GLP-1s and Sleep Apnea: the 9 Trials Behind the First Ever Approval

The pivotal trial and the basis of the approval: 469 adults with moderate-to-severe apnea and obesity, run as two parallel studies — one in people not using CPAP, one in established users. Tirzepatide cut apnea-hypopnea events by 25.3 to 29.3 an hour against 5.3 to 5.5 on placebo at 52 weeks, better than halving the index, with 17 to 20% weight loss. Daytime sleepiness, oxygen burden and sleep quality all improved. It made tirzepatide the first drug of any class approved for this condition.

PMID 38912654 ↗NCT05412004 ↗DOI 10.1056/NEJMoa2404881 ↗

The pre-specified secondary analysis, and arguably where the clinical significance lives. Systolic blood pressure fell 7 to 9 mmHg, hypoxic burden dropped substantially — a better predictor of cardiovascular death than the apnea index itself — and inflammatory and metabolic markers improved. ⚠ The cardiometabolic gains exceeded what CPAP trials historically produce, which is suggestive rather than proven: no trial has yet tested whether this translates into fewer cardiovascular events in apnea patients.

PMID 41540105 ↗NCT05412004 ↗DOI 10.1038/s41591-025-04071-1 ↗

The first dedicated trial of a GLP-1 in this condition, eight years before the approval: 359 adults with obesity and moderate-to-severe apnea who could not or would not use CPAP, on liraglutide or placebo for 32 weeks. Events fell 12.2 an hour against 6.1, with 5.7% weight loss. ⚠ A far smaller effect than tirzepatide later achieved — but it proved drug-driven weight loss lowers apnea severity independently of airway pressure, which is what made the later program worth funding.

PMID 27005405 ↗NCT01557166 ↗DOI 10.1038/ijo.2016.52 ↗

The pre-drug foundation, run inside a large diabetes trial: 264 adults randomized to intensive lifestyle intervention or education for a year. The intervention group lost 10.8 kg against 0.6 kg, and apnea events fell 5.4 an hour while rising 4.2 in controls. ★ Its lasting contribution is the exchange rate — roughly one fewer event per hour for each kilogram lost — which is the yardstick every subsequent drug trial is implicitly measured against.

PMID 19786682 ↗NCT00194259 ↗DOI 10.1001/archinternmed.2009.266 ↗

The four-year follow-up, addressing durability rather than effect. The lifestyle group held a 5.0 kg advantage and a 2.5 events-per-hour advantage at year four, with remission three times more common. ★ The implication for drug therapy is direct: what governs long-term breathing is whether weight loss is maintained, not how much came off initially — which is why the withdrawal trials on the weight-loss lists matter here too.

PMID 23633746 ↗NCT00194259 ↗DOI 10.5665/sleep.2618 ↗

The imaging study that explains the mechanism, and it is genuinely surprising. 67 adults lost a mean 9.5% of body weight and apnea events fell 31% — but the anatomic factor that predicted improvement was reduction in tongue fat, not total weight, BMI or neck circumference. That reframes the disease around where fat sits rather than how much there is, and it explains why drugs that preferentially strip visceral and ectopic fat produce larger apnea improvements than their weight effect alone would predict.

PMID 31918559 ↗DOI 10.1164/rccm.201903-0692OC ↗

★ The trial that keeps the claim honest. 30 adults randomized to CPAP, liraglutide, or both for 24 weeks. The drug delivered 6.3% weight loss and cut apnea events 33% — yet flow-mediated dilatation improved only in the CPAP arms, an early marker of blood-vessel dysfunction. Small, and the only randomized head-to-head with a vascular endpoint. It is the direct evidence against assuming a drug can simply replace airway pressure, at least within six months.

PMID 38096106 ↗NCT04186494 ↗DOI 10.1513/AnnalsATS.202309-821OC ↗

Included as adjacent context rather than an apnea trial. Heart failure with preserved ejection fraction and sleep apnea overlap heavily in obesity, and many people who qualified for the apnea trials would qualify here too. Semaglutide improved symptom scores by 7.8 points more than placebo across 529 patients, with 10.7 percentage points more weight loss. It suggests much of the cardiovascular benefit attributed to treating apnea may run through mechanisms these drugs address anyway.

PMID 37622681 ↗NCT04788511 ↗DOI 10.1056/NEJMoa2306963 ↗

The class-level pooling across roughly 900 patients from the major trials, giving a weighted average of about 11.6 fewer events per hour and 11.1% weight loss. Two subgroup findings matter: tirzepatide outperformed the rest, and people not already on CPAP responded more than those who were. That second point is easy to misread — it reflects who had room to improve, not that CPAP interferes — and it is the reference most often cited in coverage arguments.

PMID 39626095 ↗DOI 10.1093/sleep/zsae285 ↗