Not in the trial that tested it. In 71 people with compensated MASH cirrhosis, semaglutide 2.4 mg did not improve liver scarring over 48 weeks [1]. Fibrosis improved in 11% on the drug and 29% on placebo, a difference that was not significant. The approved use stops at stage F3, before cirrhosis, as the liver disease eligibility guide sets out.
The broader question of liver safety is covered in whether GLP-1 drugs are bad for the liver. This page covers the last stage of the disease only.
The cirrhosis trial
The trial enrolled adults with biopsy-confirmed MASH cirrhosis and a BMI of 27 or more. Seventy-five percent had diabetes. They were randomized 2:1, 47 to semaglutide 2.4 mg weekly and 24 to placebo. The primary endpoint was fibrosis improving by at least one stage without the steatohepatitis worsening.
That happened in 5 of 47 on semaglutide and 7 of 24 on placebo. The odds ratio was 0.28, with a 95% confidence interval from 0.06 to 1.24. Resolution of steatohepatitis did not differ either. With 71 people, the trial could miss a modest benefit.
Safety held within the trial. There were no decompensating events or deaths, and liver and kidney function stayed stable. Nausea affected 45% on semaglutide against 17% on placebo, and vomiting 17% against none.
The 2026 trial that included cirrhosis
A larger phase 2 trial enrolled 698 people with MASH and fibrosis from stage F2 to compensated cirrhosis [2]. Its main question was whether adding zalfermin to semaglutide helped, and it did not. Zalfermin 30 mg plus semaglutide improved fibrosis in 24% against 16% on placebo, p = 0.19.
Semaglutide 2.4 mg alone reached 30% against 16% at 52 weeks. The authors describe that as nominally significant, p = 0.024, and it sat outside the protected primary comparison. A cagrilintide plus semaglutide arm did not differ substantially from placebo, which matters for the CagriSema comparison.
What the cohorts add
A US veterans cohort compared GLP-1 drugs with DPP-4 inhibitors in people with fatty liver disease and diabetes [3]. In 14,606 users without cirrhosis, progression to cirrhosis ran 9.98 against 11.10 events per 1,000 person-years. The hazard ratio was 0.86. In 1,452 users who already had cirrhosis, no association appeared with any outcome.
The authors read that as a case for treating earlier in the disease. Why scarring can respond differently from liver fat is examined in the survodutide mediation analysis.
A single-center cohort looked at hepatic encephalopathy, a brain complication of advanced liver disease [4]. Of 2,557 people with cirrhosis, 139 took a weight-loss dose for six months or more. They developed it less often, 3.6% against 18.7%, with an adjusted hazard ratio of 0.33, P = 0.048.
The direction reversed in people who also had liver cancer, where the hazard ratio was 6.12, P = 0.024. The absolute cancer risk on these drugs is set out in the liver cancer cohort.
What is not yet known
Neither trial showed semaglutide reversing scarring once cirrhosis was present. Decompensated cirrhosis was excluded from both trials, so it is not measured. The encephalopathy finding comes from 139 treated people at one center. The earlier-stage evidence, where results are stronger, sits in the phase 3 liver trial.