Alcohol is the leading cause of liver-related mortality worldwide. The gut-liver axis is considered a key driver in alcohol-related liver disease. Rifaximin-α improves gut-barrier function and reduces systemic inflammation in patients with cirrhosis. We aimed to compare the efficacy and safety of rifaximin-α with placebo in patients with alcohol-related liver disease. GALA-RIF was an investigator-initiated, randomised, double-blind, placebo-controlled, single-centre, phase 2 trial done at Odense University Hospital in Denmark. Eligible participants were adults (aged 18-75 years) who had current or previous alcohol overuse (at least 1 year with ≥24 g of alcohol per day for women and ≥36 g of alcohol per day for men), biopsy-proven alcohol-related liver disease, and no previous hepatic decompensation. Patients were randomly allocated (1:1) through a web-based randomisation system to receive oral rifaximin-α (550 mg) twice daily or matched placebo for 18 months. Randomisation was done in blocks of four and stratified according to fibrosis stage and alcohol abstinence. Participants, sponsor, investigators, and nurses involved in the study were masked to the randomisation outcome. The primary endpoint was a histological decrease from baseline to 18-month treatment of at least one fibrosis stage, according to the Kleiner fibrosis score. We also assessed the number of patients with progression by at least one fibrosis stage from baseline to 18 months. Primary analyses were done in the per-protocol and modified intention-to-treat populations; safety was assessed in the full intention-to-treat population. The per-protocol population was defined as all randomly assigned patients who did not present serious protocol violations, who ingested at least 75% of the treatment, and who were not withdrawn from the study due to non-adherence (interruption of treatment for 4 weeks or more). Participants receiving at least one dose of the intervention were included in the modified intention-to-treat analyses. This completed trial is registered with EudraCT, number 2014-001856-51. Between March 23, 2015, and Nov 10, 2021, we screened 1886 consecutive patients with a history of excessive alcohol consumption and no previous hepatic decompensation, of whom 136 were randomly assigned to either rifaximin-α (n=68) or placebo (n=68). All patients were White (100%), 114 (84%) were men, and 22 (16%) were women. 133 (98%) patients received at least one dose of the intervention and were included in the modified intention-to-treat analysis; 108 (79%) completed the trial per protocol. In the per-protocol analysis, 14 (26%) of 54 patients in the rifaximin-α group and 15 (28%) of 54 patients in the placebo group had a decrease in fibrosis stage after 18 months (odds ratio 1·10 [95% CI 0·45-2·68]; p=0·83). In the modified intention-to-treat analysis, 15 (22%) of 67 patients in the rifaximin-α group and 15 (23%) of 66 patients in the placebo group had a decrease in fibrosis stage at 18 months (1·05 [0·45-2·44]; p=0·91). In the per-protocol analysis, increase in fibrosis stage occurred in 13 (24%) patients in the rifaximin-α group and 23 (43%) patients in the placebo group (0·42 [0·18-0·98]; p=0·044). In the modified intention-to-treat analysis, increase in fibrosis stage occurred in 13 (19%) patients in the rifaximin-α group and 23 (35%) patients in the placebo group (0·45 [0·20-1·02]; p=0·055). The number of patients with adverse events (48 [71%] of 68 patients in the rifaximin-α group; 53 [78%] of 68 in the placebo group) and serious adverse events (14 [21%] in the rifaximin-α group; 12 [18%] in the placebo group) was similar between the groups. No serious adverse events were deemed related to treatment. Three patients died during the trial, but none of the deaths were considered treatment related. In patients with alcohol-related liver disease, rifaximin-α might reduce progression of liver fibrosis. These findings warrant confirmation in a multicentre phase 3 trial. The EU Horizon 2020 Research and Innovation Program and The Novo Nordisk Foundation.
Read full abstract