Psychometric tests
Many new psychometric tests have emerged in recent years, such as psychometric hepatic encephalopathy score (PHES) (Ferenci et al.
2002), Stroop app (Bajaj et al.
2013), and animal naming test (ANT) (Campagna et al.
2017). Ideal screening tests for MHE should be characterized by less time-consuming, objective outcome, independent of specialists’ interpretation, simple to use and free from copyright and fees (Yoon et al.
2019).
PHES
PHES comprises of five tests and has been recognized as the gold standard for diagnosis of MHE since 1998. It is widely accepted that diagnostic of MHE is PHES score ≤ −5 (Ferenci et al.
2002; Luo et al.
2019; Singh et al.
2016). Tsai et al. found that PHES can be a useful tool for detecting patients with MHE in around one third of outpatients with cirrhosis (Tsai et al.
2015). Badea et al. showed the prevalence of MHE in Romanian cirrhosis patients. PHES of the healthy control group was significantly higher than that in the liver cirrhosis group (Badea et al.
2016). But Coskun et al. set −4 as the cutoff in Turkish PHES nomograms. They found that the PHES score in the cirrhotic group was −2.18 ± 3.3, which was significantly lower than the center of gravity (Coskun et al.
2017). To validate widely applicable norms, future multicenter studies are needed.
Stroop app
The Stroop app score, which is based on ipad or ipod, is significant for diagnosis of MHE (Bajaj et al.
2013). The method is shown in Table
1. The Stroop app has a accuracy rate of 0.74 and is suitable for MHE screening (Yoon et al.
2019). Zeng et al. demonstrated that >97.34 s for “Off” state time and > 186.63 s for “Off” state + “On” state time had the maximum area under the curve values in all patients. Meanwhile, “Off” state + “On” state time had the highest sensitivity with a cutoff of 186.63 s. Compared with PHES, the Stroop app is time saving, accessible, convenient, and accepted by patients and clinicians (Zeng et al.
2019).
Table 1
Comparison of different kinds of diagnostic methods
Psychometric tests | PHES | It comprises the number connect test (NCT)-A, NCT-B, serial dotting test, line tracing test, and digit symbol test, which are each scored from 1 to −3 | ☆take time; ☆affected by demographic factors; ☆lack ecological validity language functions; ☆only two cognitive domains |
Stroop app | “Off” and “On” task are the two components depending on the discordance or concordance of the stimuli. Patients have 2 training runs for two components. In “Off” state, pound signs (###) presented in red, green or blue, one has to respond as quickly as possible by touching the matching color, which were also randomized and not fixed to their respective positions. This continues until a total of 10 presentations. If the subject makes a mistake, the run stops and has to restart again. The “Off” state continued till the subject had achieved 5 correct runs. In the “On” state, the patient has to touch the color of the word presented which is actually the name of the color in discordant coloring. The test of cognitive processing controlling for psychomotor speed was subtracting the “Off” state time from the “On” state time | ☆complex ☆optimal cutoff is varied |
ANT | In ANT test, patients were asked to list as many animals as they could in 1 min. All repetitions and errors were excluded from the calculations | ☆no significant limitatin |
Other | ICT: An alcohol-related or control picture was presented in the centre of the screen with one of two letters superimposed on one of the four corners of the picture. Patients were instructed to press the space bar if the go cue was present, but to withhold their response if the no-go cue was present. During each trial, the picture and letter remained on screen until the participant responded or until a 1500 ms timeout had elapsed eNCT: electronic based NCT | ☆ICT requires highly functional patients |
CFF | Patient is equipped with a light shade, and the red light spot flashes at a ratio of 1:1 at a frequency of 60 Hz. At the beginning, the subject cannot recognize the flicker because the flicker frequency is fast, and then the flicker frequency gradually slows down until the subject can recognize the flicker stop. | ☆expensive; ☆time-consuming; ☆dependent on the specialist’s interpretation |
MRI | Marker, including mean kurtosis values, Six ROIs, ALFF values and default mode network, are useful biomarkers for MHE detection. | ☆lack of detection accuracy of the measured signal |
inflammatory cytokines | Na | ☆no studies on the accuracy and sensitivity of its application |
ANT
Cognitive functions related to prefrontal anterior/ cortex cortical areas are sensitive to the ANT. Patients are asked to list as many animals as they can in 1 min. Errors and repetitions should be excluded from the calculations (Table
1). Campagna et al. found that, in order to eliminate age interference, a simplified ANT was obtained, adding 3 animals for patients with <8 years of education and 6 animals if they were over 80 years old in addition (Campagna et al.
2017). Labenz et al. demonstrated that the simplified ANT may become an initial screening tool for assessment of MHE. They found that ANT was Significant lower in patients with MHE. The best discrimination between patients with and without MHE is naming <20 animals. But when the cutoff value is ≥23 animal names, 38.5% of patients could be avoided further testing for MHE, and the negative predictive value was 84% (Labenz et al.
2019).
Other tests
The inhibitory control test (ICT) represents the ability to suppress irrelevant motor or cognitive processes, and is useful for diagnosis of MHE (Di Lemma and Field
2017; Hartmann et al.
2019). The method is shown in Table
1. Gupta et al. demonstrated that ICT is correlated with disease severity, predicts the development of HE, and has excellent test–retest reliability. ICT was considered abnormal when there were ≥ 14 ICT lures. Mean ICT lures were higher and target accuracy was lower in cirrhotic patients with MHE than those without MHE. ICT had a sensitivity of 92.6% and specificity of 78.5% with an area under the receiver operating characteristic curve of 0.855 for MHE (Gupta et al.
2015).
The novel electronic number connection test (eNCT) has test–retest reliability to detect cognitive function and monitor cognitive impairment in patients with cirrhosis. Wuensch et al. found that the eNCT performance was negatively correlated with PHES performance in patients with cirrhosis. Control participants showed significantly faster eNCT completion times compared with cirrhosis patients (Wuensch et al.
2017).
Critical flicker frequency (CFF) and electroencephalography (EEG)
CFF reflects dysfunction of nerve conduction in the brain. It is an objective test that avoids the deviation caused by cultural differences. The CFF method is shown in Table
1 (Wang et al.
2013). CFF is widely acknowledged as an adjunct tool for diagnosis of MHE. However, some researchers believe that CFF should be used as an adjunct to the PHES test because of its low sensitivity for detecting MHE. They found that CFF had a diagnostic accuracy of 70.6%, specificity of 82%, sensitivity of 39% for detecting MHE (Ozel Coskun and Ozen
2017). However, for the threshold of CFF, the researchers had different choices. Barone et al. evaluated that a CFF cut-off of 39 Hz is effective in predicting survival and the first episode of MHE in cirrhosis patients who had never experienced MHE. With progression of the Child–Pugh class, the prevalence of CFF ≤39 Hz significantly increased (Barone et al.
2018). Greinert et al. found that most patients with a MELD score > 24 and CFF <43 Hz had MHE. They demonstrated that combination of CFF and MELD score may be used as a first diagnostic step to filter patients, in whom further MHE testing could be avoided. Specificity and sensitivity of a CFF cut-off of 43 Hz was 93.5% and 42.9%. (Greinert et al.
2016).
EEG has been acknowledged to confirm the presence and predict the severity of HE. Recently, Olesen et al. found increased sample entropy of the EEG in patients with MHE. α activity is gradually replaced by slowed brain oscillations typically with θ in the frequency range (4–8 Hz) in the transition from an unimpaired mental state to HE (Olesen et al.
2016). Spectral EEG is a quantitative tool for diagnosis and assessment of the response to treatment in MHE. Spectral EEG analysis showed lower mean dominant frequency and higher θ relative power but lower α relative power in patients with MHE than in patients without MHE. Singh et al. found that with spectral EEG, 96% sensitivity, 84% specificity and 90% accuracy were obtained for diagnosis of MHE (Singh et al.
2016).
Imaging
Magnetic resonance imaging (MRI) applies electromagnetic waves emitted by a graded magnetic field to acquire the internal structure of the objects (Lockwood et al.
1991; Shawcross et al.
2007). The biomarker for MHE is altered regional brain spontaneous activity. For example, the mean kurtosis values in the putamen decrease in the globus pallidus, putamen, caudate nucleus, and/or thalamus in patients with MHE (Sato et al.
2019). Chen et al. demonstrated that the amplitude of low frequency fluctuation (ALFF) values within six regions of interest (ROIs) correlated with PHES in cirrhosis patients. The ROIs contains the posterior cingulate cortex/precuneus, bilateral medial frontal cortex/anterior cingulate cortex, right lingual gyrus, the left precentral and postcentral gyrus, inferior/superior parietal and middle frontal gyrus lobule (Chen et al.
2016). Zhong et al. also found that patients with MHE had significant decreased ALFF (Zhong et al.
2016). Default mode network function, especially the medial prefrontal cortex, might also be a useful imaging marker for differentiating MHE from cirrhosis. The MHE patients showed even more decreased connectivity in medial prefrontal cortex, left superior frontal gyrus, and right middle temporal gyri when compared with non-MHE patients (Qi et al.
2014).
Different kinds of MRI play an important role in the detection of MHE. Kooka et al. demonstrated magnetic resonance spectroscopy, which shows that reduced magnetization transfer ratio in the whole brain field and an increase in glutamate/glutamine or taurine in chronic HE patients contribute to early and objective diagnosis of MHE. The levels of brain glutamine were significantly lower and the levels of brain myo-inositol were significantly higher in the control group compared with the MHE group (Kale et al.
2006; Kooka et al.
2016; Rovira et al.
2001). Altered diffusion kurtosis imaging metrics indicate brain microstructure abnormalities in MHE patients. Significantly alterations in axial diffusivity, radial diffusivity, and MD in a wide range of regions, including corpus callosum, left thalamus, were closely correlated with cognitive score (Li et al.
2019). Diffusion tensor imaging can differentiate MHE from non-MHE patients. Chen et al. found that in MD or fractional anisotropy maps, two spatially distributed white matter regions yielded 75.4–81.5% and 83.1–92.3% classification accuracy when differentiating patients with and without MHE (Chen et al.
2015).
Inflammatory cytokines
In addition to hyperammonemia, inflammation also modulates neuropsychological function in patients with MHE. For example, Circulating IL-6 is negatively associated with memory function in low-dose endotoxemia (Krabbe et al.
2005; Tsai et al.
2015), serum IL-6 and IL-17a levels are independent risk factors for MHE in HBV-infected patients (Li et al.
2015), IL-1β is a potential, independent predictor of MHE (Wunsch et al.
2013), patients with MHE have significantly higher TNF-α is significant higher in MHE patients (Srivastava et al.
2011). These inflammatory cytokines may become biomarkers for MHE diagnosis, but further researches are needed.
Comparison
Paper-and-pencil test used in PHES is the gold standard for diagnosis of MHE. But the process of diagnosis of PHES is inconvenient and affected by many factors. The diagnostic methods take time, and are affected by demographic factors, and lack ecological validity and language functions, such as verbal memory. PHES focuses on only two cognitive domains but it is not sensitive enough to detect early neurological alterations. Patients classified as without MHE by PHES have a high risk of suffering overt HE. Around 40% of patients without MHE according to PHES fail two other psychometric tests (Bajaj
2008 a; Gimenez-Garzo et al.
2017; Nardone et al.
2016; Seo et al.
2017; Wang et al.
2008). Other kinds of psychometric tests also have limitations. The ICT, which is a computerized test of response inhibition and working memory, requires highly functional patients. The Stroop app, which evaluates psychomotor speed and cognitive flexibility, is also a complex task that is applicable in highly functional patients (Bajaj et al.
2013). Tapper et al. conducted a meta-analysis to evaluate different kinds of psychometric tests. They compared ICT, Stroop app and ANT in diagnosis of MHE. They found that optimal cutoff for the Stroop app still varies. Good performance in the ICT, Stroop app or ANT is related to reduced development of HE, but longitudinal data are still limited. Studies are needed in clinically representative populations with cutoffs validated (Tapper et al.
2018). However, psychometric tests are irreplaceable now because MHE has subtle abnormalities that can be detected only using specific neuropsychometric and/or neurophysiological tools in cirrhosis patients with otherwise normal neurological examination results (Stewart and Smith
2007). Among all psychometric tests, the ANT is reasonably widespread in humans of every culture, and the influence of gender, age and education, if any, might be limited (Campagna et al.
2017).
CFF is a noninvasive, rapid, simple test for diagnosis of MHE. Compared with PHES, CFF has a positive predictive value of 93.2 ± 7.44%, specificity of 92.7 ± 7.96%, negative predictive value of 90.4 ± 8.91% and sensitivity of 91.1 ± 8.32%. CFF is excellent for diagnosis of MHE, with an area under the curve of 0.937 (Metwally et al.
2019). At one time, scientists thought EEG was an important method to predict MHE, but the detection of MHE showed limited agreement between PHES and EEG (Nardone et al.
2016). Meanwhile, it is difficult to make good use of EEG and CFF, as they are expensive, time-consuming, and dependent on specialist interpretation (Yoon et al.
2019). CFF is recommended as an adjunct (but not replacement) to psychometric testing.
Neuroimaging studies can detect diffuse abnormal metabolic activity of nerve cells, which is a typical feature of patients with MHE. MRI can provide objective and reliable imaging biomarkers that are necessary to help diagnose or identify MHE (Zhang et al.
2014). One major drawback of MRI concerns the lack of detection accuracy of the measured signal, but with technical advances, a solution to that problem is imminent (Janssen et al.
2018).
In addition to hyperammonemia, there is a parallel relationship between inflammatory cytokines and MHE, or a significant correlation between proinflammatory cytokines with MD values on diffusion tensor imaging (Srivastava et al.
2011) or PHES (Wunsch et al.
2013). However, there have been no studies on the accuracy and sensitivity of its application in the diagnosis of MHE, and more studies are expected.
Recent guidelines suggest that either alternative techniques, such as computerized tests, neurophysiological testing or EEG should be used alongside PHES for multicenter studies (Morgan et al.
2016; Vilstrup et al.
2014) (Table
1).