Amycretin is one peptide that switches on more than one receptor. It activates the GLP-1 receptor, the target the approved drugs already hit, and the amylin receptor, which they do not — and in cell-based assays it also activates the calcitonin receptor, which shares a family with amylin. Amylin is released alongside insulin and signals fullness through a different route from incretin action, so the premise is that a single molecule can pull two satiety levers at once. The same premise drives the two-molecule combination covered in the CagriSema article; this one puts both activities in one chain.
The compound has been renamed. Both 2026 phase 2 papers open with the sentence “Zenagamtide (formerly amycretin),” and the phase 3 program is registered under the new name. No product containing it appears in Drugs@FDA under either name, which means no approved indication, no approved dose, no contraindication list and no lawful United States supply.
What the preclinical work established, and what it did not
The molecule activated human, mouse and rat GLP-1, amylin and calcitonin receptors in cell-based systems. In diet-induced obese rats, 21 days of treatment cut total energy intake by 47% (95% CI for the mean, 1,044–1,390 kcal against 2,132–2,493 on vehicle) and lowered body weight by 18% (P < 0.0001) while energy expenditure was maintained rather than falling. Insulin sensitivity improved on a hyperinsulinemic euglycemic clamp, and liver histology improved mainly through less steatosis.[1]
Rat energy balance is where this class always looks best. What the preclinical package cannot say is how much of the effect survives in people who can stop taking a drug that makes them feel sick, which is the question the human trials were built to answer and mostly did not.
The first human study measured side effects, not weight
The first-in-human trial enrolled 144 adults with overweight or obesity at a single clinical research unit in San Antonio, Texas, in three parts: 48 on single oral doses from 1 mg to 25 mg, 36 on 10 days of daily dosing, and 60 on a 12-week titration reaching as high as two 50 mg tablets a day. The primary endpoint was the number of treatment-emergent adverse events. There were 364 of them in 89 (62%) of 144 participants, all mild or moderate, rising with dose; 180 (49%) of the 364 were gastrointestinal, and those landed on 72 (81%) of the 89 people who reported any event at all. No deaths occurred, and plasma concentrations tracked dose proportionally.[2]
Weight change in that trial was an exploratory endpoint, measured in the 12-week part only. A 2026 review of the program puts the oral result at up to 13.1% at 12 weeks.[3] That is a real number from a real trial, and it is the kind of number a phase 1 study is not designed to produce reliably: no efficacy hypothesis, no powering, one site.
The figure everyone quotes, and where it came from
The injectable study is the source of the headline. It randomized 125 adults aged 18 to 55 with a body-mass index of 27.0 to 39.9 — 101 to amycretin, 24 to placebo — at the same single research center, in five parts with different maintenance doses and different durations. Mean weight change was −24.3% against −1.1% at 60 mg over 36 weeks, −22.0% against +1.9% at 20 mg over 36 weeks, −16.2% against +2.3% at 5 mg over 28 weeks, and −9.7% against +2.0% at 1.25 mg over 20 weeks (P < 0.0001 for parts A–D; P = 0.0003 for part E).[4]
Three things about that paragraph deserve to travel with the number. First, the primary endpoint of this trial was also the count of treatment-emergent adverse events; weight was secondary. Second, the arms are not a clean dose-response, because the top two ran 36 weeks and the bottom one ran 20 — part of the gap between −9.7% and −24.3% is sixteen extra weeks, not dose. Third, the paper’s own summary of findings states that a large number of participants withdrew, with a high proportion of discontinuations for reasons unrelated to adverse events.[4] A trial that loses a large share of its participants reports what happened to the people who stayed, and the published abstract does not put a figure on how many left.
Read the doses instead of the percentages and the shape changes. Going from 5 mg to 60 mg — twelve times the dose — added roughly eight percentage points. Going from 1.25 mg to 5 mg, four times the dose, added about six and a half. The steep part of this curve is at the bottom, which is the same pattern the tirzepatide and semaglutide ladders show in the semaglutide results article.
Where the program was powered, the effect flattened
Two phase 2 dose-finding trials in type 2 diabetes reported in 2026 across 83 sites in 11 countries, and these were built around an efficacy endpoint: change in glycated hemoglobin at week 36. The subcutaneous trial randomized 262 of 915 screened adults 6:1 to zenagamtide or placebo and then across six doses from 0.4 mg to 40 mg. From a mean baseline of 7.8%, the estimated treatment difference against placebo ran from −0.77 percentage points (95% CI, −1.26 to −0.28; P = 0.0021) at 0.4 mg to −1.56 (95% CI, −2.05 to −1.07; P < 0.0001) at 40 mg. Serious adverse events affected 21 (8%) of 261 treated participants and no deaths occurred.[5]
A hundredfold increase in dose bought roughly double the effect, and the confidence intervals at the two ends overlap across most of their width. The oral trial in the same program tells the same story in a narrower range: 186 adults on 6, 25 or 50 mg reached treatment differences of −0.5 (95% CI, −1.04 to −0.04), −0.99 (−1.49 to −0.49) and −1.09 (−1.59 to −0.59) percentage points, so an eightfold dose step separated −0.5 from −1.09. Gastrointestinal adverse events rose with dose, at 26%, 41% and 47% against 23% on placebo, and serious adverse events occurred in 7 (4%) of 186 on drug and none on placebo.[6]
Glycemic control and weight are different endpoints and a flat HbA1c curve does not prove a flat weight curve. But these are the only trials in the program that were designed to estimate an effect rather than count side effects, and in both of them the top dose is worth much less than its milligram figure suggests. The distinction between a drug-and-glucose result and a drug-and-weight result is set out in the diabetes-versus-weight article.
Cross-trial rankings, with the intervals attached
A 2026 systematic review of 38 randomized trials and 25,816 adults without diabetes reported placebo-subtracted weight loss of −23.9% (95% CI, −29.3 to −18.5) for amycretin, against −22.1% (−24.9 to −19.3) for retatrutide, −19.0% (−21.6 to −16.4) for tirzepatide, −14.8% (−16.2 to −13.4) for subcutaneous semaglutide and −5.8% (−8.0 to −3.6) for liraglutide. Gastrointestinal adverse events ran 76.0% against 40.1% on placebo across the class, and discontinuation for adverse events 10.7% against 3.4%.[7]
The amycretin interval is ten and a half percentage points wide — more than twice the width of tirzepatide’s — because it rests on about a hundred people rather than tens of thousands. Its lower bound, −18.5%, sits inside tirzepatide’s interval. A 2026 network meta-analysis of six amylin-based trials and 4,642 participants ranks high-dose subcutaneous amycretin first at −23.95% against placebo, ahead of eloralintide at −18.01% and CagriSema at −17.18%, and its authors label the whole network sparse and low-certainty and the findings preliminary.[8] A first-place ranking drawn from six trials is a hypothesis about which molecule to test next, not a result.
Where it actually sits in development
Five zenagamtide studies are on the public trial register. Four are phase 3: one diabetes trial enrolling 1,778 participants, which began recruiting in August 2026 with a primary completion date in July 2028, and three obesity trials of 950, 650 and 400 participants, none of which had opened for recruitment, with primary completion dates running from June 2028 to January 2029. The fifth is a phase 1 study of how meal timing and water volume affect the tablet.
So the entire published human efficacy record for this molecule is two early-phase studies at one site in Texas totaling 269 people, plus two phase 2 diabetes trials that randomized 448 adults, 381 of them to drug. Nothing in phase 3 has produced a result and, on the sponsor’s own posted dates, nothing can before 2028. Anyone reading a 24% figure as a preview of a product is reading three years ahead of the evidence, and the reasons that gap matters are the ones set out in the glucagon dual agonist article, where a phase 3 top dose came in below its own phase 2.
What is not known
No cardiovascular outcome result exists, and none is in the published record. No trial of this molecule has run past 36 weeks, so nothing is established about durability, about what happens on stopping, or about the plateau discussed in the plateau article. No head-to-head against semaglutide or tirzepatide in obesity has reported. The obesity data come from one center in one city, and the diabetes data exclude anyone with a body-mass index of 50 or above. Bone density, muscle mass and the calcitonin receptor’s own effects have no published human read-out for this compound at all.
The withdrawal question is the one most worth watching. The injectable study reported both the largest weight figure in this literature and a large number of participants leaving, and only a phase 3 trial with a treatment-regimen estimand — which counts what happened to people including those who stopped — will show what the molecule does to a whole randomized group rather than to its survivors.
What a product sold under this name would be
Because neither amycretin nor zenagamtide is a component of any FDA-approved drug, no pharmacy is preparing it against an approved reference. Compounded drugs in general are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before dispensing; a substance with no approved reference product anywhere in the country sits a step further out, because there is nothing to compare a vial against and no label to check a strength against. What the unapproved channel actually delivers has been sampled, and the assay evidence is in the gray-market article.
The practical distinction between an approval, an application and a trial result is the subject of the approval-status article, and how the figures on this site are established before publication is described in the methodology. A compound three years from a phase 3 read-out has none of those things.