Potential conflict of interest: Nothing to report.
Supported by the Danish Council for Independent Research, Medical Sciences (09-073658 and 09-065565), the Aase and Ejnar Danielsen's Foundation, and the A. P. Møller Foundation for the Advancement of Medical Science.
Studies have shown decreased cerebral oxygen metabolism (CMRO2) and blood flow (CBF) in patients with cirrhosis with hepatic encephalopathy (HE). It remains unclear, however, whether these disturbances are associated with HE or with cirrhosis itself and how they may relate to arterial blood ammonia concentration and cerebral metabolic rate of blood ammonia (CMRA). We addressed these questions in a paired study design by investigating patients with cirrhosis during and after recovery from an acute episode of HE type C. CMRO2, CBF, and CMRA were measured by dynamic positron emission tomography (PET)/computed tomography (CT). Ten patients with cirrhosis were studied during an acute episode of HE; nine were reexamined after recovery. Nine patients with cirrhosis with no history of HE served as controls. Mean CMRO2 increased from 0.73 μmol oxygen/mL brain tissue/min during HE to 0.91 μmol oxygen/mL brain tissue/min after recovery (paired t test; P < 0.05). Mean CBF increased from 0.28 mL blood/mL brain tissue/min during HE to 0.38 mL blood/mL brain tissue/min after recovery (P < 0.05). After recovery from HE, CMRO2 and CBF were not significantly different from values in the control patients. Arterial blood ammonia concentration decreased 20% after recovery (P < 0.05) and CMRA was unchanged (P > 0.30); both values were higher than in the control patients (both P < 0.05). Conclusion: The low values of CMRO2 and CBF observed during HE increased after recovery from HE and were thus associated with HE rather than the liver disease as such. The changes in CMRO2 and CBF could not be linked to blood ammonia concentration or CMRA. (HEPATOLOGY 2013)
Brain energy metabolism is believed to be disturbed in patients with cirrhosis with hepatic encephalopathy (HE).1 Studies have shown decreased cerebral oxygen consumption (CMRO2) and blood flow (CBF) in patients with cirrhosis and HE when compared to patients with cirrhosis without HE or healthy subjects, whereas the latter two groups of subjects had similar values.2-5 Cross-sectional studies point to the reductions being related to the HE condition and not to the liver disease itself.4, 5 In accordance with this, one study reported that reductions in CMRO2 and CBF were restored when patients, treated with a dopamine agonist, recovered from HE.3 Another study, however, found decreased CMRO2 but unchanged CBF in patients with a history of HE when compared to patients without a history of HE or healthy controls.6 It thus remains an open question whether the reductions in CMRO2 and CBF in patients with cirrhosis and HE normalize after recovery from HE, and in the present study we therefore measured CMRO2 and CBF during and after recovery from an episode of HE in individual patients with cirrhosis.
Another key question is how changes in CMRO2 and CBF may relate to the blood concentration of ammonia. Patients with cirrhosis with HE usually have higher blood ammonia concentrations than patients with cirrhosis without HE and ammonia has been shown to be neurotoxic in both in vitro studies and experimental animal models of HE.7-9 In humans, cerebral metabolic rate of blood ammonia (CMRA) seems to be primarily determined by the arterial blood concentration of ammonia10, 11; in a cross-sectional positron emission tomography (PET) study of patients with cirrhosis with HE, patients with cirrhosis with no history of HE and healthy subjects, CMRO2 and CBF correlated negatively to the arterial blood ammonia concentration.5 The relationship between changes in blood ammonia levels and CMRO2 and CBF has, to the best of our knowledge, not been investigated in the same individuals during and after recovery from an episode of HE.
To test whether the reductions in CMRO2 and CBF during HE are related to HE and not the liver disease itself and the relationship to CMRA, we employed a paired study design with PET and computed tomography (CT) measurements of CMRO2, CBF, and CMRA in patients with cirrhosis during and after recovery from an acute episode of HE type C. We also included patients with cirrhosis with no history of HE because this enabled us to resolve whether changes in the measured parameters are predominantly associated with the HE condition or the cirrhosis.
Ten patients with cirrhosis of the liver (alcohol, n = 9; primary biliary cirrhosis, n = 1) with an acute episode of clinically overt HE type C were included (Table 1). The severity of HE was graded according to the West Haven criteria.12 All patients with HE received standard supportive medical therapy and no specific treatment for HE except for lactulose and they were hemodynamically stable and did not require any respiratory or circulatory support. As control patients, nine patients with cirrhosis (alcohol, n = 8; nonalcoholic steatohepatitis, n = 1) with no history and no current signs of HE were included.
Table 1. Patient Characteristics
Cirrhotic Patients During An Episode of HE Type C
Same Patients After Recovery from HE
Cirrhotic Patients with No History of HE
Continuous values are given as mean ± standard error of the mean or median (range).
HE, hepatic encephalopathy; CRT, continuous reaction time (13); MAP, mean arterial pressure.
One patient died before recovery.
CRT could not be performed in three patients with HE grade III.
P < 0.05 when compared to cirrhotic patients with no history of HE (Student's t test).
P < 0.05 when compared to after recovery from HE (paired t test).
All patients underwent a clinical examination in the morning before the PET/CT studies. This included a continuous reaction time test (CRT)13 to ensure that patients examined after recovery from HE and the control patients did not have minimal HE; the CRT could not be performed in three of the patients with HE grade III.
The study was approved by the Central Denmark Region Committees on Biomedical Research Ethics and performed in accordance with the Helsinki II Declaration. All patients or next of kin gave written informed consent before participating. The average radiation dose received by the participants was 3 mSv per study day. No complications to the procedures were observed.
The studies started at 8 AM and patients were instructed not to take any food for at least 8 hours before the studies but were free to drink water. An artflon catheter (Becton Dickinson, Swindon, UK) was placed percutaneously in a radial artery for blood sampling and a venflon catheter (Becton Dickinson) was placed percutaneously in a cubital vein for intravenous administration of 15O-water and 13N-ammonia. The subjects were placed on their back in the scanner (Siemens Biograph 64 Truepoint PET/CT) and positioned with the head in a neck shield to minimize head movements. A topogram was performed for optimal positioning of the head within the 21.6 cm field-of-view of the PET camera and a low-dose CT scan was performed for anatomical definition and attenuation correction of PET emission data. The study design included three dynamic PET studies each study day. First, 1,000 MBq 15O-oxygen was administered through a mouthpiece (Bennett, Pleasanton, CA) during the initial 5 seconds of a 3-minute scan. In 19 of 24 measurements, duplicate 15O-oxygen PET/CT scans were performed; CMRO2 values from same study day deviated less than 5% and therefore individual mean values were used. Second, 500 MBq 15O-water was administered intravenously during the initial 15 seconds of a 3-minute scan. Because duplicate CBF measurements deviated less than 5% in a previous study5 only one 15O-water PET/CT scan was performed each study day. Third, 700 MBq 13N-ammonia was administered intravenously during the initial 15 seconds of a 30-minute scan. PET data were reconstructed using TrueX 3D OSEM (three iterations, 21 subsets), a 256 × 256 × 109 matrix (zoom 3), and a 3-mm Gauss filter and were reconstructed using a time-frame structure of 12 × 5, 6 × 10, 3 × 20 seconds (total 21 frames) for 15O-oxygen and 15O-water, and 18 × 5, 9 x10, 2 × 30, 3 × 60, 4 × 120, 3 × 300 seconds (total 39 frames) for 13N-ammonia.
Arterial blood pH, partial pressure of carbon dioxide (pCO2), partial pressure of oxygen (pO2), O2 saturation, and hemoglobin concentration were measured before each PET recording and after the last one (ABL-555, Radiometer, Copenhagen, Denmark). No systematic changes were observed during each study day and individual mean values of the measurements were used. The arterial plasma concentration of ammonia was measured in two blood samples collected at the beginning of the 13N-ammonia PET study14 and whole-blood ammonia concentration was calculated.15 The ammonia concentration measurements from the same study day deviated less than 5% and therefore individual mean values were used.
During the 15O-water and 15O-oxygen PET recordings, radioactivity concentrations in arterial blood were measured every 0.5 seconds by an automated blood sampling system (Allog, Mariefred, Sweden) that was cross-calibrated with the PET camera. Measurements were corrected for delay and dispersion in the sampling catheter16 and for radioactive decay back to the start of the scan.
During the 13N-ammonia PET recording, 24 arterial 1-mL blood samples were collected manually at 12 × 5, 3 × 10, 1 × 30, 1 × 60, 2 × 120, 1 × 180, and 4 × 300 seconds. Blood radioactivity concentrations were measured using a well counter (Packard Instruments, Meriden, CT) that was cross-calibrated with the PET camera and concentrations were corrected for radioactive decay back to the start of the scan. Additional arterial blood samples of 2 mL were collected at 1, 2, 3, 5, 7, 10, 15, 20, 25, and 30 minutes after 13N-ammonia administration for determination of blood concentrations of 13N-ammonia, 13N-urea, and 13N-glutamine.17
Analysis of PET Data.
Analysis of the 15O-water PET data was performed using a standard clinical method implemented at our institution; each PET recording was registered to a PET template in the Talairach space.18 A parametric image of CBF (mL blood/mL brain tissue/min) was generated by fitting a standard one-tissue compartmental model to the dynamic PET recording and the time-course of arterial blood radioactivity concentration using an extraction fraction for 15O-water of 0.9.19 Figure 1 shows examples of parametric images of CBF in a patient during and after HE and a control patient.
For analysis of the 15O-oxygen PET data, a whole-brain volume of interest (VOI) was manually drawn using the combined PET/CT images and excluding skull tissue and ventricles; a brain tissue radioactivity concentration time-course was generated (kBq/mL tissue versus time in minutes). Whole-brain CMRO2 (μmol oxygen/mL brain tissue/min) was calculated by fitting a standard one-tissue compartmental model of oxygen metabolism to the time-courses of radioactivity concentrations in brain tissue and arterial blood radioactivity concentrations. Note that CMRO2 and CBF were determined independently of each other.
For analysis of the 13N-ammonia PET data, a brain tissue radioactivity concentration time-course was generated using the same VOI as for the15O-oxygen data. Kinetic parameters of cerebral 13N-ammonia uptake and metabolism were calculated by nonlinear fitting of a two-tissue compartmental model of 13N-ammonia uptake and metabolism to the 13N-radioactivity concentration time-courses in brain tissue and arterial blood.11 The model includes clearance of 13N-ammonia from blood to brain tissue (K1; mL blood/mL brain tissue/min), backflux of 13N-ammonia from brain tissue to blood (k2; per minute), and conversion of 13N-ammonia to 13N-glutamine in brain (k3; per minute). 13N-urea formed in the liver was assumed to enter and leave brain tissue by free diffusion and 13N-glutamine in blood was assumed not to enter the brain tissue in any significant amount.
The permeability surface-area product for ammonia across the blood brain-barrier (PSBBB; mL blood/mL brain tissue/min) was calculated as:
where K1/CBF equals the first-passage extraction fraction of ammonia. The metabolic clearance of blood-borne 13N-ammonia (Kmet; mL blood/mL brain tissue/min) into intracellular 13N-glutamine during steady state was calculated as:
where K1, k2, and k3 are the kinetic parameters of 13N-ammonia as described above. Finally, CMRA (μmol ammonia/L brain tissue/min) was calculated as:
where A is the mean arterial concentration of nonradioactive ammonia (μmol/L blood).
All data are presented as means ± SEM. Data were tested to be normally distributed before using the two-tailed paired t test for comparison of values within individuals during and after recovery from HE and the unpaired two-tailed t test for comparison between groups. P < 0.05 was considered statistically significant.
Clinical characteristics are summarized in Table 1. The prothrombin index and plasma concentrations of bilirubin and albumin did not change when patients recovered from HE. Prothombin index and bilirubin were significantly lower and higher, respectively, in patients with cirrhosis than the mean values in control patients (both P < 0.05). Patients with HE had a higher Child-Pugh score than control patients, but this was not the case after recovery from HE. Hemoglobin was lower in patients with HE both during and after HE when compared to control patients (P < 0.05). Arterial blood ammonia was higher during HE than after recovery (P < 0.05); both values were higher than in control patients (both P < 0.05) (Fig. 2). C-reactive protein was not significantly higher during HE when compared to after recovery and in both states the values were similar to that in control patients (all P > 0.15). This rules out severe inflammation as a cause of HE. Arterial pH was not different during HE when compared to after recovery, nor when compared to control patients, whereas pCO2 was lower during HE (P < 0.05), reflecting respiratory alkalosis due to hyperventilation. The mean arterial blood pressure (MAP) did not change after recovery from HE and was not different from the control patients (P > 0.30). Plasma sodium was significantly lower during HE than after recovery from HE (P < 0.05) and when compared to control patients (P < 0.05), whereas there was no significant difference between patients after recovery from HE and control patients (P > 0.30).
Eight patients with HE, seven after recovery from HE, and nine control patients underwent 15O-oxygen PET/CT (Fig. 3A). During HE whole-brain CMRO2 was 0.73 ± 0.08 μmol oxygen/mL brain tissue/min and it increased to 0.91 ± 0.08 μmol oxygen/mL brain tissue/min after recovery (P < 0.05). Mean CMRO2 after recovery was not different from the mean value of 1.10 ± 0.07 μmol oxygen/mL brain tissue/min in the control patients (P = 0.10). This finding is in agreement with previous findings of similar mean values in six patients with cirrhosis with no HE, being 1.34 ± 0.08 μmol oxygen/mL brain tissue/min, and in seven healthy subjects, being 1.35 ± 0.05 μmol oxygen/mL brain tissue/min (P > 0.30).5 The reason for the present mean values being slightly lower is probably due to the use of a different PET camera and improved reconstruction algorithms.
Ten patients with HE, eight after recovery from HE, and nine control patients underwent 15O-water PET/CT (Fig. 3B). During HE whole-brain CBF was 0.28 ± 0.09 mL blood/mL brain tissue/min and it increased to 0.38 ± 0.13 mL blood/mL brain tissue/min after recovery (P < 0.05). Mean CBF after recovery was not different from the mean value of 0.43 ± 0.02 mL blood/mL brain tissue/min in the control patients (P = 0.28). This finding is in agreement with previous findings of similar mean values in six patients with cirrhosis with no HE, being 0.47 ± 0.02 mL blood/mL brain tissue/min, and in seven healthy subjects, being 0.49 ± 0.03 mL blood/mL brain tissue/min (P > 0.30).5
A plot of individual values of CMRO2 against CBF values (Fig. 3C) demonstrates a positive correlation between the two parameters in the seven subjects with paired data.
Distribution of 13N-Ammonia Across the Blood Brain-Barrier and PSBBB.
Nine patients with HE, seven after recovery from HE, and eight control patients underwent 13N-ammonia PET/CT. K1 for 13N-ammonia was 0.19 ± 0.02 mL blood/mL brain tissue/min during HE and 0.19 ± 0.01 mL blood/mL brain tissue/min after recovery from HE (P > 0.30), similar to that of 0.18 ± 0.01 mL blood/mL brain tissue/min in the control patients (P > 0.30).
Data for calculation of the PSBBB for ammonia were available for nine patients with HE, seven after recovery from HE (six paired datasets), and eight control patients (Fig. 3D). The PSBBB was 0.27 ± 0.04 mL blood/mL brain tissue/min during HE and 0.25 ± 0.03 mL blood/mL brain tissue/min after recovery from HE (P > 0.30), none of which were significantly different than that of 0.23 ± 0.01 mL blood/mL brain tissue/min in the control patients (P > 0.30).
Cerebral Ammonia Metabolism.
Data for calculation of Kmet for 13N-ammonia were available for nine patients with HE, seven after recovery from HE (seven paired datasets), and six control patients. Mean Kmet was 0.15 ± 0.02 mL blood/mL brain tissue/min during HE and 0.18 ± 0.01 mL blood/mL brain tissue/min after recovery (P > 0.30). In the control patients, Kmet was 0.16 ± 0.01 mL blood/mL brain tissue/min, which was not different from the values both during and after recovery from HE (both P > 0.30).
Data for calculation of CMRA were available for nine patients with HE, seven patients after recovery from HE (seven paired datasets), and eight control patients (Fig. 3E). CMRA was 22.4 ± 4.0 μmol ammonia/L brain tissue/min during HE and 21.5 ± 1.9 μmol ammonia/L brain tissue/min after recovery (P > 0.30). In the control patients, CMRA was 14.6 ± 0.8 μmol ammonia/L brain tissue/min, which was lower than in patients both during and after recovery from HE (both P < 0.05). Figure 3F shows individual values of CMRA plotted against arterial blood ammonia.
The finding that both CMRO2 and CBF returned to control values when the patients recovered from HE shows that the reductions in CMRO2 and CBF during HE were related to the HE condition and not to the liver disease itself. It has been suggested that the reduced CMRO2 in patients with HE is not caused by limitation of the oxygen supply but reflects an actual decline in oxidative metabolism with a secondary reduction in CBF20 and the present finding of a positive relationship between CMRO2 and CBF during and after recovery from HE within individual patients (Fig. 3C) agrees with this hypothesis.
CMRA did not decrease after recovery from HE in spite of a significant decrease in arterial ammonia concentration. Furthermore, mean CMRA remained higher after recovery than in the control patients, as did the mean arterial blood ammonia concentration. The similar metabolic clearances of 13N-ammonia during and after HE with no significant difference from those observed in the control patients indicate that the kinetics of brain metabolism of blood ammonia is unaffected during HE. This is in agreement with a previous study, which also showed that patients with cirrhosis with no history of HE have similar kinetics when compared to healthy controls.11 These findings support that the arterial blood ammonia concentration is the primary determinant of CMRA, in agreement with previous cross-sectional studies,10, 11 although the relationship within individuals is complex (Fig. 3F).
Based on in vitro studies1, 8 and cross-sectional studies in patients with cirrhosis and healthy subjects,10, 11 we raised the hypothesis of a negative relationship between CMRO2 and cerebral ammonia uptake. This hypothesis was refuted, as there was an increase in CMRO2 after recovery from HE but no significant change in CMRA (Fig. 4A). Furthermore, neither CMRO2 nor CBF was related to the arterial ammonia concentration or CMRA within individuals (Fig. 4B,C). The reduction in CMRO2 could thus not be linked directly to increased CMRA but our study cannot exclude ammonia as a potential etiological factor for the development of HE.
15O-oxygen PET measures the actual oxygen consumption, but whether the decreased CMRO2 observed in patients during HE is a result of interference with oxidative metabolism by substrates such as ammonia is difficult to investigate in vivo. Cerebral ammonia metabolism may interfere with energy metabolism in several ways.1In vitro studies suggested that ammonia could inhibit tricarboxylic acid (TCA) cycle activity24 but recent studies in cell cultures and animal models of HE did not find such inhibition.25, 26 In agreement with this, a human PET study found no correlation between CMRA and brain glucose utilization, although a correlation between ammonia uptake and changes in brain metabolites on magnetic resonance spectroscopy was observed, indicating some effects of ammonia in vivo.27 Cerebral metabolites such as lactate have been found to be increased in animal models8, 21 and in cerebrospinal fluid from patients with HE,22 which may indicate increased glycolysis, but whether increased lactate is a cause or consequence of HE is uncertain.23 Thus, although we were unable to link the decrease in CMRO2 to an increase in CMRA, we cannot rule out that ammonia may be involved in the pathogenesis of HE, perhaps through subtle effects on energy metabolism. This is supported by the higher blood levels of ammonia and CMRA in the patients after recovery from HE compared to the control patients. It could also explain why metabolites from ammonia metabolism, such as glutamine and alpha-ketoglutaramate, measured in cerebrospinal fluid from patients with HE seem to correlate better with the severity of HE than blood ammonia.28
The clinical distinction between hyponatremic encephalopathy and hepatic encephalopathy is a challenge.29 Recent studies have indicated hyponatremia as an independent risk factor for patients with cirrhosis for developing HE.30, 31 In the present study plasma sodium was reduced in all patients during HE and increased significantly when the patients recovered from HE; there was no difference of mean sodium in patients after recovery from HE and the control patients, suggesting that hyponatremia indeed was related to the HE condition more than the liver disease itself. Two patients had severe hyponatremia (<114 μmol/L plasma) during HE and both suffered from HE grade III. We could not, however, link the hyponatremia to changes in CMRO2, CBF, or CMRA but this does not rule out a potential role of hyponatremia in the development of HE.
In conclusion, the decreased CMRO2 and CBF in the patients with cirrhosis with HE were associated with the HE condition and not the liver disease itself. The decreased CMRO2 is in agreement with decreased oxidative metabolism during HE with a secondary reduction in CBF, but the reduction in CMRO2 could not be linked to the blood ammonia concentration or CMRA. The present results thus do not support a direct toxic effect of hyperammonemia on brain oxidative metabolism but still does not preclude other or indirect roles of ammonia in HE.