AIM: To evaluate the association between serum concentrations of S100β in patients with cirrhosis and the presence of low grade hepatic encephalopathy (HE).

METHODS: This was a cross-sectional study. The population was categorized into four groups healthy subjects, cirrhosis without HE, cirrhosis with covert hepatic encephalopathy (CHE) and cirrhosis with overt HE. Kruskal-Wallis, Mann Whitney’s U with Bonferroni adjustment Spearman correlations and area under the ROC were used as appropriate.

RESULTS: A total of 61 subjects were included, 46 cirrhotic patients and 15 healthy volunteers. S100β values were different among all groups, and differences remained significant between groups 1 and 2 (P < 0.001), and also between groups 2 and 3 (P = 0.016), but not between groups 3 and 4. In cirrhotic patients with HE S100β was higher than in patients without HE [0.18 (0.14-0.28) ng/mL vs 0.11 (0.06-0.14) ng/mL, P < 0.001]. There was a close correlation between serum concentrations of S100β and psychometric hepatic encephalopathy score in patients with cirrhosis without HE compared to the patients with cirrhosis with CHE (r = -0.413, P = 0.019). ROC curve analysis yielded > 0.13 ng/mL as the best cutoff value of S100β for the diagnosis of HE (sensitivity 83.3%, specificity 63.6%).

CONCLUSION: Serum concentrations of S100β are higher in patients with cirrhosis than in healthy volunteers, and are further increased in the presence of hepatic encephalopathy. The results suggest that serum biomarkers such as S100β could help in the correct characterization of incipient stages of HE.

Hepatic encephalopathy (HE) reflects a spectrum of neuropsychiatric abnormalities seen in patients with liver dysfunction after exclusion of other brain diseases[1,2]. Covert hepatic encephalopathy (CHE) is the initial stage in the clinical spectrum of HE and involves subtle changes in cognitive and motor function, and electroencephalographic patterns. In contrast to patients with symptomatic HE, patients with CHE have no recognizable clinical symptoms of brain dysfunction and need to be diagnosed on the basis of neuropsychologic tests[3]. The prevalence of CHE in cirrhotics with good liver function (Child A) is 15%, while in those with advanced cirrhosis (Child B/C) it approaches 50%[4-6]. Multiple pathways are involved in the pathophysiology of HE, however, hyperammonemia plays a central role. Astroglial swelling in HE occurs if the load of glutamine accumulating during ammonia detoxification is not compensated by a decrease in other osmolytes like myo-inositol[7]. This is followed by development of low-grade cerebral edema, which is accompanied by an increased production of reactive oxygen and nitrogen species, triggering multiple protein and RNA modifications, thereby affecting brain function[8]. Associated abnormalities in the homeostasis of cerebral neurotransmitters and blood-brain barrier (BBB) disruption have not been well characterized yet. Different molecules have been addressed as markers of brain injury, some of them being linked to astrocyte damage, but to date there is no useful serologic marker that can aid the diagnosis of HE or CHE.

S100β is a 10.4 kDa protein that is primarily synthesized in the brain by the endfeet processes of the astrocytes, and it belongs to a superfamily of low molecular weight acidic calcium binding proteins of the EF-hand type[9]. This protein is primarily metabolized in the kidney and excreted in the urine, its expression across ethnic groups seems to be consistent, differences between genders are not found, and it does not seem to show circadian variation[10]. Although S100β is not entirely specific for the central nervous system (CNS), it is present in the brain at much higher concentrations than in other tissues (80%-90% of the total pool is found within the brain), and as such, this protein can be used as an early marker of brain damage[9].

Astrocytes are key in homeostasis regulation in the CNS, being activated after brain injury and releasing S100β[11]. Some studies have shown increased levels of S100β in patients with acute or chronic liver failure and HE which might reflect an early stage of intracerebral changes before the development of marked cerebral edema[12,13]. Moreover it has been proposed that increased serum concentrations of S100β could predict CHE[14]. However, evidence associating S100β levels and the presence of HE is scarce. The aim of this study was to evaluate the association between serum concentrations of S100β and the presence of low-grade HE in patients with cirrhosis.

A total of 61 subjects were included, 46 cirrhotic patients and 15 healthy volunteers. The main characteristics of the population are shown in Table 1. Median age was higher in cirrhotics compared to the healthy subjects, most patients were males. The main etiology of cirrhosis was HCV and there was a stepwise increase in Child-Pugh and MELD score according to the presence and severity of HE.

The distribution of S100β serum values according from each group is shown in Figure 1. S100β values were different among all groups, and differences remained significant between groups 1 and 2 (P < 0.001), and also between groups 2 and 3 (P = 0.016), but not between groups 3 and 4. In cirrhotic patients with HE S100β was higher than in patients without HE [0.18 (0.14-0.28) ng/mL vs 0.11 (0.06-0.14) ng/mL, P < 0.001].

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Serum concentrations of S100β (ng/mL) among groups. (1) Healthy subjects [0.05 (0.04-0.07)]; (2) Cirrhosis without HE [0.11 (0.06-0.14)]; (3) Cirrhosis with covert HE [0.16 (0.13-0.23)]; (4) Cirrhosis with overt HE [0.18 (0.14-0.28)]. Group 1 vs group 2, P < 0.001; Group 2 vs group 3, P < 0.016; Group 3 vs 4, P = 0.508. Healthy subjects vs all groups [0.14 (0.098-0.198)], P < 0.001.

The results of the HE psychometric evaluation can be observed in Table 2. There was a clear deterioration in the performance of the CFF and PHES tests in the group with cirrhosis and cirrhosis with CHE or HE, as well as worsening in the individual PHES tests with a statistically significant difference.

In the total population, S100β was found to have a moderate correlation both with PHES (rho = -0.624, P < 0.001) and CFF (rho = -0.516, P < 0.001). In the subgroup of patients with cirrhosis, S100β showed moderate correlations with PHES (rho = -0.528, P < 0.001), as well as with CTP score (rho = -0.515, P < 0.001), and MELD score (rho = -0.440, P < 0.001), but not with CFF (rho = 0.240, P = 0.108) (Figure 2).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 S100β and PHES correlations between patients with cirrhosis and covert HE and non-covert HE. r = -0.413, P = 0.019. HE: Hepatic encephalopathy; PHES: Psychometric hepatic encephalopathy score.

ROC curve analysis yielded > 0.13 ng/mL as the best cutoff value of S100β for the diagnosis of CHE and HE, with sensitivity of 83.3% and specificity of 63.6%. Area under the ROC curve (AUROC) was 0.801 (P < 0.0001).

HE and CHE are associated with decreased quality of life, impaired working performance, and patients become unfit to safely drive motor vehicles[16-21]. Cirrhotic patients with CHE are prone to develop episodes of overt HE when compared to those without CHE (incidence of 56% for those with CHE vs 8% for those without CHE at 3 years of follow up)[22]. Moreover, CHE is associated with a decreased survival, and thus it might be considered within the spectrum of complications definying advanced liver failure[23]. It is not known whether treatment of CHE may prevent overt HE or improve survival, but it is tempting to hypothesize that early identification of CHE followed by proper treatment should translate into improved quality of life, and have a favorable impact in prognosis. Psychometric tests are the preferred tool to diagnose CHE, and, since the final PHES score is adjusted to age and years of education, the differences of these two variables in our population did not have an impact in the results of our study.

S100β is not entirely specific for the CNS, but it is present in the brain at much higher concentrations than other tissues (80%-90% of the total pool is found within the brain), and as such, this protein can be used as a marker of brain damage[9]. Other places where S100β synthesis has been identified are adipose tissue[24], skin and melanoma tumors[25,26] and T-lymphocytes[27]. S100β protein is secreted by activated astrocytes and the mechanism of secretion remains unknown but astrocyte expression has been reported to be stimulated by interleukin-1 (IL-1) and by cyclic-AMP. The effects of secreted glial S100β depends on its concentration, being neurotrophic at low levels (nanomolar) and neurotoxic at high levels (micromolar)[9]. Nanomolar levels of S100β exert a stimulatory effect on astrocytes, causing glial proliferation in vitro[28]. S100β also accumulates in the extracellular space after astrocyte death or after cellular disintegration of the damaged parenchyma. Under these conditions, the S100β concentration may well be in the micromolar range and the protein may become toxic. This has originated the hypothesis that extracellular accumulation of S100β in ongoing brain insults can, in combination with other unknown factors, cause a shift in its neuronal attributes from protective to hazardous[9].

We found significant differences in the serum concentrations of S100β between healthy volunteers and patients with cirrhosis. Moreover, in patients with cirrhosis, the levels of S100β increase in the presence of CHE and overt HE. The correlation between S100β and PHES indicates that at lower PHES scores, indicating a higher degree of HE, there is an increased concentration of S100β in serum. This suggests increased permeability of the BBB as the cognitive deterioration of HE progresses, what would determine not only leaking of S100β into the blood circulation, but increased ammonia delivery to the brain, leading to astrocyte activation and swelling. All of these changes (astrocyte activation and swelling, increased BBB permeability) seem to be present in patients with cirrhosis, and, moreover, seem to be more notorious in the presence of CHE and HE. However, our results would suggest that even in the absence of overt HE and CHE, the cirrhotic patient is at a state of astrocyte activation and swelling, and has increased BBB permeability. Brain adaptation to ammonia insults during early stages of cirrhosis and portal hypertension might consitute the counteracting condition preventing CHE or overt HE[29-31]. Our understanding of the cellular mechanisms leading to astrocyte dysfunction and neurotoxicity is limited, and it is unclear whether BBB disruption is crucial in the etiopathogenesis of HE, or an inevitable consequence of the disease itself. More studies are needed to better understand the role of S100β in the origin and perpetuation of HE in cirrhotic patients.

When we compared the cirrhotic patients without HE and CHE we found that S100β could differentiate both groups of patients with a fair accuracy, as shown by an AUROC curve of 0.801. Moreover, S100β correlated with the neuropsychometric tests (PHES and CFF), and with the hepatic functional status (MELD and CTP), attesting for its possible role in the evaluation of HE. Our results suggest that S100β can be useful in the diagnosis of HE, precisely in clinical scenarios where the diagnosis of HE remains unclear because behavior changes being subtle, evaluation of highly-educated patients, or when other psychiatric abnormalities (i.e., depression) coexist or cannot be ruled out.

In conclusion, serum concentrations of S100β are higher in patients with cirrhosis when compared to healthy volunteers, and are further increased in the presence of CHE and overt HE. The results suggest the presence of astrocyte dysfunction, and maybe BBB disruption too, in patients with cirrhosis that seems to be more severe and progressive in the presence of HE. S100β showed significant differences between patients with cirrhosis but no encephalopathy at all, and the subgroup presenting with CHE, suggesting its usefulness to identify CHE. Although the AUROC curve was modest, S100β would potentially be useful in the diagnosis of low-grade HE whenever neuropsychometric tests are suboptimal, or when competing psychiatric differential diagnoses are in place.

The authors would like to thank Dr. Mick Frissen for IT support and Dr. Norberto Chavez Tapia for his assistance in the initial conceptualization of this project.