Whereas cerebral edema is a long-known complication of fulminant liver failure,1 patients with cirrhosis and hepatic encephalopathy (HE) usually show no clinical signs of overt cerebral edema and increased intracranial pressure (ICP). However, a first indication of the presence of low grade cerebral edema in patients with cirrhosis was derived from 1H-MRS studies of the human brain in vivo. These noted a reduction in the myo-inositol signal in the brain (reflecting an osmosensitive myo-inositol pool) from patients with HE, together with an increased glutamine/glutamate signal. The glial localization of these osmolytes suggested a disturbance of astrocyte volume homeostasis.2 Based on this observation and available data from animal and in vitro studies, it was hypothesized that HE in cirrhosis reflects the clinical manifestation of a low grade cerebral edema, which may worsen under the influence of precipitating factors and thereby trigger cell hydration-dependent alterations of astrocyte function.2, 3 Such alterations may affect glial-neuronal communication, whose disturbance is considered to be relevant to the pathogenesis of HE.4

Meanwhile, the existence of a low grade cerebral edema in patients with cirrhosis and HE in vivo was also demonstrated in studies on magnetization transfer ratios5, 6 and by quantitative cerebral water mapping based on a new magnetic resonance (MR) technique for fast quantitative mapping of T1 and water content.7 Also, patients with noncirrhotic portal vein thrombosis and minimal HE exhibit MR findings consistent with increased brain water8 and as noted in this issue of HEPATOLOGY, mild brain edema develops in a recently characterized animal model of hepatic encephalopathy in cirrhosis.9 MR findings consistent with a low grade cerebral edema are already found in patients with minimal HE and become more pronounced when HE severity increases,2, 8, 10 for example, after implantation of a transjugular intrahepatic portal-systemic stent2 or experimentally induced hyperammonemia.11 Conversely, these magnetic resonance spectroscopy (MRS) abnormalities can resolve after liver transplantation5 or successful medical treatment of HE.6 Thus, fluctuations of HE severity in patients with cirrhosis roughly parallel the extent of low grade cerebral edema.

In vitro studies on the mechanisms underlying this low grade cerebral edema have largely focused on the astrocyte. Astrocytes are the only compartment in the brain capable of ammonia detoxification via glutamine synthesis12 and when exposed to ammonia, can undergo so-called Alzheimer type II changes as a morphological counterpart of astrocyte swelling. Intra-astrocytic accumulation of osmotically active glutamine in liver cirrhosis and other hyperammonemic states contributes to the development of a low grade brain edema, as evidenced by consistently elevated 1H-MR spectroscopic glutamine signals in the brain of patients with cirrhosis. Further, methionine sulfoximine, an inhibitor of glutamine synthetase, prevents astrocyte swelling following an ammonia load in experimental animals.13 However, not only ammonia, but also benzodiazepines, hyponatremia and inflammatory cytokines can induce astrocyte swelling in vitro. Thus, different neurotoxins, repeatedly implicated in the pathogenesis of HE, may synergistically promote astrocyte swelling as one common pathogenetic endpoint. This may provide an explanation for another key feature of HE in patients with cirrhosis,2, 3 namely that the syndrome is precipitated by rather heterogeneous factors, such as bleeding, infections, sedatives or electrolyte disturbances.

However, it would be an oversimplification to postulate that ammonia-induced astrocyte swelling arises solely as a consequence of an accumulation of osmotically active glutamine. In cultured astrocytes, ammonia triggers the generation of oxidative stress,14, 15 also induced in response to hypoosmotic swelling, benzodiazepines16 or inflammatory cytokines such as tumor necrosis factor-α or interferons (for review see Schliess et al.17). This oxidative stress response is mediated by activation of N-methyl-D-aspartate (NMDA) glutamate receptors,15–17 but the mechanisms underlying this activation are unclear. Possible explanations are swelling- and ammonia-induced glutamate release from astrocytes17, 18 and/or deinhibition of NMDA receptors due to a depolarization-induced Mg2+ removal.17 There is a close relationship between astrocyte swelling, NMDA receptor activation and oxidative stress.17 On the one hand, astrocyte swelling induces oxidative stress through a NMDA receptor– and Ca2+-dependent mechanism,17 and on the other, NMDA receptor activation and oxidative stress trigger astrocyte swelling.19 These findings point to an auto-amplificatory signalling loop between astrocyte swelling and oxidative stress.17 NADPH oxidase isoforms20 and the mitochondrial permeability transition pore21, 22 are the most likely sources for the reactive oxygen species, which are formed in response to astrocyte swelling and ammonia.

What are the functional consequences of the low grade cerebral edema, astrocyte swelling and oxidative stress? In almost every cell type, changes in the cellular water content; i.e., cell hydration or the degree of cell swelling have been identified as an independent signal which regulates function and gene expression through osmosensing and osmosignaling pathways.23 Thus, small increases in astrocyte water content, present in hepatic encephalopathy, may have important functional consequences despite the absence of clinically overt increases of intracranial pressure. Many, but not all effects of ammonia on cultured astrocytes, can be mimicked by slight swelling of the cells in hypoosmotic media. Like ammonia exposure, hypoosmotic swelling of astrocytes activates not only NMDA receptors, but also extracellular regulated protein kinases, upregulates the peripheral type benzodiazepine receptor (PBR),24 affects multiple ion and amino acid transport systems (for review see Häussinger et al.3) and increases the pH in endocytotic vesicles,25i.e., a compartment which is involved in receptor/ligand sorting. Such comprehensive alterations in astrocyte cell biology may explain why multiple neurotransmitter systems are altered in HE. Also, the increased deposition of glycogen in astrocytes in animal models of chronic hepatic encephalopathy can be explained by cell swelling and several ammonia-induced metabolic effects, such as depression of glucose consumption in rat brains were shown to depend upon ammonia-related glutamine synthesis, but not upon the presence of ammonia per se.26 This again points to a critical role of ammonia-induced astrocyte swelling, which requires the synthesis of glutamine. However, it should be emphasized that not all ammonia effects on astrocytes can be explained by cell swelling. One example is the induction of hemoxygenase-1 in cultured astrocytes by ammonia, but not by hypoosmotic exposure.

Ammonia, inflammatory cytokines, benzodiazepines and hypoosmotic astrocyte swelling also induce nitrosative stress probably through a NMDA receptor– and Ca2+/calmodulin-dependent activation of constitutive nitric oxide synthases (NOS).17 One consequence of oxidative/nitrosative stress is a covalent modification of tyrosine residues in astrocytic proteins through nitration.15–17 Peroxynitrite is involved in ammonia-, cytokine- and swelling-induced protein tyrosine nitration (PTN),15, 17 whereas benzodiazepine-induced PTN is peroxynitrite-independent and involves the activation of the peripheral, but not of the central benzodiazepine receptor (PBR).16 PTN in astrocytes is also found in vivo in ammonia- or lipopolysaccharide- intoxicated or portocaval shunted rats.1, 15, 17 Astrocytes located near the blood brain barrier exhibit especially high levels of PTN, with unknown consequences for blood brain barrier permeability. PTN involves only distinct proteins,15–17 such as glutamine synthetase, the peripheral benzodiazepine receptor (PBR), glyceraldehyde-3 phosphate dehydrogenase and the extracellular signal regulated kinase Erk-1. PTN of glutamine synthetase affects the catalytic center of the enzyme and is associated with its inactivation. The role of PTN of the PBR is unclear; however this protein, upregulated in HE, plays a role in the mitochondrial permeability transition and the synthesis of neurosteroids, such as allopregnanolone and allotetrahydrodeoxy-corticosterone.27 Such neurosteroids have a positive GABAA-receptor modulatory activity and were identified in the brain from patients with hepatic coma.28 This could provide an explanation for the increased GABAergic tone found in patients with HE. Although PTN interferes with protein function and signal transduction, its role in the pathogenesis of hepatic encephalopathy is unclear. However, inhibition of ammonia-induced PTN by NMDA receptor antagonists, inhibitors of glutamine synthetase or NOS is associated with an amelioration of ammonia toxicity in animals.29, 30

It remains to be established whether the functional sequelae of astrocyte swelling, oxidative/nitrosative stress and other actions of neurotoxins, mainly studied in vitro or in experimental animals, also apply to human cirrhosis. If this were true, the emerging pathogenetic model for HE (Fig. 1) would offer novel sites for treatment beyond the current therapies, predominantly directed towards precipitating factors. In this model, ammonia and other neurotoxins synergistically induce a low grade cerebral edema as a result of astrocyte swelling. This occurs without a clinically overt increase in intracranial pressure, but the hydration increase is sufficient to trigger multiple alterations of astrocyte function and gene expression. These include an oxidative/nitrosative stress response with covalent protein modifications and an autoamplificatory loop between cell swelling and oxidative stress. In addition, there are direct, swelling-independent effects of neurotoxins on astrocyte and neuronal function. As a result of an altered astrocyte function, glialneuronal communication and multiple neurotransmitter systems become deranged that impact synaptic plasticity and oscillatory cerebral networks,31 which finally may account for the symptoms of HE. Multiple precipitating factors act synergistically on a common pathogenetic endpoint, i.e., glial swelling with its functional consequences. Patients without cirrhosis may tolerate such precipitating factors without developing HE symptoms, as their osmolyte systems for counteraction of cell swelling are not exhausted. In cirrhosis, however, organic osmolytes are largely depleted in order to compensate for glial glutamine accumulation and there may be little room for action of these volume-regulatory mechanisms against further challenges of cell volume. This labile situation may explain not only the rapid kinetics of HE episodes, but also the occasional appearance of clinically overt cerebral edema in end-stage cirrhosis.32

thumbnail image

Figure 1. Low grade cerebral edema and hepatic encephalopathy. Ammonia induces astrocyte swelling, which is aggravated synergistically by precipitants. Astrocyte swelling involves NMDA receptor activation, elevation of the intracellular Ca2+ concentration and the generation of oxidative and nitrosative (ROI/RNI) stress. Swelling and oxidative stress are connected through an autoamplificatory loop and trigger alterations in astrocyte function. Although astrocyte swelling is considered to be a major pathogenetic event, swelling-independent effects of neurotoxins on astrocytes are also involved. Swelling- and neurotoxin-induced astrocyte dysfunction will affect glialneuronal communication with consequences for synaptic plasticity and oscillatory electrical networks. This hypothetical model does not take into account direct neurotoxin effects on the neurons. ROI/RNI, reactive oxygen/nitrogen intermediates.

Download figure to PowerPoint


  1. Top of page
  2. References