Mazdutide GLORY-1: Approved in One Country Only

GLORY-1 established that, in 610 Chinese adults randomised to one of two mazdutide arms or placebo, body weight at weeks 32 and 48 separated clearly from placebo under a treatment-policy analysis, with low rates of discontinuation for adverse events 1. It did not establish how those figures transfer to populations outside China, because no participant was recruited elsewhere 2. Evidence tier for the trial's findings: human randomised controlled trial. Evidence tier for the molecular lineage: biochemical characterisation and a human phase 2 trial.

Mazdutide is described by its developer as a mammalian oxyntomodulin analogue that activates both the GLP-1 receptor and the glucagon receptor 3. It is also the site's worked example of a regionally approved compound: a drug whose registration rests on one country's trial population and one country's regulator. The approval scope is stated below exactly as the sources state it, because several secondary pages misstate it.

Oxyntomodulin and the lineage of dual agonists

Evidence tier for this section: biochemical and receptor pharmacology, with the compound-specific statements resting on the developer's description. Oxyntomodulin is a peptide produced from the same precursor as GLP-1 and glucagon, and it activates both of their receptors. That dual activity is the pharmacological idea behind the whole class: one ligand, two receptors, with the ratio between the two activities set by the amino-acid sequence. A native oxyntomodulin is cleared quickly, so a development programme has to alter the sequence to extend exposure.

The reason to retain glucagon-receptor activity is the one set out in the site's piece on a related compound, survodutide: glucagon receptor agonism is proposed to raise energy expenditure and to act on hepatic lipid handling, at the cost of raising blood glucose. Mazdutide differs from that compound in lineage and in programme history, not in the receptor pair it addresses. A phase 2 trial reporting 24-week results in 248 Chinese adults, with weight changes of −6.7%, −10.4% and −11.3% across three arms and +1.0% in placebo, preceded GLORY-1 6.

Trial design: eligibility, randomisation, duration

Evidence tier for this section: human randomised controlled trial, registered design. Eligible adults were aged 18 to 75 with a body-mass index of at least 28, or 24 to below 28 with at least one weight-related coexisting condition 1. The registry lists the qualifying conditions as prediabetes, hypertension, dyslipidaemia, fatty liver, weight-bearing joint pain, obesity-related dyspnoea or obstructive sleep apnoea, and excludes participants with diabetes, a history of pancreatitis, or recent use of weight-loss drugs 2.

Randomisation was 1:1:1 to the lower-dose arm, the higher-dose arm or placebo, given once weekly by subcutaneous injection for 48 weeks. Both active arms escalated from a lower starting level, and the higher arm took longer to reach its assigned level 2. The trial ran from November 2022 to April 2024 at 23 sites, all of them in China. Mean baseline body weight was 87.2 kg and mean body-mass index 31.1, a lighter and lower-BMI population than the large Western registration trials of the same pharmacological class.

Co-primary endpoints and the analysis estimand

Evidence tier for this section: human randomised controlled trial, prespecified analysis. The two primary endpoints were the percentage change in body weight from baseline and the proportion of participants with a reduction of at least 5%, both at week 32 1. The publication assessed them with a treatment-policy estimand, which analyses participants as assigned regardless of early discontinuation or the start of a new antiobesity therapy. That is the more conservative of the two framings.

The developer's release also reports an efficacy estimand, which describes participants while they were taking the assigned regimen 3. The two sets of figures differ, as expected, and the distinction is the same one that governs how any trial in this class should be read. A figure quoted without its estimand is incomplete.

Results at week 32 and week 48

Evidence tier for this section: human randomised controlled trial. At week 32 the mean percentage change in body weight under the treatment-policy estimand was −10.09% in the lower-dose arm, −12.55% in the higher-dose arm and +0.45% in placebo; the proportions with a reduction of at least 5% were 73.9%, 82.0% and 10.5% 1. At week 48 the corresponding changes were −11.00%, −14.01% and +0.30%, and the proportions with a reduction of at least 15% were 35.7%, 49.5% and 2.0%. All comparisons with placebo had P below 0.001.

The developer's release adds secondary figures that the abstract does not give: changes in waist circumference at week 48 of −9.48 cm, −10.96 cm and −1.48 cm, and changes in liver fat of −63.26%, −73.18% and +8.20% in participants with baseline liver fat of at least 5% 3. These are company-reported secondary measures from a release rather than the peer-reviewed text, and their subgroup definitions are narrower than the randomised population. Read them from the publication before relying on them.

Adverse events and discontinuation

Evidence tier for this section: human randomised controlled trial. The most frequently reported adverse events were gastrointestinal and mostly mild to moderate in severity. The incidence of adverse events leading to discontinuation of the trial regimen was 1.5% in the lower-dose arm, 0.5% in the higher-dose arm and 1.0% in placebo 1. The developer's release reports a mean heart-rate increase of 2.6 beats per minute at week 48 in both active arms and no cardiovascular safety signal 3.

Those discontinuation rates are low for the class, and the higher arm's rate was lower than placebo's. A rate that low in an escalating, injectable regimen is a feature of this trial's population and conduct, and it is the figure most worth testing for transfer. Tolerability and retention depend on prescriber practice, counselling, participant expectations and dietary patterns, all of which vary between countries.

Why one country's population limits extrapolation

Evidence tier for this section: methodological. GLORY-1 recruited only in China, at 23 sites 2. The mean baseline weight was 87.2 kg, and eligibility used lower BMI thresholds than Western obesity trials because the thresholds were set for the Chinese population. Body composition at a given BMI, the distribution of visceral fat, background diet and the prevalence of the qualifying comorbidities all differ between populations, and each can modify both the size of a weight response and the rate of gastrointestinal events.

None of this implies that the result would not replicate elsewhere. It implies that the trial cannot show that it would. The distinction matters because the weight-change percentages from this trial are routinely placed in tables next to percentages from multiregional trials as though they were interchangeable, and they are not. Differences in baseline weight alone change what a given percentage means in absolute terms.

Regulatory geography: approved in one jurisdiction

Evidence tier for this section: regulatory record as reported by the developer and a peer-reviewed approval summary; neither is the regulator's own label. Innovent's release dated 27 June 2025 states that China's National Medical Products Administration approved mazdutide for chronic weight management in Chinese adults with overweight or obesity. The release's footnote gives the indication as an adjunct to a reduced-calorie diet and increased physical activity, in adults with an initial body-mass index of at least 28, or at least 24 with at least one weight-related comorbid condition 4. The Drugs approval summary dates the first approval to June 2025 and gives the same thresholds 5.

What a multiregional trial would add

Evidence tier for this section: methodological. A multiregional phase 3 trial would randomise participants across several countries under one protocol, with prespecified analysis by region. It would show whether the effect size, the discontinuation rate and the gastrointestinal event rate hold when baseline body composition, diet and clinical practice vary. It would also supply the denominator that regulators in other jurisdictions typically ask for: a participant population resembling the people to whom a label would apply.

What this means for a researcher reading the literature

Check three things before using a mazdutide figure. Which estimand produced it. Which timepoint it describes, since the primary endpoints were assessed at week 32 and not at the end of the trial. And which population it describes, since every participant was recruited in China 12. A table that places these percentages beside multiregional results without those three qualifiers is comparing quantities that were not measured the same way.

GLORY-1 is a well-conducted trial that answers the question it was designed to answer. What it leaves open is the question of transfer, and an approval in one country, however it is described, does not close that question. It records that one regulator was satisfied by one population's data.