|
|
Posted 12/03/2003
Javier Vaquero, M.D., Chuhan Chung, M.D., Michael E. Cahill, B.A.,
Andres T. Blei, M.D.
Abstract and IntroductionAbstractHepatic encephalopathy (HE) in acute liver injury signifies a serious prognosis. Brain edema and intracranial hypertension are major causes of death in this syndrome. Comparison of HE in acute liver failure (ALF) with that of cirrhosis allows recognition of important differences and similarities. A key role for ammonia in the pathogenesis of both HE and brain edema is now firmly supported by clinical and experimental data. Additional factors, such as infection, products of the necrotic liver, and synergistic toxins, may contribute to an altered mental state. A low plasma osmolarity, high temperature, and both high and low arterial pressure may affect brain water content. A combined derangement of cellular osmolarity coupled with cerebral hyperemia can explain the development of brain edema in ALF. Increasingly, study of the mechanisms responsible for brain swelling provides critical information for understanding the pathogenesis of HE. IntroductionThe development of HE in patients with ALF signals a critical phase of the illness (also defined as fulminant hepatic failure)[1] and is associated with a reduced survival. In epidemiological studies performed in the pretransplant era, spontaneous recovery of liver function was 70% in stages I and II encephalopathy and was reduced to < 20% in stages III and IV encephalopathy.[2] Death in hepatic coma is common in patients with cirrhosis and advanced liver failure, but a unique feature of ALF is death from cerebral edema and intracranial hypertension A Clinical Overview: Comparison of HE in ALF and Cirrhosis
Encephalopathy in ALF shares features and exhibits differences
with the encephalopathy of cirrhosis. Five aspects deserve
specific consideration.
Grading of HEThe West Haven criteria, designed for clinical studies in cirrhosis,[3] have also been used in patients with ALF. However, the precise characteristics of each stage often overlap, and differences between stages I and II or between II and III can be blurred. Certain clinical features of ALF are not well-represented in this classification, especially severe agitation, which can be an initial neurological symptom in ALF and pose serious problems in management (including the need to sedate the patient with loss of neurological end points for follow-up). An excitatory behavioral phase is consistent with robust experimental findings of an increased extracellular brain glutamate in this condition.[4] Once stage IV encephalopathy is reached, the Glasgow coma scale, initially developed for patients with neurotrauma,[5] provides a numerical continuous score from 3 (worst) to 15 (best). Although it has not been formally evaluated in metabolic encephalopathies, it is better suited for examining patients in stages III and IV encephalopathy than the West Haven criteria, as was recently shown.[6] Precipitating FactorsThe pathogenic role of precipitating factors, well-recognized in the encephalopathy of cirrhosis, is often overlooked in ALF. Patients with acute liver failure may develop encephalopathy from the use of sedatives, as disturbances of sleep or agitation may be an early prodrome and are often medicated prior to arrival at a specialized center. Gastrointestinal hemorrhage, uremia, and electrolyte disturbances need to be ruled out. Infection, however, is the key precipitant to consider; the role of infection is discussed in the next section. SeizuresSeizures have traditionally been viewed as a rare event in hepatic encephalopathy. A retrospective review of electroencephalogram tracings in 94 patients with cirrhosis described epileptiform abnormalities in 14% of subjects with deep encephalopathy who did not receive a liver transplant.[7] Seizure activity has been reported in previous clinical series of ALF[8] and is a well-recognized complication of acute hyperammonemia in urea-cycle disorders.[9] In a recent controlled trial, subclinical seizure activity was detected in 10 of 22 patients enrolled as controls in a trial of prophylactic phenytoin in ALF.[10] Measurements of low oxygen saturation in the jugular vein led to the conclusion that poor cerebral perfusion and tissue anoxia were potential determinants of seizure development. At autopsy, patients in the nontreated group had greater evidence of cerebral edema. The high frequency of subclinical seizures reported in this series awaits confirmation from other centers. Brain EdemaDeath from intracranial hypertension has now been reported in patients with cirrhosis and deep hepatic encephalopathy in the setting of acute-on-chronic liver failure.[11,12] The magnetization transfer ratio, an indirect reflection of brain water content on spectroscopy, was clearly abnormal in patients with cirrhosis,[13] suggesting low-grade brain edema. The paradigm has shifted, with an increasing realization that a disturbance in brain water regulation is central to the process responsible for hepatic encephalopathy.[14,15] Nonetheless, a neurological death is a rare event in patients with cirrhosis. Cerebral PerfusionIn cirrhosis, a reduction in cerebral blood flow has been described in patients with overt[16] and minimal[17] encepha lopathy. A hyperdynamic circulatory state is a characteristic finding in liver failure, and the response of the cerebral circulation needs to be considered in this context. Recently, Guevara and colleagues[18] postulated a direct relation between the reduction in cerebral and renal blood flow in patients with cirrhosis and ascites. The decrease in perfusion of both territories was viewed as a response to systemic arterial vasodilatation, a sequence well-accepted for the renal vasoconstriction of cirrhosis.[19] The absence of signs of encephalopathy in these patients adds further credence to the view that the cerebral circulation also reacts to the generalized hemodynamic disturbance of liver failure.[20] In ALF, an initial reduction of cerebral blood flow (CBF) may reflect similar mechanisms.[21] However, a rise in CBF is prominently seen in patients with overt brain edema.[22] Pathogenesis of HE in ALF— Systemic Factors
Conceptually, hepatic encephalopathy arises from exposure of
the brain to circulating neurotoxins. In an early stage of
research in this area, the absence of a critical trophic
factor for brain function was postulated.[23]
Recently, this idea has been revived in experiments performed
in isolated liver-brain preparations.[24] However,
multiple elements point at the role of circulating toxins,
most conclusively the development of HE in the presence of a
normal liver.[25]
AmmoniaA pathogenic role for ammonia has been the focus of experimental and clinical studies. Death from cerebral edema and intracranial hypertension is well-recognized in children with urea cycle enzyme deficiencies and severe hyperammonemia.[9] In human ALF, arterial ammonia levels > 200 µg/dL were associated with cerebral herniation within 24 hours of reaching stage III-IV encephalopathy.[26] These data have been subsequently confirmed[27] and point at levels of < 150 µg/dL as a cutoff below which the risk of neurological death may be substantially decreased. Arterial sampling is important, because AV differences of ammonia can be considerable in ALF. In a recent human study, arterial concentration was 160 ± 53 versus 110 ± 35 µg/dL in the femoral vein, a significant difference.[28] Under normal circumstances (Fig. 1), the splanchnic release of ammonia, derived from the breakdown of glutamine in the intestine and the increased activity of colonic bacteria, results in 10-fold higher levels of ammonia in the portal vein. An efficient hepatic uptake mechanism, related to both urea (high capacity, low affinity) and glutamine (low capacity, high affinity) synthesis, results in tight control of ammonia levels reaching the periphery, with a hepatic extraction rate of 0.8 to 0.9.29 In ALF, hepatic vein measurements showed higher levels of ammonia (242 ± 118) than those seen in arterial blood (182 ± 80 µg/dL, n = 22).[28] In the setting of an acutely failing liver, ammonia levels in the hepatic vein are similar to those seen in the portal vein.
Muscle uptake of this increased ammonia load results in the
formation of glutamine. Whereas release and recirculation of
glutamine will result in the regeneration of ammonia,
splanchnic generation of ammonia appears saturable.[28]
Thus, the capacity of muscle to detoxify ammonia may be of
importance. Ornithine-aspartate stimulates muscle glutamine
synthetase in experimental ALF and prevents the development of
brain edema.[30] The role of amino acids in ammonia
disposal in ALF deserves further attention.
A net uptake of ammonia also occurs in the brain,[28] where it amidates both alpha-ketoglutarate and glutamate.[31] Glutamine is formed and cycles from astrocytes to presynaptic neurons, where glutamate is formed. After release into the synaptic cleft, reuptake of glutamate occurs in astrocytes. A profound alteration of this cycle has been demonstrated in experimental studies[31] and underlies the development of brain edema. Recent studies have raised the possibility that the kidney may be an important route for ammonia elimination in cirrhosis.[32] Such findings await additional confirmation. In any case, the extent of renal ammonia elimination in ALF may be affected by the development of renal failure, a common finding in this syndrome. InfectionA classic precipitant factor of the encephalopathy in chronic liver disease is the development of infection. Recent clinical observations indicate a strong association between parameters of infection and the course of encephalopathy in ALF[33] (see Bernal[34] in this issue). A recent report from the U.S. Acute Liver Failure Study group supports and extends these observations.[35] Only patients with early encephalopathy were analyzed. In a prospective evaluation of acetaminophen-induced ALF (n = 96), a positive diagnosis of infection preceded or coincided with the progression of stage I-II to deeper stages of encephalopathy in 79% of individuals. In subjects without demonstrable infection, a group that included both acetaminophen and nonacetaminophen etiologies (n = 168), a greater number of components of the systemic inflammatory response syndrome (SIRS) was associated with a stepwise progression of encepha lopathy from 25% (0 components), 34.7% (1 component), and 50% (2 to 3 components).[35] An explanation of the components of SIRS can be found in Table 1. How infection triggers encephalopathy in liver failure is poorly understood. The encephalopathy of sepsis is not similar to that of ALF.[36] Binding of cytokines to receptors in cerebral endothelial cells with subsequent signal transduction into the brain is a likely scenario.[37] Interactions of this process with other toxins, such as ammonia, have not been examined and may yield important clues to the pathogenesis of HE. The Necrotic LiverScattered reports indicate improvement of the clinical condition in ALF after total hepatectomy.[38] In two well-studied cases, intracranial pressure was reduced and liver transplantation successfully performed when a donor organ became available.[39,40] A reduction in liver-derived cytokines was suggested as a reason for this beneficial effect.[40] However, critical examination of this experience notes the development of mild to moderate hypothermia after removal of the liver. Reductions of temperature to 32 to 35°C have been associated with reductions in brain edema and intracranial pressure in both experimental models[41] and human ALF.[42] In a controlled trial of hypothermia in patients with head trauma, reduced levels of interleukin (IL)-1â accompanied body temperatures of 34°C.[43] At this time, the role of the necrotic liver in the development or progression of encephalopathy is uncertain. SynergismIn the mid-1970s, Zieve and Nicoloff[44] coined the concept of "synergistic toxins," in which a wide array of gut-derived substances potentiated ammonia's deleterious effects on the brain. These studies focused on mortality associated with ammonia administration to rats, noting a reduction of the LD50 of ammonia with the addition of short-chain fatty acids, mercaptans, and phenols. Octanoic acid had previously received attention as a putative cause of brain edema in Reye's syndrome.[45] Its role in ALF is uncertain. The impact of compounds that cross the blood-brain barrier and activate gamma-aminobutyric acid (GABA)-ergic pathways has undergone a vast change since originally proposed more than 20 years ago. Although the existence of endogenous benzodiazepine ligands in the brain of patients with ALF has been reported previously,[46] current evidence supports a potentiating effect of ammonia on GABA-induced neurotransmission.[47] These aspects are discussed in greater detail elsewhere in this issue.[48] Tryptophan is an amino acid whose levels are increased in the plasma of patients with ALF.[49] Its entry into the brain is favored by activation of the neutral amino acid carrier at the level of the blood-brain barrier in exchange for glutamine, the brain levels of which are increased as a result of ammonia detoxification in astrocytes. Tryptophan is a precursor of serotonin, but the role of serotoninergic abnormalities in the encephalopathy of ALF is uncertain. A report of increased brain quinolinic acid, a peripheral derivative of tryptophan, in human brain does not suggest a major role for this pathway in the encephalopathy of ALF.[50] Brain Edema: Part of the Spectrum of HE
For many years, the presence of brain edema was viewed as a
unique complication of ALF, a distinct entity from the
classic picture of HE. Our views on this separation have
undergone major changes in recent years. The results of in
vitro studies, animal experimentation, and human data point
at a common disturbance of water accumulation, present in
the entire spectrum of clinical manifestations (Fig. 2).
According to this view, the clinical expression of brain
edema, a rise in intracranial pressure, is prominent in ALF
but can also be detected in subjects with cirrhosis and deep
hepatic coma.[11,12] New technical developments
have allowed the estimation of an increased water content in
the brain of patients with cirrhosis,[13]
supporting the notion of low-grade brain edema.[14]
In 1999, we proposed a mechanism responsible for the
development of brain edema based on a combination of
experimental and clinical observations.[51] An
initial osmotic disturbance of the brain, when combined with
an increase in cerebral blood flow, results in this unique
complication of liver failure. Many of our views of the
pathogenesis of brain edema and intracranial hypertension
have originated from work in the rat after portacaval
anastomosis receiving an ammonia infusion. Although this
model is not one of ALF, the reliable development of brain
edema within a few hours of infusion allows the study of
factors responsible for swelling in the absence of
confounding variables seen in the setting of ALF. Such
variables may also be critically important and will be
reviewed after we espouse our basic concepts.
An Osmotic DisturbanceSelective Cellular Swelling. Brain edema represents a net increase in total brain water content. Multiple studies point at cortical astrocytes as the cellular element initially swollen in ALF. An anatomic breakdown of the blood-brain barrier is not a feature of ALF, as noted in experimental models[52,53] and after examination of cerebral capillaries in human brain.[54] Neuroanatomic studies are difficult to perform in autopsy material,[54] so animal models have been very useful for supporting a primary event in astrocytes.[52,53] Furthermore, isolated astrocytes can be induced to swell when exposed to some of the circulating toxins of liver failure.[55] The demonstration of astrocyte swelling in animals with portacaval anastomosis alone[56] supports a spectrum of changes, in which glial swelling can occur without brain edema. The term low-grade brain edema has been coined for this earlier disturbance of water homeostasis.[14] Changes in Organic Osmolytes. Direct measurements in experimental animals[57] and nuclear magnetic resonance (NMR) spectroscopic findings in humans[58] have repeatedly shown a marked increase in brain glutamine, the product of ammonia detoxification in astrocytes. Inhibition of glutamine synthetase prevents ammonia-induced swelling of isolated astrocytes[59] as well as brain edema in vivo.[60,61] The increase in brain glutamine can be fourfold to sixfold, although the limited capacity of astrocytic glutamine synthetase results in a steady high level of glutamine throughout the course of the neurological disturbance. Cells exhibit short- and long-term adaptive mechanisms to adjust for changes in osmolarity.[62] A high extracellular brain potassium has been shown to occur acutely in an experimental model of acute hyperammonemia,[63] consistent with the effects of regulatory volume decrease in isolated cells.[62] Potassium levels are increased in the jugular vein of patients with ALF,[64] suggesting an increased exit from brain tissue. In the case of chronic adaptation, reduction of the levels of myoinositol, a key intracellular organic osmolyte, is accomplished slowly over several days.[62] Such osmotic adaptation may be a factor that explains the lower frequency of brain edema in subacute or subfulminant hepatic failure.[65] Consistent with these temporal changes in osmotic adaptation is the finding of an elevated glutamine and a low brain myoinositol in patients with cirrhosis, as seen with brain NMR spectroscopy.[58] An Increase in Cerebral Blood FlowIn 1986, Ede and Williams[66] observed an increase of CBF in a subset of patients with ALF and deep HE. They proposed this increased CBF reflected the systemic vasodilatation seen in ALF. Subsequent clinical studies showed a more complex picture. In a series of 30 patients with ALF, Wendon et al[67] noted a wide range of values of CBF, with most having reduced CBF. In an American series, 24% of patients had an elevated CBF, which was associated with brain edema and a higher mortality.[22] Cerebral Anoxia in ALF? The cerebral metabolic rate for oxygen (CMRO2) can be estimated in humans by the product of CBF and the arteriovenous oxygen difference. A small cerebral arteriovenous oxygen difference (arterial-jugular vein con tent) was seen in many of the patients with low or normal CBF,[67] which is suggestive of tissue anoxia. In ALF, values of CMRO2 are low, in some cases less than those thought necessary to maintain cerebral viability.[68] However, these patients can achieve a full neurological recovery after transplantation.[22,67] Alternatively, the finding of a low CMRO2 in patients with normal CBF may be indicative of relative hyperemia, with a dissociation of CBF and the brain's metabolic needs.[69] In order to study the response of CMRO2 to alterations in CBF, Larsen and colleagues[70] measured blood flow and oxygen extraction after infusion of noradrenaline. Their findings indicate a preservation of cerebral oxidative metabolism, arguing against the concept of tissue anoxia. Failure of Cerebrovascular Autoregulation. Under normal conditions, cerebral autoregulation maintains a stable CBF in the face of fluctuations in systemic pressure. The limits of autoregulation, between 60 and 160 mmHg, can be shifted in chronic disease states such as arterial hypertension. In resistance vessels, a myogenic component of autoregulation is normally based on the rapid response of vascular smooth muscle to changes in transmural pressure. In ALF, Larsen assessed the cerebrovascular autoregulation after a noradrenaline challenge. When mean arterial pressure rose by 30 mmHg, transcranial Doppler measurements in the middle cerebral artery showed an increase of velocity of 41%.[71] This loss of autoregulation was restored within 1 day after liver transplantation or within 4 days in subjects with spontaneous recovery. Of note, the therapeutic use of hypothermia also restored cerebrovascular autoregulation in a series of 14 patients with ALF.[72] Response to Changes in PCO2. Under normal conditions, the cerebral circulation responds exquisitely to changes in hypercapnia and hypocapnia with concomitant vasodilatation and vasoconstriction, respectively. CBF changes linearly from 2 to 4% for every millimeter of mercury change in pCO2, also termed the CO2 reactivity coefficient. Hypoxia must be profound, with a pO2 less than 60 mmHg triggering cerebral vasodilatation.[73] In patients with ALF, evidence supports the existence of a dilated cerebral vasculature. Hyperventilation leading to hypocapnia results in an appropriate reduction in CBF.[74] Furthermore, it can restore cerebrovas cular autoregulation.[75] Hypercapnia, however, does not result in further increases in CBF,[74] an indicator of a markedly reduced CO2-reactivity coefficient. In an already dilated cerebral vasculature, further vasodilatory stimuli are unlikely to result in additional effects. Mechanisms. In our experimental model, a predictable and selective rise in CBF occurs prior to the development of brain edema and intracranial hypertension in the setting of stable systemic hemodynamics.[61,76] Two important observations have been made in this model that shed light on the pathogenesis of this complex phenomenon.
The link between the synthesis of glutamine and the subsequent onset of cerebral hyperemia is a critical question in this model and has not been elucidated. The finding of a high nitric oxide (NO) efflux from the sagittal sinus[76] raised a possible role for NO generated from an increased activity of neuronal NO synthase.[80] However, selective and nonselective inhibitors of this enzyme failed to prevent the rise in CBF.[81] The high CBF values seen in human ALF in deep coma highlight the importance of the search for a brain-derived signal that results in cerebral hyperemia. An increase in blood flow is associated with an increase in ammonia uptake,[82] a factor that has recently been shown to increase the likelihood of cerebral herniation in patients with ALF.[83] Oxidative and Nitrosative Stress: Pathogenic Mechanisms in Brain Edema and HE? Several recent observations support the presence of oxidative and nitrosative stress in the brain of models of HE. The formation of free radicals can be indirectly surmised from a series of clinical and experimental observations. In humans, lipofuscin pigment, reflecting the peroxidation of lipids, can be detected in Alzheimer type II astrocytes.[84] We have shown an increase in gene expression of brain heme oxygenase-1 and the reduction of Cu/Zn superoxide dismutase in rats after portacaval anastomosis, findings that support the presence of oxidative stress.[85] Activities of neuronal NO synthase are increased in this model,[86] and we have reported an increase in brain NO efflux, another free radical, in rats after portacaval anastomosis receiving an ammonia infusion.[76] The strongest evidence for this concept arises from cellular studies. The formation of free radicals can be detected in astrocytes exposed to ammonia.[87] Astrocytes exposed to ammonia also develop the mitochondrial per meability transition (MPT).[88] This effect, in which the opening of a large nonselective pore in mitochondria results in morphological and functional abnormalities, leads to defective oxidative phosphorylation and to the generation of even more free radicals. Cultured neurons did not develop the MPT when exposed to ammonia.[88] In support of a pathogenic role of glutamine, inhibition of glutamine synthetase prevented the development of the MPT in isolated astrocytes exposed to ammonia.[88] Free radicals can nitrosylate proteins, and nitrotyrosine, a stable product of this reaction, can be demonstrated in isolated astrocytes exposed to ammonia.[89] This finding can also be seen in vivo.[89] We have recently completed preliminary studies in rats after portacaval anastomosis receiving an ammonia infusion. Clear evidence of nitrotyrosine accumulation in astrocytes was noted (Fig. 3). The functional implications of these changes and their relation to the pathogenesis of HE is an evolving concept but one likely to be important in the manifestations of HE in both ALF and cirrhosis.
Clinical Pathophysiology
In the previous section, we analyzed paradigms that have
evolved from experimental models. Clinical observations
support the role of such pathophysiological mechanisms.
Plasma OsmolarityEarlier series of patients with ALF had noted hyponatremia in patients with encephalopathy.[90] In the experimental animal, hyponatremia (mean serum sodium of 117) aggravates ammonia-induced brain edema.[91] The osmotic disturbance in the brain of patients with ALF is anisosmotic, reflecting the generation of osmoles within the brain,[62] and will be potentiated by a decrease in plasma osmolarity. Patients with cirrhosis who developed intracranial hypertension after placement of a trans jugular intrahepatic portal-systemic stent-shunt (TIPS) were all hyponatremic.[11] Further support of this concept can be seen with the deterioration of mental state associated with rapid fluid shifts in hemodialysis.[92] TemperatureFever aggravates the clinical picture in ALF.[93] Fever (> 38°C) is a component of SIRS, as described earlier, and in experimental animals, an increase in cerebral blood flow and metabolic rate accompanies the hyperthermic state.[94] Preliminary evidence supports an association between temperatures above 38°C and the development of intracranial hypertension in patients with ALF.[95] Whereas fever can accompany systemic infection, septic encephalopathy is associated with an intact cerebrovascular CO2 reactivity and pressure autoregulation,[36] a conspicuous difference with the changes seen in human ALF. Arterial PressureAs a consequence of the loss of cerebrovascular autoregulation, patients with ALF are susceptible to the effects of both a reduction and an increase in arterial pressure. Cerebral ischemia can ensue as a result of the former. Cerebral hyperemia, the result of the latter, can aggravate the development of brain edema. Maintenance of an adequate level of blood pressure is critical to prevent either possibility. GlucoseA recently proposed candidate for a synergistic role in the development of intracranial hypertension in ALF is hyperglycemia. In a preliminary report, values > 12 mmol/L (> 200 µg/dL) were associated with higher values of ICP.[96] In other neurological conditions, hyperglycemia is known to aggravate the effects of brain trauma[97] and ischemia[98] in relation to the increased generation of brain lactate as a result of anaerobic glycolysis. Lactate levels are increased in the brain of experimental models[99] and human[100] ALF, although the mechanism responsible for the rise may be different.[100] The Rise in Intracranial PressureWater in the brain exists in three forms: intracellular water, blood, and cerebrospinal fluid. The latter is decreased in ALF, as seen in imaging of the brain in which shrinking of ventricular size is a common finding.[101] Swelling of the gray matter, where astrocytes constitute 30% of the cellular elements, has been recently demonstrated using NMR techniques.[102] Although cerebral blood volume is difficult to measure, the presence of cerebral vasodilatation and hyperemia suggests an increase in this compartment. With the enlargement of the cellular compartment and in the setting of a limited compliance imposed by the rigid skull, small increases in blood volume will cause an inordinate rise in ICP. The article by Jalan et al[103] in this issue provides further insight into the monitoring and management of intracranial hypertension
|
|||||||||||||
