Not All Sleep Apnea Is the Same: Why OSA Phenotypes Matter for Treatment

By Dr. Boris Zusin · Published September 8, 2026

Your AHI (apnea-hypopnea index) is the number that gets the most attention after a sleep study, but two people with the exact same AHI can look nothing alike clinically — different symptoms, different heart-disease risk, and different odds of responding to any given treatment. (For patients whose sleep study looks essentially normal despite daytime symptoms, a related condition worth knowing about is upper airway resistance syndrome.) That mismatch is exactly why sleep researchers have moved toward describing obstructive sleep apnea (OSA) in terms of "phenotypes" rather than a single severity number. The goal isn't just academic: matching treatment to the specific traits driving an individual's disease, rather than treating every case of "moderate OSA" the same way, is the whole idea behind precision medicine applied to sleep apnea.

Clinical Phenotypes: Not Everyone With OSA Feels the Same

Studies that cluster patients by their symptoms — independent of their AHI — consistently find three recognizable groups. An excessively sleepy group has pronounced daytime sleepiness and gets the clearest symptomatic benefit from CPAP or an oral appliance, but this group also carries a notably higher cardiovascular risk (roughly 1.7 to 2.4 times higher risk of a future cardiovascular event) than other OSA patients with the same AHI — a striking finding, since it means the symptom pattern itself is flagging something the AHI number alone misses. A disturbed sleep group has prominent insomnia-type complaints and, in one well-studied cohort, still had residual insomnia symptoms even after effective airway treatment — suggesting the insomnia component often needs its own targeted treatment alongside, not instead of, OSA therapy. A third, minimally symptomatic group has few daytime complaints at all and tends to see less symptomatic benefit from treatment, at least by the usual measures of "success."

What Routine Sleep Studies Miss Beyond the AHI Number

Even within a standard sleep study, several other measurements turn out to matter more for long-term risk than the AHI itself: how much and how long oxygen levels drop (hypoxic burden and time spent below 90% oxygen saturation), how events are distributed between REM and non-REM sleep — a pattern we cover in more depth in our piece on REM-predominant sleep apnea — and whether apnea is markedly worse lying on the back (positional OSA, generally defined as at least twice the event rate on the back compared to other positions). This is part of why some large trials selecting patients purely by AHI have found disappointingly little cardiovascular benefit from treatment — AHI alone doesn't capture how much oxygen-level stress a given patient's disease is actually causing. Positional OSA in particular is a distinct, separately treatable pattern; see our page on sleep position and snoring/apnea for more on that specific approach.

Anatomic and Craniofacial Phenotypes: Where Dental Evaluation Comes In

This is the phenotype domain most directly relevant to what we evaluate at a dental sleep visit. Research consistently finds certain airway and jaw traits more often in OSA patients: a narrower pharyngeal airway, a hyoid bone (the small U-shaped bone that anchors tongue and airway muscles) sitting lower than typical, increased lower-face height, a retruded or underdeveloped lower jaw, and shorter jaw lengths overall. When researchers cluster patients by combining OSA severity, body weight, and these craniofacial traits, distinct groups emerge: an obesity-predominant type where excess weight is the main driver, a skeletal type (often a retruded upper and lower jaw with a steep, "high-angle" facial pattern) in patients who aren't obese at all, and a complex type combining both obesity and unfavorable jaw anatomy along with a displaced hyoid. These aren't just academic categories — they point toward different management priorities: weight-directed care for the obesity-predominant type, and orthodontic, oral appliance, or surgical approaches that specifically address jaw anatomy for the skeletal type.

The Physiologic "PALM" Traits

Underneath the clinical and anatomic patterns above, researchers also describe four physiologic traits that can combine in any individual patient and are sometimes called by the acronym PALM:

• Pharyngeal anatomy/collapsibility — how prone the airway is to collapsing on its own, independent of everything else (this is the trait a MAD and most dental therapy directly target).
• Arousal threshold — how easily a person wakes from a partial obstruction; a low threshold means waking up before the body has a chance to stabilize breathing on its own, which paradoxically can worsen events.
• Loop gain — how unstable a person's breathing control system is; high loop gain drives a pattern of overcorrection that perpetuates repeated events.
• Upper-airway muscle responsiveness — how well the tongue and throat muscles activate to keep the airway open during sleep.

These four traits aren't evenly distributed, and they're not rare. In the research that first characterized them carefully across a mixed group of OSA patients and controls, roughly 36% had minimal genioglossus (tongue) muscle responsiveness during sleep, 37% had a low arousal threshold, 36% had high loop gain, and 26% showed more than one of these nonanatomic traits together — meaning most people with OSA have at least one physiologic contributor beyond simple airway anatomy, and a meaningful share have more than one stacked on top of each other. The same research turned up a finding that cuts against the assumption that OSA is mostly about how narrow the airway looks on exam: about 1 in 5 patients had an airway no more collapsible than a typical person without OSA at all, and in that group, loop gain was nearly twice as high as in patients with the most collapsible airways — an unstable breathing control system, not airway anatomy, was doing most of the work. Airway collapsibility itself is measured with a value called Pcrit (pharyngeal critical closing pressure): a Pcrit below roughly −2 cmH₂O generally describes a fairly stable airway, a value near 0 describes a moderately collapsible one, and a Pcrit above roughly +2 cmH₂O describes a highly collapsible airway that needs more pressure or jaw advancement to stay open overnight. Excess fat deposited specifically at the base of the tongue is one of the better-documented contributors to a high Pcrit — MRI studies comparing obese patients with and without OSA have found significantly more fat at the tongue base in those with OSA, independent of overall body weight.

What “Poor Muscle Responsiveness” Actually Means, Mechanically

It's worth unpacking what's physically happening in that fourth PALM trait, because it explains a lot about why some people's airways hold up overnight and others' don't. The tongue and throat aren't held open by bone — the pharyngeal airway has almost no rigid support of its own, so its size depends almost entirely on the tone of the surrounding muscles. The genioglossus, the main muscle that pulls the tongue forward, is the best-studied of this group: during normal wakefulness, it fires in a very specific pattern — a burst of activity right before and during each inward breath, with a lower steady background level in between, timed by the same brainstem circuitry that drives the diaphragm. It also reacts, in real time, to the airway narrowing even slightly: pressure-sensing nerve endings in the throat detect that subtle drop in pressure and reflexively tell the genioglossus to contract harder, stiffening and widening the airway before a real problem develops.

At sleep onset, that background firing drops off — part of why airway resistance rises and snoring often starts the moment someone drifts off. In a healthy airway with normal anatomy, the muscle's responsiveness to rising carbon dioxide and to that same negative-pressure reflex is enough to claw activity back up and keep the airway open through the rest of the night. The problem in OSA isn't that this reflex disappears entirely — it's that, in a subset of patients, the muscle's response to those same signals is blunted enough that it can't fully compensate for an airway that's already anatomically tight, so the airway narrows or collapses outright until a brief arousal from sleep restores muscle tone and reopens it. That's the mechanical reason “poor muscle responsiveness” and “unfavorable anatomy” so often compound each other, and why therapies that target this trait directly — hypoglossal nerve stimulation, which electrically activates the genioglossus in sync with each breath, or, at a much lower-intensity level, oropharyngeal exercise training — work by strengthening or directly driving the same muscle system this research describes, rather than by changing airway anatomy itself.

These traits matter because they explain why treatments that have nothing to do with jaw position can still work well for the right patient — a sedative-type medication for a low arousal threshold, supplemental oxygen or a medication like acetazolamide for high loop gain, or hypoglossal nerve stimulation (an implanted device that directly stimulates the nerve controlling tongue muscle tone) for poor muscle responsiveness. Those are medical or surgical options managed by a sleep physician or ENT. For a milder version of that same muscle-responsiveness problem, a lower-risk starting point some patients try first is targeted tongue and throat exercises, though the evidence base there is smaller and mostly limited to snoring and milder OSA. Understanding that all of these options exist helps explain why "just try CPAP or a MAD" isn't always the full answer for every patient.

Tying It Back to Oral Appliance Candidacy

The anatomic and physiologic phenotypes above are exactly what underlies the predictors discussed on our who responds best to a MAD page. Two additional studies reinforce that same finding from a slightly different angle: identifying an obstruction pattern centered at the base of the tongue on drug-induced sleep endoscopy (DISE), rather than a full circular collapse at the level of the palate or a side-to-side collapse pattern, meaningfully improves how well clinicians can predict who will actually respond to an oral appliance before treatment even starts. This is a big part of why a thorough dental sleep evaluation looks at more than just your AHI number — the anatomic pattern behind that number is what actually determines whether a given approach is likely to work for you.

An Honest Limitation

Phenotyping is a genuinely useful way to think about OSA, but it isn't yet a finished, standardized system you can look up in a table. Different studies define their clusters somewhat differently, most of the underlying research still needs prospective validation (confirming that sorting patients by phenotype up front actually improves outcomes, not just that the patterns exist after the fact), and much of the existing data underrepresents women and non-white populations. In practice, that means phenotyping is best used the way we use it here: as a framework for understanding why your specific case might respond differently than a friend's or family member's with a similar-sounding diagnosis, not as a rigid formula for treatment.

If you've been told you have OSA and are trying to understand why your particular case is being approached the way it is — or why a treatment that worked for someone else didn't work for you — that's exactly the kind of conversation worth having at a sleep evaluation.