Traumatic brain injury (TBI) accounts for almost one-half of all trauma fatalities and has a significant impact on mortality, morbidity, and health care costs. Understanding normal cerebral physiology and metabolic derangements due to primary and secondary injuries after TBI help to delineate cellular mechanisms of injury and identify potentially beneficial therapies. Neurological monitoring and adherence to objective, specific goals for initial resuscitation and support of the TBI patient are essential. Specific therapies such as cerebrospinal fluid drainage, diuretics, hyperventilation, barbiturates, and anticonvulsant prophylaxis have been studied. Many of these interventions have been empirically based on pathophysiologic rationale and have subsequently been implemented in the context of head-injury management protocols in an attempt to improve patient outcomes. Traditionally, studies evaluating TBI therapies have been limited by design flaws. However, larger better-designed trials have recently been undertaken, and meta-analytic techniques have also been employed to help establish the relative merit of various therapies for TBI. Unfortunately, early interest in many of these therapies has not translated into clinical benefits in well-designed trials. For these reasons, experimental therapies such as induced hypothermia, corticosteroids, indomethacin, and other neuroprotective strategies such as calcium channel blockers, antioxidants, 21-aminosteroids, antagonists of excitatory amino acids, cyclosporine, magnesium, and “Lund therapy” continue to be evaluated. The purpose of this review is to provide the reader with a comprehensive review of the pathophysiology, neurological monitoring, therapeutic goals, initial resuscitation strategies, and treatment options for the TBI patient.
Trauma results in about 150,000 deaths in the United States each year, and approximately one-half of this mortality is specifically related to head injuries. The 1-year survival rate following a head injury is lower in patients over 50 years of age, possess lower initial Glasgow Coma Scores and higher Injury Severity Scores, lack pupillary responses, reveal a hematoma on CT scan, and in those who demonstrate evidence of brainstem injury. Although mortality is the principal endpoint, the long-term functional outcome following a head injury is also crucial. Tragically, approximately 200,000 people in the U.S. live with disabilities caused by traumatic brain injuries. The economic impact of these injuries is also substantial. The initial treatment cost per severe head injury is estimated to be approximately $150,000 U.S. and a 1990 report estimates total direct and indirect lifetime costs to be $158 billion U.S for all head injuries in the U.S.
It is now generally recognized that secondary as well as primary injury contributes to the morbidity and mortality of traumatic brain injury (TBI). Brain swelling and increased intracranial pressure (ICP) promote secondary injury, and it has been hypothesized that intracranial hypertension may be the primary cause of death in some patients. Aggressive management protocols, including ICP control, have been shown to reduce the overall mortality and morbidity rate. According to data from the Traumatic Coma Data Bank (TCDB), such protocols have resulted in an overall mortality reduction following severe head injury from 50% to 36%. Despite these encouraging results, several currently applied therapies remain to be properly evaluated in well-designed trials and new pharmacological strategies to control ICP, improve cerebral oxygenation, and reduce morbidity and mortality are still needed to further improve patient outcomes.
The purpose of this review is to provide an overview of cerebral physiology under normal conditions and during TBI, provide an outline of monitoring strategies for TBI patients, review current treatment strategies to control ICP and optimize cerebral oxygenation in TBI patients, and discuss experimental therapies in the pharmacological management of TBI.
Normal Cerebral Physiology
Although the brain represents only 2% of total body weight, it utilizes 20% of the systemic oxygen consumption for the oxidation of glucose. The metabolic rate of the brain is often expressed in terms of its rate of oxygen consumption (CMRO2), which is approximately 3.5 mL/100g brain/min. Approximately 15% of cardiac output is distributed to the brain (750 mL/min), and normal cerebral blood flow (CBF) in an adult remains relatively constant at 40-50 mL/100 g brain/min. Under normal conditions, the brain has little energy reserve, so CBF is tightly coupled to CMRO2 and brain function. As cerebral metabolism and oxygen consumption increase, metabolic autoregulation results in a parallel increase in CBF and cerebral oxygen delivery (CDO2). Cerebral blood flow is tightly regulated to metabolism in order to maintain oxygen supply in balance with oxygen demand, and to protect the brain from extremes in perfusion pressure due to hypertension. Cerebral blood flow autoregulation permits cerebral vasculature to alter cerebrovascular resistance (CVR) to maintain a constant CBF despite variations in cerebral perfusion pressure (CPP). Accordingly, CBF is equal to CPP/CVR. Cerebral perfusion pressure is determined by the mean arterial pressure (MAP), and ICP, thus CPP is equal to (MAP – ICP). Under normal conditions, despite variations in MAP (between 50 and 150 mm Hg) and concurrent variations in CPP, CBF remains relatively constant due to alterations in CVR. Mean arterial pressure below 50 mm Hg may result in cerebral ischemia, while MAP above 150 mm Hg may result in capillary injury or edema.
Intrancranial pressure is measured most reliably from within the ventricle and is normally less than 15 mm Hg. The skull is a rigid compartment containing brain tissue (80%) cerebrospinal fluid (CSF) (10%), and blood volume (10%). The Monroe-Kellie doctrine states that these incompressible structures within the cranial vault are in a state of volume equilibrium, and any increase in the volume of one component must be compensated for by one or more of the other components to ensure that intracranial volume remains constant. Any disturbances in intracranial volume will alter ICP, while any increase in ICP will reduce CPP and may reduce CBF if alteration in CVR via autoregulation is not functional. Factors that normally influence cerebral autoregulation include carbon dioxide and oxygen. As PaCO2 drops to between 20-80 mm Hg , there is a linear decrease in CBF. Hypocapnia decreases CBF and hypercapnia increases CBF. In fact, CBF increases 2-3% per mm Hg increase in PaCO2 above 40 mm Hg. Hypoxemia also affects autoregulation, and when PaO2 drops below 50 mm Hg, CBF doubles. The extraction of oxygen from the cerebral blood (CEO2) is defined as the difference between arterial oxygen saturation (SaO2) and jugular venous oxygen saturation (SjvO2), thus CEO2 is equal to SaO2-SjvO2. Under normal conditions, SjvO2 is 55-71%, and CEO2 is relatively constant at 24-42%. The cerebral extraction ratio (CERO2) is defined as the ratio between cerebral oxygen consumption (CMRO2) and cerebral oxygen delivery (CDO2), CMRO2/CDO2. CERO2 is practically expressed as (SaO2-SjvO2)/SaO2. Normal CERO2 is around 35% ± 10%. Major causes of decreases in SjvO2 and thus increases in CERO2 can be due to reduced cerebral oxygen delivery (CDO2) or increased cerebral oxygen consumption (CMRO2). The normal cerebral physiology can be summarized by the equations in Table 1.
Table 1. Normal Cerebral Physiology
|CBF = CPP / CVR
CPP = MAP – ICP
CBF = (MAP – ICP) / CVR
CEO2 = SaO2 – SjvO2
CERO2 = CMRO2 / CDO2
CERO2 = (SaO2 – SjvO2) / SaO2
|CBF: Cerebral blood flow
CPP: Cerebral perfusion pressure
CVR: Cerebral vascular resistance
MAP: Mean arterial pressure
ICP: Intracranial pressure
CBF: Cerebral blood flow
CEO2: Cerebral oxygen extraction
SaO2: Arterial oxygen saturation
SjvO2: Jugular venous oxygen saturation
CERO2: Cerebral extraction ratio of oxygen
CMRO2: Cerebral metabolic consumption of oxygen
CDO2: Cerebral oxygen delivery
Pathophysiology of TBI
Neurological complications from TBI can occur as a direct result of the primary injury, or may be caused by secondary injuries that follow within minutes to days. The primary injury is typically the result of a direct initial insult and the secondary injury is caused by the subsequent cascade of biochemical changes that are triggered by ischemia and result in a disruption of the normal central nervous system balance between oxygen supply and demand (Figure 1).
Figure 1. Illustration of Secondary Neurologic Injury Cascade with the Potential Role of Indomethacin and Neuroprotective Agents
Following ischemia, cells release the excitatory amino acid glutamate and adenosine triphosphate (ATP) levels will decrease. Glutamate activates N-methyl-D-aspartate (NMDA), alpha-amino-3-hydroxy-5-methyl-4-isoxazoleproprionate (AMPA), kainate, and metabotropic receptors. Activation of these receptors results in a failure of the cellular ionic pump with rapid influxes of calcium and sodium into the cell and subsequent depolarization. Calcium has several deleterious effects including cerebral vasospasm, mitochondrial dysfunction, and activation of proteases and lipases. The effect of lipase stimulation is the formation of several arachidonic acid derivatives including thromboxane A2 (TXA2), prostacyclin (PGI2), prostaglandin G2, prostaglandin H2 and leukotrienes. The resulting effects of these arachidonic acid derivatives are platelet aggregation, vasodilation and oxygen free radical production with lipid peroxidation. Decreased ATP levels leads to a diminished ion gradient with increases in lactate and subsequent acidosis and ischemia. The entire secondary injury process is a vicious cascade of biochemical changes that leads to further spread of the ischemic injury and neurological deficits. The causes of secondary injury include cerebral edema, expanding mass lesions, and cerebral vasospasm. Other factors that may potentiate secondary injury include elevated temperature, seizures, and agitation. The loss of cellular integrity from the ionic shifts may result in cytotoxic edema and vasogenic edema may also develop due to cerebral capillary endothelial damage.
The swelling of injured tissue after TBI tends to peak within 72 hours of injury and leads to cerebral edema. The most detrimental consequence of this cerebral edema is increased ICP since it leads to significant morbidity and mortality. If the CSF or cerebral blood volume cannot be reduced to compensate for the increased brain volume, the ICP will rise. The less compliant the brain is, the higher the ICP will rise for a given amount of cerebral edema. Raised ICP compresses brain tissue and decreases CPP, further increasing cerebral injury. Cerebral injury affects autoregulation, which would normally help combat changes in pressure associated with neurological injury. Autoregulation can be either impaired, or preserved at lower levels of CBF after TBI. Thus, the injured brain may be especially susceptible to hypotension that reduces CPP and causes cerebral ischemia in the early post-injury phase.
Cerebral metabolic consumption of oxygen is globally depressed in severely head-injured patients with systemic energy expenditure decreasing from 20% to 11%. The reduction in CMRO2 is proportional to the severity of coma, as reflected by the Glasgow Coma Scale (GCS). (Table 2) (19,20)
Table 2. Glasgow Coma Scale
|Eye Opening ( /4)||Spontaneous||4|
|Best Motor Response ( /6)||Obeys commands||6|
|Abnormal flexion (decorticate rigidity)||3|
|Abnormal extension (decerebrate rigidity)||2|
|Best verbal response ( /5)||Oriented||5|
|TOTAL ( /15)||(3-15)|
There are three scenarios describing cerebral oxygenation during the period of reduced CMRO2.(Table 3) In approximately 45% of patients, there is normal coupling between the CMRO2 and CBF. (10,11) Decreased neuronal metabolism and CO2 production leads to a reduction in CBF that is proportional to the reduced CMRO2 while the CERO2 remains unchanged. In this scenario, there is intact metabolic autoregulation and a balance between cerebral oxygen consumption and delivery. The second scenario is termed “oligemic cerebral hypoxia” or “misery perfusion”. Coupling between CMRO2 and CBF is lost, and CBF is low relative to the needs of the brain. If cerebral oxygen delivery is reduced due to hypotension, systemic anemia, systemic hypoxemia, high ICP, cerebral vasospasm, or marked hypocapnia, the brain responds by increasing the extraction of oxygen to compensate, the SjvO2 will fall and the CERO2 will increase. However, a point will be reached at which further decreases in CBF cannot be compensated for by increased oxygen extraction and ischemia follows, manifested by a CERO2 above 40%.
Table 3. Cerebral Oxygenation Alterations After TBI
|Intact metabolic autoregulation||¯||¯||«||Normal coupling and extraction|
|Oligemic cerebral hypoxia (misery perfusion)||¯||¯¯||Cerebral ischemia may cause secondary injury|
|Relative cerebral hyperperfusion (luxury perfusion)||¯||¯||Excess CBV may increase ICP and cause secondary injury|
CMRO2 = Cerebral metabolic consumption of oxygen, CBF = Cerebral blood flow, CERO2 = Cerebral extraction ratio of oxygen, CBV = Cerebral blood volume, ICP = Intracranial pressure
Within 6 hours of injury, there is a global reduction in CBF to ischemic levels in 13% of patients. If there is a severe reduction of CBF for a sufficient duration of time, these ischemic changes will become irreversible and lead to brain tissue infarction. Increases in cerebral oxygen delivery after this point will not result in greater CEO2 because of permanent tissue damage. Under these conditions, the SjvO2 will actually increase and the CERO2 will fall, indicating permanent damage. To prevent cerebral ischemia, treatment strategies must focus on increasing cerebral oxygen delivery (CDO2) or reducing oxygen consumption (CMRO2). Major causes of reduced CDO2 include systemic hypotension, increased ICP and thus low CPP, systemic anemia, systemic hypoxemia, cerebral vasospasm, or marked hypocapnia. Major causes of increased CMRO2 include agitation (e.g. pain, anxiety, or delirium), fever, and seizures.
The final scenario is that called “relative cerebral hyperperfusion” or “luxury perfusion”. Coupling between CMRO2 and CBF is lost, and CBF is excessive to that required to meet the metabolic needs of the brain. This increase in cerebral blood volume may contribute to a high ICP thus reductions in cerebral blood volume or overall ICP are indicated. In this scenario, the SjvO2 is high, and the CERO2 remains low.
Neurological Monitoring in TBI Patients
Neurological monitoring in TBI patients may include physical and neurological examinations, ICP monitoring, SjvO2 monitoring, electroencephalographic (EEG) monitoring, and transcranial doppler (TCD) ultrasonography. A complete physical and neurological examination including brainstem function, GCS, and cranial nerve function are indicated to establish the extent of the injury, and to establish a baseline for neurological monitoring. The GCS is used to evaluate the level of consciousness according to eye opening, best motor response, and best verbal response. The admission or last known GCS is used to grade the severity of TBI. A GCS score of 13-14 indicates mild injury, a score of 9-12 indicates moderate injury, while a score of 3-8 indicates severe injury. The GCS has also been shown to be a useful predictor of one-year survival and cognitive recovery.
Although outcome benefits of ICP monitoring have never been assessed in a prospective, randomized clinical trial, it is nevertheless considered essential for the patient with severe TBI. Increases in ICP have been associated with worse neurological outcomes, and it has been suggested that intracranial hypertension is the primary reason for death in some patients. Moreover, the use of intensive ICP-based management protocols have been associated with significant reductions in morbidity and mortality from TBI. In one prospective, randomized controlled trial, mortality was shown to be significantly lower in patients for whom ICP was successfully controlled with barbiturate therapy. However, there have been no prospective, randomized trials to compare outcomes using different treatment thresholds for ICP. The investigators of a small, uncontrolled non-randomized trial reported an associated reduction in mortality in patients whose ICP treatment was initiated at 15 mm Hg versus treatment initiated at 20-25 mm Hg. Uncontrolled co-interventions and potential biases in the trial design may explain these results. Another recently published study involved an examination of the relationship between ICP, CPP and outcome. Juul et al. reviewed these parameters in a series of 427 patients who were enrolled in the randomized, double-blind trial of Selfotel, a NMDA receptor antagonist. They reported that treatment protocols should emphasize immediate treatment of ICP greater than 20 mm Hg and that CPP greater than 60 mm Hg had very little influence on patient outcome. The likelihood of herniation due to high ICP may also depend on the location of the intracranial mass lesion. For example, lesions of the anterior temporal lobe can cause herniation at an ICP level of less than 15 mm Hg. The Brain Trauma Foundation (BTF) guidelines contain recommendations for continuous ICP monitor placement in patients with a GCS of 8 or less with an abnormal admission CT scan, or in high-risk (age greater than 40 years, motor posturing, or systolic blood pressure under 90 mm Hg) severe head injury patients with a normal CT scan. Ventricular catheters connected to an external strain gauge are considered to be the most accurate, least costly, and most reliable method for monitoring ICP. The use of these catheters also allows for the therapeutic drainage of CSF. The BTF and European Brain Injury Consortium (EBIC) Guidelines recommend 20-25 mm Hg as the upper threshold at which treatment to lower ICP should be initiated. In summary, ICP monitoring helps determine prognosis, enables early detection of mass lesions, allows monitoring of interventions to lower ICP, enables CSF drainage to reduce ICP, and enables calculation and optimization of CPP.
Cerebral ischemia may be the single most important secondary event influencing outcome following severe TBI. (33) CPP is the physiologic variable driving CBF and energy delivery, and therefore it is closely related to ischemia. Reducing elevated ICP or artificially increasing the MAP can increase CPP. While there have been concerns that strategies to increase MAP to achieve a higher CPP may increase brain swelling and raise ICP, studies by Bouma and Bruce have shown that BP increases of as much as 30 mm Hg does not significantly alter ICP. This phenomenon appears to exist regardless of whether or not patients had intact cerebral autoregulation. No prospective, randomized, controlled studies have definitively proven that intracranial hypertension, morbidity or mortality is reduced by maintaining CPP values at or above 70 mm Hg. However, CPP has been shown to be strongly correlated to mortality, and several investigators have also demonstrated an associated decrease in patient morbidity and mortality when CPP is actively sustained above 70 mm Hg. One study involving 189 patients compared the effects of a CBF-targeted protocol that kept CPP over 70 mm Hg and PaCO2 at 35 mm Hg, versus an ICP-targeted protocol that kept CPP above 50 mm Hg and PaCO2 at 25-30 mm Hg. The CBF-targeted protocol reduced the frequency of jugular desaturation from 50.6% to 30% (p=0.006). This study was not powered to determine a difference in neurological outcome. These preliminary data suggests that a CPP of at or above 70 mm Hg may reduce both global and regional ischemia and limit secondary injury. Therefore, the BTF and EBIC Guidelines suggest that CPP be maintained above 70 mm Hg.
There is no practical method for measuring CMRO2 or CBF in the ICU setting. Although jugular venous saturation monitoring does not give quantitative information about either CMRO2 or CBF, it reflects the ratio between these two variables. By utilizing SaO2 and SjvO2, the CEO2 and CERO2 can be calculated (Table 1), and the balance between CMRO2 and CDO2 can be determined (Table 2). In this way, it is possible to know if CMRO2 and CBF are appropriately coupled, or if the injured brain is undergoing “oligemic cerebral hypoxia” or “luxury perfusion”. The occurrence of one episode of jugular venous desaturation was associated with a two-fold increase in the risk of poor neurological outcome while multiple episodes were associated with a 14-fold increase in risk. Monitoring of SjvO2 is useful because ICP and CPP are pressure monitoring strategies and may not detect cerebral ischemia. Therefore some have advocated using jugular venous bulb saturation monitoring to optimize cerebral oxygenation. It may also be especially useful in situations where ICP monitoring is not readily available or coagulation abnormalities contraindicate the use of ICP monitoring. One prospective interventional study involved a comparative outcome assessment of 353 patients with severe acute brain trauma who underwent both management of cerebral extraction of oxygen and cerebral perfusion pressure versus patients who underwent monitoring and management of cerebral perfusion pressure alone. At 6 months post injury, patient outcomes were significantly better in the group who underwent cerebral extraction of oxygen monitoring. Monitoring of SjvO2 is considered to be very useful for monitoring therapies used in TBI patients. Therapeutic strategies such as hyperventilation and indomethacin may reduce ICP at the expense of inducing cerebral ischemia, and worsening secondary brain injury and neurological outcome. These acute therapies can be safely monitored using a jugular bulb saturation monitor that will detect a drop in SjvO2 and increased CERO2 indicating cerebral ischemia. Drawbacks of jugular bulb catheter monitoring include the potential for complications, and the cost of the catheters themselves. There are technical difficulties associated with catheter placement, and anatomical structures close to the internal jugular vein can be injured during catheterization. Once placed, the position of the catheter tip within the jugular bulb should be confirmed radiographically. Complications of SjvO2 monitoring are rare, with a 4.5% incidence of carotid artery puncture, and 40% incidence of subclinical jugular thrombi. In summary, jugular bulb monitoring is used to optimize cerebral oxygenation by providing information on CERO2, and ensuring normal coupling between CMRO2 and CBF during management of TBI. The BTF Guidelines acknowledge that jugular venous oxygen saturation monitoring may be useful by identifying cerebral ischemia induced by therapies such as hyperventilation.
Electroencephalogram tracings closely correlate with CMRO2. In normal adults, infusion of barbiturates until the EEG becomes flat reduces the CMRO2 by 50% to an average of 1.5 mL/100g brain/min. (49) Therefore it is useful in sedative titration in barbiturate coma. The EEG is also considered to be the best method for detecting seizure activity, and is especially useful in patients on paralytic agents or those suspected of having non-convulsive status epilepticus. Vespa et al. recently reported their observations with the incidence and impact of nonconvulsive and convulsive seizures in a series of 94 moderate to severe TBI patients. Overall, convulsive and nonconvulsive seizures occurred in 21 patients (22%), with 6 cases of status epilepticus. All patients with status epilepticus died, compared with a mortality rate of 24% in the group without seizures (p<0.001). In more than half of the cases, the seizures were nonconvulsive and were diagnosed solely by EEG. Moreover, the mean serum phenytoin blood concentration in patients with seizures was 66.4 umol/L, which is considered to be within the usual therapeutic range for phenytoin. Thus, it can be seen that there is a significant occurrence of nonconvulsive seizures in TBI patients, and that these seizures may occur in the presence of appropriately dosed prophylactic phenytoin therapy. Although there is a valuable role for continuous EEG monitoring in the TBI patient, continuous EEG monitoring is not routinely used in all centers.
Transcranial Doppler (TCD) ultrasonography measures cerebral blood flow velocity, which is a surrogate marker of CBF. When flow through a vessel is constant, cerebral blood flow velocity increases inversely to the vessel’s cross-sectional area. Analysis of TCD waveform and blood flow velocity can indirectly provide information about CBF and ICP. The analysis of the velocities during the cardiac cycle is presented as a waveform. Current machines calculate a mean flow velocity and the pulsatility index (PI). The PI (equal to [systolic velocity – diastolic velocity]/mean velocity) is an index of vascular resistance. Therefore, this index would be elevated in patents with increased ICP. Vasospasm is defined as a mean flow velocity above 102 cm/sec. Transcranial doppler monitoring can be used to detect vasospasm, assess blood flow patterns, determine patency of cerebral vessels, and detect elevated ICP in TBI patients. The most important limitation of TCD monitoring is the technical ability of the neurologist. In summary, TCD is a non-invasive method that may be used to assess the adequacy of cerebral blood flow in an attempt to differentiate hyperemia from vasospasm and to assess the status of cerebral autoregulation in TBI patients.
Goals of Therapy for TBI
The overall goals of management for TBI patients are to reduce mortality, morbidity, and improve long-term functional outcome. Secondary brain injury can be prevented and/or limited by optimizing cerebral resuscitation.
Specific goals of therapy published in existing TBI guidelines and consensus papers include:
1) controlling ICP to less than 20-25 mm Hg;
2) maintaining CPP at or greater than 70 mm Hg or MAP above 90 mm Hg if an ICP monitor is not in place;
3) optimizing cerebral oxygenation by ensuring normal coupling between CMRO2 and CBF and maintaining CEO2 above 42% (CERO2 above 40%); and
4) preventing and/or treating cerebral and systemic complications of TBI.
Therapeutic Interventions for TBI
As with any trauma patient, the initial priorities are establishment of an adequate airway, breathing, and circulation. If the patient’s level of consciousness is impaired to the extent that the airway cannot be protected, intubation is required. Mechanical ventilation may be indicated in the setting of hypercapnic or hypoxemic respiratory failure caused by a decreased level of consciousness, atelectasis, pneumothorax, or pulmonary edema. Mechanical ventilation may enable brief hyperventilation that may be indicated for acute increases in ICP. Oxygen saturation should be maintained above 90 % and paO2 above 60 mm Hg.
Circulating volume must also be restored with packed red blood cells if the hematocrit is less than 30%, and isotonic (0.9%) saline for further fluid resuscitation. Colloids (mannitol and albumin) and hypertonic saline (3% and 7.5%) have been studied for initial resuscitation; however data from prospective trials have not shown clinically significant advantages in resuscitation or neurological outcome. Hypotonic fluids should not be used for initial fluid resuscitation, as they may lead to hyponatremia and a hypo-osmolar state that may exacerbate cerebral edema. The goal of fluid resuscitation and fluid maintenance should be euvolemia without reducing plasma osmolarity. If fluids alone do not restore MAP to 90 mm Hg, inotropic and vasopressor agents such as dopamine, phenylephrine, and norepinephrine may be required.
Adequate resuscitation is vital as SBP less than 90 mm Hg or hypoxia (apnea or cyanosis in the field or a PaO2 less than 60 mm Hg) are strong independent risk factors for mortality, and worse neurological outcome. Once initial resuscitation measures have been completed, diagnostic evaluation with a CT scan is indicated, along with a complete physical and neurological exam, serum chemistry, hematology, and toxicology screening.
General Supportive Measures
Once the TBI patient is stabilized and diagnostic tests have been completed, several routine, general supportive measures should be undertaken.
Central venous catheters, arterial catheters, jugular venous bulb catheter, and an ICP monitoring device should be placed, if indicated. The head of the bed should be elevated to 30 degrees with the patients head stabilized in a neutral position to enable adequate venous drainage and prevent venous outflow obstruction that may increase ICP. Maintenance fluids should be isotonic saline, and the serum sodium and plasma osmolarity should be maintained within the normal range. Euvolemia should continue to be the goal fluid status. Fluid therapy should be utilized to help maintain an adequate MAP above 90 mm Hg if an ICP monitor is not in place, or to a CPP > 70 mm Hg utilizing an ICP monitor. If the MAP is low, and the CVP or pulmonary capillary wedge pressure (PCWP) is adequate, vasopressors may be required to increase MAP above 90 mm Hg or to achieve a CPP of 70 mm Hg or greater.
Serum glucose should be maintained at 4-7 mmol/L as data suggests that hyperglycemia in the setting of ischemia may worsen outcome. Two retrospective reviews have attempted to describe the effect of hyperglycemia on outcome in TBI patients. Margulies et al. found that peak glucose levels were inversely related to both GCS on admission and GCS at discharge (p<0.001) in 97 patients with intracranial injuries. However these peak glucose levels may simply reflect the severity of illness and injury in this population as peak glucose was not an independent predictor of outcome in this series. Lam et al. retrospectively reviewed the relationship between serum glucose and outcome in 169 head-injured patients. Patients with lower GCS scores had higher serum glucose concentrations, and patients who remained in a persistent vegetative state or died had significantly higher glucose concentrations than those patients with a good outcome or moderate disability. In the severely injured subgroup (i.e. GCS less than 8), a serum glucose of greater than 11.1 mmol/L was associated with a significantly worse outcome (p<0.01). For these reasons, prevention of hyperglycemia and aggressive control of elevated blood glucose is required. Intravenous fluids (including medication piggybacks) should not contain dextrose, frequent blood glucose monitoring with a glucometer should be performed, and sliding scale insulin may be required to control blood glucose levels.
Oxygen saturation should be maintained above 90% and the PaO2 should be kept at 60 mm Hg or greater. Positive end expiratory pressure and inverse ratio ventilation may be required in patients with a pulmonary shunt or oxygenation difficulties, and these modalities may be used safely in TBI patients without compromising ICP.
Fever is common in head-injured patients, and increases the body’s metabolic rate by 10-13% per degree Celsius. Patients with hyperthermia have been shown to have worse neurological outcomes than those patients who remain normothermic. Measurement of rectal temperature alone may significantly underestimate brain tissue temperature by 0.5-2.0°C. For these reasons, fever prophylaxis with acetaminophen 650 mg every four hours and/or a cooling blanket should be used to maintain at least normothermia, and possibly hypothermia.
Pain, anxiety, and delirium manifested as agitation along with coughing, respiratory suctioning, and ventilator asynchrony can cause increases in cerebral oxygen consumption and ICP. Thus opioid analgesics, sedatives, and neuromuscular blocking agents may be required. There have been no large studies comparing the effects of sedative agents on outcome in patients with severe head injury. Thus, the use of analgesic, and sedative agents should be titrated based on cerebral hemodynamic and oxygenation goals (ICP, CPP, and CERO2) or by using sedation scales validated in critically-ill patients. Although opioids have been associated with increases in ICP, several studies have refuted this effect, even with newer, synthetic opioids. Until definitive evidence is available, opioids may be used in TBI patients, and ICP should be followed for possible drug-related increases. Tipps et al. recently reported a small case series in which the newer synthetic opioid, remifentanil was evaluated in the neurological intensive care unit. These investigators found that the ultra-short duration of action of remifentanil facilitated frequent neurological examinations and no patient experienced deleterious hemodynamic or neurological effects from the use of this drug. Propofol is a sedative-hypnotic agent formulated in a 1% solution in a 10% intralipid carrier. Propofol causes dose-dependent decreases in CMRO2, CBF, ICP, MAP, and CPP. The advantage of this drug is that it can be titrated easily and patients can be awoken quickly to enable neurological evaluation; however, it does not possess any analgesic properties. A recent pilot, prospective, randomized, double-blind trial compared morphine alone to morphine plus propofol in 42 patients with severe head injury. The use of neuromuscular blocking agents, benzodiazepines, pentobarbital, and CSF drainage was less in the propofol group, and the use of vasopressors was higher in the propofol group due to hypotension. Mortality and neurological outcome was no different between the groups. Until there has been a clinically significant advantage proven with propofol, other sedative agents can be used. If propofol is used, MAP should be monitored closely to ensure CPP is not compromised. Neuromuscular blocking agents may be required in TBI patients; however, they should not be routinely used. A prospective, randomized trial of 514 severely head injured patients examining the effects of prophylactic neuromuscular blockade showed that the intervention group had a longer intensive care unit stay, a higher incidence of pneumonia, and a trend toward more frequent sepsis. Complications of neuromuscular blockade including prolonged paralysis and prolonged weakness mandate that these agents be reserved. General measures are used along with specific therapeutic modalities to keep ICP less than 20 mm Hg, maintain CPP at 70 mm Hg or greater , and CERO2 below 40%.
Specific Treatment Modalities
Various specific therapeutic modalities studied in the management of TBI patients have focused on preventing or treating increases in ICP by reducing cerebral blood volume, CSF volume, or brain volume as per the Monroe-Kellie doctrine. Lowering ICP or raising MAP will increase CPP. Other therapies have focused on reducing CMRO2 or increasing CDO2 to improve cerebral oxygenation and maintain CERO2 below 40%. Specific therapies include CSF drainage, hyperventilation, mannitol, barbiturate-induced coma, and seizure prophylaxis.
The BTF Guidelines recommend ventricular catheters as the ICP monitoring method of choice because of their accuracy, ability to be recalibrated, and because of their ability to drain CSF to acutely lower ICP. Although CSF drainage is recommended as one of the first methods to acutely lower ICP, its effectiveness has never been studied in a randomized clinical trial.
Mannitol has been used to treat suspected or proven increases in ICP in TBI patients. The BTF Guidelines state that the benefits of this drug are assumed to be related to an immediate plasma-expanding effect that reduces the hematocrit, reduces blood viscosity, increases cerebral blood flow, and increases CDO2. This may explain why mannitol reduces ICP within minutes, and why its effects appear more pronounced in patients with low CPP (less than 70 mm Hg). Secondly, mannitol is thought to establish an osmotic gradient between plasma and brain cells and reducing cerebral edema and ICP by drawing water across an intact blood-brain barrier from the brain into the vascular compartment.
Mannitol can be administered at 0.25-1.0 g/kg IV q4h, with an onset of action of several minutes, a peak effect between 20-40 minutes and a variable duration of action. Mannitol administration has been associated with several neurological and systemic adverse effects. Mannitol can worsen vasogenic cerebral edema by passing across an opened blood-brain barrier and causing a reverse osmotic shift to draw water into brain cells, worsen cerebral edema, and paradoxically increase ICP. Although there are no published randomized controlled trials that have compared different modes of administration (bolus versus infusion) or different doses, preliminary data suggests that the accumulation of mannitol in the brain is most marked when mannitol is in circulation for long periods, as occurs with continuous infusions. Other potential adverse effects associated with mannitol administration may be acute congestive heart failure and pulmonary edema, hyperosmolar dehydration and hypotension with prolonged administration, hyponatremia, hypokalemia, acute renal failure when serum osmolality exceeds 320 mOsm/kg, and potential rebound increases in ICP after discontinuation of prolonged therapy.
The effect of mannitol on neurological outcome and mortality has been studied in three well-designed randomized controlled trials, and has been subjected to meta-analysis. A prospective, randomized, double-blind trial by Sayre et al. compared pre-hospital administration of 5 mL/kg of 20% mannitol over 5 minutes compared to 5 mL/kg of normal saline in 41 patients with moderate to severe head injuries. No data were provided for ICP, or disability, but relative risk (RR) for death at 2 hours was 1.75 (95% CI 0.48 – 6.38). This study failed to show a benefit from mannitol, and also failed to exclude a harmful effect on outcome. Smith et al. conducted a prospective, randomized, double-blind trial comparing the administration of ICP-directed and empiric mannitol therapy in 77 patients with severe head injuries. This study also failed to show a benefit of mannitol therapy with a RR of death at one year of 0.83 (95% CI 0.47-1.46). Schwartz et al. conducted a prospective, randomized, single-blind trial comparing 1 g/kg of 20% mannitol plus additional boluses versus a load and continuous infusion of pentobarbital in 59 patients with severe head injuries and raised ICP for less than 15 minutes. Again, this trial failed to show a beneficial effect of mannitol on outcome with a RR of death of 0.85 (95% CI 0.52-1.38).
A Cochrane Collaboration meta analysis confirmed that there are insufficient data to preclude either a harmful or beneficial effect of mannitol on mortality. Thus, despite acute reductions in ICP and improvements in CBF, mannitol has not been shown to improve neurological outcome or reduce mortality. Thus the therapeutic goals of therapy for mannitol should be to optimize control of ICP, CPP, and CERO2. The BTF Guidelines recommend mannitol for treatment of elevated ICP in bolus doses of 0.25-1.0 g/kg body weight. Mannitol can be administered to patients without ICP monitoring if there are signs of transtentorial herniation or progressive neurological deterioration not attributable to systemic pathology. Serum osmolarity should be kept below 320 mOsm/kg and adequate fluid replacement should be given to maintain euvolemia.
Furosemide is a loop diuretic that has also been shown to be effective in reducing ICP, but to a lesser extent than mannitol. Unlike mannitol, furosemide will decrease production of CSF. When used in combination with mannitol, furosemide (0.1-1 mg/kg) enhances the degree and duration of ICP reduction and it decreases the risk of rebound ICP elevation. A concern associated with the routine use of furosemide is the potential of inducing massive diuresis that would result in depletion of the intravascular volume and electrolytes. Therefore this agent should be reserved to those patients who do not have a satisfactory response to mannitol.
Hyperventilation (HV) has been widely used in the treatment and prevention of raised ICP following TBI because it is simple, and rapidly reduces ICP. Two recent surveys published in 1995 and 1996 respectively have shown that HV was used in 83% of American Centres, and 100% of United Kingdom Centres. (89,90) A recently published survey of American Centre neurosurgeons suggests that the use of prophylactic HV has dropped from 83% in 1991 to 36% in 1997. The use of HV may be dropping due to concerns highlighted in the BTF Guidelines that outline potential drawbacks of HV and summarize the results of several clinical trials re-examining HV therapy. Hyperventilation reduces ICP by causing cerebral vasoconstriction, with a subsequent reduction in CBF and CBV. A potential drawback of HV is that although it may successfully control ICP, it may reduce CBF to ischemic levels. This can be seen during jugular venous bulb monitoring as a drop in SjvO2, and rise in CERO2. Under conditions of “luxury perfusion” short-term HV could control acute increases in ICP, without adversely affected cerebral oxygenation. However, in situations of early brain injury ( less than 6 hours) during “oligemic cerebral hypoxemia” when CBF is already uncoupled and insufficient to meet CMRO2, further reductions induced by HV would be expected to further reduce CBF, induce cerebral ischemia, and potentially worsen secondary injury. For these reasons, the practice of HV is under reassessment.
Muizelaar et al. performed a prospective, randomized, unblinded trial comparing normocapnia, HV, and HV plus tromethamine (THAM) buffering over 5 days in 77 patients with severe TBI and assessed neurological outcome at 3, 6, and 12 months. (46) Overall, the study failed to show a benefit of HV or HV plus THAM on combined death or severe disability (RR = 1.14 (95% CI 0.82-1.58), RR = 0.87 (95% CI 0.58-1.28)) or mortality (RR = 0.73 (95% CI 0.36-1.49), RR = 0.89 (95% CI 0.47-1.72)) respectively. (46) Interestingly, in the subgroup with a motor GCS score of 4-5, fewer patients had a favorable outcome (good or moderately disabled) in the HV groups at 3 and 6 months (p<0.05). (46) This difference did not remain significant after 12 months, likely due to a lack of statistical power. (1,46) Based on the results of this trial, the BTF has re-evaluated the practice of HV therapy. (1) The current BTF Guidelines recommend that in the absence of increased ICP, chronic prolonged HV therapy (PaCO2 less than 25 mm Hg) should be avoided after severe TBI. (1) The use of prophylactic HV therapy (PaCO2 less than 35 mm Hg) during the first 24 hours after severe TBI should be avoided because it can compromise CPP when CBF is reduced. (1) HV therapy is most appropriately reserved for acute neurological deterioration associated with elevated ICP, or for more prolonged periods when elevated ICP is refractory to other treatments. (1) The guidelines recommend that jugular venous saturation monitoring be used to monitor for cerebral ischemia if short-term HV resulting in a PaCO2 of less than 30 mm Hg is required to acutely reduce elevated ICP. (1)
A recent study has been published that involved an evaluation of the effects of early moderate (PaCO2 of 30 ± 2 mm Hg) hyperventilation in nine patients within 12 hours of injury. (92) In this study the investigators measured CBF, CMRO2, CBV, cerebral venous oxygen content (CvO2), and CERO2. These investigators found that during hyperventilation, global CBF decreased, CBV fell, CERO2 rose, CvO2 fell, and CMRO2 remained changed. They concluded that brief, moderate hyperventilation soon after injury does not alter cerebral metabolism and is unlikely to harm the patient. However, further study is necessary to evaluate focal changes and the effects in patients with elevated ICP. (92)
If CSF drainage, mannitol, and HV fail to control an elevated ICP and improve cerebral oxygenation, high-dose barbiturate therapy has been used as a last pharmacologic resort prior to emergency non-dominant temporal lobectomy with or without craniectomy. (1,22,23) Others have used barbiturate therapy prophylactically in an attempt to improve outcome. Barbiturates reduce CMRO2, thus reducing cerebral metabolic demands, CBF, CBV, and ultimately ICP to enhance global cerebral perfusion and improve cerebral oxygenation. (1,93) At least one trial suggests that normal metabolic coupling between CMRO2 and CBF is necessary for barbiturates to be effective. (94) Barbiturate therapy has also been associated with important complications such as infection, hypothermia, and hypotension (which may reduce CPP and offset any beneficial effects from ICP reduction). (93) Despite these controversies, surveys published in 1995 and 1996 respectively have shown that barbiturates were used in 56% of UK intensive care units and 33% of American intensive care units at that time. (89,90) Since 1974, case series data have suggested beneficial effects of barbiturate therapy on raised ICP. (95,96,97,98) These preliminary data suggest that ICP can be adequately controlled with barbiturates, neurological outcome can be improved, and mortality can be reduced.
Four well designed clinical trials have been published that involved the evaluation of barbiturate therapy for the management of head injury. (29,84,99,100,101) Two studies assessed prophylactic administration of barbiturates or their use in initial therapy for ICP elevations. Ward et al. performed a randomized, controlled trial of pentobarbital versus no pentobarbital in 53 head-injured patients with acute intradural hematoma or whose best motor response was abnormal flexion or extension. (99) Although the mean ICP and temperature were lower in the barbiturate group, hypotension (SBP less than 80 mm Hg) was significantly more common in the treated patients (54% versus 7%, p < 0.05), and may have reduced CPP. (99) There was no significant difference in 1-year Glasgow Outcome Scale (GOS) score between the groups. (99) Schwartz et al. conducted a prospective, randomized, single-blind trial comparing a load and continuous infusion of pentobarbital to 1 g/kg of 20% mannitol plus additional boluses in 59 patients with severe head injuries and raised ICP for greater than 15 minutes. (84) Pentobarbital was less effective than mannitol at controlling ICP with 68% of pentobarbital patients and 39% of mannitol patients requiring a second drug for treatment of ICP (RR = 1.75 95% CI 1.05-2.92). (84) This study failed to show a difference in mortality between the two groups (RR = 1.21 95% CI 0.75 – 1.94). (84) Levy et al. performed a randomized, controlled trial in head injury patients comparing pentobarbital and etomidate for control of raised ICP. (100) This trial was stopped after 7 patients were enrolled due to patients in the etomidate group developing renal failure. (100) The final trial is a prospective, randomized, double-blind multicentre trial published by Eisenberg et al. in 1988. (29) In this study, 73 patients with severe head injury (GCS 4-8) and intractable elevations in ICP were randomized to receive high dose pentobarbital 10 mg/kg over 30 minutes, followed by 5 mg/kg/h for 3 hours, and a 1 mg/kg/h infusion versus conventional therapy (no pentobarbital). (29) Serum pentobarbital levels of 30-40 mg/L were attempted. The primary outcome in this trial was controlled ICP less than 20 mm Hg. A smaller proportion of patients in the pentobarbital group had uncontrolled ICP (68% vs. 83%, p = 0.22), although this was not statistically significant. (29) The authors reported that the likelihood of survival in barbiturate responders was 92% at one month compared with 17% for non-responders. (29) Unfortunately the design of this study and others did not enable a conclusive determination on the effect of barbiturates on death or adverse neurological outcome. (1, 29)
A recent Cochrane Collaboration meta analysis examined the effects of barbiturates on outcome after TBI found no effect on mortality or morbidity (Table 4). (101) These data concur with an earlier meta analysis by Roberts et al. which also found no beneficial effect of barbiturates (Table 4). (102) Thus, all the data to this point have failed to show a beneficial effect on outcome with barbiturates in head injury. Unfortunately, the risk of hypotension was significantly higher than no therapy and translated to a fall in BP in 1 of every 4 patients treated. (101,102) Thus, if barbiturates are to be used for refractory increases in ICP, the pentobarbital regimen by Eisenberg should be followed. (1) According to the BTF Guidelines, barbiturate therapy may be considered in hemodynamically stable salvageable severe TBI patients with elevated ICP refractory to maximal medical and surgical therapy. (1) The pentobarbital infusion should be titrated to pentobarbital levels of 30-40 mg/L or preferably burst suppression on EEG. (93,103). It is important to monitor for hypotension, and vasopressor agents should be readily available during the loading period to support MAP and maintain a CPP greater than 70 mm Hg. (93)
Table 4. Meta analyses of therapeutic interventions for TBI
|Intervention||Author, Year (ref)||Death||Death or disability
(PVS or SD)
|Barbiturates||Roberts 2000 (101)||48/105 (46%)
RR = 1.09 (0.81-1.47)
RR = 1.15 (0.81-1.64)
RR = 1.80 (1.19-2.70) NNH = 4
|Roberts 1998 (102)||37/64 (58%)
RR = 1.12 (0.81-1.54)
RR = 0.96 (0.62-1.49)
RR = 1.80 (1.19-2.70)NNH = 4
|Anticonvulsants||Schierhout 2000 (111)||95/540 (17.6%)
RR = 1.15 (0.89-1.51)
RR = 0.96 (0.72-1.39)
|66/196 (33.7%)||Early (< 7 days)
RR = 0.34 (0.21-0.54)
NNT = 10
|65/434 (15.0%)||Skin Rash
RR = 1.57 (0.57-39.88)
RR = 1.49 (1.06-2.08)
|30/76 (39.5%)||Late (> 7 days)
RR = 1.28 (0.90-1.81)
|Induced Hypothermia||Signiori 2000 (116)||31/79 (27%)
OR = 0.67 (0.38-1.17)
RR = 0.39 (0.20-0.74)
NNT = 4 (3-13)
|Corticosteroids||Alderson 1997 (128)||396/1061 (37%)
OR = 0.91 (0.74-1.12)
ARR = 1.8% (-2.5 – 5.7)
OR = 0.90 (0.72-1.11)
126/ 447 (28%)
OR = 0.92 (0.69-1.23)
OR = 1.05 (0.44-2.52)
|Alderson 2000 (129)||410/1194 (34%)
RR = 0.96 (0.85-1.08)
ARR = 1.3% (-2.5-5.2%)
RR = 1.01 (0.91-1.11)
RR = 0.94 (0.76-1.16)
RR = 1.11 (0.54-2.26)
The incidence of generalized seizures in patients suffering a head injury is 5%-10%. (10) The incidence of post-traumatic seizures (PTS) is higher in patients with a GCS less than10, cortical contusion, depressed skull fracture, subdural hematoma, epidural hematoma, intracerebral hematoma or penetrating head wound. (1) PTS have been classified into early (occurring within 7 days of injury) or late (occurring after 7 days following injury) seizures. (1) In the convalescent period following TBI, seizures may exacerbate secondary injury by increasing ICP, increasing CMRO2, CBF, CBV, lowering MAP, and potentially lowering CPP to worsen cerebral oxygenation. (10,11) Seizures may also lead to complications such as exacerbation of head injury, aspiration, nosocomial pneumonia, and a generally poorer overall outcome. For these reasons, and based on positive results from early retrospective trials, seizure prophylaxis was commonly prescribed by U.S. neurosurgeons for head-injured patients in the 1970s. (104) However, potential drawbacks of routine anticonvulsant prophylaxis can include dose-related CNS and gastrointestinal adverse effects, and idiosyncratic effects ranging from a mild rash to Steven-Johnson’s Syndrome. (1)
One of the earliest randomized, controlled trials conducted to study the effect of phenytoin prophylaxis on the development of early and late PTS prophylaxis was performed by Young et al. in 1983. (105) In this study, 244 patients with head injuries were randomized to receive phenytoin or placebo and followed for two years. Patients received an 11 mg/kg IV loading dose, followed by IM or oral dosing guided by daily plasma phenytoin sampling in an attempt to achieve target plasma phenytoin levels of 40-80 umol/L. (105) According to the authors, there was no apparent difference in the incidence of early or late seizures between the treatment (3.7% and 12.4%) and the control (3.7% and 10.8%) groups. (1,105) This lack of apparent difference may be explained by the low incidence of seizures in the placebo group, which reduced the power of the trial to detect a difference. (1,105) During the early phase of the study, serum concentrations were obtained every 24 hours, and 78% of patients had plasma phenytoin concentrations of at least 40 umol/L. (105) Moreover, none of the patients with phenytoin plasma concentrations above 48 umol/L experienced a seizure. (105)
In 1990, Temkin et al. performed the largest prospective, randomized, double blind, placebo-controlled trial evaluating the effectiveness of phenytoin for the prevention of PTS in patients with severe head injuries. (106) In this study, 404 patients were randomized to receive a phenytoin 20 mg/kg IV loading dose followed by IV, oral, or nasogastric maintenance dosing that was guided by thrice weekly plasma phenytoin concentrations in the ICU, weekly on the ward, and at 1, 3, 6, 9, and 12 months during the 24 months follow-up. The daily doses varied considerably throughout the study and ranged from 200-1200 mg/d IV or orally, and up to 2600 mg/d given via the nasogastric route. The target phenytoin concentrations were 40-80 umol/L (total) or 3-6 umol/L (free). The incidence of early PTS was found to be significantly lower in the treatment group (3.6% (95% CI 2.3-4.9) versus 14.2% (95% CI 11.6-16.8), (p<0.001), RR = 0.27 (95% CI 0.12 to 0.62), number needed to treat (NNT) 10 (95% CI 8-18)) to prevent one early seizure. No significant reduction in late PTS between groups was observed. Eighty-one percent of patients had attained free phenytoin plasma concentrations between 3.0-7.9 mmol/L following the loading dose, while at one week only 45% of measured plasma concentrations were within the same range. At the scheduled outpatient visits starting at one month, levels were at least therapeutic in 70% of patients. The drop in plasma phenytoin concentrations may have been due to early under-dosing and/or previously described pharmacokinetic alterations in severe TBI patients affecting protein binding and/or phenytoin metabolism. (107,108) There was a significantly higher incidence of rashes causing drug discontinuation in the phenytoin group (8.2 % versus 2.0%, (p<0.01), number needed to harm (NNH) of 17)). Using the initial Temkin data, for every 1000 patients treated with phenytoin, 111 seizures would be prevented and 62 patients would be expected to discontinue therapy as a result of the development of a rash. The Temkin trial had greater statistical power than the Young trial and likely showed a benefit because the population was at higher risk for seizures than those in the Young trial. The Temkin trial data have been re-analyzed to investigate the incidence of adverse effects during the first 2 weeks of therapy. (109) Based upon this new analysis, the overall incidence of adverse effects was found to be similar in the two groups (9% for phenytoin vs. 6% for placebo, p=0.52) as was the incidence of rash during the first week (0.6% vs. 0%, p=1.0). (109) Given that there is no difference in adverse effects between phenytoin and placebo, decisions on the use of 7 days of phenytoin prophylaxis should be made based on the clinical significance of the effects on seizure prevention and overall neurological outcome.
A more recent prospective, randomized, double blind trial by Temkin et al. involved a comparison of valproic acid with phenytoin for the prevention of early and late PTS in 379 patients with head injury. (110) Patients were enrolled if they had an immediate PTS, depressed skull fracture, penetrating brain injury or there was CT evidence of a cortical contusion, subdural, epidural, or an intracerebral hematoma. Patients were assigned to receive phenytoin therapy for 7 days, valproic acid therapy for 30 days, or valproic acid for 180 days. The incidence of early seizures was low and did not differ between the phenytoin and pooled valproic acid groups (1.5% versus 4.5%, p=0.14), RR = 2.9 (95% CI 0.7 – 13.3)). Due to the high upper confidence interval, the study did not have the power to exclude a potentially clinically significant inferior effect of valproic acid. The incidence of late seizures was not different between the groups, although there was a trend toward a higher mortality rate in patients receiving valproic acid vs. phenytoin (13.4% vs. 7.2%, p=0.07, RR = 2.0, 95% CI 0.9 – 4.1). For these reasons, valproic acid cannot be routinely recommended for the prevention of PTS until more data are available.
The Cochrane Collaboration group has published a recent meta analysis identifying 10 eligible randomized controlled trials involving 2,036 patients who were randomized to anticonvulsant therapy or placebo for PTS prophylaxis. (111) Data on early seizure frequency available from 4 published trials (890 patients) suggested that anticonvulsant agents reduced the incidence of early PTS but not late PTS, and failed to show a reduction in corresponding mortality (Table 4). Neither phenytoin prophylaxis nor carbamazepine prophylaxis resulted in a reduction in death and neurological disability for treated versus untreated patients (Table 4). Based on the trend towards an increase in death and neurological disability in this meta analysis, carbamazepine may actually be expected to increase death or neurological disability; however, the clinical significant of this finding is difficult to determine (Table 4). The analysis failed to rule out a clinically significant increase in skin rashes. Thus, based on the current data, the prophylactic use of anticonvulsants to prevent late PTS is not currently recommended in the BTF Guidelines and Practice Parameters by the Brain Injury Special Interest Group of the Academy of Physical Medicine and Rehabilitation. (1,112) Phenytoin is considered acceptable therapy for the prevention of early PTS; however, the current data do not support a reduction in overall mortality or death and neurological disability. (1,111) If phenytoin is used to prevent early PTS, a loading dose of 20 mg/kg, followed by an aggressive maintenance dose of 5-7 mg/kg/d (ideal body weight) should be used during the first week of therapy to attain target plasma phenytoin concentrations of 40-80 umol/L (total) or 4-8 umol/L (free) in this hypermetabolic patient population. (107,108)
Experimental Therapies for TBI
Fever is common in TBI patients, increases the body’s metabolic rate, and is associated with worse neurological outcomes in TBI patients. For these reasons, induced therapeutic hypothermia has been used experimentally and in clinical trials as a method to reduce ICP and CMRO2 in an attempt to improve neurological outcome after TBI. (1) The mechanism of protection conferred by induced hypothermia is unknown. Originally it was thought to be due to a reduction in CMRO2 (113); however, hypothermia may also influence the excessive post-traumatic release of excitatory neurotransmitters (114), and attenuate opening of the blood-brain barrier. (115,116) Potential risks of induced hypothermia include pneumonia, sepsis, coagulation abnormalities, myocardial ischemia and atrial fibrillation. (116,117)
A reduction in body temperature from 37°C to 32°C reduces CMRO2 by approximately 10%. (10) A preliminary prospective trial in severe head injury revealed that mild hyperthermia to 33.5-34.5°C significantly reduced ICP and increased CPP in patients refractory to high-dose barbiturate therapy. (118) In another study involving 33 severe head injury patients with refractory increases in ICP who received deferred moderate therapeutic hypothermia to 32-33°C, this body temperature reduction was found to reduce ICP and CBF by 40 % (p<0.004) and 26 % (p<0.021), respectively. The survival rate was also found to be higher in the treatment group (p<0.05). (40) More recently, Marion et al. completed the largest prospective, randomized, double blind trial in 82 severely head-injured patients. (119) Body temperatures were reduced to 32-33°C approximately 10 hours after their injury, and these temperatures were maintained for 24 hours. Among patients with a GCS of 5-7, hypothermia was found to be less likely associated with a poor outcome of death, vegetative state, or severe disability (GOS of 1-3) compared to normothermia at 3 months (63% versus 83%, adjusted RR = 0.2 (95% CI 0.1-0.9), NNT of 5), and 6 months [45% versus 67%, adjusted RR = 0.2 (95% CI 0.1-0.9), NNT of 5), but not at 12 months (38% versus 62%, adjusted RR = 0.3 (95% CI 0.1-1.0)].
To determine the impact of induced hypothermia on mortality and neurological outcome, the Cochrane Collaboration conducted a meta analysis of 8 randomized controlled trials. (41,116,118,119,120,121,122,123,124) Trials were stratified into two groups: those that looked at active immediate hypothermia or deferred hypothermia. (116) Meta analysis failed to show a benefit of active, immediate hypothermia versus normothermia on mortality; however, a 61% reduction in the risk of death or severe disability was observed (Table 4). This translates in to a NNT of 4 (3-13) to prevent one patient from dying or being disabled. This apparent clinical benefit should be interpreted cautiously as the result may be due to publication bias. Only one trial of 33 patients involved an assessment of deferred hypothermia and reported deaths and GOS at 6 months. (118) This trial failed to show a significant difference in the risk of death (50% versus 82%, OR = 0.21 (95% CI 0.04-1.5)) or death or severe disability (63% versus 94%, OR = 0.10 (95%CI 0.01-1.00)). (116,118) Larger trials are necessary to confirm or refute the effect of induced or deferred hypothermia on long-term outcomes. All of the above trials involved reducing body temperatures to a range of 30-34.5°C for a maximum of 48 hours.
There are no published data to suggest a target cooling temperature, or ideal duration of cooling. However, Bernard et al. recently reported that a moderate hypothermia of 30-33°C for 8 days provided a satisfactory neurological outcome in 20 (47%) of 43 severely head-injured patients. (125) In this study there was a high incidence of nosocomial pneumonia (45%) and hypokalemia on cooling induction, and a decrease in total WBC and platelet counts over 10 days. (125) Another recent trial evaluated mild hypothermia (34°C) for 48 hours involving 16 severely head-injured patients in whom ICP was maintained below 20 mm Hg with conventional therapies. (126) There was no difference in neurological outcome and the authors concluded that mild hypothermia does not convey any benefit over normothermia for those patients with ICP maintained at less than20 mm Hg using conventional therapies. (126) The potential drawbacks of induced hypothermia can include the induction of shivering which can increase CMRO2, CERO2, ICP, and lower CPP. To overcome this problem patients often require pharmacological paralysis with its inherent risks of prolonged paralysis and weakness. Controversies surrounding induced hypothermia will hopefully be resolved when a National Acute Brain Injury Study of hypothermia is published later this year. This study involves an examination of the effect of hypothermia (32-33°C) for 48 hours on GOS at 6 months in 392 severely brain-injured patients. (127)
Our ability to draw conclusions from existing studies of corticosteroids in TBI has been limited due to poor study design including a lack of control groups and blinding, inadequate sample sizes, and drug dosing inconsistencies. (128) Although corticosteroids are effective in reducing cerebral edema in patients with structural brain damage, these agents have not been shown to ameliorate increased ICP nor improve long-term outcomes. This has been thought to be due to the inability of steroids to prevent early cytotoxic edema and may also be related to corticosteroid-induced hyperglycemia, increased infection rates and/or gastrointestinal bleeding. (1,128)
Two meta analyses involving an assessment of corticosteroid trials have been reported. (128,129) In one analysis, 19 trials involving corticosteroid use for 2295 TBI patients were reviewed. (129) Mortality data extracted from 16 trials failed to show benefit on mortality, and data from 9 trials failed to show benefit on the aggregate endpoint of death or disability (Table 4). Another meta analysis that found similar results included many of the same trials, but had more stringent exclusion criteria. (128) Mortality data from 13 randomized trials involving 2073 patients failed to show benefit on mortality, and data from 10 trials failed to show benefit on the aggregate endpoint of death or disability. Both analyses failed to show a difference in infection rates and gastrointestinal bleeding between those who received corticosteroid and those who did not. It is important that although these data fail to show a significant benefit of corticosteroids on outcome, they also fail to exclude a potentially clinically significant absolute mortality reduction of up to 5.7%. (128) Even pooled together, the current trials are too small to support or refute a clinically important benefit from corticosteroids. Although a very large, multicentre trial would be required to solve this clinical dilemma, such a trial may be justified based on the clinical impact of the problem and the widely practicable treatment. (128) The current BTF Guidelines state, however, that corticosteroids are not recommended in TBI patients to lower ICP or improve outcome. (1) Despite that recommendation and the lack of evidence to support their use, the recent survey from Marion et al. suggests that 19% of U.S. Neurosurgeons would use corticosteroids in this setting. (91) The routine use of corticosteroids in TBI patients should be avoided until supported by clinical trial evidence.
Indomethacin is a non-steroidal anti-inflammatory agent with unique effects on cerebral blood flow physiology that may be of benefit in reducing elevated ICP in TBI patients. Data from animal models and randomized, controlled studies with pre-term infants have shown that IV indomethacin produces rapid, significant reductions in cerebral blood flow (CBF). (47) Controlled studies of IV indomethacin in normal volunteers show a reduction in CBF from 26-40%. (47) Case series involving severe TBI patients suggest that indomethacin IV boluses of 30-50 mg reduce ICP by 37-52%, reduce CBF by 22-26%, with a modest 14% increase in cerebral perfusion pressure (CPP). (47) The onset of effects is almost immediate, and peaks in 1-5 minutes. The duration of ICP control is at least 4 hours and may be as long as 30 hours with a continuous infusion (CI). (47) ICP returns to baseline levels after the infusion is stopped. (47) No significant adverse effects have been reported but short-term changes in sensitive physiologic indicators of cerebral ischemia are of potential concern. Using indomethacin for ICP control may be potentially harmful to patients who already have normal or high CEO2 or impaired autoregulation. In this subgroup, indomethacin may decrease CBF to ischemic levels that may not be overcome by a subsequent increase in CEO2. Indomethacin may also cause coronary ischemia in patients with ischemic heart disease, especially during anemia, hypoxemia, hypotension, systemic shunting, or impaired CEO2. For these reasons, some have suggested that indomethacin should not be used in patients with low SjvO2 (less than 60 %), low CMRO2 (less than 40 mL/100g/min), signs or symptoms of cerebral vasospasm, cardiac ischemia, renal failure, or bleeding ulcer. (130)
Important efficacy and safety issues regarding indomethacin use in TBI patients must still be addressed before it can be generally recommended for ICP control. Normal volunteer dose-response studies with IV bolus, CI, and PR dosing will characterize the pharmacodynamics of indomethacin on normal cerebral physiology. The effects of indomethacin on cerebral oxygenation in patients without luxury perfusion or those receiving therapeutic hyperventilation need to be determined. The effect of indomethacin on ICP and subsequent cerebral oxygenation must be studied, and the influence of these effects should be documented using functional neurological outcome scales. Thus, despite encouraging preliminary results, IV indomethacin should only be considered an experimental treatment for control of refractory ICP in TBI patients. Larger, well-designed randomized trials in TBI patients will provide more efficacy and safety data and delineate the effects of indomethacin alone or in combination with other proven, effective, or experimental therapies. Once these concerns have been addressed, larger outcome studies will ultimately be needed to determine the role of indomethacin for ICP control in TBI patients.
As mentioned previously, the secondary injury cascade is a complex biochemical loop that results in propagation of neurological injury. (Figure 1) Laboratory studies with numerous agents identified potential therapeutic targets for neuroprotection following TBI . (131) However, the positive effects shown in animal models have not been replicated in clinical trials. Some of these clinical investigations progressed to completion; however, many were prematurely halted due to interim “futility analyses” or due to concerns with safety. This has lead to many theories regarding why these discrepancies occur including a lack of sensitivity in outcome measures, improper timing of therapeutic intervention, an inability to translate injury mechanisms in animals to human head injuries, poor brain penetration of drugs or inadequate doses used; and insufficient safety and tolerability investigation of the agent in question prior to the study. (131) There are many questions that still need to be answered including refining our understanding of the pathophysiology of TBI and the timing of these events. This section is devoted to a discussion of neuroprotective agents both past and present.
Increases in intraneuronal calcium seen with TBI initiates a cascade of biochemical events that can eventually result in cell lysis and death. Calcium-channel blockers in experimental models attenuated the increase in intracellular calcium concentration and thereby arresting neurological damage. (18) Nimodipine is a “cerebroselective” dihydropyridine calcium channel blocker that is approved by the Food and Drug administration to decrease the neurological morbidity following aneurysmal subarachnoid hemorrhage (SAH). This agent blocks the L-type calcium channel. The postsynaptic L-type channel remains open during depolarization to allow for excessive calcium accumulation. Blocking this channel may prevent the calcium buildup and it may have vasodilatory effects on cerebral vascular smooth muscle. Two large randomized, controlled trials have been performed to study the effect of nimodipine on the outcome of severe head injury, HIT I and HIT II. (132,133) Both trials showed a modest and an insignificant increase in the proportion of favorable outcomes in patients treated with nimodipine. A reanalysis of pooled data from both the HIT I and HIT II trials using IV nimodipine found that the overall benefit of treating unselected head injured patients with nimodipine is unlikely to be clinically relevant. (134) However a subgroup analysis of the HIT II trial suggested that there could be a possible protective effect of nimodipine in patients with traumatic SAH. (133) Another prospective, double-blind trial conducted in Germany randomized 123 TBI patients with a traumatic SAH to nimodipine (2 mg/h) or a matched placebo intravenously for 7 to 10 days, followed by oral treatment (360 mg daily) until day 21. (135) Patients treated with nimodipine had a significantly less unfavorable outcome (death, vegetative survival, or severe disability) at 6 months than placebo-treated patients (25% vs. 46%, p=0.02, OR 0.39, 95% CI 0.18-0.86). Therefore, due to these promising preliminary results, nimodipine is currently being investigated in a Phase III trial in patients with traumatic SAH with GCS 4-15.
Activation of the presynaptic N-type calcium channel in the nerve terminals is the leading cause of excitatory amino acid release during ischemia. Therefore, blockade of this channel might reduce this excitotoxic process. The omega-conopeptide SNX-111, obtained from a carnivorous marine snail, is a potent antagonist of the voltage-gated N-type calcium channel and has demonstrated significant neuroprotective effects against neuronal injury. (136,137,138) However, the limiting adverse effect of this agent was hypotension, which could possibly lower CPP and affect outcome. This trial was halted due to a futility analysis indicating that the drug could not be safely used outside a controlled clinical trial setting. Following injury, release or formation of highly reactive free radicals may produce secondary damage to cellular membranes. This type of injury occurs through a process called lipid peroxidation, which is accelerated in the presence of free reactive iron. (18) Endogenous defense mechanisms and antioxidants were identified that scavenged free radicals. These systems include; superoxide dismutase (SOD), tocopherols, ascorbic acid, and retinoic acid. However, after injury the endogenous levels of these systems are depleted which permits progressive destruction of the cellular membrane. In the laboratory, scavenging of the superoxide anion SOD or polyethylene glycol (PEG)-conjugated SOD (PEG-SOD) has been shown to be beneficial following TBI. (139) A Phase II study included 104 patients randomly assigned to receive either placebo or PEG-SOD (2000, 5000, or 10,000 U/kg) intravenously as a bolus, an average of 4 hours after injury. At 3 months, 44% of patients in the placebo group were vegetative or had died, while only 20% of patients in the group receiving PEG-SOD were in these outcome categories (p <0.03); similarly at 6 months, the same outcomes were 36% and 21%, respectively (p=0.04). (140) From this study it appeared that PEG-SOD was a promising therapy. However, in a follow-up prospective, randomized, double-blind, placebo-controlled multicentre study of 463 patients treated within 8 hours of injury, no statistically significant difference in neurological outcome or mortality was observed between patients treated with PEG-SOD and those receiving placebo. (141) Therefore the use of PEG-SOD in clinical practice was abandoned.
A novel group of compounds, the 21-aminosteroids (“lazaroids”), also emerged as potent inhibitors of oxygen free radical-induced, iron-catalyzed lipid peroxidation. One of these, tirilazad mesylate (U-74006F), was selected for clinical investigation in various models of neurological injury. Preclinical models found that tirilazad improved neurological outcome when given between 5 minutes and 3 hours after injury. (142,143) The Phase III trial of tirilazad evaluated a cohort of 1120 head-injured patients. (144) Six-month outcomes for the tirilazad and placebo-treated groups for the Glasgow Outcome Scale categories of both good recovery and death showed no significant difference. However, subgroup analysis suggested that tirilazad mesylate may be effective in reducing mortality rates in males suffering from severe head injury with accompanying traumatic SAH. The investigators also described problems with imbalances in prognostic variables such as pretreatment hypotension, pretreatment hypoxia, and the incidence of epidural hematomas. (144) This study further emphasized the heterogeneous nature of TBI and the need for better randomization strategies to allow for a more homogenous study sample.
Excitotoxic mechanisms due to over-activity of the amino acid neurotransmitters glutamate and aspartate may be responsible for brain damage after injury. Glutamate activates the NMDA receptor that results in failure of the cellular ionic pump. Pharmacological antagonists of excitatory amino acids represent the most highly investigated class of agents following TBI. The competitive NMDA-receptor antagonist, CGS19755 (Selfotel) has been evaluated in the clinical setting with data reported from two phase III clinical trials. (145) A total of 693 TBI patients with a GCS of 4-8 and at least one reactive pupil were randomized to placebo or an intravenous infusion of 5 mg/kg of Selfotel once daily for 4 days. Due to possible increased deaths and serious brain-related adverse effects in the treatment arms of the two head injury trials and in two stroke trials, the Safety and Monitoring Committee stopped the trial prematurely. An interim efficacy analysis also indicated that it would be futile to continue the trial because the likelihood of demonstrating success with the treatment arm was almost nil. In the final analysis there was no statistically significant difference in mortality between the treatment groups. Cerestat (CNS 1102), a noncompetitive NMDA-receptor antagonist has also been evaluated in human TBI and did not show a positive effect. (131) CP-101,606 is a postsynaptic antagonist of the glutamate-mediated NR2B subunit of the NMDA receptor. A safety evaluation of CP-101,606 in patients with moderate or mild TBI or hemorrhagic stroke has been reported. Patients began receiving treatment within 12 hours of brain injury. The drug/placebo was administered by intravenous infusion for up to 70 hours. All the patients tolerated the drug, and there were no clinically significant cardiovascular, psychotropic, or hematological complications. (146) This agent is currently in Phase III testing and results are not yet available.
Some of the other agents currently being investigated include cyclosporine and magnesium. Cyclosporine transiently inhibits the opening of the mitochondrial permeability transition pore and maintains calcium homeostasis in isolated mitochondria. (147) Animals who received cyclosporine either 5 minutes before injury or immediately following injury show a significant reduction in the amount of cortical damage 7 days after TBI. (148) Another animal model administrated a 20 mg/kg intraperitoneal bolus of cyclosporine or vehicle 15 minutes post injury found that all animals receiving cyclosporine demonstrated a significant reduction in lesion volume. (147) Cyclosporine was also shown to inhibit calcium-induced axonal damage when administered 30 minutes after injures. (149) These investigations suggest that there is a therapeutic window for the use of drugs targeting mitochondria and energy regulation in traumatic brain injury. (149) However, it is yet to be seen if these positive laboratory results will translate into significant clinical benefit.
Magnesium is another agent now being investigated for its potential neuroprotective properties. An animal model has been used to demonstrate that magnesium plays a role in neuroprotection following severe diffuse traumatic axonal brain injury. Magnesium was shown to significantly improve motor outcome when administered up to 24 hours after injury. As would be expected, early treatments provided the most significant benefit and repeated administration beyond 24 hours post injury did not provide additional effects. (150) Magnesium appears to be promising but needs further evaluation before being tested in clinical trials.
Due to the ongoing high morbidity and mortality associated with conventional treatment of severe TBI, novel therapeutic strategies are being developed. One of these strategies is “Lund Therapy” of posttraumatic brain edema, named after the Lund University Hospital in Sweden. (151) This therapy attempts to prevent cerebral hypoxia while simultaneously counteracting transcapillary filtration. The Lund treatment protocol includes the following: 1) preservation of normal colloidal-absorbing force; 2) reduction in intracapillary pressure through a reduction in systemic blood pressure by antihypertensive therapy (metoprolol combined with clonidine); and 3) simultaneous, moderate constriction of precapillary resistance vessels with low-dose thiopental and dihydroergotamine. A cohort of 53 consecutive severe TBI patients admitted from 1989-1994 who were treated with Lund Therapy was compared with a historical control group of 38 severe TBI patients treated from 1982-1986. Fewer patients in the Lund Therapy group died (8% vs. 47%, p<0.0001), and the ratio of patients with a favorable outcome was significantly higher in the Lund Therapy group (p<0.001). (151) Due to methodological weaknesses of this study design, these hypothesis-generating results will need to be confirmed in a randomized clinical trial. (151)
The future of pharmacological neuroprotection is unclear at the present time. Several compounds have been tested and none have been shown effective in the clinical setting. The need for clinically effective agents is apparent especially in light of the devastating morbidity and mortality following severe TBI. It is the hope of many clinicians that these previous negative studies have taught us more about this injury and possible ways to plan clinical trials to avoid these same problems in the future.
TBI causes significant mortality, morbidity, and contributes substantially health care costs. Understanding normal cerebral physiology and altered physiology due to primary and secondary injury is essential to the identification of potentially useful therapies. Unfortunately, many of the specific therapies evaluated to date have not been shown to provide clinically important benefits.
CSF drainage and diuretics have failed to show a benefit on neurological outcome or mortality. Despite acute reductions in ICP and improvements in CBF with mannitol therapy, there are insufficient data to preclude either a harmful or beneficial effect on mortality. Hyperventilation may acutely control ICP; however, it has not been shown to improve outcome, and may worsen outcome if it is used routinely for long periods of time. Barbiturates may also improve surrogate physiologic markers in TBI, but they have failed to improve outcome, and are associated with a risk of hypotension in one out of four patients that may adversely affect CPP. Anticonvulsant prophylaxis has been shown to prevent early seizures after TBI, but failed to show a beneficial effect on outcome. Induced hypothermia reduces ICP, CBF, improves CPP, and meta analysis data have suggested that when used for less than 48 hours induced hypothermia is associated with a reduction in death or disability in TBI patients. Controversies regarding induced hypothermia may be resolved when a National Acute Brain Injury Study of hypothermia is published later this year. Meta analyses have failed to show a beneficial effect on mortality or neurological outcome in TBI patients treated with corticosteroids; however, a small but important beneficial effect of corticosteroids cannot be ruled out. Although a large clinical trial evaluating corticosteroids may be warranted, at this point routine corticosteroid therapy cannot be recommended.
Due to the disappointing results with specific therapies for TBI to date, specific experimental therapies and neuroprotective agents will continue to be evaluated. To date, many promising experimental neuroprotective strategies have not been proven beneficial in the clinical setting, but further studies are needed. Studies comparing overall management strategies that include therapies not shown to be harmful would also be beneficial and may enhance the effect size shown in the experimental group. Thus, until more data are available, specific goals of therapy published in TBI guidelines and consensus articles should be followed: 1) controlling ICP less than 20-25 mm Hg; 2) maintaining CPP at or above 70 mm Hg or MAP above 90 mm Hg in the absence of an ICP monitor; 3) optimizing cerebral oxygenation by ensuring normal coupling between CMRO2 and CBF and maintaining CEO2 below 42% (CERO2 less than 40%); (1,32) and 4) preventing and/or treating cerebral and systemic complications of TBI is also essential in these patients.