If a drug improves a fatty liver, the interesting question is whether it did anything the weight loss would not have done by itself. A post hoc analysis of a phase 2 trial in MASH tried to answer it directly, taking participants (n = 170) with stage F2 to F3 fibrosis and paired biopsies and apportioning each liver outcome between a weight-dependent and a weight-independent path [1]. The comparison between the marketed drugs on liver endpoints is in tirzepatide against semaglutide.
The pattern that emerged is cleaner than these analyses usually produce. Liver fat behaved like a weight problem: 77.4% of the effect on the FibroScan attenuation measure and 58.2% of the effect on MRI-measured fat fraction ran through weight reduction. So did most of the effect on MASH itself, at 66.7% for resolution and 71.8% for improvement.
Fibrosis behaved differently. Only 36.3% of the effect on fibrosis improvement was attributed to weight, leaving the majority to a direct action of the drug. The blood markers pointed the same way, with weight mediating between 16.5% for AST and 38.6% for the Enhanced Liver Fibrosis score. The authors suggest glucagon receptor agonism as the route, which is what distinguishes survodutide from a pure GLP-1 drug, and places it among the multi-receptor agents surveyed in the network meta-analysis of nineteen drugs.
Three further limits keep this in proportion. It is post hoc, so the questions were chosen after the results were known. It pools the dose arms, discarding the dose-response information that is the main output of a phase 2 trial. And 170 participants with biopsies at both ends is a small number on which to divide six separate effects in two.
Survodutide is investigational and cannot be bought, so nothing here is a recommendation about anything currently available. What it contributes is a mechanistic claim worth testing properly: that in the liver, the fat and the scarring may respond to different things, and a drug that only removes weight may do less for the second. That would matter for the marketed drugs too, whose liver results are usually reported without this separation — as are their effects on body composition generally, covered in what these drugs do to muscle and in the pooled weight figures.