Retatrutide Cleared Two More Phase 3 Trials — and Missed the Thing That Mattered Most

On 23 July 2026, Eli Lilly reported positive topline results from two more pivotal Phase 3 trials of retatrutide, its investigational GIP, GLP-1 and glucagon triple hormone receptor agonist. The headline numbers were large: up to 20.8% average weight reduction in TRIUMPH-2 and up to 22.6% in TRIUMPH-3.¹ The number that mattered most was not in the headline, and most coverage buried it. Retatrutide remains an investigational compound with no approval from the FDA, the EMA or any other regulator; nothing below is medical advice, dosing guidance or a statement of availability, and the research-use-only reference materials this site supplies are laboratory chemicals, not medicines.

Updated 1 October 2026. TRIUMPH-2 was published in The Lancet on 29 September 2026 and presented at the EASD annual meeting in Milan.¹¹ This revision adds its peer-reviewed results, including the primary (treatment-regimen) analysis, the TRIUMPH-1 publication in the New England Journal of Medicine, and the current regulatory position. TRIUMPH-3 remains topline-only.

What did Lilly actually announce?

Two trials, both read out at 80 weeks, both hitting their primary endpoint.¹ TRIUMPH-2 enrolled 1,152 adults with obesity or overweight and type 2 diabetes, randomised 1:1:1:1 to retatrutide 4 mg, 9 mg, 12 mg or placebo. Mean baseline weight was 106.4 kg, BMI 38.2 kg/m², A1C 7.7%. TRIUMPH-3 was larger and sicker: 1,949 adults with severe obesity, defined as BMI ≥35, plus established cardiovascular disease, with or without type 2 diabetes, randomised 1:1:2 to 9 mg, 12 mg or placebo. Mean baseline weight was 111.4 kg and BMI 40.4 kg/m².

Retatrutide is the third generation of a lineage that began with an incretin peptide isolated from lizard venom and now runs through single, dual and triple receptor agonists — a history we cover in the origins of the GLP-1 class. For what the molecule is and how it differs from its predecessors, see what retatrutide is and our comparison of semaglutide, tirzepatide and retatrutide.

The 23 July announcements were topline only: Lilly released point estimates without p-values.

The full TRIUMPH-2 dataset followed on 29–30 September 2026, published in The Lancet and presented in an EASD symposium in Milan.¹¹¹⁰ Its primary analysis uses a different estimand from the headline figure (see below). TRIUMPH-3 has not yet been published; everything said about it here still rests on Lilly’s topline release.

22.6% the largest average weight reduction reported in TRIUMPH-3 — in a population selected for severe obesity and existing heart disease

How much weight did the trials report?

The figures below are efficacy-estimand results: what the model estimates would have happened had every participant remained on the assigned intervention for the full 80 weeks. The treatment-regimen estimand, which accounts for people who stopped, is typically several points lower. Lilly did not lead with it in the July announcement; the peer-reviewed TRIUMPH-2 paper reports it as the primary analysis (second table below).

Efficacy-estimand results from Lilly’s topline release of 23 July 2026.¹⁹ Doses are reported here strictly as trial facts — what was administered under supervision in a registrational study — and not as guidance of any kind. Retatrutide is investigational and unapproved; research-use-only reference materials are not clinical products and carry no human use.

What the peer-reviewed TRIUMPH-2 paper reports

Treatment-regimen estimand, The Lancet, 29 September 2026.¹¹ Trial facts only; not guidance of any kind.

The 20.8% headline and the 18.8% primary result are the same trial and the same dose, analysed two ways. The efficacy estimand assumes everyone stays on treatment and is helped by a placebo arm that lost 4.0% rather than 5.1%; the treatment-regimen estimand counts what happened to everyone randomised. Against placebo, the highest dose gained 13.8 percentage points. Of 1,152 participants, 965 (84%) completed study drug.¹¹

Tolerability tracked the class. In TRIUMPH-2, diarrhoea affected 27.4–33.6% of retatrutide participants versus 13.2% on placebo, and nausea 13.7–28.0% versus 8.0%. Dysesthesia — an unusual sensory side effect that has followed retatrutide since Phase 2 — appeared in 4.5–7.3% versus 0.7% on placebo. Discontinuation due to adverse events ran 3.8% to 11.6% across the retatrutide arms against 4.9% on placebo, and 9.8% to 13.5% in TRIUMPH-3 against 4.8%.¹ The Lancet paper adds hypotension (1%, 5% and 6% across the three doses versus under 1% on placebo) and reports permanent discontinuation due to adverse events or death in 4%, 12% and 8% versus 5%. Seven participants died during the trial, six on retatrutide and one on placebo; investigators judged all deaths unrelated to the study intervention.¹¹

Why is the cardiovascular result the real story?

TRIUMPH-3 deliberately enrolled people with established cardiovascular disease. That design choice invites an obvious question, and the trial did not answer it. On the three-point MACE composite — cardiovascular death, myocardial infarction, stroke — there were 27 events on retatrutide against 23 on placebo, a hazard ratio of 1.12 with a 95% confidence interval of 0.64 to 1.96.¹ The point estimate numerically favours placebo. On the wider five-point composite, which adds all-cause death, heart failure and revascularisation, the direction reversed: 44 events against 52, hazard ratio 0.82, confidence interval 0.55 to 1.22. Both intervals comfortably contain 1. Neither tells you anything definitive.

Lilly’s own framing was careful. The company said retatrutide “meaningfully reduced certain cardiovascular risk factors”, citing a 37.0% reduction in triglycerides, 16.5% in non-HDL cholesterol, 9.3 mmHg in systolic blood pressure, 19.0 cm in waist circumference and 51.2% in high-sensitivity C-reactive protein at the highest dose.¹ Risk factors. Not events. FierceBiotech reported that peak weight loss “fell short of previous trials” while the trial failed to prove the medication reduces cardiovascular risk.³ Lilly noted that MACE occurred less frequently than anticipated in both arms, and BioPharma Dive reported that the trials were not designed to definitively establish whether retatrutide lowers the risk of heart attack or stroke.⁴ William Blair’s Andy Hsieh called it difficult to draw solid conclusions given the small event count and short follow-up.⁴

Risk factors are not events. A compound that moves triglycerides, blood pressure and hsCRP has changed a lab report. Whether it changes what happens to a patient is a different question — and TRIUMPH-3 was not built to answer it.

Why were the numbers below earlier retatrutide trials?

TRIUMPH-1, reported on 21 May 2026 in 2,339 adults with obesity or overweight and no diabetes requirement, produced 28.3% at 80 weeks on the 12 mg arm under the efficacy estimand, and 25.0% on the primary treatment-regimen estimand when the trial was published in the New England Journal of Medicine on 29 September 2026.⁵¹² TRIUMPH-4, reported on 11 December 2025 in 445 adults with overweight or obesity and knee osteoarthritis, produced 28.7% at 68 weeks.⁶⁷ Against those, 20.8% and 22.6% look like a step down. Compared on the same estimand the gap narrows: 25.0% in TRIUMPH-1 against 18.8% in TRIUMPH-2 at the highest dose is about six points, against the 7.5 points that 28.3% versus 20.8% suggests.

The caveat is real and cuts the other way. People with type 2 diabetes consistently lose less weight on incretin therapies than people without it, and a cohort with severe obesity plus established cardiovascular disease is not comparable to a healthier obesity-only population. RBC Capital Markets called the TRIUMPH-2 and TRIUMPH-3 results “as expected”, noting that these patients typically lose less given their comorbidities.³ Cross-trial comparison is the weakest form of comparison available; different populations, different baselines, different follow-up. The honest position is that the magnitudes are lower and that there is a plausible, well-documented reason why.

What does the Q1 2027 filing tell you?

Lilly now plans to submit a Biologics License Application to the FDA in the first quarter of 2027, pending completion of the chemistry, manufacturing and controls data package.¹⁸ A filing had previously been expected later in 2026. CNBC reported that the company needed more time to assemble manufacturing and quality-control data for the agency.⁴ That is a delay, not an acceleration, and it is worth stating plainly because a great deal of secondary commentary read the announcement as a step towards imminent approval. It is not. Kenneth Custer, who leads Lilly’s cardiometabolic business, said the package supports global submissions across obesity, knee osteoarthritis and obstructive sleep apnea.¹² Submissions are applications, not authorisations. Lilly reaffirmed the Q1 2027 timeline on 29 September 2026.¹⁴

The pathway itself is contested. According to BioSpace, the FDA has classified retatrutide as a drug rather than a biologic because it does not count 40 or more alpha amino acids; that would route it to a New Drug Application with five years of exclusivity instead of a Biologics License Application with twelve. Lilly sued in September 2024; courts vacated the biologic designation decision but returned the question of whether retatrutide is “analogous to a protein” to the agency, and Lilly has appealed. The litigation was unresolved as of August 2026, and Lilly says it will file a BLA.¹³

None of this alters the compound’s legal status in Europe or the United States. Retatrutide has no marketing authorisation anywhere. Material sold by any vendor as retatrutide is not a licensed medicine, cannot lawfully be supplied for human use, and is not equivalent to clinical trial supply — a distinction we set out in our piece on research-grade retatrutide.

An honest read of the evidence

Five things are true at once, and most coverage picked two.

The weight-loss result is genuine and large. A 22.6% average reduction over 80 weeks in a severe-obesity, established-CVD population is a substantial finding on its own terms, and the A1C reductions in TRIUMPH-2 are consistent with what a triple agonist would be expected to do. Second, the cardiovascular endpoint did not deliver. The three-point MACE composite numerically favoured placebo, both composites had confidence intervals that span the null, and the trial was underpowered for the question by design. Third, the magnitudes are below TRIUMPH-1 and TRIUMPH-4, with a defensible population-based explanation but no way to verify it from topline data. Fourth, the regulatory timeline moved back, on manufacturing grounds rather than efficacy grounds as far as anyone has stated publicly. Fifth, and most limiting for TRIUMPH-3: it is still a press release. No p-values, no full protocol, no peer review, no independent statistical review. TRIUMPH-2 is no longer in that category: its full dataset was published in The Lancet on 29 September 2026, and on its primary analysis the headline moved from 20.8% to 18.8%.¹¹¹

The broader pipeline point is worth noting without overstating it. Triple agonism is one of several directions the field is taking; amylin-based approaches such as cagrilintide are being developed on a different mechanistic premise, and none of these programmes has yet demonstrated a hard cardiovascular outcome benefit in this class of trial. The field is generating impressive surrogate endpoints faster than it is generating event data. The most direct cross-trial comparison so far is Lilly’s own amylin-plus-tirzepatide combination: 23.3% at 48 weeks in type 2 diabetes in a phase 2b trial, against 20.8% for retatrutide at 80 weeks on the same efficacy estimand, with caveats on size, design and tolerability set out in our analysis of EloraTZP versus retatrutide.

Glucagon-based dual agonists are another branch of the same pipeline; survodutide’s phase 3 MRI substudy is reviewed in our survodutide body-composition analysis.

To restate what should not need restating: retatrutide is investigational. It is not approved by the FDA, not approved by the EMA, and not approved by any other regulator. Nothing on this page is medical advice, a use recommendation or a dosing instruction, and the doses reported above describe what was administered in supervised registrational trials — nothing more. Condor Research supplies compounds strictly as research-use-only reference materials for in vitro and preclinical laboratory work. A research-use-only reference material is not a medicine, is not manufactured to pharmaceutical standards for human administration, and is not intended for human or veterinary use under any circumstances.

Condor Research · Scientific desk
info@condorresearch.com