Best Practice & Research Clinical Anaesthesiology Vol. 21, No. 4, pp.

517538, 2007
doi:10.1016/j.bpa.2007.09.001 available online at http://www.sciencedirect.com

6 Prevention and treatment of intracranial hypertension

Jan-Peter A.H. Jantzen *
MD, PhD, DEAA

Dr med habil, Professor at Johannes Gutenberg University Medical School at Mainz, Professor at University of Texas, Southwestern Medical School at Dallas, and Head Department of Anaesthesiology, Intensive Care and Pain Management., Academic Teaching Hospital Hannover Nordstadt, Haltenhoffstrasse 41, D-30167 Hannover, Germany

Intracranial pressure (ICP) is the pressure exerted by cranial contents on the dural envelope. It comprises the partial pressures of brain, blood and cerebrospinal uid (CSF). Normal intracranial pressure is somewhere below 10 mmHg; it may increase as a result of traumatic brain injury, stroke, neoplasm, Reyes syndrome, hepatic coma, or other pathologies. When ICP increases above 20 mmHg it may damage neurons and jeopardize cerebral perfusion. If such a condition persists, treatment is indicated. Control of ICP requires measurement, which can only be performed invasively. Standard techniques include direct ventricular manometry or measurement in the parenchyma with electronic or beroptic devices. Displaying the time course of pressure (high-resolution ICP tonoscopy) allows assessment of the validity of the signal and identication of specic pathological ndings, such as A-, B- and C-waves. When ICP is pathologically elevated at or above 2025 mmHg it needs to be lowered. A range of treatment modalities is available and should be applied with consideration of the underlying cause. When intracranial hypertension is caused by hematoma, contusion, tumor, hygroma, hydrocephalus or pneumatocephalus, surgical treatment is indicated. In the absence of a surgically treatable condition, ICP may be controlled by correcting the patients position, temperature, ventilation or hemodynamics. If intracranial hypertension persists, drainage of CSF via external drainage is most effective. Other rst-tier options include induced hypocapnea (hyperventilation; paCO2 < 35 mmHg), hyperosmolar therapy (mannitol, hypertonic saline) and induced arterial hypertension (CPP concept). When autoregulation of cerebral blood ow is compromised, hyperoncotic treatment aimed at reducing vasogenic edema and intracranial blood volume may be applied. When intracranial hypertension persists, second-tier treatments may be indicated. These include forced hyperventilation (paCO2 < 25 mmHg), barbiturate coma or experimental protocols such as tris buffer, indomethacin or induced hypothermia. The last resort is emergent bilateral decompressive craniectomy; once taken into consideration, it should be performed without undue delay.

* Tel.: 49 511 9701580; Fax: 49 511 9701012. E-mail address: jan-peter@jantzen.net 1521-6896/$ - see front matter 2007 Elsevier Ltd. All rights reserved.

518 J.-P. A. H. Jantzen

Key words: intracranial pressure; ICP; intracranial hypertension; hydrocephalus; ICP treatment; intracranial hypertension; ICP measurement; ICP monitoring; ICP etiology; ICP treatment; intracranial hypertension; Lund; CPP; intracranial osmolarity; cerebral edema; vasogenic cerebral edema; cerebral edema treatment; cerebrum; cerebrum blood ow; cerebrum blood volume; cerebral blood ow; cerebral circulation; cerebral vessels; CO2 reactivity.

The human brain is embedded in rigid structures dura mater and skull which protect it against traumatic injury. This protection may, however, become counterproductive when trauma is severe or following neurological surgery. Due to minimal compliance, space-occupying processes in the head cannot increase volume, as in abdominal emergencies, but will increase pressure. For the same reason the skull functions as a Starling resistor, limiting cerebral perfusion pressure whenever pressure within the cranium exceeds central venous pressure. Monitoring intracranial pressure (ICP) is of utmost value for early detection of such processes, and is almost exclusively used in patients receiving neurosurgical intensive care treatment. In the early postoperative period, when the patient is on the ventilator, comatose or sedated, ICP monitoring helps to detect postoperative or posttraumatic complications such as pneumocephalus, rebleeding, hydrocephalus/hygroma or edema if undetected, could result in herniation and death. PHYSIOLOGY OF INTRACRANIAL PRESSURE Global ICP is the pressure exerted on the dura mater by the contents of the cranial vault. Pressure is built up entirely by arterial inux, hence ICP decreases to central venous pressure (CVP) level when arterial perfusion ceases. Under controlled ventilation, the ICP curve looks quite similar to the CVP curve (Figure 1). ICP reects the sum of three partial pressures: ICP pcerebrum pblood pCSF This sum depends on posture and may increase signicantly e.g. during coughing or sneezing. Such an increase is, however, transient and not considered to be intracranial hypertension. Beyond that, ICP remains rather constant. When one partial pressure increases for example pcerebrum with the development of cerebral edema the other partial pressures (notably pCSF) will change in the opposite direction, keeping ICP constant. This interactivity considering volumes, not partial pressures was described by Burrows in 1846 and is referred to as the MonroKellie-hypothesis. Once reserve spaces are exhausted i.e. when most of CSF is displaced into the extracranial compartment any further increase of intracranial content will raise ICP. The pressure volume curve, following an exponential course, reects that relation. The skulls capability to accommodate mass challenges is termed elastance (DP/DV; reciprocal of compliance). Elastance is highest in children and lowest in older patients due to cerebral atrophy. The signal obtained with an ICP monitor is more complex. The ICP wave form is made up of at least three independent components: pulse wave, ventilatory wave and vasogenic wave(s). As a resultant of arterial perfusion, ICP uctuates with the hemodynamic cycle; in ventilated patients this is superimposed on the pressure cycle produced by the ventilator. The ICP number displayed by monitoring devices, is the mean pressure, calculated (depending on the software used) over one or more hemodynamic and ventilatory cycles. It is the mean value of a complex, cyclic wave form. The relationship between the arterial pressure input signal and the ICP output signal, the intracranial

Prevention and treatment of intracranial hypertension 519

Figure 1. Intracranial pressure (ICP) versus time. ICP uctuates in concert with ventilation (capnogram, end-tidal carbon dioxide pressure, etpCO2) and pulse rate (arterial blood pressure, AP) (pig model).17 The lack of increase in the cardiogenic ICP amplitude during inspiration (positive pressure ventilation) indicates physiological intracranial elastance, or sufcient compensatory reserve. PAP, pulmonary artery pressure; IOP, intraocular pressure; CVP, central venous pressure.

transfer function, denes the frequency response for the cranial cavity. This response reects the physical properties of cranial contents. The derived variable ICP is the average of all oscillations, derived from the intracranial contents. The frequency of these oscillations reects the fundamental frequency of cranial contents or their modication by physiological or pathological impact. This impact changes the intracranial transfer function and thus the ICP wave-form response to arterial input (Figure 2). PATHOLOGY In the supine position ICP should be somewhere below 10 mmHg. Any persistent (i.e. lasting more than 5 minutes) increase to !20 mmHg is considered intracranial hypertension. Intracranial hypertension may be mild (2029 mmHg), moderate (3040 mmHg) or severe (>40 mmHg). While transient intracranial hypertension even as high as 100 mmHg! is tolerated well by the healthy brain, a persistently high ICP is detrimental in itself, exerting pressure on neurons and shearing forces on the brain. Beyond that, as one determinant of cerebral perfusion pressure (CPP), which equals mean

520 J.-P. A. H. Jantzen

Arterial pressure wave

Cranium

Intracranial pressure wave

Mechanical properties of cranial content Mechanical properties of Transfer function arterial vessel walls Cerebral vasculature venous vessel walls Mechanical properties of

Figure 2. The relationship between the arterial pressure input signal and the intracranial pressure (ICP) output signal.26

arterial pressure (MAP) minus ICP, ICP may limit cerebral blood ow (CBF CPP: cerebrovascular resistance, CVR) once MAP has fallen below the lower limit of autoregulation. Extended phases of very high ICP, known as Lundberg A-waves, are ominous signs of intracranial pathology, heralding poor outcome. Lundberg B- and C-waves are less well dened and less indicative of poor prognosis. Under pathological conditions, ICP is not necessarily homogeneous within the cranium owing to anatomical compartmentalization.1 The falx cerebri and the tentorium cerebelli resist an even distribution of pressure exerted by mass lesions. The ensuing pressure gradients may be as high as 30 mmHg and produce damaging shearing forces and subsequent herniation. Associated mass movements may block CSF circulation with a subsequent further increase in ICP. To identify the cause of intracranial hypertension, analysis of the time course is helpful. An ICP increase within seconds is frequently a stress response, elicited by coughing, ghting the ventilator, endotracheal suctioning, or simply inadequate positioning of the head (Figure 3). An increase within minutes may be caused by intracranial bleeding (arterial rebleeding at the site of surgery, ruptured aneurysm, epidural hematoma) or other vascular accidents (e.g. acute occlusion of the sagittal sinus). Cerebral swelling early posttraumatic intracranial hyperemia due to vasoparalysis may also develop rapidly. Its signicance (or existence), however, is unclear. Intracranial hypertension subsequent to capillary or venous bleeding at the site of surgery, pneumocephalus, CSF outow obstruction hygroma or hydrocephalus or cerebral edema may take hours or days to develop. Some space-occupying intracranial pathologies, e.g. focal contusions or low-pressure-hydrocephalus, may not increase ICP at all, notably in elderly patients. The adverse effects of intracranial hypertension mandate early intervention. Prerequisite to lowering an increased ICP is measuring it; all ICP-monitoring techniques are, however, invasive and require a clear indication. INDICATIONS FOR MONITORING In general terms, ICP should be measured whenever manifest or impending intracranial hypertension is assumed, but must be limited to patients that may benet from

Prevention and treatment of intracranial hypertension 521

mmHg 70 60 50 40 30 20 10

Head rotation Head rotation Head flexion Jugular venous compression right left left right

bilateral

Figure 3. Effect of head position and jugular compression on intracranial pressure (ICP). Hulme A, Cooper R. In: Beks J, Bosch DA, Brock M (eds): Intracranial pressure III. Springer, Berlin Heidelberg New York, 1976.

ICP management. This excludes, for example, patients with infaust prognosis. For patients with severe head trauma, indications are proposed by the Brain Trauma Foundation (BTF).2 According to current (2007) recommendations, there is level-II evidence to support ICP monitoring in comatose patients Glasgow Coma score (GCS) 8 with pathological ndings in the cranial computed tomography (CT) scan. Level II evidence suggests that intracranial pressure (ICP) should be monitored in all salvageable patients with a severe traumatic brain injury (TBI; GCS score of 38 after resuscitation) and an abnormal CT scan. An abnormal CT scan of the head is one that reveals haematoma, contusion, swelling, herniation, or compressed basal cisterns. Level III evidence suggests that ICP monitoring is indicated in patients with severe TBI with a normal CT scan if two or more of the following features are noted at admission: age over 40 years, unilateral or bilateral motor posturing, or systolic blood pressure (BP) 90 mmHg. Beyond these recommendations concerning TBI, indications are less well dened, and there is ongoing debate as to the effect of ICP monitoring on outcome. Following intracranial surgery, the most common complications are pneumocephalus and surgical bleeding. Pneumocephalus is present in almost every patient having undergone craniotomy and may persist for up to 2 weeks.3 Bleeding occurs usually at the site of the operation and is facilitated by poor hemostasis. Both conditions are readily detected in awake patients displaying clinical signs of intracranial hypertension, but not in patients requiring postoperative ventilatory support. In the authors institution, indications are rather liberal in comatose patients on controlled ventilation. Besides neurotrauma, ICP monitoring is considered in patients with malignant media infarction, in higher Hunt-and-Hess grade (IIIIV) subarachnoid hemorrhage, intraventricular and intraparenchymal hematoma, or in Reye syndrome. ICP monitoring following major craniotomy is less frequently indicated, mainly owing to advancements in microneurosurgical techniques.

522 J.-P. A. H. Jantzen

MEASUREMENT TECHNIQUES ICP monitoring was introduced by Jenny and Guilleaume in 1951 and propagated as a useful clinical tool by Nils Lundberg in the 1960s. In general, ICP may be measured anywhere within the skull or subarachnoid compartment. A variety of measuring sites have been used and abandoned over the last four decades, probes being placed in epidural, subdural, subarachnoidal or intraparenchymal spaces. Under physiological conditions (i.e. with unobstructed CSF circulation), and the patient in the supine position, zero-point at the tragus, accurate measurement of ICP may be obtained in the lumbar subarachnoid space, a technique pioneered by Heinrich Quincke in 1905. Today, according to many guidelines, the gold standard for monitoring ICP is direct measurement in a lateral ventricle. This is also a standard technique in experimental research.4 Cannulation of a ventricle and connecting a communicating tube to an external metering device is a most versatile approach. It allows for continuous and drift-free monitoring of ICP, ICP control by drainage of CSF, assessment of elastance, and analysis of CSF for markers of glial damage (S-100, NSE), excitotoxicity (glutamate) or infection. However, this approach is invasive and carries the risk of complications (Figure 4). Positioning the catheter requires a burr hole of 8 mm and penetration of brain tissue to a depth of 5 cm (Figures 5 and 6); up to three attempts to a maximum depth of 5 cm are considered acceptable, which denitely is not atraumatic. With severely increased ICP the ventricles may be compressed, making cannulation difcult or even impossible. Intraventricular placement of the catheter fails in about 5% of cases. External ventricular drainage The technique of external ventricular drainage (EVD) was pioneered by Carl Wernicke at the end of the 19th century. In our institution, CSF drainage is frequently used in patients with aneurysmal subarachnoid hemorrhage (SAH) and traumatic brain injury, especially when TBI is complicated by traumatic subarachnoid hemorrhage (tSAH). In these patients, circulation of CSF between the intra- and extracranial compartments

Figure 4. Complications associated with ventricular catheter placement. From Chesnut RM, Marshall LF: Intracranial pressure monitoring and cerebrospinal uid drainage. In: Benumof JL (ed) 1992, Clinical Procedures in Anesthesia and Intensive Care, Philadelphia, Lippincott with permission.

Prevention and treatment of intracranial hypertension 523

Figure 5. Contemporary techniques of intracranial pressure (ICP) monitoring. Parenchymal probe (left), ventricular catheter (right). From Schweitzer JS, Bergsneider M, Becker DP: Intracranial pressure monitoring. In: Cottrell JE, Smith DS (eds) 1994, Anesthesia in Neurosurgery, 3rd ed, St. Louis Mosby with permission.

may be blocked; drainage of CSF is the most effective means of controlling ICP. Of patients with SAH and ventricular blood, 66% show neurological improvement when treated with EVD.5 The ventricular catheter is connected to a drainage bag via a three-way stopcock. The position of the collecting bag or the highest point of the line in relation to the tragus determines the set value of ICP. When this is reached, CSF spills over, maintaining the set ICP. It must be kept in mind that height of the collecting bag relative to the tragus reects cmH2O, while ICP is measured as mmHg. Accordingly, if 15 mmHg is the intended spillover pressure, the collecting bag must be positioned about 20 cm above the tragus. Great caution must be exercised to adapt the setting whenever the patient is repositioned. Otherwise inadvertent (iatrogenic) intracranial hypertension or ventricular collapse may ensue, both complications being equally dangerous. Whenever an exact reading of ICP is required the direction of the stopcock is changed such that the ventricular catheter is connected to the pressure transducer. The most frequently observed complications of EVD are malfunction and infection. Other complications include disconnection, CSF leakage or malpositioning of the catheter tip. Malfunction must always be suspected when cardiogenic or ventilatory oscillations of the ICP curve are diminished or lost. This happens in 25% of patients with ventricular blood, and in 15% of other patients.5 The catheter may be cleared by injection of 0.1 mL saline. If that fails twice, no more attempts should be made because of increased risk of infection; the catheter then needs to be removed. In 3% of EVD cases surgical repositioning is required. The rate of infection is time-dependent, being rather low for up to 4 days and increasing signicantly after day 5; overall, the average rate of ventriculitis is 10%. In neurotrauma with open skull fractures the rate may be as high as 40%. Neither routine ventricular catheter exchange nor prophylactic antibiotic use for ventricular catheter placement is recommended to reduce infection.6 When antimicrobial treatment is indicated, combined intravenous and intrathecal administration is most effective (cephalosporin and aminoglycoside, according to the antibiogram). Naturally, EVD should be removed as early as possible. Factors to be considered before removal include the following:     clinical condition (comatose versus awake); CSF volume per 24 h; ICP; CCT.

524 J.-P. A. H. Jantzen

When ICP is below 20 mmHg for 24 h without treatment, CSF drainage is <50 mL/ day, and no hydrocephalus is present, an EVD can generally be removed. Following TBI, most variables indicate that the 5th day is the worst. This must be kept in mind when the timing of weaning is considered. When removal criteria are not fullled after 1 week, exchanging the catheter should be considered. Although the overall incidence of complications is low (hematoma requiring surgical removal <0.5%), less-invasive monitoring techniques are increasingly used. Epidural measurement is popular in Germany (Spiegelberg probes ## 1 and 2), but is considered less accurate than subdural and subarachnoid sites e.g. in the BTF-guidelines.7 One manufacturer offers an epidural ICP monitor that may be used to calculate intracranial elastance semi-automatically (Spiegelberg; Compliance-Monitor CMP 27.1). This is valuable information; however, its impact on patient outcome has not yet been studied. Parenchymal probes Tissue probes (e.g. Integra Camino 1104B, Codman microsensor) are less invasive than ventricle catheters, require a burr hole of only 4 mm and penetrate the tissue by <15 mm (Figure 7). The signal obtained is somewhat dampened as compared to the ventricular catheter, and the system, once installed, cannot be re-zeroed or corrected for drift. Drift is, however, rarely signicant with the latest generation of probes. Once calibrated, measurements are accurate and independent of the patients position. The monitor may be disconnected and reconnected, e.g. before transportation or diagnostic procedures, without the need for recalibration. In patients with intact CO2 responsiveness of cerebral vessels and ICP management by end-tidal carbon dioxide pressure (etpCO2) control, this may be achieved by closed-loop feedback. In one such computer system combined with feedback-managed CSF drainage ICP serves as controlled variable and minute ventilation as regulated variable (Continuous ICP, Millennium medical group, Overland Park, Ks, USA). The technical standard for ICP monitors American National Standards Institute (ANSI)/Association for the Advancement of Medical Instrumentation (AAMI) NS28:1988/(R)2006 mandates an accuracy of 2 mmHg/10% within the range of 020/20100 mmHg. For probe placement see Figures 6 and 7. Display of Intracranial Pressure Undoubtedly, ICP is more than a number.8 The number usually displayed on a monitor is mean ICP, which is, depending on the software, somewhere between systolic and diastolic ICP, averaged over a number of ventilatory and hemodynamic cycles. This value is used for calculation of CPP and evaluation of treatment. It does not give helpful information on intracranial compensatory reserve (elastance) or prognosis. Accordingly, the course of ICP as a function of time should be displayed on a suitably sized highresolution monitor, allowing detailed analysis of curve shape and time course (IC tonoscopy/tonography). Detailed analysis of the ICP curve unveils information that is lost when only the digital number (mean ICP) is taken into account. When, during positive pressure ventilation, the pulse pressure amplitude increases with inspiration, increased elastance may be assumed (see Figure 1). Secondary indices can be derived when computer-based curve analysis is available. Such indices may be used to assess e.g.

Prevention and treatment of intracranial hypertension 525

Figure 6. Placement of a ventricular catheter. From Chesnut RM, Marshall LF: Intracranial pressure monitoring and cerebrospinal uid drainage. In: Benumof JL (ed) 1992, Clinical Procedures in Anesthesia and Intensive Care, Philadelphia, Lippincott with permission.

compensatory reserve and autoregulatory competence,8 but though promising, any impact of this derived information on patient outcome still needs to be demonstrated. THRESHOLD FOR INTERVENTION Persistent intracranial hypertension is associated with poor prognosis; in some clinical conditions such as Reye syndrome it is the primary cause of death. Particularly

526 J.-P. A. H. Jantzen

Figure 7. Placement of a parenchymal probe (Integra Camino). From Chesnut RM, Marshall LF: Intracranial pressure monitoring and cerebrospinal uid drainage. In: Benumof JL (ed) 1992, Clinical Procedures in Anesthesia and Intensive Care, Philadelphia, Lippincott with permission.

when associated with low blood pressure, high ICP may reduce CPP to an extent that jeopardizes cerebral perfusion. In order to avoid cerebral ischemia ICP needs to be controlled so as to maintain CPP >50 mmHg (range 5070 mmHg). Since CPP can be managed by manipulation of arterial pressure to a great extent, the risk of herniation is what denes the ICP intervention threshold. The goal is to balance the risks of herniation against the iatrogenic risks of over-treatment. For the patient with TBI the recommended threshold for intervention is a mean ICP of >20 mmHg (level-II evidence)9; for other patient populations with intracranial hypertension this threshold is less clear. When ICP monitoring is performed subsequent to surgical removal of a space-occupying mass, the threshold for intracranial hypertension is lower, e.g. 20 mmHg. In addition, the presence or absence of pathological waveforms (e.g. Lundberg waves) must be taken into consideration. Certain limitations must be taken into account. The ICP as displayed by the monitor depends on the technology used, and even more so on the site of the probe, i.e. proximity to the lesion. When CPP is of concern, the blood pressure transducer must be zeroed at the same level as the ICP monitor such as to reect mean arterial pressure in the circle of Willis not in the aorta. Taking the accuracy of monitors (ICP and MAP) into account it becomes clear that a calculated CPP of 60 in reality is anywhere between 50 and 70 mmHg at best! This should be considered when ICP-lowering strategies CPP concept versus Lund concept (see below) are evaluated (or disputed).

Prevention and treatment of intracranial hypertension 527

It is important to be aware that thresholds such as those given above always refer to the expected reaction of a majority of patients to a standardized perturbation of physiological processes. The clinicians decision, in contrast, must always be tailored to the needs of his individual patient which are determined by individual physiology and pathology and beyond any standardizing efforts! MEANS TO LOWER RAISED INTRACRANIAL PRESSURE Treatment of the patient with intracranial hypertension must take the cause of the hypertension into account. In some cases the cause may be quite remote from the brain, such as with abdominal compartment syndrome. Conditions like that require surgical correction and are beyond the scope of this review. Assessment starts with checking the patients position, ventilation and hemodynamic status, and clearing the airway; maintaining gas exchange and treating arterial hypotension is essential before anything else is considered. Meticulous attention must be paid to correct positioning. The previous dogma of head-up position was abandoned years ago, particularly in the US. Putting the patient in an anti-Trendelenburg position will lower both ICP and MAP. The maximum CPP is probably achieved in most patients at an angle of 30 . However, with respect to stability, nursing care, and other aspects, the optimum angle is about 15 , which is the standard in this authors institution. Even more important is the position of the head and cranio-cervical axis: The axis must be straight, the head must not be rotated, and the neck must not be exed (Figure 3). Adequate ventilatory support implies that the patient does not ght the ventilator. This is achieved by adapted modes of assisted ventilation (e.g. biphasic positive airway pressure) and analgesia and sedation. Because evidence-based medicine has not yet identied perfect ventilation or perfect analgesia and sedation, both can only be adapted to the needs of the individual patient as considered appropriate by the treating physician. When PEEP or recruitment is indicated to improve oxygenation, the effects on ICP are generally negligible; negative inotropic effects of PEEP, however, need attention if MAP decreases. Another means to improve oxygenation, which may be indicated when TBI is associated with acute lung injury, is the prone position. In our experience, turning the patient prone is not detrimental to ICP or CPP. If (re)positioning is performed gently and judiciously, ICP does not increase signicantly. The prone position may, however, unmask latent hypovolemia, which would adversely affect CPP. Accordingly, normovolemia is prerequisite to turning the patient prone. Analgesia and sedation should be titrated to achieve tolerance of ventilation. At our institution, sedation is provided with midazolam (24 mg/h), and analgesia with with sufentanil or (preferably) sufentanil plus S-ketamine. Prior to nursing procedures likely to provoke a stress response such as a endotrachel suctioning any adverse hemodynamic reaction should be preempted. We prefer to administer lidocaine at a dose of 1 mg/kg IV at approximately 3 min before the nursing intervention. If the ICP is still pathologically elevated, although CPP is !60 mmHg, ventilation is normocapnic, and patient position is correct, a surgically treatable cause (hematoma, hygroma, pneumocephalus, space-occupying contusion) must be excluded. This is usually done by cranial CT scan. When the CCT scan is negative, the most likely cause of intracranial hypertension is cerebral edema. Edema may be intracellular (cytotoxic edema) or extracellular (vasogenic edema); however, in most patients a combination of both is present. A perifocal vasogenic edema in patients with neoplastic lesions is responsive to corticosteroids; hyperosmolar agents are drugs of rst choice in cytotoxic

528 J.-P. A. H. Jantzen

edema. Intracranial hypertension may also be treated symptomatically. This may include hemodynamic augmentation, induced hypocapnea (hyperventilation), hypothermia, barbiturate coma, tris buffer or decompressive craniectomy. First-tier treatment (e.g. mannitol, hyperventilation) is usually well tolerated, but second-tier approaches (e.g. forced hyperventilation, barbiturate coma, craniectomy) are associated with signicant side-effects. HEMODYNAMIC MANAGEMENT OF INTRACRANIAL HYPERTENSION Cerebral perfusion pressure concept Intracranial blood volume (CBV) relates to cerebral blood ow (CBF), which is controlled by autoregulation. Autoregulatory mechanisms respond to changes in MAP. Within the range of autoregulation (MAP z 50150 mmHg), an increase in MAP initiates arteriolar vasoconstriction, which reduces CBV and consecutively ICP. When ICP is increased, e.g. because of cerebral edema, CPP drops. In order to maintain CBF, autoregulation lowers cerebrovascular resistance (CVR). This is achieved by vasodilation associated with an increase in CBV. The latter raises ICP, initiating a vicious cycle (Figure 8). The philosophy of the CPP concept is to reverse that cycle.10In case of intracranial hypertension, autoregulation is triggered by pharmacologically increasing MAP. Drugs of choice are norepinephrine (0.1 mg/kg/min) or dopamine (410 mg/kg/min). Subsequently, CBV is reduced by vasoconstriction and thus, ICP is lowered. In patients with TBI, the early goal was to achieve a MAP level of >70 mmHg. CPP management depends on intact autoregulation, which, however, may be disturbed under global or regional pathological conditions. In fact, in patients with severe TBI, the autoregulation is dysfunctional in most cases. Accordingly, autoregulatory competence should be tested before the CPP concept is initiated. This test is done with a small dose of epinephrine IV (0.1 mg/kg). When the induced increase in MAP is followed by a decrease in ICP, global autoregulation may be considered intact. When ICP increases in parallel with MAP, it may be unwise to apply the CPP concept. In our department, this test is carried out routinely. Patients showing autoregulatory decits, are not treated according to the CPP concept, but are switched to therapeutic approaches with the Lund concept. In patients with defective autoregulation after SAH with vasospasm, inducing arterial hypertension expectedly raises ICP. This increase is accompanied by an increase in CPP and brain-tissue oxygenation (pTIO2), justifying the concept of induced arterial

Figure 8. Vicious cycle of intracranial hypertension: high intracranial pressure (ICP) decreases cerebral perfusion pressure (CPP). This activates autoregulation. Cerebral vascular resistance (CVR) is lowered by vasodilation. Increased vascular diameters increase cerebral blood volume (CBV), which contributes to a further increase in ICP. This cycle can be reversed by increasing CPP through arterial blood pressure elevation.

Prevention and treatment of intracranial hypertension 529

hypertension.11 In vasospastic patients, maintaining cerebral blood ow does have priority over lowering ICP. Hyperoncotic treatment This treatment modality for patients with TBI has been pioneered in the Swedish university town of Lund.12 It is widely used in Scandinavia, but is less popular in other parts of Europe, and almost unknown in the US. There are no clinical studies comparing the Lund with the CPP concept, and published reports are anecdotal or of low evidence. The concept assumes a pivotal importance of vasogenic edema and colloid-osmotic pressure. The latter gains importance when the bloodbrain barrier (BBB) is disrupted and uid exchange cannot be effectively controlled by transcapillary crystalloid osmotic pressure. In order to reduce transmural/hydrostatic pressure in cerebral vessels, arterial pressure and cardiac output are decreased with b-receptor antagonists and a2-receptor agonists. In order to reduce capillary pressure, precapillary arterioles are constricted with dihydroergotamine (DHE). In addition, DHE will constrict cerebral veins, signicantly reducing CBV (60% of CBV is represented by the venous vascular bed!), and consequently ICP.13 The ICP-lowering effect of DHE (4 mg/kg intravenously) can be striking (Figure 9). With neurotrauma, it is important to recognize that DHE, in contrast to other vasoconstrictors, does not reduce CBF. The importance of colloid-osmotic pressure (COP) with respect to cerebral edema is poorly dened. Improved outcome with higher COP has as yet been shown in animal models, only.14 Our experience indicates that using the Lund concept may indeed lower ICP when increasing CPP and plasma osmolarity is not effective. Specic complications of DHE like ergotism16 must, however, be kept in mind.

Figure 9. Management of intracranial hypertension according to the Lund concept. Intracranial pressure (ICP) increases despite induced arterial hypertension (CPP concept), mannitol administration (Mannit), and buffering (Tris). Then the ICP decreases rapidly to dihydroergotamine (DHE). MAP, mean arterial pressure; CPP, cerebral perfusion pressure; SvjO2, jugular venous oxygen saturation.

530 J.-P. A. H. Jantzen

In cases that do not respond to hemodynamic management alone, ventilatory and pharmacological means are indicated for controlling ICP. These therapeutic modalities are classied as rst- or second-tier treatment, depending on the balance of desired versus undesired effects. Accordingly, second-tier treatment is effective, but invasive and burdened with side-effects. MANAGEMENT APPROACHES TO REFRACTORY INTRACRANIAL HYPERTENSION First-tier treatment Hyperventilation CO2 is a powerful cerebral vasodilator (Figure 10). The sensitivity of cerebral vessels to changes in paCO2 (CO2 reactivity) is an important physiological control system. Blood ow is directed away from temporarily less active regions towards regions with higher metabolic activity. On the average, lowering paCO2 by 1 mmHg reduces CBF by 4%. Hyperventilation lowers paCO2, resulting in respiratory alkalosis and subsequent vasoconstriction; accordingly, that effect subsides when pH equilibrium is re-established. When vasoconstriction is pronounced, intracranial blood volume will decrease and lower ICP; however, the iatrogenically increased cerebrovascular resistance (CVR) may reduce cerebral blood ow, possibly below the ischemic threshold. This is particularly important in known low-CBF conditions, such as severe TBI or vasospasm. To avoid cerebral ischemia, ventilation should be adjusted to a paCO2 of not less than 30 mmHg. If a lower partial pressure (2025 mmHg) is deemed necessary to control ICP, additional cerebro-metabolic monitoring is indicated. This may include sjvO2, pTiO2, near-infrared spectroscopy (NIRS) or microdialysis. In the authors department sjvO2 was the standard until the mid-1990s, then it was largely replaced by pTiO2. NIRS is currently under investigation, microdialysis is limited to study protocols. When induced hypocapnea is to be monitored by end-tidal capnometry, the arterialend-tidal pCO2 gradient must be taken into account. This gradient mainly reects ventilatory dead space and is about 5 mmHg (Figure 11). In hemodynamically unstable patients and those with multiple injuries a (much) higher gradient should be expected, measured, and taken into account. Once the arterialend-tidal gradient has stabilized, ventilatory control may be facilitated by a closed-loop feedback system.17

Figure 10. Effects of hypercapnea on intracranial pressure (ICP). ICP changes parallel uctuations of endtidal carbon dioxide pressure (etpCO2) (pig model).17

Prevention and treatment of intracranial hypertension 531

paCO2

>
Diffusion
02 mmHg

pACO2

>>
Dead space
5 mmHg

petCO2

Figure 11. The arterialend-tidal pCO2 gradient. paCO2, arterial partial pressure of CO2; pACO2, alveolar partial pressure of CO2; petCO2, end-tidal partial pressure of CO2.

CO2 reactivity is prerequisite to hyperventilation therapy. Compared to autoregulation of CBF, CO2 reactivity is more stable and is maintained even in the traumatized or otherwise injured brain. With major intracranial derangement, however, CO2 reactivity may be lost, and hyperventilation becomes ineffective. Such a condition usually heralds poor prognosis. It must be kept in mind that controlled hyperventilation and forced hyperventilation in particular is more a rescue maneuver than a standard treatment. Hyperosmolar therapy When ICP is high despite adequate hemodynamic and ventilatory management, the administration of mannitol is indicated. Alternatively, glycerol (10%) or sorbitol (40%) may be used; however, mannitol 20% has become a standard. Although there is ongoing debate as to the mechanisms of mannitol-induced ICP control, increasing plasma osmotic pressure, and thus the osmotic gradient between plasma and brain tissue, is one factor. Improved rheology subsequent to reduced viscosity and an increase in MAP may further contribute to the benecial effect. The osmotic gradient is believed to drain uid from the brain and make cells shrink. This mechanism implies that the effect is time-limited. As soon as a new osmotic equilibrium is reached, uid translocation will stop. It also becomes clear that this mechanism requires an intact blood brain barrier. If this barrier is damaged regionally or locally no osmotic gradient will be established. On the contrary, the osmotic agent may cross the defective barrier, enter the cells and increase the pressure there. This rebound effect may draw plasma into the cells, increasing edema. In this authors experience, that effect is overrated in the literature and of minor clinical signicance. However, mannitol should be administered judiciously. A single dose should not exceed 0.251.0 g/kg and should not be repeated more than twice. Normotension and normovolemia must be maintained at all times. Plasma osmolarity should not increase beyond 320 mOsm/L, the threshold to renal tubular necrosis. In the absence of ICP monitoring, mannitol should primarily be considered a rescue medication. Another way to increase plasma osmolarity is the administration of hypertonic saline (HS). The use of HS for ICP control resulted from initial studies on small-volume resuscitation. Hypertonic saline solutions were used in pre-hospital trauma care and improved outcome most notably in patients with TBI. These ndings stimulated research on the effects on increased ICP.18 Proposed benecial effects of HS may arise from more than one mechanism. Like mannitol, HS produces an osmotic gradient, resulting in shrinkage of brain tissue and reduction in ICP. Furthermore, HS augments volume resuscitation and increases circulating blood volume, MAP and CPP. Animal and human studies suggest that HS is a potential therapeutic agent to assist with medical treatment of patients with TBI. It may have a place as osmotherapy to reduce brain size, predominately of the uninjured brain, and has several potential advantages over mannitol; e.g. the osmotic threshold for renal tubular necrosis is 360 mOsm/L with

532 J.-P. A. H. Jantzen

HS, increasing the margin of safety. HS may be considered a therapeutic adjunct to the medical management of cerebral edema or intracranial hypertension, awaiting denitive evidence to support routine use.19 Beyond effects on ICP, HS may improve cerebral perfusion by reversing endothelial swelling of capillaries. This occurs within low-ow areas after TBI. Second-tier treatment When intracranial hypertension persists, a surgically treatable cause must be assessed by CT scan. If such a cause hematoma, hygroma, hydrocephalus, contusion is excluded, other forms of treatment must be considered. Forced hyperventilation Hypoventilation down to a paCO2 of 2025 mmHg may be readily achieved, but the effect may be short-lived and carries the risk of cerebral ischemia. It is mostly a measure to buy time for further diagnostic procedures or with impending herniation to survive the trip to the theater. Barbiturate coma Barbiturates are known to suppress the cerebral metabolic rate for oxygen (CMRO2), terminate convulsions, scavenge free oxygen radicals, and obtund a cerebral hyperthermic response to ischemia. Only when cerebrovascular CO2 reactivity is maintained, is inducing barbiturate coma an option. Another mandatory prerequisite is hemodynamic stability. Barbiturates have not been shown to improve outcome, but they do lower the ICP. The treatment should be monitored by electroencephalogram (EEG), which is superior to plasma-level-guided administration. The desired EEG rhythm is a burst-suppression pattern. Barbiturate dosages are 5 mg/kg thiopental over 30 min, followed by an infusion of 5 mg/kg/h. Less experience has been gained with hydroxybarbiturates (e.g. methohexital) in comparison to thiopental. However, they may be advantageous with respect to accumulation, hemodynamic or immunological suppression. Alternatively, propofol may be used at a dosage not exceeding 5 mg/kg/h. Awareness of the rare, but potentially lethal complication, the propofol infusion syndrome, is necessary. Some second tier options notably forced hyperventilation and barbiturate coma may be bypassed (See note 4 in the algorithm). This applies primarily to younger patients with limited primary injury and excessively high ICP when time is brain. Decompressive craniectomy In consideration of the physiology of ICP, eliminating the rigid envelope (craniectomy, duraplasty) should be a most effective way to reduce ICP. The procedure is invasive, however, and not all neurosurgeons are ready to perform such radical surgery of uncertain benet. Most reports concerning intracranial hypertension subsequent to malignant media infarction or TBI are anecdotal or case series. However, recent data appear promising enough to justify prospective evaluation.20 Currently, a multicenter study is under way in Europe which is trying to identify the position of decompressive craniectomy in TBI: RESCUEicp.21 In our institution bilateral craniectomy is performed when all less-invasive treatment modalities have failed and the patients condition allows for major surgery. This is particularly the case in younger patients with initially reactive pupils and no brain-stem injury. If surgical decompression is taken into

Prevention and treatment of intracranial hypertension 533

consideration it should be performed without undue delay. Timing of the indication is not a matter of hours or days, but is a consequence of the patients response to rsttier or other less-invasive treatments. When these treatments have failed and craniectomy is considered useful, it should be performed promptly. Delaying surgery for wait-and-see, may harm the patient and discredit the procedure. Other second-tier treatment Hypothermia. There is convincing evidence from preclinical studies that mild hypothermia is associated with improved neurological outcome following regional cerebral ischemia, TBI, or cardiac arrest. Such evidence was conrmed in patients with cardiac arrest; the situation is less clear with respect to aneurysm surgery or TBI. Mild hypothermia was, however, shown to lower ICP in patients with TBI. The underlying mechanism is not fully understood. Hypothermia-induced reduction of CMRO2 may be a factor, such that under conditions of articial ventilation with xed minute volume a decrease in CMRO2 will result in hypocapnea. This effect my go undetected when the blood-gas analyzer is not adjusted for in-vivo temperature. In the authors institution mild hypothermia is induced in patients with TBI and SAH. Temperature is lowered by convective cooling, aiming at a bladder temperature of 3536  C for 48 h following trauma or hemorrhage. The major objective of this protocol is not necessarily to achieve hypothermia, but rather to avoid inadvertent hyperthermia. Tris buffer. Acidosis is detrimental in patients with cerebral ischemia. One aspect is that acidosis removes the Mg2 lock from the NMDA receptor, facilitating excitotoxicity. When the pH drops below 7.0, acid-sensitive ion channels a group of calciumpermeable acid-sensing ion channels may be activated, allowing intracellular Ca2 to increase further.22 In contrast, low pH protects energy-dependent glutamate porter systems and thus delays onset of excitotoxicity. Tris buffer corrects intracellular acidosis and increases the buffering capacity of CSF. Accordingly, it has been used in combination with controlled hyperventilation. In cerebral ischemia it reduces tissue lactate concentration, edema and infarct size. Tham (trishydroxymethylaminomethane) was shown to reduce ICP in patients with TBI.23 However, it was also shown that the decrease in ICP is accompanied by a decrease in cerebral tissue oxygenation in some patients. Due to unpredictable effects on MAP15, the net effect on CPP and PTiO2 requires close observation. In our institution, Tham (1 mg/kg intravenously) is given ex juvantibus to patients with intracranial hypertension who do not respond to rst-tier treatment. Steroids. Steroids have been in use for neurosurgical patients for decades. Drugs, dosages and indications have varied with time. After steroids were eliminated from the neurotraumatologists armamentarium following a negative recommendation by the BTF18 and the CRASH trial24, they sneaked back via the spinal-trauma discussion; 10% of patients with TBI have an associated spinal trauma. With respect to intracranial hypertension following TBI, there is no indication for steroids. However, classical indications, such as perifocal edema surrounding intracranial tumors or edema associated with cerebral abscess, remain valid. Indomethacin. The non-selective cyclo-oxygenase inhibitor indomethacin is known to reduce ICP following TBI or intracranial surgery. ICP is assumed to fall subsequent to

534 J.-P. A. H. Jantzen

Nordstadt ICP-Algorithm

GCS < 8

CCT
Surgical mass

Intubation, ventilation, art. line, cv line, normovolemia hemodynamic stabilization No surgical mass

Surgery ICU ICP monitoring indicated ? no

1

ICU

yes ICP > 20-25 mmHg ?

Observation / Standard ICU care

no

yes

Rev. Trendelenburg Pos.: 15+ Head and craniocervical axis straight Ventilation: paCO2 35-36 mmHg saO2 > 95% Temperature < 36,5 MAP > 70 mmHg ICP > 20-25 mmHg ? no yes

Ext. ventr. drain Parenchymal probe CSF drainage

ICP > 20-25 mmHg? no yes Clinical assessment? COP?

Hyperventilation paCO2 30 mmHg ICP > 20-25 mmHg? no yes

Hyperosmolar treatment Mannitol 0,25-1,0 g/kg; n < 3x ICP > 20-25 mmHg? no yes

Autoregulation-test Norepinephrin 0,2 g IV

Figure 12. Nordstadt intracranial pressure (ICP) algorithm for the management of intracranial hypertension.

Prevention and treatment of intracranial hypertension 535

Autoregulation-test Norepinephrin 0,2 g IV ICP ? no CPP-Concept

2

yes Lund-Concept3

ICP > 25 mmHg ? no yes Clinical assessment? CCT/TCD? Osmolarity?

Forced hyperventilation paCO2 20-25 mmHg cerebral oximetry ICP > 25 mmHg ? no yes

Barbiturate coma Thiopental 5 mg/kg EEG: Burst suppression ICP > 25 mmHg? no yes

Thiopental 5 mg/kg*h

Temp. 35C Tris-buffer 1 mg/kg DHE 4 g/kg Indomethacin 0,5 mg/kg

ICP > 25 mmHg? no yes

Case conference

Clinical assessment? CCT / TCD / MRI Terminate treatment

Withdraw treatment

Decompressive craniectomy

ICU / Rehabilitation

Donor program

1: Primary injury must allow acceptable quality of life, contraindications be observed. GCS < 8 and pathological CCT (Hematoma, edema, contusion, midline shift). Or: GCS < 8 and age > 40 / SAP < 90 mmHg / motor posture (2 of 3). 2: MAP > 80 mmHg. Norepinephrin 0,1 g/kg*min or dopamin 4-10 g/kg*min 3: CPP > 50 / MAP > 70 mmHg. DHE 4 g/kg, albumin, metoprolol and clonidine as required. 4: Alternative pathway (see text for details)

Figure 12. (Continued).

536 J.-P. A. H. Jantzen

a decrease in CBV caused by vasoconstriction of small resistance vessels. This may be due to inhibition of vasodilator arachidonic-acid metabolites.25 Uncertainty persists as to whether or not the decrease in CBF may render the brain ischemic. However, cerebral ischemia has so far not been shown in clinical studies, and some authors ascribe neuroprotective properties to indomethacin. For the time being, the routine use of indomethacin has to await further clinical studies. Treatment algorithm The sequence and use of rst- and second-tier treatment as practised in this authors institution is summarized in our management of intracranial hypertension algorithm (Figure 12). Concluding remarks Measurement of ICP is the essential component of most brain-monitoring systems. After almost 50 years it has stood the test of time. Despite enthusiastic attempts to introduce potential alternatives, notably monitors to assess cerebral metabolism, intracranial tonometry has proved to be robust and reliable. It is only moderately invasive and may be performed in many hospitals, including those without neurosurgical services.

Practice points  intracranial pressure (ICP) is the pressure exerted on the dural envelope by contents of the cranium  ICP is non-uniform because of hydrostatic gradients and compartmentalization of the intracranial space  under pathological conditions, regional pressure differences may be !30 mmHg; the place of the ICP probe must therefore be taken into consideration when interpreting measured values  ICP may be regarded as the sum of three partial pressures, reecting three partial volumes: brain tissue (80%), CSF (812%) and blood: ICP pcerebrum pCSF pblood  normal ICP is approximately 510 mmHg; values of 2030 mmHg/30 40 mmHg indicate mild/moderately increased ICP, values above 40 mmHg severely increased ICP  under physiological conditions the ICP is kept constant within a narrow range; the increase of the partial pressure of one component (brain tissue, CSF) is compensated by the contradirectional change of the intracranial blood volume (MonroKellie hypothesis)  the ICP is one determinant of cerebral perfusion pressure (CPP MAP ICP) and thus is an indirect determinant of cerebral blood ow (CBF CPP:CVR)  continuously increased ICP injures the brain, directly and indirectly (perfusion damage); in patients with traumatic brain injury or aneurysmal subarachnoid hemorrhage, continuously raised ICP is an independent indicator of poor prognosis

Prevention and treatment of intracranial hypertension 537

 control of pathologically increased ICP presupposes its measurement; this can only be accomplished invasively. Established sites for placement of ICP probes include the epidural, ventricular and intraparenchymal space. Measurement techniques are based on the principle of the communicating tube (e.g. in the lateral ventricle), electronically or ber-optically  direct ICP measurement with a catheter placed into the lateral ventricle (the ofcial gold standard) allows for pressure control by CSF drainage, assessment of intracranial elastance (DP/DV) as well as microbiological and metabolic analysis  placement of that catheter is traumatic (burr hole diameter 4.4 mm, depth of brain passage 4.5 cm) and may be complicated by bleeding (1.4%) and contamination/infection (3%)  ICP measurement with a ber-optical or electronic sensor in brain parenchyma is less traumatic (burr hole 2.8 mm, depth of brain passage 1.5 cm) and is associated with fewer complications  not monitoring ICP while treating intracranial hypertension can be deleterious and result in poor outcome  ICP measurement is indicated whenever intracranial hypertension is assumed; the result of ICP monitoring may prompt cranial CT scanning, but CCT is by no means a substitute for ICP monitoring

Research agenda  a randomized clinical trial of ICP monitoring to assess its impact upon outcome is desirable, but probably unlikely to be performed  technical standard for ICP monitors should include in-vivo clinical ICP drift measurement; in-vitro testing of devices does not necessarily reect clinical performance  the difference in pressure between ventricular and parenchymal ICP measurement needs to be elucidated; both positive and negative differences have been reported in clinical studies  the most suitable site for parenchymal monitoring needs to be identied: near to or remote from a cerebral lesion?  further improvement in ICP monitoring technology should focus on developing multiparametric ICP devices, possibly combining ventricular CSF-drainage and parenchymal-ICP-monitoring options  ICP may closely relate to the risk of herniation; a method to estimate herniation pressure should be developed  ICP monitors that allow for high-resolution display (tonoscopy) are needed; tonoscopy must allow wave-form analysis and computer-based calculation of derived variables, e.g., ratio of pulse-pressure amplitude to mean ICP. The relevance of derived variables for outcome assessment should be examined in prospective studies

538 J.-P. A. H. Jantzen

REFERENCES
*1. Mindermann T, Reinhardt H & Gratzl O. Signicant lateralisation of supratentorial ICP after blunt head trauma. Acta Neurochirurgica 1992; 116: 6061. *2. Brain Trauma Foundation. Indications for intracranial pressure monitoring. Journal of Neurotrauma 2007; 24(supplement 1): S37S44. 3. Reasoner DK, Todd MM, Scamman FL et al. The incidence of pneumocephalus after supratentorial craniotomy. Anesthesiology 1994; 80: 10081012. 4. Jantzen J-P, Hennes HJ & Wallenfang T. A porcine model for direct and continuous ICP measurement. European Journal of Anaesthesiology 1992; 9: 269. 5. Milhorat TH. Acute hydrocephalus after aneurysmal subarachnoid hemorrhage. Neurosurgery 1987; 20: 1520. 6. Lozier AP, Sciacca RR, Romanoli M et al. Ventriculostomy-related infection: a critical review of the literature. Neurosurgery 2002; 51: 170182. *7. Brain Trauma Foundation. Intracranial pressure monitoring technology. Journal of Neurotrauma 2007; 24(supplement 1): S45S54. *8. Czosnyka M, Smielewski P, Timofeev I et al. Intracranial pressure: More than a number. Neurosurgical Focus 2007; 22: E10 17. *9. Brain Trauma Foundation. Intracranial pressure thresholds. Journal of Neurotrauma 2007; 24(supplement 1): S55S58. *10. Rosner MJ, Rosner SD & Johnson AH. Cerebral perfusion pressure: management protocol and clinical results. Journal of Neurosurgery 1995; 83: 949962. 11. Muench E, Horn P, Bauhuf C et al. Effects of hypervolemia and hypertension on regional cerebral blood ow, intracranial pressure, and brain tissue oxygenation after subarachnoid hemorrhage. Critical Care Medicine 2007; 35: 18441851. *12. Asgeirsson B, Grande PO & Nordstrom CH. A new therapy of post-trauma brain oedema based on haemodynamic principles for brain volume regulation. Intensive Care Medicine 1994; 20: 260267. 13. Asgeirsson B, Grande PO, Nordstrom H et al. Cerebral haemodynamic effects of dihydroergotamine in patients with severe traumatic brain lesions. Acta Anaesthesiologica Scandinavica 1995; 39: 922930. 14. Drummond JC, Patel PM, Cole DJ & Kelly PJ. The effect of the reduction of colloid oncotic pressure, with and without reduction of osmolality, on post-traumatic cerebral edema. Anesthesiology 1998; 88: 9931002. 15. Duthie SE, Goulin GD, Zornow MH et al. Effects of THAM and sodium bicarbonate on intracranial pressure and mean arterial pressure in an animal model of focal cerebral injury. Journal of Neurosurgical Anesthesiology 1994; 6: 201208. 16. Gupta VL & Mjrndal TO. Gangrene and renal failure caused by dihydroergotamine used to treat raised intracranial pressure following head trauma. Acta Anaesthesiologica Scandinavica 1996; 40: 389391. 17. Jantzen J-P. Closed loop feed back control of petCO2. An experimental study in pigs. Journal of Neurosurgical Anesthesiology 2007; 19: 346. *18. Brain Trauma Foundation. Hyperosmolar therapy. Journal of Neurotrauma 2007; 24(supplement 1): S14S20. 19. White H, Cook D & Venkatesh B. The use of hypertonic saline for treating intracranial hypertension after traumatic brain injury. Anesthesia & Analgesia 2006; 102: 18361846. 20. Timofeev I, Kirkpatrick PJ, Corteen E et al. Decompressive craniectomy in traumatic brain injury: outcome following protocol-driven therapy. Acta Neurochirurgica. Supplement 2006; 96: 1116. 21. Hutchinson PJ, Corteen E, Czosnyka M et al. Decompressive craniectomy in traumatic brain injury: the randomized multicenter RESCUEicp study. Acta Neurochirurgica. Supplement 2006; 96: 1720. 22. Xiong ZG, Zhu XM, Chu XP et al. Neuroprotection in ischemia: blocking calcium-permeable acidsensing ion channels. Cell 2004; 118: 687698. 23. Pfenninger E, Lindner KH & Ahnefeld FW. An infusion of THAM (trishydroxymethylaminomethane) as therapy to lower increased intracranial pressure in acute craniocerebral injuries. Anaesthesist 1989; 38: 189192. 24. Sauerland S & Maegele M. A CRASH landing in severe head injury. Lancet 2004; 364: 12911292. 25. Rasmussen M. Treatment of elevated intracranial pressure with indomethacin: Friend or foe? Acta Anaesthesiologica Scandinavica 2005; 49: 341350. *26. Lin ES, Poon W, Hutchinson RC et al. Systems analysis applied to intracranial pressure wave forms and correlation with clinical status in head injured patients. British Journal of Anaesthesia 1991; 66: 476482.