Retatrutide for Sleep Apnea: What the NEJM Trials Found
Retatrutide sleep apnea results from NEJM: breathing events fell 34.3 an hour on 9 mg vs 9.9 on placebo over 80 weeks. Doses, Zepbound and CPAP, explained.
In the two Phase 3 retatrutide sleep apnea trials published in NEJM on October 21, 2026, people without diabetes on 9 mg had 34.3 fewer breathing pauses and shallow breaths per hour of sleep after 80 weeks, against 9.9 fewer on placebo. People with type 2 diabetes improved less, by 24.7 an hour on 9 mg and 25.4 on 12 mg against 10.3 on placebo. Retatrutide isn't approved for sleep apnea or anything else, and no trial has tested whether it lets anyone give up CPAP.
Key takeaways
- In people without diabetes, 9 mg and 12 mg cut breathing events by 34.3 and 31.7 an hour over 80 weeks, against 9.9 on placebo.
- People with type 2 diabetes lost less weight and got smaller drops: 24.7 and 25.4 an hour on 9 and 12 mg, against 10.3 on placebo.
- About half the people on 9 mg without diabetes ended in the mild-or-better zone. The other half didn't get there.
- With placebo subtracted, the results land close to Zepbound's approved sleep apnea trials, which were shorter and left out diabetes. Nobody has compared the two directly.
- No study has tested stopping CPAP on retatrutide, and PAP users agreed to pause their machines for about a week before each sleep test.
Both sleep apnea studies sat inside bigger obesity trials, so this page sticks to the sleep results. Weight results, dosing and the full design of each parent trial are on the TRIUMPH-1 and TRIUMPH-2 pages.
What the retatrutide sleep apnea trials measured
The main yardstick was the apnea-hypopnea index (AHI): how many times an hour, on average, breathing stops or turns shallow enough to drop oxygen by at least 4%. It was counted in an overnight lab sleep study (polysomnography). Lilly's medical information summary of the trials uses the usual bands: 5 to under 15 is mild, 15 and up is moderate, 30 and up is severe. Everyone needed an AHI of at least 15 to get in.
The primary endpoint was the change in AHI at week 80 on 9 mg and 12 mg, compared with placebo. The 4 mg comparison was a secondary result. Both trials also tracked:
- the percent change in AHI
- the share of people whose AHI fell by half or more
- the share who ended with an AHI under 5, or 5 to 14 with an Epworth score of 10 or less (a 0 to 24 questionnaire about dozing off during the day). Zepbound's label calls this remission or mild non-symptomatic sleep apnea
- sleep apnea-specific hypoxic burden, a measure of how deep and how long the oxygen drops run per hour of sleep
- a questionnaire on how much poor sleep gets in the way of daily life (PROMIS sleep-related impairment)
- body weight
There are two ways of counting. NEJM's main analysis for the sleep results is a hybrid-treatment estimand. It counts everyone randomized, but if a person's treatment failed for lack of effect, their later values are reset to where they started. Lilly's June headline of "up to 36.1 events per hour (60.6%)" uses the efficacy estimand instead, which assumes everyone stayed on the drug. This page uses the hybrid numbers unless it says otherwise.
Results by dose in both trials
TRIUMPH-1-OSA enrolled people without diabetes, about 60 per group. Numbers below are from Lilly's medical information summary of the NEJM results, at week 80, using the hybrid-treatment estimand (weight uses the treatment-regimen estimand). The placebo differences are from the NEJM abstract.
| TRIUMPH-1-OSA, week 80 | 4 mg | 9 mg | 12 mg | Placebo |
|---|---|---|---|---|
| Change in AHI, events an hour | -22.9 | -34.3 | -31.7 | -9.9 |
| Difference from placebo | -13.0 | -24.4 | -21.9 | |
| Change in AHI, percent | -38.5% | -57.3% | -53.3% | -11.1% |
| AHI down by half or more | 45.9% | 65.7% | 61.4% | 23.9% |
| AHI under 5, or 5 to 14 without daytime sleepiness | 26.3% | 51.6% | 42.3% | 12.5% |
| Hypoxic burden, percent change | -56.4% | -75.0% | -70.2% | -25.5% |
| Body weight, percent change | -12.5% | -22.2% | -20.6% | -4.8% |
TRIUMPH-2-OSA enrolled people with type 2 diabetes, 109 to 116 per group, with the same sources and methods.
| TRIUMPH-2-OSA, week 80 | 4 mg | 9 mg | 12 mg | Placebo |
|---|---|---|---|---|
| Change in AHI, events an hour | -19.2 | -24.7 | -25.4 | -10.3 |
| Difference from placebo | -8.9 | -14.4 | -15.1 | |
| Change in AHI, percent | -27.6% | -39.5% | -40.9% | -13.3% |
| AHI down by half or more | 39.1% | 52.3% | 52.5% | 24.4% |
| AHI under 5, or 5 to 14 without daytime sleepiness | 24.3% | 39.5% | 38.9% | 12.6% |
| Hypoxic burden, percent change | -43.9% | -62.8% | -64.1% | -30.0% |
| Body weight, percent change | -12.1% | -17.9% | -18.5% | -5.6% |
Lilly's summary says the 4 mg dose wasn't part of the trials' statistical testing hierarchy, and most of its comparisons are marked as not passing it, so read that column as descriptive. The 9 mg and 12 mg doses met the primary endpoint in both trials.
Our read: on the two higher doses, people were about 3 to 4 times as likely as on placebo to end in the mild-or-better zone (51.6% against 12.5% on 9 mg in TRIUMPH-1-OSA). That's a big effect. It still leaves roughly half of them short of it.
Why 9 mg did as well as 12 mg
In TRIUMPH-1-OSA, 9 mg came out ahead of 12 mg on every row, weight included (22.2% against 20.6%). With about 60 people per group, NEJM's confidence intervals for the two doses overlap almost completely (-34.1 to -14.7 for 9 mg, -32.2 to -11.5 for 12 mg). That reads as a tie. It doesn't show that the higher dose works worse.
Why the diabetes trial showed smaller drops
People with type 2 diabetes lost less weight, 17.9% and 18.5% on 9 and 12 mg, and their AHI fell less with it. By our arithmetic, the 9 mg group in TRIUMPH-2-OSA started at an average of 58.2 events an hour and finished near 33.5, still in the severe band on average. The same weight gap shows up in TRIUMPH-2 as a whole, which the retatrutide diabetes page covers.
Who was in TRIUMPH-1-OSA and TRIUMPH-2-OSA
TRIUMPH-1-OSA randomized 243 adults at 22 centers in 3 countries, and TRIUMPH-2-OSA randomized 448 at 49 centers in 5 countries, according to Lilly's summary. Everyone had a BMI of at least 27 and moderate to severe obstructive sleep apnea, and the groups were heavy: average weight 119.6 kg and 115.9 kg in NEJM, with average BMI of about 40 to 44 by group.
Lilly's baseline table adds detail:
- Most had severe sleep apnea: 73% to 82% of each group in TRIUMPH-1-OSA and 69% to 77% in TRIUMPH-2-OSA had an AHI of 30 or more.
- Most were men. Women made up 36% to 37% of each group in TRIUMPH-1-OSA and 38% to 43% in TRIUMPH-2-OSA.
- Average age was 46 to 49 without diabetes and 54 to 56 with it.
- 69% to 85% of each group were Hispanic or Latino, 25% to 41% were American Indian or Alaska Native, and 0% to 2% were Asian.
- About 3 in 10 were using PAP at the start (26% to 33% per group). Average Epworth sleepiness scores ran from 9.2 to 12.0 out of 24.
The registry entries for both trials excluded people who treated their apnea with a dental appliance or any device other than PAP. Men outnumbered women in every group, and Asian participants were 2% or fewer per group, so there's little data on them.
How it compares with Zepbound for sleep apnea
Zepbound (tirzepatide) is the approved comparison. The FDA approved it on December 20, 2024 for moderate to severe obstructive sleep apnea in adults with obesity, at 10 or 15 mg weekly, based on SURMOUNT-OSA (NEJM, 2024). That was two 52-week trials. Study 5 took people who couldn't or wouldn't use PAP, Study 6 took people already on PAP, and both left out type 2 diabetes. The table below sets those results beside TRIUMPH-1-OSA. It's a cross-trial comparison: different people, different lengths, different ways of handling dropouts.
| Zepbound, Study 5 (no PAP) | Zepbound, Study 6 (on PAP) | Retatrutide 9 mg, TRIUMPH-1-OSA | Retatrutide 12 mg, TRIUMPH-1-OSA | |
|---|---|---|---|---|
| Length | 52 weeks | 52 weeks | 80 weeks | 80 weeks |
| Starting AHI, drug group | 52.9 | 46.1 | 59.6 | 56.2 |
| AHI change minus placebo | -20.0 | -23.8 | -24.4 | -21.9 |
| AHI down by half or more, drug vs placebo | 61.2% vs 19% | 72.4% vs 23.3% | 65.7% vs 23.9% | 61.4% vs 23.9% |
| AHI under 5, or 5 to 14 with Epworth 10 or less | 42.2% vs 15.9% | 50.2% vs 14.3% | 51.6% vs 12.5% | 42.3% vs 12.5% |
| Weight change, drug vs placebo | -17.7% vs -1.6% | -19.6% vs -2.3% | -22.2% vs -4.8% | -20.6% vs -4.8% |
| How missing data was handled | Everyone dosed, multiple imputation | Everyone dosed, multiple imputation | Hybrid-treatment estimand | Hybrid-treatment estimand |
Sources: Zepbound prescribing information, section 14.2 (Table 9); Lilly's medical information summary of the retatrutide trials. That's tirzepatide data in the first two columns, not retatrutide.
With each placebo subtracted, the AHI gaps sit within a few events of each other. Retatrutide's group started more severe and was treated for 28 weeks longer, so nothing here shows that it beats Zepbound for sleep apnea. TRIUMPH-5, the head-to-head trial against tirzepatide, has weight as its primary endpoint, not sleep apnea, and its primary completion is estimated for November 2026. One thing retatrutide does have that Zepbound's sleep apnea approval lacks is a trial in type 2 diabetes. Weight, side effects and self-pay prices are compared on the retatrutide vs Zepbound page.
Can retatrutide replace a CPAP?
We found no registered retatrutide trial that tests that, and the published results can't answer it. People already on PAP could join the retatrutide trials if they planned to keep using it, and the registry entries say they had to agree to pause it for about 7 days before each sleep study. So the AHI numbers show how people breathed with the machine off for a week. They don't show whether it was safe to stop for good, and Lilly's summary doesn't report separate results for the PAP users.
Four more things in the data argue against reading the results as "no more CPAP":
- About half the people on 9 mg without diabetes (51.6%) reached an AHI under 5, or mild apnea without daytime sleepiness. In the diabetes trial it was 39.5%.
- By our arithmetic, the 9 mg group in TRIUMPH-1-OSA averaged about 25 events an hour at week 80 (59.6 at the start, down 34.3). That's still moderate.
- The TRIUMPH-1 registry entry lists the share of people reaching an AHI under 5 as a secondary outcome, and Lilly's summary doesn't report it.
- The trials had 4 weeks of follow-up after the last dose and report no sleep study after stopping retatrutide, so nobody knows how fast apnea returns once the drug and the weight loss stop.
The approved drug has the same gap. Zepbound's label says its sleep apnea trials "did not evaluate the timing or appropriateness of PAP discontinuation in patients who were previously compliant with PAP therapy." The FDA called PAP the standard of care for moderate to severe sleep apnea when it approved Zepbound, and the National Heart, Lung, and Blood Institute calls a PAP machine the most common treatment.
Our read: big weight loss on any drug is a good reason to ask your sleep doctor about a repeat sleep study. It isn't a reason to put the machine in the closet on your own.
Side effects in the sleep apnea trials
They looked like retatrutide's side effects elsewhere: mostly stomach trouble, usually mild to moderate and most common during dose increases, per Lilly's summary. In TRIUMPH-1-OSA, diarrhea hit 30%, 44% and 35% of people on 4, 9 and 12 mg against 15% on placebo, and nausea 27%, 20% and 23% against 10%.
Stopping because of side effects was close to placebo. In TRIUMPH-1-OSA it was 4 people (7%) on 4 mg, none on 9 mg and 5 (8%) on 12 mg, against 4 (6%) on placebo. In TRIUMPH-2-OSA, 9 mg was highest at 10 people (9%), against 4 (4%) on placebo. One case of pancreatitis was confirmed, on 12 mg in TRIUMPH-1-OSA. Four people died across both trials: three on placebo and one on 9 mg, from bleeding in the upper digestive tract. Lilly's summary doesn't say whether that death was linked to the drug. The full picture is on the retatrutide side effects page.
Everyone in these trials got Lilly's own retatrutide at a set dose under a protocol. Vials sold online under the name aren't Lilly's product, and none of them was tested here. If you're weighing retatrutide for sale online, that page covers what's actually sold and what third-party test reports can and can't tell you.
When could retatrutide be approved for sleep apnea?
Sleep apnea is part of Lilly's plan. In its Q2 2026 results on August 5, 2026, Lilly said its clinical data package was complete to support global registrations for obesity, obstructive sleep apnea and knee osteoarthritis pain.
Lilly last said, on September 29, 2026, that it plans to submit retatrutide to the FDA in the first quarter of 2027 as a Biologics License Application (BLA). On October 2, 2026, the FDA decided that retatrutide is a drug, not a biological product, so an application would need to be filed under section 505 of the Federal Food, Drug, and Cosmetic Act. Lilly's appeal on that question is pending in the Seventh Circuit, and we found no ruling as of October 31, 2026.
Lilly's October 2026 FAQ gives no launch date and says launch "will depend on the completion of these trials and the regulatory approval process." Our FDA approval page tracks each step. For now, Zepbound is the drug with an FDA sleep apnea approval, at a maintenance dose of 10 or 15 mg a week on its label, and a sleep doctor can tell you whether it fits with the treatment you already use.
Frequently asked questions
Does retatrutide help sleep apnea?
In two Phase 3 trials, yes. Adults with moderate to severe obstructive sleep apnea had far fewer breathing pauses per hour of sleep after 80 weeks on retatrutide than on placebo, with or without type 2 diabetes. The results were published in NEJM on October 21, 2026, but retatrutide isn't approved for sleep apnea or anything else yet.
How much does retatrutide lower AHI?
In people without diabetes (TRIUMPH-1-OSA), AHI fell by 22.9, 34.3 and 31.7 events an hour on 4, 9 and 12 mg, against 9.9 on placebo, from a starting average of 58.6. In people with type 2 diabetes (TRIUMPH-2-OSA), it fell by 19.2, 24.7 and 25.4, against 10.3, from 55.9. Lilly's higher headline of 36.1 an hour comes from a different analysis that assumes everyone stayed on the drug.
Can retatrutide replace a CPAP?
No trial has tested that. People who used PAP in the retatrutide trials agreed to pause it for about 7 days before each sleep study, and Lilly's summary doesn't report whether anyone stopped for good. Even on 9 mg, about half of people without diabetes didn't reach the mild-or-better mark, so any change to PAP belongs with a sleep doctor and a repeat sleep study.
Is retatrutide better than Zepbound for sleep apnea?
Nobody knows, because no trial has compared them. With each trial's placebo subtracted, retatrutide 9 mg cut AHI by 24.4 events an hour over 80 weeks, and Zepbound cut it by 20.0 and 23.8 in its two 52-week trials. The groups, trial lengths and analysis methods differ, and only Zepbound is approved for sleep apnea.
Does retatrutide work for sleep apnea with type 2 diabetes?
Yes, but less than without diabetes. In TRIUMPH-2-OSA, 448 adults with type 2 diabetes had AHI drops of 24.7 and 25.4 events an hour on 9 and 12 mg, against 10.3 on placebo, and they lost less weight than the group without diabetes. Zepbound's sleep apnea trials excluded people with type 2 diabetes.
When could retatrutide be approved for sleep apnea?
There is no date. Lilly said in August 2026 that its data package was complete to support registrations for obesity, obstructive sleep apnea and knee osteoarthritis pain, and last said, on September 29, 2026, that it planned to submit to the FDA in the first quarter of 2027. The FDA then decided on October 2, 2026 that retatrutide is a drug, not a biological product, and Lilly's appeal on that question had no ruling we could find as of October 31, 2026.
Sources
- Retatrutide Once Weekly for Treatment of Obstructive Sleep Apnea (TRIUMPH-1-OSA and TRIUMPH-2-OSA). The New England Journal of Medicine, 2026. doi.org
- What is the impact of retatrutide in patients with obesity or overweight and obstructive sleep apnea?. Lilly Medical Information, 2026. medical.lilly.com
- Lilly's triple agonist, retatrutide, drove substantial improvements in weight, A1C, knee osteoarthritis pain, and obstructive sleep apnea, demonstrating its remarkable potential to treat obesity and its complications. Eli Lilly and Company (PR Newswire), 2026. prnewswire.com
- A Study of Retatrutide (LY3437943) in Participants Who Have Obesity or Overweight (TRIUMPH-1), NCT05929066. ClinicalTrials.gov, 2026. clinicaltrials.gov
- A Study of Retatrutide (LY3437943) in Participants With Type 2 Diabetes Mellitus Who Have Obesity or Overweight (TRIUMPH-2), NCT05929079. ClinicalTrials.gov, 2026. clinicaltrials.gov
- Zepbound (tirzepatide) injection: prescribing information. DailyMed, U.S. National Library of Medicine, 2026. dailymed.nlm.nih.gov
- FDA Approves First Medication for Obstructive Sleep Apnea. U.S. Food and Drug Administration, 2024. fda.gov
- Tirzepatide for the Treatment of Obstructive Sleep Apnea and Obesity (SURMOUNT-OSA). The New England Journal of Medicine, 2024. doi.org
- A Study of Retatrutide (LY3437943) Compared to Tirzepatide (LY3298176) in Adults Who Have Obesity (TRIUMPH-5), NCT06662383. ClinicalTrials.gov, 2026. clinicaltrials.gov
- Sleep Apnea: Treatment. National Heart, Lung, and Blood Institute, 2025. nhlbi.nih.gov
- Eli Lilly and Company Q2 2026 results, Form 8-K exhibit 99.1. U.S. Securities and Exchange Commission, 2026. sec.gov
- Lilly's triple agonist, retatrutide, delivered substantial weight loss and A1C reduction, underscoring its potential promise for people with obesity and type 2 diabetes. Eli Lilly and Company (PR Newswire), 2026. prnewswire.com
- FDA decision letter on retatrutide (Eli Lilly v. Kennedy, 7th Cir. No. 26-1301, Doc. 46-2). U.S. Court of Appeals for the Seventh Circuit (CourtListener), 2026. storage.courtlistener.com
- Eli Lilly and Company v. Robert Kennedy Jr., No. 26-1301 (docket). CourtListener, 2026. courtlistener.com
- What to know about retatrutide. Eli Lilly and Company, 2026. lilly.com
Dr. Juan Pablo Frías, MD
Physician-Scientist & Clinical Trialist • Endocrinology, Diabetes & Metabolism
Dr. Frías earned his M.D. from Vanderbilt University School of Medicine and completed his endocrinology fellowship at UC San Diego. A principal investigator on landmark incretin trials (including retatrutide, tirzepatide and semaglutide), he reviews all guides on Retatrutide360 for clinical accuracy, dosing safety and evidence fidelity.
Related guides & clinical tools
Continue reading our evidence-backed guides and use our syringe calculation tools.
Retatrutide vs Zepbound
Retatrutide vs Zepbound: the Zepbound label's weight and sleep apnea results next to TRIUMPH data, plus what 12 self-pay fills cost as of October 2026.
TRIUMPH 1 Retatrutide Results
TRIUMPH 1 retatrutide results from the NEJM paper: 25.0% average weight loss on 12 mg over 80 weeks, serious side effects by dose, who stopped and why.
TRIUMPH-2 Results
TRIUMPH 2 retatrutide results from The Lancet: 20.8% weight loss and a 1.5-point A1C drop on 12 mg over 80 weeks in type 2 diabetes, with dropout rates by dose.
Is Retatrutide FDA Approved? Approval Status and Timeline
Is retatrutide FDA approved? No. Lilly plans to file in Q1 2027, and the FDA says it must go in as a drug. See the review clock and legal ways to get it now.
Turn any dose into syringe units
Vial, water and dose in, exact syringe units out. Calculates reconstitution volumes and needle tick marks for 5 mg to 60 mg vials with zero guesswork.