A cheap diabetes pill improved fatty liver disease in a 154-person trial

Metabolic dysfunction-associated steatohepatitis — MASH, the inflammatory form of fatty liver disease — has spent two decades as a condition with no approved drug in most of the world. This month produced movement on two fronts that could hardly be more different: a randomised trial of a generic diabetes tablet, and a regulatory opinion on a drug designed for the disease from the start.

The trial

The BMJ published a multicentre, double-blind, randomised, placebo-controlled trial of dapagliflozin, a sodium-glucose cotransporter 2 inhibitor, on 4 June [s1].

It ran at six tertiary hospitals in China from 23 November 2018 to 28 March 2023, and enrolled 154 adults with biopsy-diagnosed MASH, with or without type 2 diabetes [s1]. Participants were randomly assigned to 10 mg of oral dapagliflozin or matching placebo once daily for 48 weeks [s1].

The primary endpoint was MASH improvement — defined as a decrease of at least 2 points in the non-alcoholic fatty liver disease activity score (NAS), or a NAS of 3 or less — without worsening of liver fibrosis, meaning no increase in fibrosis stage [s1].

On that endpoint, 53% (41 of 78) of the dapagliflozin group improved against 30% (23 of 76) on placebo, a risk ratio of 1.73 (95% CI 1.16 to 2.58; P=0.006) [s1]. The mean difference in NAS was -1.39 (95% CI -1.99 to -0.79; P<0.001) [s1].

Two secondary endpoints moved in the same direction. MASH resolution without worsening of fibrosis occurred in 23% (18 of 78) on dapagliflozin against 8% (6 of 76) on placebo, a risk ratio of 2.91 (95% CI 1.22 to 6.97; P=0.01) [s1]. Fibrosis improvement without worsening of MASH occurred in 45% (35 of 78) against 20% (15 of 76), a risk ratio of 2.25 (95% CI 1.35 to 3.75; P=0.001) [s1].

Discontinuation for adverse events was 1% (1 of 78) on dapagliflozin and 3% (2 of 76) on placebo [s1]. Analyses used the intention-to-treat dataset, and the trial is registered as NCT03723252 [s1].

What 154 people can and cannot establish

The effect sizes are large and the confidence intervals exclude no effect, but the trial is small by the standards of liver-outcome research. The lower bound of the risk ratio for MASH resolution — 1.22 — sits close to 1, on 18 versus 6 events. Small event counts produce wide intervals and unstable point estimates, and a risk ratio of 2.91 built on 24 total events should be read as directional rather than precise.

Histological endpoints are also surrogate endpoints. The trial reports biopsy scores at 48 weeks; it does not report cirrhosis, liver transplantation, hepatocellular carcinoma or death, and 48 weeks is not long enough to. Whether histological improvement in a SGLT2 inhibitor trial translates into fewer hard liver events over a decade is not something this trial was built to answer.

The setting is a single country, six tertiary hospitals, and participants had biopsy-confirmed disease — a group already selected by being sick enough and willing enough to undergo liver biopsy twice [s1].

The regulatory track, running in parallel

Six weeks after the trial appeared, the European Medicines Agency's Committee for Medicinal Products for Human Use met on 16-19 June and recommended granting a conditional marketing authorisation for Rezdiffra (resmetirom), for the treatment of adults with non-cirrhotic metabolic dysfunction-associated steatohepatitis [s2]. The agency's own summary notes there is currently no authorised treatment for MASH in the European Union [s2]. The applicant is Madrigal Pharmaceuticals EU Limited, and the therapeutic indication as listed covers MASH with liver fibrosis [s2].

A CHMP positive opinion is a recommendation, not an authorisation; the European Commission decision follows separately, and the EMA lists the product as pending an EC decision [s2]. "Conditional" marketing authorisation means the evidence package is incomplete and further data are required after approval — a category the agency uses when unmet need is judged to outweigh residual uncertainty.

The June meeting recommended 13 new medicines in total, of which three were new non-orphan medicines, two orphan medicines, six biosimilars and two generic, hybrid or informed-consent medicines; there were no negative opinions on new medicines [s2].

Two different bets on the same disease

The two developments are not in competition, and it would be a mistake to read the trial as an argument against the new drug or vice versa. They represent different strategies. Resmetirom was developed for MASH and carries a MASH-specific evidence package assessed by a regulator [s2]. Dapagliflozin is an established, widely available, off-patent-adjacent diabetes and cardiorenal drug that a single 154-person trial suggests may also improve liver histology [s1].

Neither this article nor the trial constitutes guidance about treatment. Dapagliflozin is not approved for MASH in any jurisdiction on the basis of this trial, and the trial itself does not claim it should be.

What to watch

Whether the dapagliflozin result is replicated in a larger, longer, multi-country trial with fibrosis-stage endpoints and adequate power — the standard that resmetirom's package had to meet. And whether the European Commission follows the CHMP opinion on resmetirom, and on what conditions [s2].

Sources

• [s1] Effect of dapagliflozin on metabolic dysfunction-associated steatohepatitis: multicentre, double blind, randomised, placebo controlled trial. The BMJ, 4 June 2025. https://doi.org/10.1136/bmj-2024-083735
• [s2] Meeting highlights from the Committee for Medicinal Products for Human Use (CHMP) 16-19 June 2025. European Medicines Agency, 20 June 2025. https://www.ema.europa.eu/en/news/meeting-highlights-committee-medicinal-products-human-use-chmp-16-19-june-2025