Spectrum of neurocognitive impairment in cirrhosis: Implications for the assessment of hepatic encephalopathy


  • Jasmohan S. Bajaj,

    Corresponding author
    1. Divisions of Gastroenterology, Hepatology, and Nutrition, Virginia Commonwealth University and McGuire VA Medical Center, Richmond, VA
    • Division of Gastroenterology, Hepatology, and Nutrition, Virginia Commonwealth University and McGuire VA Medical Center, 1201 Broad Rock Boulevard, Richmond VA 23249
    Search for more papers by this author
    • fax: 804-675-5816.

  • James B. Wade,

    1. Divisions of Clinical, Department of Psychiatry, Virginia Commonwealth University and McGuire VA Medical Center, Richmond, VA
    Search for more papers by this author
  • Arun J. Sanyal

    1. Divisions of Gastroenterology, Hepatology, and Nutrition, Virginia Commonwealth University and McGuire VA Medical Center, Richmond, VA
    Search for more papers by this author

  • Potential conflict of interest: Dr. Bajaj is a consultant for, advises, and received grants from Salix and Ocera. Dr. Sanyal is a consultant for and received grants from Salix. He also received grants from Sanofi-Aventis, Intercept, and Roche.

Cirrhosis and chronic liver disease adversely affect neurocognitive functioning.1 These deficits range from neurological complications such as hepatic myelopathy to cognitive and mental status changes such as hepatic encephalopathy (HE).2 These neurocognitive difficulties can also severely restrict the patient's functioning and result in morbidity and mortality.3–6 The most prominent neurocognitive complication of cirrhosis is HE, which reflects a spectrum of neuropsychiatric abnormalities seen in patients with liver dysfunction after exclusion of other known brain disease.7 The current system for studying HE is based on a subjective clinical classification largely based on mental status changes, the West Haven scale.2, 7–9

In addition to the mental status changes detected by clinical scales such as the West Haven scale, there are significant neurocognitive disturbances in those with normal mental status, which is characterized by impaired neuropsychological and perceptual motor dysfunction.10 The current system of HE classification does not take into account the continuum of this neurocognitive dysfunction in HE, which forms a spectrum of neurocognitive impairment in cirrhosis (SONIC).7, 11, 12 This spectrum spans a patient's performance from normal to overt HE to coma. The continuous nature of these impairments is supported by the presence of psychometric and neurophysiological impairments in patients even before they reach overt HE,13 poorer performance on tests in patients with treated overt HE than those without overt HE,1, 14 and persistence of psychometric deficits even after adequate resolution of an episode of overt HE.15 In addition, the worse clinical outcomes in patients with poor psychometric performance as a continuous measure have also been described.16 This review focuses on the available techniques and systems for characterizing HE and discusses the need for an approach for gauging severity of early HE as a continuum based on relevant clinical outcomes.

The overall neurological complications of cirrhosis can be HE-related or independent of HE, and it is important to review these separately.

Neurological Complications in Cirrhosis Not Associated with HE

Patients with cirrhosis can benefit from a detailed neurological examination; traditional teaching is that patients with difficulties in consciousness who also exhibit focal neurological deficits do not have HE. However, there is a high prevalence of focal deficits in patients with cirrhosis that is unrelated to HE, which may be due to (1) prior alcohol use resulting in cirrhosis and focal deficits, such as neuropathy and cerebellar signs17; (2) direct effects of cirrhosis on the nervous system, such as hepatic myelopathy and extra-pyramidal signs18, 19; (3) deficits unrelated to liver disease, such as residual deficits of prior strokes; and (4) diseases that affect the brain and liver concurrently, such as Wilson's disease.20 Therefore, the mere presence of focal deficits does not specifically exclude HE. Existing focal neurological deficits are also dependent on the depth of examination. A study by Krieger et al.21 revealed that 50% of patients with HE had “focal neurological signs”; these signs were mostly observed in those with a history of prior overt HE. However, most patients that have been included in recent studies do not have any specific focal neurological signs.

Hepatic myelopathy is diagnosed as a spastic paraparesis and hyperreflexia without sensory loss; it does not improve with HE therapies but can reverse after liver transplantation.22, 23 Less than 1% of cirrhotic patients also present with acquired hepatolenticular degeneration with extrapyramidal and cerebellar symptoms. Unlike hepatic myelopathy, this responds to HE therapy.18

Although the above syndromes are important in our overall understanding of the neurological complications of cirrhosis, HE remains the most prevalent of these complications.


BDT, Block Design test; CFF, critical flicker frequency; DST, digit symbol test; EEG, electroencephalogram; HE, hepatic encephalopathy; HESA, hepatic encephalopathy scoring algorithm; ICT, inhibitory control test; MHE, minimal hepatic encephalopathy; NCT-A/B, number connection tests A/B; PHES, Psychometric Hepatic Encephalopathy Score; RBANS, repeatable battery for assessment of neuropsychological status.

HE-Related Neurocognitive Impairment

HE is clinically divided into normal or overt HE. However when psychometric or neurophysiological tests are also used, it can be divided into normal, minimal HE (MHE), and overt HE.7, 24 This is because minimal HE cannot be diagnosed using just the clinical examination without these specialized tests.

Clinical Classification of HE into Normal and Overt HE.

The well-known West Haven criteria have been used in several studies, but these criteria suffer from a lack of consistency in their application (Table 1).9, 24 There is a lack of reproducibility apart from the extremes of consciousness.24 Recent studies have shown that there is good agreement among observers in grading patients who are in a coma and those who are completely alert.7, 25 The Glasgow Coma Scale is used to further classify patients with HE who are in a coma.24, 26 The clinical hepatic encephalopathy staging scale (CHESS) is promising for mental status assessment but is currently undergoing validation trials.27 Although the clinical scales are easy to apply, there remains a large gray zone in the quantification of changes between the two extremes of normal and coma (Fig. 1).8, 25

Table 1. West Haven Criteria for the Diagnosis of Hepatic Encephalopathy
StageDistinguishing Features
0No abnormality detected
ITrivial lack of awareness
Euphoria or anxiety
Shortened attention span
Impairment of addition or subtraction
IILethargy or apathy
Disorientation for time
Obvious personality change
Inappropriate behavior
IIISomnolence to semi-stupor
Responsive to stimuli
Gross disorientation
Bizarre behavior
IVComa, unable to test mental state
Figure 1.

Agreement of clinical scales in HE. There is insufficient agreement between observers in the clinical scales of HE except in the extremes. The large population of HE that exists between coma and appearing clinically normal is prone to misclassification using pure clinical scales, which leads to subjectivity.

Differentiation Between Stage 0 and Stage 1 of the West Haven Criteria.

Stage 0 encompasses patients with normal cognitive function and minimal HE, the only definition being that patients have no current clinical signs and symptoms of overt HE.24 This differentiation between stage 0 and stage 1 of the West Haven criteria, which is critical for inclusion or exclusion into research trials or therapy, is clouded in uncertainty due to the nonspecificity of signs and symptoms of stage 1 HE (Fig. 2).9 The problem is this: What profile constitutes symptoms of HE in patients who are otherwise ambulatory? This is a difficult question when using pure clinical scales, as evidenced anecdotally by clinicians and through systematic studies.

Figure 2.

Characterization of HE stages using clinical and psychometric/neurophysiological tests. Using clinical and specialized testing, patients evaluated for HE are classified as normal, minimal HE, or overt HE. Depending on the specialized tests used and the availability of population norms, the diagnosis of MHE versus normal can vary between populations. In addition, the important distinction between stage I overt HE using West Haven criteria and minimal HE is often blurred due to the inability of the clinical scales to accurately define this stage. This furthers the subjectivity of this categorical approach in the initial stages of HE. The assessment of stage II, when patients begin to exhibit disorientation, is fairly reliable between raters.

Subcategorizing Stage 0 into Normal and MHE.

In addition to the difficulties faced in dividing patients into normal (stage 0) and stage 1, there is even more controversy surrounding the division of stage 0 into normal and MHE. MHE is defined as cognitive dysfunction without clinical signs of overt HE.13, 28 The diagnosis of MHE is only possible through specialized psychometric and neurophysiological measures.29 These methods are sensitive and reproducible to a large extent, but the applications across several populations are limited because of lack of norms and lack of appropriate language forms.29

Tests for the Diagnosis of MHE

Cognitive dysfunction in patients with MHE is characterized by attention deficit, working memory problems, and defects in executive functions such as response inhibition.1, 30 There is no evidence of long-term memory or language function decline in patients with MHE. Therefore, testing strategies for MHE are traditionally focused on these neurocognitive domains. Details of individual psychometric and neurophysiological tests are listed in Table 2, and the specific domains are shown in Fig. 3. Although these tests are sensitive, their specificity is in question, because patients with other metabolic and traumatic insults to the brain can have similar deficits. Therefore, the test performances have to be interpreted in the context of the patient's overall medical history, examination findings, and socio-economic status.

Table 2. Logistical Issues with Currently Available Psychometric Tests
TestDomains TestedUnited States NormsCopyrightSpecialized Expertise (Psychology/Neurology) NeededTime for Administration InterpretationSpecific Comments
  1. The NCT-A, NCT-B, DST, LTT, and SDT are parts of the Psychometric Hepatic Encephalopathy Score (PHES). The expense for each test depends on local availability and the need for copyrighted test materials and computers. Up-front costs may be minimal if these tests are used often.

  2. Abbreviations: CDR, cognitive disease research; LTT: line tracing test; MDF, mean dominant frequency; SDT, serial dotting test.

Paper-pencil psychometric tests
NCT-APsychomotor speed+NoNo30–120 secondsPoor specificity
NCT-BPsychomotor speed, set shifting, divided attention-YesYes1–3 minutesMore specific than NCT-A but not pathognomonic for any disorder
BDTVisuo-spatial reasoning, praxis, psychomotor speed+YesYes10–20 minutesCan be used for dementia testing as well
DSTPsychomotor speed, attention+YesYes2 minutesTends to be very sensitive and is an early indicator
LTTPsychomotor speed, visuo-spatialYesNo10 minutesOutcomes are errors and time; tests a balance between speed and accuracy
SDTPsychomotor speedYesNo1–4 minutesOnly tests psychomotor speed
RBANSVerbal/visual/working memory, visuo-spatial, language, psychomotor speed+YesYes35 minutesHas been primarily studied in dementia and brain injury; current trials with HE are underway
Computerized psychometric tests
ICTResponse inhibition, working memory, vigilance, attentionLimited normsNoNo15 minutesRequires highly functional patients; familiarity with computers may be necessary
CDRAttention, episodic and working memoryYesNo15–20 minutes
Sternberg paradigm or scan testWorking memory, vigilance, attentionYesNo10–15 minutes
Neurophysiological tests
EEG MDF and spectral indexGeneralized brain activityLocal normsNoYesDifferent rangesCan be performed in comatose patients
Visual evoked potentialsInterval between visual stimulus and activityLocal normsNoYesDifferent rangesHighly variable and poor overall results
Brainstem auditory evoked potentialsResponse in the cortex after auditory click stimuliLocal normsNoYesDifferent rangesInconsistent response with HE testing/prognostication
P300 cognitive evoked potentialsInfrequent stimulus embedded in irrelevant stimuli is studiedLocal normsNoYesDifferent rangesGood diagnostic potential, but requires patient cooperation
CFFVisual discrimination and general arousalNoNo10 minutesRequires highly functional patients
Figure 3.

Psychometric testing in order of complexity and functions tested. From the bottom up, there is an increasing complexity of functions tested by the various methods. Clinical scales primarily establish orientation, whereas psychometric tests start from the basic reaction times and proceed to complex executive functions. The tests must be used judiciously given the background of the particular patient and domains that need to be tested. The tests also may worsen in the reverse order with worsening HE.

Psychometric Tests

Paper-Pencil Batteries.

The traditional batteries used are paper-pencil batteries, which have the benefit of being portable, easily translatable, and with alternate forms to prevent learning effects. The recommended battery is the portosystemic encephalopathy syndrome test or Psychometric Hepatic Encephalopathy Score (PHES).1, 7, 10, 30 This battery consists of five tests: the number connection tests A and B (NCT-A/B), digit symbol test (DST), serial dotting test, and line tracing test. The NCT-A (or trails test A) tests for psychomotor function, whereas the NCT-B analyzes divided attention and executive function. The DST is a test of attention and processing speed. Serial dotting and line tracing are also tests of processing speed. All these tests are compared against age- and education-matched controls, and from these five tests, six scores are generated (the line tracing test has a score for time and errors), which are added to give a composite. The total score is a summation of the number of standard deviations (>−1 to <−3) from the age-corrected mean values. Scores between −5 to −18 are considered abnormal. This battery has been validated in Europe, but does not have norms for the United States.31

The working group on HE recommendeds that the PHES—or, in its absence, any two of the following four tests: NCT-A, NCT-B, DST, or Block Design test (BDT)—must be abnormal to be called MHE.7 The BDT tests for constructional praxis and visuo-motor processing. Currently, the BDT and DST are available as components of the Wechsler's Adult Intelligence Scale.32

The International Society for Hepatic Encephalopathy and Nitrogen Metabolism recommended the use of PHES or Repeatable Battery for Assessment of NeuroPsychological Status (RBANS).33 The RBANS consists of five domains: immediate memory, delayed memory, language, attention, and visuo-spatial skills.34, 35 Composite and individual cognitive domain scores are generated at the end of the test. The RBANS has been used for evaluating dementia and in unselected cirrhotic patients, but a systematic experience with RBANS in HE is currently lacking.35, 36

Computerized Psychometric Tests.

Psychometric tests administered with a computer have the advantage of being automated; however, a familiarity with computers is often needed in some tests. Three testing systems—the inhibitory control test, the drug research test, and the Sternberg paradigm—have been used recently to diagnose cognitive dysfunction in cirrhosis.

Cognitive Drug Research.

With over 50 parallel forms of each task, the cognitive drug research system (Cognitive Drug Research Ltd, Goring-on-Thames, United Kingdom) is widely used for the assessment of cognitive function in clinical trials in the United Kingdom.37 It has five domains: power of attention, continuity of attention, speed of memory, quality of working memory, and quality of episodic memory. This system was validated in outpatients with cirrhosis; performance on key domains worsened after transvenous intrahepatic portosystemic shunt placement and improved after liver transplantation.37

Inhibitory Control Test.

The inhibitory control test (ICT) is a modification of the continuous performance test in which patients are instructed to respond to alternating presentations of X and Y (called targets) and to avoid responding to nonalternating presentations called lures that are interspersed within the test.38 The ICT assesses vigilance, sustained attention, response inhibition, and working memory.39, 40 ICT lures can differentiate between cirrhotic patients with and without cognitive dysfunction. ICT performance also predicts the development of overt HE, worsens after transvenous intrahepatic portosystemic shunt placement, and improves after therapy.39, 40 A recent study revealed that the ICT was associated with motor vehicle crashes.16 The ICT is available for free download at www.hecme.tv.

Scan Test.

The scan test is a computerized digit-recognizing task based on the Sternberg paradigm that measures the mean reaction times and the percentage of errors in recognition of patterns of digits.41, 42 The scan test is influenced by visual stimulus encoding, altered memory scanning, sustained attention, and psychomotor processing defects.42 Performance on this test was associated with survival; however, it was not shown to be influenced by the glutamine challenge, and therefore the test needs further validation.41, 43

Neurophysiological Tests

A compendium of neurophysiological tests has been used to characterize cognitive dysfunction in cirrhosis ranging from simple electroencephalogram (EEG) to complex oddball paradigms for evoked potentials and the critical flicker frequency. The advantages of neurophysiological testing are the absence of learning effects and the relative specificity of the response.7 The need for expensive equipment, low sensitivity, and the lack of accompanying behavioral information are drawbacks of these methods.


The simple EEG has some value in determining the advanced stages of HE with the characteristic “triphasic waves.”44 In the earlier stages of HE, the mean dominant frequency of the EEG and spectral EEG analysis are useful and can predict the development of overt HE and liver-related death, at least in patients with advanced liver disease.11 A recent study reaffirmed the use of bispectral index of the EEG in advanced overt HE. However, the EEG has low concordance with PHES, in part due to its restriction to studying cortical activity.31

Evoked Potentials.

The latency between application of a stimulus and the brain's ability to sense it can be measured as evoked potentials. Some evoked potential components can even be useful in a deep coma when the EEG is suppressed.45 The specific evoked potentials tested are visual, somatosensory, and auditory. Visual evoked potentials, which can be flash, pattern-reversed, or motion-elicited, have been examined in HE. Pattern visual evoked potentials are insensitive to change in HE and require active patient cooperation, making them only useful for early stages of HE.45 Somato-sensory evoked potential abnormalities were seen in 48% with abnormal inter-peak latencies N20-N65, but there was no correlation with psychometric tests in later studies.46


Brain electrical responses can be measured in response to stimuli that are administered visually or through auditory means. An auditory P300 is when a patient receives an infrequent stimulus embedded in a series of otherwise irrelevant frequent stimuli (oddball paradigm); the subject is asked to identify the rare stimuli (oddballs) by pressing a button or to keep a mental count of their occurrence. The brain shows a response typically 300 milliseconds after the oddball stimulus; delay in brain response (i.e., after a 300-millisecond gap) indicates dysfunction. Most studies of this technique have included stage 1 HE patients along with MHE patients, and most of the changes were observed in alcoholics.47, 48 Visual P300 responses were found to be abnormal in 78% compared with psychometric abnormalities in only 41% of patients but were related to a decline in hepatic metabolic capacity; these are not recommended due to their inconsistent results.11, 49 Auditory P300 does have good diagnostic potential and can be used when available for evaluation of neurophysiological function in cirrhosis.11, 47, 48, 50

Critical Flicker Frequency.

Critical flicker frequency (CFF) evaluates the maximum frequency at which a flickering light source can still be perceived to flicker and measures a threshold for the fusion of these lights.14 The threshold is achieved by increasing and decreasing frequency over several trials. The CFF score has been used as a continuous variable in several studies, which have shown good sensitivity in the diagnosis of overt and MHE.14, 51 It has been validated for use in Europe and India. A CFF of 39 Hz showed a 100% separation between overt HE versus controls/nonovert HE.14 Separation between MHE versus no MHE was lower, with a sensitivity of 55% and specificity of 100% in the earlier studies.14 More recent studies have shown increased sensitivity of the CFF in MHE.52, 53 The CFF, however, does require patient cooperation and binocular vision.

Blended Scales for HE

Because the West Haven criteria (Table 1) are of limited sensitivity in evaluating HE, scales that blend psychometric/neurophysiological measures with clinical measures have been developed.9

The portosystemic encephalopathy index developed by Conn et al.9 includes the clinical assessment of mental state, the trail-making test score, EEG, asterixis, and arterial ammonia. However, this index has been criticized because the number connection tests were not controlled for age and education, and some criteria were inter-dependent.7

The hepatic encephalopathy scoring algorithm (HESA) is a blended scale consisting of clinical and neuropsychological measures.25, 54 The HESA divides patients with HE into four grades and is a modification of the West Haven criteria. This scale provides an assessment of mental status and neuropsychological state from early overt HE to coma. Experience with HESA in a multicenter trial of hepatic replacement therapy demonstrated its applicability and the need for better criteria than the existing West Haven criteria. Agreements on the HESA were highest in both extremes, but there remained a significant difference in assessments between sites in the interim. In addition, HESA alone has not been proven to be sensitive for the diagnosis of minimal HE, because the cognitive tests used are relatively simple.54, 55

A simple division of HE into low-grade or high-grade HE has been proposed by Haussinger and colleagues.56 Although such a method is easy and practical, it has not been adopted widely, because it may be an oversimplification of a multifaceted problem.

Approaching the Spectrum of Neurocognitive Impairment in Cirrhosis as a Continuum Rather Than Dividing Patients into Normal, Minimal, and Overt HE

A continuous rather than categorical approach to gauge the neurocognitive impairment in HE is required due to (1) an inherent subjectivity of the clinical assessments in the earlier and middle stages of the disease; (2) no consensus regarding the multiplicity of tools to diagnose cognitive dysfunction; (3) a lack of population norms for most tools in the United States, which invalidates data interpretation; and (4) imperfect prediction of important clinical and psycho-social outcomes based on the test performances.

The continuous nature of this neurocognitive impairment is missed by the current system of arbitrary cutoffs, most often >2 standard deviations impaired beyond comparable controls.7 In therapeutic HE trials, the decision for HE reversal versus persistence can therefore often boil down to a few seconds or limited change in raw scores on the individual tests, which can result in important changes in patient classification. The current system could also be improved by focusing on hard outcomes such as development of overt HE and traffic accidents. Prior studies have shown that MHE patients are more likely to develop overt HE and be involved in traffic accidents; however, this is by no means a foolproof prediction, because the majority of MHE patients do not develop overt HE and are not involved in traffic accidents.6, 16, 57 A better method would be to first focus on the outcomes and then determine which criteria predicted those outcomes with accuracy. It is also likely that the considerable uncertainty between the various methods to divide patients into normal, minimal HE, and overt HE would be overcome by approaching neurocognitive performance in cirrhosis as a whole as this continuum. The proposed spectrum of neurocognitive impairment in cirrhosis (SONIC) requires validation in a large, prospective study. This study would involve administration of several measures assessing psychometric and neurophysiological domains (because none of the available tests is capable as a stand-alone tool) while permitting the development of clinically relevant outcomes.

In conclusion, there are several tests and grading methods for HE. The current system of grading patients into separate categories of normal, minimal HE, and overt HE methods can increase subjectivity. Treating this neurocognitive dysfunction as a continuum (i.e., the spectrum of neurocognitive impairment in cirrhosis) with patient-specific outcomes could improve the understanding of this disease and its application in patients' daily lives.