BIOANALYSISPAPER
A peer reviewed mediation analysis — one that separates how much of a drug's effect runs through weight loss versus other pathways — of the survodutide phase 2 trial in MASH, a severe form of fatty liver disease, finds that roughly two thirds to
A new peer-reviewed analysis of the phase 2 trial of survodutide in MASH finds that the drug's measured benefit on liver scarring and inflammation came mostly through a pathway other than weight loss, even though its effect on liver fat tracked the pounds shed.
MASH, the current formal name for what used to be called NASH, is a severe form of fatty liver disease in which fat buildup, inflammation, and scarring combine. Weight loss can help, but in MASH it often leaves residual scarring. That clinical gap is the space a drug with a separate hepatic effect would have to close.
The paper, published in Hepatology, is a post-hoc causal mediation analysis of the 170-participant phase 2 trial registered as NCT04771273, in which adults with fibrosis stages F2 and F3, meaning significant but pre-cirrhosis liver scarring, received 48 weeks of survodutide or placebo. The original primary results were reported in the New England Journal of Medicine in 2024. The mediation analysis asks a separate question: of the drug's measured effect on each liver endpoint, how much traveled through percent change in body weight, and how much through a direct hepatic pathway the model attributes to the drug itself?
Survodutide is an injectable dual GLP-1 and glucagon receptor agonist under investigation for obesity and MASH. GLP-1 agonism drives appetite suppression and weight loss; glucagon agonism is the arm the authors point to as the likely source of the extra hepatic effect.
A mediation analysis separates an observed treatment effect into a share that runs through a chosen intermediate variable and a residual share that does not. In this model, treatment was the exposure, percent change in body weight the mediator, and the covariates were baseline body weight, type 2 diabetes status, and fibrosis stage. The share of each endpoint's total effect that the model routed through a non-weight pathway: 66.7% for MASH resolution without worsening of fibrosis, 71.8% for MASH improvement without worsening of fibrosis, and 36.3% for improvement in fibrosis without worsening of MASH. For steatosis (liver-fat) endpoints, weight loss carried more of the effect: 58.2% of MRI-measured proton density fat fraction and 77.4% of FibroScan's controlled attenuation parameter, a measure from an ultrasound-based liver scan. Non-invasive inflammation and fibrosis markers sat between, with 16.5% for AST, a liver-injury enzyme, and 38.6% for the Enhanced Liver Fibrosis (ELF) score, a blood-test composite, attributed to the non-weight pathway.
The authors' interpretation: the inflammation and fibrosis endpoints show weight-independent work, consistent with a direct hepatic effect from the glucagon-receptor arm of the drug. The fat-content endpoints are mostly a downstream read of weight loss.
That interpretation comes with three caveats. The analysis is post-hoc, the survodutide dose arms were pooled, and 170 paired biopsies is a small evidentiary base. Causal mediation gives a modeled answer, not a direct mechanism measurement. The result is hypothesis-generating for the GLP-1/glucagon class, not a confirmation that glucagon agonism is the lever on scarring.
A larger, dose-stratified trial will decide whether the mechanism reading is real or a small-sample artifact.