Brain dysfunction occurs in about 30–80% of patients with cirrhosis and is referred to as hepatic encephalopathy (HE), which is a complex behavioural, neuropsychiatric disorder. Clinically, HE manifestations can range from mild, subclinical disturbances in cognition, emotional regulation, or behaviour (minimal HE [mHE]) to gross disorientation and coma, a condition referred to as overt HE (OHE) (Montagnese et al. 2022). Some of the clinical features of HE overlaps with other brain conditions such a neurodegeneration in the mild forms and with delirium in those with more acute forms. Worldwide, there are about 10 M people with decompensated cirrhosis and about 30–40% of these patients will be hospitalised with OHE each year (Casadaban et al. 2015; Tapper et al. 2019). In Europe, about 200 K people suffer from decompensated cirrhosis and it is estimated that about 65 K patients per year are hospitalised with OHE, with the consequent economic burden on health systems, in addition to the attendant loss of work force (Moon et al. 2020). Both the burden of HE is increasing, and also charges per hospital stay (Elsaid et al. 2020; Louissaint et al. 2022). Importantly, when patients with cirrhosis develop an episode of OHE, they are prone to repeated episodes, the chance of complete reversibility drops, their health-related quality of life declines and the risk of death is abruptly increased up to 20–30% at 1-year (Arguedas et al. 2003; Tapper et al. 2016).

Currently, mHE is diagnosed using tests such as psychometric hepatic encephalopathy score (PHES) or critical flicker frequency, which are time-consuming to perform and interpret. Consequently, they cannot be routinely applied in clinical practice. This highlights the need to identify biomarkers for mHE and OHE that would enable early diagnosis and improved treatment.

Biomarkers are measurable and objectively evaluable characteristics that indicate a normal or pathological biological process or the response to a therapeutic intervention (Biomarkers Definitions Working Group 2001). Their study has significantly advanced in recent decades due to technological progress, molecular biology, and omics approaches, as well as the integration of these fields. Biomarkers are of particular interest in clinical practice for detecting and monitoring various diseases. They can serve as tools for early diagnosis, disease severity stratification, prognosis assessment, and predicting therapeutic response (Mayeux 2004; Jain 2017). Furthermore, studying disease-specific biomarkers is valuable for improving the development of novel therapies and evaluating their efficacy within clinical trials (Rolan 1997; Biomarkers Definitions Working Group 2001).

In the spectrum of HE, the research on specific biomarkers holds special relevance from multiple perspectives. On the one hand, it is crucial for understanding the pathophysiology of the condition from its early stages and its progression to more severe phases, facilitating the development of therapeutic targets and early-phase clinical trials. On the other hand, the need to identify a reliable biomarker is even more critical in mHE, as these biomarkers could provide a valuable tool for the early diagnosis in a simple and rapidly applicable manner. This would enable earlier treatment, significantly enhancing its effectiveness in preventing progression to OHE. Even when OHE is already present, biomarkers could prove useful in these stages to assess disease severity or identify patients who would benefit from prophylactic measures.

Ammonia is a product of amino acid metabolism that is known to accumulate in the blood of patients with cirrhosis due to the reduced function of the urea cycle, which is uniquely located in the liver. In health, its circulating levels are tightly controlled. In liver disease, hyperammonemia is commonly observed and recognized to play a pivotal role in the pathogenesis of HE (Felipo and Butterworth 2002; Bosoi and Rose 2009). Therefore, emerging evidence supports the utility of ammonia for risk stratification. However, ammonia measurement is complex and requires careful sample handling, rapid transport to the laboratory in refrigerated conditions and the values obtained are highly variable making it difficult to compare across laboratories. Therefore, the role of ammonia in guiding HE treatment is still unclear and there is equipoise in its use in clinical practice.

Many blood-based biomarkers reflecting these neurotoxic effects of ammonia and liver disease can be measured in the blood allowing the development of new biomarkers to diagnose cirrhosis patients at risk. It is also clear that the effect of ammonia on the brain is modulated by severity of systemic inflammation, which is commonly observed in patients with cirrhosis (Albillos et al. 2014, 2022). Therefore, markers of systemic inflammation and neuronal function may serve as additional biomarkers.

The aim of the present narrative review is to provide the latest evidence on biomarkers of HE beyond ammonia. Tables 1 and 2 present a summary of possible biomarkers for hepatic encephalopathy based on different kind of parameters measured, which will be discussed in the following sections.

Table 1 Summary of possible biomarkers for hepatic encephalopathy based on different kind of parameters measuredTable 2 Summary of biomarkers depending on their capacity for diagnosis and risk and severity prognosis of hepatic encephalopathyAmmonia and its metabolism

Ammonia concentrations in blood and brain are regulated by its synthesis and degradation. In the blood of cirrhotic patients, the primary source of ammonia is the deamination of glutamine by intestinal glutaminase (Damink et al. 2002). In the brain, this deamination occurs in neurons (Márquez et al. 2013). Therefore, glutaminase activity has been studied as a factor related to the development of OHE. For the first time, Romero-Gómez et al. (2010) described variations in a microsatellite region in the promoter of the glutaminase gene associated with OHE in the Spanish population. Specifically, the presence of two long alleles of this microsatellite (≥ 14 repeats; 198 to 210 base pairs) was associated with higher glutaminase activity and with the development of OHE. Although the association between long alleles and OHE is corroborated by another independent study in the Caucasian population (Mayer et al. 2015), these results were not replicated in the East Asian population, indicating that the genetic predisposition given to these microsatellites is not universal and must be validated for each population (Ahn et al. 2017).

Ammonia can be detoxified in the brain and muscle by incorporating it into glutamine via glutamine synthetase. In the liver and intestinal mucosa, glutaminase degrades glutamine to glutamate and ammonia, incorporated into the urea cycle for degradation. The ability to metabolize glutamine to glutamate and ammonia by the intestinal mucosa can be assessed through oral glutamine challenge (OGC). In cirrhotic patients, intestinal glutaminase activity is increased and correlates with mHE (Romero-Gómez et al. 2004), and OGC results in a significant increase in blood ammonia levels, which does not occur in control subjects or those with liver transplants (Oppong et al. 1997). Similarly, an altered response to OGC in patients with mHE is associated with an increased risk of developing OHE (Romero-Gómez et al. 2002). These studies indicate that, beyond blood ammonia levels, it is crucial to investigate different aspects involved in ammonia metabolism as risk factors for the development of OHE and their utility as biomarkers.

Prediction models based on clinical parameters

As previously mentioned, ammonia levels in cirrhotic patients are useful for stratifying the risk of developing OHE. In patients with stable cirrhosis, a 1.4 times ammonia level upper limit of normal (AMM-ULN) or more has been shown to define the risk of future hospitalization with OHE (Tranah et al. 2022; Balcar et al. 2023). Integrating other biochemical and clinical data can enhance its utility as a biomarker for diagnosis and determinate the risk of developing mHE and OHE. Regarding biochemical parameters, albumin levels have been identified as potentially helpful in diagnosing mHE, although their sensitivity is low when considered alone (Demirciler 2023). In this context, the AMMON-OHE model has been developed, which includes variables such as AMM-ULN, sex, diabetes, albumin, and creatinine to identify the risk of developing the first episode of OHE (Ballester et al. 2023). The model achieved a C-index of 0.844 was validated with an external cohort, and is currently available for use in clinical practice (link: https://ammon-ohe.shinyapps.io/ammon-ohe/).

Other scores based on clinical parameters have been developed, such as the CCHE (Clinical Covert Hepatic Encephalopathy) (Labenz et al. 2019), and BABS (Bilirubin–Albumin–Beta-Blocker–Statin) scores (Tapper et al. 2018) (Tables 2 and 3).

Table 3 Construction of a Risk Score for HE with the BABS (bilirubin–albumin–beta-blocker–statin) score

The CCHE score was created to predict covert HE (which comprises mHE and HE grade 1) in cirrhotic patients using variables such as serum albumin levels, clinically detectable ascites, a history of OHE, and scores from the simplified animal naming test and the activity subdomain of the Chronic Liver Disease Questionnaire (Labenz et al. 2019). The CCHE score generates two cutoff points, stratifying patients into low-risk (< 53.5), intermediate-risk (53.5 ≤ CCHE score ≤ 57.5), and high-risk (> 57.5) categories for developing covert HE (see formula in Table 1). The scoring system demonstrates good sensitivity, specificity, and positive and negative predictive values (90%, 91%, 85%, and 94%, respectively).

The BABS score was developed to stratify the risk of developing OHE using biochemical variables (bilirubin and albumin levels) and the patient medication use (non-selective beta blockers and statins) (Tapper et al. 2018) (Tables 2 and 3). Two predictive models were developed for this score, using baseline or longitudinal data. The baseline-data model stratifies the 5-year risk of HE into low (< −10), medium (−9 to 20), and high (≥ 21). A score ≤ −10 is associated with a 27% risk, while a score > −10 corresponds to a risk > 49%. The longitudinal-data model stratifies the 1-year risk of HE into low (< 0), medium (1–20), and high (≥ 21). A score ≤ 0 is associated with a 6% risk, while scores ≥ 1 correspond to a 25% risk.

Biomarkers based in metabolic profile

Beyond ammonia metabolism, other metabolic pathways are disrupted during cirrhosis, mHE, and OHE, leading to changes in metabolite levels in both serum and cerebrospinal fluid (CSF). For example, severe OHE is associated with increased levels of aromatic amino acids (AAA) and methionine in CSF (Cascino et al. 1982).

Studies conducted to distinguish cirrhotic patients with or without mHE based on their serum metabolic signature, showed that mHE patients exhibited increased levels of glucose, lactate, and trimethylamine-N-oxide, primarily, along with elevated levels of glycerol and methionine to a lesser extent. In contrast, patients without mHE were characterized by elevated levels of low-density lipoprotein, choline, alanine, α-acid glycoproteins, valine, acetoacetate, isoleucine, leucine, and glycine. Based on these changes, they developed a model with sensitivity and specificity of 87% and 95%, respectively (Jiménez et al. 2010).

Other studies revealed alterations in 72 metabolites in CSF from patients with OHE, most of which were associated with ammonia metabolism, energy pathways, methylation pathways, and aromatic amino acids, along with an increase in bile acids acids (Weiss et al. 2016), related with glymphatic system, which has been shown to be impaired in OHE animal models (Hadjihambi et al. 2019; Hsu et al. 2021). From this study, carnitine, 5-hydroxyindoleacetic acid and uracil were identified as being related to the severity of OHE, as they showed positive and negative correlations with the West Haven score scale and Glasgow coma scale, respectively.

Biomarkers based on systemic inflammation

Systemic inflammation is another consequence of liver cirrhosis and a significant factor in the development of OHE. It has been demonstrated that systemic inflammation can modulate the toxic effects of ammonia on the brain (Shawcross et al. 2004) and that high systemic inflammatory response syndrome score, rather than ammonia in the blood, correlate with grades 3 and 4 of OHE (Shawcross et al. 2011). These studies support findings from other research groups that suggest a synergistic effect of systemic inflammation and ammonia levels in inducing neurological dysfunction in chronic liver disease (Felipo et al. 2012b; Montoliu et al. 2015).

In the early stages preceding OHE, there is already involvement of the immune system (Yadav et al. 2016). The levels of interleukin (IL) 6 and IL-18 are more than twice in patients with mHE compared to patients without mHE (Montoliu et al. 2009). These levels correlate with the severity of mHE, and it was observed that patients with mHE had IL-6 levels exceeding 11 pg/mL (Montoliu et al. 2009). Other studies have also demonstrated that the levels of cytokines IL-6 and IL-17, as well as the factor STAT3, are elevated and independently associated with mHE (Luo et al. 2012; Wu et al. 2016; Gairing et al. 2022). Moreover, in patients with mHE, IL-17 levels in plasma exceed 49 pg/mL (Li et al. 2015). Subsequently, more detailed characterization of immunological changes in patients with mHE was conducted (Mangas-Losada et al. 2017). Patients with mHE showed an increase in pro-inflammatory intermediate monocytes (CD14++CD16+), activated B and CD4+ T lymphocytes, and autoreactive CD4+CD28− T lymphocytes. These immunological changes promote a pro-inflammatory environment characterized by elevated serum levels of pro-inflammatory cytokines such as IL-6, IL-21, IL-17, IFN-γ, IL-12, IL-18, TNF-α, IL-1β, IL-22, and IL-15, as well as chemokines CCL20, CXCL13, and CX3CL1. There is also an expansion of Th22 and follicular Th lymphocytes and increased activation of Th17 lymphocytes (see Figure 6 in Mangas-Losada et al. 2017). In this study, serum levels of IgG, IL-15, CXCL13, IL-6, and CX3CL1 achieved diagnostic values with an area under the receiver operating characteristic (AUROC) greater than 0.75 (Mangas-Losada et al. 2017).

In patients with OHE, elevated levels of some of these cytokines have been observed and correlated with the severity of the OHE, such as IL-18 (Onal et al. 2011; Komala et al. 2020), TNF-α (Odeh et al. 2004; Goral et al. 2011), and IL-6 (Luo et al. 2013). These findings support the hypothesis that immune system alterations in cirrhotic patients are potential biomarkers for the progression of OHE.

Oxidative and nitrosative stress

Both hyperammonemia and inflammation can induce oxidative stress, which may mediate the neurological alterations seen in mHE and OHE (Görg et al. 2013). The presence of oxidative stress has been demonstrated in the blood and brain of patients with mHE and OHE (Görg et al. 2010; Giménez-Garzó et al. 2015, 2018). Specific markers of oxidative stress in the blood of patients with mHE, such as the oxidized/reduced glutathione ratio, reduced glutathione levels, malondialdehyde, and 3-nitrotyrosine, correlate with the severity of mHE and with attention and motor coordination impairments in these patients (Montoliu et al. 2011; Gimenez-Garzó et al. 2015). Serum levels of 3-nitrotyrosine are a marker of oxidative stress. Under oxidative stress conditions, nitric oxide reacts with superoxide to produce peroxynitrite, which in turn reacts with tyrosine to form 3-nitrotyrosine (Reiter et al. 2000; Pietraforte et al. 2003; Pacher et al. 2007). Levels of 3-nitrotyrosine are independently associated with mHE (Felipo et al. 2013) and have shown good diagnostic value for mHE, with an AUROC of 0.96, establishing a cutoff point of 14 nM, achieving 83% specificity and 94% sensitivity (Montoliu et al. 2011). Subsequent studies confirmed the diagnostic value of 3-nitrotyrosine in mHE, with AUROC values of 0.85, and sensitivity and specificity percentages of 85% and 82.5%, respectively, at a cutoff point of 14.15 nM (Salman et al. 2021).

Nitrosative stress is also implicated in neuronal alterations in patients with OHE (Genesca et al. 1999). The activation of guanylate cyclase by nitric oxide is altered in the brains of subjects with OHE, leading to impaired cyclic guanosine monophosphate (cGMP) formation, which contributes to the deterioration of cognitive functions during liver failure and hyperammonemia (Corbalán et al. 2002; Erceg et al. 2005a, b). cGMP homeostasis is disrupted in patients with liver cirrhosis, evidenced by increased blood levels but reduced lymphocyte levels of cGMP (Rodrigo et al. 2004; Montoliu et al. 2005). The disturbance in cGMP homeostasis in both the brain and blood of cirrhotic patients suggests that these blood alterations may reflect brain changes and could be associated with mHE. Studies in this area demonstrated that both cGMP levels and nitric oxide-induced guanylate cyclase activation in lymphocytes are elevated in patients with mHE, correlating with the severity of the condition (Montoliu et al. 2007).

Neuroinflammation and central nervous system-derived components

The central nervous system (CNS) is a privileged tissue, isolated from the rest of the body by BBB, making it difficult to access. Postmortem analysis of this tissue aids in understanding the neuropathology associated with OHE, but obtaining biopsy samples from patients is not workable. Due to this difficulty, the analysis of CNS-derived components has been proposed to study its pathological and physiological state, primarily in CSF and blood. Several studies demonstrate the presence of neuroinflammation in patients with OHE (Cagnin et al. 2006; Dennis et al. 2014). Ammonia levels and systemic inflammation mediate this neuroinflammation. Ammonia directly affects microglia (Zemtsova et al. 2011) and systemic inflammation increases the permeability of the BBB. This increased permeability allows the infiltration of immune cells and inflammatory factors into the central nervous system (Kebir et al. 2007; Reboldi et al. 2009; Huppert et al. 2010; Rochfort et al.