Indomethacin prevents the development of experimental ammonia-induced brain edema in rats after portacaval anastomosis

Indomethacin Prevents the Development of Experimental AmmoniaInduced Brain Edema in Rats After Portacaval Anastomosis CHUHAN CHUNG, JEANNE GOTTSTEIN, AND ANDRES T. BLEI

Patients with fulminant hepatic failure (FHF) die with brain edema, exhibiting an increased cerebral blood flow (CBF) at the time of cerebral swelling. Mild hypothermia prevents brain edema in experimental models and in humans with FHF, an effect associated with normalization of CBF. To study the effects of alterations of CBF on the development of brain edema, we administered intravenous (IV) indomethacin to rats receiving an ammonia infusion after portacaval anastomosis. This model predictably develops brain edema and a marked increase in CBF at 3 hours of infusion. Brain water was measured with the gravimetry technique; CBF was monitored with both laser Doppler flowmetry and radioactive microspheres, whereas intracranial pressure (ICP) was monitored with a cisterna magna catheter. Coadministration of indomethacin prevented the increase in CBF seen with ammonia alone (110 ⴞ 19% vs. ⴚ2 ⴞ 9%) as well as the increase in brain water (80.86 ⴞ 0.12% vs. 80.18 ⴞ 0.06%) and the increase in ICP. Plasma ammonia and brain glutamine levels were markedly elevated in the ammonia-infused group and unaffected by indomethacin. However, ammonia uptake by the brain was significantly reduced by indomethacin. Levels of 6-ketoPGF1␣, a stable metabolite of prostacyclin, were reduced in the cerebrospinal fluid (CSF) of indomethacin-treated animals. As with mild hypothermia, avoiding cerebral vasodilatation with indomethacin will prevent the development of brain edema in this hyperammonemic model. Cerebral vasoconstriction reduces cerebral ammonia uptake and, if selective to the brain, may be of benefit in FHF. (HEPATOLOGY 2001;34:249-254.) Brain edema is a leading cause of death in fulminant hepatic failure (FHF). Patients with FHF who develop brain edema and intracranial hypertension exhibit an increase in cerebral blood flow (CBF).1-3 Changes in cerebral perfusion have been documented with different techniques, are not related to alterations in systemic hemodynamics,4 and contribute to intraAbbreviations: FHF, fulminant hepatic failure; CBF, cerebral blood flow; PCA, portacaval anastomosis; ICP, intracranial pressure; LDF, laser Doppler flowmetry; MAP, mean arterial pressure; IV, intravenous; CSF, cerebrospinal fluid; Veh, saline vehicle. From the Hepatology Section and Department of Medicine, VA Chicago Health Care System, Lakeside Division and Northwestern University, Chicago, IL. Received February 23, 2001; accepted May 21, 2001. Supported by a Merit Review from the Veterans Administration Research Service and the Stephen B. Tips Memorial Fund at Northwestern Memorial Hospital. Address reprint requests to: Andres T. Blei, M.D., Department of Medicine (111E), VA Chicago Health Care System, Lakeside Division, 333 E. Huron St., Chicago, IL 60611. Fax: 312-695-5998. Copyright © 2001 by the American Association for the Study of Liver Diseases. 0270-9139/01/3402-0006$35.00/0 doi:10.1053/jhep.2001.26383

cranial hypertension by increasing cerebral blood volume. Although some subjects with FHF show a reduced CBF,5,6 the majority of these patients do not exhibit brain edema.1-2 An increase in cerebral perfusion has been described in experimental models of severe hyperammonemia.7 Recent studies link elevated plasma ammonia levels to the development of cerebral herniation in FHF.8,9 Insight into the importance of cerebral perfusion arises from work in an experimental model of brain edema, i.e., rats with portacaval anastomosis (PCA) infused with ammonium acetate. In this preparation, brain swelling and an increase in intracranial pressure (ICP) predictably develop after 3 hours of ammonia infusion in the absence of acute liver injury, allowing the performance of pathophysiologic studies in an otherwise stable model.10 Using the radioactive microsphere technique, we have shown a selective increase in CBF at 3 hours of infusion,11 without concomitant changes in systemic hemodynamics or flow to other regional beds. In this model, the increase in cerebral perfusion appears related to an intracerebral signal generated after the detoxification of ammonia to glutamine in astrocytes. Administration of methionine-sulfoximine, an irreversible inhibitor of glutamine synthetase, ameliorated the increase in CBF seen in this model.12 The nature of the intracerebral signal has not been elucidated. Our initial postulate of increased nitric oxide synthesis in the brain12,13 led to experiments using nonselective and selective inhibitors of nitric oxide synthase isoforms. Under these conditions, the rise in CBF was still present.14 We proposed that the development of cerebral hyperemia is necessary for the appearance of cerebral swelling in this model. Mild hypothermia (33°C and 35°C) was shown to prevent the development of ammonia-induced brain edema11 as well as forestall brain water accumulation in a model of FHF.15 Clinical studies confirm the protective effect of hypothermia to 32°C in patients with FHF and severe intracranial hypertension.3 The mechanisms by which hypothermia reduces ICP may be multiple,16 but an effect on cerebral perfusion has been well documented in our experimental model11 as well as in humans with FHF.3 High values of CBF seen in both the experimental and human condition are normalized by hypothermia. We designed an experiment in which indomethacin, a known cerebral vasoconstrictor,17 was coadministered to PCA rats receiving an ammonia infusion. The drug is a nonselective inhibitor of endothelial cyclooxygenase and has minimal penetration into the brain.18 If a selective effect on cerebral blood flow influences the development of brain edema in this model, such findings would support the contention that cerebral hyperemia plays a critical role in the development of brain edema in FHF.

249

250 CHUNG, GOTTSTEIN, AND BLEI

HEPATOLOGY August 2001

This study was performed in male Sprague-Dawley rats (Charles River, Wilmington, DE) weighing 300 to 400 g. Once received, the animals were housed in cages, fed a regular chow diet with free access to water, and kept on a 12:12 light/dark cycle. After 1 week, they were anesthetized with methoxyflurane and an end-to-side PCA was constructed using a continuous suture technique (7-0 Surgilene; Davis & Geck, Danbury, CT). The anastomosis was completed within 15 minutes. The abdomen was sutured in 2 layers and rats were returned to individual cages where feeding and water ingestion promptly resumed.

was measured gravimetrically using a density gradient of bromobenzene-kerosene pre-calibrated with K2SO4, as previously described.15 The equilibration point of the samples was read at 2 minutes and averaged; conversion from specific gravity to brain water was calculated19 and the results were expressed as the percentage of water content. The other cerebral hemisphere was stored at ⫺70°C for subsequent measurement of glutamine content. Gray matter of the cerebral cortex was homogenized in 8 volumes of 1.2 mol/L HClO4. The supernatant was neutralized with 15% KOH in 0.1 mol/L K2HPO4 and centrifuged, and the supernatant was used to measure glutamine enzymatically.20 Plasma ammonia levels were measured enzymatically using a commercially available kit (Sigma, St. Louis, MO).

Experimental Preparation

Experimental Design

MATERIALS AND METHODS Performance of Portacaval Anastomosis

After 24 hours had elapsed, animals again received methoxyflurane anesthesia; ketamine was administered intramuscularly (100 mg/kg). Catheters (PE-50; Intramedic, Parsippany, NJ) were placed in the femoral artery and both femoral veins for intravenous (IV) infusions. Through a scalp incision, 3 small burr holes, 2 to 3 mm in diameter, were drilled through the skull. One hole was used for placement of a PE-10 catheter into the cisterna magna, as described,10 and one was used for identification of the sagittal sinus. The third hole was used to place a laser Doppler probe (Probe 407; Perimed, Stockholm, Sweden) just above the surface of the dura allowing for continuous measurement of blood velocity using laser Doppler flowmetry (LDF). Flowmetry was performed using a Periflux laser Doppler System 5000 monitor (Perimed). A tracheotomy was performed and the rat was artificially ventilated (Harvard, model 683; South Natick, MA). The arterial and intracisternal catheters were connected to a pressure transducer (Ohmeda, Oxnard, CA) and values of mean arterial pressure (MAP) and ICP were continuously inscribed in a recorder (Sensormedics, Yorba Linda, CA); zero was placed at the mid-level of the body. Recordings of LDF were continuously monitored and stored in a computer using specific software (Perisoft; Perimed) and analyzed for relative percentage changes from baseline. Percentage changes of CBF were analyzed at 30, 90, 120, 150, and 180 minutes of infusion. Body temperature was monitored with an intra-abdominal thermistor and maintained at 36.8 to 37.2°C with the aid of a heating blanket. Experimental Procedure Once the surgery was completed, and stable pressure and flow tracings were noted, ammonium acetate (55 ␮mol/kg/min) or a control infusion of saline was infused intravenously at a rate of 0.038 mL/min for 180 minutes. Ten minutes after the start of the ammonia infusion, half of the animals received a bolus dose of indomethacin (10 mg/kg, IV) diluted in saline, followed by a continuous IV infusion of indomethacin (0.1 mg/kg/h). The remaining half received a saline bolus followed by a continuous infusion of saline. Within a 3-hour period after initiating the ammonia infusion, animals exhibited a significant increase in CBF, as previously described11,12: a point where the study was terminated. Before sacrifice, arterial blood was drawn (0.5 mL) to determine arterial blood gases and oxygen content (Instrumentation Laboratories, Lexington, MA) and plasma was stored at ⫺70°C for later measurement of plasma ammonia. A similar volume was obtained from the sagittal sinus for measurement of blood gases and oxygen content. Samples (40 ␮L) were obtained from the cerebrospinal fluid (CSF) via the cisterna magna catheter and stored at ⫺70°C for later measurement of 6-keto PGF1␣ levels, a stable metabolite of prostacyclin, using a commercially available Enzyme Immunoassay Kit (Cayman Chemicals, Ann Arbor, MI). Once the experiments were completed, the animals were immediately decapitated, and the brain quickly removed. One cerebral hemisphere was kept at 4°C and processed within 30 minutes. The hemisphere was cut into coronal slices and 10-mg samples were obtained from the gray matter of the cortex. Water content of each specimen

Two sets of experiments were performed. Experiment A. In the first experiment, we evaluated the effect of

indomethacin on relative changes of cerebral perfusion and brain water in PCA rats receiving an ammonia infusion using LDF. Thirty rats were divided into 4 groups: control ⫹ vehicle (Veh) (n ⫽ 8); ammonia ⫹ Veh (n ⫽ 7); control ⫹ indomethacin (n ⫽ 7); ammonia ⫹ indomethacin (n ⫽ 8). Animals were sacrificed at 180 minutes of infusion. Experiment B. In the second experiment, we sought to confirm the main findings of experiment A by measuring absolute values of CBF using the microsphere technique, as previously described.8,9 Before sacrifice, 80,000 microspheres (15 ⫾ 2 ␮m) labeled with 85Sr (New England Nuclear, Boston, MA) were injected into the left ventricle for measurements of CBF. At the same time arterial blood was collected for 1 minute from the femoral artery at a rate of 0.786 mL/min. The brain and kidneys were desiccated overnight in an oven at 160°C, homogenized, and placed in counting tubes. Radioactivity in tissues and arterial blood was measured using a gamma counter (Captus 2000; Capintec, NJ). Agreement between right and left kidney flows ensured the adequacy of the microsphere injection. Regional blood flow (brain, kidney) ⫽ 0.786 ⫻ organ counts/arterial blood counts. All animals received the same ammonium acetate (55 ␮mol/kg/ min) infusion. One half of the animals received indomethacin as described in experiment A. The other half were administered saline vehicle. Twelve rats were divided into 2 groups (n ⫽ 6): NH3 ⫹ saline, NH3 ⫹ indomethacin. All animals were sacrificed at 180 minutes of infusion. Statistical Analysis One-way analysis of variance was used to analyze differences between the 4 groups in experiment A. Differences between group means were calculated using the Tukey’s test. Values are reported as mean ⫾ SE. In experiment B, the Student’s unpaired t test was used to compare differences between the 2 groups. The Animal Care and Use Committee at VA Chicago Health Care System Lakeside Division approved this study. RESULTS Experiment A General Aspects. Body weights in the 4 groups were similar: control ⫹ Veh (359 ⫾ 16 g), ammonia ⫹ Veh (342 ⫾ 16), control ⫹ indomethacin (371 ⫾ 15), and ammonia ⫹ indomethacin (348 ⫾ 13). Baseline arterial pressures and heart rates were similar in all groups. MAP and heart rate were similar in all 4 groups at the time of sacrifice (Table 1). ICP rose significantly (P ⬍ .05) in the ammonia ⫹ Veh group as compared with all other groups; the small absolute increase in ICP reflects the termination of the experiment before a larger increase occurs. Arterial blood gases were within normal limits at the time of sacrifice (Table 1).

HEPATOLOGY Vol. 34, No. 2, 2001

251

CHUNG, GOTTSTEIN, AND BLEI TABLE 1. General Characteristics—Experiment A

MAP (mm Hg)

Heart Rate (beats/s)

ICP (mm Hg)

Final Arterial Blood Gases

Group

Basal

Final

Basal

Final

Basal

Final

pH

pCO2

pO2

Control ⫹ Veh (n ⫽ 8) Ammonia ⫹ Veh (n ⫽ 7) Control ⫹ indomethacin (n ⫽ 7) Ammonia ⫹ indomethacin (n ⫽ 8)

90 ⫾ 3 89 ⫾ 4

83 ⫾ 2 84 ⫾ 3

286 ⫾ 5 257 ⫾ 9

268 ⫾ 9 249 ⫾ 15

1.5 ⫾ 0.2 1.8 ⫾ 0.3

2.3 ⫾ 0.2 5.8 ⫾ 1.6*

7.40 ⫾ 0.01 7.45 ⫾ 0.01

40 ⫾ 1 39 ⫾ 2

92 ⫾ 1 92 ⫾ 2

87 ⫾ 2

82 ⫾ 2

280 ⫾ 10

260 ⫾ 11

1.6 ⫾ 0.3

2.2 ⫾ 0.2

7.41 ⫾ 0.01

38 ⫾ 1

96 ⫾ 3

85 ⫾ 1

78 ⫾ 2

283 ⫾ 9

250 ⫾ 11

1.6 ⫾ 0.2

3.3 ⫾ 0.5

7.39 ⫾ 0.01

41 ⫾ 1

91 ⫾ 1

NOTE. Values are expressed as mean ⫾ SE. One-way ANOVA with Tukey’s test. * P ⬍ .05 ammonia ⫹ Veh vs. all other groups.

Plasma Ammonia and Brain Glutamine. Predictably, both groups of animals that received an ammonia infusion had significantly (P ⬍ .01) higher levels of plasma ammonia than the 2 groups that received the control infusion (Table 2). Brain glutamine levels were also significantly (P ⬍ .01) increased in the ammonia-infused animals compared with the control-infused groups (Table 2). Control animals had received a PCA and thus exhibit higher plasma ammonia and brain glutamine than normal rats: in our laboratory, the average is 38 ␮mol/L and 4 ␮mol/g, respectively, in normal animals. LDF and Brain Water. Figure 1 depicts 3 tracings, which represent the measurement of LDF during 180 minutes of infusion. At 30 minutes of infusion, LDF was significantly (P ⬍ .05) reduced in the control ⫹ indomethacin group (⫺30 ⫾ 6%) as compared with the control ⫹ Veh group (⫺6 ⫾ 7%) (Fig. 2). No significant differences were seen in LDF between the 4 groups at 60 or 90 minutes of infusion. However, at 120 minutes of infusion LDF rose significantly (P ⬍ .01) in the ammonia ⫹ Veh group (42 ⫾ 12%) as compared with the control ⫹ Veh group (⫺6 ⫾ 7%) (Fig. 2). Values of LDF in the indomethacin groups at 120 minutes of infusion were as follows: control ⫹ indomethacin group (⫺19 ⫾ 4%), ammonia ⫹ indomethacin group (⫺19 ⫾ 11%). At 180 minutes of infusion, the increase in LDF seen in the ammonia ⫹ Veh group (110 ⫾ 19%) was significantly (P ⬍ .01) greater than all other groups (Table 2). All indomethacin groups showed no increase in LDF (Fig. 2). The decrease in flow seen initially with indomethacin appeared to mitigate with time and at 180 minutes of infusion the values of LDF seen in the indomethacin groups were not significantly different from those seen in the control-infused group (Fig. 2). The ammonia ⫹ Veh group had a significantly (P ⬍ .01) lower value of arteriovenous oxygen difference (A-V ⌬O2) (2.7 ⫾ 0.5 vol%) as compared with the other groups (Table 2).

In ammonia ⫹ Veh animals brain water rose significantly (P ⬍ .01) as compared with all other groups (Fig. 2). 6-Keto PGF1␣ Levels. In the 2 groups that received indomethacin, there was a significant (P ⬍ .01) decrease in 6-keto PGF1␣ levels measured in the CSF as compared with the other 2 groups, suggesting inhibition of prostacyclin synthesis (Fig. 3). Ammonia infusion did not result in an increase in the levels of this stable metabolite of prostacyclin. Experiment B General Aspects. All baseline variables measured in the 2 groups were similar. Values of MAP and arterial blood gases were also similar at the time of sacrifice. A significant (P ⫽ .001) increase in ICP was seen in the NH3 ⫹ saline group (Table 3). At 180 minutes, CBF was significantly (P ⬍ .01) higher in the NH3 ⫹ saline group (Table 3). Ammonia Uptake by the Brain. No significant differences were seen in the arterial plasma ammonia levels. However, there were significant differences (P ⫽ .01) seen in the sagittal sinus ammonia values between the NH3 ⫹ saline group (634 ⫾ 67 ␮mol/L) versus the NH3 ⫹ indomethacin group (1,037 ⫾ 117 ␮mol/L). As a result, ammonia uptake (␮mol/min) (calculated as CBF x [arterial-sagittal sinus ammonia]), was significantly reduced in the indomethacin group (Table 3). 6-Keto PGF1␣ Levels. As in experiment A, the levels of 6-keto PGF1␣ were significantly (P ⫽ .04) reduced in the NH3 ⫹ indomethacin group (136 ⫾ 24 pg/mL) as compared with the NH3 ⫹ saline group (258 ⫾ 50 pg/mL). In 4 additional experiments, we assessed the permeability of the blood-brain barrier during ammonia infusion in PCA rats (n ⫽ 4) using Evans blue dye. Animals were infused with ammonia for 180 minutes as in the previous 2 experiments. Just prior to sacrifice, 100 mg/kg Evans blue dye (Sigma) was injected IV as a 10% solution. The animals were sacrificed by decapitation and the brains removed. The brains were divided

TABLE 2. Specific Measurements—Experiment A

Group

Final Arterial NH3 (␮mol/L)

Final Brain Glutamine (␮mol/g)

Final Arteriovenous Oxygen Difference (vol%)

Final % Change in LDF

Control ⫹ Veh (n ⫽ 8) Ammonia ⫹ Veh (n ⫽ 7) Control ⫹ indomethacin (n ⫽ 7) Ammonia ⫹ indomethacin (n ⫽ 8)

108 ⫾ 15 940 ⫾ 43* 150 ⫾ 25 970 ⫾ 130*

9.7 ⫾ 0.5 23.2 ⫾ 1.1* 12.4 ⫾ 1.1 24.0 ⫾ 1.2*

6.6 ⫾ 0.4 2.7 ⫾ 0.5† 7.0 ⫾ 0.8 6.7 ⫾ 0.5

⫺6 ⫾ 5 110 ⫾ 19† ⫺8 ⫾ 4 ⫺2 ⫾ 9

NOTE. Values are expressed as mean ⫾ SE. One-way ANOVA with Tukey’s test. * P ⬍ .01 ammonia-infused groups vs. control-infused groups. † P ⬍ .01 ammonia ⫹ Veh vs. all other groups.

252 CHUNG, GOTTSTEIN, AND BLEI

HEPATOLOGY August 2001

FIG. 3. Levels of 6-keto PGF1␣ in the CSF of all 4 groups. Control and ammonia-infused animals had significantly higher values than the indomethacin-treated rats (*P ⬍ .01). There was no difference between the first 2 groups.

FIG. 1. Laser Doppler tracings of continuous measurement of relative changes in CBF over the 3 hours of the experiment. Note the increase in CBF in the ammonia-infused group in contrast to the reduction in cerebral perfusion seen with indomethacin alone.

into 1-mm-thick coronal sections. Visible examination and comparison with other tissues noted absence of staining with Evans blue within all areas of brain tissue. DISCUSSION

In this study, we have shown that coadministration of indomethacin with an ammonia infusion to rats after PCA prevents the development of cerebral hyperemia, significantly decreases the degree of cerebral edema, and thwarts the increase in ICP seen in this model. Effects on cerebral perfusion were confirmed using both relative and absolute measurements of CBF. The reduction in CBF seen with indomethacin was seen in both control as well as ammonia-infused rats and is in accordance with the known effects of the drug.17 The dose of indo-

methacin used in this study has been shown to affect CBF without concomitant changes in neuronal activity, as reflected by local cerebral glucose utilization.21 However, the mechanisms responsible for indomethacin-mediated cerebral vasoconstriction are complex and include non–prostaglandin-related pathways. Other inhibitors of cyclooxygenase, such as aspirin and ibuprofen, will not result in cerebral vasoconstriction.22 Alternative mechanisms include interference with the regulation of cerebrovascular tone mediated by extracellular pH,23 a reduction of cerebral temperature,24 as well as a direct non–prostaglandin-mediated vasoconstrictive effect.25 Our study was not designed to dissect this complexity, although we did show evidence of decreased prostanoid formation without effects on body temperature. Indomethacin prevented the marked increase in CBF seen with ammonia infusion. The rise in blood flow of this model appears to originate from an intracerebral signal generated once ammonia is detoxified to glutamine in astrocytes.12 Proposed explanations for this signal have tried to link the observed increase in extracellular brain glutamate (seen in the brain of experimental models of FHF26) with the increase in CBF. Indeed, inhibition of glutamate N-methyl-D-aspartate receptors with memantine was shown to a have beneficial effect on ICP in hyperammonemic rats.27 The increased out-

FIG. 2. Relative changes in LDF over the time of the experiment (left panel). Control ⫹ indomethacin had a significant reduction in LDF at 30 minutes (s ⫽ P ⬍ .05). Ammoniainfused animals had a significant increase after 120 minutes (*P ⬍ .01 vs. other groups). Brain water was significantly increased (right panel) in the ammonia-infused animals (*P ⬍ .01 vs. other groups).

HEPATOLOGY Vol. 34, No. 2, 2001

CHUNG, GOTTSTEIN, AND BLEI

TABLE 3. CBF Measured With Radioactive Microspheres Group

NH3 ⫹ Veh (n ⫽ 6) NH3 ⫹ indomethacin (n ⫽ 6)

Final ICP (mm Hg)

Final CBF (mL/min/100 g)

Ammonia Uptake (␮mol/min)

4.6 ⫾ 0.3

204 ⫾ 29

1.86 ⫾ 0.22*

2.8 ⫾ 0.3*

76 ⫾ 4*

0.24 ⫾ 0.06

* P ⬍ .01 vs. the other group.

put of nitric oxide from the brain12 in the ammonia-PCA model led to the hypothesis of N-methyl-D-aspartate receptor–mediated activation of nitric oxide synthase in postsynaptic neurons as the mechanism responsible for cerebral vasodilatation.13 However, in our laboratory administration of N␻-nitro-L-arginine and 1-(2-trifluoromethylphenyl) imidazole, nonselective and selective inhibitors of neuronal nitric oxide synthase, could not prevent the development of hyperemia.14 It has also been proposed that glutamate-mediated activation of cyclooxygenase in astrocytes may result in the generation of vasodilatory prostaglandins such as PGE1 and PGE2.28 The poor penetration of indomethacin into brain tissue17 makes the astrocyte a less-likely target for indomethacin’s effect on CBF. We did observe, however, a reduction in the CSF levels of 6-keto PGF1␣ with indomethacin, a sampling site that suggests nonvascular effects of the drug. Because control animals receiving indomethacin alone also experienced a reduction in CBF, together with a decrease in the level of 6-keto PGF1␣, changes in prostacyclin metabolites are not specific to the effects of ammonia in this model. Prevention of cerebral hyperemia with indomethacin averted the development of ammonia-induced brain edema. This finding reproduces the sequence seen with mild hypothermia.11 Levels of cerebral glutamine were markedly increased in the ammonia ⫹ indomethacin group, but brain water was not increased. The marked decrease in ammonia uptake with indomethacin (more than 85%) was an unexpected observation. A decrease in the capillary surface exchange area could be postulated, as arterial pH and ammonia levels were unchanged by the drug. A decreased ammonia uptake can be reconciled with the high levels of brain glutamine in the ammonia ⫹ indomethacin group in view of the saturation of glutamine synthetase, a high-affinity, low-capacity detoxifying system.29 The increase in CBF is necessary for the development of cerebral swelling in this model. This increase in flow is observed with both pentobarbital11 and ketamine10 anesthesia. A breakdown of the blood-brain barrier was not observed, as assessed by Evans blue staining. The combination of an osmotic stimulus (glutamine formation in astrocytes) with an increase in CBF explains the considerably faster development of ammonia-induced brain edema in PCA rats as compared with normal animals. In the latter paradigm, total CBF does not increase, and brain edema develops within 6 hours of the same ammonia infusion,30 in contrast to the 3 to 3.5 hours of this preparation. An increase of CBF in normal animals is only seen at very high ammonia levels, 2,000 ␮mol/L.7 Although ammonia levels in our study (900 ␮mol/L) are higher than those seen in human disease, the increase in cerebral perfusion is a reproducible, selective event, with a clear parallel to the increase in CBF seen in human FHF.1-3 Comparisons between human and rat ammonia levels may not be completely valid, because PCA alone in the rat results in ammonia levels (approximately 150 ␮mol/L) that are seen in human FHF.

253

Indomethacin has been administered to individuals with elevated ICP as a result of brain trauma with encouraging results.17,24 However, we do not recommend the use of indomethacin to prevent the development of brain edema and cerebral hyperemia in human FHF. The administration of the drug at an early stage of the neurologic picture in FHF could result in renal vasoconstriction and pose risks of gastrointestinal hemorrhage. We did not administer indomethacin once edema had developed. The drug has been administered in the late stages of evolution of the neurologic picture to an individual with FHF and intracranial hypertension, where high ICP waves were repeatedly aborted with an IV bolus of indomethacin.31 The sum of experimental and clinical observations, with both indomethacin and hypothermia, suggests that pharmacologic manipulation of the cerebral circulation may be an important end point in the management of human disease. A safe pharmacologic agent to selectively constrict the cerebral circulation may allow the design of studies to prevent, rather than treat, the development of brain edema in FHF. REFERENCES 1. Aggarwal S, Kramer D, Yonas H, Obrist W, Kang Y, Martin M, Policare R. Cerebral hemodynamic and metabolic changes in fulminant hepatic failure: a retrospective study. HEPATOLOGY 1994;19:80-87. 2. Wendon JA, Harrison PM, Keays R, Williams R. Cerebral blood flow and metabolism in fulminant liver failure. HEPATOLOGY 1994;19:1407-1413. 3. Jalan R, Damink SW, Deutz NE, Lee A, Hayes PC. Moderate hypothermia for uncontrolled intracranial hypertension in acute liver failure. Lancet 1999;354:1164-1168. 4. Larsen FS, Strauss G, Knudsen GM, Herzog TM, Hansen BA, Secher NH. Cerebral perfusion, cardiac output, and arterial pressure in patients with fulminant hepatic failure. Crit Care Med 2000;28:996-1000. 5. Almdal T, Schroeder T, Ranek L. Cerebral blood flow and liver function in patients with encephalopathy due to acute and chronic liver diseases. Scand J Gastroenterol 1989;24:299-303. 6. Durham S, Yonas H, Aggarwal S, Darby J, Kramer D. Regional cerebral blood flow and CO2 reactivity in fulminant hepatic failure. J Cereb Blood Flow Metab 1995;15:329-335. 7. Dempsey RJ, Kindt GW. Experimental acute hepatic encephalopathy: relationship of pathological cerebral vasodilation to increased intracranial pressure. Neurosurgery 1982;10:737-741. 8. Clemmesen JO, Larsen FS, Kondrup J, Hansen BA, Ott P. Cerebral herniation in patients with acute liver failure is correlated with arterial ammonia concentration. HEPATOLOGY 1999;29:648-653 9. Madl C, Kramer L, Gendo A, Funk G, Bauer E, Zauner C. Prognostic accuracy of sensory evoked potentials and arterial ammonia in predicting development of cerebral edema and death in patients with fulminant hepatic failure [Abstract]. Gastroenterology 2000;118:2400A. 10. Blei AT, Olafsson SM, Therrien G, Butterworth RF. Ammonia-induced cerebral edema and intracranial hypertension in rats after portacaval anastomosis. HEPATOLOGY 1994;19:1431-1444. 11. Cordoba J, Crespin J, Gottstein J, Blei AT. Mild hypothermia modifies ammonia-induced brain edema in rats after portacaval anastomosis. Gastroenterology 1999;116:686-693. 12. Master S, Gottstein J, Blei AT. Cerebral blood flow and the development of ammonia induced brain edema in rats after portacaval anastomosis. HEPATOLOGY 1999;30:876-880. 13. Blei AT, Larsen FS. Pathophysiology of cerebral edema in fulminant hepatic failure. J Hepatol 1999;31:771-776. 14. Larsen FS, Gottstein J, Blei AT. Cerebral hyperemia and neuronal nitric oxide synthase in rats with ammonia-induced brain edema. J Hepatol 2001;34:548-554. 15. Traber PG, Ganger DR, Dal Canto M, Blei AT. Effect of body temperature on brain edema and encephalopathy in the rat after hepatic devascularization. Gastroenterology 1989;96:885-891. 16. Kataoka K, Yanase H. Mild hypothermia—a revived countermeasure against ischemic neuronal damages. Neurosci Res 1998;32:103-117. 17. Harrigan MR, Tuteja S, Neudeck BL. Indomethacin in the management of elevated intracranial pressure: a review. J Neurotrauma 1997;14:637650.

254 CHUNG, GOTTSTEIN, AND BLEI 18. Slavik RS, Rhoney DH. Indomethacin: a review of its cerebral blood flow effects and potential use for controlling intracranial pressure in traumatic brain injury patients. Neurol Res 1999;21:491-499. 19. Marmarou A, Poll W, Shulman K, Bhagavan H. A simple gravimetric technique for the measurement of cerebral edema. J Neurosurg 1978;49: 530-537. 20. Bergmeyer HU (ed). Methods of Enzymatic Analysis. New York: Academic, 1984. 21. McCullough J, Kelly PAT, Grome JJ, Pickard JD. Local cerebral circulatory and metabolic effects of Indomethacin. Am J Physiol 1982;243: H416-H423. 22. Markus HS, Vallance P, Brown MM. Differential effect of three cyclooxygenase inhibitors on human cerebral blood flow velocity and carbon dioxide reactivity. Stroke 1994;25:1760-1764. 23. Wang Q, Paulson OB, Lassen NA. Indomethacin abolishes cerebral blood flow increase in response to acetazolamide-induced extracellular acidosis: a mechanism for its effect on hypercapnia? J Cereb Blood Flow Metab 1993;13:724-727. 24. Jensen K, Ohrstrom J, Cold GE, Astrup J. The effects of indomethacin on intracranial pressure, cerebral blood flow and cerebral metabolism in patients with severe head injury and intracranial hypertension. Acta Neurochir (Wien) 1991;108:116-121.

HEPATOLOGY August 2001 25. Norins NA, Wendelberger K, Hoffman RG, Keller PA, Madden JA. Effects of indomethacin on myogenic contractile activation and responses to changes in O2 and CO2 in isolated feline cerebral arteries. J Cereb Blood Flow Metab 1992;12:866-872. 26. Butterworth RF. Hepatic encephalopathy and brain edema in acute hepatic failure: does glutamate play a role? HEPATOLOGY 1997;25:10321034. 27. Vogels BA, Maas MA, Daalhuisen J, Quack G, Chamuleau RA. Memantine, a non-competitive NMDA receptor mediated antagonist improves hyperammonemia-induced encephalopathy and acute hepatic encephalopathy rats. HEPATOLOGY 1997;25:820-827. 28. Harder DR, Alkayed NJ, Lange AR, Gebremedhin D, Roman RJ. Functional hyperemia in the brain: hypothesis for astrocyte-derived vasodilator metabolites. Stroke 1998;29:229-234. 29. Butterworth RF, Girard G, Giguere JF. Regional differences in the capacity for ammonia removal by brain following portacaval anastomosis. J Neurochem 1988;51:486-490. 30. Takahashi H, Koehler RC, Brusilow SW, Traystman RJ. Inhibition of brain glutamine accumulation prevents cerebral edema in hyperammonemic rats. Am J Physiol 1991;261:H825-H829. 31. Clemmesen JO, Hansen BA, Larsen FS. Indomethacin normalizes intracranial pressure in acute liver failure: a twenty-three-year-old woman treated with indomethacin. HEPATOLOGY 1997;26:1423-1425.