Survodutide's phase 3 is published. The 16.6% headline is not the paper's primary result.

What actually happened on June 7

Survodutide, also identified as BI 456906, is a once-weekly injected peptide that activates both the glucagon receptor and the GLP-1 receptor.[7] Until this summer its human record stopped at phase 2. On June 7, 2026 two phase 3 trials were published on the same day: SYNCHRONIZE-1 in The New England Journal of Medicine and SYNCHRONIZE-MASLD in Nature Medicine.[1][5]

That is a genuine change in the evidence base, and it is the reason this note exists. It also arrived with a reporting problem attached. The topline announcement six weeks earlier had put a single number in circulation, and the peer-reviewed primary analysis reports a different one.

The sponsor's release states the trial met its co-primary endpoints under both estimands, and that is consistent with the published paper.[2][1] Nothing here is a contradiction. It is a choice about which of two pre-specified questions to lead with.

Why the gap exists

An estimand is a question, not a spin

Since 2019 the ICH E9(R1) addendum has required trials to pre-specify an estimand: an explicit statement of the treatment effect being estimated, including how the analysis handles events like stopping the drug or taking a different one.[3] Two estimands on one dataset can produce two honest, differently sized answers.

The SYNCHRONIZE-1 authors describe their primary one plainly. The treatment-regimen estimand "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 everyday terms it counts what happened to everyone who was randomized, including the people who came off the drug.

We are not going to define the efficacy estimand for you, because the material we read does not. The company release attributes 16.6% to it and does not spell out its handling rules, and the published abstract names only the primary analysis.[2][1] Broadly, an efficacy estimand of this type describes the effect under treatment as intended rather than as taken, which is why it is usually the larger figure. Treat the 3.6-point gap as the size of that difference in this trial, not as evidence that anyone was hiding anything.

One practical consequence: if you compare survodutide's 16.6% against another drug's published primary result, you are not comparing like with like. Our own note on cagrilintide ran into the same problem from the other direction, where the sponsor reported both "regardless of adherence" and "assuming adherence" figures for every arm.

The placebo arm is the part nobody quoted

Under the published primary analysis, the placebo group lost 5.4% of body weight over 76 weeks, and 46.3% of people on placebo reached the ≥5% threshold that formed one of the trial's two primary endpoints.[1] Under the efficacy estimand, the sponsor put placebo at 3.2% and the responder rate at 38.8%.[2]

Both survodutide doses beat placebo decisively, and the paper reports P<0.001 for every comparison.[1] The point is the size of the reference line. Every participant in this trial also received counseling for lifestyle modification.[1] Nearly half of the people who got only that, plus a placebo injection, lost at least 5% of their body weight. Any figure quoted without its comparator arm is a number without a scale.

Separate trials, not one pooled population. Registry status checked August 3, 2026.[4][5][10][12][7]

The higher dose bought 0.8 points, and cost 8.8 points of gastrointestinal events

SYNCHRONIZE-1 tested two doses. Going from 3.6 mg to 6.0 mg moved mean weight loss from −12.2% to −13.0% under the primary analysis, a difference of 0.8 percentage points, with overlapping confidence intervals.[1] On the trial's other primary endpoint the lower dose did marginally better: 72.6% of the 3.6 mg group reached ≥5% against 71.9% at 6.0 mg.[1]

The tolerability difference is larger than the efficacy difference. Gastrointestinal symptoms, described as typically mild to moderate, were reported in 80.9% at 3.6 mg, 89.7% at 6.0 mg and 47.9% on placebo.[1] The sponsor separately reported that 19% of survodutide participants discontinued because of gastrointestinal adverse events, against 2.9% on placebo.[6] No deaths were reported in the trial.[1]

This is not a new pattern for the molecule. The 2024 dose-finding phase 2 trial ran 386 treated participants across 43 centres in 12 countries for 46 weeks and only 60.4% completed the treatment period, at similar rates on drug and on placebo.[7] Adverse events occurred in 91% of survodutide recipients against 75% on placebo, primarily gastrointestinal in 75% against 42%.[7]

Note also which doses were carried forward. Phase 2 topped out at 4.8 mg and reported −14.9% there, its best result, against −13.2% at 3.6 mg and −2.8% on placebo.[7] Phase 3 tested 3.6 mg and 6.0 mg, skipping the dose that performed best in dose-finding. The published design paper states the doses and permits flexibility within them; it does not, in the material we read, explain that selection.[9]

The liver trial is the more interesting result

SYNCHRONIZE-MASLD randomized 216 adults with obesity and at-risk metabolic dysfunction-associated steatotic liver disease 2:1 to survodutide 6.0 mg (n=146) or placebo (n=70) for 48 weeks. Its co-primary endpoints were a ≥30% reduction in liver fat content measured by MRI-PDFF and percentage change in body weight; both were met.[5]

This paper does what the obesity release did not: it reports both estimands side by side. A ≥30% liver-fat reduction occurred in 84.2% versus 24.3% under the efficacy estimand, and 68.5% versus 28.6% under the treatment-regimen estimand (P<0.0001 for both). Weight change was −12.2% versus −1.0% and −8.7% versus −1.4% respectively.[5]

Read those four pairs together and the estimand gap stops being an abstraction. The same trial supports "84% of patients" or "69% of patients" depending on the question asked, and it supports 12.2% or 8.7% weight loss. The authors list the 48-week duration among the trial's limitations.[5]

The phase 2 predecessor is worth keeping in view because it used biopsies rather than imaging. In 293 adults with biopsy-confirmed MASH and fibrosis stage F1 through F3, histologic improvement in MASH without worsening of fibrosis occurred in 47%, 62% and 43% of the 2.4, 4.8 and 6.0 mg groups against 14% on placebo; improvement in fibrosis by at least one stage occurred in 34%, 36% and 34% against 22%.[8] The dose response was not monotonic — the middle dose looked best on the primary endpoint — and the fibrosis differences are narrower than the steatohepatitis ones.[8]

Alongside the obesity trial the sponsor reported an MRI substudy finding reductions of up to 34% in visceral fat and up to 63.1% in liver fat, plus a pre-specified body-composition analysis on lean mass.[6] Those figures are sponsor-reported and, as far as we can find, not yet in a peer-reviewed publication, so we are citing them as such and not treating them as established.

What is still unknown

Three things are missing, and each of them is the kind of gap that a headline percentage hides.

The BMJ analysis is still the best available context for the class it does cover. At one year against lifestyle modification alone it put Tirzepatide at −14.9% and cagrilintide-semaglutide at −14.8% on moderate-to-high certainty evidence, and it found that larger benefits generally came with greater harms and more discontinuation.[13] That last finding is the one to carry into any comparison involving a drug with a 19% gastrointestinal discontinuation rate.

Approved is not the same as investigational

Survodutide has collected regulatory attention without regulatory approval. The sponsor's June 2026 release lists FDA Fast Track designation in May 2021, FDA Breakthrough Therapy designation in September 2024, acceptance to the EMA's PRIME scheme in November 2023, and breakthrough designations from China's NMPA in June 2024 and Taiwan's FDA in September 2024.[6]

Every one of those is a process designation. They govern how quickly and closely a regulator engages with a development programme. None of them is a finding that the drug works or is safe, and none of them permits marketing. The same release states it directly: "Survodutide is an investigational agent and has not been approved for use; its efficacy and safety has not been established."[6] We found no announced regulatory submission, in any jurisdiction, as of this writing.

Which brings the question back to the vial. Our own price index tracked 16 survodutide listings across 15 vendors on August 1, 2026. A research-market product labelled survodutide shares a peptide name with the molecule in these trials. It is not the pen used in SYNCHRONIZE-1, it did not come from that supply chain, and it does not inherit the identity, purity, formulation or evidence of the material that produced these numbers. The honest summary is that survodutide now has real phase 3 human data, that the data are entirely sponsor-run, that its cardiovascular safety trial has not reported, and that it is approved nowhere.

Questions people are asking