ARiE Trial – Rifaximin for encephalopathy

ARiE Trial

Antibiotics With or Without Rifaximin for Acute Hepatic Encephalopathy in Critically Ill Patients With Cirrhosis: A Double-Blind, Randomized Controlled (ARiE) Trial

Kulkarni et al. Am J Gastroenterol 2024; doi: 10.14309/ajg.0000000000002575

Clinical Question

  • In critically ill adults with overt acute hepatic encephalopathy (HE) admitted to ICU, does the addition of rifaximin to broad spectrum antimicrobials affect time to resolution of HE?


  • Rifaximin is an orally administered broad-spectrum antimicrobial which is not absorbed systemically. It reduces local ammonia production in the gut and therefore reduces systemic ammonia absorption
  • It is known that rifaximin reduces the risk of recurrence of overt acute HE in patients with cirrhosis when used in addition to a non-absorbable dissacharide (such as lactulose)
  • A recent meta-analysis suggests rifaximin may decrease mortality and duration of acute HE when used with lactulose, but with low-certainty of evidence (Zacharias et al, 2023)
  • It is unclear if rifaximin improves time to resolution of acute HE, particularly when broad spectrum antimicrobials are already being administered


  • Randomised, parallel group trial
  • Computer generated randomisation in 1:1 ratio
  • Concealment via sequentially numbered sealed envelopes
  • Authors describe a non-inferiority trial in parts of the discussion but trial methodology does not support this
  • Sample size calculation based on a power of 90%, an alpha error of 5% and margin of 15% for resolution of HE
  • Intention to treat and per-protocol analysis carried out


  • Single centre study from Hyderabad, India
  • Participants recruited between 28th February 2022 and 3rd April 2023


  • Inclusion:
    • Adults aged 18-75 with overt HE (West Haven score of 2 or greater) admitted to ICU with a background of cirrhosis
      • Cirrhosis was defined by a clinical history of decompensation (ascites, HE or variceal bleeding) with supportive imaging or biopsy findings
    • Exclusion:
      • Acute liver failure (jaundice to encephalopathy <4 weeks)
      • Urea on presentation >150mg/dl (8.3 mmol/l)
      • Underlying chronic renal failure (eGFR <60 for >3 months)
      • Hepatocellular carcinoma
      • History of alcohol consumption within 1 week
      • Death within 24 hours of admission (before randomisation)
    • 260 patients assessed for eligibility, 184 randomised
      • 76 excluded consistent with criteria above
      • No loss to follow-up
      • Crossover in 5 patients (3 discontinued rifaximin, 2 commenced)
    • No major difference in baseline characteristics –

Comparing antibiotics only group to rifaximin + antibiotics:

  • Age 48 vs 47 years
  • Sex 88% vs 89% male
  • Etiology primarily alcohol (64% vs 68%) followed by NASH (14% vs 17%) and viral (10% vs 7%)
  • AKI on presentation in 61% vs 50%
    • LOLA used in 41% vs 50% (noting above)
  • ACLF criteria met in 50% vs 54%
  • SOFA 8.9 vs 8.8, MELD 29.6 vs 29.1
  • 32% vs 28% were mechanically ventilated
  • 45% vs 40% were diagnosed with infection
  • Antimicrobials –
    • Carbapenems 36% vs 40%
    • Cephalosporin/β-lactamase inhibitors 39% vs 34%
    • Penicillin/β-lactamase inhibitor 16% vs 21%
  • Note grade of encephalopathy – higher percentage of low-grade (II) encephalopathy in antimicrobial only group
    • II 29% vs 20%
    • III 60% vs 70%
    • IV 11% vs 9%


  • Rifaximin and broad spectrum antibiotics
    • Rifaximin
      • 550mg twice daily orally or via NG
    • Broad spectrum antimicrobials
      • carbapenems, cephalosporin/β-lactamase inhibitors, penicillin/β-lactamase inhibitors, glycopeptides, fluroquinolones and macrolides were considered broad-spectrum
    • Length of course not specified


  • Broad-spectrum antibiotics (as above)

Management common to both groups

  • Authors report all patients received lactulose either enterally or via enema titrated to stool frequency (2-3/day)
    • Total doses not reported
  • Treatment of underlying precipitant for HE was instigated as per hospital policy (e.g. protocolised septic screen, terlipressin and albumin for AKI, screening for intracerebral pathology)


  • Primary outcome:
    • Two grade reduction and/or complete resolution in HE:
    • 46.7% (95% CI 33.8-63) in intervention vs 44.6% (95% CI 32-70) P = 0.84 over 4.1 vs 3.6 days P=0.27 – no significant difference
  • Secondary outcomes:
  • Comparing antimicrobials + rifaximin vs. antimicrobials only,
    • No significant difference in:
      • In-hospital mortality
        • 50% (CI 37-67) vs 62% (CI 47-80) HR 0.7; P = 0.06
      • Duration of hospital stay
        • 10 days vs 9 days, P =0.18
      • Incidence of nosocomial infection
        • 13% vs 6.5% P=0.21
      • Change in endotoxin level
        • -0.01 EU/ml vs +0.08EU/ml P= 0.64
  • Subgroup analysis
    • In patients with decompensated cirrhosis (n=88), there was a statistically significant improvement in mortality (56.5% vs 33.3% P=0.003) in patients given rifaximin + antimicrobials, compared with antimicrobials along (a significant difference in nosocomial infections was noted between these groups)

Authors’ Conclusions

  • The addition of rifaximin to broad-spectrum antimicrobials does not result in early HE resolution in critically ill patients with cirrhosis and does not reduce the incidence of infection or endotoxin levels.
  • In patients with decompensated cirrhosis rifaximin may reduce in-hospital mortality and rates of nosocomial infection, while in ACLF patients rifaximin has no effect


  • The study authors ask a pragmatic question – is rifaximin required in HE when already on broad-spectrum antimicrobials?
  • Performed in population of interest to critical care physicians, significant numbers of patients requiring mechanical ventilation with high SOFA scores
  • Both groups reportedly received lactulose, an already established treatment
  • The aetiology of underlying liver disease was primarily alcohol, and to a lesser extent non-alcoholic steatohepatitis, similar to many developed nations
  • Authors were careful to protocolise and describe infection screens and analysis of infective causes and describe a rigorous approach to diagnostic work-up of patients with HE


  • Single centre
    • Use of broad-spectrum antimicrobials will vary significantly from nations with lower rates of antimicrobial resistance, and empiric use of these agents outside of variceal bleeding is not established practice as conducted in this trial
  • No placebo group introduces a significant risk of bias
  • The authors state they chose a non-inferiority trial design in the discussion but trial methodology, sample size/ power calculation and statistical analysis reporting does not clearly support this
  • HE assessed only once/ 24 hours could underestimate between group differences, missing a true difference in HE resolution times
  • Trial participants without AKI also received l-Ornithine l-Aspartate (LOLA), another theorised treatment for HE (50% in the rifaximin group vs 41% in the control group) – potential confounder
  • The ACLF subgroup analysis seems to indicate no statistical difference between the intervention group on mortality, but of note the ACLF intervention group appeared to have higher grades of encephalopathy and higher MELD and SOFA scores. In addition the improvement in mortality in decompensated cirrhosis groups was attributed to fewer nosocomial infections however with little physiological basis for how rifaximin would result in this

The Bottom Line

  • I will continue to use lactulose as my current standard treatment for HE in critically ill patients
  • Where lactulose is not tolerated or HE is refractory to this therapy, I consider rifaximin as a relatively safe adjunct therapy that may improve HE resolution and recurrence, particularly in those with decompensated cirrhosis


External Links

Summary author: Christopher Smith @ChrisCritCare

Peer review editor: Aniket Nadkarni; @AniketNadkarni1

Image by dtiner76 from Pixabay

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.