Serum S100B Levels in Patients with Liver Cirrhosis and Hepatic Encephalopathy

Hepatic encephalopathy (HE) is one of the main complications of liver cirrhosis (LC) and is classified into minimal hepatic encephalopathy (MHE) and overt hepatic encephalopathy (overt HE). S100B is expressed mainly in astrocytes and other glial cells, and S100B has been reported to be associated with various neurological disorders. The present study aimed to investigate the diagnostic ability of serum S100B to discriminate the grade of HE and the parameters correlated with serum S100B levels. Additionally, we investigated whether serum S100B levels can be used to predict 1-year mortality in cirrhotic patients. In total, 95 cirrhotic patients were consecutively enrolled and divided into the following three groups: (i) without any types of HEs; (ii) with MHE; and (iii) with overt HE. The diagnosis of MHE was made by the Mini-Mental State Examination (MMSE) and Psychometric Hepatic Encephalopathy Score (PHES). Among the three groups, there were no significant differences in serum S100B levels regardless of HE severity. The clinical parameters correlated with serum S100B levels were age, serum bilirubin, and creatinine levels. The Model for End-Stage Liver Disease (MELD) score showed a significant positive correlation with serum S100B levels. The relationship between serum S100B levels and MELD score was maintained in 48 patients without any type of HE. Additionally, hyperammonemia, low cholesterol levels, and the combination of serum S100B levels ≥ 35 pg/mL with MELD score ≥ 13 were factors for predicting 1- year mortality. In conclusion, serum S100B level was not useful for differentiating the severity of HE. However, we found that serum S100B levels can be affected by age, serum bilirubin, and creatinine in cirrhotic patients and are associated with MELD scores. Additionally, serum S100B levels showed the possibility of predicting 1-year mortality in cirrhotic patients. These findings suggest that serum S100B levels may reflect liver dysfunction and prognosis in liver disease.


Introduction
S100B is a protein belonging to the S100 protein family that was first reported by B.W. Moor in 1965 [1]. The S100 proteins, a family of calcium (Ca 2+ )-binding cytosolic proteins, are composed of 25 known members, namely, S100A1-18, hair hyaluronan, keratin fibrin, reptin, S100B, S100P, S100Z and S100G [2]. These proteins have diverse intracellular and/or extracellular functions, including the regulation of cell proliferation, differentiation, apoptosis, migration, energy metabolism, calcium balance, protein phosphorylation, and inflammation [3][4][5][6]; therefore, altered expression of S100 proteins may be associated with disease development. S100B has been reported to be expressed mainly in astrocytes and other glial cells, such as oligodendrocytes and Schwann cells. It has been reported to be associated with various neurological disorders [7]. However, S100B has also been detected in definite nonneural cell types, such as dendritic cells, certain lymphocyte subpopulations, chondrocytes, Langerhans cells, melanocytes, adrenal medulla satellite cells, skeletal muscle satellite cells, and adipocytes [8][9][10][11]. Additionally, we have previously reported that S100B is expressed in the liver of mice and that the interaction with the receptor for advanced glycation end products (RAGE) is observed in cholestatic liver injury by bile duct ligation [12]. Another experimental study has reported that serum S100B levels increase in the setting of local ischemia and perfusion of the rat liver without intracranial injury, suggesting that the liver may be a possible source of S100B [13].
To date, serum levels of S100B protein have been extensively studied in certain conditions, such as neural tissue injury, but rarely in relatively stable cirrhotic patients without any types of hepatic encephalopathy (HE). LC is the end stage of various liver diseases and a major cause of morbidity and mortality worldwide. HE is one of the main complications of LC and develops in 30-45% of patients with cirrhosis [14]. HE is defined as brain dysfunction caused by liver insufficiency and/or portosystemic shunting. Its manifestations include a wide spectrum of neurological or psychiatric abnormalities, such as inappropriate behavior, disorientation, confusion, slurred speech, stupor, and coma [15]. Depending on the severity of symptoms, HE can be classified into minimal hepatic encephalopathy (MHE) and overt hepatic encephalopathy (overt HE). Several studies have suggested that elevated serum S100B levels may be a diagnostic marker of overt HE in fulminant hepatic failure or cirrhosis [16,17]. However, little is known about the roles of serum S100B levels in diagnosing MHE or predicting 1-year mortality in cirrhotic patients.
In the present study, we aimed to investigate the diagnostic ability of serum S100B to discriminate MHE and the parameters correlated with serum S100B levels in cirrhotic patients, depending on whether HE is present. Second, we investigated whether serum S100B levels can be used to predict 1-year mortality in cirrhotic patients.

Patients
This prospective observational study included all consecutive adult patients with LC admitted to the Hallym University Sacred Heart Hospital in the Republic of Korea from July 2013 to July 2014. The diagnosis of LC was made by clinical, laboratory, and radiologic features. Patients with a history of malignancy or brain diseases, such as stroke, trauma, or psychiatric disease, were excluded. A total of 95 patients were enrolled, and all patients were divided into the following three groups: (i) without any types of HEs; (ii) with MHE; and (iii) with overt HE. To assess the severity of HE, the West Haven criteria were used (Table 1).

Diagnosis of MHE and Overt HE
Patients with marked changes in consciousness and behavior corresponding to grades 3-4 of the Western Haven criteria were diagnosed as overt HE. Patients classified as grade 0-2 of the Western Haven criteria underwent MMSE. When the MMSE score was 24 or more, PHES was additionally performed to check for MHE. The PHES was calculated based on the results of the following five paper-and-pencil tests: digit symbol test (DST), number connection test A (NCT-A), number connection test-B (NCT-B), serial dotting test (SDT), and line tracing test (LTT). MHE was diagnosed when the PHES was −5 or less, and patients who satisfied both MMSE score ≥ 24 and PHES > −5 were demonstrated to be free of HE (Figure 1).

Measurement of Serum S100B Levels
Blood samples for analysis of S100B were collected in EDTA tubes, and the plasma was separated and stored at −80 • C until analysis. Serum S100B protein levels were measured using a commercially available enzyme-linked immunosorbent assay (Biovendor, LLC, Chandler, NC, USA). Serum samples were added to the wells of the ELISA plate and incubated. The plates were then washed and incubated with avidin-conjugated horseradish peroxidase-conjugated substrate solution, and HCL (1.0 M) was added to stop the reaction. The optical density (OD) was measured spectrophotometrically at 490 nm on a microplate reader. All serum samples were tested in duplicate, and the average was reported. The minimum detectable concentration was 15 pg/mL. The intra-and interassay precision values were 3.3% and 7.7%, respectively. The serum S100B assays were conducted at the Ilsong Institute of Life Science at Hallym University. The assays were performed by one technician who was blinded to the study groups.

Statistical Analysis
Statistical analyses were performed using SPSS version 24 (IBM Corp, Armonk, NY, USA). Quantitative variables are expressed as the mean ± standard deviation (SD). ANOVA with a Scheffe post hoc test was used to compare three groups. Correlations between bio-chemical factors and S100B were evaluated using Pearson's correlation. A p value less than 0.05 was considered significant, and all p values were two-tailed. Predictive factors for 1-year mortality were analyzed with uni-and multivariate logistic regression models. Variables with p < 0.05 in the univariate model were entered into the multivariate stepwise logistic regression model (forward selection and likelihood ratio) to assess the independent predictive effect of the variables of interest.

Patient Characteristics
During the study period, 95 patients with LC were prospectively included. The mean age of the participants was 52.77 ± 10.23 years, and 70 (73.7%) patients were male. The etiologies of LC included viral (n = 24), alcoholic (n = 63), and other causes (n = 8). The severity of liver dysfunction was described by the Model for End-Stage Liver Disease (MELD) and the Child-Turcotte-Pugh (CTP) classification. In 22 patients (23.2%), MHE was diagnosed, and overt HE was observed in 25 (26.3%) patients. Moreover, 48 patients were classified as being free of HE. The baseline clinical characteristics of the three groups are shown in Table 2.

Serum S100B and Ammonia Levels According to the Degree of HE
Because S100B has been demonstrated to be a potential marker of brain injury, we evaluated the correlation between serum S100B levels and the degree of HE. The serum S100B levels were 38.1 ± 17.2, 36.5 ± 13.9, and 43.5 ± 19.7 pg/mL in the non-HE, MHE, and overt HE groups, respectively (Table 2). In patients with overt HE, serum S100B levels were slightly high, but there were no significant differences among the three groups. Serum ammonia levels were a better predictor for overt HE than S100B ( Figure 2). Serum ammonia levels were significantly higher in the overt HE group than in the non-HE group; however, serum ammonia level could not discriminate patients with MHE from the non-HE group or the overt HE group. Comparision of serum S100B and ammonialevels according to the severity of HE; (A): There was no significance differences in serum S100B among the three groups. (B): Serum ammonia levels showed significant difference only between the overt HE group and the non-HE group.

Parameters Correlated with Serum S100B Levels
To identify clinical parameters associated with increased S100B levels, patients were divided into three groups according to serum S100B levels: <25 pg/mL, 25~<34 pg/mL, and ≥34 pg/mL. The group with serum S100B ≥34 pg/mL was characterized as follows: older; higher levels of bilirubin, creatinine, and MELD score; lower levels of platelet; and relatively fewer males (Table 3). The relationships of serum S100B levels with clinical and biochemical parameters are shown in Table 4. Serum total cholesterol, platelet, and albumin levels negatively correlated with serum S100B levels. In contrast, age, serum bilirubin, uric acid, lactate dehy-drogenase (LD), alanine aminotransferase (ALT), INR, and creatinine levels showed a positive correlation with serum S100B levels. Among these, serum creatinine, bilirubin, and INR are the main parameters of the MELD scoring formula. For this reason, it was presumed that a significant relationship was observed between the MELD score and serum S100B levels. To avoid the multicollinearity between individual parameters (i.e., INR, bilirubin, and creatinine) and the MELD score, the multivariate linear regression analysis was performed, excepting the MELD score (Table 5). Age, serum bilirubin, and creatinine levels were revealed to be significantly related to serum S100B levels.

Serum S100B Level and 1-Year Mortality
Among 95 patients, 21 patients (22.1%) died after 1 year. To identify 1-year mortalityassociated factors, clinical characteristics were compared between surviving (n = 74) and non-surviving patients (n = 21) ( Table 6). The non-surviving patients were characterized as follows: older; higher levels of S100B, ammonia, white blood cell (WBC), bilirubin, and INR; and lower levels of cholesterol and albumin. The MELD and CTP scores were significantly higher in the non-surviving group compared to the surviving group. There was no significant difference in the distribution of HE between the two groups. Multivariate analysis demonstrated that high ammonia levels, low cholesterol levels, and a combination of high MELD score and serum S100B levels were independently associated with 1-year mortality ( Table 7). Because S100B showed a higher association with MELD than the severity of HE, we investigated whether S100B contributes to predicting 1-year mortality. Using the ROC curve, the cutoff values of the MELD score and serum S100B level related to 1-year mortality were determined to be 13 and 35 pg/mL, respectively. We divided the patients into the following four groups: (i) S100B < 35 pg/mL and MELD score < 13; (ii) S100B ≥ 35 pg/mL and MELD score < 13; (iii) S100B < 35 pg/mL and MELD score ≥ 13; and (iv) S100B ≥ 35 pg/mL and MELD score ≥ 13. The Kaplan-Meier survival curve showed significantly higher 1-year mortality rates when the S100B level and MELD score were higher than the cutoff value ( Figure 3A). To clarify the predictive power of S100B for 1-year mortality, we divided study subjects into three groups depending on the levels of serum S100B (<25 pg/mL, 25 ≤ S100B < 35 pg/mL, and 35 pg/mL ≤). The higher the S100B levels, the higher mortality rates were observed ( Figure 3B).  When the S100B level and MELD score were high, 1-year mortality rates were significantly higher than other groups. (B): Even without the MELD score, the higher the S100B levels, the higher mortality rates were observed.

Clinical Significance of Serum S100B in the Non-HE Group
To clarify the clinical significance of serum S100B levels, we investigated the relationship between serum S100B and biochemical parameters in 48 patients without any types of HE. Total bilirubin and the MELD score showed a positive correlation with serum S100B levels ( Table 8). Among the 48 patients, seven patients (14.6%) died during 1 year. Interestingly, a significant portion of non-surviving patients (71.4%) showed high serum S100B levels and MELD scores above the cutoff value previously mentioned (Figure 4). However, when S100B (≥35 pg/mL) or MELD score (≥13) was applied separately, there was no statistical significance.

Discussion
In the present study, serum S100B level was not useful for differentiating the severity of HE. However, serum S100B levels were associated with age and the main parameters of the MELD score, such as serum bilirubin and creatinine levels in cirrhotic patients. Additionally, serum S100B levels showed the potential in the prediction of 1-year mortality in cirrhotic patients.
LC is a life-threatening disorder leading to 1.03 million deaths per year worldwide [18]. Patients with decompensated LC characterized by variceal bleeding or ascites have a 10fold higher risk of death than the general population [19]. The CTP and MELD scores have been widely used as noninvasive tools for predicting the prognosis of LC. Because the CTP scoring system includes rather subjective variables, such as the degree of evaluation of ascites and HE, the MELD score is considered a more reproducible and accurate tool than the CTP score. In the present study, serum S100B levels showed a positive correlation with MELD scores, and the relationship was more distinct than that with CTP scores.
The biological half-life of S100B has been reported to be in the range of 30-132 min, depending on the clinical situation [20][21][22]. Similar to other small proteins, S100B is presumed to be eliminated by degradation in the proximal tubules in the kidney. Therefore, impaired renal function may affect the serum levels of S100B. Although a significant difference in renal function was not observed among our study groups, the serum S100B level was correlated well with the serum creatinine level, which is one of the main parameters of the MELD scoring system. Regarding the elimination of S100B, Jönsson et al. reported that S100B levels were not influenced by a moderate decrease in glomerular filtration rate (GFR) [22]. At the same time, Gross et al. demonstrated that serum S100B was negatively related to creatinine clearance [23]. In cirrhotic patients, the serum creatinine level is not an accurate marker for estimating renal function due to various factors, such as reduced muscle mass, increased tubular secretion of creatinine, and impaired production of creatine, which is the precursor of serum creatinine. Therefore, to establish the relationship between S100B levels and renal dysfunction in cirrhotic patients, further study using other renal biomarkers, such as cystatin C, is necessary.
In the present study, total serum bilirubin level was associated with serum S100B levels. Similar to our result, Okumus et al. found a significant correlation between serum bilirubin and S100B levels [24]. However, it is unclear how serum bilirubin levels affect serum S100B levels. Serum bilirubin levels have been considered an important factor in the diagnosis and prognosis of patients with liver disorders. Almost all prognostic scores for liver diseases include bilirubin in their calculations. These findings suggest that liver dysfunction can affect serum S100B levels. Additionally, the present study showed that age was associated with serum S100B levels. Similar to our data, Gross et al. found that serum S100B was positively related to age [23]. On the other hand, Portela et al. demonstrated that a correlation between age and S100B was not evident in individuals >20 years of age but was present in individuals <20 years of age [25]. Therefore, further study is needed to clarify the effect of age on serum S100B levels. To date, most of the research on the clinical use of S100B has been conducted in patients with neurologic disorders or some malignancies [26][27][28][29]. Regarding liver diseases, there have been a few studies reporting serum S100B as a neuromarker of HE in patients with fulminant hepatitis [16,17,30]. Especially, Toney et al. investigated that individual neuromarkers derived from specialized cell types within the brain, such as neuron-specific enolase (NSE) present in neurons, S100B present in astrocytes, and myelin basic protein (MBP) in oligodendrocytes, would be associated with the development of HE in pediatric acute liver failure [16]. They found that only S100B was associated with HE. Additionally, they showed no relationship between S100B and the severity of HE, similar to the present study. There is another study using samples from 54 subjects in the US Acute Liver Failure Study Group. In that study, Vaquero et al. found that S100B was increased in subjects with (i) stage 1-2 HE who did not progress, (ii) stage 1-2 HE who progressed to a severe encephalopathy, (iii) stage 3-4 HE who survived and (iv) stage 3-4 HE who died or required transplantation [31]. However, a significant difference in S100B level according to the severity of HE was not observed. Hepatic stellate cells, which share many functional and morphologic characteristics with glial cells, were suggested as another relevant source of serum S100B in the setting of liver failure.
Taken together, obvious astrocyte swelling and damage developed in acute liver failure may release S100B protein; thus, elevated serum S100B may be a marker suggesting severe HE. However, based on the results of the present study, the elevated serum S100B levels observed in fulminant hepatic failure may be the result of not only astrocyte injury but also liver damage itself.
HE is a typical cirrhosis-related complication and is associated with repeated readmission and, more importantly, high rates of mortality [32,33]. Although MHE is the mildest form in the HE spectrum, it is associated with low health-related quality of life (QOL) and worse survival [34,35]. Because patients with MHE have no clinically detectable neurological-psychiatric abnormalities, neuropsychological or electrophysiological tests are required to diagnose MHE [36]. Under real-world conditions, these diagnostic proce-dures are cumbersome and time-consuming for clinicians. Therefore, the identification of serologic markers for diagnosing MHE will benefit the current clinical practice for cirrhotic patients. Unfortunately, in the present study, serum S100B levels did not show diagnostic value for MHE or overt HE. Ammonia levels also failed to show a direct correlation with the severity of HE.
However, in the present study, serum S100B levels had a significant relationship with the MELD score regardless of the presence of HE. It is uncertain how serum S100B levels are elevated in cirrhotic patients without brain injury. Cell damage followed by apoptosis or necrosis, which occurs in the liver, is presumed to be the possible source of blood S100B protein. S100B is known to play various roles through intracellular and extracellular functions [37]. In our previous experimental study, we found that hepatic stellate cells, the key players in liver fibrosis, are activated by recombinant S100B protein treatment [12]. Recently, another study has elucidated the role of S100B in hepatocellular carcinoma (HCC) [38]. Yan et al. showed that S100B expression is correlated with hypoxia and the immune response in a human HCC cell line. Therefore, S100B may be involved in various liver diseases. To clarify the mechanism of S100B in liver diseases, further research is needed.
The present study indicated that the combination of a high MELD score and S100B levels, high ammonia levels, and low cholesterol levels were associated with 1-year mortality in cirrhotic patients. Recently, Tranah et al. reported that plasma ammonia is an independent predictor of hospitalization with liver-related complications and mortality in clinically stable outpatients with cirrhosis through a prospective cohort study [39]. As a gut-derived neurotoxin, hyperammonemia resulting from impaired metabolism in chronic liver diseases leads to cerebral edema and intracranial pressure elevation [40,41]. Additionally, ammonia is known to be related to the pathogenesis of liver-related complications, including liver cell damage, immune dysfunction, sarcopenia, and portal hypertension [42]; thus, together with S100B, ammonia may be an interesting indicator of LC. Similar to our findings, Feng et al. reported that decreased total cholesterol is significantly associated with reduced survival in cirrhotic patients [43].
The results of the present study should be interpreted considering its limitations. First, the sample size was relatively small, but this study was conducted prospectively and used proper neuropsychological tests to discriminate MHE and overt HE. In the present data, the prevalence of MHE and mortality-related factors in cirrhotic patients were similar to those of recently published studies, reflecting the adequacy of the present study. The sample size of the non-HE group was also too small to perform survival analysis. Nevertheless, our results showed that S100B is worthy of further study in cirrhotic patients without any types of HE. Second, the value of serum S100B in the age-matched healthy control group was not investigated. Furthermore, seasonal variation (summer/winter) or genetic factors (S100B single nucleotide polymorphism rs9722 and S100B haplotype T-G-G-A) could affect the levels of serum S100B [44,45]. Therefore, in liver disease, subsequent research on the clinical value of S100B should include a normal control group and be designed considering the source of variation affecting the levels of serum S100B concentrations. Lastly, in the present study, MHE was diagnosed using MMSE and PHES. Recently, two abnormal tests among neuropsychological assessments (i.e., PHES, critical flicker frequency test, inhibitory control test, and Stroop test et al.) have been required to diagnose MHE [46]. Unfortunately, the present study was designed in early 2013. At that time, no agreement existed as to what combination of tests should be carried out to confirm MHE. We followed a general approach to the diagnosis of MHE based on Ferenci et al. [47]. Therefore, subsequent research using more various neuropsychological assessment tools is necessary.
In summary, this is a negative study showing that serum S100B is not useful for MHE diagnosis; however, the present study found that serum S100B levels can be affected by age, serum bilirubin, and creatinine in cirrhotic patients, and associated with MELD score. Additionally, serum S100B levels showed the possibility of predicting 1-year mortality in cirrhotic patients. In the future, basic and clinical studies are necessary to identify the effect of liver dysfunction on serum S100B levels in various liver diseases.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of Hallym University Sacred Heart Hospital (2013-I103).

Informed Consent Statement:
Informed consent was obtained from all subjects or legal representatives involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available because they are property of the Institute for Liver and Digestive Diseases, Hallym University, Republic of Korea.

Conflicts of Interest:
The authors declare no conflict of interest.