TRIUMPH-3: retatrutide in severe obesity and heart disease

TRIUMPH-3 tested retatrutide in people with severe obesity who had already had a heart attack, a stroke or serious artery disease in the legs. Lilly announced topline results on 23 July 2026. Weight loss was large. The trial also counted heart events, but it was far too small to say whether the drug prevents them. The peer-reviewed paper has not been published.

What the trial did

To join, participants needed a BMI of 35 or more and at least one of: a previous heart attack, a previous stroke, or symptomatic peripheral arterial disease. People who had had a heart attack, stroke, artery procedure or hospital stay for heart failure in the previous 90 days were excluded, along with the usual exclusions for pancreatitis and medullary thyroid cancer.^[2] Registered as NCT05882045.

Two things make this trial different from TRIUMPH-1. It included people with type 2 diabetes alongside people without. And it only tested the two higher doses, with twice as many people assigned to placebo as to each drug arm. Everyone on retatrutide started on a low dose stepped up every four weeks to a target of 9 mg or 12 mg.^[1] The average starting weight was 111.4 kg and the average BMI 40.4.

Weight loss

Figures at 80 weeks.^[1] As with TRIUMPH-2, Lilly's announcement gave only the efficacy estimand, the larger of its two standard figures; the treatment-regimen results, which count people who stopped, will be lower. The gap between the two doses was small, about one percentage point.

Risk markers moved the right way. At the top dose, triglycerides fell by an average of 37.0%, non-HDL cholesterol by 16.5%, systolic blood pressure by 9.3 mmHg, waist circumference by 19.0 cm and the inflammation marker hs-CRP by 51.2%.^[1]

The heart-event numbers, and why they prove nothing yet

Because everyone in TRIUMPH-3 had heart disease, the trial counted major adverse cardiovascular events (MACE) as a pre-planned extra analysis. There were 44 events of the broad kind (death from any cause, heart attack, stroke, heart-failure event or artery procedure) in the combined retatrutide groups and 52 on placebo, a hazard ratio of 0.82. For the narrower kind (cardiovascular death, heart attack or stroke) there were 27 on retatrutide and 23 on placebo, a hazard ratio of 1.12.^[1]

A hazard ratio below 1 points towards fewer events on the drug; above 1, towards more. Both of these come with a "95% confidence interval", the range the true answer could plausibly lie in. For the broad measure it was 0.55 to 1.22; for the narrow one, 0.64 to 1.96.^[1] Both ranges include 1. In plain terms: these numbers are consistent with retatrutide cutting heart events substantially, having no effect, or increasing them. Lilly itself noted that events "occurred less frequently than anticipated" in every group, which makes the estimate even less precise.

This is not a flaw. TRIUMPH-3 was designed to measure weight, and 1,949 people over 80 weeks is nowhere near enough to measure heart attacks. Proving that a weight-loss drug prevents them takes about ten thousand people followed for several years. That trial exists, it is called TRIUMPH-Outcomes, and it is due to complete around 2029.^[3] Until it reports, nobody can say retatrutide reduces heart attacks or strokes. See the TRIUMPH-Outcomes page.

Side effects and drop-outs

Diarrhoea, nausea and constipation led. Decreased appetite was recorded as an adverse event in 13.5% and 14.5% of the two retatrutide groups against 3.0% on placebo. Because the trial included people with diabetes, it also recorded high blood sugar as an event: 3.9% and 3.1% on retatrutide against 13.4% on placebo, the one line where placebo did worse. Dysesthesia affected 6.4% on both doses versus 1.3% on placebo.^[1]

Drop-outs because of side effects were the highest of any TRIUMPH trial after TRIUMPH-4: 9.8% on 9 mg and 13.5% on 12 mg against 4.8% on placebo.^[1] Roughly one person in seven on the top dose stopped. See discontinuation rates.

The announcement gave no heart-rate or rhythm figures. That is a bigger gap here than in the other trials, because this is the population in which a heart-rate rise, seen in Phase 2, would matter most.^[4] See heart rate and rhythm.

What is still unknown

• The treatment-regimen results and full safety tables, including heart rate and rhythm in a high-risk group.
• Whether the weight loss translates into fewer heart events: TRIUMPH-Outcomes, around 2029.
• Whether people with and without diabetes responded differently within the trial.