Experimental models of hepatic encephalopathy: ISHEN guidelines


  • These guidelines were prepared by the Commission on Experimental Models of Hepatic Encephalopathy appointed by the International Society for Hepatic Encephalopathy and Nitrogen Metabolism (ISHEN). The content was discussed and approved in the 13th ISHEN Symposium, Padova, Italy, 28 April to 1 May 2008.

Dr Roger F. Butterworth, Neuroscience Research Unit, Hôpital St-Luc (CHUM), University of Montreal, 1058, St-Denis, Montreal, QC, Canada H2X 3J4
Tel: +1 514 890 8000 ext. 35740
Fax: +1 514 412 7253
e-mail: roger.butterworth@umontreal.ca


Objectives of the International Society for Hepatic Encephalopathy and Nitrogen Metabolism Commission were to identify well-characterized animal models of hepatic encephalopathy (HE) and to highlight areas of animal modelling of the disorder that are in need of development. Features essential to HE modelling were identified. The best-characterized animal models of HE in acute liver failure, the so-called Type A HE, were found to be the hepatic devascularized rat and the rat with thioacetamide-induced toxic liver injury. In case of chronic liver failure, surgical models in the rat involving end-to-side portacaval anastomosis or bile duct ligation were considered to best model minimal/mild (Type B) HE. Unfortunately, at this time, there are no satisfactory animal models of Type C HE resulting from end-stage alcoholic liver disease or viral hepatitis, the most common aetiologies encountered in patients. The commission highlighted the urgent need for such models and of improved models of HE in chronic liver failure in general as well as a need for models of post-transplant neuropsychiatric disorders. Studies of HE pathophysiology at the cellular and molecular level continue to benefit from in vitro and or ex vivo models involving brain slices or exposure of cultured cells (principally cultured astrocytes) to toxins such as ammonia, manganese and pro-inflammatory cytokines. More attention could be paid in the future to in vitro models involving the neurovascular unit, microglia and neuronal co-cultures in relation to HE pathogenesis.

Members of the commission were invited by the International Society for Hepatic Encephalopathy and Nitrogen Metabolism (ISHEN) President (Dr Piero Amodio, Padua, Italy) to develop recommendations for animal models of hepatic encephalopathy (HE) with a view to standardization of procedures and a move towards a consensus on the best-characterized models available at the time of writing this document and to identify areas of animal modelling that are in need of development. An early decision was taken to include only models of Type A, B and C HE as defined by the Vienna consensus document on classification of HEs (Table 1). Other liver disorders such as Wilson's disease, urea cycle enzymopathies and Reye syndrome that frequently result in cerebral dysfunction were not included.

Table 1.   Classification of hepatic encephalopathy syndromes
Type AHepatic encephalopathy (HE) associated with acute liver failure (ALF) resulting from severe inflammatory and/or necrotic liver disease of rapid onset.
Type BHE resulting from portacaval shunting in the absence of parenchymal liver disease.
Type CHE that accompanies chronic liver failure (cirrhosis). Aetiology of cirrhosis resulting from alcohol, viral infection, biliary obstruction, drugs or toxins. Portal hypertension arises as a result of cirrhosis; the high portal pressure stimulates the opening of embryonic venous channels. Portal-systemic shunts then permit toxins of intestinal origin (examples include ammonia, manganese, cytokines) to bypass the liver into the systemic circulation.

A first draft of the consensus document was prepared based on an update of review articles on animal models of HE by Blei et al. (1) and Chamuleau (2) and the document was circulated to commission members by email for comments and modifications. Dr Norenberg was requested to provide a concise assessment of the importance of cell culture models. Dr Albrecht was requested to cover models based on brain slice preparations. An iterative procedure followed whereby commission members were invited to make further corrections. A revised document was presented by Dr Butterworth at the ISHEN meeting in Padua in April 2008 where all participants were invited to submit further corrections/comments leading to the present document.

Hepatic encephalopathy is a neuropsychiatric disorder that results from impaired liver function. In humans, HE occurs in one of the three forms that have recently been reclassified (Table 1).

Studies of the pathogenesis of a wide range of human neuropsychiatric disorders including neurodegenerative diseases (e.g. Alzheimer disease, Parkinson disease), stroke, epilepsy and psychiatric disease (e.g. psychosis, depression) have been significantly enhanced by the use of animal models. Pathophysiological mechanisms involved in metabolic encephalopathies such as vitamin deficiencies and HE have also benefited from the use of animal models. However, HE has proven difficult to model since liver diseases in humans have many varied aetiologies (alcoholic, viral, toxic, autoimmune, ischaemic, genetic), involve varying degrees of portal-systemic shunting, damage to other end-organs and in the majority of cases are associated with precipitating factors such as a gastrointestinal bleed, infection, protein loading, hypoglycaemia, hyponatraemia, use of sedatives, etc. It is not possible to reproduce all (or even many) of these aspects of HE. It is therefore essential to choose an animal model that is appropriate to the research objectives and the measurements to be undertaken. There are currently no satisfactory animal models of HE resulting from alcoholic cirrhosis, viral hepatitis or acetaminophen hepatotoxicity, the three major aetiologies encountered in human HE.

Choice of animal species

A wide variety of animal species have been used for the study of HE including large animals (dogs, goats, pigs, rabbits) and rodents (rats, mice). Large animals have the advantages of facilitating repeated sampling of body fluids, biopsies, the use of techniques such as visual/auditory evoked potentials and electroencephalography (EEG). More material is available from large animals and assessment of neuropsychiatric status may be facilitated. Specific research questions (e.g. evaluation of liver support systems for HE) benefit from the use of large animals. In spite of these multiple advantages, relatively few investigators now use large animals. Reasons for this include the costs of animals, animal housing, materials and salaries for personnel. Heightened ethical concerns may also be an issue.

A review of the recent literature reveals that the laboratory rat and, increasingly, the mouse are the most common species currently used for the studies in HE. Reasons for this include the following:

  • (i)Availability of databases (molecular studies), atlases (anatomical studies) and a wide literature on neurobehavioral, pathological, biochemical methods and findings.
  • (ii)The mouse and rat genomes are now fully cloned and characterized. There is an urgent need for more mouse models of HE for molecular genetic studies.
  • (iii)Antibodies and molecular probes are often commercially available.
  • (iv)Relatively low costs of animals and animal housing.

Animal models of hepatic encephalopathy in acute liver failure (Type A hepatic encephalopathy)

Features proposed as essential for an animal model of HE in acute liver failure (ALF) include the following:

  • (i)Reproducible clinical picture to facilitate staging of encephalopathy.
  • (ii)Progression of symptoms to include brain oedema and its complications (intracranial hypertension, brain herniation).
  • (iii)Potential for reversibility.
  • (iv)Hyperammonaemia/increased brain ammonia, glutamine.
  • (v)Well-characterized hepatic and brain pathology.
  • (vi)Minimal hazard to personnel from toxins and infectious agents.

Few models satisfy all these criteria. All animal models of ALF lead to hypothermia, hypoglycaemia and other systemic complications. It is therefore of paramount importance to control temperature and glycaemia routinely and effectively and to monitor other organ systems and provide supportive care.

In view of the absence of animal models of ALF related to viral hepatitis, animal models of ALF have been developed principally along two lines, namely exclusion of the liver (anhepatic models) from the circulation or administration of a hepatotoxin (Table 2).

Table 2.   Animal models of Type A hepatic encephalopathy
 Species used
Anhepatic models
 Hepatic devascularizationRat, rabbit, pig
 HepatectomyRat, pig
 Portacaval anastomosis+ammoniaRat
Hepatotoxin models
 GalactosamineRat, rabbit, guinea pig
 AcetaminophenRat, dog, pig

Anhepatic models of Type A hepatic encephalopathy

In hepatic devascularization, the liver is left in situ and its afferent circulation diverted (portacaval anastomosis) and interrupted (hepatic artery ligation). In total hepatectomy, the liver is removed generally as a two-stage operation and the splanchnic circulation is rerouted. The course of HE after either hepatic devascularization or hepatectomy is relatively rapid. Neither procedure manifests a potential for recovery.

Swelling of astrocyte foot processes has been described in both devascularized (3) and hepatectomized (4) rats. A progressive increase in intracranial pressure has been noted in hepatic devascularized rats where EEG changes are similar to those described in ALF in humans.

A model of ammonia-precipitated brain oedema in the context of liver failure uses end-to-side portacaval anastomosis followed 24 h later by ammonia infusions (5).

Hepatic devascularized rats manifest a reproducible progression of encephalopathy and brain oedema and have been used extensively for studies on the pathogenesis and treatment of ALF. Animals are hyperammonaemic with brain ammonia in the 1–5 mM range (6). The model has been applied to studies of brain metabolism (7), neurotransmission abnormalities, gene expression and brain inflammation in ALF. Brain oedema and HE in hepatic devascularized animals respond to hypothermia, ammonia-lowering agents and anti-inflammatory drugs.

Toxin models of Type A hepatic encephalopathy

A range of hepatotoxic substances has been used to create ALF in experimental animals including galactosamine, acetaminophen, thioacetamide and azoxymethane. Other toxins known to lead to ALF include phosphorus, nitrosamines and CCl4 but HE resulting from these latter toxins has not been well characterized. The CCl4 model is useful in demonstrating astrocytic response at the level of RNA synthesis (8). Galactosamine is directly hepatotoxic whereas acetaminophen and thioacetamide must be converted to toxic metabolites in the liver via the microsomol P450 system. The pathological nature of the hepatitis produced by each of these toxins varies; bridging necrosis and a lymphocytic inflammatory infiltrate have been described in the thioacetamide rat (9) but not in the galactosamine rabbit (10). Massive centrolobular necrosis is the pathological hallmark of acetaminophen-induced liver injury.

The galactosamine-treated rabbit

Traber et al. (11) reported marked swelling of the pericapillary astrocyte foot processes and swelling of processes surrounding axons and dendrites. Cerebral oedema was restricted to grey matter. Neurons appeared normal; the blood–brain barrier remained grossly intact but permeability to α-aminoisobutyric acid (normally impermeable) was increased in basal ganglia (12). This model has been used for evoked potential studies.

The galactosamine-treated rat

This model is not as well characterized from the hepatic and brain standpoints as the galactosamine-treated rabbit. Severity of HE and mortality are variable (13). Dixit and Chang (14) described cerebral cortical and cerebellar oedema together with swelling of pericapillary astrocytes. Brain tissue necrosis and breakdown of the blood–brain barrier have also been reported (15), features that are not routinely observed in human ALF.

The thioacetamide-treated rat

Thioacetamide treatment results in hepatocellular necrosis, bridging necrosis and lymphocytic infiltrate without cholestasis (9). Rats develop renal failure, hypoglycaemia, hypotension and weight loss if supportive fluids, glucose and external heating are withheld (16). The blood–brain barrier is intact in this model. The thioacetamide model has been used to delineate changes in neuromodulatory functions of astrocytes in relation to progression of HE (17).

The azoxymethane-treated mouse

This relatively new model has been characterized from the liver and brain standpoints (18), and used for studies of HE mechanisms involving gene manipulations. Mice rapidly develop encephalopathy and brain oedema.

Acetaminophen-treated animals

Acetaminophen administration to dogs has been used as a model for evaluation of artificial liver support systems (19). The rat model of ALF because of acetaminophen results in variable HE and brain oedema. Neuropathology has not been performed in this model.

Animal models of hepatic encephalopathy in chronic liver impairment

Features considered essential for an animal model of HE in chronic liver impairment (Type B and C HE) include the following:

  • (i)Chronic liver disease with portal-systemic shunting.
  • (ii)Range of symptoms of encephalopathy from minimal hepatic encephalopathy (MHE) to coma.
  • (iii)Alzheimer Type II astrocytosis at coma stages of encephalopathy.
  • (iv)Hyperammonaemia, increased brain ammonia/glutamine.
  • (v)Precipitating factor.
  • (vi)Clinical response to established treatments.

Animal models of hepatic encephalopathy based on portal-systemic shunting (Type B hepatic encephalopathy)

Portacaval anastomosis has been performed in rats, rabbits, dogs and pigs (Table 3). Plasma ammonia correlates with EEG changes and neurological status in dogs with portacaval anastomoses (20). Attempts to perform portacaval anastomosis in rabbits may lead to high mortality (21).

Table 3.   Animal models of Type B (and C) hepatic encephalopathy
 Species used
Portacaval anastomosisRat, dog, rabbit, pig
Congenital portacaval shuntsDogs, cats
Graded portal vein stenosisRat
Portacaval anastomosis+ammoniaRat
Bile duct ligation*Rat

Dogs and cats with congenital portacaval shunts offer naturally occurring models (Table 3). Dogs develope psychomotor dysfunction, motor signs and hyperammonaemia, reduced hepatic function (22) and are susceptible to high protein diets.

Graded portal-vein stenosis affords a model of MHE in the rat. This procedure is easier to perform than the end-to-side portacaval anastomosis and leads to loss of activity, altered day–night rhythms, hyperammonaemia and increased brain ammonia/glutamine.

Side-to-side portacaval anastomosis in the rat, although challenging from a surgical standpoint, offers a potential for reversibility (23).

Hepatic encephalopathy resulting from portacaval anastomosis in the rat

End-to-side portacaval anastomosis in the rat affords a popular model of Type B HE. Liver dysfunction is limited to liver atrophy and loss of perivenous hepatocytes. This model is widely used to study the effects of portacaval shunting and reduced hepatic capacity on brain function. Stenosis may occur leading to variable results; liver/body weight ratio and testicular weight may be used as a surrogate index of degree of stenosis. Pair feeding of control animals may be necessary with studies of long-term shunted animals to control for anorexia and decreased food intake (24). Shunted animals manifest hyperammonaemia and increased brain ammonia/glutamine, alterations of day–night and circadian rhythms, hypokinesia, reduced memory and learning ability, altered reflexes and testicular atrophy. Altered aromatic/branched chain amino acid ratios, decreased brain glucose utilization, oxidative/nitrosative stress and altered multiple neurotransmitter function similar to MHE in human cirrhotic patients have been reported. Portacaval shunted rats are sensitive to ammonia administration, which leads to severe encephalopathy (coma) (25, 26).

Animal models of hepatic encephalopathy resulting from decompensated liver cirrhosis (Type C hepatic encephalopathy)

An optimal animal model of Type C HE does not exist at the present time. Alcohol administration does not produce micronodular cirrhosis in the small laboratory animal. A common procedure used to produce cirrhosis in experimental animals makes use of the hepatotoxin CCl4. Monitoring makes use of a daily body weight method. Problems include inconsistent lesions animal to animal. The presence of ascites may limit neurobehavioral assessment.

Obstruction of the common bile duct induces a reproducible model of biliary cirrhosis in rats (Table 3). In contrast to portacaval anastomosis, bile duct-ligated (BDL) animals have liver failure, developing jaundice, portal hypertension (27), portal-systemic shunting (28), bacterial translocation and immune system dysfunction (29). BDL rats are hyperammonaemic (30) but show only low-grade encephalopathy (decreased locomotor activities) (31). However, BDL rats are infected, they do not eat and lose weight and pair feeding of control groups may be necessary in order to avoid confounds owing to diminished food intake (32). Feeding ammonium salts to BDL rats provides a model of acute-on-chronic HE, which reproduces the human neuropathology (Alzheimer Type II astrocytosis) and alterations of brain osmolytes that are characteristic of human Type C HE as well as low-grade brain oedema, inflammation and deficits of motor coordination (32).

Animal models of pure acute/chronic hyperammonia

These models, generally limited to rats and mice, are designed to study the effects of hyperammonia per se (in the absence of liver dysfunction) on brain function. Methods include the feeding of high-ammonia diets or urease. These models have been successfully used to demonstrate alterations of multiple neurotransmitter systems in brain (e.g. glutamate receptor-mediated signal transduction) (33, 34). Hyperammonaemic animals manifest disturbances in learning and memory. Experiments are inexpensive and simple to perform.

Use of cultured neural cells for the study of hepatic encephalopathy mechanisms

These studies are valuable in the study of the effects of exposure to particular neural cells to toxic substances that are known to be increased in brain in acute or chronic liver failure. Toxins studied so far include ammonia, manganese, octanoic acid and cytokines. The majority of studies to date have made use of primary cultures of rat cortical astrocytes (35). Some studies use C-6 glioma cells. Few studies so far have made use of cultures of other neural cells (neurons, cerebrovascular endothelial cells, microglia) or cocultures. Studies using these latter systems are greatly encouraged.

Cultured astrocytes exposed to ammonia manifest many of the pathological, metabolic and neurochemical changes observed in human HE (35). Cultured neurons exposed to ammonia reproduce alterations in signal transduction observed in human HE (36). Advantages of the cultured cell approach are its low cost, excellent control of the cellular environment and reproducibility. The use of cell cultures offers a unique opportunity for unravelling complex pathogenetic pathways. Disadvantages include potential differences in cellular/molecular properties between cultured cells and their in situ counterparts, particularly regarding expression of certain proteins. Extrapolation of findings to the in vivo situation should be done with caution.

There is a need for cell culture models of the neurovascular unit for the study of its role in HE mechanisms.

Use of brain slices for the study of hepatic encephalopathy mechanisms

Brain slices have classically been used to study mechanisms in a wide range of experimental neurological disorders. Brain slices have a relatively well-preserved cytoarchitecture permitting the study of cell–cell interactions. In vitro experiments make use of brain slices from normal animals exposed to toxins implicated in HE (e.g. ammonia). Ex vivo experiments make use of brain slices from animals with HE resulting from experimental acute or chronic liver failure. Disadvantages of brain slices are their relatively short-lived viability (l–2 h). Several investigators have used brain slices to study brain oedema mechanisms (37), astrocyte neuropathology (38), long-term potentiation (39), nerve cell damage (40) and cell signalling mechanisms (41) in relation to toxin exposure in vitro and ex vivo in slices from animal models of HE.

Recommendations of the commission

It is important to bear in mind that the recommendations of the commission are based on available information reported in the biomedical literature at the time of the Padua ISHEN meeting in 2008 and that, by their very nature, these recommendations are subject to revision as further models of HE become characterized and existing models undergo modifications. As stated in this review, there are no satisfactory animal models of HE resulting from alcoholic cirrhosis, viral hepatitis or acetaminophen hepatotoxicity, the three most common aetiologies encountered in patients. With the above provisos, the commission makes the following recommendations:

  • 1Well-characterized animal models of HE in ALF (Type A HE) that meet the majority of criteria are:
  • (a)The hepatic devascularized rat.
  • (b)The rat with thioacetamide-induced toxic liver injury.
  • 2Well-characterized animal models of HE in chronic liver impairment that meet the majority of criteria are:
  • (a)The rat with end-to-side portacaval anastomosis (Type B HE) (MHE).
  • (b)The BDL rat (some aspects of Type C HE).
  • 3There is a need for improved animal models of Type C HE in rodents.
  • 4There is a need for improved mouse models of all types of HE in the mouse to facilitate studies of the molecular genetics of HE.
  • 5There is a need for models of post-orthotopic liver transplantation neurological/psychiatric disorders.
  • 6Studies of molecular and cellular mechanisms related to HE may be supplemented by studies in:
  • (a)Hyperammonaemic animals (with intact liver function) in order to assess the role of ammonia in vivo (per se).
  • (b)Cultured neural cells (particularly glial cells) exposed to pathophysiologically relevant concentrations of toxins (e.g. ammonia, manganese, cytokines) known to accumulate in brain in liver failure.
  • (c)Brain slice preparations from normal animals exposed in vitro to toxins or ex vivo slice preparations from HE animal models.
  • 7There is a need for studies in cultured cerebrovascular endothelial cells, microglia, neurons and mixed cultures in order to study cell–cell interactions and the role of the neurovascular unit in HE.
  • 8It is proposed that efforts be made to standardize procedures in order to limit variability of the preparations. This could be achieved, for example, by exchange of research trainees between laboratories.