Can a GLP-1 Treat Sleep Apnea?

Obstructive sleep apnea has had essentially one durable treatment for four decades, and the long running problem with positive airway pressure has never been whether it opens the airway but whether the mask stays on past month three.

That is the context in which two randomized trials called SURMOUNT-OSA landed in June 2024, and in December of that year the FDA approved tirzepatide for moderate to severe obstructive sleep apnea in adults with obesity, making it the first prescription medication ever approved for the condition rather than for the symptoms around it.

What follows covers how the trials were built, how far the apnea hypopnea index moved, what the protocol meant by disease resolution, and the several things the design genuinely could not answer. This is educational content about published research and an approved indication rather than medical advice, and nobody should start, stop or change a prescribed therapy, including positive airway pressure, on the strength of an article.

The apnea hypopnea index is the average number of breathing interruptions recorded per hour of sleep during a formal sleep study, and it is the measure both trials were built around. An apnea is a near complete cessation of airflow and a hypopnea is a partial reduction, and the index counts both together.

The American Academy of Sleep Medicine grades severity in bands, where fewer than 5 events per hour is considered normal, 5 to under 15 is mild, 15 to under 30 is moderate, and 30 or more is severe. SURMOUNT-OSA required an index of at least 15 for entry, which restricted enrollment to the moderate and severe end of that scale.

This framing matters because a change of 20 events per hour means something different depending on where a participant started. Both trial populations entered with a mean index near 50 events per hour, deep in the severe band, so a reduction of that size moves a substantial share of people across category boundaries rather than merely improving a number within one.

The trials were split by whether participants were already using positive airway pressure, which is the design decision that makes the results interpretable.

Trial 1 enrolled 234 adults who were not using positive airway pressure and were not planning to start it, while Trial 2 enrolled 235 adults who were already established on the therapy and continued using it throughout, so tirzepatide was tested there as an addition to existing treatment rather than as a substitute for it. Both trials ran 52 weeks, both were double blind and placebo controlled, and both used tirzepatide titrated to a maximum tolerated dose of either 10 or 15 mg weekly. Entry required a BMI of at least 30 alongside the sleep criterion.

Running the two populations separately rather than pooling them answers a question a single mixed trial would have blurred, because someone already on positive airway pressure has their airway mechanically held open during sleep, which makes a drug effect measured on top of that a different quantity from one measured in an untreated airway.

Tirzepatide reduced the apnea hypopnea index by 20.0 more events per hour than placebo in Trial 1 and by 23.8 more events per hour in Trial 2, with both trials meeting their primary endpoint at P less than 0.001.

Those are placebo adjusted differences, which is the correct way to report a randomized result but not the most intuitive way to picture it. Measured against baseline instead, Trial 1 participants on tirzepatide saw a mean reduction of 55.0 percent in the index compared with 5.0 percent on placebo.

The categorical result translates most directly into something a person can picture. Up to 51.5 percent of tirzepatide treated participants met the protocol definition of disease resolution, meaning an apnea hypopnea index below 5 events per hour, or an index of 5 to 14 with an Epworth Sleepiness Scale score of 10 or lower. Starting from a mean near 50 events per hour, about half finished the year at a level that would not be diagnosed as sleep apnea at all.

The placebo arms provide the calibration that makes those figures readable, because a 5.0 percent reduction across a full year confirms the index does not simply drift downward on its own under trial monitoring, which is the alternative explanation a single arm study would have left wide open.

Weight loss is the most plausible mechanism and the trials cannot fully separate it from anything else, which is the honest position rather than a hedge.

Body weight fell 18.1 percent on tirzepatide versus 1.3 percent on placebo in Trial 1, with Trial 2 landing close to 20 percent, so the magnitude sits in the same range the earlier SURMOUNT trials had already established for this compound at these doses. The mechanical link between that and the airway is well described, because fat deposition in the parapharyngeal tissues narrows the airway directly while abdominal adiposity reduces lung volume and the caudal traction that helps hold the upper airway open during sleep. Removing close to a fifth of body weight plausibly acts on all of those at once.

Whether tirzepatide does anything to the airway beyond what its weight loss accomplishes is a separate question that SURMOUNT-OSA was not built to answer, because establishing a weight independent effect would require either a comparator matched for weight loss achieved by other means or a powered mediation analysis, and neither was part of the design.

This distinction has a practical edge rather than being merely academic, because if the benefit runs through weight then the withdrawal data from the broader GLP-1 literature becomes directly relevant, and weight regain after discontinuation has been documented consistently enough that the airway would be unlikely to escape following the weight back up.

Several secondary endpoints shifted in the same direction, which is what gives the primary result its internal coherence.

Hypoxic burden fell, and that matters because it captures the depth and duration of oxygen desaturation rather than only the count of events, making it the dimension most closely tied to cardiovascular strain. Systolic blood pressure fell as well, and hsCRP, a marker of systemic inflammation, declined in line with what the weight loss trials for this drug class have consistently reported.

Sleep related patient reported outcomes also improved, which is not a trivial addition given that the gap between a sleep study number and how someone actually feels during the day is the usual complaint about apnea treatment.

Positive airway pressure remains the standard of care, and SURMOUNT-OSA does not challenge that because it never tested the comparison.

No arm of either trial compared tirzepatide against positive airway pressure, and Trial 2 was explicitly designed with participants continuing their therapy throughout, so the only comparative question the data touches is whether adding the drug to existing treatment produces further improvement, which it did by 23.8 events per hour against placebo.

What the trials genuinely establish is that a pharmacological option now exists for a condition that previously had only a mechanical one, and that it works in people who are not using a machine as well as in people who are. How that fits alongside existing therapy for any individual is a clinical decision involving a sleep physician and repeat testing, given that an untreated airway during the interval is not a neutral state.

The approval language is narrow in a way that deserves noting, because it covers moderate to severe obstructive sleep apnea in adults with obesity and therefore does not extend to apnea at a normal BMI, to mild apnea, or to central sleep apnea, which has a different mechanism entirely.

SURMOUNT-OSA is a well built pair of trials that answered a narrow question clearly, showing that in adults with obesity and moderate to severe obstructive sleep apnea, 52 weeks of tirzepatide reduced measured breathing events by around 20 to 24 more per hour than placebo, with roughly half of treated participants finishing at a level that would no longer be diagnosed as apnea.

The parts left open deserve equal firmness, because the trials did not compare the drug against positive airway pressure, did not establish whether anything beyond weight loss is doing the work, did not measure cardiovascular events, and did not follow anyone past the treatment year.

Obstructive sleep apnea is a diagnosed medical condition with real cardiovascular consequences when untreated, and it is managed with sleep studies and a clinician rather than privately. What changed in December 2024 is that the list of options got longer, which is a genuinely different thing from any single option becoming unnecessary.