Survodutide's headline was 16.6%. Its primary analysis said 13.0%.

Why this note exists

Survodutide is at the stage where a compound stops being obscure and starts being quoted. Two Phase 3 reports landed in peer-reviewed journals on the same day in June 2026, no FDA approval was identified, and sixteen of the 112 vendors in pepmg's index already carry it — 17 listings, fifty-first of the 83 compounds in the index by vendor coverage, in the index generated August 15, 2026.[14]

The interesting thing about this file is not that the evidence is weak. It is that the evidence is unusually good, and the number most people will repeat still isn't the one the trial was designed around. That gap has a name, it is disclosed in plain sight in both publications, and it is worth understanding once, because every obesity drug now reports both.

A note on what kind of evidence this is: every efficacy and safety figure below is human trial data, and each one carries its design, its analysis population and its participant count. This note reports no animal or in vitro data at all. Doses appear only as the trials administered them, attributed to the publication that reported them, and pepmg does not convert a trial dose into a protocol for a research vial.

The trial

What SYNCHRONIZE-1 actually did

SYNCHRONIZE-1 was a Phase 3, randomized, double-blind, placebo-controlled trial in adults with a body-mass index of 30 or higher, or 27 or higher with at least one obesity-related complication, excluding diabetes.[1] Participants were assigned 1:1:1 to once-weekly subcutaneous survodutide at a dose adjusted up to 3.6 mg, up to 6.0 mg, or placebo, all with lifestyle-modification counseling.[1] The registry records the trial as running from November 25, 2023 to February 20, 2026, quadruple-masked, with 726 participants enrolled; the published report analyses 725 randomized participants.[2][1]

There were two primary endpoints, both measured from baseline to week 76: percent change in body weight, and achieving a weight reduction of at least 5%.[1] At baseline the mean age was 47.1, mean BMI was 37.9, mean body weight was 108.8 kg, and 294 participants (40.6%) were men.[1]

Mean percent change in body weight from baseline as reported in the primary publication. At least 5% weight reduction was reached by 72.6%, 71.9% and 46.3% of participants respectively, P<0.001 for all comparisons with placebo.[1]

The gap

An estimand is a question, and this trial published two answers

An estimand is the precise statement of what a trial is measuring — including what it does with everything that happens after randomization: people who stop the drug, people who start a different one, people who never reach the target dose. Choose differently and the same dataset yields a different number.

SYNCHRONIZE-1 prespecified the treatment-regimen estimand as its primary efficacy analysis. The published report defines it as one "which incorporates the effects of any early discontinuation of survodutide or placebo, the use of protocol-prohibited obesity medications, and a prolonged dose-escalation period."[1] In plain terms: what happened to the people assigned to this drug, including the ones for whom it did not go to plan.

The sponsor's topline announcement on April 28, 2026 reported the trial under the efficacy estimand — the effect if participants had stayed on assigned treatment throughout. That release states that participants "experienced sustained weight loss of up to an average of 16.6% after 76 weeks using the efficacy estimand, a statistically significant decrease versus 3.2% in the placebo arm (p<0.0001)," and that "up to 85.1% of adults treated with survodutide" achieved at least 5% weight reduction "using the efficacy estimand, versus 38.8% in the placebo arm."[3] The same release says the trial met its co-primary endpoints under both estimands, which is the fact that keeps this from being a scandal.[3]

Read the placebo column, not the drug column. Under the efficacy estimand the placebo group loses 3.2% and 38.8% of it reaches the 5% threshold; under the primary analysis the same placebo group loses 5.4% and 46.3% reaches it.[1][3] The two estimands move the comparator as much as they move the drug, which is exactly why a placebo-adjusted difference calculated across two different analyses would be meaningless. The 13.4-point and 7.6-point figures above are each internally consistent; mixing a drug number from one with a placebo number from the other is not.

The reason the two diverge as far as they do is legible in the tolerability data rather than in the statistics. In SYNCHRONIZE-1 the most common adverse events were gastrointestinal, typically mild to moderate, and they occurred in 80.9% of the 3.6 mg group, 89.7% of the 6.0 mg group and 47.9% of the placebo group; no deaths were reported.[1] The sponsor's release adds that discontinuations happened more frequently during the dose-escalation phase.[3] The published report does not attribute the gap between estimands to any single one of the three factors its definition names, and neither does this note.

The second trial

The liver trial printed both numbers side by side, which is the honest format

SYNCHRONIZE-MASLD, published in Nature Medicine the same week, is the cleaner illustration because its abstract reports each result under both estimands without being asked. It enrolled 216 adults (131 female, 85 male) with obesity and at-risk metabolic dysfunction-associated steatotic liver disease, randomized 2:1 to once-weekly subcutaneous survodutide 6.0 mg (n = 146) or placebo (n = 70) for 48 weeks.[4] Both co-primary endpoints were met.[4]

Liver fat content assessed by MRI proton density fat fraction, baseline to week 48. All four figures are from the same 216 participants.[4]

The authors state their own limitations in the abstract: the trial ran 48 weeks, and participants were recruited only in the United States and Spain.[4] An Author Correction to this paper was published on August 25, 2026; readers should read it alongside the original, and this note does not characterize what it changed.[5]

The record behind it

The Phase 2 trials, including the completion rate

The Phase 3 program rests on two Phase 2 trials that are both peer-reviewed and both worth reading for their attrition as much as their effect sizes.

The dose-finding obesity trial, published in The Lancet Diabetes & Endocrinology in 2024, enrolled 387 participants and treated 386 across four survodutide doses and placebo.[6] Analysed by planned treatment, mean weight change at week 46 was −6.2% at 0.6 mg, −12.5% at 2.4 mg, −13.2% at 3.6 mg and −14.9% at 4.8 mg, against −2.8% on placebo.[6] Adverse events occurred in 281 of 309 survodutide recipients (91%) and 58 of 77 on placebo (75%), primarily gastrointestinal in 75% and 42% respectively.[6] 233 of 386 treated participants — 60.4% — completed the 46-week treatment period.[6] That is the context in which a choice of estimand stops being a technicality.

The MASH trial, published in The New England Journal of Medicine in 2024, randomized 293 adults with biopsy-confirmed MASH and fibrosis stage F1 through F3 to survodutide 2.4, 4.8 or 6.0 mg or placebo for 48 weeks.[7] Histologic improvement in MASH without worsening of fibrosis occurred in 47%, 62% and 43% of the survodutide groups against 14% on placebo; improvement in fibrosis by at least one stage occurred in 34%, 36% and 34% against 22%.[7] Nausea was reported by 66% against 23%, diarrhea 49% against 23% and vomiting 41% against 4%; serious adverse events were 8% against 7%.[7] The authors' own conclusion was that the result warranted further investigation in Phase 3, and it is worth noting that the highest dose was not the best-performing one on the primary endpoint.[7]

Regulatory status

Two FDA designations, no approval, and a 5,531-person outcomes trial nobody has seen

Survodutide is not approved in any country. The sponsor's own announcement carries the sentence in full: it "is an investigational agent and has not been approved for use; its efficacy and safety has not been established."[3] FDA's Drugs@FDA dataset, queried on August 31, 2026, returns no record for survodutide.[13]

What it does have is two FDA designations: Fast Track in May 2021 and Breakthrough Therapy in September 2024.[3] Both are decisions about how an application will be handled, not findings that a drug works, and both are granted long before a Phase 3 result exists. They belong in a note like this only so that nobody mistakes them for the thing they are adjacent to.

SYNCHRONIZE-CVOT is by a wide margin the biggest study in the program, and its registry record shows the primary completion date already passed.[8] Nothing has been published from it. A cardiovascular outcomes trial is where a drug in this class either acquires or fails to acquire its most consequential safety and benefit claims, and a trial with no reported results is unreported — not negative, and not positive.

Two Phase 3 liver-outcome trials run much longer: LIVERAGE, in MASH with moderate or advanced fibrosis, with an estimated 1,800 participants and a primary completion date of December 2031, and LIVERAGE-Cirrhosis, with an estimated 1,590 and a primary completion date of June 2029.[9][10] A registry query on August 31, 2026 returned 26 registered survodutide studies in total.[12]

The market

There is no approved product for a research vial to resemble

The usual version of this paragraph says a trial result does not transfer to a vial bought online. Survodutide has a sharper version of the same point available, and it comes from the sponsor's own registry entries.

As of August 31, 2026, a Phase 1 study was recruiting 80 participants with overweight or obesity for the sole purpose of comparing how two different formulations of survodutide are taken up by the body, with a primary completion date of January 2027; a second registered formulation-comparison study had not yet begun recruiting.[11][12] The company that makes the molecule is still characterizing its own formulations. There is no approved reference product for survodutide anywhere in the world, which means there is no marketed strength, formulation, excipient set or stability profile for anything else to be compared against. Sequence, salt form, purity, sterility and stability are all separate, open questions for any vial, and pepmg makes no claim about the contents of any vendor's.

The size data are their own small illustration of the distance. In the index generated August 15, 2026, the 17 survodutide listings run to 10 mg on nine of them, 6 mg on two, 12 mg on two, one 100 mg bulk quantity, and three that state no size at all.[14] The maintenance dose in SYNCHRONIZE-1 was up to 6.0 mg once weekly, subcutaneously, in adults meeting specific entry criteria, with lifestyle counseling, monitoring and a dose-escalation schedule — reported here as the trial reported it, and not as a protocol.[1]

Three things this note is not saying

First, it is not saying the sponsor misreported anything. Both estimands were prespecified, the trial met its endpoints under both, the announcement labeled its figure as the efficacy estimand in the sentence that contained it, and the full peer-reviewed report followed within six weeks.[1][3] Against the general standard of this field, that is close to exemplary.

Second, it is not saying survodutide does not work. A 725-person randomized double-blind Phase 3 trial found a statistically significant, clinically substantial weight reduction against placebo on its primary analysis, and a second Phase 3 trial found large reductions in liver fat.[1][4] Those are real results in real people.

Third, it is not saying 16.6% is a false number. It is a published figure under a named analysis, and it answers a legitimate question about what happens to people who stay on the drug.[3]

What it is saying is narrower, and it generalizes past this compound: the headline and the primary analysis differ by 3.6 percentage points on the drug arm and by 5.8 on the placebo-adjusted difference, and that gap is usually not spin — it is the trial telling you how the result changes once everyone who could not stay on the drug is counted. Read which estimand a number came from before you compare it to anything.

Questions people are asking