Mazdutide (IBI362)

Identity and regulatory status

One structure is registered for mazdutide, and three different molecular formulae are attached to it. The FDA Global Substance Registration System holds the registration under UNII MB76Z4IBZ5, with CAS number 2259884-03-0, INN 12034 and USAN code JK-37, and describes a single 34-residue chain, HGQGTFTSDYSKYLDEKKAKEFVEWLLEGGPSSG, carrying 2-aminoisobutyric acid at position 2, a C-terminal glycinamide, and an acyl group on the lysine at position 20 built from a twenty-carbon diacid attached through a gamma-glutamyl spacer and two AEEA linkers. Composing that structure gives C210H322N46O67 and an average mass of 4563.14 g/mol. PubChem's entry does not match it: CID 167312357 reads C207H317N45O65 and about 4476 g/mol, a deposit whose systematic name carries only three of the four serines the registered subunit requires. GSRS then disagrees with its own subunit, its molecular formula property reading C210H328O69N46 against a calculated weight of 3790. The backbone is an oxyntomodulin analogue rather than a GLP-1 analogue (GSRS UNII MB76Z4IBZ5; PMID 41028652).

Eli Lilly originated the molecule as LY3305677 and licensed Chinese rights to Innovent Biologics, which is why the same compound appears in the literature under three codes. China's National Medical Products Administration approved it in June 2025 for long-term weight management in adults with a body-mass index of at least 28, or at least 24 with one or more weight-related comorbidity, and again in September 2025 for glycaemic control in adults with type 2 diabetes. Shirley's 2025 review in Drugs records both approvals and the proprietary name under which the product is marketed in China (PMID 41028652). A query of the FDA Drugs@FDA database by active ingredient returns no record for mazdutide.

That regulatory position is unusual for a compound of this class: approved and marketed in one jurisdiction, investigational everywhere else. A ClinicalTrials.gov v2 API query on query.term=mazdutide returned 38 registered studies on 18 August 2026; widening the term to mazdutide OR IBI362 OR LY3305677 returned 40, two of which register other molecules from the same two sponsors and were excluded by hand. Sponsorship splits cleanly along the licence. Innovent runs the Chinese registration programme under the IBI362 code, and Lilly runs the first-in-human work in the United Kingdom, a small phase 1 in Germany, a high-dose phase 1 and a phase 2 in the United States, and a completed phase 2 proof-of-concept study in alcohol use disorder. Anything described as a mazdutide result should be checked against which of those two programmes produced it, because they enrolled different populations.

The glucagon half of the molecule

Oxyntomodulin is a proglucagon fragment that engages the GLP-1 receptor and the glucagon receptor together. Ji and colleagues frame the design of mazdutide against that biology in the introduction to their 2022 phase 1b report, citing the observation that glucagon receptor agonism raises energy expenditure and the observation that several earlier GLP-1 and glucagon dual agonists had not shown weight-loss superiority over GLP-1 receptor agonism alone. Adding glucagon receptor activity is the design premise of the molecule. It is not a demonstrated mechanism of any measured clinical outcome, and the published trials did not test it as one.

No peer-reviewed characterisation of mazdutide's receptor pharmacology appears in PubMed. The only retrievable numeric affinities sit outside the journal literature. The IUPHAR/BPS Guide to Pharmacology records, under ligand 13924, a human glucagon receptor Ki of 1.46x10^-8 M (pKi 7.8) and a human GLP-1 receptor Ki of 2.3x10^-8 M (pKi 7.6), both sourced to an Eli Lilly patent, US9938335B2, and both carrying an empty assay description. The in-vitro binding data and the mouse energy-expenditure data that the phase 1b report cites come from a 2021 American Diabetes Association conference abstract, number 682-P by Chen, Mezo, Coskun and colleagues, published in a Diabetes supplement, which likewise has no methods section. A PubMed search combining mazdutide with energy expenditure returns five records, of which four are narrative reviews and one is a multi-omics study in male db/db mice.

The consequence is a specific gap. The only retrievable numeric receptor-affinity values sit in a Lilly patent and a conference abstract, not in a peer-reviewed characterisation, and neither source publishes the assay that produced them. Statements about how mazdutide's activity is balanced between the two receptors, and statements about the size of any energy-expenditure effect, therefore cannot be checked against a described method. The preclinical file that does exist in PubMed is small and recent: a high-fat-diet mouse model of steatotic liver disease, the db/db cognition study, an MRI study of liver fat and iron in a diet-induced rodent model, and a 2026 Science Advances paper on tubular glucagon receptor signalling in which mazdutide appears in the framing rather than as the experimental agent.

The Chinese phase 3 programme

Four phase 3 trials have been published, all in Chinese adults, all sponsored by Innovent. GLORY-1 (NCT05607680) randomised 610 participants with a body-mass index of at least 28, or 24 to under 28 with a weight-related condition, to mazdutide 4 mg, mazdutide 6 mg or placebo for 48 weeks. Ji and colleagues reported in the New England Journal of Medicine in 2025 that mean percentage change in body weight at week 48 was −11.00% at 4 mg, −14.01% at 6 mg and +0.30% on placebo, with 35.7%, 49.5% and 2.0% of participants respectively reaching a reduction of at least 15%.

GLORY-2 (NCT06164873) tested a higher dose in a heavier population. Gao and colleagues reported in JAMA in 2026 that among 461 Chinese adults with a body-mass index of at least 30 who received treatment, mean percentage change in body weight at week 60 was −16.65% with 9 mg against −1.50% with placebo. The same trial recorded vomiting in 53.1% of the mazdutide group against 1.3% on placebo, nausea in 46.9% against 3.2%, and diarrhoea in 39.4% against 6.5%. Discontinuation attributable to adverse events was 2.9% against 0%.

Two diabetes trials appeared back to back in Nature in 2026. DREAMS-1 (NCT05628311) randomised 320 adults with type 2 diabetes controlled inadequately by diet and exercise alone; Zhu and colleagues reported HbA1c change at week 24 of −1.57% at 4 mg and −2.15% at 6 mg against −0.14% on placebo. DREAMS-2 (NCT05606913) randomised 731 participants already on background oral agents against dulaglutide 1.5 mg; Guo and colleagues reported least-squares mean treatment differences in HbA1c at week 28 of −0.24% and −0.30%, and in body weight of −3.78% and −5.76%.

Where the non-Chinese data lives

The largest non-Chinese trial to report is NCT06124807, a Lilly phase 2 conducted under a master protocol at United States sites, enrolling 179 adults with obesity or overweight, mean baseline weight 107.6 kg, 125 of them recorded as White and 34 as Black or African American. Results were posted to ClinicalTrials.gov in April 2026. No journal publication of the trial exists. Least-squares mean percentage change in body weight at week 32 was −7.33% in the 3/6 mg arm, −15.6% at 10 mg, −18.1% at 16 mg and −0.85% on placebo; at week 48 the corresponding figures were −10.54%, −19.2%, −22.3% and −0.047%.

The same posting carries findings that no mazdutide publication contains. Among participants with baseline liver fat of at least 5% measured by MRI proton density fat fraction, least-squares mean relative change at week 48 was −67.4% in the 16 mg arm against −11.7% on placebo, in 31 and 36 participants respectively. Treatment-emergent anti-drug antibodies were detected in 11 of 51 participants at 16 mg, 10 of 47 at 10 mg, 3 of 32 in the 3/6 mg arm and 5 of 47 on placebo. No deaths occurred; serious adverse events were reported in one to two participants per arm.

Earlier Lilly work is less accessible still. The first-in-human single-ascending-dose study (NCT02972645, 66 participants) and the multiple-ascending-dose study (NCT03325387, 56 participants) were both conducted in the United Kingdom and completed in 2017 and 2018; a 24-participant phase 1 in adults with type 2 diabetes ran in Germany (NCT03928379). None has posted results and none appears in PubMed under any of the three compound codes. The high-dose phase 1 in 32 United States adults did publish: Bhattachar and colleagues reported in Diabetes, Obesity and Metabolism in 2025, from NCT05623839, that mean percentage change in body weight at week 20 was −20.0% and −21.0% in the two 16 mg escalation cohorts against −0.1% on placebo.

Heart rate, and why it is measured

Both published phase 1 reports measured heart rate as a safety endpoint. In the 2022 multiple-ascending-dose phase 1b in Chinese adults with overweight or obesity, with eight participants receiving mazdutide and four receiving placebo in each cohort, mean change from baseline in heart rate reached 17.4 beats per minute in the 9 mg cohort and 11.6 in the 10 mg cohort, against 4.3 for pooled placebo (PMID 36247927). Ji and colleagues recorded ventricular extrasystoles in one mazdutide participant and one placebo participant, both mild, alongside a fall in systolic blood pressure of up to 12 mm Hg.

The high-dose phase 1 quantified it differently and more starkly. Bhattachar and colleagues reported mean pulse rate rising from 70 to 85 beats per minute in one cohort and 76 to 84 in the other by day 134, while the placebo group fell from 72 to 64. An increase of at least 10 beats per minute on two consecutive visits occurred in 91.7% of the 24 participants receiving mazdutide and 37.5% of the eight receiving placebo. No adverse electrocardiogram findings were recorded, and five participants had transient laboratory changes including raised transaminases.

No cardiovascular outcome trial for mazdutide is registered. Among the registered studies, one phase 4 trial of coronary plaque in 116 participants is listed as not yet recruiting with primary completion in October 2027, and a phase 2 study in heart failure with preserved or mildly reduced ejection fraction is recruiting. Whether the recorded heart rate change carries any clinical consequence over years of exposure is therefore an open question that the existing register is not designed to answer.

What the register holds that the literature does not

None of the four published Chinese phase 3 trials has posted results to ClinicalTrials.gov; they went to journals instead. That leaves the registry entries as protocol records only, so secondary endpoints not carried into the published abstracts cannot be checked against a posted results table. The reverse holds for the American phase 2, which posted results and has no paper. A reader tracing a mazdutide figure has to know which of the two conventions applies to the trial it came from.

Several completed trials have reported nothing anywhere. DREAMS-3 (NCT06184568), the first head-to-head comparison against semaglutide 1 mg in 349 Chinese adults with type 2 diabetes and obesity, completed in September 2025; Luo and colleagues published its design and baseline characteristics in Contemporary Clinical Trials in 2026, but no outcome data have appeared. The phase 2 proof-of-concept study in alcohol use disorder (NCT06817356), 308 participants at United States sites, reached primary completion in March 2026 with no results posted and no publication.

Citation hygiene in the secondary literature around this compound is uneven, including in peer-reviewed venues. The 2026 phase 2 report in Med on mazdutide 9 mg gives its registration as NCT01904913, which is not a valid ClinicalTrials.gov record; the trial it describes is NCT04904913. Kamrul-Hasan and colleagues, pooling nine randomised trials and 2292 participants in a 2026 meta-analysis, graded the certainty of the obesity weight-loss estimates as very low. Nong and colleagues, in a 262-trial network meta-analysis in the BMJ, placed mazdutide among emerging agents at 13.1% to 14.6% weight loss with very low to low certainty.

What is not known

Longest published exposure is 60 weeks, in GLORY-2. No durability, discontinuation-rebound or long-term safety data exist, and no cardiovascular outcome trial is registered, so whether the recorded heart rate increase matters over years is untested. Body composition has never been characterised, meaning the proportion of the recorded weight change attributable to fat mass against lean mass is unknown for this compound. Every published efficacy trial but one enrolled Chinese adults: the single published non-Chinese trial is a 32-participant phase 1, the larger United States phase 2 exists only as a ClinicalTrials.gov results posting, and the two United Kingdom first-in-human studies from 2017 and 2018 and a 24-participant phase 1 in Germany have neither posted results nor been published, so the foundational human pharmacokinetic and tolerability dataset is not publicly retrievable. The only retrievable receptor-affinity values sit in a patent and a conference abstract, neither of which publishes its assay. The registries do not agree on the molecular formula or the molecular weight. Nine registered trials in populations of specific interest — metabolic dysfunction-associated steatohepatitis, obstructive sleep apnoea, adolescents, heart failure with preserved ejection fraction, renal impairment, polycystic ovary syndrome, alcohol use disorder, hypertension, early dementia — are ongoing or completed without reporting. Published trials excluded people with type 1 diabetes and offer no data on pregnancy, lactation, or adults over 75. Mazdutide holds a marketing authorisation in China only; it is not approved by the FDA and no European authorisation was located. Material sold outside a registered trial or the Chinese supply chain has not been subject to the identity, purity or sterility controls that apply to either.

Questions