A single molecule combining GLP-1 and amylin reports phase 2 results in both forms

The incretin field has spent three years combining hormones. Tirzepatide pairs GLP-1 with GIP. The next contested pairing is GLP-1 with amylin, and zenagamtide is an attempt to deliver both from one molecule rather than a co-formulation of two.

Two phase 2 trials published simultaneously in The Lancet report it in both delivery forms — a once-weekly injection and a once-daily tablet — in people with type 2 diabetes. Both were funded by Novo Nordisk [s1][s2].

The injection

Between 7 August and 27 December 2024, 262 (29%) of 915 screened individuals were randomly assigned to subcutaneous zenagamtide (n = 225) or placebo (n = 37), then further randomised across six doses: 38 participants to 0.4 mg, 36 to 1.5 mg, 37 to 5 mg, 38 to 10 mg, 38 to 20 mg and 38 to 40 mg, against 37 on placebo [s1]. In total 261 participants were exposed to treatment [s1].

Mean HbA1c across all groups at baseline was 7.8% (SD 0.8) [s1]. At week 36, estimated change ranged from −0.9% with zenagamtide 0.4 mg, an estimated treatment difference against placebo of −0.77% (95% CI, −1.26 to −0.28; p = 0.0021), to −1.7% with zenagamtide 40 mg, a difference of −1.56% (−2.05 to −1.07; p < 0.0001) [s1].

Every dose separated from placebo, including the lowest, which is the finding a dose-ranging study exists to produce.

Most adverse events were gastrointestinal and mild to moderate in severity [s1]. Serious adverse events were reported in 21 (8%) of 261 participants: four with 0.4 mg, three with 1.5 mg, two with 5 mg, five with 10 mg, three with 20 mg, one with 40 mg, and three with placebo [s1]. No deaths occurred [s1].

Those serious-event counts do not rise with dose — the highest dose recorded the fewest, at one [s1]. With 36 to 38 participants per arm, this is a distribution consistent with chance rather than a dose-response signal in either direction, and it should not be read as evidence that 40 mg is safer than 10 mg.

The tablet

The same screening pool produced the oral trial: 186 (20%) of 915 screened participants were randomly assigned to oral zenagamtide — 54 to 6 mg, 51 to 25 mg and 51 to 50 mg — or placebo (30 participants) [s2].

From a baseline mean range of 7.9–8.1%, estimated mean HbA1c change at week 36 was −0.9% with 6 mg (estimated treatment difference against placebo −0.5% [95% CI, −1.04 to −0.04]; p = 0.033), −1.3% with 25 mg (−0.99% [−1.49 to −0.49]; p = 0.0001), and −1.4% with 50 mg (−1.09% [−1.59 to −0.59]; p < 0.0001) [s2].

The dose-response is cleaner here, and the lowest oral dose is the weakest result in either trial — a treatment difference of −0.5% whose confidence interval reaches −0.04, and a p-value of 0.033 [s2].

The tolerability data is where the oral trial earns its separate publication. The most common adverse events were gastrointestinal, occurring in 14 (26%) of 54 participants on 6 mg, 21 (41%) of 51 on 25 mg, 24 (47%) of 51 on 50 mg, and seven (23%) of 30 on placebo [s2].

That is a clear dose-dependent gradient, from 26% to 47%, against a placebo rate of 23% [s2]. At the lowest dose the gastrointestinal rate is barely above placebo — and so is the efficacy. At the highest, roughly one in two participants had gastrointestinal adverse events for an additional 0.59 percentage points of HbA1c reduction over the 6 mg dose [s2].

Serious adverse events were reported in seven (4%) of 186 participants receiving zenagamtide — two on 6 mg, two on 25 mg, three on 50 mg — and none on placebo [s2]. No deaths occurred [s2].

What the two trials do and do not establish

They establish that a single molecule agonising both receptors lowers HbA1c against placebo at 36 weeks, by both routes, with a safety and tolerability profile the authors describe as consistent with other GLP-1-based and amylin-based therapies [s1][s2].

Several things they do not establish are worth stating.

This is phase 2. Both trials are dose-ranging studies in a few hundred people, designed to select a dose for phase 3 rather than to demonstrate that the drug should be used.

There is no active comparator. Both trials compare against placebo [s1][s2]. Nothing here says how zenagamtide performs against semaglutide or tirzepatide, which is the question that decides whether it matters. An HbA1c reduction of 1.7% against placebo is not informative about a field where effective drugs already exist.

Neither is a weight trial. These are HbA1c trials in type 2 diabetes, and the abstracts report glycaemic endpoints [s1][s2]. Amylin is of interest partly for satiety and body composition, and that case is not made by this data.

Neither is an outcomes trial. HbA1c is a surrogate. Whether the drug prevents cardiovascular or kidney events is unaddressed.

The screening ratios are notable. 262 of 915 screened in one trial and 186 of 915 in the other [s1][s2]. Heavy screening produces a selected population, which usually flatters both efficacy and tolerability relative to ordinary practice.

What to watch

Whether phase 3 selects a dose from the middle of the range, where the oral trial's efficacy is near-maximal and gastrointestinal events sit at 41% rather than 47% [s2]. Whether any trial compares zenagamtide against an existing incretin rather than placebo. And whether the once-daily oral form holds up, since an effective oral incretin would change access more than another injection would.

This article describes published trial results. Zenagamtide is investigational and not approved. It is not medical advice.

Sources

• [s1] Efficacy and safety of once-weekly subcutaneous zenagamtide, a novel unimolecular GLP-1 and amylin receptor agonist, in type 2 diabetes, The Lancet, 2026;408(10555):621–635.
• [s2] Efficacy and safety of once-daily oral zenagamtide, a novel unimolecular GLP-1 and amylin receptor agonist, in type 2 diabetes, The Lancet, 2026;408(10555):607–620.