++
Hypoxemia, usually defined as a state when the partial pressure of oxygen in arterial blood (Pao2) is less than approximately 50 to 60 mm Hg, is associated with cerebral vasodilation.55 Conversely, hyperoxemia and its accompanying hypocapnea have a vasonstrictive effect, as demonstrated by Floyd et al,56 in a group of healthy volunteers, Nakajima et al57 evaluated this phenomenon in patients with cerebrovascular disease, finding that areas of the brain with impaired cerebrovascular reserve were not adversely affected by hyperoxia.
++
The optimal Pao2 in a brain-injured patient is presently unclear. There are data to support hyperoxic therapy, as well as data to suggest that such an approach is deleterious.58 Furthermore, the bedside decision about Pao2 management is coupled to cerebrovascular reserve issues, previously discussed. Thus a low Pao2 that would normally be tolerated through vasodilation may be less well tolerated if vasodilatory reserve is compromised with, for example, carotid occlusion, brain edema, or anemia.
++
Fiskum et al,59 among, others have reported in laboratory studies that hyperoxic therapy promotes the generation of free radicals and that such oxidative stress causes mitochondrial injury, which will act to impair neurologic recovery. This notion from in vitro considerations is supported by in vivo studies in rodents and dogs that demonstrate worse neurologic outcomes when hyperoxia is used before or after an ischemic insult.60-62 Empirical support for the notion of hyperoxic toxicity has been provided through a study by Kilgannon et al,58 who reported that hyperoxia was independently associated with increased in-hospital mortality after resuscitation for cardiac arrest.
++
Conversely, with the advent of reports supporting the feasibility and reliability of brain tissue partial pressure of oxygen (Po2) monitoring,63 data are accumulating to suggest that normoxemic therapy, in the context of cerebral tissue hypoxia, may promote ischemic injury.63-65 Many recent reports from a nonrandomized retrospective and prospective series of both traumatic and SAH patients have described that a partial pressure of oxygen (Po2) in brain tissue equaling less than 20 to 30 mm Hg is associated with worsened neurologic outcome (Fig. 85-3).63,66-72 Notably, however, these studies did not examine for the effects of hyperoxia, which is the negative situation identified by Fiskum et al (described previously)59; rather, these studies point out the value of avoiding tissue hypoxia, perhaps at the cost of developing systemic hyperoxia but not intracranial hyperoxia. However, one side observation that falls out of these studies is the potential (theoretical) impact of avoiding hyperoxia; brain tissue oxygen monitoring allows one to provide that minimal fraction of inspired oxygen (Fio2) which permits the optimal (not too high or too low) oxygen level in brain tissue. Moreover, the results of Jaeger et al73 suggest the notion that an alternate way to use brain tissue Po2 data is to help identify the autoregulatory optimum for a given patient.
++
++
Given these potentially conflicting therapeutic priorities, it seems that the most sensible approach at this time is as follows: In the presence of a brain tissue Po2 monitor, adjust physiologic parameters to keep oxygen tension in brain tissue (Pbro2) at greater than 20 mm Hg. This may entail the use of Fio2 at greater than 60% with concomitant risk of pulmonary oxygen toxicity.74 It seems that this risk can be incurred for 1 to 2 days; however, continued dependence on toxic pulmonary oxygen concentrations should produce a time-dependent imperative for caregivers to decrease Fio2, even if this means using higher airway pressure or allowing Pbro2 to decrease after a few days. As bedside computing technology improves, the added benefit of measuring ORx, as described by Jaeger et al,73 to optimize blood pressure may be an added use of brain tissue Po2 monitoring.
++
In the absence of a Pbro2 monitor, the clinician is left to base therapy on assumptions about brain oxygenation. If the clinician believes that many brain areas are well perfused and at risk for hyperoxia, then pulmonary management should aim for a Pao2 just sufficient to produce oxygen saturation (Sao2) greater than 95%. Conversely, if there is elevated ICP and/or areas of brain hypoperfusion, then a reasonable empiric approach would be to utilize an Fio2 of 0.60. This will maximize Pao2 and Pbro2 without a significant risk of acute pulmonary injury.
++
The partial pressure of carbon dioxide in arterial blood (Paco2) is determined by the balance between CO2 production and elimination. CO2 production is determined by use of the metabolic rate and respiratory quotient (ie, CO2 production divided by O2 consumption, which is ordinarily 0.8). Factors that may increase CO2 production are hyperthyroidism, fever, elevated catecholamine levels, exercise, sepsis, and some pharmacologic stimulants. The respiratory quotient is affected by energy metabolism; intake of calories in excess of needs results in lipogenesis, a CO2-producing process. The net effect is that more CO2 is produced than oxygen is consumed.75
++
Carbon dioxide elimination is determined by use of minute ventilation and dead space measurements. There is a linear relationship between minute ventilation and Paco2 such that one can describe a simple proportion to predict the Paco2 that will result with a given change in minute ventilation. Dead space effects are more complex. There are 2 types of dead space: anatomic and physiologic. Anatomic dead space is that portion of the airways that do not participate in gas exchange because they are not proximate to pulmonary capillaries. Such structures include the mouth, trachea, bronchi, and other large airways. Notably, anatomic dead space is roughly halved via endotracheal intubation and halved again by conversion from translaryngeal intubation to tracheostomy. In contrast, physiologic dead space is that portion of non–gas-exchanging ventilation that occurs in alveoli that are suboptimally perfused. Thus physiologic dead space will be increased in 2 ways: by anything that increases the amount of gas in alveoli without a commensurate increase in alveolar perfusion, or by anything that may decrease perfusion to alveoli without a commensurate decrease in ventilation. Physiologic situations associated with elevated physiologic dead space include use of PEEP in compliant lungs, pulmonary emboli, or shock. A more detailed overview of this physiology can be found in West.75
++
In the healthy brain CBF varies linearly with a Paco2 between about 20 and 60 mm Hg.76 The mechanism of effect is thought to be related to the effects of Paco2 on cerebrospinal fluid (CSF) pH.77 Thus patients who are chronically hypercapneic and sustain pH adjustment in the CSF may not be hyperemic. These patients, of course, may be expected to sustain even more profound decreases in CBF with decrements in Paco2 to equivalent levels.
++
Generally, the Paco2-mediated changes in CBF have no neurologic import in health. However, in the context of head injury or other causes of ICP elevation, the effects can be profound because the changes in CBF induce changes in intracranial blood volume. In a brain with little capacitance for such a change in intracranial contents, the change in Paco2 can have a significant impact on the intracranial pressure.
++
Thus for many years hyperventilation was embraced as a mainstay of the treatment of intracranial hypertension.78 However, such therapy was observed to produce a significant cerebral oligemia; the lowered ICP79 often developed from a low CBF baseline and a hyperventilation-induced ischemic burden developed (Fig. 85-4).79 Conversely, in patients with brain injury, elevated Paco2 was noted at times to lead to both high ICP and high CBF, producing a therapeutic quandary. Moreover, adding to the dilemma were observations from the basic science literature of some neuroprotective side effects associated with hypercapnic cerebral acidosis.80
++
++
Relatively recent studies debunked the previously accepted theory that hyperventilation is an automatic and necessary element of treatment for head injury. Muizelaar et al81 conducted a prospective randomized study of the efficacy of hyperventilation in traumatic brain injury (TBI). Their outcome data showed a persuasively negative impact of hyperventilation (Fig. 85-5) such that it has been abandoned as a routine therapy in traumatic brain injury. However, there are some situations where it is still accepted. Some authors suggest that brain oxygen monitoring by either jugular oximetry or tissue Pbro2 can be used to guide the use of hyperventiliation.82 Direct CBF measuring techniques could also be used. Notably, such information can allow the clinician to identify whether the patient has an element of hyperemia contributing to the elevated ICP. Logically, this situation seems an appropriate use for hyperventilation therapy, but this notion has not undergone rigorous scrutiny.
++
++
Temperature has a profound effect on the brain. Fever is convincingly associated with worsened outcomes including greater release and toxicity of neurotoxic amino acids, mismatch between flow and metabolism, oxidative stress, and many other likely unknown processes.83,84 In normal brain, with every 1°C reduction in brain temperature, hypothermia produces a 7% reduction in the cerebral metabolic rate for oxygen.85 The consequent result is a decrease in the consumption of energy metabolites and an increase in time until an hypoxic stress produces high-energy phosphate depletion, thus increasing the time that the hypoxia can be tolerated.86 This reduction in cerebral metabolic rate for oxygen cannot explain the neuroprotective effect of mild hypothermia, which must be due to synergism of many physicochemical mechanisms. If one compares the neuroprotective efficacy of hypothermia with that produced by an anesthetic producing an equivalent decrement in cerebral metabolic rate, one always observes greater protection by hypothermia. This is thought to be somehow related to the differential effects of hypothermia and anesthetics on the compartments of brain energy metabolism.87 Anesthetics reduce metabolic processes related to the work of the neuron (ie, neurotransmitter synthesis and metabolism), whereas hypothermia also affects the compartment responsible for constitutive activities of the cell (ie, membrane integrity, ionic concentration homeostasis, and so on). In addition, there are other biochemical processes that contribute to hypothermic protection. For example, with mild hypothermia there is a substantial blunting of the release of neurotoxic dicarboxylic amino acids such as glutamate and aspartate.88
++
It is thus not surprising that there are countless case reports and basic science studies showing the neuroprotective potential of hypothermia across a broad range of neurologic insults. It is of interest that clinical studies do not uniformly show comparable efficacy.
++
Hypothermia has been studied and clinically used for much of the 20th century. This interest arose from anecdotes describing miraculous recovery from drowning and other brain ischemia situations in cold environments. Deep hypothermic conditions have been used for many years for neuroprotection during cardiac surgery and during therapeutically induced deep hypothermic cardiac arrest for a variety of procedures.89,90
++
Used at less extreme levels, hypothermia has also been reported to be neuroprotective, although not uniformly so in recent studies. This discussion will focus on moderate hypothermia (30°C-34°C).
++
There are many reports of neuroprotection with hypothermia in traumatic brain injury. However, all these reports are single-institution studies. When hypothermia was examined in multi-institutional studies, protection could not be demonstrated.91 However, notably, in a study by Clifton et al, neuroprotection was reported if the patient arrived already hypothermic, and rapid rewarming may have contributed to some of the negative findings.92,93 This supports the notion that speed of induction and suspension of hypothermia may not have been uniformly applied across the participating institutions in the multi-institutional studies. Clifton et al93 make a persuasive argument in this regard, asserting that significant degradation of the signal-to-noise ratio may have made detection of hypothermic neuroprotection very difficult. The variables contributing to this, which they documented, are the extensive practice variation that occurs across the United States in the approach to management of head trauma, many of which likely have an impact on outcome, and differences in admission temperature. This latter factor may have important geographic and climactic origins. Moreover, hypothermia requires an attentive multidisciplinary approach in order for it to be rapidly and safely induced. (The problem of clinical heterogeneity in the context of failed neuroprotection studies has been discussed in an editorial by Kofke.94) Hypothermia is associated with coagulopathy, immune suppression, and worsened pneumonia, among other effects.
++
Induction of moderate hypothermia (28°C-32°C) before cardiac arrest has been used successfully since the 1950s to protect the brain against the global ischemia that occurs during some open-heart surgeries. Data from Cheung et al provide recent biochemical support for this with their suggestion that about 30 minutes is a neurophysiologically identified safe limit for deep hypothermic arrest.95 In additional unpublished studies at the University of Pennsylvania, Cheung's group has further correlated this 30-minute time limit for deep hypothermic arrest with retrograde perfusion of the brain between evoked potentials and biomarkers for brain injury taken from blood flowing from the ischemic human brain. Induction of hypothermia after return of spontaneous circulation (ROSC) following cardiac arrest has been associated with improved functional recovery and reduced cerebral histologic deficits in various animal models of cardiac arrest.96-99 These and similar studies led to concurrent publication in the New England Journal of Medicine of 2 single-institution studies showing a neuroprotective effect of moderate hypothermia applied to patients sustaining out-of-hospital cardiac arrest.100,101 These patients sustained a return of spontaneous circulation, but on initial examination were not responsive and thus were randomized to the protocol. The protection was observed when the target hypothermic temperature was successfully achieved up to 6 hours after the return of spontaneous circulation (Fig. 85-6). A summary of these 2 studies is presented in Table 85-3.
++
++
++
On the basis of the published evidence to date, the Advanced Life Support (ALS) Task Force of the International Liaison Committee on Resuscitation (ILCOR) made the following recommendations in October 2002102:
++
- Unconscious adult patients with spontaneous circulation after out-of-hospital cardiac arrest should be cooled to 32°C to 34°C for 12 to 24 hours when the initial rhythm was ventricular fibrillation.
- Such cooling may also be beneficial for other rhythms or in-hospital cardiac arrest.
++
Although these data seem compelling in support of the use of hypothermia after global brain ischemia in various contexts, there are some potential issues to consider in applying these observations to the perioperative situation. These studies were very selective in their entry criteria. Thus up to 92% of cardiac arrests were excluded. Moreover, the 2 studies were not blinded, a practice that is thought to be essential to such investigations; this was clearly impossible for this specific therapy. In addition, Darby103 suggests that nonuniform neurologic entry criteria were used. There has been some doubt as to whether they studied normothermia or simply prevention of fever, as the normothermia group had an increase in core temperature of up to 38°C, as is often seen after cardiac arrest. It is also important to note that the therapy is not trivial to implement and is associated with some morbidity. Hypothermia was associated with higher systemic vascular resistance, lower cardiac output, and higher blood glucose, and 22% of hypothermic patients developed complications, especially pneumonia, although statistical significance was not achieved.
++
One controversial issue is whether findings from animal experiments and published clinical studies are enough to extend the use of therapeutic mild hypothermia to patients who remain comatose after cardiac arrest from any rhythm, after in-hospital or perioperative cardiac arrest, and after cardiac arrest in children. Moreover, many perioperative cardiac arrests have noncardiac causes (eg, hemorrhage, anesthesia), and because the use of therapeutic hypothermia has not been studied to a significant extent in this population, its relative risks and benefits are unknown. Kofke suggests in an editorial that in cases like this, the value of the potential benefit, preservation of neural function, is so important as to merit the risk from uncertainty.104 Further research is needed to determine the optimal duration of therapeutic hypothermia, optimum target temperature, and rates of cooling and rewarming. Animal data suggest that the sooner cooling is initiated after reperfusion from cardiac arrest, the better the outcome, although an impressive therapeutic benefit was seen in clinical studies when cooling was delayed for several hours. One important finding from these studies and from those investigating traumatic brain injury is that normothermia should be restored only slowly, as rebound hyperthermia is common and should be avoided.
++
Focal ischemia can be categorized as temporary or permanent. Temporary ischemia occurs often during aneurysm clipping surgery when a large cerebral artery may be temporarily occluded to facilitate clipping of the aneurysm. Typically this lasts only a few minutes, but occasionally it can last longer than 15 minutes such that risks arise for the development of ischemic injury. Ample animal data support the potential value of hypothermia in this context105-107 and was the reason that most neuroanesthesiologists in the United States routinely used it prophylactically in patients undergoing cerebral aneurysm clipping up until 2005. At that time, another multi-institutional study, the Intraoperative Hypothermia for Aneurysm Surgery Trial (IHAST)108 in over 1000 patients randomized to moderate hypothermia or normothermia, was unable to detect a difference in either stroke rate or cognitive deficits. Thus the use of prophylactic moderate hypothermia for this clinical context has been largely abandoned.
++
However, 1 criticism, similar to some of the issues with the Clifton TBI study,91 is that the IHAST study, by examining all aneurysm patients without measuring the degree of intraoperative ischemia, may also have had a signal-to-noise problem. Only a small subset of the study group may actually have had a risk of stroke-inducing ischemia. Arguing against this in this case was the scrupulous oversight given to the physiologic support of all of the patients at the various sites and the power analysis that was done on the pilot data patients who were presumably reflective of the entire group. Notably, the pilot data came from a single institution study that showed a protective effect.109 Nonetheless, this negative study supports not using prophylactically applied moderate hypothermia in unselected cases.
++
So, left unresolved in my opinion are 2 issues: whether it would be useful to apply hypothermia in a case that is expected to be complex with a high likelihood of significant focal ischemia, and whether rapid induction of moderate hypothermia could be of value in the patient in whom significant focal ischemia is actively occurring. I believe that a neuroprotective effect in this specific situation may have been missed in the IHAST study. Taking into account the data from many neuroprotective animal studies, and the possible therapeutic gain versus expected outcome and risks of the intervention, hypothermia seems a reasonable approach in cases where temporary focal ischemia is thought to be very likely.104
++
An opposite situation is that of fever, which, in both animal and clinical studies, has been convincingly associated with exacerbation of neurologic outcome in the injured brain. Many neuroscience intensive care unit (NeuroICU) patients have recurrent problems with significant fever in excess of 39°C in the absence of identifiable sepsis. Indeed, in a prospective quantification of 428 consecutive patients admitted to neurovascular or neurotrauma ICUs, 46.7% of patients had at least 1 febrile episode.110 This is most likely a sequela of the neurologic process, thought by 1 group of authors to at times reflect autonomic dysfunction111 induced by the neurologic injury.112 SAH is an independent risk factor for development of fever without identifiable cause.113 The notion that fever kills neurons is gaining widespread acceptance and is based on many clinical studies showing the improvement in neurologic outcome associated with prevention of fever. Normalization of fever in SAH patients with microdialysis probes was observed to improve the interstitial metabolic profile.114
++
In a study by Castillo et al,114a the mortality rate in patients who were hyperthermic within the first 72 hours after stroke was 15.8%, compared to a 1% mortality rate in patients who were normothermic during that time. Hyperthermia that occurred within the first 24 hours after stroke, without respect to infectious or noninfectious origin, was independently related to a larger infarct volume, and worse neurologic deficits and dependency 3 months postinjury.114a Azzimondi and Bassein115 reported that fever of 37.9°C or above proved to be an independent risk factor predicting a worse outcome, and patients with high fever were far more likely to die within the first 10 days than those with lower temperatures. A prospective study by Reith et al116 of 390 consecutive cases of acute stroke classified patients into 3 admission-temperature groups: hypothermic (36.5°C or less), normothermic (36.6°C-37.5°C) and hyperthermic (above 37.5°C). They showed that admission body temperature was highly correlated with initial stroke severity, size of the infarct, mortality rate, and poor outcome. For a 1°C difference in body temperature, the relative risk of poor outcome was more than doubled. This was supported by Ginsberg and Busto,117 who reported that in stroke patients, fever that occurs soon after stroke onset is most strongly associated with poor outcome. Body temperature was significantly higher in patients who died within 3 days after admission compared with the rest of the study population. Moreover, in ICH patients surviving the first 72 hours after hospital admission, the duration of fever is associated with poor outcome and seems to be a prognostic factor in these patients.118 In SAH fever is associated with vasospasm and poor outcome independent of hemorrhage severity or the presence of infection.
++
Pharmacologic methods can be used to control fever. Such modalities include acetaminophen119 and ibuprofen,120 although there is some evidence suggesting that ibuprofen is not efficacious in the context of ischemic stroke.121 Acetaminophen's value can be limited by hepatotoxicity. It is also of theoretical concern that acetaminophen is an oxidizing agent that even at subhepatotoxic levels may decrease glutathione to lessen potentially helpful free-radical scavenging processes.122 Ibuprofen is associated with gastric ulceration and bleeding.123
++
An alternate method is to use a hypothermia blanket or an indwelling hypothermia catheter.124 Both techniques are based on a servo-controlled system wherein the cooling bath temperature decreases when the patient's temperature starts to rise. In our NeuroICU we have generated illustrations of temperature curves indicating that it is possible to virtually eliminate fever from the pathophysiology of severe brain injury. Notably, this has created a new vital sign, Tbathmin, which indicates the minimum bath temperature that was needed to prevent fever in a given patient. Ascertaining a low Tbathmin may indicate a need to initiate an investigation for infection.
++
In non-neurologic critical care populations, recent randomized trials have yielded conflicting results regarding the importance of glycemic control. The initial "Leuven" study125 in postcardiac surgery patients indicated a mortality improvement with tight glucose control in the 80 to 110 gm% range. This was a single-center study of a homogeneous patient population with specially trained nursing staff. However, subsequent studies in medical and more heterogeneous groups and muticenter studies have been unable to demonstrate a beneficial effect from tight glucose control.126-128 Van den Berghe et al have nicely reviewed the issues in interpreting these conflicting studies with speculation as to contributing factors.129 To date, none of the studies of tight glucose control have focused on patients with cerebral ischemia and, due to considerations reviewed elsewhere in this chapter, glucose management with neurologic disease is somewhat enigmatic.130
++
Hyperglycemia has been associated with exacerbation of brain damage with both head trauma and cerebral ischemia.131-134 This is not a straightforward issue, however, because of the aforementioned conflicting clinical studies as well as pathophysiology considerations. Clearly, neuronal damage after global cerebral ischemia is exacerbated with hyperglycemia.132,135-138 Some studies have suggested that a blood glucose greater than 120 gm% is deleterious in stroke patients.131 However, subsequent studies with subhuman primates subjected to global ischemia have suggested a threshold of around 180 gm%.138 It is apparent that blood glucose greater than 400 gm% causes striking worsening of neurologic outcome with global ischemia.132,137
++
With focal cerebral ischemia, the role of blood glucose is a good deal less clear. Animal and human studies have shown that hyperglycemia either worsens, does not affect, or lessens brain damage.131,139-156 One report by Prado et al153 in rats suggested that the discriminating factor in whether or not brain damage is worsened with hyperglycemia is the presence of collateral flow. Areas of the brain with minimal collaterals were not affected or were improved with hyperglycemia. Brain areas with a continued trickle of anaerobic hyperglycemic flow sustained worsened brain damage. Presumably the continued substrate supply in oligemic (not ischemic) areas allowed greater accumulation of organic acids in the cells, leading to worsening of brain damage.133,144 Unfortunately, these observations are difficult to apply clinically to specific patients with focal ischemia.
++
Even if low levels of hyperglycemia were deleterious, it would not be straightforward to treat. Aggressive therapy of hyperglycemia would impose a risk of producing hypoglycemia, with deleterious effects arising therefrom.157 Moroever, Oddo et al158 report a paradoxical increase in interstitial lactate on microdialysis monitoring associated with tight glucose control, suggesting a direct deleterious effect of tight glucose control. The genesis of this increased lactate in the context of low precursor glucose is enigmatic, and thus the interpretation of these microdialysis data is unsettled. Therefore, given the data in the general critical care literature, it seems that a reasonable approach is to aim for a blood glucose of approximately 150 to 180 gm% in all acutely ill hyperglycemic patients at risk of cerebral ischemia, but to be less stringent in blood glucose control in less acutely ill patients who are expected to be in the ICU for less than 3 days. In any event, to avoid having the blood glucose swing to less than 80 gm% or greater than 400 gm%, it should not be allowed to undergo wide variations in concentration. Thus an insulin infusion should be used in acutely ill patients at risk of cerebral anaerobic metabolism who develop hyperglycemia greater than 200 gm%. In these patients, assess glucose levels frequently and titrate to maintain blood glucose at about 150 to 180 gm%. Once the hyperacute phase has resolved, the intensity of glucose monitoring can be lessened somewhat to match the lowered acuity. A sliding-scale insulin paradigm can be used; apply somewhat less stringent goals as the patient is readied for discharge from the ICU setting.
++
Hyperglycemia has not been shown to have deleterious or protective effects in 2 animal models of status epilepticus.159,160 The model used in the report by Swan and Meldrum160 produced limbic system damage whereas the report by Kofke et al159 used a model producing substantia nigra damage. Nigral damage in this model is associated with hypermetabolic lactic acidosis,161 which should have been exacerbated with hyperglycemia. The fact that hyperglycemia did not exacerbate nigral damage suggests one of the following: either metabolic acidosis may not be an important factor in the development of brain damage after seizure, or the lactic acidosis associated with hyperglycemic exacerbation of ischemic brain damage is not the true pathogenetic culprit but is, rather, an epiphenomenon.
+++
Cardiovascular Issues
+++
Blood Pressure Effects on ICP–Plateau Waves and Determination of Blood Pressure Optimum
++
Lundberg in 1960162 monitored ICP in hundreds of patients, identifying characteristic patterns of pressure waves. One of these patterns is that of plateau waves, known to be associated with increased cerebral blood volume (CBV).163 Such waves occur when the ICP abruptly increases to nearly systemic levels for about 15 to 30 minutes, occasionally accompanied by neurologic deterioration. Rosner and Becker164 provided observations and a synthesis of the data, which convincingly suggest that intracranial blood volume dysautoregulation is responsible for plateau waves. These investigators induced mild heterogeneous head trauma in cats and intensively monitored the animals after the insult. In normally fluctuating blood pressure, the development of plateau waves is preceded by decrements of mean-arterial blood pressure to a range of approximately 70 to 80 mm Hg. In normal brain tissue, cerebral blood volume increases due to autoregulatory vasodilation and blood pressure decreases. Moreover, the increase in CBV is nonlinear; as blood pressure decreases to below 80 mm Hg, CBV increases exponentially.165 In a setting of abnormal intracranial compliance with the ICP, a small reduction in blood pressure, although in the normotensive range, produces an exponential rise in CBV. This phenomenon is represented by the "elbow" of the classical curve of the intracranial pressure–intracranial volume relationship. Thus a small decrease in blood pressure introduces an exponential change in CBV upon a dramatically increased ICP such that ICP will increase abruptly and to a significant extent. Plateau waves spontaneously resolve with a hypertensive response or with hyperventilation, which acts to oppose the increase in CBV. Clearly, to develop a plateau wave, some portion of the brain must have normally reactive vasculature in some other brain areas with a mass effect and elevated ICP, a situation of heterogeneous autoregulation. Such data suggest that in patients with an elevated ICP in the 80 to 100 mm Hg range, not only should plateau waves be prevented and treated, but also it is probably important to maintain MAP. However, this may be an overly simplistic conclusion based on more recent investigations into the notion of a CPP optimum; this issue is detailed below.
++
Conversely, systemic hypertension can also increase ICP. Ordinarily, in cases where blood pressure is within the normal autoregulatory range and there is a normal intracranical pressure, changes in blood pressure have no effect on ICP. However, with brain injury and associated vasoparalysis, blood pressure increases are presumed to mechanically produce autoregulatory cerebral vasodilation, which will increase ICP.166 This observation forms the basis for the notion, detailed below, of using such ICP variations to quantitate autoregulation.167
++
Any consideration of hemodynamics and ICP also has to account for the veins. Blood coursing through the brain runs through arteries, capillaries, veins, the sagittal and other dural sinuses, and then on to the internal jugular and other extracranial veins. In the context of a closed intracranial space, the relationship of these vessels to the tissue and CSF surrounding them becomes important. Notably, several investigators, in laboratory preparations, have observed a distinct drop-off in intraluminal pressure for blood moving from cerebral cortical veins to the sagittal sinus. This phenomenon is most evident when ICP is elevated, and indicates the presence of a vascular waterfall at a point just proximal to the sagittal sinus, whereby the extraluminal high-pressure CSF is acting to impede flow from cortical veins to the sagittal sinus.168-171 In fact, Nemoto169 and Nakagawa et al168 have further observed that the cerebral venous pressure (CVP) tends to be consistently higher than the ICP. The implications of these observations are the following: the elevated ICP begets increased CVP; the increased CVP promotes and exacerbates brain edema, which may have been the initial cause of the intracranial hypertension; and this then leads to a positive feedback cycle wherein increased ICP increases CVP, which increases ICP.169,172 Thus any other factors that may promote brain edema or otherwise increase ICP in this tenuous situation (eg, high extraventricular drain, systemic hypertension173 or hypo-osmolarity) may initiate such a positive feedback process.
++
Notably, the Lund group172 incorporates consideration of these issues of cerebral venous pathophysiology into its treatment approach and suggests that hypertension-induced exacerbation of brain edema increases ICP. The increase in ICP then acts to occlude venous outflow, increasing venous pressure, which in turn acts to further worsen the brain edema, constituting a positive feedback cycle initially started by arterial hypertension.
++
It thus appears that both increasing and decreasing blood pressure can increase ICP, suggesting the presence of a CPP optimum for ICP. In the absence of any patient-specific physiologic information, this optimum is probably about 80 to 100 mm Hg.
++
These considerations underlie a current controversy with respect to blood pressure management in the context of elevated ICP. One argument is that blood pressure should be maintained at high levels to ensure adequate CBF and minimize the probability of developing plateau waves. The contrary argument is to use ample fluids and consequent low blood pressure to primarily promote CBF rather than perfusion pressure. It is this author's opinion that the preferred approach would be to induce the lowest blood pressure that allows sufficient CBF as indicated by repeated (preferably bedside) measurements. Recent investigations indicate that such bedside assessment capability should soon be widely available.51
++
Recent advances in transcranial Doppler ultrasonograpy have allowed insights into the dynamic, nearly instantaneous assessment of cerebral autoregulation in critically ill patients. This approach uses the correlation coefficient, determined in real time at the bedside, to make inferences regarding autoregulation. A high correlation of blood flow velocity with ICP suggests poor autoregulation; a lack of correlation is normal. Czosnyka et al174 observed in TBI patients a U-shaped curvilinear relationship in flow velocity versus arterial blood pressure (ABP), with worse autoregulation (ie, high correlation) at ABP readings of less than 75 mm Hg and greater than 125 mm Hg. The authors noted that increasing ABP also increased ICP, further indicating a correlation-based marker of dysautoregulation, the so-called pressure reactivity index, or PRx.52,167,175 Dynamic time domain analysis of cerebrovascular autoregulation using transcranial Doppler (TCD) sonography, ICP, brain tissue oxygen partial pressure (Pbo2), or near-infrared spectroscopy (NIRS) is a current topic of investigation with promising reports of potential efficacious and valid bedside use.51,167,176-178 For example, Joshi et al,52 using the TCD-based autoregulation assessment mean velocity index (Mx) in cardiac surgery patients, reported an association of disturbed Mx with postoperative stroke; autoregulation feasibility in this context was demonstrated with NIRS.179,180
++
The ICP pressure-reactivity index (PRx) is another way to dynamically evaluate autoregulation in a number of conditions.167,175,181 PRx is a quantitation of the aforementioned description of abnormal dynamic correlation of ICP changes and ABP changes. Reports indicate that PRx correlates well with other autoregulation indices.52,175,182,183 Steiner et al167 reported on the use of PRx monitoring in TBI patients to determine the optimal CPP in this population. Patients with better autoregulation in the optimal range as defined by PRx had better outcomes. Moreover, patients with dysautoregulation related to higher ABP with corresponding ICP elevation also had worse outcomes, suggesting that autoregulation monitoring, to ensure adherence to an individual's optimal CPP, may be an outcome-altering ICU measure. Zweifel et al175 report congruent observations. Notably, PRx, as with TCD-based autoregulation studies, also appears to undergo a U-shaped curvilinear relationship with variations in CPP; in TBI patients it is abnormally high (ie, ICP varies with ABP) at both low (ischemic) and high (hyperemic) CPP. Moreover, further complementing this information are observations of abnormally high oxygen extraction fraction (OEF) and low OEF at these respective ABP extremes. This is underscored by reports of a significant ischemic burden in TBI patients,79,184-186 suggesting a delicate balance between hypotension-associated hypoperfusion and hypertension-associated edema/ICP exacerbation, both of which will worsen regional ischemia. Taken together, these autoregulation studies introduce the hypothesis that there is an individualized ABP optimum in TBI patients,175 which should be a therapeutic goal. Autoregulation monitoring is also being reported using a brain tissue oxygen pressure reactivity index (ORx)73 and an NIRS-rendered cerebral oximetry index (COx), suggesting that noninvasive autoregulation assessment will be a feasible bedside modality that can be used to deliver the optimal CPP.
+++
Blood Pressure Management
++
Blood pressure management is an important issue in most NeuroICU patients and a matter of some concern. Systemic hypertension may exacerbate cerebral edema or intracranial hemorrhage, or have deleterious cardiopulmonary effects such as pulmonary edema or myocardial ischemia. Conversely, blood pressure reduction can lead to insufficient perfusion even at a pressure in the normal range of autoregulation. Moreover, mild blood pressure decreases have been implicated in the genesis of plateau waves. Several important principles apply to the management of blood pressure in NeuroICU patients. These principles are discussed below.
++
When blood pressure is high, a fundamental initial question must be whether or not the pressure is elevated due to normal homeostatic mechanisms that are acting to maintain adequate perfusion. For example, in conditions with inadequate brainstem perfusion, a compensatory hyperadrenergic state may occur, leading to increased blood pressure, which acts to supply sufficient perfusion to maintain aerobic metabolism in the brainstem. If a decision is made to decrease blood pressure, brainstem failure and death may ensue.
++
Animal data with cerebral ischemia models provide strong support for the notion that sympatholytic drugs should be used to decrease blood pressure if cerebral ischemia is a possibility. When compared to hemorrhage-induced hypotension, ischemic damage occurred to a lesser extent with the use of ganglionic blockade with hexamethonium,36 central adrenergic blockade with α2 agonists,37 and angiotensin-converting enzyme inhibition.38 Hemorrhaged controls were noted to sustain an increase in exogenous catecholamine concentrations. To test the hypothesis that these catecholamines contributed to brain damage, some of the animals treated with hexamethonium also received intravenous (IV) catecholamine infusions. Reversal of hexamethonium brain-protective effect was observed in these animals (Fig. 85-7).36 Similarly, brain protection has been observed in laboratory studies with preischemic188 and preseizure189 treatment with reserpine, a drug that depletes presynaptic catecholamine stores. Finally, there is a report by Neil-Dwyer et al190 regarding subarachnoid hemorrhage patients given therapy with phentolamine and/or propranolol or with no sympatholytic agents (Fig. 85-8). Subjects who received sympatholytic therapy had significantly better neurologic outcome than controls. In addition, β-adrenergic blocking drugs have not been reported to produce cerebral vasodilatation or increased ICP.191-193
++
++
++
Other drugs that may have brain-protective effects are the calcium channel antagonists, also available for antihypertensive therapy. Nimodipine and nicardipine, developed specifically for brain-protection purposes, have been assessed in numerous studies, with several reports of their having conferred protection versus vasospasm and ischemic brain damage.194-200 Based solely on these observations, these drugs thus become reasonable choices for antihypertensive therapy. Nimodipine and nicardipine are vasodilators; they can modestly increase ICP201,202 and may be expected to produce compensatory catecholamine release.203 Thus, based on the previous discussion, this may obviate some of their protective qualities. Nimodipine moreover has been observed to decrease brain tissue Po2.204 Whether this phenomenon can alter outcome is not known.
++
Peripheral vasodilators such as nitroprusside, nitroglycerin, and hydralazine all have the potential to induce cerebral vasodilatation and thus cause hyperemic intracranial hypertension.205-208 Moreover, they are associated with a compensatory increase in peripheral catecholamines and renin,203 factors which theoretically may worsen ischemic brain damage.36,188 However, the lack of bradycardia and bronchoconstriction associated with the use of these agents may make them the optimal treatment choice in some patients with these conditions. If one of these drugs is selected to treat a patient at risk of neurologic deterioration from ischemia or high ICP, close clinical observation is indicated. Any deterioration would mandate discontinuation of the drug. Such concerns are important for deciding among these 3 drugs. Although hydralazine is convenient to use, it cannot be reversed at the receptor, and its effects can last hours. Thus it may be preferable to use nitroprusside in such situations, as adverse effects can be treated quickly simply by discontinuing the infusion.
++
It should be clear that the choice of antihypertensive agent in a patient at risk of cerebral ischemia is not straightforward. Therapeutic urgency, sympatholytic and brain-protective side effects, and potential to increase ICP are all important considerations in the choice of an antihypertensive drug.
++
If it is deemed that blood pressure has to be decreased very quickly, (eg, within minutes), nicardipine in 100- to 500-μg boluses is very effective and safely titratable.209,210 Once the blood pressure is reduced, a maintenance regimen of nicardipine or another drug can then be started. Alternatively, sodium nitroprusside can be started, although it has a variety of deleterious side effects (elaborated on previously) that might move it down the preference list for an antihypertensive agent in a neurologic patient. Sodium nitroprusside, however, may be the most potent and reliable antihypertensive drug available.
++
Plateau waves, first reported by Lundberg162 and associated with neurologic deterioration, were demonstrated by Risberg et al to be associated with cerebral vasodilatation.163 Rosner and Becker reported that mild reduction in mean arterial blood pressure to the 60 to 80 mm Hg range can be associated with plateau waves.164 As mentioned previously, the decrease in blood pressure presumably prompts vasodilatation in normally autoregulating tissue. The increase in cerebral blood volume, which is an exponential function as compared with cerebral perfusion pressure,164 superimposed on the exponential ICP (versus the intracranial volume relationship), then is associated with an explosive hyperemic increase in ICP: a plateau wave. This dynamic then introduces a concern for using any antihypertensive agent (aside from specific, direct cerebrovascular effects): When CPP falls to about 80 mm Hg or less, normal autoregulatoy mechanisms may also increase ICP.
++
Clearly, any time hypotensive therapy is used in a patient with altered intracranial compliance, edema, or ischemia, very close and recurrent observation of the patient is mandatory. Deterioration should prompt consideration that one of the above processes is occurring. Introduction of corrective therapy should be accompanied by a reconsideration of the need to decrease blood pressure.
+++
Hypotension and Induced Hypertension
++
Hypotensive patients require treatment for their low blood pressure as well as ongoing effort to ascertain the cause of the hypotension. In head trauma patients, consider other injuries that might result in hemorrhage or spinal shock. For instance, loss of blood flow to the brainstem can be associated with hypotension that can be quite difficult to treat. In addition, usual non-neurologic causes of hypotension in an ICU, such as pneumothorax, sepsis, and cardiogenic causes, should also be considered.
++
When contemplating therapy to increase blood pressure, as may be done with a patient with vasospasm, seriously consider the cost-benefit. Excessive increases in blood pressure can exacerbate cerebral edema.211-215 This presumably occurs in brain areas with dysautoregulation and blood–brain barrier (BBB)–disruption, such that increasing blood pressure, rather than producing vasoconstriction with no change in rCBF, causes vascular distention, increased rCBF, and transudation of fluid across the damaged BBB. In addition, increases in blood pressure impose a risk for producing or exacerbating intracranial hemorrhage.
++
Catecholamines used to increase blood pressure are neurotransmitters or are chemically similar to neurotransmitters. It is thus to be expected that if they cross the BBB, neural effects will arise secondary to their use. Normally, exogenously administered catecholamines do not cross the BBB and have no effect on CBF or metabolism.216 However, catecholamine infusion in the presence of BBB disruption has been shown to lead to increases in blood flow and metabolism.217 In subarachnoid hemorrhage patients, catecholamine infusions produce a variety of disparate and unpredictable effects on CBF218 and adrenergic blockade confers neurologic protection.190 Finally, catecholamines have direct neurotoxic potential, as indicated by data showing neurotoxicity with application directly to the cortex in vivo.219 Unfortunately, catecholamines are the only clinically accepted routine means to pharmacologically increase blood pressure in NeuroICU patients.
++
Increasing preload to the heart is one nonpharmacologic method of increasing blood pressure. The use of crystalloid or colloid infusion is generally associated with hemodilution. When this type of infusion is contemplated, consider the effects on a given patient's status. The hemodilution may improve flow to areas where microcirculation is compromised. However, it may be associated with increased CBV and hyperemic intracranial hypertension if hematocrit decreases excessively with compensatory vasodilatation.
++
Whether to use crystalloid or colloid for this purpose remains controversial. The BBB is functionally an osmometer.220-225 Thus the added trivial increase in osmolarity with colloid is not a sufficient reason to use it. It makes sense, and is supported by animal studies, that iso-osmolar or slightly hyperosmolar fluids should be used to reduce the possibility of increasing brain edema secondary to fluid administration.
++
Some advocate the use of induced systemic hypertension with high ICP to prevent plateau waves. As blood pressure increases, incremental increases in vasoconstriction occur in normally reactive tissue and consequently decrease CBV and thus ICP. However, the advantage of this therapy may be offset by increased edema in injured brain regions.
++
Subarachnoid hemorrhage is an entity particularly notable for catecholamine effects, some of which will be described. However, catecholamine effects also occur with other intracranial processes including increased ICP, stroke, head trauma, or any situation of compromised midbrain-hindbrain oxygen delivery.
++
Serum catecholamine levels increase dramatically after SAH. Notably, levels peak at the same time as the peak incidence of post-SAH vasospasm, with symptom development corresponding to serum catecholamine levels.226-229 This phenomenon has led to the notion that hypothalamic injury with excess catecholamine release may be an important factor in the genesis of post-SAH spasm and stroke,229,230 observations that may be relevant to other intracranial processes previously elaborated. Several lines of evidence further support this hypothesis:
++
The cerebral vasculature is invested somewhat with adrenergic nerves. With SAH, the number of adrenergic receptors in the cerebral vessels decreases.230 This suggests that denervation hypersensitivity may be occurring such that the increase in humoral catecholamines with SAH produces spasm in hyperreacting vessels.
Catecholamine release after SAH is sufficient to produce electrocardiographic changes226 with ventricular wall motion abnormalities227 and myocardial injury.228
Treatment of humans with SAH with β- and α-adrenergic antagonists is associated with an improvement in neurologic outcome190 (Fig. 85-8) and electrocardiographic abnormalities.226
In animal models, selective destruction of hindbrain adrenergic nuclei with cephalad projections prevents the development of vasospasm.229 Moreover, laboratory studies indicate an important role for vasopressin in vasospasm, as vasospasm cannot be produced in vasopressin-deficient rats.230
As discussed previously, animal data with cerebral ischemia models provide strong support for the notion that catecholamines can exacerbate cerebral ischemia. When compared to hemorrhage-induced hypotension, ischemic damage was decreased with hypotension induced through the use of ganglionic blockade with hexamethonium,36 central adrenergic blockade with α2 agonists,37 and angiotension-converting enzyme inhibition.38 Hemorrhaged controls were noted to sustain an increase in exogenous catecholamine concentrations. To test the hypothesis that these catecholamines contributed to brain damage some of the animals treated with hexamethonium also received IV catecholamine infusions. Reversal of the hexamethonium brain protective effect was observed in these animals (Fig. 85-7).36
Brain protection has been observed in laboratory studies with preischemic188 and preseizure189 treatment using reserpine, a drug that depletes presynaptic catecholamine stores.
Application of catecholamines directly to nonischemic cortical tissue has also been observed to have neurotoxic potential.240 In addition, intravenous administration can exacerbate brain swelling after head trauma, although this is most likely a direct effect of blood pressure on a dysautoregulating brain (Fig. 85-8) rather than a manifestation of biochemical neurotoxicity.213
+++
PEEP and Intracranial Hypertension
++
Positive end-expiratory pressure (PEEP) can increase ICP.242 Two mechanisms for this can be posited. The first is through impedance of cerebral venous return, which will increase cerebral venous pressure and ICP. The second mechanism is through decreased blood pressure with a reflex increase in cerebral blood volume, which will increase ICP. The data of Huseby et al243 suggest that cerebral venous effects only occur with very high PEEP.
++
Shapiro244 demonstrated increases in ICP in head-injured humans with intracranial hypertension with application of PEEP. Examination of Shapiro's data indicates that the most profound decreases in CPP occurred in patients with PEEP-induced decrements in mean arterial pressure consistent with the notion put forth by Rosner and Becker164 that decreases in blood pressure increase CBV, which in turn increase ICP. Aidinis et al,245 in studies with cats, confirmed these observations in a more controlled setting. In addition, they assessed the role of pulmonary compliance, finding that decreased pulmonary compliance with oleic acid injections lessens the effect of PEEP to increase ICP. Such observations indicate that in situations where PEEP is likely to be needed—situations that are often accompanied by decrements in pulmonary compliance—any adverse effects on ICP are less likely to manifest. This phenomenon may be related to observations that hemodynamic effects of PEEP are less apparent with noncompliant lungs246 such that hypotensive-mediated increases in CBV do not occur.
++
The intuitive notion that PEEP increases cerebral venous pressure to increase ICP is not as straightforward as some may indicate. In order for PEEP to increase CVP to levels that will increase ICP, the cerebral venous pressure must be equal to at least the ICP (based on previously described notions of a cerebral venous vascular waterfall). Thus the higher the ICP, the higher PEEP must be in order to have such a direct hydraulic effect on ICP. This concept was nicely proved by Huseby et al243 in dog studies in which PEEP was increased progressively along with different starting levels of ICP (Fig. 85-9A). It is important to note that the investigators prevented the development of PEEP-induced decrements in blood pressure, thus avoiding any reflex-mediated increases in cerebral blood volume. They suggested a hydraulic model to better demonstrate this concept (Fig. 85-9B). Thus, for example, if all of a 10 cm H2O PEEP application was transmitted to the cerebral vasculature (which is unlikely given the decreased pulmonary compliance associated with the need for such PEEP), then ICP will only be affected if it is 10 cm H2O (7.7 mm Hg) or less and does not increase to a level higher than the applied PEEP. Such observations are consistent with the notion that there is a Starling resistor regulating cerebral venous outflow.247
++
+++
Optimal Hemoglobin Level
++
Anemia is generally well tolerated neurologically except at extreme levels. This indicates the enormous cerebrovascular reserve that, in health, is in place to compensate for blood loss and similar physiologic stresses. The observations of Borgstrom et al248 in rodents indicate that decreasing Hb levels produce an increase in CBF. This is initially due primarily to decreased viscosity, but as Hb continues to decrease to below 10 gm%, active vasodilatation arises (Fig. 85-10). If the brain vasculature is already maximally vasodilated because of other stresses, such as hypoxemia or low cerebral perfusion pressure, then the anemic stress may not be well tolerated and may produce hypoxic/ischemic brain damage.
++
++
Supporting observations of anemia-associated cerebral vasodilatation have been noted in a number of studies. Floyd et al noted this phenomenon in humans after cardiac surgery (Fig. 85-11).249 Dexter modeled the competing issues of anemia and cerebral vasodilatation mathematically.249a His data approximated those of laboratory249 and empiric observations that an Hb less than 10 gm% associated with vasodilatation can be expected to be deleterious in conditions of altered cerebrovascular reserve. Kim and Kang's250 observations of post-gastrointestinal (GI) hemorrhage anemia-associated stroke support Dexter and Hindman's calculations. Other supporting data for the relationship of anemia and vasodilatation come from observations in brain-injured and SAH patients: low Pbro2 was associated with anemia,251 and after transfusion, brain tissue Po2 increased to exceed an Hb of 10.252 One of the approaches to managing low Pbro2 is to transfuse to an Hb greater than 10 gm%, based on observations that such therapy can significantly improve brain oxygenation in the context of severe brain injury.
++
++
Notwithstanding reports that 7 gm% is optimal in a general critical care population,253 these above-mentioned observations taken altogether suggest that transfusion to a goal of 10 gm% is reasonable in the context of impaired cerebrovascular reserve.
++
It should be noted, however, that considerable controversy continues as evidence accumulates of either no improvement or deleterious general254-257 and neurologic258-260 effects associated with transfusion, especially if older blood is used.261 Thus a balance must be sought between risks and benefits such that transfusion is only used when there is good empiric evidence supporting its use or, lacking that, which is the usual situation, when a strong physiologic rationale exists to support the optimal decision.