The human trials

Amycretin's evidence base is young and shallow. From the hormone biology worked out over decades, the drug reached its first human trials, and by mid-2025 two early-phase papers and a preclinical paper had appeared. The phase 3 decision was announced days before the Lancet papers, and the sponsor based it on regulatory feedback rather than on a new result. The confirmatory trials began enrolling in 2026, and the phase 2 diabetes results were published that July.

Notice how compressed this is: amycretin went from its first published human results to an enrolling phase 3 programme in about a year, on the strength of early signals. First dosing was earlier still, in May 2022. That is a normal, promising trajectory, but it means the confirmatory evidence, the kind that supports approval, has not been generated yet. Everything below is dose-finding.

The 2026 papers are the first amycretin results in people with type 2 diabetes, and the first from a multi-country trial rather than a single center. They ran for 36 weeks across 83 sites in 11 countries. Their main measure was HbA1c, the blood test that averages blood sugar over about three months. In the oral cohort, 186 participants randomly assigned, estimated HbA1c change at week 36 was about -1.4% on 50 mg daily. In the subcutaneous cohort, 262 randomly assigned, it was about -1.7% on 40 mg weekly. Do not subtract the placebo figure to get the difference: both abstracts state their own treatment difference, -1.09% (95% CI -1.59 to -0.59) for oral 50 mg and -1.56% (-2.05 to -1.07) for subcutaneous 40 mg.

Neither 2026 abstract reports a weight figure. That number exists only in the sponsor's ADA release from June 2026, which gives a mean body weight reduction of up to 14.6% on the 40 mg dose against 2.1% with placebo, from a mean baseline of 99.2 kg in 262 adults. Now watch what happens if you do the obvious arithmetic. Subtracting the placebo figure from the drug figure gives a gap of about twelve and a half points, and that gap is not a number any source states. A table note in the same release gives the dose-response model instead: an estimated mean change of up to -14.5%, with a stated treatment difference against placebo of -11.81 points (95% CI -15.37 to -8.25). The sponsor's own difference is smaller than the subtraction. Where a source states its difference, quote the difference.

One more line from that release keeps the weight figure in proportion. The sponsor notes that the higher dose groups were only exposed to the maintenance dose for a short period, 20 mg for 8 weeks and 40 mg for 4 weeks. It offers that as a reason the glycaemic result is impressive. Read as a fact about exposure it also means the top-dose weight figure describes a few weeks at the target dose on top of many months of climbing toward it.