concerns about that study have prompted a re-evaluation of hypothermia in head-injured patients, and trials are presently being conducted.

Hyperthermia has been shown to be deleterious to outcome after standardized ischemic events.^{[514] [543] [564] [565]} Hyperthermia before, during, and after ischemic events should be prevented or treated.

GLUCOSE.

The withholding of glucose-containing solutions in situations in which cerebral ischemia may occur is now an established practice. This practice is based on numerous demonstrations in animal models of brain and spinal cord ischemia that elevation in plasma glucose before episodes of either complete or incomplete ischemia results in an aggravation of neurologic injury.^{[453] [454] [455] [566] [567] [568] [569]} However, it should be noted that most investigations have involved adult animals and that there is less certainty regarding the adverse effects of hyperglycemia in immature subjects such as neonates.^[570] Furthermore, it should be noted that only some^{[571] [572]} and not all^{[573] [574] [575] [576] [577]} of the investigations in humans have provided confirmation of an independent effect of serum glucose on neurologic outcome. A recurrent theme in the discussion of these studies is that the elevation in glucose levels may be a result of the stress associated with a severe insult, either ischemic or traumatic, rather than its cause. In addition, the inevitable questions of whether and how quickly immediate insulin treatment of an elevated plasma glucose level reduces the risk to normoglycemic levels have

842

SEIZURES, PH.

Normalization of systemic pH, prevention and treatment of seizures, which dramatically increase CMR, and control of ICP and CPP are all, though mundane and lacking the appeal of a pharmacologic silver bullet, important elements in brain protection and resuscitation.

INTRAVASCULAR VOLUME/HEMATOCRIT MANIPULATION.

Although hemodilution has not proved effective in studies of human stroke, both laboratory and human data support the practice, and it is an established part of management of the ischemia associated with vasospasm. However, the data do not currently justify routine hemodilution (a hematocrit of 30% to 35% is the theoretical optimum) in patients in whom focal ischemia might occur in the operating room.^[578] On the other hand, the potentially deleterious effects of hemoconcentration should help further suppress the out-of-date notion that neurosurgical patients should be "run dry." An increased hematocrit, because of viscosity effects, reduces CBF.^[85] It is our unsubstantiated opinion that in anticipation of a procedure wherein incomplete ischemia might occur, such as carotid endarterectomy, a hematocrit in excess of 55% should be lowered by preoperative phlebotomy.

It is quite apparent from the discussion just presented that the efficacy of pharmacologic brain protection is, at best, limited. By contrast, the potential for exacerbation of preexisting cerebral injury, either ischemic or traumatic, is substantial. Accordingly, effort should be focused on maintenance of physiologic parameters (perfusion pressure, oxygenation, normocapnia, temperature management, control of hyperglycemia, and seizure prophylaxis) within the appropriate range and less on pharmacologic agents to reduce cerebral injury.

DEFERRING ELECTIVE PROCEDURES AFTER STROKE.

The risks of extension of cerebral infarction in the event of subsequent anesthesia and surgery have not been studied systematically. In patients who have suffered a stroke, CBF undergoes marked changes. Areas of both high and low CBF occur, and stabilization of regional CBF and CMR is apparent after about 2 weeks.^[579] Loss of normal vasomotor responses (CO₂ responsiveness, autoregulation) in the early postinsult period is very common,^{[580] [581] [582]} and changes persist beyond 2 weeks in a small percentage of stroke victims.^{[581] [582]} BBB abnormalities, as reflected by accumulation of CT contrast or brain scan isotopes, are still present 4 weeks after the insult,^[583] and histologic resolution of large infarcts is not complete for several months. The available information does not allow a definite statement regarding how long elective procedures should be deferred. A 6-week delay should give some assurance of the probable recovery of autoregulation, CO₂ responsiveness, and BBB integrity. However, the size and location of the infarction should be weighed. A small infarction in silent cortex may give wider latitudes than a large lesion that has resulted in a paresis that is still resolving. A recent prospective study suggested that in patients with nondisabling stroke, early carotid endarterectomy can be safely performed within 2 weeks of the stroke.^[584] However, in patients with larger strokes, pending other information, it seems reasonable to defer elective surgery for at least 4 weeks after a cerebral vascular accident and preferably for 6 weeks from the point at which a stable postinsult neurologic state has been achieved.