ABSTRACT
Background
Covert hepatic encephalopathy (CHE) is associated with cognitive impairment and adverse clinical outcomes; however, randomized evidence supporting therapeutic intervention remains limited.
Aim
To evaluate the efficacy of rifaximin (RFX) as a treatment for CHE.
Methods
In this multicentre, open‐label randomized controlled trial, patients with CHE associated with liver cirrhosis were randomized (1:1) to receive RFX or no treatment and followed for 12 weeks. The primary endpoint was the change in Stroop test performance. Secondary endpoints included NCT‐B scores, serum ammonia levels, cirrhosis‐related adverse events, and gut microbiota composition.
Results
Fifty patients were randomized and completed follow‐up. Stroop test performance improved significantly in the RFX group (p = 0.006) but not in controls (p = 0.400), with a trend toward greater improvement with RFX (Δ Stroop test: −4.45 ± 7.12 vs. −0.98 ± 5.74 s; p = 0.056). Among patients not receiving synthetic disaccharides at baseline, improvement was significantly greater with RFX (Δ Stroop test: −3.73 ± 5.96 vs. −0.80 ± 5.86 s; p = 0.049). No significant changes were observed in NCT‐B scores or serum ammonia levels. Cirrhosis‐related adverse events were significantly reduced in the RFX group (p = 0.006). Overall, gut microbial diversity did not differ between groups; however, RFX selectively altered specific taxa, including loss of the [Eubacterium] brachy group.
Conclusions
RFX improved cognitive performance assessed by the Stroop test and reduced cirrhosis‐related adverse events in patients with CHE. These randomized data support RFX as an effective therapeutic option and highlight the Stroop test as a sensitive endpoint for treatment response.
Keywords: covert hepatic encephalopathy (CHE), gut microbiota, liver cirrhosis (LC), minimal hepatic encephalopathy (MHE), rifaximin (RFX), stroop test
Rifaximin improves Stroop test performance and reduces cirrhosis‐related adverse events in patients with covert hepatic encephalopathy. These effects occur without changes in overall gut microbial diversity but are accompanied by selective alterations in specific taxa, suggesting a potential therapeutic strategy targeting the gut–liver–brain axis.
1. Introduction
Hepatic encephalopathy (HE) constitutes a major complication of acute liver failure and liver cirrhosis (LC), and may further present as a neuropsychiatric syndrome secondary to portosystemic shunt formation. It encompasses a broad spectrum of neurological and psychiatric manifestations, ranging from subtle subclinical abnormalities to coma, and is associated with increased morbidity and a substantial reduction in health‐related quality of life [1, 2, 3, 4]. Ammonia plays a central role in the neurotoxicity underlying HE [5, 6, 7], although accumulating evidence suggests that additional mechanisms, including systemic inflammation and oxidative stress, also contribute to its pathogenesis [8]. Moreover, LC and portal hypertension are characterized by profound alterations in the intestinal microbiota, including an increased abundance of ammonia‐producing bacteria, further promoting the development of HE [9, 10].
The West Haven Criteria (WHC) and the International Society of Hepatic Encephalopathy (ISHEN) classification systems have been widely used to categorize the severity of HE [1, 11]. Covert HE (CHE) represents the subclinical end of this spectrum and is characterized by impairments in attention, vigilance, psychomotor speed, and executive function in the absence of overt clinical signs. Accordingly, the diagnosis of CHE relies on neuropsychological, neurophysiological, or neuroimaging assessments [7, 12]. Among available tools, the Stroop test has recently been recognized as a sensitive and practical diagnostic method for CHE, and validated age‐adjusted cutoff values have facilitated its application in clinical practice [13, 14, 15, 16]. The ISHEN consensus recommends that CHE be diagnosed using nationally and culturally validated neuropsychological tests, including the Stroop test, selected according to availability and local expertise [17].
Despite the absence of overt symptoms, CHE has substantial clinical consequences. Patients with CHE demonstrate impaired performance on driving simulators, and up to one‐third are reportedly involved in motor vehicle accidents within 5 years [18, 19, 20]. CHE is also associated with an increased risk of falls, anorexia, muscle weakness, and sleep disturbances, all of which markedly impair quality of life [21, 22, 23]. Importantly, CHE confers a high risk of progression to overt HE and is associated with a shorter time to death or liver transplantation, underscoring its prognostic significance and the potential value of effective intervention at an early stage [21, 22, 24].
Rifaximin (RFX) is a minimally absorbed, gut‐selective antibiotic with broad‐spectrum antimicrobial activity, a favourable safety profile, and a low risk of bacterial resistance [25]. Current practice guidelines from the American Association for the Study of Liver Diseases (AASLD) and the European Association for the Study of the Liver (EASL) recommend RFX for the treatment of overt HE [3, 26, 27]. Previous studies have demonstrated that RFX improves cognitive function and modulates gut microbiota and metabolite profiles in patients with HE [28, 29]. In addition, a recent randomized controlled trial demonstrated that RFX provides clinical benefit in patients with more advanced liver disease, including those with severe cirrhosis and ascites [30]. However, high‐quality randomized evidence supporting therapeutic intervention in CHE, particularly with RFX, remains limited. This gap in evidence represents an unmet clinical need, as early treatment may prevent neurological deterioration and improve long‐term outcomes.
Therefore, we conducted a multicentre, open‐label randomized controlled trial to evaluate the efficacy of RFX initiated at the covert stage of HE, prior to progression to overt HE.
2. Methods
2.1. Study Design and Participants
This multicentre, open‐label, randomized controlled trial enrolled patients with CHE. CHE was diagnosed on the basis of abnormal performance on either the Number Connection Test B (NCT‐B) or the Stroop test using the Neuro‐Psychological Tests for iPad software (version 3.1; Otsuka Pharmaceutical Co. Ltd., distributed by the Japan Society of Hepatology). The cutoff values were defined according to previously reported and established criteria in Japan [16]. The Stroop test consisted of two runs in the on‐state. Among 60 screened candidates, 50 patients who met the eligibility criteria were randomly assigned to the RFX treatment group or the no‐treatment control group and followed for 12 weeks [Figure 1]. Among the 50 enrolled patients, 49 had LC, whereas one had idiopathic portal hypertension (IPH) with porto‐systemic shunts and secondary morphological changes of the liver. At baseline, eight patients were receiving synthetic disaccharides; two had a history of overt HE, and six had a history of hyperammonemia. This study was designed as an exploratory pilot randomized trial in a clinical area with limited prospective data; therefore, a formal sample size calculation was not performed. The sample size was determined on the basis of feasibility across participating centres and prior studies.
FIGURE 1.
Flow diagram of patient enrollment, randomization, and follow‐up. Among 60 patients assessed for eligibility, 50 met the inclusion criteria and were randomized to the rifaximin or control group. All randomized patients completed the 12‐week follow‐up.
Eligible participants were Japanese adults aged 20–74 years diagnosed with CHE in the setting of LC with a serum albumin level ≤ 3.5 g/dL or a portosystemic shunt. Participants were required to be able to attend scheduled outpatient visits or undergo hospitalization during the study period. Written informed consent was obtained from all participants. When a participant lacked decision‐making capacity, written consent was obtained from a legally authorized representative, with patient assent obtained whenever possible. Concomitant use of synthetic disaccharides, including lactulose, was permitted but not required, and no restrictions were placed on baseline serum ammonia levels.
Patients were excluded if they had psychiatric or neurological disorders that could interfere with neuropsychological testing, including depression, masked depression, schizophrenia, or dementia; a diagnosis of malignancy regardless of treatment status; or severe cirrhosis‐related complications, such as refractory ascites requiring repeated paracentesis, bleeding from ruptured oesophageal or gastric varices, spontaneous bacterial peritonitis, or severe electrolyte disturbances affecting neuropsychiatric function. Additional exclusion criteria included a history of hypersensitivity to RFX; severe renal impairment or end‐stage renal disease requiring dialysis; uncontrolled cardiovascular conditions (including myocardial infarction, heart failure, angina pectoris, or clinically significant arrhythmias); initiation or dose adjustment of carnitine or zinc supplementation within 4 weeks prior to randomization or during the 12‐week study period; pregnancy, lactation, or intention to become pregnant; inability to take oral medications; and concomitant use of prohibited medications, including ethinylestradiol or cyclosporine. Patients who had initiated RFX‐containing antibiotics or proton pump inhibitors within 12 weeks prior to enrollment were also excluded. Finally, patients deemed unsuitable for safe participation by the attending physician were excluded.
2.2. Randomization and Interventions
Patients assigned to the RFX group received RFX at a dose of 400 mg three times daily (total daily dose, 1200 mg) for 12 weeks. Patients in the control group received no RFX during the study period. No changes to background therapies for LC were mandated by the study protocol, except for the restrictions specified in the exclusion criteria.
2.3. Endpoints
The primary endpoint was improvement in Stroop test performance from baseline to week 12. Secondary endpoints included changes in the NCT‐B, serum ammonia levels, the occurrence of cirrhosis‐related adverse events (overt HE, falls, sleep disturbances, and gastrointestinal bleeding), and alterations in gut microbiota composition. Sleep disturbances were assessed using the Athens Insomnia Scale (AIS) score.
2.4. Statistical Analysis
Continuous variables are presented as mean ± standard deviation or median with interquartile range, as appropriate, and categorical variables are expressed as number and percentage. Baseline characteristics were compared between groups using the Student's t‐test or the Mann–Whitney U test for continuous variables, and the chi‐square test or Fisher's exact test for categorical variables, as appropriate. Within‐group changes from baseline to week 12 were assessed using the paired t‐test or the Wilcoxon signed‐rank test. Between‐group differences in changes were evaluated using the Student's t‐test or the Mann–Whitney U test, as appropriate. The incidence of cirrhosis‐related adverse events was compared between groups using Fisher's exact test. All statistical tests were two‐sided, and a p value < 0.05 was considered statistically significant. All analyses were conducted according to the intention‐to‐treat principle. Statistical analyses were undertaken using the JMP11 software package (SAS Institute, Cary, NC, USA). Graphs and figures were created using GraphPad Prism version 6 (GraphPad Software, San Diego, CA, USA).
2.5. Gut Microbiota Analysis
Faecal samples collected at baseline and week 12 were subjected to 16S rRNA gene amplicon sequencing performed by TechnoSuruga Laboratory Co. Ltd. (Shizuoka, Japan). DNA extraction and polymerase chain reaction amplification were performed as previously described [31, 32]. Amplicon sequencing was conducted using the MiSeq i100 platform (Illumina, San Diego, CA, USA). The obtained sequence data were processed using the QIIME2 platform (version 2024.5; https://qiime2.org). Sequences were quality‐filtered and dereplicated to generate amplicon sequence variants (ASVs), followed by taxonomic assignment using the SILVA reference database (silva‐138.2‐ssu‐nr99‐341f‐806r amplicon‐2024.5‐classifier.qza). Alpha‐ and beta‐diversity analyses were also performed within the QIIME2 framework. Alpha diversity indices, including Pielou's evenness, Faith's phylogenetic diversity, the number of observed features, and the Shannon entropy index, were analysed using linear mixed‐effects models to account for the longitudinal study design. Time (baseline and week 12), treatment group (RFX vs. control), and their interaction (Time × Treatment) were included as fixed effects, whereas subjects were modelled as random effects to account for within‐subject correlation. Models were fitted using robust standard errors, and the Time × Treatment interaction was used to assess differential changes over time between treatment groups. All statistical analyses were performed using Stata/SE version 12 (StataCorp, College Station, TX, USA), and a two‐sided p value < 0.05 was considered statistically significant. Beta diversity was analysed using Bray–Curtis, Jaccard, and weighted UniFrac distance metrics in QIIME 2. Principal coordinates analysis (PCoA) was performed to visualize microbial community structures at baseline and week 12, with 95% confidence ellipses used to illustrate group‐level distributions. Longitudinal changes were assessed by calculating within‐subject distances from baseline to week 12 for each metric, which were subsequently compared between treatment groups. Taxon‐level analyses were performed using two complementary approaches. First, MaAsLin2 was applied to identify RFX‐associated changes in the relative abundance of individual bacterial taxa. Second, to assess longitudinal changes in taxon prevalence, mixed‐effects logistic regression models were applied to taxa with a prevalence greater than 25% across all samples, with subjects treated as random effects. In addition, Fisher's exact test was used to compare the prevalence of individual bacterial taxa between treatment groups at each time point, with correction for multiple testing applied where appropriate.
2.6. Ethical Considerations
This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Ethics Committee of Kumamoto University (Approval No. 2546). The trial was registered with the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (UMIN000050125).
3. Result
All 50 randomized patients completed the 12‐week observation period. In the RFX group, all patients completed the full 12‐week course of RFX treatment [Figure 1]. Baseline characteristics, including liver function and neuropsychological test scores, did not differ significantly between the RFX and control groups [Tables 1 and 2]. With respect to the primary endpoint, no significant change in Stroop test performance from baseline to week 12 was observed in the control group (p = 0.400), whereas a significant improvement from baseline was observed in the RFX group (p = 0.006) [Figure 2A,B]. Similarly, the change in Stroop test performance tended to be greater in the RFX group than in the control group (Δ Stroop test: −4.45 ± 7.12 vs. −0.98 ± 5.74 s; p = 0.056) [Figure 2C]. In a subgroup analysis of patients who were not receiving synthetic disaccharides at baseline, the improvement in Stroop test performance was significantly greater in the RFX group than in the control group (Δ Stroop test: −3.73 ± 5.96 vs. −0.80 ± 5.86 s; p = 0.049) [Figure 2D]. Although only eight patients were receiving synthetic disaccharides at baseline (2 in the control group and 6 in the RFX group), resulting in limited statistical power, no significant difference in the change in Stroop test performance was observed between the two groups (Figure 2E). In contrast, no significant changes from baseline to week 12 were observed in either the control group or the RFX group for the secondary endpoints of the NCT‐B, the Athens Insomnia Scale (AIS) score, or serum ammonia levels [Figure 3]. Notably, the incidence of cirrhosis‐related adverse events was significantly higher in the control group than in the RFX group (7/26 [26.9%] vs. 0/24 [0%]; p = 0.006), including overt HE in two patients, falls in three patients, traffic accidents in two patients, and gastrointestinal bleeding in one patient [Table 3]. No adverse events related to RFX, including infections, were observed during the study period.
TABLE 1.
Baseline clinical characteristics of patients by treatment group.
| Control (N = 26) | RFX (N = 24) | |
|---|---|---|
| Age, years, median (IQR) | 55.5 (32–73) | 61.0 (46–74) |
| Sex (male/female) | 16/10 | 17/7 |
| BMI, kg/m2 | 26.0 (18.5–43.9) | 25.7 (17.0–37.9) |
| Child–pugh class (A, B, C), n (%) | 6 (23.1%), 18 (69.2%), 2 (7.7%) | 4 (16.7%), 17 (70.8%), 3 (12.5%) |
| MELD score, median (IQR) | 11.0 (7–21) | 13.5 (6–20) |
| Aetiology (HBV/HCV/alcohol/MASLD/other) | 1/3/10/6/6 | 4/1/7/9/3 |
| Ascites (yes/no) | 7/19 | 6/18 |
| Oesophagogastric varices (yes/no) | 10/16 | 9/15 |
| Portal vein thrombosis (yes/no) | 1/25 | 2/22 |
| Concomitant medication (yes/no) | ||
| Spironolactone | 9/17 | 12/12 |
| Furosemide | 9/17 | 14/10 |
| BCAA | 11/15 | 14/10 |
| Synthetic disaccharides | 2/24 | 6/18 |
| L‐carnitine | 2/24 | 1/23 |
| Zn | 1/25 | 2/22 |
| Stroop test, sec, median (IQR) | 27.9 (19.5–78.3) | 29.3 (19.1–54.0) |
| NCT‐B, sec, median (IQR) | 68.7 (23.6–139.9) | 79.6 (48–145) |
| Handgrip strength, kg, median (IQR) | 29.3 (18–47.4) | 33.0.0 (16–45) |
| AIS score, median (IQR) | 4.0 (1–20) | 2.5 (0–17) |
| AIS score ≥ 6, n (%) | 11 (44%) | 7 (29.2%) |
| History of traffic accidents, n (%) | 1 (3.8%) | 0 (0%) |
| History of falls, n (%) | 1 (3.8%) | 1 (4.2%) |
Note: Data are presented as median (interquartile range) or number (%), unless otherwise indicated.
Abbreviations: AIS, Athens Insomnia Scale; ALD, alcohol‐associated liver disease; BCAA, branched‐chain amino acids; BMI, body mass index; CHE, covert hepatic encephalopathy; HBV, hepatitis B virus; HCV, hepatitis C virus; IQR, interquartile range; MASLD, metabolic dysfunction–associated steatotic liver disease; MASLD, metabolic dysfunction–associated steatotic liver disease; MELD, model for end‐stage liver disease; NCT‐B, Number Connection Test B; OHE, overt hepatic encephalopathy; RFX, rifaximin.
TABLE 2.
Baseline laboratory characteristics of patients by treatment group.
| Control (N = 26) | RFX (N = 24) | |
|---|---|---|
| Haemoglobin (g/dL) | 12.0 ± 2.0 | 11.9 ± 2.2 |
| Platelets (1 × 104/L) | 12.3 ± 5.6 | 9.7 ± 4.0 |
| Albumin (g/dL) | 3.1 ± 0.3 | 3.0 ± 0.3 |
| Total bilirubin (mg/dL) | 2.1 ± 2.3 | 2.0 ± 1.2 |
| Alanine aminotransferase (U/L) | 34.9 ± 19.3 | 31.5 ± 23.9 |
| Aspartate aminotransferase (U/L) | 51.3 ± 31.2 | 51.2 ± 26.3 |
| Total cholesterol (mg/dL) | 161.3 ± 47.6 | 156.1 ± 34.6 |
| Serum creatine (mg/dL) | 0.78 ± 0.2 | 1.00 ± 0.4 |
| Serum sodium (mEq/L) | 138.6 ± 3.1 | 138.0 ± 3.0 |
| Serum potassium (mEq/L) | 4.1 ± 0.3 | 4.2 ± 0.48 |
| Serum ammonia (μg/dL) | 80.6 ± 57.7 | 83.5 ± 36.2 |
| Prothrombin activity (%) | 69.3 ± 16.4 | 51.4 ± 18.1 |
Note: Data are presented as mean ± standard deviation.
FIGURE 2.
Effects of rifaximin (RFX) on Stroop test performance over 12 weeks. (A) Stroop test completion times at baseline and week 12 in the control group. Each line represents an individual patient. No significant change was observed (paired t‐test, p = 0.400). (B) Stroop test completion times at baseline and week 12 in the RFX group. Each line represents an individual patient. A significant improvement was observed (paired t‐test, p = 0.006). (C) Comparison of changes in Stroop test performance (Δ Stroop test time) between the control and RFX groups. Box plots show the median and interquartile range, with whiskers indicating the full range (Mann–Whitney U test, p = 0.056). (D) Between‐group comparison of changes in Stroop test performance among patients not receiving synthetic disaccharides at baseline. Box plots show the median and interquartile range, with whiskers indicating the full range (Mann–Whitney U test, p = 0.049). (E) Between‐group comparison of changes in Stroop test performance among patients receiving synthetic disaccharides at baseline. Box plots show the median and interquartile range, with whiskers indicating the full range (Mann–Whitney U test, p = 0.669).
FIGURE 3.
Effects of RFX on secondary endpoints over 12 weeks. (A) Number Connection Test B (NCT‐B) completion times at baseline and week 12 in the control group. (B) NCT‐B completion times at baseline and week 12 in the RFX group. (C) Serum ammonia levels at baseline and week 12 in the control group. (D) Serum ammonia levels at baseline and week 12 in the RFX group. (E) Athens Insomnia Scale (AIS) scores at baseline and week 12 in the control group. (F) AIS scores at baseline and week 12 in the RFX group. In panels A–F, each line represents an individual patient. No significant changes from baseline to week 12 were observed in either group for NCT‐B performance, serum ammonia levels, or AIS scores (paired t‐test).
TABLE 3.
Incidence of cirrhosis‐related adverse events during the 12 weeks.
| Control | Rifaximin | p | |
|---|---|---|---|
| OHE, n (%) | 2 (7.7%) | 0 (0%) | 0.166 |
| Falls, n (%) | 3 (11.5%) | 0 (0%) | 0.086 |
| Traffic accidents, n (%) | 2 (7.7%) | 0 (0%) | 0.166 |
| GI bleeding, n (%) | 1 (3.9%) | 0 (0%) | 0.332 |
| Any adverse event, n (%) | 7 (26.9%) | 0 (0%) | 0.006 |
Note: Adverse events occurring during the 12‐week observation period were analysed. Each event was counted once per patient. P‐values were calculated using Fisher's exact test.
Abbreviations: GI, gastrointestinal; OHE, overt hepatic encephalopathy.
Analysis of alpha diversity indices, including Pielou's evenness, Faith's phylogenetic diversity, observed features, and Shannon entropy, showed no significant main effects of time or treatment. A significant Time × Treatment interaction was observed only for the number of observed features (β = −21.5, p = 0.034), indicating a reduction in microbial richness at week 12 in the RFX‐treated group, whereas no significant interactions were detected for the other indices [Figure 4A]. Beta diversity analysis using Bray–Curtis, Jaccard, and weighted UniFrac distances demonstrated substantial overlap between the control and RFX groups at both baseline and week 12, with no evident separation in PCoA plots [Figure 4B]. Consistently, within‐subject distance‐to‐baseline changes did not differ significantly between groups for any beta diversity metric (all p > 0.05), indicating the absence of broad community‐level restructuring [Figure S1]. Taxon‐level analyses revealed selective RFX‐associated changes. Although MaAsLin2 analysis identified no significant differences in relative abundance, prevalence‐based analyses using mixed‐effects logistic regression identified four taxonomic groups with significant treatment effects. At baseline, the prevalence of the [Eubacterium] brachy group (family Anaerovoracaceae) did not differ between groups (p = 0.37); however, by week 12, this taxon was completely absent in the RFX group (0/23) but remained detectable in the control group (14/21), resulting in a significant difference after correction for multiple testing (q = 5.4 × 10−4) [Figure 4C]. Three additional taxa—the order Christensenellales, the genus Anaerotruncus, and the family XIII AD3011 group—also showed significantly reduced prevalence at week 12 in the RFX‐treated group [Figure 4C]. Exploratory analyses of bacterial taxa associated with lipopolysaccharide‐mediated inflammation or ammonia production showed no significant changes in the abundance of Enterobacteriaceae in either group. The genus Streptococcus exhibited a tendency toward reduced abundance at week 12 in the RFX group, although this did not reach statistical significance (p = 0.13) [Figure S2].
FIGURE 4.
Effects of RFX on gut microbiota composition in patients with covert hepatic encephalopathy. (A) Alpha diversity indices of the gut microbiota at baseline (week 0) and week 12 in the control and RFX groups, including Pielou's evenness, Faith's phylogenetic diversity, the number of observed features, and Shannon entropy. Data were analysed using linear mixed‐effects models with subjects treated as random effects, and the β coefficients and p values shown represent the Time × Treatment interaction for each index. A significant Time × Treatment interaction was observed for the number of observed features (β = −21.5, p = 0.034), indicating a reduction in microbial richness at week 12 in the RFX group. In contrast, no significant interactions were detected for Pielou's evenness (β = −0.012, p = 0.495), Faith's phylogenetic diversity (β = 0.61, p = 0.577), or Shannon entropy (β = −0.18, p = 0.251). (B) Principal coordinates analysis (PCoA) plots on the basis of Bray–Curtis, Jaccard, and weighted UniFrac distance metrics showing gut microbial community structures at baseline (week 0) and week 12 in the control and RFX groups. PC1 and PC2 represent the first and second principal coordinates, respectively, with percentages indicating the proportion of variance explained. Each point represents an individual sample, and colours indicate treatment group and time point. Ellipses denote 95% confidence intervals for each group. No clear separation of microbial community structures was observed between the control and RFX groups at either time point across all distance metrics. (C) Prevalence of selected bacterial taxa at baseline (week 0) and week 12 in the control and RFX groups. Presence/absence data are shown for the [Eubacterium] brachy group (family Anaerovoracaceae), the order Christensenellales, the genus Anaerotruncus, and the family XIII AD3011 group. Longitudinal changes were assessed using mixed‐effects logistic regression models with subjects treated as random effects. Fisher's exact test was used to compare prevalence between groups at each time point, with correction for multiple testing applied where appropriate. All four taxa showed reduced prevalence at week 12 in the RFX group.
4. Discussion
In this multicentre, open‐label randomized controlled trial, we demonstrated that RFX treatment significantly improved cognitive performance, assessed by the Stroop test, in patients with CHE associated with LC or a portosystemic shunt. Whereas no significant change in Stroop test performance was observed in the control group over the 12‐week observation period, patients treated with RFX showed a significant improvement from baseline. These findings provide randomized controlled evidence that RFX effectively ameliorates cognitive dysfunction in CHE—an area in which high‐quality therapeutic data have historically been scarce. Notably, among patients who were not receiving synthetic disaccharides at baseline, the change in Stroop test performance differed significantly between the two groups, suggesting that RFX confers incremental benefit in patients naïve to other standard therapies. This observation highlights the potential role of RFX as an independent therapeutic option in the management of CHE. In contrast, no significant improvement in NCT‐B scores was observed in either group. This discrepancy likely reflects differences in the cognitive domains assessed by these tests. The Stroop test is particularly sensitive to subtle impairments in attention, executive function, and psychomotor processing—hallmark features of CHE—and therefore may represent a more responsive endpoint for detecting treatment‐related cognitive improvement in this population [14]. Serum albumin was used as an inclusion criterion to identify patients at increased risk of CHE, an approach supported by previous studies, including a Japanese multicentre study demonstrating that serum albumin ≤ 3.2 g/dL may serve as an objective biomarker of CHE [16]. In the present study, a slightly higher threshold was applied to include a broader at‐risk population, consistent with the exploratory design of this trial.
Serum ammonia levels did not change significantly from baseline to week 12 in either group, indicating that the cognitive improvement observed with RFX was not directly associated with measurable reductions in systemic ammonia levels. This finding is consistent with previous reports suggesting that the pathophysiology of CHE is a multifactorial disorder mediated not only by hyperammonemia but also by systemic inflammation, neuroinflammation, and alterations in the gut microbiota. The dissociation between cognitive improvement and stable ammonia levels supports the concept that therapeutic strategies targeting multiple HE‐related pathways may be required.
Importantly, the incidence of cirrhosis‐related adverse events, including OHE, falls, traffic accidents, and gastrointestinal bleeding, was significantly higher in the control group than in the RFX group (7/26 [26.9%] vs. 0/24 [0%]; p = 0.006). Notably, these differences were observed within a relatively short 12‐week period. These results suggest that RFX treatment may not only improve cognitive performance but also reduce clinically relevant complications associated with CHE. The reduction in falls and traffic accidents is particularly noteworthy, as cognitive impairment related to CHE has been shown to compromise driving performance, motor coordination, and daily functioning. This benefit may be mediated not only by improvements in cognitive function but also by modulation of gut‐derived factors implicated in the gut–liver–brain axis. Taken together, these findings reinforce the broader clinical value of addressing CHE proactively to improve patient safety, quality of life, and healthcare utilization.
In the present study, we extended our investigation beyond clinical and neuropsychological outcomes to include an analysis of the gut microbiota in patients with CHE. Despite the significant improvement in cognitive performance with RFX treatment, 16S rRNA‐based sequencing revealed no significant differences in overall gut microbial diversity or community structure between the RFX and control groups. These findings indicate that RFX does not induce broad restructuring of the gut microbiota, but rather exerts selective effects within the dysbiotic microbial environment characteristic of cirrhosis. At the taxon level, prevalence‐based analyses identified several bacterial groups that were selectively reduced in the RFX‐treated group at week 12. Notably, the [Eubacterium] brachy group and members of the Christensenellaceae family showed significant reductions in prevalence following RFX treatment. Although the functional roles of these taxa in HE remain incompletely understood, previous reports have linked Eubacterium species to inflammatory conditions, suggesting that their reduction may reflect attenuation of a pro‐inflammatory gut milieu [33, 34]. The biological significance of changes in Christensenellaceae in this cohort is less clear and should be regarded as exploratory. In addition, bacterial taxa previously associated with endotoxin or ammonia production, including Enterobacteriaceae and Streptococcus, showed a trend toward reduced abundance in the RFX group, although these changes did not reach statistical significance. This pattern is consistent with the observed dissociation between cognitive improvement and unchanged serum ammonia levels, supporting the notion that RFX may influence CHE through mechanisms beyond ammonia reduction, such as modulation of gut‐derived inflammatory signalling. Given the substantial temporal variability of the gut microbiota in patients with cirrhosis, the detection of systematic, RFX‐associated changes in specific microbial taxa suggests a targeted effect on the gut–liver–brain axis. Overall, the beneficial effects of RFX in CHE appear to be mediated not by global microbiota remodelling, but by selective modulation of functionally relevant microbial subsets, a mechanism that may not be fully captured by serum ammonia measurements alone. Taken together, the cognitive benefits of RFX occurred without global changes in gut microbial diversity or serum ammonia levels, but were associated with selective alterations in specific anaerobic bacterial taxa, suggesting that targeted modulation of functionally relevant microbial subsets may contribute to the pathophysiology and treatment response of CHE.
Several limitations of this study warrant consideration. First, the study population was limited to Japanese adults, which may affect the generalizability of the findings to other populations. In addition, the open‐label design may have introduced potential bias, although objective neuropsychological assessments were employed as primary endpoints. Second, the sample size was relatively small, potentially limiting statistical power for secondary outcomes. Although a formal sample size calculation was not performed, clinically meaningful differences were observed in key outcomes, supporting the robustness of the findings. However, the modest sample size may have limited the ability to detect more subtle differences. Third, the 12‐week observation period did not allow a comprehensive evaluation of long‐term outcomes; extended follow‐up of this cohort is ongoing. In addition, microbiota analyses were based on 16S rRNA sequencing, which provides limited taxonomic and functional resolution and does not directly assess microbial metabolic pathways, including ammonia generation, endotoxin production, or bile acid metabolism, nor does it enable evaluation of the gut resistome or antimicrobial resistance genes. The relative‐abundance nature of these data and the substantial interindividual variability characteristic of cirrhosis may also have influenced the detection of RFX‐associated microbiota changes. Future studies incorporating shotgun metagenomics, metabolomics, functional analyses, and resistome profiling will be necessary to clarify the mechanistic relationships between microbial modulation and cognitive benefit.
In conclusion, RFX significantly improved Stroop test performance and reduced cirrhosis‐related adverse events in patients with CHE associated with LC or a portosystemic shunt, even in the absence of significant changes in serum ammonia levels. Importantly, although previous randomized trials of RFX have primarily focused on clinical outcomes such as complications or survival in patients with advanced cirrhosis, our findings demonstrate that RFX improves cognitive function and reduces clinically relevant adverse events in CHE. These findings provide randomized controlled evidence supporting RFX as an effective therapeutic option for CHE and underscore the importance of cognitive‐focused endpoints when evaluating treatment efficacy in this condition. Collectively, this study offers meaningful clinical and mechanistic insights into the management of CHE and may help inform future guideline recommendations regarding therapeutic intervention at the covert stage of HE.
Author Contributions
Haruki Uojima: investigation. Etsuko Iio: investigation. Toshinori Toyota: data curation, formal analysis, investigation, validation. Satoshi Miuma: investigation. Hiroki Inada: conceptualization, writing – review and editing, writing – original draft, investigation, methodology, data curation, formal analysis, project administration, visualization, validation. Katsuya Nagaoka: investigation. Sotaro Kurano: investigation. Shiho Miyase: investigation. Takahiro Mizuta: investigation. Daiki Maeda: investigation. Satoshi Narahara: investigation. Kentaro Tanaka: investigation. Takao Miwa: investigation. Hisashi Hidaka: investigation. Hiroko Setoyama: investigation. Yoko Yoshimaru: investigation. Masahito Shimizu: investigation. Takehisa Watanabe: investigation. Jiro Nakayama: formal analysis, data curation. Yasuhito Tanaka: conceptualization, methodology, funding acquisition, writing – original draft, writing – review and editing, project administration, supervision. Shuichiro Iwasaki: investigation. Kazuhiro Sugi: investigation.
Funding
This work was supported by the Japan Agency for Medical Research and Development, JP24fk0210103, JP25fk0210172. Japan Society for the Promotion of Science, JP24K18908.
Conflicts of Interest
Yasuhito Tanaka received honoraria from AbbVie GK, Gilead Sciences Inc., Chugai Pharmaceutical Co. Ltd., ASKA Pharmaceutical Holdings Co. Ltd., GlaxoSmithKline PLC, AstraZeneca, Eisai, and H.U. frontier. He received research funds from OTSUKA Pharmaceutical Co. Ltd., AbbVie GK, FUJIREBIO Inc., Sysmex Corp, GlaxoSmithKline PLC., Gilead Sciences Inc., and AstraZeneca. Masahito Shimizu received honoraria from ASKA Pharmaceutical Holdings Co.
Supporting information
Figure S1: Within‐subject changes in beta diversity from baseline to week 12. Box plots showing within‐subject distance‐to‐baseline changes from week 0 to week 12 for (A) Bray–Curtis, (B) Jaccard, and (C) weighted UniFrac distance metrics in the control and RFX groups. Each dot represents an individual subject. Comparisons between groups were performed using Welch's t‐test. No significant differences in within‐subject beta diversity changes were observed between the control and RFX groups for any metric.
Figure S2: Changes in bacterial taxa associated with endotoxin or ammonia production. Box plots show the median and interquartile range, with individual data points representing each subject. (A) Relative abundance of Enterobacteriaceae and (B) Streptococcus at baseline (week 0) and week 12 in the control and RFX groups. P values for within‐group comparisons are shown. No statistically significant changes were observed, although a trend toward reduced Streptococcus abundance was noted in the RFX group.
Acknowledgements
This work was supported by a grant‐in‐aid from the Research Program on Hepatitis from the Japan Agency for Medical Research and Development (AMED JP24fk0210103, JP25fk0210172) and the JSPS KAKENHI (grant number JP24K18908).
Inada H., Toyota T., Uojima H., et al., “Rifaximin Improves Cognitive Performance and Reduces Cirrhosis‐Related Adverse Events in Covert Hepatic Encephalopathy: A Randomized Controlled Trial,” Alimentary Pharmacology & Therapeutics 64, no. 3 (2026): 379–389, 10.1111/apt.70712.
Handling Editor: Daniel Huang
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Ferenci P., Lockwood A., Mullen K., Tarter R., Weissenborn K., and Blei A. T., “Hepatic Encephalopathy—Definition, Nomenclature, Diagnosis, and Quantification: Final Report of the Working Party at the 11th World Congresses of Gastroenterology, Vienna, 1998,” Hepatology 35 (2002): 716–721. [DOI] [PubMed] [Google Scholar]
- 2. Atluri D. K., Prakash R., and Mullen K. D., “Pathogenesis, Diagnosis, and Treatment of Hepatic Encephalopathy,” Journal of Clinical and Experimental Hepatology 1, no. 2 (2011): 77–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. American Association for the Study of Liver Diseases; European Association for the Study of the Liver , “Hepatic Encephalopathy in Chronic Liver Disease: 2014 Practice Guideline by the European Association for the Study of the Liver and the American Association for the Study of Liver Diseases,” Journal of Hepatology 61, no. 3 (2014): 642–659. [DOI] [PubMed] [Google Scholar]
- 4. Kroupina K., Bémeur C., and Rose C. F., “Amino Acids, Ammonia, and Hepatic Encephalopathy,” Analytical Biochemistry 649 (2022): 114696. [DOI] [PubMed] [Google Scholar]
- 5. Vaquero J., Chung C., and Blei A. T., “Brain Edema in Acute Liver Failure. A Window to the Pathogenesis of Hepatic Encephalopathy,” Annals of Hepatology 2, no. 1 (2003): 12–22. [PubMed] [Google Scholar]
- 6. Dbouk N. and McGuire B. M., “Hepatic Encephalopathy: A Review of Its Pathophysiology and Treatment,” Current Treatment Options in Gastroenterology 9, no. 6 (2006): 464–474. [DOI] [PubMed] [Google Scholar]
- 7. Katayama K. and Kakita N., “Possible Pathogenetic Role of Ammonia in Liver Cirrhosis Without Hyperammonemia of Venous Blood: The So‐Called Latency Period of Abnormal Ammonia Metabolism,” Hepatology Research 54, no. 3 (2024): 235–243. [DOI] [PubMed] [Google Scholar]
- 8. Ochoa‐Sanchez R. and Rose C. F., “Pathogenesis of Hepatic Encephalopathy in Chronic Liver Disease,” Journal of Clinical and Experimental Hepatology 8, no. 3 (2018): 262–271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Chen Y., Yang F., Lu H., et al., “Characterization of Fecal Microbial Communities in Patients With Liver Cirrhosis,” Hepatology 54 (2011): 562–572. [DOI] [PubMed] [Google Scholar]
- 10. Qin N., Yang F., Li A., et al., “Alterations of the Human Gut Microbiome in Liver Cirrhosis,” Nature 513 (2014): 59–64. [DOI] [PubMed] [Google Scholar]
- 11. Bajaj J. S., Cordoba J., Mullen K. D., et al., “Review Article: The Design of Clinical Trials in Hepatic Encephalopathy—An International Society for Hepatic Encephalopathy and Nitrogen Metabolism (ISHEN) Consensus Statement,” Alimentary Pharmacology & Therapeutics 33 (2011): 739–747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Rose C. F., Amodio P., Bajaj J. S., et al., “Hepatic Encephalopathy: Novel Insights Into Classification, Pathophysiology and Therapy,” Journal of Hepatology 73 (2020): 1526–1547. [DOI] [PubMed] [Google Scholar]
- 13. Bajaj J. S., Thacker L. R., Heuman D. M., et al., “The Stroop Smartphone Application is a Short and Valid Method to Screen for Minimal Hepatic Encephalopathy,” Hepatology 58, no. 3 (2013): 1122–1132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Bajaj J. S., Heuman D. M., Sterling R. K., et al., “Validation of EncephalApp, Smartphone‐Based Stroop Test, for the Diagnosis of Covert Hepatic Encephalopathy,” Clinical Gastroenterology and Hepatology 13 (2015): 1828–1835. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. European Association for the Study of the Liver , “EASL Clinical Practice Guidelines on the Management of Hepatic Encephalopathy,” Journal of Hepatology 77, no. 3 (2022): 807–824. [DOI] [PubMed] [Google Scholar]
- 16. Kondo Y., Iwasa M., Kawaratani H., et al., “Proposal of Stroop Test Cut‐Off Values as Screening for Neuropsychological Impairments in Cirrhosis: A Japanese Multicenter Study,” Hepatology Research 51, no. 6 (2021): 674–681. [DOI] [PubMed] [Google Scholar]
- 17. Bajaj J. S., Lauridsen M., Tapper E. B., et al., “Important Unresolved Questions in the Management of Hepatic Encephalopathy: An ISHEN Consensus,” American Journal of Gastroenterology 115 (2020): 989–1002. [DOI] [PubMed] [Google Scholar]
- 18. Bajaj J. S., Hafeezullah M., Hoffmann R. G., et al., “Navigation Skill Impairment: Another Dimension of the Driving Difficulties in Minimal Hepatic Encephalopathy,” Hepatology 47, no. 2 (2008): 596–604. [DOI] [PubMed] [Google Scholar]
- 19. Lauridsen M. M., Thacker L. R., White M. B., et al., “In Patients With Cirrhosis, Driving Simulator Performance is Associated With Real‐Life Driving,” Clinical Gastroenterology and Hepatology 14, no. 5 (2016): 747–752. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Bajaj J. S., Hafeezullah M., Hoffmann R. G., and Saeian K., “Minimal Hepatic Encephalopathy: A Vehicle for Accidents and Traffic Violations,” American Journal of Gastroenterology 102, no. 9 (2007): 1903–1909. [DOI] [PubMed] [Google Scholar]
- 21. Román E., Córdoba J., Torrens M., et al., “Minimal Hepatic Encephalopathy is Associated With Falls,” American Journal of Gastroenterology 106, no. 3 (2011): 476–482. [DOI] [PubMed] [Google Scholar]
- 22. Yoshimura E., Ichikawa T., Miyaaki H., et al., “Screening for Minimal Hepatic Encephalopathy in Patients With Cirrhosis by Cirrhosis‐Related Symptoms and a History of Overt Hepatic Encephalopathy,” Biomed Rep 5, no. 2 (2016): 193–198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Miyaaki H., Hiraoka A., Haraguchi M., et al., “Proposal for New Sleep Disorder Criteria in Patients With Chronic Liver Disease: Influence of Liver‐Related Complications,” Hepatology Research 52, no. 4 (2022): 364–370. [DOI] [PubMed] [Google Scholar]
- 24. Patidar K. R., Thacker L. R., Wade J. B., et al., “Covert Hepatic Encephalopathy is Independently Associated With Poor Survival and Increased Risk of Hospitalization,” American Journal of Gastroenterology 109, no. 11 (2014): 1757–1763. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. DuPont H. L., “Therapeutic Effects and Mechanisms of Action of Rifaximin in Gastrointestinal Diseases,” Mayo Clinic Proceedings 90, no. 8 (2015): 1116–1124. [DOI] [PubMed] [Google Scholar]
- 26. Vilstrup H., Amodio P., Bajaj J., et al., “Hepatic Encephalopathy in Chronic Liver Disease: 2014 Practice Guideline by the American Association for the Study of Liver Diseases and the European Association for the Study of the Liver,” Hepatology 60, no. 2 (2014): 715–735. [DOI] [PubMed] [Google Scholar]
- 27. Bass N. M., Mullen K. D., Sanyal A., et al., “Rifaximin Treatment in Hepatic Encephalopathy,” New England Journal of Medicine 362, no. 12 (2010): 1071–1081. [DOI] [PubMed] [Google Scholar]
- 28. Bajaj J. S., Heuman D. M., Sanyal A. J., et al., “Modulation of the Metabiome by Rifaximin in Patients With Cirrhosis and Minimal Hepatic Encephalopathy,” PLoS One 8, no. 4 (2013): e60042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Yu X., Jin Y., Zhou W., et al., “Rifaximin Modulates the Gut Microbiota to Prevent Hepatic Encephalopathy in Liver Cirrhosis Without Impacting the Resistome,” Frontiers in Cellular and Infection Microbiology 11 (2022): 761192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Thévenot T., Elkrief L., Bureau C., et al., “Effect of Rifaximin in Patients With Severe Cirrhosis and Ascites: A Randomized Double‐Blind Placebo‐Controlled Trial,” Journal of Hepatology 83, no. 6 (2025): 1320–1327. [DOI] [PubMed] [Google Scholar]
- 31. Takahashi S., Tomita J., Nishioka K., Hisada T., and Nishijima M., “Development of a Prokaryotic Universal Primer for Simultaneous Analysis of Bacteria and Archaea Using Next‐Generation Sequencing,” PLoS One 9 (2014): e105592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Hisada T., Endoh K., and Kuriki K., “Inter‐and Intra‐Individual Variations in Seasonal and Daily Stabilities of the Human Gut Microbiota in Japanese,” Archives of Microbiology 197 (2015): 919–934. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Mukherjee A., Lordan C., Ross R. P., and Cotter P. D., “Gut Microbes From the Phylogenetically Diverse Genus Eubacterium and Their Various Contributions to Gut Health,” Gut Microbes 12, no. 1 (2020): 1802866. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Sugawara Y., Kanazawa A., Aida M., Yoshida Y., Yamashiro Y., and Watada H., “Association of Gut Microbiota and Inflammatory Markers in Obese Patients With Type 2 Diabetes Mellitus: Post Hoc Analysis of a Synbiotic Interventional Study,” Biosci Microbiota Food Health 41, no. 3 (2022): 103–111. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1: Within‐subject changes in beta diversity from baseline to week 12. Box plots showing within‐subject distance‐to‐baseline changes from week 0 to week 12 for (A) Bray–Curtis, (B) Jaccard, and (C) weighted UniFrac distance metrics in the control and RFX groups. Each dot represents an individual subject. Comparisons between groups were performed using Welch's t‐test. No significant differences in within‐subject beta diversity changes were observed between the control and RFX groups for any metric.
Figure S2: Changes in bacterial taxa associated with endotoxin or ammonia production. Box plots show the median and interquartile range, with individual data points representing each subject. (A) Relative abundance of Enterobacteriaceae and (B) Streptococcus at baseline (week 0) and week 12 in the control and RFX groups. P values for within‐group comparisons are shown. No statistically significant changes were observed, although a trend toward reduced Streptococcus abundance was noted in the RFX group.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.