Semaglutide in MASH: What a 52-Week Biopsy Trial Found

A randomised trial in Diabetes, Obesity and Metabolism, published 17 August 2026, set out to answer a straightforward question: does adding an SGLT2 inhibitor to semaglutide improve liver biopsy results in people with biopsy-proven MASH and type 2 diabetes? On the face of it, the combination arm beat semaglutide alone on every key histology percentage reported. Yet the study still came up short on its primary endpoints.

That tension, the one between numbers that lean one way and the statistical bar for calling it evidence, is the most instructive part of the paper. It is also the nuance that often gets sanded off in summaries.

One framing note before getting into the details: these are approved medicines being studied in patients. Peptra Labs supplies semaglutide-class compounds as reference material for laboratory research only, and nothing here should be read as guidance for human use.

What the trial did

Run in Japan, the study was a 52-week, open-label, randomised, parallel-group trial. Adults with biopsy-confirmed metabolic dysfunction-associated steatohepatitis and type 2 diabetes were assigned to either semaglutide plus luseogliflozin or semaglutide alone, both given at antidiabetic doses.

The backbone of the evidence was paired liver biopsies, one at baseline and one at week 52, read by pathologists who were blinded to treatment assignment. That is the hard version of this kind of trial. Biopsy endpoints are invasive and slow, which is why MASH studies so often lean on imaging or blood-based surrogates. When paired-biopsy trials are done cleanly, they tend to carry extra weight for that reason.

Three primary histological endpoints were set in advance: resolution of MASH without worsening fibrosis, at least a one-point improvement in the non-alcoholic fatty liver disease activity score without worsening fibrosis, and at least a one-stage improvement in fibrosis without worsening MASH.

The trial appears under UMIN000045003 and jRCTs061210009.

The numbers

Sixty participants were randomised, 24 to the combination arm and 36 to semaglutide alone.

At 52 weeks, every pre-specified histology endpoint moved in the combination arm’s favour on a purely numerical read. MASH resolution without worsening fibrosis was reported in 34.9% of the combination group versus 19.4% with semaglutide alone. A one-point-or-more activity score improvement came in at 75.5% versus 55.6%. A one-stage-or-more fibrosis improvement was 26.9% versus 13.9%.

Put side by side, those pairs look like a win. The authors do not present them as one.

Why the paper is more cautious than the percentages

The authors state that, in the pre-specified full analysis set, adding luseogliflozin at antidiabetic doses did not significantly improve the pre-specified histological endpoints compared with semaglutide alone. In the pre-specified per-protocol analysis, the activity score endpoint reached nominal significance.

Most of the meaning sits in three phrases.

“Full analysis set” means analysing participants as randomised, including those who stopped early or drifted from the protocol. That approach is the more conservative one, and it protects the value of randomisation.

“Per-protocol” narrows the view to people who completed the trial as planned. It can make a treatment look better, because the group that finishes a study cleanly is often not the same as the group that drops out, and the comparison stops being purely random.

And “nominally significant” signals that a p value crossed a threshold without correcting for the fact that several endpoints were tested. With three primary endpoints and two different analysis sets, that adjustment matters.

One other practical detail matters here: the split was 24 versus 36 participants. In samples that small, a change of just a few individuals can swing a headline percentage. That is a feature of the study’s size, not an issue with how the results were reported.

What did improve

This was not a blanket null. The authors report that the combination reduced body weight, HbA1c, aminotransferases, and FibroScan-derived liver stiffness, with no new safety issues flagged.

Those are real changes in measures that clinicians and researchers track. Still, they are not the biopsy-based primary endpoints the trial was built around, and paired-biopsy designs exist precisely because tissue is the standard against which surrogate markers get judged.

How this fits with the broader semaglutide MASH story

Semaglutide has a long paper trail in this disease area, so a 60-participant add-on trial lands in a context that readers should keep in view.

A placebo-controlled trial of subcutaneous semaglutide in nonalcoholic steatohepatitis appeared in the New England Journal of Medicine in 2021. A phase 3 trial in metabolic dysfunction-associated steatohepatitis followed in June 2025, also in the same journal.

This new trial is not a fresh referendum on semaglutide itself. It is narrower: does layering an SGLT2 inhibitor on top deliver extra histological benefit at antidiabetic doses? It is a reasonable question, studied at a smaller scale, and the answer on the pre-specified primary analysis is not yet.

How to read this kind of result

This pattern repeats across clinical research. A paper reports several endpoints, some locked in as primary and others treated as supportive. The results section contains percentages that look favourable, and the conclusion steps back from making a broad claim.

Both can be true at once. Those percentages are what happened in this sample. The conclusion reflects what the dataset can support as a general statement. Coverage that quotes the first while skipping the second is not wrong about the numbers, but it misstates what the study actually found.

A similar issue comes up when post hoc or composite outcomes are used to count how many people hit multiple targets at once. Just as in those cases, what a composite endpoint does and does not show depends heavily on rules chosen before the data are examined.

For laboratory work

In a research setting, the bigger takeaway is the design, not the clinical verdict.

A paired-biopsy trial, read by blinded pathologists with analysis sets defined in advance, remains a reference standard in this field. It is the template that surrogate-marker studies are judged against. It also serves as a reminder that if a combination arm is meant to answer a specific add-on question, it needs to be sized to that question, not simply built by stacking two active agents.

For labs working across the incretin class, our tirzepatide reference material documentation covers identity and purity fields, and our ranking of the most-studied research peptides tracks where the published evidence currently sits for each compound.

References

• Miyake T, Yoshida O, Furukawa S, et al. Efficacy of Luseogliflozin Added to Semaglutide in Patients With Metabolic Dysfunction-Associated Steatohepatitis and Type 2 Diabetes Mellitus: A Randomised, Open-Label Trial. Diabetes, Obesity and Metabolism, 17 August 2026. doi 10.1111/dom.71233
• Phase 3 Trial of Semaglutide in Metabolic Dysfunction-Associated Steatohepatitis. New England Journal of Medicine, 5 June 2025
• Newsome PN, Buchholtz K, Cusi K, et al. A Placebo-Controlled Trial of Subcutaneous Semaglutide in Nonalcoholic Steatohepatitis. New England Journal of Medicine, 25 March 2021

The products referenced on this site are supplied for laboratory research use only. They are not medicines and are not intended for human or veterinary use. This article summarises published research for informational purposes and is not medical advice. Statements about third-party studies belong to their authors.