Hypoxia-ischemia, hypoglycemia, and status epilepticus induce energy failure with consequent brain damage (1,2,3). The significant differences in the time course and distribution of brain damage that result from these three insults are depicted in Table 17.1 and are reviewed by Auer and Sutherland (4). Brain consumes disproportionate amounts of oxygen, has scant glycogen stores, and tolerates hypoxia and hypoglycemia less well than most organs (5). Neurons need constant oxygen and glucose supply to maintain ion and neurotransmitter gradients.

Oxygen pressure is not uniform throughout the brain. It is higher in gray matter than in white matter, as is blood flow and glucose utilization. Brain glucose and oxygen consumption per gram of tissue are higher in neonatal than adult brain (4). The first effect of experimental cerebral hypoxia is an increase of intracellular pH. Subsequently, intracellular calcium content rises as consequence of calcium release from the endoplasmic reticulum. ATP concentration begins to fall, and when 50% to 70% of neuronal ATP is lost, the sodium pump fails and voltage-controlled ion channels open, permitting Na⁺, K⁺, Ca⁺⁺ and Cl^– to flow down their concentration gradients and release stored neurotransmitters (6). Water follows because of increased osmolality, and cells swell. Neuronal intracellular calcium concentration can then increase by up to four orders of magnitude. These huge increases in cytosolic [Ca⁺⁺] activate lipases, proteases, and other catabolic enzymes.

Changes in oxygen pressure have rapid, direct effects on membrane ion channels, mostly phosphorylation-related (7). Some ion channels are down-regulated, reducing ion flux and cellular energy demands (8). Others ion channels are up-regulated, promoting depolarization and cell death (9). Hypoxia also induces a variety of molecules called hypoxia-inducible factors (HIF). HIFs are induced more slowly than the effects on ion channels. They are expressed in all tissues and activate transcription of genes that increase systemic O₂ delivery or provide cellular metabolic adaptation under conditions of hypoxia. Thus, they activate the transcription of the gene for erythropoietin, the genes for glycolytic enzymes, and genes involved in angiogenesis. Juul has reviewed the use of erythropoietin to protect against neonatal ischemic brain injury (10); a recent study reported that it decreased brain injury when given to very young rats before hypoxic-ischemic encaphalopathy (HIE) (11). Further investigation is warranted, but the dismal results of human neuroprotective drug trials to date imply that clinical use of these drugs may be far in the future (12). The factors that mediate the induction of HIFs are under intense investigation (13). They are related to the phenomenon of ischemic tolerance, or preconditioning. This term refers to the observation that a brief period of cerebral ischemia confers transient tolerance to a subsequent ischemic insult to the brain (14). The mechanisms underlying ischemic tolerance are not fully understood. Kirino has suggested two possibilities: (a) A cellular defense function against ischemia may be enhanced by the mechanisms inherent to neurons. They may arise by post-translational modification of proteins or by expression of new proteins via a signal transduction system to the nucleus. These cascades of events could strengthen the influence of survival factors or could inhibit apoptosis; (b) a
cellular stress response and synthesis of stress proteins can lead to an increased capacity for health maintenance inside the cell (14). The mechanisms of ischemic tolerance or preconditioning may ultimately become relevant for therapy of HIE (15).

The mature brain requires both oxygen and glucose but can adapt to reduced serum glucose levels (16). The fetal and neonatal brain also uses ketone bodies as an energy source (17); these metabolic pathways can be increased by diet and fasting (18). In vitro studies show that low glucose media increase hypoxic injury and that this effect correlates well with tissue ATP content (19).

Two types of cell death have been recognized: apoptosis and necrosis. Biologists reported extensive cell death in developing organisms in the 19th century but did not consider cell death to be a central part of normal development (26).
Lockshin’s 1965 PhD thesis on silkworm metamorphosis introduced the phrase “programmed cell death” (PCD) (27). The Australian pathologist John Kerr studied human liver disease and noted an orderly form of cell death characterized by cell shrinkage with nuclear compaction and no inflammation. He first called this shrinkage necrosis (28). Peripheral areas where shrinkage necrosis predominated sometimes surrounded a central zone of coagulation necrosis. Kerr and colleagues later replaced “shrinkage necrosis” with the Greek word apoptosis (29). They saw mitosis and apoptosis as opposing forces, postulating that apoptosis inhibited tumor growth and development. Wyllie reported DNA fragmentation in cells undergoing apoptosis (30). Gel electrophoresis revealed “DNA ladders.” Apoptosis or PCD became a focus of attention, and studies of its cellular control mechanisms in the nematode C. elegans led to a Nobel Prize (31). Vertebrates produce too many neurons; programmed cell death prunes the numbers of neurons and glia. Mice with caspase gene knock-outs (caspases are proteases that trigger some forms of PCD) have gross central nervous system (CNS) malformations (32). Pathologists accepted the idea of qualitatively different forms of cell death, but they initially limited this to two types.

Lucas and Newhouse and Olney defined neuronal cell death caused by excessive glutamate receptor stimulation or “excitotoxicity” (33,34). It was characterized by swelling and bursting of cells, release of lysosomal enzymes, and subsequent inflammation. They called this phenomenon necrosis; any cell may suffer necrosis in the older sense of cell death followed by inflammation; excitotoxic cell death is a special form of necrosis limited to neural tissue with glutamate receptors. Standard teaching about HIE came to include the idea that glutamate leakage from presynaptic stores increased calcium and sodium influx and aggravated the brain injury (35). The characteristics of necrosis and apoptosis are summarized in Table 17.2. Coagulation necrosis refers to the death and disruption of a group of cells and is usually due to ischemia. It can often be detected by magnetic resonance imaging (MRI) in that inflammation and increased water content become visible in the T2 and spin-echo sequences, and contrast enhancement is common, due to breakdown of the blood brain barrier. Apoptosis is not associated with inflammatory changes and is generally invisible to MRI. Its demonstration in the postmortem brain can be difficult because of technical problems and because apoptotic cells are often cleared in minutes or hours, and can disappear within a few days (36). Karyorrhexis can indicate apoptosis but is not specific. The DNA digestion pattern recognized in apoptotic thymocytes by Wyllie gave rise to the TUNEL (terminal deoxynucleotidyl transferase dUTP nick end-labeling) method for identification of apoptosis (37). However, the specificity and sensitivity of the TUNEL method has been increasingly doubted (38), especially in postmortem brain tissue. Kerr defined apoptosis by electron microscopy; this technique is presently the best way to distinguish types of cell death (28). The various forms of PCD have been characterized by Clarke (39) (Table 17.3).

Cell death morphology depends upon tissue, cell type, and age of the animal; severity and duration of asphyxia; and the density and characteristics of cellular receptors. These factors are reviewed in Chapter 6. Mixtures of necrosis and PCD may be found in the same section or same cell, with intermediate forms of cell death that include attributes of both PCD and necrosis (40,41). Inhibition of one form of cell death may appear to stimulate another form (42).

The histologic manifestations of anoxic-ischemic brain injury take time to become evident (4,46). Sometimes, there is partial clinical recovery after anoxic events, followed by dramatic clinical deterioration. This syndrome occurs with strangulation (55) and with many forms of anoxia, including near drowning (56,57). It is most dramatic after carbon monoxide poisoning; this phenomenon is considered
in Chapter 10. Most patients with delayed deterioration, whether their initial injury was due to anoxia or carbon monoxide poisoning, develop extensive deep white matter injury—“Grinker’s myelinopathy”(58). However, some patients, such as Dooling and Richardson’s 11-year-old patient (55) and other pediatric cases (59,60) have only gray matter lesions.

The goals of treatment for cardiac arrest and postanoxic coma whatever its cause are to stabilize cardiac function, prevent further brain injury including delayed encephalopathy, and to maximize recovery of function. Cardiac arrest typically occurs at unexpected and inopportune times. Therefore, advance protocols or treatment outlines are desirable. Numerous treatment regimens have been used for cerebral salvage. These include induced hypothermia, barbiturate coma, and intracranial pressure monitoring to control cytotoxic cerebral edema. None of these has been effective in improving the ultimate outcome (75,85). The neurologist attending a near-drowning victim should keep in mind that hypoglycemia and hyperglycemia can cause further neurologic damage. Hyperthermia should be avoided and seizures controlled, with phenytoin being the preferred anticonvulsant. Animal studies suggest that vasopressin may be superior to epinephrine for resuscitation after cardiac arrest, and some suggest using both drugs (86). However, a human study failed to show a clear difference in outcome when either vasopressin or epinephrine was used (87). A study of 68 pediatric patients found that high-dose epinephrine was no better than and possibly inferior to standard-dose epinephrine given after an initial unsuccessful standard epinephrine dose (88). Neither magnesium sulfate nor diazepam given after out-of-hospital resuscitation improved outcome (89). Moderate hypothermia (12 to 24 hr at 32° to 34°C) improved outcome of comatose survivors of out-of-hospital cardiac arrest in two small European studies (90,91). The details associated with hypothermia are important—such as use of paralysis and sedation and possible interaction between hypothermia and various other drugs. Volume restriction is generally unwise; Ginsberg (92) advises the use of albumin infusions not only to increase plasma volume and decrease blood viscosity (93), but also to bind fatty acids and other toxins. Glucocorticoids worsen the effects of HIE in vivo and oxygen-glucose deprivation in vitro in that they induce hyperinsulinemia and hyperglycemia and have complex effects on glucose transport and energy metabolism (94,95). They also impair neuronal and astrocytic glucose uptake and inhibit astrocytic uptake of excitatory amino acids promiscuously released during anoxic insults (96).

Although there are no current standards of practice or uniformly accepted regimens at this time, I recommend the following:

The concept of brain death developed once it became possible to maintain vital functions in patients with massive brain injury. Studies during the 1960s suggested that patients without evidence of brain electrical activity did not recover unless their condition was due to drug overdose. However, those studies included few children and no neonates.

In 1987, the Ad Hoc Committee on Brain Death from the Children’s Hospital, Boston, defined brain death as follows:

Recommendations made in 1987 by a special task force appointed to set guidelines for determining brain death in children have been published (102). Although it is generally recognized that particular caution should be exerted when diagnosing brain death in small children, the task force further emphasized this age distinction by recommending different brain death criteria for infants between 7 days and 2 months of age, between 2 months and 1 year, and those older than 1 year. The period of observation before declaration of brain death in the youngest group should be such that two examinations and EEGs to document electrocerebral silence are performed, separated by at least 48 hours. In the group from 2 months to 1 year of age, the interval between the two examinations and EEGs can be reduced to 24 hours. The clinical examination of the pediatric patient suspected of being brain dead is summarized in Table 17.6 (103,104,105). A repeat examination and EEG are not necessary in this group if radionuclide angiography demonstrates absent cerebral blood flow (106). In children older than 1 year, the task force recommended the period of observation be a minimum of 12 hours, unless corroborating tests added further support to the diagnosis of brain death. When the extent and reversibility of brain damage are difficult to assess because of the type of insult (e.g., hypoxic-ischemic encephalopathy), the observation period should be extended to at least 24 hours. The confounding problem today is that most children die in PICUs and receive many medications in the last days of life. It is likely that “therapeutic and falling” levels of barbiturates and other anticonvulsants do not invalidate ordinary clinical and EEG criteria for brain death (107), but drug metabolism can be very slow in the brain dead patient, and high levels of other drugs such as opioids are often present. EEG and evoked potentials may occasionally become unobtainable, only to return hours or days later (108).

An important aspect in diagnosing brain death is the documentation of apnea. During this procedure, it is vital to prevent hypoxemia. Administration of 100% oxygen for 10 minutes is recommended before withdrawal of respiratory support. A catheter should be inserted into the endotracheal or tracheostomy tube, and oxygen be continued at 6 L per minute during the test. The arterial pCO₂ level
should be allowed to increase to 60 mm Hg. Patients who are hypothermic or receiving medications that suppress respiration cannot be reliably tested using this procedure (70).

Some authorities have therefore challenged these recommendations, and a survey of PICUs shows substantial variability, even within the same unit, with respect to criteria used by clinicians for the diagnosis of brain death (109). In a clinical and neuropathologic study of brain death, Fackler and coworkers found no support for employing distinct brain death criteria for infants between 2 months and 1 year of age (110). Other investigators question the validity of relying on the EEG to confirm brain death because EEG activity is occasionally seen after brain death (111). Conversely, phenobarbital levels above 25 to 35 μL can suppress EEG activity in neonates (112). The brainstem auditory-evoked response cannot be used as a confirmatory laboratory criterion of brain death. Its absence is not predictive of brain death, and persistence of peak I has occasionally been seen in brain dead infants (113).

Demonstration of cerebral circulatory arrest is the most convincing proof of brain death and is not influenced by drugs or hypothermia. In principle, this can be done by transcranial Doppler ultrasonography in the hands of an experienced technician (114), by radionuclide angiography (115), or by spiral CT (116). However, these methods are less precise than angiography or MRI. Ruiz-Garcia and colleagues compared EEG, radionuclide angiography, and evoked potentials in 125 brain dead Mexican children. There was considerable concordance, but only 61 of 76 children who were subjected to all three studies had positive results on all three (115). Although complete absence of cerebral blood flow is considered irrefutable evidence of brain death, it is necessary to consider that cerebral blood flow is extremely low in normal term or preterm newborns (117).

Although CT angiography (118) and MRI and MR angiography (119) offer even more precise demonstration of cerebral circulatory arrest, they require moving the patient to the radiology suite. High-resolution brain CT scans with contrast are quite satisfactory for this purpose. If there are no flow voids in a technically satisfactory MRI of the brain, the patient is brain dead.

I believe that as these more sophisticated imaging techniques are applied to the clinical evaluation of brain death in children, the criteria for making this diagnosis will become refined and perhaps simplified.

Hyperoxia is iatrogenic, the result of increased oxygen content of inspired air or hyperbaric oxygen therapy. The question is when and how hyperoxia under normal atmospheric pressure causes neuronal or glial cell death. Hyperoxia is well known to cause cell death in the eye and lung. Hu and coworkers found that both hypoxia and hyperoxia caused cell death with apoptotic features in 7-day-old rat cerebral cortex; hyperoxia after hypoxia did not prevent cell death (120). Several animal studies show mild deleterious effects of normobaric hyperoxia used for resuscitation, and hyperbaric oxygen treatment can cause seizures and permanent brain injury in adults (121,122). Children breathing 100% oxygen develop hyperintense cerebrospinal (CSF) signs on FLAIR (FLuid Attenuated Inversion Recovery) sequences. These partially or completely disappeared when FiO₂ was reduced to 30% (123). The controversies with respect to the value of oxygen in resuscitation have been reviewed by Saugstad (124).

Hyperglycemia aggravates ischemic brain injury. Rats made hyperglycemic before cerebral ischemia have greater brain damage and mortality (167). This is understandable from Myers’s studies showing that feeding prior to cardiac arrest resulted in postischemic lactic acidosis (168). Even if hyperglycemia is induced after the ischemic period, recovery of function is impaired compared with normoglycemic controls (169). Therefore, normoglycemia is aimed for in acutely ill patients and patients undergoing cardiac surgery. In adult patients, hemiparesis, confusion, and movement disorders may be seen with either hypoglycemia or hyperglycemia (170).

Infants are more susceptible to hypothermia than older children. As body temperature falls progressively, many complications ensue (184,185). Electrolyte abnormalities and disseminated intravascular coagulation are common in both sustained hyper- and hypothermia (186). Hypokalemia is the most common electrolyte abnormality seen with hypothermia and represents a shift of potassium into cells. If extra potassium is given, severe complications may arise when the patient is rewarmed (187). Hypothermia accompanied by vomiting and coma are common manifestations of alcohol intoxication in prepubertal children in cold climates (188). Several rare syndromes of periodic hypothermia exist (189,190), including periodic hypothermia and hyperhidrosis, sometimes associated with callosal abnormalities and called Shapiro’s syndrome (191). These periodic diencephalic disorders may respond to clonidine or cyproheptadine and may be related to migraine (192).

Calcium is the major extracellular divalent cation. Both high and low serum calcium levels are associated with neurologic symptoms. Total calcium in serum is found in three forms: protein bound (therefore nondiffusible, 30% to 55% of total); chelated (i.e., diffusible but nonionized, 15% of total); and ionized (remaining percentage). Generally, the appearance of neurologic symptoms correlates well with levels of ionized calcium of 2.5 mg/dL or less. The concentration of CSF calcium is normally approximately one-half that of serum calcium and represents the result of a secretory process, rather than the movement of diffusible and ionized calcium from the serum. Changes in the CSF concentration are relatively small, although large alterations in serum calcium values overcome homeostatic mechanisms.

Phosphate depletion and hypophosphatemia has been associated with a rare and poorly defined encephalopathy (272,273,274). This is usually a complication of total parenteral nutrition (273,274). Patients may show tremor, confusion, agitation, ophthalmoplegia, and coma, and some are areflexic. Hypophosphatemia appears to contribute to central pontine myelinolysis in a few patients with Wernicke’s encephalopathy, including patients with hyperemesis gravidarum as the primary problem (275). Hypophosphatemia and encephalopathy have been seen in anorexic patients receiving hyperalimentation.

The various disorders of iron and other metals are covered in Chapter 1.

In acute liver failure, the morphologic changes in the brain are dominated by astrocytic alterations, notably astrocytic swelling and cytotoxic brain edema (325). With progression of brain edema, intracranial pressure increases and ultimately results in cerebral herniation. In chronic liver failure, the principal microscopic abnormalities include enlargement and an increased number of protoplasmic astrocytes. These cells (Alzheimer II cells) are astrocytes with an enlarged, pale nucleus and a marked diminution in glial fibrillary acidic protein. They are found throughout the cerebral cortex, basal ganglia, brainstem nuclei, and the Purkinje layer of the cerebellum (322,324). They also are seen in human immunodeficiency virus (HIV) encephalopathy (see Chapter 7). Neuronal changes are generally not seen. Less often, central pontine myelinolysis has been noted in children with hepatic failure (326).

According to current consensus, HE is multifactorial and in part represents a failure of glioneural communication and cooperation (327,328,329). The two most important factors in its pathogenesis are increased plasma and brain concentrations of ammonia. The brain:blood ratio of ammonia concentrations is markedly elevated, probably a consequence of a disrupted blood–brain barrier (324). In the brain, ammonia is converted to glutamine, which cycles from astrocytes to neurons, where it is further converted to glutamate. After release of glutamate into the synaptic cleft, its reuptake occurs in astrocytes. It has been postulated that the increased synthesis of glutamine depletes available amounts of α-ketoglutarate and reduces the concentration of high-energy phosphates, thus slowing the reactions in the Krebs tricarboxylic acid cycle. Decreased oxygen consumption and glucose metabolism, however, are secondary to HE rather than causative (328).

Although in the past enhanced GABAergic neurotransmission was cited as a cause for hepatic encephalopathy, the evidence for this effect is far from convincing. In a vast majority of HE models, there are no alterations of GABA content in the brain tissue and/or extracellular space, and most of the substances in brain and other tissue with GABAergic properties can be traced to the ingestion of benzodiazepines (324).

HE is associated with multiple secondary biochemical abnormalities. MRI studies have shown increased signal in the globus pallidus in T1-weighted images, and an increased concentration of brain manganese is believed responsible for these findings. Brain copper concentrations also are increased, and as is the case in Wilson disease, the accumulation of the toxic metals can alter astrocyte function and morphology (324).

Evidence for the synergistic role of other neurotoxins such as mercaptans, short-chain fatty acids, and phenols as well as the generation of false neurotransmitters such as octopamine is currently less strong (328). Additionally, liver failure induces profound multisystem disturbances, which, in turn, can further impair neurologic function (330). Of significance are gut-derived bacteria and their toxic products, which are known to injure the liver and cause systemic illness (331,332,333). Serum levels of proinflammatory cytokines are increased in hepatic encephalopathy, and the severity of encephalopathy has been correlated with serum levels of tumor necrosis factor α (334).

HE can occur in two forms: acutely, as in fulminant hepatic failure, and as a chronic, progressive encephalopathy. In children, acute hepatic failure is primarily responsible for clinically important HE (335). The most common predisposing causes are acute infectious hepatitis; Wilson disease (336); the ingestion of various drugs such as valproic acid, acetaminophen, isoniazid, or halothane; and toxins, notably mushroom poisoning (337). In infancy, galactosemia, fructosemia, or tyrosinemia can present as fulminant hepatic failure (see Chapter 1). In the past, Reye syndrome and hemorrhagic shock syndrome presented with fulminant liver failure (see Chapter 7).

The onset of the encephalopathy usually coincides with a deterioration of the general clinical condition. The principal signs and symptoms of hepatic coma are related to disorders of consciousness. Neurologic symptoms in hepatic failure generally can be divided into six groups: those due to hypoglycemia; sepsis; intracranial bleeding resulting from coagulopathies; renal failure; electrolyte disturbances, notably hyponatremia, hypokalemia, and hypocalcemia; and cerebral edema. The various stages of HE are outlined in Table 17.13.

It is of utmost importance that the first signs of encephalopathy are recognized, as the progression from stages I to IV can be exceedingly rapid. Because the first evidence of encephalopathy may be outbursts of violent agitation or uncharacteristic behavior, the early stage of HE is frequently misdiagnosed. Hyperventilation can develop during stages II and III and can lead to alkalosis, low serum pCO₂, and a further deterioration of mental status. A fine tremor, and more characteristically coarse flapping movements, termed asterixis, can be present in stages I and II, respectively, whereas decorticate and decerebrate postural responses accompany stage IV of HE. Choreiform movements, a fluctuating rigidity of the limbs, parkinism, dystonia, and periods of noisy delirium are particularly frequent asian indian (338,339).

Poddar and colleagues reported 67 Asian-Indian children with fulminant hepatic failure treated without transplantation, 63 of whom were thought to have viral hepatitis (340). All children with stage I or II encephalopathy survived. Seventeen of 36 children with stage III or IV encephalopathy died. Ascites, peritonitis, and chemical evidence of severe hepatocellular dysfunction all increased the risk of death. Cerebral edema is a prominent part of the clinical picture of acute HE and is the principal cause of death, with brainstem herniation found in up to 80% of patients dying in fulminant hepatic failure (341).

Although severe liver disease is a prerequisite for the appearance of HE, ascites, jaundice, edema, or hepatomegaly do not invariably accompany the neurologic involvement. In fact, frequently, as irreversible liver failure supervenes, previously elevated serum transaminase levels decrease rapidly, the coagulopathy worsens, the initially enlarged liver shrinks, and the total bilirubin climbs while the conjugated portion decreases.

In a majority of patients, the EEG shows paroxysmal and diffuse bursts of high-voltage slow-wave activity, a pattern that is not specific for HE but is highly indicative of one of the metabolic encephalopathies. Triphasic waves, characteristic for HE, are common in adults but rare in children. Clinical or EEG evidence of seizures is associated with a poor outcome (342).

The advent of successful liver transplantation, which offers a success rate of between 55% and 89%, has revolutionized the management, treatment, and prognosis of children with liver failure and HE (343,344). Liver transplantation is the definitive treatment in patients with acute or chronic hepatic failure, and in several major centers, including our own, results have been encouraging both in terms of survival rate and post-transplant complications. The decision of which patients to transplant and when surgery is to be done is beyond the scope of this text. Suffice it to say that the mortality of patients in stage IV HE is 63% to 80%. In particular, patients who have developed cerebral edema fare badly and are not suitable candidates for liver transplant (344). In the experience of the Children’s Hospital of Pittsburgh, 70% of children who developed cerebral edema as demonstrated by computed tomography (CT) scan died, and 15% were left with severe to profound neurologic deficits. The remainder were left with moderate deficits that prevented them from an independent lifestyle (344).

The child with HE requires meticulous medical management until either the liver resumes adequate function or a replacement organ is found. Management of the precipitating event, dietary protein restriction, avoidance of constipation, and alteration of the intestinal flora are the major aspects of the therapeutic regimen (345). The therapeutic value of flumazenil, a benzodiazepine antagonist, appears to be minimal in the pediatric population (346). Response to flumazenil is transient, and the medication may precipitate an anaphylactic reaction.

Most commonly, the neurologist becomes involved in the treatment of a child with HE who has developed cerebral edema. Because cerebral edema mainly occurs on a cytotoxic basis, corticosteroids are not recommended for its treatment and were found to be ineffective in controlled trials. The best course of management is by fluid restriction and assisted hyperventilation. Hyperventilation is widely used and seems helpful early in the course of the illness. Cerebral blood flow tends to increase as encephalopathy develops if ventilation is not controlled, and this relative hyperemia contributes to intracranial hypertension (347). Late in the course of hepatic failure, cerebral blood flow tends to decrease, and hyperventilation at this stage can be harmful (348). Mannitol and related osmotic diuretics are helpful, but prolonged use of mannitol can cause hyperosmolality and aggravate renal problems (349,350). Minimizing stimulation (lights, sound, endotracheal suctioning) avoids sharp increases in intracranial pressure that are liable to become sustained and recalcitrant to therapeutic measures. Short-acting narcotics (fentanyl) can be administered to further blunt intracranial pressure increase from stimulation. Hypothermia and barbiturate coma have been advocated but should only be used if intracranial pressure is monitored.

Extradural or subdural monitoring devices are increasingly used for children in stage III or IV (350). These allow management of intracranial pressure and permit documentation of cerebral hypoperfusion and the rapid fluctuations of intracranial pressure encountered during the transplant procedure (351). Placement requires an experienced neurosurgeon and often necessitates simultaneous administration of fresh frozen plasma. In the experience of Blei and coworkers, a potentially fatal hemorrhage was the most common complication of cerebral pressure monitoring in fulminant hepatic failure (352).

This is not the place to discuss the complexities of artificial liver support (353) and the details of managing severe hepatic failure. For a comprehensive discussion of the therapy of hepatic failure, the interested reader is referred to reviews by Jalan (354), Sherlock (349), and DeVictor and coworkers (350).

Suffice it to say that hepatic transplantation is no solution unless intracranial pressure is controlled. Sustained increases in intracranial pressure resulting in diminished cerebral perfusion pressure (less than 40 torr for more than 2 hours) are generally accepted as a contraindication to liver transplantation (355). In addition, the cause of hepatic failure and a variety of other prognostic indicators must be considered (356). In particular, the longer the interval between the onset of jaundice and the development of HE, the worse the outcome (357). As a rule, somatosensory-evoked potentials are superior to EEG in terms of prognosis, and a lack of a thalamocortical potential presages a poor outcome (358).

The molecular basis for uremic encephalopathy remains complex and poorly understood; it is generally accepted that several toxins are responsible. Urea is the most studied of these. However, it has been known for some time that the severity of cerebral symptoms correlates poorly with levels of serum urea, and hemodialysis sometimes reverses symptoms without lowering blood urea (395). Creatinine, p-cresol, the guanidines, organic acids,
phosphates, and secondary hyperparathyroidism also are believed to contribute to the encephalopathy. Parathormone is believed to be responsible for some aspect of the various neurologic symptoms encountered in uremia, notably the peripheral neuropathy and the myopathy (396). Cerebral blood flow studies have shown a defect in oxygen use. In part, this defect might be caused by nonspecific increases in brain permeability and disordered membrane function, which could allow toxic products, possibly a variety of organic acids, to enter the brain. These acids could alter the function of the sodium-potassium ion pump. Disorders in blood and CSF electrolytes can aggravate the clinical picture, as can bouts of acute hypertensive encephalopathy (397).

The principal neurologic symptoms of uremia are abnormalities in mental status, tremor, myoclonus, asterixis, convulsions, and muscle cramps (239,398). As a rule, the more rapidly renal failure develops, the more prominent the motor symptoms, which may include unclassifiable twitches that are bilateral but asynchronous, and seizures. The motor disorder is usually relatively symmetrical. Peripheral nerve involvement is common in patients with uremia. Most frequently, it takes the form of a polyneuropathy. This can be a symptomatic mixed motor and sensory neuropathy, or it can be subclinical, detected only by nerve conduction studies. In one report, 76% of uremic children had a significantly reduced peroneal motor nerve conduction velocity without any clinical evidence of neuropathy (399). When symptoms develop, they begin with sensory abnormalities in the lower extremities. The condition can progress slowly to total flaccid quadriplegia. Nerve biopsy can reveal primarily an axonal neuropathy, progressive axonal neuropathy with secondary demyelination, or predominantly demyelinating neuropathy (400). Less commonly, patients develop a mononeuropathy, cranial nerve palsies, and choreoathetosis. Restless legs syndrome is seen in a large proportion of uremic patients (401). Signs of hypocalcemia and hypomagnesemia are often present. On rare occasions, one can encounter a primary myopathy (402).

EEGs are generally slowed and sometimes include spike-wave discharges in patients without clinical seizures (403). Rhythmical EEG discharges without clinical seizures are more common in renal failure than in children without renal disease; many of these patients do not demonstrate paroxysmal discharges when monitored (404).

In hypertensive encephalopathy, such as occurs with acute glomerulonephritis, patients develop symptoms and signs of increased intracranial pressure, with headache, vomiting, disturbance of vision, and papilledema. Seizures and transient focal cerebral syndromes, including hemiparesis and cortical blindness, also are common.

As a rule, developmental quotients of children who develop chronic renal failure before 1 year of age are more affected than those of children who go into uremia after 3 years of age (405). In the experience of McGraw and Haka-Ikse, more than 50% of patients with chronic renal failure present since infancy had significant developmental delay (406). This is accompanied by a significant reduction in the head circumference.

Neuroimaging studies of the brain of patients with end-stage uremia reveal a high incidence of cerebral atrophy, suggesting an adverse effect of uremia on brain development (407). MRI of patients with cortical blindness demonstrates increased signals in occipital white matter and cortex on T2-weighted images. With treatment, these tend to resolve over the ensuing weeks (408). The neurologic symptoms in uremia have been reviewed by Fraser and Arieff (409) and Smogorzewski (410).

Treatment of uremia involves correction of electrolyte disturbance and maintenance of normal plasma composition. These have been greatly assisted by the use of dialysis. In some instances, neurologic symptoms can become aggravated after peritoneal dialysis or hemodialysis. Some workers have suggested that urea in the brain does not equilibrate freely with urea in blood, and therefore water enters the brain along an osmotic gradient. This is generally referred to as the dialysis-dysequilibrium syndrome. Gradual changes in blood electrolytes and earlier dialysis prevent some neurologic complications. In general, motor symptoms tend to improve once blood urea levels are lowered, whereas sensory symptoms tend to remain fixed. The sensory neuropathy does, however, respond dramatically to renal transplant (411). The correction of anemia by means of recombinant erythropoietin improves intellectual function (412).

Successful renal transplantation is associated with acceleration in head growth and improved intellectual functioning. Improvement can continue for more than 1 year after the transplant (413,414). Nevertheless, prospective studies of children with moderate to severe congenital renal disease indicate that both cognitive and motor developmental delay is common. Children who develop uremia in the first year of life often have microcephaly and significant cognitive impairment even if dialyzed and later transplanted (406,415). The younger the child, the greater the risk of this complication (416). This delay in brain growth reflects, in part, a toxic effect of uremia on brain growth and maturation, and, in part, chronic malnutrition, the various metabolic disturbances, and also any antecedent brain malformation. In the series of Bock and coworkers, approximately one-half of infants with congenital
renal disease maintained normal development. In the remainder, development was delayed or deteriorated. Neither the cause for the renal disease nor its severity influenced the neurologic or cognitive status (417). Successful renal transplantation is associated with head growth and intellectual improvement (418,419). However, not all studies find better cognitive outcome following transplantation than that following dialysis (420).

Treatment of convulsions in uremic patients depends on the cause of the seizures. Seizures accompanying the dysequilibrium states are usually self-limiting and often can be prevented by close supervision of dialysis. Chronic seizures can be treated with phenobarbital or phenytoin, with recognition that serum protein binding, particularly for phenytoin, is reduced in uremia. As a result, therapeutic as well as toxic effects of the drug are encountered at lower serum levels than in patients who have normal renal function. Nevertheless, anticonvulsant activity can usually be achieved with the usual doses because the free fraction of phenytoin remains unchanged (421). No modification of carbamazepine clearance or bioavailability has been observed. In renal failure, the free fraction of valproate increases two- to threefold. However, the intrinsic metabolism of the drug is reduced so that the actual clearance remains normal. The metabolism of the various benzodiazepines is unaffected. Renal clearance of levetiracetam is up to 70% reduced in severe renal impairment; clearance of the drug generally correlates with creatinine clearance. Both phenytoin and phenobarbital are known to hasten the metabolism of corticosteroids and the immunosuppressant drugs cyclosporine and FK506. This causes ineffective immunosuppression in renal transplant recipients and reduced cadaver allograft survival (422). Hence, Wassner and coworkers have suggested that anticonvulsants not be administered to patients right after transplant unless absolutely essential, and if they are given, corticosteroid dosage should be increased accordingly (422). Alternatively, a benzodiazepine can be used. The various drug interactions are considered by Cutler (423). (See Chapter 14 for a discussion of anticonvulsant therapy of patients with renal failure.)