Value of the critical flicker frequency in patients with minimal hepatic encephalopathy†
Article first published online: 28 MAR 2007
Copyright © 2007 American Association for the Study of Liver Diseases
Volume 45, Issue 4, pages 879–885, April 2007
How to Cite
Romero-Gómez, M., Córdoba, J., Jover, R., del Olmo, J. A., Ramírez, M., Rey, R., de Madaria, E., Montoliu, C., Nuñez, D., Flavia, M., Compañy, L., Rodrigo, J. M. and Felipo, V. (2007), Value of the critical flicker frequency in patients with minimal hepatic encephalopathy. Hepatology, 45: 879–885. doi: 10.1002/hep.21586
Potential conflict of interest: Nothing to report.
- Issue published online: 28 MAR 2007
- Article first published online: 28 MAR 2007
- Manuscript Accepted: 13 DEC 2006
- Manuscript Received: 8 JUN 2006
- Spanish Ministry of Health to the Spanish Network of Hepatic Encephalopathy Research. Grant Number: 03/155-2002
Minimal hepatic encephalopathy (MHE) is mainly diagnosed using psychometric tests such as the psychometric hepatic encephalopathy score (PHES). Despite the clinical and social relevance of MHE, psychometric testing is not widespread in routine clinical care. We assessed the usefulness of the critical flicker frequency (CFF), for the diagnosis of MHE and for the prediction of the development of overt episodes of HE. The normal range of PHES in the Spanish population was evaluated in a control group. Subsequently, 114 patients with cirrhosis and 103 healthy controls underwent both PHES and CFF tests. A diagnosis of MHE was made when the PHES was lower than −4 points. Patients were followed-up every 6 months for a total of 1 year. CFF did not correlate with age, education, or sex in the control group. The mean CFF was significantly lower in patients with MHE versus non-MHE or controls. Mean CFF correlated with individual psychometric tests as well as PHES (r = 0.54; P < 0.001). CFF <38 Hz was predictive of further bouts of overt HE (log-rank: 14.2; P < 0.001). There was a weak correlation between mean CFF and Child-Pugh score but not with model for end-stage liver disease score. In multivariate analysis using Cox regression, CFF together with Child-Pugh score was independently associated with the development of overt HE. Conclusion: CFF is a simple, reliable, and accurate method for the diagnosis of MHE. It is not influenced by age or education and could predict the development of overt HE. (HEPATOLOGY 2007;45:879–885.)
Hepatic encephalopathy (HE) is a major complication of cirrhosis and is associated with a poor prognosis.1, 2 Minimal hepatic encephalopathy (MHE) is the first stage in the clinical spectrum of HE.3, 4 MHE is associated with an impaired quality of life,5 and patients become unfit to safely drive a motor vehicle.6 Moreover, patients with MHE and altered oral glutamine challenge have a shortened life span.7 Although no definitive treatment is currently available, evaluating MHE in patients with cirrhosis could be recommended because of its prognostic value. Nevertheless, in a recent Spanish questionnaire, only 41% of hepatologists test for MHE in routine clinical practice.8 The major obstacles to the evaluation of MHE include: (1) a lack of consensus diagnostic criteria; (2) the fact that most psychometric tests used in the diagnosis of MHE are time-consuming; (3) a lack of normal distribution values corrected for educational level and age; (4) the high cost of neurophysiological assessment; and (5) the need for experienced personnel and specialized equipment.
The psychometric hepatic encephalopathy score (PHES) includes a battery of 5 psychometric tests that have been found useful for the diagnosis of MHE. In a recent consensus meeting,9 the PHES was recommended as the gold standard in the diagnosis of MHE because: (1) it covered the spectrum of cognitive aspects that are affected in HE; (2) normative age-corrected data are available; and (3) it is inexpensive. The neurophysiological critical flicker frequency (CFF) analysis is used to test cerebral cortex function and has been applied for the study of several neurological disorders, including multiple sclerosis10 and Alzheimer's disease.11 CFF analysis has been found to be sensitive and objective in the quantification of low-grade hepatic encephalopathy,12 but no data are available regarding the usefulness of this tool in the management of patients with cirrhosis.
The aims of the present study were: (1) to assess the usefulness of PHES as a diagnostic tool for MHE in a population of Spanish patients with cirrhosis; (2) to evaluate CFF as a tool for the assessment of MHE compared with PHES; and (3) to determine the usefulness of PHES and CFF in the prediction of overt HE and of survival.
Patients and Methods
Suitable patients were invited to participate. Written informed consent was obtained from each patient, and the local ethics committee approved the study protocol. Patients with cirrhosis being managed in the outpatient clinics of four Spanish hospitals between November 1, 2004, and April 1, 2005, were considered eligible for recruitment into the study. Cirrhosis was diagnosed via liver biopsy or the presence of biochemical, ultrasonographic, or endoscopic features of portal hypertension and/or liver dysfunction. The study comprised 114 consecutive patients with cirrhosis, including 32 females (28.1%) and 82 males (71.9%). The mean age of the overall study group was 57 ± 10 years. The etiology of cirrhosis was alcohol in 48 patients (42.1%), HCV in 48 (42.1%), HBV in 4 (3.5%), autoimmune cirrhosis in 5 (4.4%), and cryptogenic cirrhosis in 9 (7.9%). There were 57 patients (50.9%) in Child-Pugh class A, 36 (30.7%) in class B, and 21 (18.4%) in class C. The average model for end-stage liver disease score was 11.7 ± 5.7. All patients showed normal neurological signs on examination: alert, without flapping, ataxia, or dysarthria. No patient had been prescribed lactulose, lactitol, or neomycin in the previous 3 months and none was receiving antibiotic or sedatives in the week prior to the study. Patients with alcohol-induced liver disease had to be abstinent for at least 3 months prior to the start of the study. Patients with biochemical evidence of renal disease, respiratory disease, cardiac failure, or severe malnutrition were excluded. Ninety-six patients underwent routine upper gastrointestinal endoscopy to detect esophageal varices. Esophageal varices were absent in 28 cases (29.2%), small varices were detected in 30 (31.3%), medium-sized varices in 24 (25%), and large varices in the other 14 patients (14.6%). Previous complications of cirrhosis were recorded. Hepatic encephalopathy was found in 16 patients (14%), ascites in 50 (43.9%), variceal bleeding in 16 (14%), hepatocellular carcinoma in 6 (5.3%), and 26 patients did not suffer from previous decompensation.
Control Group 1.
We recruited 880 people from the Spanish general population to establish the distribution of PHES scores. The subjects were drawn from the populations of Sevilla, Barcelona, Alicante, and Valencia and included healthy individuals or those receiving attention for common diseases such as arthrosis (n = 14), fibromyalgia (n = 15), cholelithiasis (n = 17), type 2 diabetes mellitus (n = 9), and hypertension (n = 18). No control subject had been receiving treatment for these disorders. Patients who were suffering from chronic liver, renal, or cardiac diseases; were suffering from psychiatric or neurological diseases; had a daily alcohol intake of >50 g/d; or were consuming drugs for any illness were also excluded. Age, sex, education level, daily alcohol consumption, employment, and ethnic origin were recorded. Following the above exclusions, a control group of 757 individuals was used to establish the normal distribution of PHES values in the Spanish population.
Control Group 2.
CFF values were obtained in 103 healthy subjects from the general population of Sevilla, Barcelona, and Valencia. The group consisted of 53 females and 50 males with a mean education level of 7.5 ± 4.2 years of formal education and an average age of 52.1 ± 11.4 years. This control group was used to establish distribution data for CFF values, influence of age and education on CFF, and the correlation between PHES and CFF in these ostensibly healthy subjects. PHES was corrected using normality tables extracted from control group 1, and no individual had a score lower than −4 (range, −2 to +5).
MHE was diagnosed based on a battery of psychometric tests that has been demonstrated to be suitable for the diagnosis of minimal HE. This battery comprises the digit symbol test (DST), the number connection test A (NCT-A), the number connection test B (NCT-B), the serial dotting test (SDT), and the line drawing test (LDT).13 The patients were invited to perform all 5 tests together with the critical flicker frequency measurement during the same session.
The results of the five tests included in the PHES score (NCT-A in seconds, NCT-B in seconds, DST in points, SDT in seconds, and LDT adding errors plus seconds) were compared with the normal distribution scores obtained from control group 1. Age and education had independent influence on all five test scores. Normality tables (means, standard deviations, and ranges) were constructed after multiple linear regressions for each test (Table 1). We compared expected versus observed results and derived the score as the number of standard deviations of difference between the two values for each test. The sum of all of them allowed us to calculate PHES score. Patients were classified as having MHE when the score was less than −4 points. Spanish normality tables have recently been published14 and are freely available (www.redeh.org).
|Digit symbol test||0.73||DST = 56.8 − 0.614 × age + 1.317 × education||10.45|
|NCT-A||0.48||NCT-A = 26.772 + 0.596 × age − 1.519 × education||21.30|
|NCT-B||0.56||NCT-B = 31.638 + 1.856 × age − 2.809 × education||44.74|
|Serial dotting||0.30||SDT = 61.346 + 0.236 × age − 1.279 × education||23.30|
|Line drawing||0.52||LDT = 51.303 + 1.043 × age − 0.99 × education||25.62|
Critical Flicker Frequency
The CFF was measured in a quiet, semidarkened room without distracting noises. A portable, battery-powered analyzer was used (Hepatonorm Analyzer; R&R Medi-Business Freiburg GmbH, Freiburg, Germany). The analyzer evokes an intrafoveal light stimulus with defined pulses of light at a wavelength of 650 nm, luminance of 270 cd/m2, and luminous intensity of 5.3 mcd. The frequency of the red light, which is initially generated as a high-frequency pulse (60 Hz) and which gives the patient the impression of a steady light, was reduced gradually until the patient had the impression that the steady light had changed to a flicker. The patient registered this change by pressing a hand-held switch. The process was repeated at least 5 times to ensure that the patient understood the procedure. Later, the procedure was repeated 10 times and from these data, the mean and the standard deviation values for each patient were calculated. CFF measurement was not reliable in 9 patients (7.8%) and 3 controls (2.9%) owing to visual defects or inability to understand the fundamentals involved in the performance of this test. These patients were excluded from the study.
Each patient was followed-up every 6 months at our outpatient clinic until he/she developed an episode of clinical HE, had a liver transplantation, died, or was observed to be free of any of these events at the end of the study. Five patients were lost during follow-up, thus no data on them were available. The patients were assessed for neuropsychiatric symptoms, including alterations in behavior, mood, orientation, and flapping tremor. If any of these signs or symptoms appeared, the patients returned to our clinic for further monitoring. No specific treatment was prescribed if patients showed MHE. Patients undergoing liver transplantation were censored as alive at the day of transplantation. Variceal bleeding,15 ascites (confirmed via paracentesis or ultrasound), hepatorenal syndrome,16 and spontaneous bacterial peritonitis17 were also diagnosed according to consensus definitions.
The results of the psychometric tests were expressed as PHES score after correction for age and education level derived from the normal distribution tables. The Fisher exact test or the chi-square test was used to assess differences between qualitative variables. The Levene test was used in the evaluation of differences in variance. ANOVA and Student t test were used to compare the mean CFF values with respect to presence or absence of cirrhosis and, as well, the presence of MHE versus patients with cirrhosis versus control subjects. Preliminary analyses provided data for receiver-operating characteristic curves to determine the optimum cut-off of CFF values for the diagnosis of MHE. The best sensitivity and specificity was found at 38 Hz. Spearman or Pearson coefficients were used to compare quantitative variables. Multiple linear regression analyses were performed to identify independent variables associated with the psychometric test performance. Kaplan-Meier analysis was used for univariate analysis and Cox regression for multivariate analysis of variables associated with survival and overt HE risk. SPSS version 11.5 for Windows (SPSS, Chicago, IL) was used for all statistical analyses and in the generation of the figures.
PHES in Healthy Controls.
We were able to perform NCT-A and DST in 884 patients, NCT-B in 792 patients, and SDT and LDT in 757 patients. Thus, overall PHES was available from 757 individuals. Mean age was 48.6 ± 12.6 years, and the mean education was 10.2 ± 4.4 years. The control group comprised 439 females and 445 males. Of these, 113 (12.8%) consumed >10 g/d of alcohol, and 771 (87.2%) were abstainers. In the univariate analysis, all five tests included in the PHES correlated with age and education level. NCT-A and NCT-B differed with respect to sex. The NCT-A in males was 36.2 ± 21.7 versus 40.4 ± 24.9 in females (P < 0.015). NCT-B in males was 87.7 ± 52.9 versus 97.1 ± 56.3 in females (P < 0.02). Alcohol consumption >10 g/d was associated with an impairment in the performance of DST (37.6 ± 123.8 versus 41.6 ± 15.4; P < 0.01) and SDT (53.7 ± 22.4 versus 60.7 ± 24.9; P < 0.006).
In the multiple linear regression analysis using age, education level, sex, hospital of recruitment, and alcohol consumption as dependent variables in each of the tests, only age and education level remained independent variables in the model (Table 1).
Prevalence of MHE.
MHE, as diagnosed by PHES, was detected in 35 out of 114 patients with cirrhosis (30.7%). Twelve of 57 patients (21%) were in Child-Pugh class A and 23 out of 57 (40.3%) patients were in class B or C (P = 0.02). MHE was found in 15 out of 48 patients with alcohol-related cirrhosis (31.2%) and in 20 of 66 patients with non–alcohol-related cirrhosis (30.3%) (P value not significant).
Factors Associated With Critical Flicker Frequency.
In the 103 subjects from control group 2, CFF did not correlate with age (r = −0.078; P value not significant), educational level (r = 0.056; P value not significant), or sex (males: 43.2 ± 3.7 versus females: 42.1 ± 3.5 P value not significant). The mean ± SD of PHES in this control group was 0.99 ± 1.47 (range, −2 to +5).
In patients with cirrhosis, CFF correlated weakly with Child-Pugh score (r = −0.22; P = 0.018) but not with model for end-stage liver disease score (r = 0.11; P = ns). The alcohol-based etiology of cirrhosis did not influence the mean of CFF values (39.6 ± 5.3 versus 38.7 ± 3.8; P = 0.32).
CFF correlated with NCT-A (r = −0.51; P = 0.0001) and LDT (r = −0.56; P = 0.001). There was a significant correlation with NCT-B (r = −0.45; P = 0.001), DST (r = 0.31; P = 0.001), and SDT (r = −0.30; P = 0.001). The Spearman correlation coefficient between CCF and PHES score was 0.54 (P < 0.0001) (Fig. 1).
Diagnosis of MHE Using CFF.
Mean CFF was significantly different (P = 0.001) in patients with cirrhosis with evidence of MHE (35.6 ± 4.1 Hz) compared with patients with cirrhosis without MHE (40.5 ± 3.7 Hz) or compared with controls (42.7 ± 3.6 Hz) (Fig. 2). The mean CFF in patients with MHE was not significantly different with respect to alcoholic (n = 15) or nonalcoholic (n = 20) etiology of cirrhosis (35 ± 4.7 versus 36.8 ± 4.3; P = 0.23).
The receiver-operating characteristic curve analysis of CFF for the diagnosis of MHE showed an area-under-the-curve value of 0.79 (95% CI 0.695-0.88) (Fig. 3).
At the cutoff of 38 Hz, there were 58 patients with normal PHES and CFF, 27 with abnormal PHES and CFF, 8 with altered PHES with normal CFF, and 21 with altered CFF and normal PHES (Table 2). At this cutoff, the specificity was 77.2% and the sensitivity was 72.4%, with an observed agreement of 75%. Of the patients with cirrhosis, 31% (35/114) had an abnormal PHES score, 42% (48/114) had an abnormal CFF score, and 49% (56/114) had an abnormal PHES and/or CFF. In the healthy control group, CFF was lower than 38 Hz in 3 out of 103 (2.9%) cases. Using the cut-off of 39 Hz proposed by Kircheis et al.,12 we observed a lower specificity (61.4%) with a slightly better sensitivity (76.2%), and the number of healthy patients showing altered CFF increased to 10 out of 103 (9.7%).
In patients with a mean CFF score >38 Hz, the frequency of MHE was 12.1%. In patients with a mean CFF score between 36 Hz and 38 Hz, the probability of having MHE was 42.3%; in patients with a mean CFF score <36 Hz, the probability of having MHE was 72.7%.
CFF in the Prediction of the Risk of Developing Overt HE and Survival: Univariate and Multivariate Analysis.
Time zero was the initial PHES and CFF assessments and the end points (in April 2006) were death, overt HE, liver transplantation, or loss to follow-up. The median follow-up period was 10.2 months. Sixteen patients died: 15 from liver-related complications and 1 from non–liver-related complications. There were overt episodes of HE in 21 patients, 7 patients had variceal bleeding, 17 had ascites, 1 had hepatorenal syndrome, 2 had spontaneous bacterial peritonitis, and 6 developed hepatocellular carcinoma. HE-related complaints were constipation in 3 patients, sedative drug requirement in 3 patients, infection in 5 cases, diuretic treatment in 3 patients, digestive tract bleeding in 4 patients, and spontaneous HE in 3 patients. In the overall series of patients, the probability of survival after an overt HE episode was 22% at 1 year. According to Kaplan-Meier univariate analysis, CFF (log-rank 15.08; P < 0.0001), Child-Pugh classification (log-rank 15.17; P < 0.0001), and MHE via PHES (log-rank 14.5; P < 0.0001) were associated with lower survival and higher risk of overt HE (Fig. 4). The probability of developing overt HE was higher in patients with altered CFF and MHE (OR 11.3; 95% CI 2.2-59.2). The percentage of patients who developed overt HE was 45% when either tool indicated abnormal scores, 36.4% in patients with MHE and normal CFF, 31.3% in patients without MHE but altered CFF, and 3.3% in patients with normal values in both tests (P < 0.0001). Taking CFF alone, in patients with Child-Pugh class B or C, the risk of overt HE during follow-up was 6.2 (95% CI 1.96-19.6) when CFF was below 38 Hz. In a Cox multivariate regression analysis, the independent predictors of overt HE were Child-Pugh class (hazard ratio 1.53; 95% CI 1.23-1.58; P = 0.041) and abnormal CFF (<38 Hz) (hazard ratio 4.37; 95% CI 1.58-12.5; P = 0.0024) (Table 3).
|Overt HE (n = 21)||Exitus (n = 16)|
|Child-Pugh class A (n = 56)|
|PHES abnormal (n = 13)||1 (7.7%)||—|
|CFF abnormal (n = 17)||2 (11.8%)||—|
|PHES + CFF abnormal (n = 9)||0 (0%)||—|
|PHES + CFF normal (n = 35)||1 (2.8%)||—|
|Child-Pugh class B/C (n = 53)*|
|PHES abnormal (n = 20)||13 (65%)||9 (45%)|
|CFF abnormal (n = 21)||13 (61.9%)||7 (31.8%)|
|PHES + CFF abnormal (n = 13)||10 (76.9%)||6 (46.2%)|
|PHES + CFF normal (n = 25)||1 (4%)||3 (12%)|
In univariate analysis, the variables associated with survival were altered CFF <38 Hz (hazard ratio 3.03; 95% CI 1.02-8.3; P = 0.0012), MHE via PHES (hazard ratio 3.7; 95% CI 1.26-11.1; P = 0.04), Child-Pugh (hazard ratio 9.5; 95% CI 2.09-43.3; P = 0.001). There were no statistically significant associations with sex, size of esophageal varices, platelets, creatinine, alkaline phosphatase, alcoholic etiology, or prothrombin activity. In a Cox multivariate regression analysis, the independent predictor of mortality was Child-Pugh class (hazard ratio 1.8; 95% CI 1.32-2.46; P = 0.0002).
The main findings of this study were: (1) significant correlation of the mean CFF with the psychometric tests of PHES and (2) CFF together with Child-Pugh class predicts the development of overt bouts of HE in follow-up. CFF is reliable, simple, easy to apply, and is not influenced by age or education level. CFF is a well-established neurophysiological technique that measures the ability of the central nervous system to detect flickering light, and which is directly influenced by cortical activity.18 The appropriate cutoff to identify abnormal CFF is still not defined. In the study by Kircheis et al.,12 the threshold was 39 Hz when comparing healthy subjects to patients with cirrhosis and encephalopathy. In the current study, the receiver operating characteristic curves revealed better sensitivity and specificity for the diagnosis of MHE using the threshold of 38 Hz. Moreover, impaired CFF (<38 Hz) correlated better with risk of developing overt HE than did 39 Hz. Furthermore, 10 healthy controls had shown a CFF lower than 39 Hz. As such, we propose the lower level of 38 Hz be used as the cutoff for “altered” CFF. Nonetheless, nearly a quarter of MHE patients had a mean CFF >38 Hz, and another fifth of non-MHE patients with cirrhosis had an altered CFF value. This could be explained on the basis of different brain pathways and functions being assessed by each of the methods. CFF appears to detect a broad spectrum of neuropsychological abnormalities ranging from visual signal processing (retinal gliopathy) to cognitive functions. Moreover, PHES explores visual perception, construction, and visual/spatial orientation together with motor speed, accuracy, concentration, and attention. Thus, the two methods could be complementary. Indeed, a lack of concordance between neurophysiological studies (mainly exogenously evoked potentials) and psychometric tests has been reported,3, 19 supporting the hypothesis that each method explores different brain functions.
In this study, the prevalence of MHE diagnosed by PHES was 30% in patients with cirrhosis. As previously reported, the prevalence of MHE depends on liver function5, 20, 21 and the type of psychometric tests or neurophysiological studies used together with the existence of an appropriate healthy control group. MHE has been diagnosed by several criteria, including the combination of subtests of the Whechsler Adult Intelligence Scale (block design test and digit symbol test) together with the number connection test,22 simple reaction time,23 or electroencephalography using automated analysis.5 Nevertheless, MHE is characterized by a decrease in psychomotor speed, together with deficits in attention and visual-constructive abilities. PHES is a reliable tool for the diagnosis of MHE, because it is able to detect the majority of these MHE-related psychometric impairments.24 The tests included in the PHES battery were carefully selected, and the score is able to distinguish patients with MHE from those with overt hepatic encephalopathy as well as from healthy controls.25 The weakness of this tool is the need for data on education-level and age-adjusted distributions. In the present study, PHES scores were adjusted using a cohort of 757 people drawn from general populations from different regions of Spain.14 As such, the prevalence of MHE in the current study would be a close approximation to that expected using appropriately adjusted tests.
Neurophysiological tools such as electroencephalography or exogenously evoked potentials including brainstem-evoked potential, visual-evoked potentials, somatosensory-evoked potentials, or endogenous P300 wave have been widely used for the diagnosis of MHE. However, these tools have lower sensitivity than psychometric tests and require specialized equipment with trained personnel. The CFF test, on the other hand, is easier and more sensitive. Spectral electroencephalography has been shown to have prognostic value when combined with psychometric tests,26 but with a low level concordance. Hence, CFF could contribute toward detecting some features of MHE that would be misdiagnosed by PHES, thus pre-empting the disadvantage associated with neurophysiological tools.
Altered CFF predicts risk for the development of overt HE but not survival. This is in accordance with previous studies that reported a lack of influence of MHE on survival.7, 20 However, in a previous study, psychometric tests in combination with oral glutamine challenge (a simple tool to analyze intestinal ammonia production) predicted survival in patients with cirrhosis.7 Further studies are needed to determine whether CFF in combination with oral glutamine challenge could also predict survival. More than two thirds of individuals at Child-Pugh class B or C were observed to be at risk of developing overt HE in the first year of follow-up when CFF was found to be altered. Conversely, only 4% of patients without PHES and/or CFF alterations developed overt HE. Both tools showed a strong ability to define risk for overt HE in patients with liver dysfunction and, as such, could be used in the prioritization of subjects selected for liver transplantation, because overt HE is related to a low survival rate. In our cohort, the survival rate was only 22% at 1 year after overt HE. This result is concordant with the data of Bustamante et al.,1 who reported a survival rate ranging from 10% to 73%, depending on the degree of liver and renal dysfunction.
In summary, CFF is a simple, reliable tool for the diagnosis of MHE. The data are not influenced by education level or age and could be useful in the management of patients with cirrhosis. A value of CFF below the cutoff of 38 Hz defines a high risk for overt HE over the first year of follow-up in patients with MHE at Child-Pugh classes B or C.
We thank Juan Carlos Quero and Stephan Curran for the critical evaluation of the manuscript and Peter R. Turner for editorial assistance.
- 8Diagnosis and treatment of hepatic encephalopathy in Spain: results of a survey of hepatologists [in Spanish]. Gastroenterol Hepatol 2006; 29: 1–6., , , , .
- 20The prognostic significance of sub-clinical hepatic encephalopathy. Am J Gastroenterol 2000; 95: 2029–2034., , , , , , et al.Direct Link:
- 25The PSE-Test: an attempt to standardize neuropsychological assessment of latent portosystemic encephalopathy (PSE). In: RecordC, Al MardiniH, eds. Advances in Hepatic Encephalopathy and Metabolism in Liver Disease. Newcastle-upon-Tyne: Medical Faculty of the University of Newcastle-upon-Tyne, 1997: 489–494., , , , , , et al.