Retatrutide Phase 3 Results
Every reported figure from the four TRIUMPH trials that have read out — each one labelled with the population, duration and estimand it belongs to.
Four Phase 3 trials have reported. Under the efficacy estimand, mean body-weight reduction at the highest dose was 28.3% (TRIUMPH-1, 80 weeks), 20.8% (TRIUMPH-2, 80 weeks), 22.6% (TRIUMPH-3, 80 weeks) and 28.7% (TRIUMPH-4, 68 weeks). Placebo groups changed by 2.1% to 4.0%.
One trial, TRIUMPH-1, has been published in a peer-reviewed journal. The other three figures come from sponsor announcements.
Start here: why one trial reports two numbers
If you have seen TRIUMPH-1 reported as both 28.3% and 25.0% and concluded that somebody is being dishonest — neither is wrong. They answer different questions, and the difference has a name: the estimand.
| Efficacy estimand | Treatment-regimen estimand | |
|---|---|---|
| Question it answers | What happens if people stay on the treatment? | What happens to everyone assigned to it, whatever they then do? |
| Developer's definition | Efficacy “had all randomized participants remained on study intervention… without initiating prohibited weight management treatments” | “The average treatment effect regardless of adherence to study intervention or initiation of prohibited weight management treatments” |
| TRIUMPH-1, 12 mg | −28.3% | −25.0% |
| TRIUMPH-1, placebo | −2.2% | −3.9% |
| Typically quoted by | Press releases and headlines | The NEJM abstract |
Neither number is the “real” one. The efficacy estimand describes the treatment's potential under adherence; the treatment-regimen estimand describes what happened across the whole randomised group, discontinuations included. A reader who does not know this will think a source is lying when it is not.
Rule of thumb
When you see a weight-loss percentage for any medicine in this class, ask which estimand it uses. If the source does not say, treat the number as incomplete. On this site, every efficacy figure is labelled.
TRIUMPH-1 — obesity without diabetes
- Peer-reviewed
- Phase 3
- NCT05929066
An 80-week randomised, double-blind, placebo-controlled master trial in adults with obesity or overweight and at least one weight-related comorbidity, without diabetes. It randomised 2,339 participants 1:1:1:1 to retatrutide 4 mg, 9 mg, 12 mg or placebo. Mean baseline weight was 112.7 kg with a BMI of 40.0. Topline results were announced on 21 May 2026, and the trial was published in the New England Journal of Medicine on 29 September 2026 (DOI 10.1056/NEJMoa2604169).
| Endpoint | 4 mg | 9 mg | 12 mg | Placebo |
|---|---|---|---|---|
| Body weight change | −19.0% | −25.9% | −28.3% | −2.2% |
| Absolute change | −21.4 kg | −29.2 kg | −31.9 kg | −2.5 kg |
| Waist circumference | −16.3 cm | −21.8 cm | −24.1 cm | −3.6 cm |
| Achieved ≥25% reduction | 27.8% | 52.9% | 62.5% | 2.2% |
| Achieved ≥30% reduction | 15.3% | 37.9% | 45.3% | 0.5% |
| Achieved ≥35% reduction | 5.9% | 20.8% | 27.2% | 0.3% |
The 4 mg weight result was a key secondary endpoint, not the primary. Treatment-regimen estimand at 80 weeks: −17.6%, −23.7%, −25.0% and −3.9% respectively.
The 104-week extension
A pre-specified extension enrolled 532 participants with a baseline BMI of 35 or above who completed week 80 on study drug without discontinuing or permanently reducing dose, then escalated them blinded to a maximum tolerated dose of 9 or 12 mg. Mean baseline weight in this subset was 121.7 kg, BMI 42.8.
Efficacy estimand at 104 weeks: −27.9% (4 mg→MTD), −29.5% (9 mg→MTD), −30.3% (12 mg→MTD), and −19.2% for the group switched from placebo.
Why the extension figures need care
This was not a random subset. By design it enrolled people who had already completed 80 weeks without discontinuing or needing a permanent dose reduction — that is, those tolerating treatment well. A result from a group selected for tolerating a treatment is not interchangeable with a result from everyone who started it. The headline “30.3% at two years” is accurate for that group and should not be read as a general figure.
TRIUMPH-2 — obesity with type 2 diabetes
- Sponsor announcement
- NCT05929079
An 80-week randomised, double-blind, placebo-controlled trial that randomised 1,152 participants 1:1:1:1. Mean baseline weight 106.4 kg, BMI 38.2, A1C 7.7%. Topline results announced 23 July 2026.
| Endpoint | 4 mg | 9 mg | 12 mg | Placebo |
|---|---|---|---|---|
| Body weight change | −12.7% | −19.1% | −20.8% | −4.0% |
| A1C change (from 7.7%) | −1.4% | −1.6% | −1.5% | −0.2% |
A detail we are not smoothing over
The A1C result is not ordered by dose: 9 mg produced a larger reduction (−1.6%) than 12 mg (−1.5%). We report it as stated rather than presenting a tidy dose–response story. Weight reduction in this trial was lower than in TRIUMPH-1, which is consistent with the general observation that weight reduction tends to be smaller in people with type 2 diabetes.
TRIUMPH-3 — severe obesity with cardiovascular disease
- Sponsor announcement
- NCT05882045
An 80-week randomised, double-blind, placebo-controlled trial in adults with severe obesity (BMI ≥35) and established cardiovascular disease, with or without type 2 diabetes. It randomised 1,949 participants 1:1:2 to 9 mg, 12 mg or placebo. Mean baseline weight 111.4 kg, BMI 40.4.
| Endpoint | 9 mg | 12 mg | Placebo |
|---|---|---|---|
| Body weight change | −21.6% | −22.6% | −3.2% |
| Absolute change | −23.9 kg | −25.3 kg | −3.5 kg |
At the highest dose, reported changes in cardiovascular risk markers were: triglycerides −37.0%, non-HDL cholesterol −16.5%, systolic blood pressure −9.3 mmHg, waist circumference −19.0 cm, and hsCRP −51.2%.
The cardiovascular event data — read this carefully
TRIUMPH-3 also reported cardiovascular events. This is the most misread result in the programme, in both directions.
| Analysis | In-study (pre-specified) | On-treatment (not pre-specified) |
|---|---|---|
| MACE-5 death, MI, stroke, HF event, revascularisation |
HR 0.82 (95% CI 0.55–1.22) 44 vs 52 events |
HR 0.73 (95% CI 0.47–1.12) |
| MACE-3 CV death, MI, stroke |
HR 1.12 (95% CI 0.64–1.96) 27 vs 23 events |
HR 0.92 (95% CI 0.51–1.65) |
These data establish neither benefit nor harm
Every confidence interval above crosses 1.0. In plain terms, each result is compatible with a reduction in events, no difference, or an increase — the trial cannot distinguish between those possibilities.
- TRIUMPH-3 was powered to detect weight change, not cardiovascular outcomes.
- Event counts were small (44 vs 52; 27 vs 23), and Eli Lilly noted events occurred less frequently than anticipated in both arms.
- The on-treatment analysis was not pre-specified, which lowers the weight it can carry.
So HR 0.82 is not evidence of cardiovascular benefit, and HR 1.12 is not evidence of harm. Anyone presenting either as a finding is overreading the data. The question is being addressed by TRIUMPH-Outcomes (NCT06383390), an event-driven trial of about 10,000 participants with primary completion expected in February 2029.
TRIUMPH-4 — obesity with knee osteoarthritis
- Sponsor announcement
- NCT05931367
A 68-week randomised, double-blind, placebo-controlled trial in 445 adults with obesity or overweight and knee osteoarthritis, without diabetes, randomised 1:1:1 to 9 mg, 12 mg or placebo. 84.0% had a baseline BMI of 35 or above. Co-primary endpoints were WOMAC pain subscale and percent body-weight change. Announced 11 December 2025 — the programme's first Phase 3 readout.
| Endpoint | 9 mg | 12 mg | Placebo |
|---|---|---|---|
| Body weight change | −26.4% | −28.7% | −2.1% |
| WOMAC pain (from 6.0 points) | −4.5 pts | −4.4 pts | −2.4 pts |
| WOMAC physical function | −4.1 pts | −4.2 pts | −2.1 pts |
| Achieved ≥25% weight reduction | 47.7% | 58.6% | 1.3% |
| Achieved ≥30% weight reduction † | 30.5% | 39.4% | 0.8% |
| ≥70% reduction in WOMAC pain † | 73.0% | 67.7% | 26.2% |
† Not controlled for multiplicity. The percentage expressions of WOMAC change widely quoted elsewhere (−75.8%, −74.3%) come from a post-hoc analysis that was not pre-specified; we report the point change as the primary result.
Note that pain relief was not dose-ordered: 9 mg slightly exceeded 12 mg on both WOMAC measures. Note too the placebo group's 2.4-point pain improvement — a large placebo response is common in pain trials and is exactly why a control arm exists.
Why these four trials should not be ranked against each other
It is tempting to line up 28.3%, 20.8%, 22.6% and 28.7% and conclude something about where retatrutide “works best”. That inference is not available from these data.
- Different populations. Diabetes status, cardiovascular disease and osteoarthritis all differ across the four.
- Different baselines. Mean baseline weight ranged from 106.4 kg to 112.7 kg.
- Different durations. TRIUMPH-4 ran 68 weeks; the others 80.
- Different randomisation ratios. TRIUMPH-3 used 1:1:2 and studied only two doses.
These were designed to answer separate questions in separate populations, not to be compared with one another.