Introduction

Liver cirrhosis (LC) patients present with osmoregulation abnormalities in the brain caused by ammonia (NH3) hypermetabolism. This NH3 hypermetabolism results in astrocyte enlargement and cerebral edema and is thus deeply involved in the pathogenesis of hepatic encephalopathy (HE) (Rovira et al. 2001; Mardini et al. 2011; Iwasa and Takei 2015). Hyponatremia, meanwhile, is a complication commonly seen in LC patients that contributes to reduced quality of life and poor vital prognosis (Heuman et al. 2004; Solà et al. 2012). Hyponatremia is reported to act as an exacerbating factor for cerebral edema and HE because it reduces osmotic pressure (Ahluwalia et al. 2015a; Iwasa et al. 2014). The involvement of inflammation in the pathogenesis of HE has also been suggested by the high rate of HE exacerbation in patients who present with concomitant systemic inflammatory response syndrome (SIRS), and by the correlation between electroencephalographic findings and blood inflammatory cytokine levels in LC patients (Shawcross et al. 2011; Montagnese et al. 2011). In addition to these clinical findings, in vitro experimental systems have shown that exposing astrocytes to NH3, low sodium (Na) levels, and inflammatory cytokines induces cell enlargement (Häussinger and Schliess 2005; Reinehr et al. 2007). This suggests that elevated NH3, together with electrolyte abnormalities and inflammation, is intricately involved in the onset of HE (Romero-Gómez et al. 2015).

HE is diagnosed based on psychometric testing abnormalities and has a wide range of stages from minimal HE to overt HE and coma (Vilstrup et al. 2014). Grade I HE is technically overt HE, but it is often difficult to clinically determine Grade I, and there have also been recent proposals to refer to minimal HE and Grade I HE collectively as “covert HE.” (Vilstrup et al. 2014; Bajaj et al. 2011) Opinions are divided regarding whether to intervene therapeutically in minimal HE and covert HE patients. Treatment is usually indicated when overt HE is seen or when the LC patient has a history of overt HE (Kappus and Bajaj 2012; Bajaj 2010). However, once a patient develops overt HE, the life expectancy of LC patients is poor (Romero-Gómez et al. 2004; Amodio et al. 2001). Identifying factors associated with the onset of HE is therefore considered an important challenge.

This study examined the relationships between covert HE and blood NH3, Na, and C-reactive protein (CRP) in LC patients who underwent psychometric testing. It also investigated the effects elevated NH3, hyponatremia, elevated CRP, and decreased hepatic functional reserve on the development of overt HE.

Subjects and methods

Subjects

The subjects were 40 LC patients without ≥ Grade II HE (20 men, 20 women, mean age: 66 ± 9 years). This group of LC patients was the same as that used in a clinical study on HE diagnosis by magnetic resonance imaging (MRI) in 2008 (Sugimoto et al. 2008). The cause of LC in these patients was hepatitis B virus infection in 3 patients, hepatitis C virus infection in 30 patients, alcoholism in 5 patients, and primary biliary cirrhosis in 2 patients. Child-Pugh classifications of liver severity were Class A in 12 patients, Class B in 25 patients, and Class C in 3 patients. This group included 3 patients with a history of overt HE, and the 3 patients with alcoholism had all stopped drinking from at least 3 months before the start of the study.

Neuropsychological tests and laboratory examinations

The Trail Making Test A (number connection test) and the Digit Symbol Substitution Test were used to test psychometric function. Abnormal results in both tests indicated covert HE, while normal results in both tests indicated no HE. Plasma NH3, serum Na, and high-sensitivity CRP were measured early in the morning on an empty stomach on the same day as the psychometric function tests. The relationships between covert HE and blood NH3, Na, and CRP in patients who underwent psychometric testing were examined.

Follow-up

Patients were asked to visit the hepatology outpatient department once a month to be examined for the presence or absence of disorientation and asterixis and to diagnose them with overt HE, if present. Drugs involved in the onset of HE, such as branched-chain amino acid formulations and synthetic disaccharides, were continued without changing the dosages. It investigated the effects elevated NH3, hyponatremia, elevated CRP, and decreased hepatic functional reserve on the development of overt HE. This study was in compliance with the Declaration of Helsinki.

Statistical analysis

Results are expressed as means ± standard deviation. The chi-squared test and Mann-Whitney U test were used to test for differences between the two groups. Spearman’s or Pearson’s correlation coefficients were used to examine correlations between variables. We used receiver-operating characteristic (ROC) curves to determine the optimum cut-off of blood NH3, Na, and CRP for predicting overt HE onset. The Kaplan-Meier method and the log-rank test were used to investigate the risk overt HE onset. Analyses were performed using a Cox proportional hazards model. The analysis of variance was performed using Tukey-Krame test for multiple comparisons among the four groups.

Results

Assessment of HE severity and clinical test values

Following psychometric testing, 10 patients were deemed to have covert HE, and 20 patients were deemed to have no HE. No significant differences were seen between the two groups in age, sex, cause of LC, history of overt HE, liver severity, or blood NH3 level (covert HE: 57 ± 49 μmol/L vs. no HE: 37 ± 23 μmol/L; Fig. 1a) (Sugimoto et al. 2008). By contrast, the Na level was 137 ± 4 mEq/L in the covert HE group and significantly higher, at 141 ± 2 mEq/L, in the no HE group (p < 0.01; Fig. 1b). CRP was significantly higher in the covert HE group than in the no HE group (1.85 ± 1.93 mg/dL vs. 0.59 ± 0.93 mg/dL; p < 0.05; Fig. 1c). However, none of the patients with an elevated CRP level had clear infection foci of spontaneous bacterial peritonitis, upper respiratory infection, urinary infection, or other infections. In all patients, additional cross-correlation examinations between blood NH3, Na, and CRP showed no significant correlations between NH3 and Na or between NH3 and CRP, but a strong negative correlation was found between Na and CRP (r = −0.527, p < 0.001; Fig. 2).

Fig. 1
figure 1

Comparison of plasma ammonia (NH3) (a), serum sodium (Na) (b) and serum high-sensitivity C-reactive protein (CRP) (c) levels in covert hepatic encephalopathy (HE) patients and no HE patients. The covert HE group has significantly lower serum Na and significantly higher serum CRP levels

Full size image
Fig. 2
figure 2

Correlations between blood test values in all patients. Correlations between plasma ammonia (NH3) and serum Na (a), NH3 and serum high-sensitivity CRP (b), and Na and CRP (c) are examined. There are no correlations between NH3 and Na and between NH3 and CRP, but a significant negative correlation is seen between Na and CRP

Full size image

Factors in the prediction of the risk of developing overt HE

Next, the associations of psychometric testing values and clinical test values with the onset of overt HE were analyzed. During the median observation period of 11 months, 10 patients developed overt HE and remaining 30 patients did not develop. NH3 was significantly higher in the overt HE group than in the no HE group (68 ± 46 μmol/L vs. 37 ± 25 μmol/L; p < 0.05) (Fig. 3a), although there were no significant differences between the two groups in serum Na (overt HE: 139 ± 4.0 mEq/L vs. no HE: 140 ± 2.0 mEq/L) and CRP level (overt HE: 1.19 ± 1.46 mg/dL vs. no HE: 0.68 ± 0.90 mg/dL) (Fig. 3b, c). The NH3 ROC curve analysis for predicting overt HE onset showed a cut-off of 50 μmol/L offered 78.3 % sensitivity and 62.5 % specificity (supplementary Figure 1). For prediction of overt HE onset, a cut-off value of 139 mEq/L for Na or 0.25 mg/dL for CRP resulted in a sensitivity and specificity of 87.0 % and 37.5 % or 52.2 % and 62.5 %, respectively (supplementary Figure 1). HE patients deemed to have covert HE had a high rate of development of overt HE (Fig. 4a, p < 0.001). Among the blood test values, an NH3 level > 50 μmol/L contributed to the development of overt HE (Fig. 4b, P < 0.05). Patients with a Na level < 139 mEq/L and CRP level > 0.25 mg/dL also had a tendency to develop overt HE, although these values were not significant (Fig. 4c, p = 0.122; Fig. 4d, p = 0.06). In the relationship between the severities of hepatic functions and the occurrence of overt HE, a Child-Pugh classification of B or C was not associated with HE onset (Fig. 4e, p = 0.22).

Fig. 3
figure 3

Comparison of plasma ammonia (NH3) (a), serum sodium (Na) (b) and serum high-sensitivity C-reactive protein (CRP) (c) levels in overt hepatic encephalopathy (HE) patients and no HE patients. NH3 was significantly higher in the overt HE group than in the no HE group

Full size image
Fig. 4
figure 4

Comparison of the presence or absence of overt HE, stratified by presence or absence of covert HE (a), plasma NH3 > 50 μmol/L or ≤50 μmol/L (b), serum Na < 139 mEq/L or ≥139 mEq/L (c), serum CRP > 0.25 mg/dL or ≤0.25 mg/dL (d), and Child-Pugh classification of A or Child-Pugh classification of B or C (e). Factors contributing to the onset of overt HE are covert HE and NH3 > 50 μmol/L

Full size image

Univariate analyses using a Cox proportional hazards model with presence or absence of covert HE and NH3, Na, and CRP levels as variables showed p values of <0.01, < 0.01, 0.123, and 0.053 for each of the variables, respectively. The p values for presence or absence of covert HE and NH3 were significant (Table 1). Multivariate analysis using the presence or absence of covert HE, NH3, and CRP as variables showed significance for the presence of covert HE and the NH3 level, with hazard ratios of 4.65 and 1.02, respectively (Table 1).

Table 1 Risk factors of overt HE in patients with liver cirrhosis
Full size table

We evaluated whether length of time to the onset of overt HE is different according to risk factor. The length of time of patients with covert HE, elevated NH3, hyponatremia, and elevated CRP were 7.7 ± 10.0, 5.8 ± 7.8, 8.8 ± 10.7, and 8.0 ± 8.8 months, respectively and no difference was found.

Discussion

This study demonstrated that psychometric testing results, particularly the presence of covert HE, and serum Na and CRP levels were associated with development of overt HE. In LC patients with hyponatremia, cerebral edema induced by reduced extracellular osmolarity is already known to be a potential exacerbating factor for HE onset (Iwasa and Takei 2015). Bajai et al. recently reported that short-term administration of tolvaptan in LC patients with hyponatremia simultaneously raises the Na level and improves brain volume on brain MRI and psychometric testing, and they reiterated the importance of Na control in the management of HE (Ahluwalia et al. 2015b). Meanwhile, other reports claimed that inflammation also has a stimulatory effect on HE. Patients with concomitant SIRS frequently experience exacerbations of HE (Shawcross et al. 2011), and in minimal HE, it has already been reported that CRP and IL-6 levels are high (Shawcross et al. 2007). Bacterial translocation occurs frequently in LC patients. These patients have a high NH3 level and may develop HE via local neuroinflammation in the brain (Luo et al. 2015). In the present study, no direct correlation was seen between blood NH3 and CRP, suggesting the need for examinations using arterial NH3, IL-6, and tumor necrosis factor alpha (Jain et al. 2013).

No study appears to have examined whether hyponatremia and mild inflammation contribute to the onset of overt HE in covert and minimal HE patients. The present study identified covert HE and elevated blood NH3 as factors associated with the onset of overt HE, but serum Na or CRP were not found to be factors. These results support the hypothesis that HE is essentially enlargement of astrocytes and cerebral edema caused by sustained NH3 metabolism, and that osmotic pressure abnormalities and neuroinflammation due to low Na are exacerbating factors. It has been suggested that, in addition to traditional NH3 countermeasures, hyponatremia countermeasures, such as water intake restrictions and adjustment of diuretics (Iwasa and Takei 2015), and bacterial translocation countermeasures, including intestinal sterilization, are needed to control overt HE (Dhiman et al. 2014). Furthermore, while many patients in the present study developed overt HE during the early stage of follow-up, LC associated with Grade I HE may have been included as a cause of this.

This study had several limitations. This study examined a small sample at a single facility, the latest computer systems were not used to test psychometric function (Kato et al. 2004), sensitive inflammatory indices such as endotoxins and inflammatory cytokines were not measured, and the subjects were not limited to patients with minimal HE. Nonetheless, the results showing that psychometric testing result abnormalities and hyperammonemia are linked to subsequent development of HE are important, and together with the finding that electrolyte abnormalities and mild inflammation contribute to the pathogenesis of HE, these findings may serve as extremely valuable information in routine clinical settings.