Review article: the current pharmacological therapies for hepatic encephalopathy


  • NMB has received honoraria as a speaker and consultant for Salix Pharmaceuticals, Inc.

Dr N. M. Bass, UCSF Liver Transplantation Service, School of Medicine, University of California, San Francisco, Box 0538, Rm 357-S, San Francisco, CA 94143-0538, USA. E-mail:


Effective treatment options for hepatic encephalopathy are limited. Based on the principle that intestinal-derived ammonia contributes to the pathogenesis of hepatic encephalopathy, current therapeutic approaches are directed at reducing bacterial production of ammonia and enhancing its elimination.

Non-absorbable disaccharides are first-line therapy for hepatic encephalopathy, but published clinical studies evaluating their safety and efficacy are limited. Alternative therapies such as benzodiazepine receptor antagonists, branched-chain amino acids, and l-ornithine-l-aspartate also have limited clinical data supporting their use.

Studies of antibiotics indicate that they are effective in the treatment of hepatic encephalopathy, but adverse effects and concerns about long-term safety have limited the widespread use of most.

Rifaximin is a minimally absorbed antibiotic that concentrates in the gastrointestinal tract and is excreted mostly unchanged in faeces. It has been studied extensively in the treatment of hepatic encephalopathy and appears to confer therapeutic benefits greater than those of placebo and non-absorbable disaccharides and at least comparable with those of systemic antibiotics. Rifaximin was also well tolerated in patients with hepatic encephalopathy and is not associated with clinical drug interactions or clinically relevant bacterial antibiotic resistance.

In conclusion, non-absorbed antibiotics such as rifaximin offer a favourable benefit–risk ratio in the treatment of hepatic encephalopathy and may help to improve patient outcomes.


Hepatic encephalopathy (HE) is a neuropsychiatric disorder resulting from complications of liver disease that can significantly impact quality of life. Clinical manifestations of HE range from subtle neuropathic abnormalities to coma and include disturbances in consciousness, altered neuromuscular activity and electroencephalogram (EEG) abnormalities.1, 2 The pathogenesis of HE is thought to involve the accumulation of ammonia because of poor hepatic function and portosystemic shunting.

Diagnosis of HE is based on clinical presentation in patients with chronic liver disease, and severity of HE can be classified as stage 0–4, depending on symptoms (Table 1).1, 2 The portosystemic encephalopathy (PSE) index, often measured in studies of HE, is calculated from clinical HE parameters including mental state, asterixis, time to completion of number connection test, EEG mean cycle frequency and blood ammonia concentration.3 Diagnosis of subclinical HE is more difficult than that of overt HE, and up to 80% of patients with cirrhosis present with subclinical disease.1, 2

Table 1.   Neuropsychiatric staging of hepatic encephalopathy
StageConsciousnessPersonality and intellectNeurologicAmmonia levelEEG findings
  1. EEG, electroencephalogram.

  2. Reproduced with permission from Abou-Assi et al.1

SubclinicalNormalNormalImpaired psychomotor testingNormalNormal
Stage 1Insomnia, disturbed sleep patternConfusion, forgetfulness, agitationTremor, constructional apraxia, uncoordinationSlightly abnormal
Stage 2LethargyDisorientation, bizarre behaviourAsterixis, ataxia↑↑Slowing of triphasic waves
Stage 3Somnolence, but patient may be arousableDisorientation, aggressionAsterixis, hyperactive reflexes, positive Babinski's reflex↑↑↑Slowing of triphasic waves
Stage 4Coma, unresponsiveComaDecerebrate posture↑↑↑↑Slow waves (2–3 cycles per second)

The prevalence of chronic liver disease in the United States is c. 5.5 million cases.4 The true incidence and prevalence of HE is difficult to establish given the considerable differences in severity of overt disease and the difficulty in diagnosing subclinical disease.5 Clinical prognosis after developing overt HE is poor, with 1-year survival estimated at 42% and 3-year survival estimated at 23%.6 However, most clinical manifestations of moderate disease are reversible with effective treatment.7

Therapeutic approaches in HE are directed at reducing bacterial production of ammonia and/or enhancing its elimination. The objective of this review is to evaluate current pharmacologic therapies available for the management of HE.

Non-absorbable disaccharides

Non-absorbable disaccharides are recommended as first-line pharmacotherapy in current guidelines.2, 7 Disaccharides remain undigested until they reach the colon, where they function to inhibit bacterial ammonia production and trap ammonia as non-diffusable ammonium in the intestinal lumen.1, 7

Published studies evaluating the safety and efficacy of non-absorbable disaccharides are limited. Two trials of lactulose vs. neomycin determined that the treatments were comparable in efficacy.8, 9 A small meta-analysis determined that lactulose and lactitol were equally effective in the treatment of HE.10 Approval of lactulose for the treatment of HE in the United States was based on a small, open-label clinical study.11

A meta-analysis of 22 randomized trials highlighted the lack of data supporting the efficacy of non-absorbable disaccharides.12 Primary endpoints included the number of patients without improvement in HE and all-cause mortality in 10 clinical studies of non-absorbable disaccharides vs. placebo or no intervention and 12 clinical studies of non-absorbable disaccharides vs. antibiotics (including rifaximin, neomycin, ribostamycin and vancomycin). Non-absorbable disaccharides compared with no intervention or placebo did not significantly affect mortality or significantly reduce the risk of no clinical improvement of HE in a small number of high-quality clinical studies (i.e. adequate randomization and blinding). Non-absorbable disaccharides were inferior to antibiotics in reducing risk of no clinical improvement and in lowering blood ammonia concentrations but did not differ from antibiotics with respect to having no impact on all-cause mortality. The authors concluded that current evidence is insufficient to support or refute the use of non-absorbable disaccharides for treatment of HE.


The goal of antibiotic therapy in the treatment of HE is to reduce the mass of enteric bacteria that produce ammonia.7 Although the efficacy of systemic antibiotics has been demonstrated, adverse effects and concerns about safety have generally limited their widespread use.2

Neomycin and metronidazole

Antibiotics with activity against urease-producing bacteria (e.g. neomycin, metronidazole) are effective in reducing intestinal ammonia production. However, the safety and efficacy of these antibiotics in treatment of HE have not been adequately demonstrated. One randomized, controlled trial evaluated the efficacy of neomycin vs. placebo in patients with acute HE and demonstrated no significant difference in time to improvement of disease symptoms with neomycin 6 g/day (n = 20) vs. placebo (n = 19).13 Two additional studies found no significant difference between lactulose and neomycin, concluding that both therapies are effective for HE.8, 9 However, neither study had adequate design to statistically determine equivalence. No additional benefit vs. placebo was observed in study of neomycin-lactulose combination therapy (N = 80), and the therapy was not well tolerated.14 Additionally, one small trial compared the efficacy of metronidazole with neomycin in 18 patients with mild, moderate, or severe HE.15 Both treatments improved mental state, asterixis and EEG measurements although neither antibiotic significantly improved mean blood ammonia levels.15

Systemic absorption of antibiotics used to treat HE may result in serious adverse effects. Neomycin can directly impact the small-bowel mucosa resulting in intestinal malabsorption with doses as low as 3 g/day.2, 16, 17 Neomycin is poorly absorbed following oral administration, but cumulative systemic absorption following prolonged use can cause hearing loss and renal failure.2 Metronidazole is also associated with serious adverse effects including peripheral neuropathy,7 and its elimination by hepatic oxidation may increase risk of toxicity in patients with chronic liver disease.18


Rifaximin is an oral, minimally absorbed antibiotic (<0.4%) with broad-spectrum in vitro activity against enteric bacteria.19–22 The tolerability profile of rifaximin is comparable to placebo,22 and no clinical drug interactions have been reported.21 Because of the lack of systemic accumulation with rifaximin,20 no dosing adjustments are required in patients with hepatic insufficiency.

Rifaximin was approved in the United States in 2004 for non-dysenteric diarrhoea caused by Escherichia coli and is licensed in Europe, Latin America, Asia and Africa for several indications, including HE. Multiple published studies have evaluated the safety and efficacy of rifaximin in the treatment of HE3, 23–39 and representative data are discussed below.

A prospective study was conducted to explore appropriate doses of rifaximin for the treatment of HE.3 Patients with stage 1–3 HE received rifaximin 600 mg/day (n = 18), 1200 mg/day (n = 19), or 2400 mg/day (n = 17) for 7 days. A significant reduction in the mean PSE index from baseline was observed with rifaximin 1200 or 2400 mg/day, but not 600 mg/day (Table 2). These results suggest that rifaximin ≥1200 mg/day is more beneficial than 600 mg/day in the treatment of HE. The studies included in this review have all employed a daily dose of 1200 mg.

Table 2.   Mean change from baseline in PSE index in patients with hepatic encephalopathy treated with rifaximin 600, 1200 or 2400 mg/day
Rifaximin (mg/day)n*PSE index (%)
Change from baseline, mean ± s.d. (range) 95% CI
  1. CI, confidence interval; PSE, portosystemic encephalopathy; s.d., standard deviation.

  2. *Signifies number of patients for whom data were available.

  3. Adapted from Williams et al.24

60014−6.4 ± 13.7 (−25.0, 25.0)−14.0, 1.2
120016−10.3 ± 13.7 (−28.6, 32.1)−17.4, −3.1
240016−10.7 ± 14.9 (−39.3, 14.3)−17.8, −3.6

Rifaximin efficacy in HE has been assessed in a randomized, double-blind study.3 Patients with HE, cirrhosis and lactulose or lactitol intolerance received rifaximin 1200 mg/day (n = 48) or placebo (n = 45) for 14 days. Rifaximin significantly improved asterixis compared with placebo (P < 0.01) but did not effectively improve mental state, the primary endpoint of the study. The lack of statistical difference between rifaximin and placebo may have resulted from the fact that the majority of patients had mild disease at baseline. In this study, the tolerability profile of rifaximin was comparable with that of placebo3 and supported tolerability findings from studies in patients with travellers’ diarrhoea.22 This favourable safety and tolerability profile differentiates rifaximin from other antibiotics and non-absorbable disaccharides in treatment of HE.

Comparison with other antibiotics

Rifaximin has been compared with other antibiotics (e.g. neomycin, paromomycin) in the treatment of HE.23, 30–35 In a randomized, double-blind study, rifaximin 1200 mg/day (n = 15) was compared with neomycin 3 g/day (n = 15) in 30 patients with cirrhosis and stage 1–3 HE.30 After 21 days of treatment, neuropsychiatric symptoms and blood ammonia concentrations were significantly reduced vs. baseline in both groups; however, reduction in blood ammonia concentrations was significantly greater with rifaximin treatment vs. neomycin (Figure 1). Although no patient administered rifaximin experienced adverse events, 26% of patients administered neomycin showed increases in blood urea nitrogen and plasma creatinine levels, and 33% reported nausea, abdominal pain and vomiting.

Figure 1.

 (A) Number connection test performance and (B) blood ammonia concentrations in patients with hepatic encephalopathy treated with rifaximin 1200 mg/day or neomycin 3 g/day. *P < 0.001 vs. baseline; P < 0.025 vs. baseline; P < 0.005 vs. neomycin. Data from Pedretti et al.30

In another randomized trial, patients with cirrhosis and chronic stage 1–2 HE received rifaximin 1200 mg/day (n = 25) or neomycin 3 g/day (n = 24) for 14 consecutive days per month for 6 months.32 Both rifaximin and neomycin significantly reduced mean blood ammonia concentrations and reduced neuropsychiatric symptoms at end of treatment (Figure 2).32

Figure 2.

 Blood ammonia concentrations in patients with hepatic encephalopathy treated with rifaximin 1200 mg/day or neomycin 3 g/day. *P < 0.001 vs. baseline. Data from Miglio et al.32

Comparison with non-absorbable disaccharides

Rifaximin has also been compared with non-absorbable disaccharides.23, 26–29, 37, 39 Overall, both rifaximin and disaccharides reduced blood ammonia levels and improved neuropsychiatric symptoms. However, rifaximin was associated with earlier, more marked improvements and was better tolerated than the non-absorbable disaccharides.

Rifaximin 1200 mg/day (n = 50) was compared with lactitol 60 g/day (n = 53) administered for 5–10 days in patients with HE stage 1 (30%), 2 (49%), or 3 (21%).26 Both treatment groups exhibited improvement in neurologic parameters and reductions in blood ammonia concentrations. However, the overall PSE index improved more with rifaximin than with lactitol (Table 3), a finding attributed to greater improvement in EEG results and blood ammonia levels with rifaximin. Comparable percentages of patients exhibited clinical improvement at the end of treatment (82% for rifaximin and 80% for lactitol).

Table 3.   Improvement in hepatic encephalopathy parameters in patients treated with rifaximin 1200 mg/day or lactitol 60 g/day
 Rifaximin 1200 mg/day (n = 50)Lactitol 60 g/day (n = 53)P-value*
  1. HE, hepatic encephalopathy; PSE, portosystemic encephalopathy.

  2. * Rifaximin vs. lactitol at end of treatment.

  3.  Including number connection test.

  4. Data from Mas et al.26

Blood ammonia (g/dL)121701241090.008
Mean PSE index0.610.140.550.210.01
Mean improvement in HE (%)70620.008

In patients with stage 1, 2 or 3 HE, rifaximin 1200 mg/day (n = 20) significantly improved blood ammonia concentrations, EEG results, overall HE score, asterixis score and mental state compared with lactulose 60 g/day (n = 20) (P < 0.05 for each assessment).28 No drug-related adverse events were reported with rifaximin, whereas GI effects, including nausea (n = 5), flatulence and diarrhoea (n = 13), and abdominal cramps (n = 3), were reported with lactulose.

Rifaximin also compared favourably with lactulose in a study of patients with mild HE.29 Patients received rifaximin 1200 mg/day (n = 20) or lactulose 120 mg/day (n = 20) for the first 2 weeks of each month for 90 days. Both treatments improved mental state vs. baseline, and improvements were significantly greater with rifaximin therapy than with lactulose therapy after 60 and 90 days of treatment (P < 0.05 and P < 0.02, respectively). Reductions in PSE index were significantly greater with rifaximin vs. lactulose at 15, 30, 60 and 90 days (P < 0.05). No drug-related adverse effects were reported with rifaximin, whereas GI effects, including abdominal pain (n = 10) and nausea (n = 5), were reported with lactulose.

In another study, patients with cirrhosis and HE (n = 58) received rifaximin 1200 mg/day or lactulose 30 g/day for 15 days.27 Study endpoints included improvement in mental status, asterixis, Reitan number connection test, EEG, and blood ammonia levels and were assessed at baseline and days 3, 6, 9, 12 and 15. Compared with lactulose, rifaximin significantly improved all clinical parameters at various time points (P < 0.05 for each significant time point), excluding the Reitan test and asterixis. Rifaximin exhibited a more favourable safety profile than lactulose with no reports of diarrhoea, dyspepsia, or anorexia.

Cost effectiveness of treatment

One single-centre, retrospective chart review compared the frequency of hospitalizations and related outcomes in patients with HE treated with lactulose 60 cc/day for ≥6 months followed by treatment with rifaximin 1200 mg/day for ≥6 months.40 Rifaximin significantly lowered the mean number of reported hospitalizations (0.5) compared with lactulose (1.6) (P < 0.001). In addition, mean time spent hospitalized was significantly lower during rifaximin treatment (0.4 weeks) vs. lactulose (1.8 weeks) (P < 0.001). These findings translated into hospitalization charges four times lower with rifaximin ($14 222 per day) than lactulose ($56 635 per day) (charges calculated in 2005 dollars).40

In contrast, another study employed decision analysis with Markov modelling to determine the cost effectiveness of six treatment strategies (including no treatment, lactulose, lactitol, neomycin, rifaximin and rifaximin salvage in patients with intolerance or no response to lactulose) in patients with subclinical to stage 2 HE.41 Using discounted cost per life years gained as the primary outcome, the authors determined that rifaximin was not cost effective as first-line HE therapy. However, they concluded that rifaximin salvage therapy in patients failing lactulose was a highly cost-effective treatment option in mild to moderate HE.41

In summary, multiple clinical studies have demonstrated rifaximin is at least comparable with non-absorbable disaccharides in treatment of HE, is more effective in lowering blood ammonia concentrations and demonstrates a favourable tolerability profile. However, these results should be interpreted in the context of the limitations imposed by the small sample sizes and unblinded design of many of the studies.

Other therapies

Sodium benzoate

Sodium benzoate reduces serum ammonia levels by increasing ammonia excretion in urine;1 however, limited data support the benefits of sodium benzoate therapy for HE. One prospective study compared the efficacy of sodium benzoate (n = 38) with lactulose (n = 36) in patients with cirrhosis or surgical portosystemic anastomosis who presented with an HE exacerbation.42 Sodium benzoate improved symptoms of HE in 80% of patients compared with 81% receiving lactulose (20%). A similar incidence of adverse events was observed between the two groups.42

Branched-chain amino acids

A primary function of the liver is to regulate amino acid supply to peripheral tissues. The balance of physiologic amino acid concentrations is altered in patients with liver disease with an increased ratio of aromatic to branch-chain amino acids.43, 44 It has been suggested that restoring the appropriate balance of amino acid levels might benefit HE patients.43 Anecdotal reports have provided some support for this notion;45 however, results from controlled studies indicate there is no consensus concerning the benefits of branched-chain amino acid (BCAA) therapy in HE.43, 46–48 One study compared BCAA therapy with dietary protein supplementation in patients with cirrhosis.46 Oral BCAA supplements (n = 17) and dietary protein (n = 20) were equally effective in restoring the nitrogen balance from negative to positive in all patients. A meta-analysis reviewed clinical trials evaluating the efficacy of BCAA therapy in patients with cirrhosis and acute HE.47 Although pooled analysis of five clinical studies demonstrated significant mental recovery from high-grade HE in patients treated with BCAAs, two studies reported an increased mortality risk. A more recent analysis reviewed controlled BCAA studies for the treatment of chronic HE.48 Only two trials could be evaluated, and authors concluded that large, multicentre studies are needed to confirm the use of BCAA therapy in HE.

Dopamine agonists

Altered dopaminergic transmission has also been implicated in HE pathogenesis.2 However, there is a paucity of data on the benefit of dopaminergic agonists in HE therapy. One randomized study compared the efficacy of the dopamine agonist bromocriptine vs. placebo in seven patients with cirrhosis and chronic PSE.49 In this small study, bromocriptine was not superior to placebo, and three patients experienced constipation during treatment.

Benzodiazepine receptor antagonists

Benzodiazepines exert depressant effects on the central nervous system by binding to the γ-aminobutyric acid (GABA)–benzodiazepine receptor complex.7 It has been suggested that ‘endogenous benzodiazepines’ may cause neuroinhibitory effects in patients with HE.2 Antagonism of this effect with the benzodiazepine receptor antagonist flumazenil has been evaluated. Intravenous flumazenil (n = 265) improved neurologic scores in 18% of patients with stage 3 HE and 15% of patients with stage 4 HE, compared with 4% and 3% of patients, respectively, treated with placebo (n = 262).50 Flumazenil also improved EEG recordings vs. placebo. The authors concluded that flumazenil is beneficial only in patients with cirrhosis and severe HE;50 further studies are necessary to determine the benefits of flumazenil in patients with subclinical, mild, or moderate HE.

Other therapies

l-ornithine-l-aspartate lowers serum ammonia levels by providing substrates for the intracellular metabolic conversion of ammonia to urea and glutamine.1, 2 Results from controlled trials suggest that ornithine-aspartate reduces ammonia levels and provides therapeutic benefits in patients with chronic mild to moderate HE.51, 52 One study compared ornithine-aspartate 20 g/day for 7 days with placebo in 126 patients with cirrhosis, hyperammonaemia (>50 μmol/L) and chronic HE.51 Ornithine-aspartate significantly improved venous ammonia concentration (P < 0.01), mental status (P < 0.001) and PSE index (P < 0.01). Adverse events consisted of mild GI disturbances and were reported in 5% of patients administered ornithine-aspartate. A second study evaluated the efficacy of oral ornithine-aspartate 18 g/day for 14 days in 66 patients with cirrhosis, hyperammonaemia, and stable, chronic HE.52 Ornithine-aspartate significantly improved blood ammonia concentrations, mental state grade and PSE index compared with placebo (P < 0.05 for each comparison), and no adverse events were reported. These findings point to l-ornithine-l-aspartate as a promising HE treatment that deserves further evaluation.

Levocarnitine has been suggested to lower blood ammonia levels by enhancing metabolic energy production.1, 53 In 150 patients with mild or moderate HE, levocarnitine significantly reduced serum ammonia levels and improved mental status compared with placebo (P < 0.05).54 However, in an earlier clinical study, following rectal ammonium administration in patients with cirrhosis, levocarnitine provided no significant protection against increases in ammonia levels compared with placebo.55

One controlled study evaluated treatment acarbose in 107 patients with mild or moderate HE.56 Acarbose 150–300 mg/day significantly decreased blood ammonia levels and improved number connection test scores vs. placebo (P < 0.01). However, acarbose was associated with adverse events including abdominal bloating/pain, flatulence and increased frequency of bowel movements. The mechanism whereby acarbose improves clinical manifestations of HE is unknown, and its use as a therapeutic agent requires further study.


Hepatic encephalopathy is a complex disease with clinical manifestations ranging from subtle neuropathic abnormalities to coma. Limited treatment options are available for HE patients, underscoring the need for safe, well-tolerated therapies with documented efficacy. Based on the principle that accumulation of unmetabolized ammonia contributes to HE pathogenesis, therapeutic approaches are targeted at reducing bacterial production of ammonia and enhancing its elimination.

Non-absorbable disaccharides are recommended as first-line pharmacotherapy; however, published studies evaluating their safety and efficacy are limited. Therapies such as sodium benzoate, dopamine agonists and benzodiazepine receptor antagonists are alternative options for treatment of HE, but studies confirming the efficacy and safety of these agents are lacking. Antibiotic therapy has shown considerable potential for the treatment of HE, but concerns about safety following systemic absorption have generally limited their use.

The non-absorbed antibiotic rifaximin has been extensively studied, and available evidence indicates that rifaximin provides greater therapeutic benefits than placebo and non-absorbable disaccharides and at least comparable benefits with other antibiotics in the treatment of HE. The favourable tolerability profile of rifaximin clearly differentiates it from lactulose, the use of which is limited by unpredictable, severe diarrhoea. Additionally, rifaximin lacks the toxicity of aminoglycoside antibiotics such as neomycin, which can cause hearing loss and renal injury.

The clinical studies reviewed herein should be considered in the context of the challenges in conducting clinical studies in HE – a therapeutic area characterized by lack of validated measurements for assessing clinical symptoms, and a very ill patient population with multiple comorbidities and medication regimens. Thus, in addition to the need for effective, well-tolerated therapies, further study is required to determine the optimal method for assessing the benefits of treatment in patients with HE.


Grant/research support (Salix Pharmaceuticals, Inc.).

Guarantor of Submission

N. M. Bass, MD.