Hepatic encephalopathy (HE) is a complex neuropsychiatric syndrome that arises as a consequence of acute or chronic liver dysfunction. It is characterized by a broad spectrum of clinical manifestations, ranging from subtle cognitive impairments to deep coma [1]. Its presence signifies advanced hepatic impairment and represents one of the most serious complications in patients with cirrhosis; it is associated with a marked decline in quality of life, increased risk of hospitalization, and high mortality [2, 3]. Clinically, HE has traditionally been classified into minimal HE (MHE) and overt forms, with increasing recognition of the impact of recurrent episodes and progression to chronicity. In recent years, the pathophysiological understanding of HE has expanded beyond hyperammonemia to include mechanisms such as neuroinflammation, blood-brain barrier (BBB) dysfunction, and altered neurotransmission—an interplay reflecting the dynamic communication between the central nervous system (CNS), hepatic metabolism, and the intestinal environment [1, 4]. This dynamic interaction is strongly influenced by intestinal factors such as the gut microbiota composition, the production of short-chain fatty acids (SCFAs), the modulation of bile acid metabolism, and the maintenance of intestinal barrier integrity, which together determine the levels of microbial metabolites, endotoxins, and ammonia that reach the liver and systemic circulation, ultimately affecting neural homeostasis and inflammatory pathways.

Within this framework, the gut-liver axis has gained prominence as a bidirectional communication network that links the gastrointestinal tract, intestinal microbiota, and liver via immunological, humoral, and neural pathways [5]. Under physiological conditions, this axis maintains immunometabolic homeostasis through tight junction (TJ) integrity, balanced microbial diversity, and regulated mucosal immunity; however, in the context of chronic liver disease (CLD), gut dysbiosis, increased intestinal permeability and bacterial translocation lead to the systemic release of microbiota-derived products such as endotoxins, ammonia, and nitrogenous metabolites, further exacerbating hepatic inflammation and neurological dysfunction [6, 7]. Factors such as small intestinal bacterial overgrowth (SIBO), alterations in the production of urea-splitting bacterial enzymes, and changes in intestinal pH also contribute to enhanced ammonia generation and absorption, aggravating hyperammonemia and the neuropsychiatric burden of HE. Notably, SIBO—specifically hydrogen production—has been significantly associated with HE, as it promotes excessive intestinal ammonia production and is correlated with impaired liver function, thereby playing a direct and independent role in the development and severity of this condition [8].

Multiple studies have demonstrated that gut microbiota-targeted interventions, such as probiotics, prebiotics, nonabsorbable antibiotics, and dietary modifications, have shown clinically meaningful benefits, including reductions in the frequency of HE episodes and improvements in cognitive function [8, 9]. This has paved the way for targeted and personalized therapeutic strategies that may transform the clinical management of this severe hepatic complication [10].

The most widely accepted classification differentiates between MHE and overt HE, which are further subdivided on the basis of severity (grades I to IV) and clinical pattern (episodic, recurrent, or persistent) [11–13]. MHE lacks evident clinical signs and is diagnosed through neuropsychological testing, such as the Psychometric Hepatic Encephalopathy Score and critical flicker frequency [14, 15]. In contrast, overt HE presents with alterations in consciousness, behavior, and neuromotor function, typically graded using tools such as the West Haven criteria (Table 1) [12, 16].

Epidemiological data reveal that MHE affects approximately 30–40% of patients with cirrhosis, while recurrent or persistent forms develop in up to 25% of cases, significantly increasing mortality and health care utilization [18–20]. Hospitalizations related to HE have risen by nearly 50% over the past decade, with a higher incidence observed in older adults and men, likely because of the higher prevalences of cirrhosis and alcohol-associated liver disease in these groups [21]. Even a single HE episode results in a 15–20% one-year mortality risk, which increases with recurrence and is closely associated with progressive neurocognitive decline [19]. These trends underscore the necessity for comprehensive and targeted therapeutic strategies. Beyond recurrence reduction, rifaximin therapy has been associated with significant improvements in health-related quality of life (HRQL), as evidenced by enhanced scores across all domains of the CLD Questionnaire (CLDQ), including fatigue, emotional function, and general well-being [22, 23]. Although MHE is often underrecognized because of the absence of overt clinical signs, it still compromises daily functioning, such as attention, driving, and work performance, especially in younger individuals and women, where underdiagnosis is common [16]. These findings underscore the importance of early detection and classification, not only for clinical prognostication but also for the implementation of strategies that prevent disease progression.

In clinical practice, the diagnosis of HE requires the exclusion of structural, metabolic, toxic, or infectious neurological causes [12]. Laboratory workup includes liver function tests, bilirubin levels, INR, renal function, electrolytes, and ammonia, although the latter does not always correlate with severity [24, 25]. Neuroimaging and electroencephalography are used when clinical suspicion suggests alternative diagnoses [26, 27].

The classic model of HE focuses on ammonia accumulation, which interacts with systemic inflammation, astrocytic dysfunction, BBB impairment, and glial-mediated neuroinflammation [28, 29]. Ammonia is a byproduct of the effects of gut bacterial urease activity on urea [28]. Under normal conditions, the liver detoxifies ammonia via the urea cycle, but in liver failure or portosystemic shunting, this pathway is impaired [30], allowing ammonia to cross the BBB as NH₃ or NH₄⁺ via specific transporters [31]. In the brain, astrocytes convert ammonia to glutamine via the activity of glutamine synthetase. Excess glutamine causes astrocytic swelling, cytotoxic edema, and dysfunction [32]. These astrocytes exhibit Alzheimer type II morphology, with nuclear enlargement and cytoplasmic vacuolization [28, 33]. Mitochondrial glutamine overload exacerbates oxidative stress and impairs energy production [30]. Ammonia enhances GABAergic tone through neurosteroids that modulate GABA-A receptors, causing CNS depression [2]. It also suppresses glutamatergic signaling by interfering with glutamate uptake and energy metabolism [30]. Immunologically, ammonia activates microglia, which release tumor necrosis factor alpha (TNF-α), interleukin 1 beta (IL-1β), IL-6, and reactive oxygen species (ROS), further damaging the BBB and promoting neuroinflammation [34].

Recent research has highlighted bilirubin as a potential contributor to HE beyond its role as a marker of liver dysfunction [35]. Unconjugated bilirubin crosses the BBB, disrupts neuronal and mitochondrial membranes, and induces oxidative stress by impairing antioxidant defenses [29, 31, 36]. It inhibits mitochondrial complexes I and IV, leading to energy failure that worsens ammonia-induced glial toxicity, both toxins activate microglia and cytokine release, amplifying inflammation and neuronal injury (Figure 1) [31, 34]. Moreover, bilirubin inhibits ornithine transcarbamylase, impairing ammonia detoxification [28], whereas ammonia exacerbates cholestasis and alters transporter expression, establishing a vicious cycle [27]. While these findings provide a useful framework for understanding the multifactorial pathogenesis of HE, the proposed mechanisms of bilirubin still need to be confirmed by well-designed translational and clinical studies.

Emerging strategies increasingly focus on leveraging the gut-liver axis to restore microbial balance, reinforce the epithelial barrier, and modulate host signaling. Among these, microbiota-based therapies are pivotal. Fecal microbiota transplantation (FMT) has shown promise in the THEMATIC trial, where cirrhotic patients receiving donor microbiota had improved psychometric scores, reduced ammonia levels, and exhibited durable shifts toward eubiotic communities enriched in Bifidobacterium and Lachnospiraceae [90]. These clinical findings are supported by mechanistic evidence indicating that FMT enhances secondary bile acid production and restores bile acid-microbiota homeostasis, thereby modulating FXR and TGR5 signaling, which are critical for maintaining intestinal and neurological function in patients with cirrhosis [91, 92]. However, variability in donor selection, risk of pathogen transmission, and uncertain long-term engraftment remain significant limitations [93, 94]. Importantly, despite encouraging evidence, FMT is not yet considered a standard therapeutic option in current guidelines, such as those from EASL 2022 [12].

Beyond FMT, probiotic, prebiotic, and synbiotic strategies modulate microbiota composition and function. For example, targeted strains such as Escherichia coli Nissle 1917 have improved colonization resistance, reduced ammonia levels, and enhanced cognition in MHE without adverse events [95]. This strain exerts its effects by outcompeting urease-producing pathogens, increasing epithelial nitrate respiration, and promoting local anti-inflammatory cytokine expression, and its safety profile has been validated in multiple controlled settings [93, 94]. In a recent comparative study, Wang et al. [96] demonstrated that probiotics, rifaximin, and lactulose were safe and effective for MHE, with recovery rates of 58.8%, 45.5%, and 57.1%, respectively, and that all treatments partially restored gut microbial composition by enriching taxa associated with anti-inflammatory and cognitive benefits. Here, it is important to clarify that TJ proteins maintain epithelial barrier integrity; when their expression is reduced by dysbiosis, paracellular permeability increases, facilitating the translocation of harmful metabolites—a key target of these therapies. Rifaximin suppressed pathogenic overgrowth and indirectly modulated bile acid metabolism and FXR/TGR5 activation by altering the intestinal microbiota composition, whereas lactulose enhanced ammonia excretion through luminal acidification and nitrogen trapping [97, 98]. Mechanistically, Bifidobacterium species metabolize lactulose into acetate and lactate, feeding butyrate-producing bacteria and strengthening the mucosa [99]. These fermentation products increase regulatory T-cell activity, inhibit NF-κB-mediated inflammation, and promote epithelial restitution, as supported by decompensated cirrhosis models and meta-analyses demonstrating reduced hospitalization and mortality [93, 100, 101]. Despite promising advances with microbiota-based interventions such as probiotics and FMT, their clinical application remains limited by variability in patient response, lack of standardized protocols, and uncertainty about their long-term stability. Further large-scale, long-term, comparative, and cost-effectiveness studies are needed before they can be fully integrated into clinical practice.

Dietary interventions complement microbiota-based approaches. Protein-adjusted, vegetarian-based regimens supplemented with prebiotics and branched-chain amino acids increase SCFA production, upregulate the expression of barrier markers such as TJ proteins, and lower the risk of HE recurrence [102]. SCFAs, particularly butyrate, stimulate mucosal immunity, regulate TJ assembly via histone deacetylase inhibition, and serve as energy sources for colonocytes, collectively improving barrier function and reducing systemic endotoxemia [93, 101].

In addition to microbiota modulation, barrier-targeted strategies directly aim to strengthen the epithelial barrier and limit microbial translocation. Nutraceuticals such as epigallocatechin-3-gallate (EGCG) upregulate mucin genes, restore TJ proteins, and reduce endotoxemia in HE models, with histological and behavioral improvements [103]. EGCG also activates the Nrf2 pathway to limit oxidative stress and neuroinflammation, although poor oral bioavailability remains a challenge [93, 98]. Additionally, a combination of Lactobacillus rhamnosus, glutamine, and zinc has been shown to increase enterocyte turnover, restore innate immune defense, and reduce lipopolysaccharide (LPS) translocation and the expression of proinflammatory cytokines such as IL-6 and TNF-α [94, 100, 101]. Another group of agents with growing relevance in cirrhosis is alpha-glucosidase inhibitors (AGIs) used in patients with diabetic cirrhosis to protect barrier function by slowing carbohydrate fermentation, thus lowering luminal stress and permeability while improving postprandial glycemic control [104]. Experimental studies have shown that miglitol, an AGI, can alter the intestinal microbial composition and the production of SCFAs, leading to increased butyrate levels and upregulation of hepatic CYP7A1, the rate-limiting enzyme in bile acid synthesis [105]. Its translational potential is supported by clinical data. In a nationwide cohort study of patients with diabetes and compensated cirrhosis, prolonged use of AGI was associated with a significantly lower risk of decompensation, hepatocellular carcinoma, HE, and mortality compared to non-users [106].

Receptor-based interventions represent another frontier. FXR agonists modulate bile acid synthesis and inflammation while supporting barrier integrity by upregulating the expression of TJ proteins such as claudin-1 and promoting SHP-mediated suppression of CYP7A1 [107–109]. FXR activation also increases urea cycle enzyme expression, enhancing hepatic ammonia metabolism, although overactivation can cause pruritus and dyslipidemia [92]. Intestinal FXR signaling induces FGF19, which not only suppresses hepatic bile acid synthesis but also improves hepatic glucose and nitrogen metabolism, contributing to reduced hyperammonemia [97, 110, 111]. TGR5 agonists complement FXR modulation by attenuating neuroinflammation through caspase-8/NLRP3 inhibition in microglia and increasing intracellular cAMP levels, which suppresses NF-κB signaling [112–114]. However, their limited CNS penetration is being addressed by newer synthetic analogs [93, 98]. Similarly, aryl hydrocarbon receptor (AhR) activation by microbial tryptophan metabolites such as indole-3-aldehyde upregulates IL-22 levels, enhances mucosal tolerance, strengthens the barrier, and modulates astrocytic and microglial responses, although prolonged AhR stimulation may affect CYP1A1 expression and carry oncogenic risks [93, 115, 116]. In parallel, bacterial-derived benzodiazepine-like substances have emerged as additional neurotoxins implicated in HE. These compounds can cross the BBB, modulate GABA-A receptors, and induce astrocytic swelling, thereby amplifying CNS depression and neurotransmitter imbalance. Experimental evidence supports their contribution to the neurobehavioral phenotype of HE, particularly in synergy with ammonia [47].

Multimodal approaches are gaining traction as the results show synergy when combining these strategies. In post-transjugular intrahepatic portosystemic shunt (TIPS) patients, triple therapy with rifaximin, lactulose, and FXR agonists reduced the HE incidence and readmissions, highlighting the value of simultaneously targeting microbial composition, barrier integrity, and bile acid signaling [98, 117, 118]. Accordingly, current guidelines increasingly recommend tailored protocols integrating microbiota diagnostics, nutraceuticals, and receptor modulators [34]. Personalized strategies enable patient stratification by microbial signatures, bile acid profiles, and ammonia metabolism rates, optimizing treatment [94, 119]. The role of the TIPS has also been reevaluated. While effective for portal hypertension, it may increase the risk of HE because of the shunting of gut toxins. New TIPS technologies aim to balance hemodynamic benefits with microbiota preservation [118]. Observational studies report variations in HE post-TIPS based on pressure changes and biomarkers of endotoxemia [120].

From a clinical perspective, comparative efficacy and cost-effectiveness are important in guiding therapeutic adoption. A comprehensive network meta-analysis by Dhiman et al. [121], which included 25 randomized trials with over 1,500 participants, confirmed that rifaximin and lactulose are the most effective agents for reversing MHE, while lactulose and L-ornithine L-aspartate were superior for preventing overt HE. Probiotics showed benefits but scored lower in efficacy compared to these agents, although they remain attractive in many settings due to their lower cost and accessibility.

Finally, emerging perspectives include designer probiotics engineered to express glutaminase inhibitors or SCFA-producing enzymes, with the goal of reducing intestinal ammonia production and strengthening barrier function, although regulatory challenges related to genetic stability and biosafety persist [93]. Next-generation TGR5 agonists with improved brain penetration and selective AhR modulators for neuroprotection are under development [52, 116, 122, 123]. The integration of host-microbial omics promises real-time therapeutic monitoring and more precise modulation of the gut-liver-brain axis [93, 98]. Another emerging line of research involves the use of bacteriophages to selectively target harmful gut microbes involved in liver disease [41]. Preclinical studies have shown that, in alcoholic hepatitis, certain strains of Enterococcus faecalis that produce the cytolysin exotoxin aggravate the severity of the disease and mortality. In experimental models, oral administration of phages specifically targeting these cytolytic strains was able to counteract liver damage, while phages against non-cytolytic strains had no effect [124]. Similarly, studies on diet-induced steatohepatitis demonstrated that the elimination of ethanol-producing Klebsiella pneumoniae with adapted phages prior to fecal transplantation prevented the development of diet-induced steatohepatitis [125]. However, to date, no studies have been conducted to evaluate phage therapy in the context of HE. To translate these findings into clinical practice, challenges related to safety, regulatory approval, and long-term microbial stability will need to be addressed. Nevertheless, phage-based strategies represent a promising complement to existing therapies targeting the microbiota and barrier within the gut-liver-brain axis.

The future of HE management is shifting toward a personalized, technology-integrated model centered on the gut-liver-brain axis. While FMT has gained recognition as a therapeutic strategy for modulating dysbiosis and restoring microbial balance, its widespread application continues to face challenges such as no standardized protocols, donor variability, and incomplete data on long-term engraftment and safety outcomes [126]. In addition to conventional microbiota transfer, novel strategies involve exploring host-microbe interactions at the molecular level. For instance, Li et al. [127] proposed targeting cellular pathways, such as modulating autophagy via ST3GAL2-regulated sialylated glycosphingolipids, to influence gut-liver metabolic crosstalk and mitigate neuroinflammation. However, these molecular targets still require robust clinical validation before being adopted into routine care [127]. Understanding host metabolic pathways, including bilirubin metabolism, opens new therapeutic windows for modulating oxidative stress and inflammation in liver-brain axis disorders such as HE [128].

Concurrently, digital health tools and artificial intelligence (AI)-driven innovations are transforming HE monitoring and management. Mobile applications using machine learning algorithms for stool pattern recognition and consistency scoring have demonstrated potential in guiding dynamic lactulose titration—enhancing ammonia clearance, improving adherence, and reducing the recurrence of overt episodes [129, 130]. Moreover, AI applications in hepatology have expanded beyond diagnostics; they now integrate clinical, behavioral, and microbiome-derived data to generate predictive models that support early decision-making and risk assessment in HE [131]. These platforms empower patients through real-time self-monitoring and support clinician decision-making, reinforcing the paradigm of responsive, patient-centered care [132]. Recent validation studies confirm that such tools facilitate individualized therapeutic adjustments, reduce hospitalization risk, and improve longitudinal outcomes in cirrhotic patients [133]. For example, Qiu et al. [134] demonstrated the use of AI-based prognostic models to predict 28-day outcomes in patients with liver failure, underscoring their applicability in risk stratification for HE patients.

In parallel, diagnostic innovation is emerging as a cornerstone of next-generation HE strategies. Functional neuroimaging modalities, particularly functional magnetic resonance imaging (fMRI), allow for early detection of MHE, which often precedes overt HE but significantly impairs quality of life and cognitive function. fMRI enables the visualization of subtle changes in neural connectivity, improving risk stratification and supporting earlier therapeutic intervention in at-risk patients with cirrhosis [85]. Furthermore, the integration of host-microbial multiomics—including metagenomics, transcriptomics, metabolomics, and proteomics—has begun to reveal distinct molecular signatures associated with HE progression, offering a blueprint for individualized treatment selection and therapeutic monitoring [93, 98]. Bajaj et al. [135] reported that gut microbiome profiles could effectively exclude HE in ambiguous clinical cases, highlighting how omics data can enhance diagnostic specificity and reduce overtreatment. Similarly, Sah et al. [136] proposed a future in which AI and omics converge to redefine diagnostic frameworks for cirrhosis-related syndromes, emphasizing their complementary potential in tailoring interventions. However, the implementation of diagnostics based on omics and AI technologies faces significant obstacles. Multi-omics platforms remain costly and technically demanding, with limited availability outside specialized research centers, restricting their routine clinical application. Standardization of sample processing, data integration, and interpretation also lags technological advances, posing challenges for reproducibility and clinical decision-making. In terms of AI, predictive models rely on representative, high-quality datasets, but current studies often include small or region-specific cohorts, limiting their generalizability. Furthermore, integrating AI tools into existing clinical workflows requires not only robust validation but also regulatory approval, digital infrastructure, and training for clinicians to ensure safe and effective use [137, 138].

Nonetheless, these technological advances must be accompanied by comprehensive public health strategies to address persistent care gaps. For instance, Miwa et al. [139] highlighted that MHE remains widely underrecognized by both health care professionals and patients, underscoring the need for standardized diagnostic pathways and educational initiatives to promote timely detection and intervention. This is particularly crucial in high-risk contexts, such as post-TIPS placement, where integrated strategies that combine early biomarkers, neurocognitive testing, and microbiota modulation may reduce the incidence of HE and improve patient outcomes [140]. The trajectory of HE management is evolving toward an integrated framework that combines molecularly targeted therapies, precision microbiota modulation, AI-enabled digital tools, and advanced neuroimaging. Realizing this vision will require ongoing clinical validation of emerging therapies, infrastructure for implementing omics-based diagnostics, and the upskilling of health care providers to interpret and apply novel technologies effectively [132]. Coordinated efforts across research, clinical practice, and policy will be essential to translate these innovations into measurable improvements in patient care, prognosis, and quality of life.

HE represents a multifactorial neuropsychiatric disorder intricately linked to the gut-liver-brain axis, in which gut dysbiosis, increased intestinal permeability, microbial metabolite production, and systemic inflammation converge to disrupt hepatic and neurological homeostasis. Advances in our understanding of this axis have redefined the pathophysiological framework of HE, shifting the focus from isolated hyperammonemia toward a dynamic, integrative model that includes microbial, immunological, and metabolic pathways. Clinical and experimental evidence supports microbiota-targeted therapies—such as rifaximin, probiotics, and FMT—as promising interventions that can modulate the underlying mechanisms of HE. Furthermore, emerging therapeutic strategies involving receptor agonists, barrier-enhancing nutraceuticals, and AI-driven monitoring tools illustrate a shift toward personalized and mechanistically guided management. Nonetheless, the translation of these innovations into widespread clinical practice requires rigorous validation, optimization of diagnostic tools, and equitable access to novel technologies. Ultimately, embracing a multidimensional, microbiome-centered paradigm will be essential for improving outcomes and quality of life in patients affected by this severe hepatic complication.