Many white mannequin heads with masks on display

Can a Pill Treat Sleep Apnea?

A daily pill cut sleep-test apnea scores more than placebo in two trials. The score is not the outcome: what the trials show, and what they cannot.

In two 52-week phase 3 trials, once-daily oral orforglipron lowered the apnea-hypopnea index by 12 to 13 events an hour more than placebo, in people both with and without positive airway pressure.

The appeal of a once-daily pill over a nightly breathing mask needs no explaining. Two trials of the oral drug orforglipron did lower the sleep-test count of breathing interruptions by more than placebo. But that count is not a health outcome: when a large trial used a mask to push the score near normal, heart attacks and strokes did not fall.

What this means

In two 52-week trials of 712 adults with moderate to severe sleep apnea and a body mass index of at least 27, a once-daily oral drug, orforglipron, lowered the apnea-hypopnea index by 12 to 13 events an hour more than placebo, in people using a breathing machine and in people who were not. The index is a sleep-test number: this paper reports no weight-loss figure and no heart, stroke or mortality outcome, and CPAP drove the same number close to normal in a large trial without reducing cardiovascular events. What a lower index means for how someone sleeps or how long they live is still untested for this drug.

What the two trials measured

Both trials ran under a single master protocol, NCT06649045, and together randomized 712 adults with moderate to severe obstructive sleep apnea and a body mass index of at least 27. Everyone took one tablet a day for 52 weeks, assigned 1:1 to orforglipron at the highest tolerated dose — equivalent to a 9, 14.5 or 17.2 mg tablet — or to placebo. Trial 1 enrolled people who were not using positive airway pressure; trial 2 enrolled people who were using it. Enrollment ran from November 11, 2024, to July 23, 2025, at 66 sites in nine countries, and final follow-up was July 30, 2026.

The primary result

In trial 1, which randomized 363 people (mean age 48.4 years, mean BMI 37.4, mean starting AHI 50.7 events an hour, 31.4% women), the index fell by 17.8 events an hour on orforglipron and by 4.5 on placebo. The between-group difference was -13.2 events an hour (95% CI -17.7 to -8.7; P<.001). In trial 2, with 349 people (mean age 51.1 years, BMI 38.6, starting AHI 57.6, 26.9% women), it fell by 17.3 against 5.3, a difference of -12.0 (-17.1 to -6.9; P<.001). Both numbers describe a change in a surrogate measure: the abstract reports no link between that change and any symptom, a cardiovascular event or death.

Set against each trial’s own starting value, those falls are about 35% and 30% of the index, against about 9% on placebo. Both percentages are arithmetic from the printed means; the abstract prints changes, not percent change.

Why the index is not the outcome

The index counts breathing interruptions per hour of sleep. It is a surrogate endpoint: a test result standing in for the outcomes patients would notice, and the primary outcome was the index itself. The measure closest to how a night feels, sleep-related impairment, was one of seven multiplicity-controlled secondary endpoints: 6 of the 7 were statistically significant in favor of orforglipron, and the abstract does not say which one was not, or give any of their values.

The trial that makes that gap matter

CPAP is the obvious comparison, because it is very good at moving this number. In SAVE, 2,717 adults aged 45 to 75 with moderate to severe apnea and existing coronary or cerebrovascular disease were assigned to CPAP plus usual care or to usual care alone, with a mean follow-up of 3.7 years. AHI fell from 29.0 to 3.7 events an hour on CPAP. The primary cardiovascular composite still occurred in 17.0% of the CPAP group and 15.4% of the usual-care group (HR 1.10, 95% CI 0.91 to 1.32; P=0.34). Symptoms and quality of life did improve.

SAVE enrolled people who already had heart or blood-vessel disease, so a null result there does not show that no apnea treatment helps anyone. It does show that a large fall in this index, on its own, is not evidence of fewer cardiovascular events.

What the abstract does not report

The abstract gives no weight-loss figure for either trial. That gap matters: if much of the index change arrived with weight loss, the drug’s own contribution to keeping the airway open is smaller than the headline difference suggests, and this paper gives no number that would let a reader divide the two. In orforglipron’s own obesity trial, ATTAIN-1, participants lost 7.5%, 8.4% and 11.2% of body weight at the three doses after 72 weeks, against 2.1% on placebo, among 3,127 adults. That is a different trial in a different population and cannot fill this gap.

Nor does the abstract give per-arm denominators for the index analysis beyond the randomized totals, so how people who stopped treatment were counted is not visible. In trial 2, where everyone was already on PAP, it does not say whether PAP use held steady through the year; a difference measured on top of PAP depends on that.

Side effects, as printed

Gastrointestinal adverse events were reported by 123 people on orforglipron (68.0%) and 68 on placebo (37.4%) in trial 1, and by 125 (70.6%) and 63 (36.6%) in trial 2. The abstract gives no severity breakdown, no count of people who stopped treatment because of them and no serious-event totals, so those questions stay open.

Where it sits among the incretins

Orforglipron is the newest incretin drug tested in obstructive sleep apnea; its phase 3 program follows tirzepatide’s. Tirzepatide reported larger falls in the same two settings: 20.0 events an hour more than placebo without PAP (95% CI -25.8 to -14.2) and 23.8 more with PAP (-29.6 to -17.9), with all prespecified key secondary endpoints significant. That paper carries a correction notice whose contents could not be read, so its figures stand as first published. No trial has compared the two drugs head to head in sleep apnea, and a gap of this size between separate trials cannot be credited to potency, to weight loss or to the people each trial enrolled.

Funding, and what we could read

Only the abstract of the paper was available to us at press time, so the safety tables, the per-arm weight figures and the paper’s own funding statement were not read. The abstract carries no funding statement and no conflict-of-interest information; the trial registry, NCT06649045, lists Eli Lilly and Company as lead sponsor. Conflicts of interest were not examined, so nothing here asserts or clears them.

Terms explained

  • Apnea-hypopnea index (AHI) — the number of times an hour that breathing stops (apnea) or becomes too shallow (hypopnea) during sleep, as scored on a sleep study. A higher number means more interruptions.
  • Obstructive sleep apnea — a condition in which the upper airway repeatedly narrows or closes during sleep, cutting off airflow and dropping blood oxygen; it is diagnosed with a sleep test.
  • PAP / CPAP — positive airway pressure: a machine that pushes air through a mask to hold the airway open. CPAP delivers that pressure continuously.
  • Surrogate endpoint — a laboratory or test measure used in a trial in place of the outcomes patients care about, such as heart attack, stroke or death. A treatment can move a surrogate without changing those outcomes.
  • Hypoxic burden — a way of weighting each oxygen dip during sleep by how deep and how long it is, rather than simply counting events; one of the seven secondary endpoints in these trials.

Sources

  1. Primary source (abstract only): Malhotra A, Gottlieb DJ, Woehrle H, Dennehy EB, Calamai G, et al. “Orforglipron for Obstructive Sleep Apnea: Results From 2 Randomized Clinical Trials (ATTAIN-OSA).” JAMA, published online November 9, 2026. https://jamanetwork.com/journals/jama/fullarticle/2856406 (doi:10.1001/jama.2026.24080; PMID: 43002161). 712 randomized across two trials; primary outcome change in AHI at week 52.
  2. Trial registry: ATTAIN-OSA master protocol, ClinicalTrials.gov identifier NCT06649045. https://clinicaltrials.gov/study/NCT06649045. 712 participants; primary outcome change in AHI; lead sponsor Eli Lilly and Company.
  3. Challenge (abstract): McEvoy RD, Antic NA, Heeley E, et al. “CPAP for Prevention of Cardiovascular Events in Obstructive Sleep Apnea.” N Engl J Med 2016;375(10):919-931. https://pubmed.ncbi.nlm.nih.gov/27571048/ (doi:10.1056/NEJMoa1606599; PMID: 27571048). SAVE; 2,717 adults; mean follow-up 3.7 years.
  4. Landscape comparator (abstract): Malhotra A, Grunstein RR, Fietze I, et al. “Tirzepatide for the Treatment of Obstructive Sleep Apnea and Obesity.” N Engl J Med 2024;391(13):1193-1205. https://pubmed.ncbi.nlm.nih.gov/38912654/ (doi:10.1056/NEJMoa2404881; PMID: 38912654). Correction notice on record (N Engl J Med 2024;391(15):1464; doi:10.1056/NEJMx240005); its contents were not read.
  5. Adjacent obesity trial (abstract): Wharton S, Aronne LJ, Stefanski A, et al. “Orforglipron, an Oral Small-Molecule GLP-1 Receptor Agonist for Obesity Treatment.” N Engl J Med 2025;393(18):1796-1806. https://pubmed.ncbi.nlm.nih.gov/40960239/ (doi:10.1056/NEJMoa2511774; PMID: 40960239). ATTAIN-1; 3,127 adults; 72 weeks.