Brain edema and inflammatory activation in bile duct ligated rats with diet-induced hyperammonemia: A model of hepatic encephalopathy in cirrhosis


  • Potential conflict of interest: Nothing to report.

  • Presented in part as a poster at 2005 Meeting EASL, Paris.


Studies of the pathogenesis of hepatic encephalopathy are hampered by the lack of a satisfactory animal model. We examined the neurological features of rats after bile duct ligation fed a hyperammonemic diet (BDL+HD). Six groups were studied: sham, sham pair-fed, hyperammonemic, bile duct ligation (BDL), BDL pair fed, and BDL+HD. The BDL+HD rats were made hyperammonemic via an ammonia-containing diet that began 2 weeks after operation. One week later, the animals were sacrificed. BDL+HD rats displayed an increased level of cerebral ammonia and neuroanatomical characteristics of hepatic encephalopathy (HE), including the presence of type II Alzheimer astrocytes. Both BDL and BDL+HD rats showed activation of the inflammatory system. BDL+HD rats showed an increased amount of brain glutamine, a decreased amount of brain myo-inositol, and a significant increase in the level of brain water. In coordination tests, BDL+HD rats showed severe impairment of motor activity and performance as opposed to BDL rats, whose results seemed only mildly affected. In conclusion, the BDL+HD rats displayed similar neuroanatomical and neurochemical characteristics to human HE in liver cirrhosis. Brain edema and inflammatory activation can be detected under these circumstances. (HEPATOLOGY 2006;43:1257–1266.)

Hepatic encephalopathy (HE) is a severe neuropsychiatric complication in both acute and chronic liver failure. Although the pathophysiology of this disorder is incompletely understood, there is agreement on the important role of neurotoxins in its development, especially ammonia.1 Neuropathologically, HE in chronic liver disease is characterized by astrocytic rather than neuronal changes.2 Histopathology studies of brain sections from patients with cirrhosis who died in hepatic coma show the presence of changes known as Alzheimer type II astrocytosis. These astrocytes display a specific characteristic of swollen cytoplasm containing a large pale nucleus, with prominent nucleolus and patches of heterochromatin associated with the nuclear envelope.2 Recently, a new pathophysiological hypothesis has been developed, emphasizing the role of low-grade brain edema in the pathogenesis of HE in chronic liver disease.3 Following this theoretical model, the presence of a low-grade astrocyte swelling could have important functional consequences despite the absence of clinically overt increases of intracranial pressure.3

We propose a new animal model for the study of HE that occurs in chronic liver failure. Chronic liver disease was modeled in rats via ligation of the common bile duct, followed by the addition of ammonia in their diet to produce hyperammonemia. Bile duct ligation (BDL) is very effective in causing jaundice and impairment of liver function. BDL has also been shown to induce fibrosis, portal hypertension, portal-systemic shunting, and immune system dysfunction.4–9 As a way to mimick HE that occurs in humans, ammonia was added as a precipitant factor to induce hyperammonemia and overt encephalopathy in rats.

Impairment in the glutamate-nitric oxide-cyclic guanosine monophosphate (cGMP) pathway is considered an important aspect of ammonia neurotoxicity.10, 11 In a previous study,12 we found that bile duct ligation plus hyperammonemic diet (BDL+HD) rats showed the same modulation pathway that had been previously described in human HE.13, 14 BDL+HD rats and patients with HE showed a reduced activation of guanylate cyclase by nitric oxide in cerebellum and an increased activation in the cerebral cortex. Other models (hyperammonemia alone or BDL alone) are not capable of reproducing the changes seen in human HE.

Our aim was to characterize the model of BDL plus hyperammonemic diet (BDL+HD) for the study of acute HE that occurs on chronic liver disease. This characterization is done at three levels: biochemical, neuropathological, and behavioral. A second objective was to determine whether this model reproduces the osmolyte pattern and the brain water disturbances (low-grade edema) in HE occurring in patients with liver cirrhosis.


HE, hepatic encephalopathy; BDL, bile duct ligation; HD, hyperammonemic diet; BDL+HD, bile duct ligation plus hyperammonemic diet; PF, pair fed; HPLC, high-performance liquid chromatography; IL-6, interleukin-6; GFAP, glial fibrillary acidic protein; GS, glutamine synthetase; TNF-α, tumor noecrosis factor alpha.

Materials and Methods


Male Wistar rats (200–250 g) were housed under standard conditions and fed control diet (Global Diet 2014, Harlan Teklan, IN) for 2 weeks after surgery. Six groups were used in this study: sham, sham pair-fed (PF), hyperammonemic diet (HD), BDL, BDL PF, and BDL+HD.

Bile Duct Ligation.

Under general anesthesia with diazepam (3 mg/kg) and ketamine (100 mg/kg) administered intraperitoneally, a midline incision was made. The common bile duct was localized, doubly ligated, and cut between these two ligatures. In sham animals, a midline incision was performed, but without BDL.

Induction of Hyperammonemia.

As previously described, HD and BDL+HD groups were given the ammonium-containing diet during the 3 weeks after operation until their sacrifice.15 During the third week, two groups of control PF rats were included: Sham PF were fed an amount of control diet equal to the amount of ammonium-containing diet eaten by HD rats; BDL PF were BDL rats fed an amount of control diet equal to the amount of ammonium-containing eaten by BDL+HD rats.

Animals were killed in the morning (9:00 AM), 3 weeks after the intervention. The animals were killed by decapitation. Cerebellum and cerebral cortex were rapidly removed and stored at −80°C. The experiments were approved by the Animal Care Committee at the Medicine School of University Miguel Hernández. Mortality of BDL animals was 15%. Body weight and temperature was measured each week from the first day after surgery until the day of killing.

Biochemical Characterization

Blood samples were obtained at death from at least eight animals in each group. Biochemical determinations [glucose, urea, creatinine, sodium, alanine aminotransferase, gamma glutamyltransferase, alkaline phosphatase, bilirubin, and total proteins] were made using an autoanalyzer Kodak Ektachem 700 (Orthoclinical diagnostics, Rochester, NY).

Determination of Amino Acids in Serum.

Aminoacids in serum were analyzed using a Waters reverse-phase high-pressure liquid chromatography (HPLC) system with fluorescence detection and precolumn o- phthalaldehyde derivatization (Waters Corp, Milford, MA). The column used was a reverse-phase C18, 5-μm particle size, 250 × 4.6 mm (Waters). The chromatography was performed using a gradient at a constant flow rate of 1 mL/min.

Determination of Ammonia.

Ammonia was measured in plasma and cerebral cortex. Blood (150 μL) was taken from the tail vein in the morning (9:00 AM), the third week after surgery. The cerebral cortex was homogenized and deproteinized in 5 volumes of ice-cold, 6% trichloroacetic acid, and kept on ice for 15 minutes. After centrifugation at 12,000g for 10 minutes at 4°C, the supernatants were collected, neutralized with 2 mol/L KHCO3, and centrifuged at 12,000g for 10 minutes at 4°C. Ammonia was measured on the neutralized supernatants by fluorimetry (Fluoroskan Ascent Labsystems, Helsinki, Finland). Assays were performed in Costar 96-well UV plates (Corning Costar Corporation, Cambridge, MA).

Nitrites/Nitrates Assay.

Nitrites and nitrates were measured in plasma samples by the Griess method16 after nitrate reduction into nitrite. Blood (150 μL) with EDTA was taken from the tail vein in the third week after surgery and centrifuged at 450g for 5 minutes. Before measurement, plasma samples were incubated for 20 minutes at 37°C in 0.5 mol/L phosphate buffer with 10 U/mL nitrate reductase, 0.1 mmol/L flavin adenine dinucleotide and 1 mmol/L reduced nicotinamide adenine dinucleotide phosphate, to convert nitrate to nitrite.

Interleukin-6 and Tumor Necrosis Factor Alpha Measurement.

Blood samples were collected in the third week after surgery in Vacutainer tubes containing K3-EDTA (1 mg/mL), centrifuged at 450g for 5 minutes at 4°C, and frozen at −80°C until assay. Interleukin-6 (IL-6) was determined by the ELISA Biotrak system (Amersham Biosciences, UK). Tumor necrosis factor alpha (TNF-α) was determined by the BD OptEIA test (BD Biosciences, San Diego, CA).

Neuroanatomical Characterization

For anatomical studies, 10 sham, 10 HD, 9 BDL, and 10 BDL+HD rats were killed and their brains removed from the skull and then fixed by immersion in 4% paraformaldehyde. Examiners were blind about the type of animal. The brains were serially sectioned at 50 μm in a sliding microtome (Microm, Heidelberg, Germany) coupled to a freezing unit. Slides of 250 μm were mounted and stained with thionin or cresyl violet. Selected sections from the four groups, including the frontal and parietal cortices, were immunostained in parallel with glial fibrillary acidic protein (GFAP) rat polyclonal (Calbiochem, La Jolla, CA) and glutamine synthetase (GS) monoclonal antobodies (Chemicon, Temecula, CA). Sections were incubated with the primary antibody diluted 1:100 (GFAP) or 1:1,000 (GS) in phosphate-buffered saline (pH 7.3–7.4) containing 1% Triton X-100, 5% goat serum, 1% bovine serum albumin (BSA), and 0.25% gelatin, at 4°C, overnight in continuous agitation. They were incubated with the appropriate biotinylated secondary antibody, followed by the standard Vectastain ABC kit (Vector, Burlingame, CA) and reacted with DAB. Sections were mounted on gelatinized slides, air dried, dehydrated, and coverslipped. A selected number of GFAP-reacted sections were counterstained with cresyl violet. For quantitative analysis, GFAP-immunoreactive astrocytes were plotted in 500-μm-wide rectangles covering the entire thickness of the cortex (from layers I to VI), and their number per millimeter tissue was calculated using the Neurograph software (Microptic, Barcelona, Spain). In total, four rectangles per animal were placed (2 in the frontal and 2 in the parietal cortices).

Brain Edema and Osmolytes

Extraction of Organic Osmolytes.

Analysis was done in eight animals from each group. Brain samples were weighed just before thawing; after that they were homogenized in cold 0.6 mol/L HClO4 (2 mL/sample) and neutralized with 25% KHCO3 w/v (approx. 0.5 mL/sample). The measurement of organic osmolytes was carried out by the HPLC method using an Agilent liquid chromatograph model 1100 (Agilent, Palo Alto, CA). Separation was performed at 80.0°C. on a Sugar Pak-1 column (Waters Corp., Milford, MA) of 6.5 × 300 mm. All reference samples were analyzed in duplicate. The concentration was calculated from the peak areas. The unit was mmol/kg wet tissue. The retention times obtained were: myo-inositol, 11.8 minutes; taurine, 13.5 minutes; urea, 17.4 minutes; glutamine, 20.5 minutes; glycine, 23.5 minutes; and creatine, 29.6 minutes.

Brain Edema.

Water content in the cerebral cortex was determined from measurements of tissue specific gravity using a gravimetric method as previously described by Marmarou et al.17 After decapitation, brain hemispheres of eight animals from each group were cut into coronal slices, and samples approximately 2 mm2 were taken from the cerebral cortex. The samples were placed into a kerosene/bromobenzene gradient column, and the equilibration point was recorded within 2 minutes. Previously, the columns were calibrated with four K2SO4 solutions of different concentrations with known specific gravities. The specific gravity of each sample was calculated from the linear equation derived from the flotation levels of the K2SO4 standards, and conversion to brain water was done according to the formula reported by Marmarou et al.17 Results were expressed as percentage of water content. Five measures were made for each animal, and values were averaged.

Behavior Studies

Motor Activity and Coordination Tests.

The tests were performed in 11 sham, 7 sham PF, 9 BDL, 8 BDL PF, 10 HD, and 12 BDL+HD rats. Two motor coordination tests were performed: rotarod (in the morning) and beam walking (in the afternoon). Tests were performed on the 6th day in ammonium diet. Locomotor activity was assessed on the 7th day after the start of the diet.


In this test the ability of rats to stay on a rotating drum is evaluated. An accelerating rotarod18 was used (Ugo Basile, Comerio, Italy). Two consecutive days before testing, each rat was placed in the rotarod switched off for 3 minutes. The rotarod speed increased from 4 to 40 rpm over 300 seconds. The time at which each animal fell off the rungs was recorded, with a maximum cut-off of 600 seconds.

Beam Walking Test.

In this test, the ability of rats to pass through a narrow beam to reach a dark box is evaluated. To force rats to pass through the beam, a white light illuminates the beginning of the beam. The wooden square beam (2 × 100 cm) was elevated to a height of 1 m above the floor. In this test, the time to cross the beam and the number of forelimb and hindlimb foot faults were recorded. A fault was defined as any foot slip off the top surface of the beam or any limb use on the side of the beam. The day of the test, four trials were performed before recording the results. The goal was to accustom the rats to the beam and to let them to know the existence of the dark box at the end of the beam. The interval between trials was 5 minutes.

Motor Activity.

Motor activity was evaluated by putting the rats individually in transparent plastic cages (24 cm high, 48 cm long, 30 cm wide) with sawdust in the bottom. Locomotor activity was measured by counting crossovers through a line dividing the cages into two compartments. Vertical activity was measured by the number of rearings. Both cross-overs and rearing were counted for 60 minutes just after placing the rats in the experimental cage.

Statistical Methods

Results are expressed as mean ± SD. One-way ANOVA was used for intergroup comparisons, followed by post-hoc Newman-Keuls-test. When two groups of animals were compared, analysis was performed with unpaired Student t test. Nonparametric Mann-Whitney test and Kruskal-Wallis ANOVA were used for comparisons between groups that did not follow a normal distribution. A confidence level of 95% was accepted as significant. Analyses were performed with the SPSS 10.0 statistical package (SPSS Inc., Chicago, IL).


Rats of each group had similar weights at the time of surgery. Growth patterns of rats can be seen in Fig. 1. At death, most BDL and BDL+HD rats had ascites and evident signs of portal hypertension. In the histological analysis of the liver at death, 70% of animals had cirrhosis, 20% showed advanced fibrosis, and 10% had developed moderate fibrosis.

Figure 1.

Growth pattern of rats. Animals showed loss of weight after the start of ammonia diet or pair feeding. *P < .05 between BDL and sham PF vs. sham. **P < .001 between BDL PF and BDL+HD vs. sham. There were no significant differences between BDL PF and BDL+HD. BDL, bile duct ligation; BDL+HD, bile duct ligation plus hyperammonemic diet; PF, pair fed; HD, hyperammonemia diet.

Biochemical Characterization

HD and BDL+HD rats were hyperammonemic in comparison with sham animals (Table 1), but only BDL+HD animals showed increased ammonia in the cerebral cortex (Table 1). Biochemical and amino acid patterns were similar in BDL and BDL+HD rats (Table 1 and Supplemental Table 1).

Table 1. Plasma Biochemical Measurements and Cerebral Ammonia
  • NOTE. Values are the mean ± standard deviation of duplicate samples from at least 7 rats per group.

  • Abbreviations: ANOVA, post-hoc Newman-Keuls; BDL, bile duct ligation; BDL+HD: bile duct ligation plus hyperammonemic diet; PF: pair fed; HD: hyperammonemic diet.

  • *

    P < .05 for the difference from sham and sham PF animals.

  • a

    P < .05 for the difference from BDL-PF animals.

  • b

    P < .05 for the difference from HD animals.

  • c

    P < .05 for the difference from BDL, BDL PF animals.

  • d

    P < .05 for the difference from all other groups.

Plasma ammonia (μmol/L)112 ± 3095 ± 26185 ± 54*a122 ± 7070 ± 25170 ± 39*a
ALT (U/L)34 ± 1429 ± 726 ± 866 ± 20*b61 ± 17*b77 ± 41*b
GGT (U/L)15 ± 717 ± 616 ± 751±19*b51 ± 15*b60±13*b
Alkaline phosphatase (U/L)172 ± 62144 ± 62153 ± 40456 ± 140*b301 ± 153*b343 ± 198*b
Total protein (g/dL)6.3 ± 0.46.4 ± 0.35.8 ± 0.45.9 ± 0.36.0 ± 0.35.1 ± 0.5
Bilirubin (mg/dL)0.1 ± 0.00.2 ± 0.10.1 ± 0.010.8 ± 2.7*b8.7 ± 4.8*b9.1 ± 2.4*b
Glucose (mg/dL)207 ± 47197 ± 48200 ± 31160 ± 47117 ± 15135 ± 21
Urea (mg/Dl)37 ± 1256 ± 2870 ± 13*c35 ± 631 ± 763 ± 6*c
Creatinine (mg/dL)0.3 ± 0.00.3 ± 0.00.3 ± 0.00.3 ± 0.10.3 ± 0.00.2 ± 0.1
Na (mEq/L)146 ± 5148 ± 8148 ± 4146 ± 4149 ± 5150 ± 8
Plasma osmolarity (mOsm/L)310 ± 12317 ± 10320 ± 15307 ± 12310 ± 10319 ± 12
Brain ammonia (μmol/g)0.16 ± 0.100.20 ± 0.100.25 ± 0.130.29 ± 0.180.18 ± 0.050.73 ± 0.16d

To study inflammation-related parameters, plasmatic nitrites/nitrates, tumor necrosis factor alpha (TNF-α), and IL-6 were measured. BDL animals showed higher values in plasmatic nitrites/nitrates (sham, 16,2 ± 7.4; HD, 12.4 ± 2.7; BDL, 30.1 ± 14.8; BDL+HD, 45.0 ± 20.0; P = .001), TNF-α (sham, 49 ± 12; HD, 33 ± 7; BDL, 234 ± 61; BDL+HD, 187 ± 39; P < .05), and IL-6 (sham, 30.8 ± 5.3; HD, 31.0 ± 12.7; BDL, 44.0 ± 7.2; BDL+HD, 41.6 ± 9.5; P < .01). Results can be seen in Fig. 2. Body temperature did not show significant changes in any group. At the start of the study, mean temperature was 35.6 ± 0.6°C, and at death, mean body temperature was sham, 37.5 ± 1.4°C; HD, 36.1 ± 0.4°C; BDL, 37.0 ± 0.9°C; BDL+HD, 35.3 ± 0.9°C (P > .05).

Figure 2.

Plasma concentration of inflammatory mediators. (A) IL-6. (B) TNF-α. (C) nitrites/nitrates. Values are the mean Data from BDL+HD, BDL and hyperammonemia (1 week) are obtained from the present study ± standard deviation from at least seven rats per group. *P < .05 for the difference from sham, sham PF, and HD. **P < .01 for the difference from sham, sham PF, and HD. aP < .01 for the difference between BDL PF and BDL. bP < .01 for the difference between BDL PF and BDL+HD. ANOVA, post-hoc Newman-Keuls; BDL, bile duct ligation; BDL+HD, BDL plus hyperammonemic diet; PF, pair fed.

Neuropathological Characterization

Alzheimer Type II Astrocytes.

We analyzed the selected brain regions for the presence of Alzheimer type II astrocytes, the neuropathological hallmark of hepatic encephalopathy.19, 20 These astrocytes would become evident with thionin and cresyl violet stains. In BDL and sham rats, astrocytes presented a round or oval nucleus, spotted chromatin, and small nucleolus, without evidence of Alzheimer type II astrocytes (Fig. 3A). We only found examples of Alzheimer type II astrocytes in BDL+HD rats (Fig. 3B, Supplemental Fig. 1). In this group, some astrocytes had larger nuclei compared with controls, with very diluted chromatin (pale in appearance), and sometimes, prominent nucleolus (Fig. 3B). In HD rats, the nucleus of the astrocytes was similar to sham and BLD rats; however, in some astrocytes they were larger than normal and with pale chromatin. Findings such as lobulation of the nuclei were seldom found. Nuclear morphology of astrocytes was similar in all the cortical regions analyzed, although it was more evident in those layers where cellular density was lower (molecular layer of the dentate gyrus, stratum oriens, radiatum and lacunosum-moleculare of CA3-CA1 fields of the hippocampus, and layer I in the neocortex).

Figure 3.

Photomicrographs showing astrocytes immunoreactive to GFAP in coronal sections of the parietal cortex of sham (A,C), BDL (D), HD (E), and BDL+HD (B,F) rats. Sections shown in A and B were counterstained with cresyl violet. In cresyl violet–counterstained sections, astrocytes immunoreactive to GFAP in BDL+HD rats (arrows in B) show a large nucleus with pale chromatin as in Alzheimer type II astrocytosis. In BDL+HD (B,F) rats, GFAP immunoreactive astrocytes (arrows) appear less intensely stained than in sham rats (arrows in A and C). Same magnification for A-B, and for C-F. BDL, bile duct ligation; HD, hyperammonemic diet; BDL+HD, bile duct ligation plus hyperammonemic diet.

GFAP and GS Immunocytochemistry.

In all the regions analyzed, immunocytochemically stained astrocytes showed a typical morphology such as described previously.21, 22 In BDL, HD, and BDL+HD rats, GFAP-immunoreactive astrocytes were present in the same cortical layers and they had the same fibrous appearance as in controls (Fig. 3C–F). No differences were seen in the density of immunoreactive astrocytes, and frontal and parietal cortices were similar in all groups (Supplemental Fig. 2). GS-positive cells were found in all regions analyzed. The hippocampus presented GS-positive astrocytes distributed in all its regions, albeit with density differences. In a qualitative analysis, BDL+HD rats showed a decrease in the number of GS astrocytes, although those present displayed larger size and a stronger immunoreactivity to GS compared with controls.

Brain Osmolytes and Brain Edema

In BDL+HD rats, mean brain glutamine increased by 87%, and mean myoinositol decreased by 38% with respect to sham operated rats (Table 2). BDL rats had minor and nonsignificant changes in brain glutamine and myoinositol compared with controls. Examples of HPLC chromatograms of brain extracts from different groups are shown in Fig. 4. Changes in other organic osmolytes can be also seen in Table 2.

Table 2. Brain Organic Osmolytes
  • NOTE. Osmolytes (mmol/kg wet tissue) were measured by HPLC (see Materials and Methods). Values are mean ± standard deviation of duplicate samples from at least 8 rats per group. Values that are significantly different from sham rats are indicated as follows.

  • *

    P < .05 for the difference between BDL+HD and all the other groups.

  • Abbreviations: ANOVA, post-hoc Newman-Keuls; BDL, bile duct ligation; BDL+HD, bile duct ligation plus hyperammonemic diet; PF: pair fed; HD, hyperammonemia diet.

Glutamine7.1 ± 0.66.4 ± 0.98.1 ± 0.97.4 ± 1.47.4 ± 0.312.4 ± 4.0*
Myoinositol6.6 ± 1.25.6 ± 0.96.2 ± 0.55.3 ± 0.86.0 ± 0.24.1 ± 0.8*
Taurine5.1 ± 0.65.0 ± 0.85.5 ± 0.46.1 ± 0.46.0 ± 0.26.1 ± 0.6
Urea4.1 ± 1.36.7 ± 1.04,8 ± 0.66.0 ± 1.76.3 ± 1.19.4 ± 2.8*
Glycine7.1 ± 1.17.2 ± 1.17.7 ± 0.67.2 ± 2.77.0 ± 0.78.6 ± 2.5
Creatine9.9 ± 1.79.6 ± 1.69.3 ± 0.59.4 ± 1.59.3 ± 0.79.6 ± 1.0
Figure 4.

High-performance liquid chromatography chromatograms of brain extracts from control (A), BDL (B), and BDL+HD (C) rats. BDL, bile duct ligation; BDL+HD, bile duct ligation plus hyperammonemic diet.

BDL+HD rats had an increase in the brain water content that indicated low-grade brain edema (Fig. 5). There were no differences in brain water content between BDL, HD, and sham rats (sham, 79.84% ± 0.07%; HD, 79.91% ± 0.04%; BDL, 79.73% ± 0.12%; BDL+HD, 80.36% ± 0.13%; P = .007).

Figure 5.

Brain water in cortex (%). (*) indicates statistically significant differences between BDL+HD with the other group of rats (P < .05; ANOVA and Newman-Keuls test). BDL, bile duct ligation; BDL+HD, BDL plus hyperammonemic diet; PF, pair fed; HD, hyperammonemic diet.

Behavior Studies


Sham rats stayed on the rotarod 192 ± 13 seconds, similarly to sham-PF rats (185 ± 14 seconds) and to HD rats (175 ± 12 seconds). Both BDL rats and BDL-PF rats showed a mild impairment of motor coordination (132 ± 14 seconds for BDL rats and 131 ± 13 seconds for BDL-PF rats, P < .01 as compared with sham rats, sham-PF rats, and P < .05 as compared with HD rats), and BDL+HD exhibited a more important impairment of motor coordination (72 ± 7 seconds, P < .0001 as compared with sham, sham-PF, and HD rats) (Fig. 6A).

Figure 6.

Result of motor coordination tests: rotarod (A) and beam walking test (B–C). (A) **P < .01 for a difference between BDL and BDL-PF as compared with sham, or sham-PF or HD rats. ***P < .001 for a difference between BLD+HD as compared with sham, or sham-PF or HD rats. (B) aP < .05 for a difference between BDL-HD as compared with sham PF rats and HD rats. **P < .01 for a difference between BDL+HD as compared with sham rats. (C) ***P < .001 for a difference between BLD+HD as compared with each of the other groups. BDL, bile duct ligation; BDL+HD, BDL plus hyperammonemic diet; PF, pair fed; HD, hyperammonemic diet.

Beam Walking Test.

The time to pass across the beam was significantly higher in BDL+HD (23 ± 3 seconds) as compared with sham rats (10 ± 1 seconds) (P < .01), sham-PF rats (14 ± 1 seconds) (P < .05), and HD rats (12 ± 2 seconds) (P < .05). In BDL rats and BDL-PF rats, a trend toward an increase was seen but the difference was not statistically significant (Fig. 6B). The number of foot faults displayed by BDL+HD was significantly higher than in the other groups (P < .001 in each comparison) (Fig. 6C).

Motor Activity.

The spontaneous motor activity in a novel environment was significantly lower in BDL rats (P < .05) (Fig. 7A-B), but still more impaired in BDL+HD rats (P < .01) (Fig. 7A–B). Crossovers in 60 minutes were as follows: sham, 39 ± 6; Sham-PF, 40 ± 6; BDL, 25 ± 2; BDL-PF, 27 ± 2; HD, 41 ± 6; BDL ± HD, 17 ± 3. Rearing in 60 minutes was as follows: Sham, 100 ± 14; Sham-PF, 125 ± 10; BDL, 64 ± 7; BDL-PF, 70 ± 6; HD, 99 ± 8; BDL+HD, 43 ± 7.

Figure 7.

Results of motor activity tests. (A) Crossovers in 60 minutes. (B) Rearings in 60 minutes. *P < .05 for a difference between BDL or BDL-PF as compared with sham, sham-PF, and HD rats. **P < .01 for a difference between BDL+HD as compared with sham, sham-PF, and HD rats. BDL, bile duct ligation; BDL+HD, BDL plus hyperammonemic diet; PF, pair fed; HD, hyperammonemic diet.


The model of BDL+HD in rats reproduces the main characteristics of acute HE previously observed in human liver cirrhosis (Table 3). Striking similarities were seen between BDL+HD model and human episodic HE, in terms of brain ammonia, amino acid pattern, neuropathology, neurochemical pattern of organic osmolytes, presence of brain edema, and activation of the inflammatory system. Other models commonly used to study episodic HE in cirrhosis have not demonstrated this complete set of similarities (Table 3).

Table 3. Comparison Between Animal Models of Chronic HE and Human HE
 BDL+HDBDLHyperammonemic Diet (1 week)Chronic Hyperammonemia (≥3 weeks)Portacaval AnastomosisHuman Episodic HE
  1. NOTE. Data from BDL+HD, BDL and hyperammonemia (1 week) were obtained from the current study.

Brain ammonia==10
Fischer ratio==Not reported(42)(43)
Alzheimer type II astrocytesYesNoNoNot reportedNo(44)Yes
Glutamine/myoinositol↑↑/↓=/==/=10/Not reported↑/↓(40)↑↑/↓ (23;24;26)
Brain edemaYesNoNoNot reportedNo(40;41)Yes(45–47)
Activation inflammatory systemYesYesNoNot reportedNot reportedYes(34)

We have observed cytoarchitectonic alterations in the cerebral cortex, such as the presence of Alzheimer type II astrocytes, similar to those found in human portosystemic encephalopathy. Moreover, we observed an increase of brain glutamine and a decrease of myoinositol in BDL+HD rats. This metabolic pattern of changes in brain organic osmolytes is characteristic of HE that occurs in liver cirrhosis.3, 23–26 The difference in the proportion of brain organic osmolytes is important to the homeostasis of brain water. BDL rats had minor or insignificant changes in glutamine and myoinositol levels, ultimately showing no signs of brain edema. On the contrary, BDL+HD rats displayed a distinct increase in brain glutamine levels and a decrease in myoinositol. The decrease in the amount of myoinositol is not enough to compensate for the glutamine accumulation, and therefore, stimulates a mild but significant increase in brain water. This is the same pathogenesis of HE observed in human liver cirrhosis: encephalopathy progresses, the proportion of glutamine to myoinositol changes, and finally, when overt encephalopathy is established, the osmoregulatory effect malfunctions as the level of myoinositol decreases and levels of brain glutamine increase.24, 26 With this model, we can confirm the role of the osmoregulatory response to myoinositol in the development of brain edema in animals with chronic liver disease.

BDL animals did not have obvious signs of infection, fever, or signs of sepsis at death. Nonetheless, these animals showed activation of pro-inflammatory cytokines, similar to those found in human cirrhosis.27–29 Patients with cirrhosis without signs of active infection show increased levels of nitrite/nitrate that correlate with endotoxemia30 as well as high levels of TNF-α and IL-6.27–29 These changes in pro-inflammatory cytokines can even be seen in early stages of the disease, and they are a consequence of liver cirrhosis and hemodynamic impairment.27, 30 Patients with cirrhosis are frequently in a subclinical inflammatory state. In spite of an absence of signs of active infection, it is possible to detect the presence of bacterial DNA fragments in serum of asymptomatic patients. These fragments reflect repeated episodes of bacterial translocation and are capable of activating cell-mediated immune response and nitric oxide production in peritoneal macrophages.31, 32 The BDL+HD model points out the possible interaction between inflammation and hyperammonemia in liver cirrhosis. BDL rats showed the same degree of activation of the inflammatory system as BDL+HD animals; however, their behavioral changes are mild. Conversely, in BDL+HD animals, the addition of ammonia as a precipitant factor, over a background that combines liver disease and activation of inflammatory mediators, provokes a greater behavioral impairment. The importance of inflammation in the development of HE in liver cirrhosis has recently been underscored.33, 34 This model is useful in the study of how inflammatory mediators may exert synergistic effects with toxins, such as ammonia, in the pathogenesis of HE.

The lack of an ideal animal model of HE has greatly hindered progress in defining the primary pathophysiology responsible for the mediation of abnormalities of nitrogen metabolism in cirrhosis.35 Until now, there has not been a satisfactory animal model of chronic liver disease with HE and abnormalities in nitrogen metabolism, as seen in humans.35, 36 The animal models currently used to study hepatic encephalopathy are the hyperammonemic rat without liver disease37, 38 and the portacaval anastomosis model.39 The first one reproduces the effects of the ammonia, the main toxin implied in the pathogenesis of HE, and the second seems appropriate as an equivalent to minimal human HE.36 These two models lack a very important factor implied in the genesis of HE—hepatocellular injury. Moreover, neither has shown the presence of low-grade brain edema.40, 41

The model proposed here has several interesting features: First, it is a reproducible model. Most of the animals reach the desired state of liver impairment and hyperammonemia; 90% develop cirrhosis or advanced liver fibrosis. Second, this model can be considered appropriate for the study of acute HE in chronic liver disease. Third, results can be obtained at a low cost and using an easy surgical intervention. Finally, BDL+HD rats have a strong impairment in motor coordination and activity. This observation correlates with some of the typical signs found in patients with HE developed during chronic liver disease. Nonetheless, a more complete neurophysiological characterization of the model would be necessary, by means of electroencephalogram spectral analysis and evoked responses.

The current model could be useful in the study of the molecular basis for neurological alterations in HE. As we demonstrated in a previous work, BDL+HD is a good model to study alterations in the modulation of soluble guanylate cyclase by nitric oxide in brain.12 This model has not tested the effect of other substances involved in the pathogenesis of HE in cirrhosis, such as GABA or manganese. The relationship between ammonia and other brain toxins in liver disease could be the scope of future studies using this model.

In conclusion, the model proposed here reproduces the main neuropathological features of human HE caused by liver cirrhosis. Moreover, BDL+HD rats also have the characteristic chemical brain changes leading to the appearance of low-grade brain edema, and show activation of the inflammatory system. This model offers the chance to study these key events in the pathogenesis of HE in liver cirrhosis and could be useful in developing a greater knowledge of the pathophysiology and treatment for this frequent complication associated with liver cirrhosis.


The authors thank Cristina Alenda and Artemio Payá for the pathological evaluation of the animals' liver, and Ximena Sylveira and José Antonio Pérez de Gracia for their technical assistance. We are also grateful to José Such for helpful discussions.