Ammonia and glutamine have been identified as key. Cerebral Metabolism of Ammonia and Amino Acids in Patients With Fulminant Hepatic Failure

Transcription

1 GASTROENTEROLOGY 2001;121: Cerebral Metabolism of Ammonia and Amino Acids in Patients With Fulminant Hepatic Failure GITTE IRENE STRAUSS,* GITTE MOOS KNUDSEN, JENS KONDRUP,* KIRSTEN MØLLER, and FIN STOLZE LARSEN* Departments of *Hepatology, Neurology, and Infectious Diseases, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark Background & Aims: High circulating levels of ammonia have been suggested to be involved in the development of cerebral edema and herniation in fulminant hepatic failure (FHF). The aim of this study was to measure cerebral metabolism of ammonia and amino acids, with special emphasis on glutamine metabolism. Methods: The study consisted of patients with FHF (n 16) or cirrhosis (n 5), and healthy subjects (n 8). Cerebral blood flow was measured by the 133 Xe washout technique. Blood samples for determination of ammonia and amino acids were drawn simultaneously from the radial artery and the internal jugular bulb. Results: A net cerebral ammonia uptake was only found in patients with FHF ( mol 100 g 1 min 1 ). The cerebral glutamine efflux was higher in patients with FHF than in the healthy subjects and cirrhotics, vs and mol 100 g 1 min 1, respectively (P < 0.05). Patients with FHF who subsequently died of cerebral herniation (n 6) had higher arterial ammonia concentrations, higher cerebral ammonia uptake, and higher cerebral glutamine efflux than survivors. Intervention with short-term mechanical hyperventilation in FHF reduced the net cerebral glutamine efflux, despite an unchanged net cerebral ammonia uptake. Conclusions: Patients with FHF have an increased cerebral glutamine efflux, and shortterm hyperventilation reduces this efflux. A high cerebral ammonia uptake and cerebral glutamine efflux in patients with FHF were associated with an increased risk of subsequent fatal intracranial hypertension. Ammonia and glutamine have been identified as key substrates involved in the development of cerebral edema and intracranial hypertension in fulminant hepatic failure (FHF). Hyperammonemia is a consistent finding in patients with FHF and may adversely affect the brain by interfering with neuronal membrane potentials, 1 by altering the interstitial concentrations of glutamate, 2 and by causing disturbances in cerebral blood flow and oxidative metabolism. 3,4 Ammonia (NH 3 ) crosses the blood-brain barrier easily by passive diffusion. We have previously shown that high arterial concentration of ammonia is correlated to cerebral herniation in patients with FHF. 5 Animal studies indicate that ammonia is eliminated from the brain through formation of glutamine. 6 Detoxification of ammonia takes place in perivascular astrocytes. 7 Consequently, astrocytes undergo ammonia-induced metabolic and morphologic derangement, 8 leading to cell swelling, increased intracranial pressure, and death. According to the glutamine hypothesis, accumulation of glutamine results in a shift of water from the extracellular space into the astrocyte. 9 A consistent finding of previous studies of patients with FHF is low or normal plasma concentrations of glutamate, branched-chain amino acids (BCAA) (valine, leucine, and isoleucine), and tryptophan (an aromatic amino acid [AAA]), whereas almost all other plasma amino acid concentrations are elevated Furthermore, Record et al. 11 found that 15 of 18 amino acids were increased in the cerebrospinal fluid and that 15 of 21 amino acids were increased in the frontal cortex at autopsy of patients with FHF. However, although many studies of the cerebral metabolic rate of amino acids and ammonia have been performed in animal models of acute liver failure, 14,15 none have so far been performed in patients with FHF. Thus, the aim of the present study was to measure cerebral metabolism of ammonia and amino acids, with special emphasis on glutamine, in patients with FHF. Because acute mechanical hyperventilation can reduce the intracranial pressure, we also investigated the cerebral metabolism of ammonia and amino acids during intervention with short-term mechanical hyperventilation. Abbreviations used in this paper: AAA, aromatic amino acid; BCAA, branched chain amino acid; CBF, cerebral blood flow; CMR, cerebral metabolic rate; FHF, fulminant hepatic failure by the American Gastroenterological Association /01/$35.00 doi: /gast

2 1110 STRAUSS ET AL. GASTROENTEROLOGY Vol. 121, No. 5 Table 1. Clinical and Laboratory Data (Arterial Concentrations) Age (yr) Sex (M/F) Hgb (mmol/l) INR Bilirubin ( mol/l) ALT (U/L) Creatinine (mmol/l) ph PaCO 2 Temp (kpa) SvJO 2 ( C) Controls (n 8) / Cirrhosis (n 5) 50 9 a 5/ a FHF (n 16) a 5/ a a a a a a b Hgb, hemoglobin; INR, international normalization ratio; ALT, alanine aminotransferase; Temp, temperature. a P 0.05 vs. control (ANOVA). b P 0.05 vs. control. Patients and Methods The regional Scientific Ethical Committee (KF /98) approved the study, and informed consent was obtained from the patients (healthy subjects) or patient s next of kin. The study included 3 groups (Table 1): 16 patients with FHF, 5 patients with cirrhosis of the liver, and 8 healthy subjects. FHF Patients with FHF were intubated and mechanical ventilation instituted (Servo 900C; Siemens, Solna, Sweden), when stage 3 4 hepatic encephalopathy appeared. 16 The cause of FHF was acetaminophen (n 12), non A E hepatitis (n 3), or acute hepatitis B (n 1). The study was conducted 10 (range, 4 24) hours after institution of mechanical ventilation. Fourteen of the patients were sedated with midazolam (median, 6; range, 2 10 mg/h), and the other 2 were not sedated. According to standard recommendation, 16 all patients were given intravenous N-acetylcysteine (6.25 mg kg 1 h 1 ) and 20% glucose (85 ml/h 17 g/h). All medication remained unchanged before and during the entire study period. Mean arterial pressure, pulse, and arterial blood gases had been stable for at least 30 minutes before the study. None of the patients had clinical signs of intracranial hypertension before or during the study (pronation seizures, sluggish pupillary reaction to light, significant pupillary asymmetries, or systemic hypertension). Cerebral herniation was diagnosed clinically when there was the following: (1) no brain stem reflexes, (2) no signal in the middle cerebral artery as measured by Transcranial Doppler sonography, (3) dilated unresponsive pupils, (4) pronation seizures, and/or (5) systemic hypertension. In 3 patients, intracranial pressure monitoring was initiated 1 2 days after participation in the study. Controls All patients with cirrhosis were recruited from the outpatient clinic and classified as Child-Pugh A. All had been abstaining from alcohol at least 6 months. None of the patients received ammonia-lowering agents. All healthy subjects underwent a physical examination, electrocardiogram, and blood tests before inclusion in the study. All controls were investigated after an overnight fast in supine position and were given glucose 5% (340 ml/h 17 g/h) intravenously for hours before cerebral blood flow (CBF) measurement and blood sampling. Catheters Catheters used for monitoring of mean arterial pressure and blood sampling were placed in the right radial artery, and retrogradely in the internal jugular bulb. The left cubital vein was used for 133 Xe infusion. CBF Measurements Controls were studied during spontaneous ventilation only, whereas patients with FHF were studied both during mechanical normo- and hyperventilation. CBF was measured either by the Kety-Schmidt technique with 133 Xe as the tracer or with single photon emission computed tomography, known to yield comparable global CBF values. To ensure steady state conditions during hyperventilation, 133 Xe infusion was started after minutes of hyperventilation. Hyperventilation was maintained throughout the period of investigation and blood sampling, i.e., 40 minutes during the Kety Schmidt technique and 10 minutes during single photon emission computed tomography. At the end of sampling, ventilator settings were returned to baseline levels. A detailed description of these methods is given elsewhere. 17 Amino Acids During each CBF measurement, 3 ml of blood was withdrawn simultaneously from the radial artery and the internal jugular bulb. The blood samples were immediately transferred to ice-cold heparinized tubes, placed on ice, and within 5 minutes centrifuged at 3000 rpm for 15 minutes at 4 C. The resultant heparinized plasma was precipitated with sulfosalicylic acid (6%) containing the internal standard for the analysis, norleucine. The sample was then left on ice for 1 hour. Thereafter, the sample was centrifuged at 3000 rpm for 30 minutes at 4 C, and the supernatant was frozen and stored at 80 C until analysis. Amino acids were separated by ion-exchange high-performance liquid chromatography with fluorescence detection (Waters HPLC system, Milford, MA) using a postcolumn derivatization. Standard curves were linear in the whole concentration range except for glutamine. There-

3 November 2001 CEREBRAL AMINO ACID FLUXES IN FHF 1111 fore, duplicate samples were diluted 10 times for the determination of glutamine. For all amino acid measurements, the coefficient of variation was less than 5%. Ammonia Arterial and venous blood samples for measurements of plasma ammonia concentration were withdrawn simultaneously. One milliliter of blood was collected in a chilled sterilized tube containing 15 IU of lithium-heparin and afterwards immediately placed on ice. Samples were centrifuged at 5 C and analyzed within 20 minutes by use of microdiffusion, quantitation by reaction with bromphenol blue, and spectrophotometry at 600 nm (Kodak Ektachem 700 Analyzer, Clinical Chemistry Slide [NH 3 /AMON]; Eastmann Kodak Co., Rochester, NY). Whole blood ammonia concentration in mol/l was calculated by Conn s formula, 18 NH whole-blood NH plasma and converted from g/100 ml into mol/l. Calculations Cerebral metabolic rates (CMRs) of amino acids and ammonia are calculated as CMR x CBF avd x where CBF is the cerebral blood flow and avd x is the arteriovenous difference of an amino acid or ammonia. The net cerebral nitrogen production subsequently incorporated into glutamine that could not be explained by the cerebral ammonia uptake (non-ammonia Gln N ), is evaluated as non-ammonia Gln N (2 CMR gln ) CMR ammonia where 2 accounts for the 2 nitrogen atoms of glutamine, CMR gln is the cerebral efflux of glutamine, and CMR ammonia is the cerebral uptake of ammonia. The ratio between BCAAs and AAAs, i.e., the Fischer ratio, 19 is calculated as Fischer ratio (valine isoleucine leucine) /(tyrosine phenylalanine). Statistics Unless otherwise stated, the results are presented as mean SD. Normality was tested by the Kolmogorov Smirnov test. Kruskal Wallis analysis of variance (ANOVA) on the Ranks using Dunn s method for multiple comparison was used for comparison between several groups. The Mann Whitney U test was used for comparison of unpaired data when 2 groups were compared, and the Wilcoxon test for paired data (SigmaStat 2.0 Statistical software; Jandel Scientific, San Rafael, CA). A one-sampled test was applied when determining significance from zero. Bonferroni correction factor 5 was applied to P values to adjust for multiple comparison between 2 unpaired groups and paired data. P 0.05 was considered significant. Results Cerebral blood flow was 39 8mL 100 g 1 min 1 in patients with FHF (n 16), 63 9mL 100 g 1 min 1 in healthy subjects, and ml 100 g 1 min 1 in patients with cirrhosis (P 0.05, ANOVA). Ammonia Arterial ammonia concentration was significantly higher in patients with FHF ( mol/l) than in healthy controls (45 10 mol/l) and patients with cirrhosis (67 27 mol/l). There was a net cerebral uptake of ammonia in patients with FHF, whereas the net cerebral uptake of ammonia was not different from zero in healthy controls and patients with cirrhosis (Table 2). The cerebral net extraction fraction of ammonia was 22% 8% in FHF, 4% 23% in healthy subjects, and 4% 18% in patients with cirrhosis (P 0.003, ANOVA). In patients with FHF, cerebral ammonia uptake increased with increasing arterial ammonia concentration (Figure 1A and B). Arterial Concentrations of Amino Acids Except for serine and tryptophan, all the arterial plasma concentrations of amino acids in patients with FHF were significantly different from those of healthy subjects (Table 2). In general, the concentrations were higher, except for the BCAAs and glutamate, which were significantly lower in FHF than in healthy subjects (Table 2). There was no difference in the arterial amino acid concentration between patients with cirrhosis and healthy subjects. Because of technical difficulties in highperformance liquid chromatography separation, ornithine was not measured in patients with cirrhosis. The Fischer ratio was significantly lower in patients with FHF than in healthy subjects and patients with cirrhosis, vs and mol/l, respectively (P 0.05, ANOVA). CMRs In patients with FHF, ph was in the artery and in the internal jugular bulb, in patients with cirrhosis and , and in healthy subjects and , respectively (NS, ANOVA). Cerebral glutamine efflux was significantly higher in patients with FHF than in

4 1112 STRAUSS ET AL. GASTROENTEROLOGY Vol. 121, No. 5 Table 2. Arterial Concentration and CMR of Amino Acids, Ammonia, and Amino Acid-Nitrogen in Healthy Controls, Patients With Cirrhosis, and Patients With FHF Controls (n 8) Cirrhosis (n 5) FHF (n 16) Hyperventilation FHF (n 14) Artery CMR Artery CMR Artery CMR Artery CMR Taurine c Aspartic acid d a OH-proline d Threonine a c a f Serine Asparagine b Glutamate d Glutamine a a c a,c a,g Proline b c a a,g Glycine b a c a Alanine b a d a g Citruline d a,g Valine BCAA d a,f Cystine a d a Methionine c a,f Isoleucine BCAA c a a,g Leucine BCAA d a a,g Tyrosine AAA c a g Phenylalanine AAA c Tryptophan AAA Histidine c Ornithine c Lysine c Arginine d Sum a c a g NOTE. Results are presented as mean SD. Concentrations are in mol/l and CMR in mol/(100 g min). Minus denotes cerebral net efflux. For amino acid-nitrogen, each amino acid was weighted with its number of nitrogen atoms. a P 0.01 vs. 0. b P 0.05 vs. 0. c P 0.05 vs. control and cirrhosis. d P 0.05 vs. control. e P 0.05 vs. control and FHF. f P 0.05 vs. FHF (normoventilation) (paired comparison). g P 0.01 vs. FHF (normoventilation) (paired comparison). healthy controls and patients with cirrhosis (Table 2). Glutamine efflux corresponding to ammonia uptake was significantly higher in FHF than in controls and cirrhosis (Figure 2). The non-ammonia Gln N efflux tended to be higher in patients with FHF compared with the control groups (P 0.11). In patients with FHF, the net cerebral efflux of threonine and tyrosine and the net cerebral uptake of Figure 1. (A) Relationship between the arterial and venous ammonia concentration showing the line of identity and the regression line (slope and is 1) for the cerebral ammonia extraction. (B) Scatterplot between arterial ammonia and net cerebral ammonia uptake in patients with FHF.

5 November 2001 CEREBRAL AMINO ACID FLUXES IN FHF , between survivors and nonsurvivors. The plasma concentration of amino acids tended to be higher in nonsurvivors than in survivors, but the major part did not reach statistical significance (Table 3). The arterial ammonia concentration (P 0.002), the net cerebral ammonia uptake (P 0.02), and the cerebral glutamine efflux (P 0.02) was significantly higher in nonsurvivors than in survivors (Figure 3). The non-ammonia Gln N efflux tended to be higher in nonsurvivors compared with survivors (P 0.10). There was no statistically discernible difference in the cerebral flux or gradient of any of the other measured amino acids between survivors and nonsurvivors (Table 3). Figure 2. The mean cerebral glutamine-nitrogen efflux in patients with FHF during normoventilation (I) and hyperventilation (II), in patients with cirrhosis of the liver and in healthy subjects (ANOVA on Ranks, P 0.05). Dashed areas shows the maximal cerebral glutamine efflux (mean) that can be explained by cerebral ammonia uptake. BCAAs was significantly different from zero (Table 2). When examining gradients alone, the pattern was the same. Hemoglobin remained unchanged from the artery to the venous blood in all 3 groups. Hyperventilation Only 14 patients with FHF could be evaluated during intervention with mechanical hyperventilation. PaCO 2 decreased from to (P 0.05), ph increased from to (P 0.05), and CBF decreased from 40 8to32 5 ml 100 g 1 min 1 (P 0.05). The arterial concentration of all amino acids and ammonia remained unchanged during hyperventilation (Table 2). The cerebral ammonia uptake remained unchanged during hyperventilation, despite a decrease in cerebral glutamine efflux (Figure 2). Changes in net cerebral fluxes during hyperventilation of the other measured amino acids is given in Table 2. Fischer ratio remained unchanged during hyperventilation ( ). Outcome Eleven of 16 patients with FHF fulfilled King s College criteria for liver transplantation. Ten patients survived, of whom 2 underwent emergency liver transplantation on day 1 and 3, respectively, after participation in the study. Six patients died due to cerebral herniation on day after participation in the study (see Patients and Methods). The 2 transplanted patients are omitted from the following data analysis. There was no difference in arterial ph, vs , or venous ph, vs Discussion This study reports for the first time data on the cerebral fluxes of amino acids and ammonia in patients with FHF. Except for glutamine and ammonia, the results do not indicate any large differences in the cerebral fluxes of amino acids between patients with FHF, healthy subjects, and patients with cirrhosis, respectively. The net cerebral glutamine efflux was higher in patients with FHF than in healthy subjects and patients with cirrhosis (Table 2). In addition, there was a striking cerebral uptake of ammonia in patients with FHF, but not in healthy subjects and cirrhotic patients. The major part of this increased cerebral glutamine efflux in patients with FHF could not be explained by the increased cerebral ammonia uptake (Figure 2). Moreover, we also showed that patients who died of cerebral complications had higher cerebral ammonia uptake and cerebral glutamine efflux than patients who survived (Figure 3). Interestingly, we also found that short-term mechanical hyperventilation of patients with FHF resulted in a reduced cerebral glutamine efflux, whereas the cerebral ammonia uptake remained unchanged (Figure 2). Ammonia Ammonia is a weak base with a pk a of in aqueous solution at 37 C. Thus, at physiologic ph, only a small fraction of the ammonia will exist as the diffusible gas that readily crosses the blood-brain barrier. According to the Henderson Hasselbalch equation, the difference in ph in the present study could account for only 10% of the measured arteriovenous difference of ammonia in FHF. In addition, in healthy subjects and cirrhotic patients we were unable to detect a cerebral ammonia uptake, despite the same differences in ph values. Thus, the cerebral ammonia uptake in patients with FHF cannot only be explained by the ph difference between the artery and internal jugular vein.

6 1114 STRAUSS ET AL. GASTROENTEROLOGY Vol. 121, No. 5 Table 3. Arterial Concentrations and CMRs of Amino Acids, Ammonia, and Amino Acid-N in Patients With FHF, Survivors (n 8) vs. Cerebral Herniation (n 6) Survivors Cerebral herniation Artery CMR Artery CMR Taurine Aspartate OH-proline Threonine Serine Asparagine a Glutamate Glutamine d a,d Proline d Glycine d a d Alanine d a d Citruline a Valine BCAA a,d Cystine a,d Methionine Isoleucine BCAA d Leucine BCAA d d Tyrosine AAA d b Phenylalanine AAA b Tryptophan AAA Histidine Ornithine Lysine Arginine Sum d a,d Ammonia d c a,d AA-N d a,d Fischer s ratio b NOTE. Results are presented as mean SD. Concentrations in mol/l and CMR in mol/(100 g min). Minus denotes cerebral release. For amino acid nitrogen, each amino acid was weighted with its number of nitrogen atoms. Fischer s ratio (Val Ile Leu)/(Tyr Phe). a P 0.05, b P 0.01, c P 0.001, vs. survivors. d P 0.05 vs. zero. In accordance with previous studies of patients with FHF, we found an increase in arterial plasma concentrations of ammonia. 5 There is substantial experimental evidence that ammonia is involved in the development of cerebral edema and intracranial hypertension in FHF. 20,21 Ammonia interferes with neuronal function and neurotransmission. 22 Furthermore, ammonia inhibits glutaminase, which is the enzyme that leads to the deamidation of glutamine. The activity of this enzyme is high in the neurons and is dependent on the phosphate concentration. 23 However, glutaminase is also located in the synaptosomes as well as in the astrocytes. 23 In the neurons, the primary inhibitor of glutaminase is glutamate, whereas ammonia only Figure 3. Boxplot showing the median, 10th, 25th, 75th, and 90th percentiles as vertical boxes with error bar of arterial ammonia concentration (P 0.002), net cerebral ammonia uptake (P 0.02), and net cerebral glutamine efflux (P 0.02) in survivors compared with nonsurvivors.

7 November 2001 CEREBRAL AMINO ACID FLUXES IN FHF 1115 slightly inhibits the activity of glutaminase. 23 Hogstad et al. 23 showed that exposure of glutaminase in neurons to 5 mmol/l ammonia only decreased the activity of glutaminase slightly, i.e., to 83% 6% (mean SE) of its maximal activity. Glutaminase in the synaptosomes is strongly inhibited by ammonia, whereas ammonia has no inhibitory effect on glutaminase activity in the astrocytes. 23 The activity of glutaminase in the astrocytes is low compared with its activity in the neurons, i.e., it requires much higher substrate concentration to obtain half maximal activity compared with glutaminase in the neurons. In the present study, we found a high cerebral glutamine efflux in patients with FHF. This is supported by experimental models showing that moderate hyperammonemia induced in rats by portacaval shunting or liver ischemia results in cerebral cortical uptake of ammonia, accumulation of glutamine, and a glutamine efflux. 24 In addition, infusion of ammonia into normal animals also results in glutamine accumulation, swelling of astrocytes, increased brain water, and intracranial hypertension. 20,21,25 27 We suggest that the major part of this cerebral glutamine efflux originates from the astrocytic compartment because glial swelling is the primary neuropathological feature in FHF patients 28 and in experimental models. 20 However, it is possible that the neuronal compartment, through the slight inhibition of glutaminase by ammonia, and the extracellular compartment may have contributed to the high cerebral glutamine efflux observed in the present study. The present study does not, however, allow for a conclusion on this point. In the present study, patients with the highest levels of ammonia seemed to have the highest risk of subsequent cerebral herniation. Ammonia per se does not result in cerebral edema. Administration of MSO to normal rats and rats with portacaval anastomosis receiving ammonia infusion decreases cerebral glutamine levels 27,29 and prevents the development of cerebral edema in normal rats, 25 decreases astrocytic swelling, 30 and ameliorates brain edema in portacaval anastomosis rats 20,27 despite persistently high ammonia concentrations. However, hyperammonemia seems to be a key event that contributes to the development of cerebral edema, as demonstrated by Rose et al., 31 because cerebral edema was prevented in rats with acute liver failure treated with L-ornithine-Laspartate, an agent that effectively reduces arterial ammonia levels Alterations in cerebral perfusion also seem to be important for the development of cerebral edema and intracranial hypertension. 34 Several studies have pointed towards high CBF in patients with clinical signs of cerebral edema and intracranial hypertension In patients with FHF without cerebral edema and intracranial hypertension, most studies have reported reduced values of CBF. 35,38 40 The latter is in accordance with CBF obtained in the present study in patients with FHF; this study was performed well before clinical signs of intracranial hypertension developed. Accordingly, this study is not in contrast to previous studies, and we have little doubt that increases in CBF play a major role for the development of cerebral edema and intracranial hypertension. A recent experimental study 29 supports this view; mild hypothermia decreased CBF and prevented formation of cerebral edema, despite a similar increase in brain glutamine as in the normothermic rats. Because ammonia alone does not affect CBF, 30 the increase in CBF is suggested to be mediated by a signal within the brain, initiated by an increased glutamine synthesis. In a recent experimental study, 41 brain water starts to increase 1 hour before cerebral hyperemia evolves. Amino Acids Because separation is difficult in whole blood, amino acids were measured in plasma, although the exchange between brain and blood of amino acids takes place from both compartments (plasma and erythrocytes). 42,43 The plasma and whole blood concentrations for most amino acids are the same in humans, 43,44 with the exception of whole blood/plasma concentration ratio of taurine, glutamate, and aspartate, which is reported to be 3 5 for taurine and glutamate 43,44 and 20 for aspartate. 44 Thus, metabolic rates of these amino acids should be interpreted with caution. In the present study, plasma concentrations of amino acids in healthy subjects are in accordance with whole blood concentrations of amino acids in humans as measured by Eriksson et al., with the exception of taurine, glutamate, and aspartate. 45 Calculation of cerebral fluxes on the basis of plasma concentration and whole blood flow requires complete equilibration between erythrocytes and plasma. In the present study, the blood samples were allowed to equilibrate for at least 5 minutes before centrifugation. Uptake of glutamine into the erythrocytes is resolved into 2 components: a saturable and a linear component (diffusion), giving the equation V V max S K m S K d S where V is the unidirectional influx ( mol/ml cell water/min), V max is the maximal velocity ( mol/ml cell water/min), K m the concentration at half-maximum velocity ( mol/l), [S] is the substrate concentration

8 1116 STRAUSS ET AL. GASTROENTEROLOGY Vol. 121, No. 5 ( mol/l), and K d is the rate constant for the linear component (min 1 ). According to Picó et al., 46 the kinetic constants of glutamine across the erythrocyte membrane at 37 C in starved rats are: V max, mol/l cell water/min; K m, 116 mol/l; and K d, min 1. If the mean venous concentration (2315 mol/l) from the present study is used in the above formula, the unidirectional influx (V) from plasma to the erythrocyte is mol/l/min, and because the arteriovenous difference of glutamine was 163 mol/l, complete equilibration between plasma and erythrocytes had occurred at the time of centrifugation in the present study. Another factor is the dry matter content in whole blood, which is 15%. Because hematocrit was lower in FHF than in healthy subjects, there will be a tendency towards an overestimation of the cerebral flux of glutamine in FHF. However, it cannot completely explain the observed difference in cerebral glutamine fluxes between groups in the present study. Infusion of glucose (17 g/h) to healthy subjects is followed by a decrease in the arterial concentrations of BCAAs, AAAs, and threonine, whereas the other amino acids remained unchanged. 47 Infusion of lower glucose concentrations (8.5 g/h) did not have any effect on the arterial concentrations. In our study, all 3 groups received glucose (17g/h). Thus, the effect of glucose on the arterial concentrations of the aforementioned amino acids is similar for all 3 groups. N-acetylcysteine was, however, only administered to patients with FHF, and may have caused the higher concentration of cystine in this group compared with healthy subjects and patients with cirrhosis. In accordance with previous studies of patients with FHF, we found an increase in almost all amino acids with the exception of BCAAs in patients with FHF ,48 Likewise, the arterial plasma concentrations of amino acids in the healthy controls from our study are similar to previous reports of healthy subjects Although there was no statistical difference in the arterial concentrations between survivors and nonsurvivors for most of the measured amino acids, they tended to be higher in patients who subsequently died of cerebral herniation (Table 3). However, the arterial concentration of the AAAs, tyrosine and phenylalanine, were significantly higher in nonsurvivors (Table 3). This may indicate differences in the brain synthesis of neurotransmitters in survivors compared with nonsurvivors. 52 The cerebral uptake of neutral amino acids across the blood-brain barrier is quite sensitive to competition. Thus, increased uptake of AAAs could be expected in FHF. This is supported by the finding of an increased extraction fraction of AAAs in experimental models, and it has been proposed that both ammonia and glutamine stimulate the cerebral uptake of AAAs. 53,54 However, Knudsen et al. 10 showed that this was caused by an increased retention of aromatic amino acids caused by a diminished back-flux out of the brain rather than increased uptake. The data in the present study support this view because we found no evidence of increased net cerebral uptake of AAAs, neither in nonsurvivors nor in survivors. Low arterial concentrations of BCAAs is a consistent finding in FHF, which could be the result of high turnover of these essential amino acids or the result of decreased oral intake. The low levels of BCAAs have been suggested to be involved in the development of HE, 19 and some studies have even shown that treatment with BCAAs could improve the stage of HE. 19,55 In the present study, we found an increased cerebral uptake of leucine and isoleucine, despite low arterial concentrations, which could be a compensatory cerebral mechanism for the increased cerebral glutamine efflux. However, the present data do not allow for conclusions regarding a possible influence of BCAAs on HE. Our finding of a net cerebral glutamine efflux in healthy controls is in accordance with previous human studies. 49,51 However, the sum of measured amino acids and the net cerebral nitrogen balance (AA-N) did not differ from zero (Table 2). In contrast, in cirrhotic patients and patients with FHF, we found a net cerebral nitrogen efflux that differed from zero (Table 2). In both these groups, the cerebral glutamine efflux caused a resulting negative amino acid-n (AA-N) balance. Patients with cirrhosis of the liver are usually malnourished. 56 It is therefore reasonable to assume that they have a negative cerebral AA-N balance caused by a combination of fasting and malnutrition. This is supported by Pell et al., 57 who showed a net cerebral glutamine efflux in fed and fasted sheep and that the cerebral nitrogen balance became negative in sheep after more than 3 days of fasting, despite an unchanged net uptake of BCAAs, suggesting that cerebral net protein degradation took place after prolonged fasting. Because glutamine is the major nitrogen carrier out of the brain, fasting will, according to Pell et al., 57 result in a net cerebral glutamine production and efflux. Contrary to patients with cirrhosis, patients with FHF were previously healthy persons, not malnourished before hospital admission. However, in patients with FHF, symptoms such as nausea and vomiting had been present for some days before appearance of stage 3 4 HE. Thus, part of the cerebral glutamine efflux/negative nitrogen balance in patients with FHF could be caused by fasting.

9 November 2001 CEREBRAL AMINO ACID FLUXES IN FHF 1117 Although the high cerebral ammonia uptake in patients with FHF probably contributed to the high cerebral glutamine efflux, this did not explain the total increase in glutamine efflux. Tracer studies with [13N]ammonia have shown that blood-borne ammonia is metabolized almost exclusively ( 95%) into the amide group of glutamine in the brains of normal rats, 58 as well as in brains of acutely (urease-treated) and chronically (12 14-week portacaval-shunted) hyperammonemic rats. 59 However, if the normal amount of cerebral nitrogen production incorporated into glutamine is set to 3.87 mol 100 g 1 min 1 (mean value of healthy subjects in the present study), and this value and the cerebral ammonia uptake is subtracted from the cerebral glutamine-n efflux, an efflux of mol 100 g 1 min 1 remains unexplained in the patients with FHF. Thus, factors other than an increased cerebral metabolism of ammonia may be responsible for the increased net cerebral glutamine efflux in FHF, such as a shift of cerebral glutamine pool or the toxic effect of hyperammonemia itself, which could affect the brain protein turnover. First, a change in the cerebral glutamine pool cannot be completely excluded, but this would not offer a complete explanation because we also found a significant efflux of threonine and tyrosine and the patients were in a clinical steady state condition during each measurement. Second, however, these results support an increased net cerebral protein breakdown. Subrahmanyam et al. 60 found that infusion of ammonium resulted in a decreased protein content in cerebral cortex, cerebellum, and brain stem of experimental animals, but did not determine whether the effect was caused by decreased protein synthesis or increased cerebral protein breakdown, or a combination of these, because no tracer studies were carried out. Furthermore, loss of glial fibrillary-acidic protein in cerebral gray matter has been shown in humans with HE by Sobel et al., 61 as well as in an experimental model. 62 Thus, a cerebral protein breakdown seems to occur in conditions with hyperammonemia. Hyperventilation results in an increase in ph. When taking the relative small change in ph between normoventilation and hyperventilation in the present study, together with an pka for glutamine on ,it is unlikely that this could have influenced the results to a greater extent. Acute institution of mechanical hyperventilation often reduces intracranial pressure mainly through a decrease in CBF. Furthermore, Ede et al. 63 showed that mechanical hyperventilation could postpone the development of cerebral edema and intracranial hypertension. In the present study, mechanical hyperventilation was only performed as a short-term intervention to evaluate the effect on amino acid metabolism. We found that mechanical hyperventilation changed the cerebral nitrogen balance towards normal, primarily because of a decreased cerebral glutamine efflux (Table 2) and secondarily an increased uptake of BCAAs, despite an unchanged cerebral ammonia uptake. This suggests that mechanical hyperventilation, through an induced alkalosis, may result in less net protein degradation and thereby a decreased non-ammonia cerebral glutamine efflux (Figure 2). Thus, studies examining the relationship between ph and protein metabolism have shown that alkalosis may increase the rate of protein synthesis. 64 Naturally, this study does not allow for any conclusion on this point. In conclusion, the results of the present study suggest that a high arterial ammonia level is associated with an increased risk of subsequent fatal intracranial hypertension. Our data indicate that the higher cerebral ammonia uptake in patients with FHF who subsequently died of cerebral herniation results in increased cerebral glutamine metabolism and edema formation. Because the cerebral glutamine efflux could not be explained solely by cerebral ammonia uptake, we suggest that ammonia itself induces cerebral protein breakdown. References 1. Raabe W. Effects of NH 4 on the function of the CNS. Adv Exp Med Biol 1990;272: Butterworth RF. Effects of hyperammonaemia on brain function. J Inherit Metab Dis 1998;21(suppl 1): Gjedde A, Lockwood AH, Duffy TE, Plum F. Cerebral blood flow and metabolism in chronically hyperammonemic rats: effect of an acute ammonia challenge. Ann Neurol 1978;3: Cooper AJ. Ammonia metabolism in normal and portacavalshunted rats. Adv Exp Med Biol 1990;272: 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: Cooper AJ, Lai JC. Cerebral ammonia metabolism in normal and hyperammonemic rats. Neurochem Pathol 1987;6: Norenberg MD, Martinez-Hernandez A. Fine structural localization of glutamine synthetase in astrocytes of rat brain. Brain Res 1979;161: Norenberg MD, Baker L, Norenberg LO, Blicharska J, Bruce-Gregorios JH, Neary JT. Ammonia-induced astrocyte swelling in primary culture. Neurochem Res 1991;16: Brusilow SW, Traystman RJ. Letter to the editor. N Engl J Med 1986;314: Knudsen GM, Schmidt J, Almdal T, Paulson OB, Vilstrup H. Passage of amino acids and glucose across the blood-brain barrier in patients with hepatic encephalopathy. Hepatology 1993;17: Record CO, Buxton B, Chase RA, Curzon G, Murray-Lyon IM, Williams R. Plasma and brain amino acids in fulminant hepatic failure and their relationship to hepatic encephalopathy. Eur J Clin Invest 1976;6:

10 1118 STRAUSS ET AL. GASTROENTEROLOGY Vol. 121, No Clemmesen JO, Kondrup J, Ott P. Splanchnic and leg exchange of amino acids and ammonia in acute liver failure. Gastroenterology 2000;118: Rosen HM, Yoshimura N, Hodgman JM, Fischer JE. Plasma amino acid patterns in hepatic encephalopathy of differing etiology. Gastroenterology 1977;72: Dejong CH, Deutz NE, Soeters PB. Cerebral cortex ammonia and glutamine metabolism in two rat models of chronic liver insufficiency-induced hyperammonemia: influence of pair feeding. J Neurochem 1993;60: Swain M, Butterworth RF, Blei AT. Ammonia and related amino acids in the pathogenesis of brain edema in acute ischemic liver failure in rats. Hepatology 1992;15: Williams R, Wendon J. Indications for orthotopic liver transplantation in fulminant liver failure. Hepatology 1994;20:S5 10S. 17. Strauss GI, Moller K, Holm S, Sperling BK, Knudsen GM, Larsen FS. Transcranial Doppler sonography and internal jugular bulb saturation during hyperventilation in patients with fulminant hepatic failure. Liver Transpl 2001;7: Conn HO. Studies on the origin and significance of blood ammonia. II. The distribution of ammonia in whole blood, plasma and erythrocytes of man. Yale J Biol Med 1966;39: Fischer JE, Rosen HM, Ebeid AM, James JH, Keane JM, Soeters PB. The effect of normalization of plasma amino acids on hepatic encephalopathy in man. Surgery 1976;80: Blei AT, Olafsson S, Therrien G, Butterworth RF. Ammonia-induced brain edema and intracranial hypertension in rats after portacaval anastomosis. Hepatology 1994;19: Ganz R, Swain M, Traber P, DalCanto M, Butterworth RF, Blei AT. Ammonia-induced swelling of rat cerebral cortical slices: implications for the pathogenesis of brain edema in acute hepatic failure. Metab Brain Dis 1989;4: Cooper AJ, Plum F. Biochemistry and physiology of brain ammonia. Physiol Rev 1987;67: Hogstad S, Svenneby G, Torgner IA, Kvamme E, Hertz L, Schousboe A. Glutaminase in neurons and astrocytes cultured from mouse brain: kinetic properties and effects of phosphate, glutamate, and ammonia. Neurochem Res 1988;13: Dejong CH, Kampman MT, Deutz NE, Soeters PB. Cerebral cortex ammonia and glutamine metabolism during liver insufficiencyinduced hyperammonemia in the rat. J Neurochem 1992;59: 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 H Vogels BA, van Steynen B, Maas MA, Jorning GG, Chamuleau RA. The effects of ammonia and portal-systemic shunting on brain metabolism, neurotransmission and intracranial hypertension in hyperammonaemia-induced encephalopathy. J Hepatol 1997;26: 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: Kato M, Hughes RD, Keays RT, Williams R. Electron microscopic study of brain capillaries in cerebral edema from fulminant hepatic failure. Hepatology 1992;15: Cordoba J, Crespin J, Gottstein J, Blei AT. Mild hypothermia modifies ammonia-induced brain edema in rats after portacaval anastomosis. Gastroenterology 1999;116: Willard-Mack CL, Koehler RC, Hirata T, et al. Inhibition of glutamine synthetase reduces ammonia-induced astrocyte swelling in rat. Neuroscience 1996;71: Rose C, Michalak A, Rao KV, Quack G, Kircheis G, Butterworth RF. L-ornithine-L-aspartate lowers plasma and cerebrospinal fluid ammonia and prevents brain edema in rats with acute liver failure. Hepatology 1999;30: Staedt U, Leweling H, Gladisch R, Kortsik C, Hagmuller E, Holm E. Effects of ornithine aspartate on plasma ammonia and plasma amino acids in patients with cirrhosis. A double-blind, randomized study using a four-fold crossover design. J Hepatol 1993; 19: Kircheis G, Nilius R, Held C, Berndt H, Buchner M, Gortelmeyer R, Hendricks R, Kruger B, Kuklinski B, Meister H, Otto HJ, Rink C, Rosch W, Stauch S. Therapeutic efficacy of L-ornithine-L-aspartate infusions in patients with cirrhosis and hepatic encephalopathy: results of a placebo-controlled, double-blind study. Hepatology 1997;25: Larsen FS. Cerebral circulation in liver failure: Ohm s law in force. Semin Liver Dis 1996;16: Wendon JA, Harrison PM, Keays R, Williams R. Cerebral blood flow and metabolism in fulminant liver failure. Hepatology 1994; 19: Larsen FS, Pott F, Hansen BA, Ejlersen E, Knudsen GM, Clemmesen JO, Secher NH. Transcranial Doppler sonography may predict brain death in patients with fulminant hepatic failure. Transplant Proc 1995;27: Jalan R, Damink SW, Deutz NE, Lee A, Hayes PC. Moderate hypothermia for uncontrolled intracranial hypertension in acute liver failure. Lancet 1999;354: Larsen FS, Hansen BA, Ejlersen E, Secher NH, Clemmesen JO, Tyqstuip N, Knudsen GM. Cerebral blood flow, oxygen metabolism and transcranial Doppler sonography during high-volume plasmapheresis in fulminant hepatic failure. Eur J Gastroenterol Hepatol 1996;8: 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: 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: Larsen FS, Gottstein J, Blei AT. Cerebral hyperemia and nitric oxide synthase in rats with ammonia-induced brain edema. J Hepatol 2001;34: Drewes LR, Conway WP, Gilboe DD. Net amino acid transport between plasma and erythrocytes and perfused dog brain. Am J Physiol 1977;233:E320 E Felig P, Wahren J, Raf L. Evidence of inter-organ amino-acid transport by blood cells in humans. Proc Natl Acad Sci USA 1973;70: Hagenfeldt L, Arvidsson A. The distribution of amino acids between plasma and erythrocytes. Clinica Chimica Acta 1980;100: Eriksson LS, Law DH, Hagenfeldt L, Wahren J. Nitrogen metabolism of the human brain. J Neurochem 1983;41: Pico C, Pons A, Palou A. Regulation of rat erythrocyte L-glutamine, L-glutamate and L-lysine uptake by short-term starvation. Int J Biochem 1992;24: Elia M, Neale G, Livesey G. Alanine and glutamine release from the human forearm: effects of glucose administration. Clin Sci 1985;69: Fischer JE. Amino acids in hepatic coma. Dig Dis Sci 1982;27: Grill V, Bjorkman O, Gutniak M, Lindqvist M. Brain uptake and release of amino acids in nondiabetic and insulin-dependent diabetic subjects: important role of glutamine release for nitrogen balance. Metabolism 1992;41: Felig P, Wahren J, Ahlborg G. Uptake of individual amino acids by the human brain. Proc Soc Exp Biol Med 1973;142: Lying-Tunell U, Lindblad BS, Malmlund HO, Persson B. Cerebral blood flow and metabolic rate of oxygen, glucose, lactate, pyruvate, ketone bodies and amino acids. Acta Neurol Scand 1980; 62: Michalak A, Butterworth RF. Selective increases of extracellular

11 November 2001 CEREBRAL AMINO ACID FLUXES IN FHF 1119 brain concentrations of aromatic and branched-chain amino acids in relation to deterioration of neurological status in acute (ischemic) liver failure. Metab Brain Dis 1997;12: Hilgier W, Puka M, Albrecht J. Characteristics of large neutral amino acid-induced release of preloaded L-glutamine from rat cerebral capillaries in vitro: effects of ammonia, hepatic encephalopathy, and gamma-glutamyl transpeptidase inhibitors. J Neurosci Res 1992;32: James JH, Ziparo V, Jeppsson B, Fischer JE. Hyperammonaemia, plasma aminoacid imbalance, and blood-brain aminoacid transport: a unified theory of portal-systemic encephalopathy. Lancet 1979;2: Beaubernard C, Delorme ML, Opolon P, Boschat M, Morin J, Oryszcyn MP, Franco D. Effect of the oral administration of branched chain amino acids on hepatic encephalopathy in the rat. Hepatology 1984;4: Hamberg O. Regulation of urea synthesis by diet protein and carbohydrate in normal man and in patients with cirrhosis. Relationship to glucagon and insulin. Dan Med Bull 1997;44: Pell JM, Bergman EN. Cerebral metabolism of amino acids and glucose in fed and fasted sheep. Am J Physiol 1983;244:E282 E Cooper AJ, McDonald JM, Gelbard AS, Gledhill RF, Duffy TE. The metabolic fate of 13N-labeled ammonia in rat brain. J Biol Chem 1979;254: Cooper AJ, Mora SN, Cruz NF, Gelbard AS. Cerebral ammonia metabolism in hyperammonemic rats. J Neurochem 1985;44: Subrahmanyam K, Prasad MS, Rangavalli G, Muralidhar K, Sadasivudu B. Functional relationship of ammonia to DNA, RNA and protein in brain. Neuroscience 1985;15: Sobel RA, DeArmond SJ, Forno LS, Eng LF. Glial fibrillary acidic protein in hepatic encephalopathy. An immunohistochemical study. J Neuropathol Exp Neurol 1981;40: Neary JT, Whittemore SR, Zhu Q, Norenberg MD. Destabilization of glial fibrillary acidic protein mrna in astrocytes by ammonia and protection by extracellular ATP. J Neurochem 1994;63: Ede RJ, Gimson AE, Bihari D, Williams R. Controlled hyperventilation in the prevention of cerebral oedema in fulminant hepatic failure. J Hepatol 1986;2: Fuller SJ, Gaitanaki J, Sugden PH. Effects of increasing extracellular ph on protein synthesis and protein degradation in the perfused working heart. Biochem J 1989;259: Received March 29, Accepted July 26, Address requests for reprints to: Gitte I. Strauss, M.D., Department of Hepatology A-2121, Rigshospitalet, University of Copenhagen, Blegdamsvej 9, DK-2100 Copenhagen, Denmark. gstrauss@nru.dk; fax: (45) Supported by Rigshospitalet; University of Copenhagen; The Laerdal Foundation for acute medicine; Savvaerksejer Jeppe Juhl and wife Ovita Juhls Foundation; The Novo Nordisk Foundation; The King Christian the 10th Foundation; The AP-Møller Foundation; The Danish Hospital Foundation for Medical Research, Region of Copenhagen, The Faroe Islands, and Greenland; The Danish Medical Association Research Fund; and The Beckett Foundation. The authors thank Gerda Thomsen, Karin Stahr, Mie Poulsen, Nine Scherling, and Liselotte Hansen for skillful technical assistance.