What do the MASH histological endpoints actually mean, and why are they always paired?

Every trial in metabolic dysfunction-associated steatohepatitis reports two co-primary endpoints with oddly constructed names: "resolution of steatohepatitis without worsening of fibrosis" and "improvement in fibrosis of at least one stage without worsening of steatohepatitis". I understand each half separately. What I do not understand is why they are always paired with a "without worsening" clause, and why the two endpoints are reported separately rather than combined.

The clause looks defensive, as though someone expects a treatment to improve one thing at the expense of the other. Is that a real phenomenon in liver disease, or a regulatory belt-and-braces convention?

I would also like to know how to read the numbers. ESSENCE reported resolution rates in the sixties for semaglutide against the thirties for placebo, which sounds enormous, but a placebo resolution rate in the thirties also sounds enormous for a chronic fibrotic disease. If a third of people on placebo resolve their steatohepatitis, either the disease is far more labile than I thought or the measurement is noisier than I thought, and I suspect the second.

Expanding on the fibrosis column, because "improvement of at least one stage" hides an important asymmetry that the table cannot show.

Fibrosis stages are not equal steps. F0 to F1 is the appearance of perisinusoidal fibrosis; F2 is periportal with septa; F3 is bridging; F4 is cirrhosis. The clinical stakes rise steeply and non-linearly, and the risk of liver-related outcomes is concentrated at F3 and F4. So a trial reporting "37% achieved one-stage improvement" is reporting a mixture of F2-to-F1 transitions, which are of modest clinical importance, and F3-to-F2 transitions, which matter a great deal.

Things to look for in the supplementary tables of any of these papers:

• The breakdown of improvement by baseline stage. If the effect is concentrated in F2 participants, the headline overstates the clinically important benefit.
• Progression to cirrhosis as a separate reported outcome. This is the endpoint that regulators and hepatologists care most about and it is often underpowered but directionally informative.
• Whether the fibrosis endpoint was assessed by consensus of multiple readers or by a single central reader, and whether readers were blinded to timepoint. Unblinding to sequence biases towards showing improvement.

The F3 subgroup is also where the two endpoints most often diverge. Advanced fibrosis is less reversible over a 72-week window, so a trial enrolling mostly F3 will report weaker fibrosis numbers than one enrolling mostly F2, independently of the drug. This is a large source of apparent between-trial differences.

Finally, the reason all of this is being replaced as fast as anyone can validate an alternative: paired biopsy is invasive, carries a small but real complication rate, and is the main reason these trials are slow and expensive. Any drug programme in this space is running non-invasive markers in parallel and hoping one qualifies as a surrogate.

On the weight-mediation question for the liver endpoints, since it is the obvious follow-up to the dissociation noted in the accepted answer.

Weight loss improves steatohepatitis in a dose-dependent way, and this was established well before incretins entered the field: bariatric and intensive-lifestyle cohorts show that losses above roughly 10% produce resolution in a majority and fibrosis improvement in a substantial minority. So an agent producing about -10.5% mean weight loss would be expected to improve MASH histology even if it had no direct hepatic action.

That makes the interesting question not "does it work" but "does it work more than the weight loss predicts". Ways the trials address it, none conclusive:

• Comparing histological response across achieved-weight-loss strata within the active arm. Informative but observational, since achieved weight loss is not randomised.
• Comparing agents with similar weight effects and different mechanisms, or agents with no weight effect at all. Resmetirom is the useful anchor here precisely because it moves fibrosis without moving weight.
• For glucagon-containing agents, the mechanistic case for a direct hepatic effect is stronger, since glucagon receptor agonism increases hepatic fatty acid oxidation independently of weight. Survodutide's phase 2b results are consistent with a contribution of that kind, though the trial cannot isolate it [1].

My reading of the current evidence: most of the resolution effect for GLP-1-based agents is plausibly weight-mediated, the fibrosis effect is smaller than weight loss alone would predict from bariatric cohorts, and a direct hepatic contribution for glucagon co-agonists is likely but unquantified. Anyone with actual liver disease needs a hepatologist rather than a trial table, and none of these agents is available for that use outside approved indications and trials.