OSA Endotyping as Targets for Treatment. How Far Have We Come?

ABSTRACT

Obstructive sleep apnoea (OSA) is highly prevalent and is associated with significant impairments in quality of life and comorbidity. The most common treatment employed is continuous positive airway pressure (CPAP), especially in moderate and severe cases. The fundamental pathophysiology of the disorder relates to increased collapsibility of the oropharyngeal airway, but the growing recognition that multiple endotypic factors contribute to this collapsibility has opened the possibility of alternative treatment possibilities, especially in less severe cases. In addition to the most important endotype of upper airway narrowing, additional non‐anatomical endotypes include upper airway muscle responsiveness, loop gain, and the arousal threshold. Recent reports have identified mechanisms to identify these endotypes, including those suitable for application in clinical practice. These developments have encouraged a focus on non‐CPAP therapies, including oral appliances, pharmacotherapy, and neurostimulation. Recent reports have identified effective pharmacotherapies directed at each non‐anatomical endotype. While CPAP is likely to remain the most important and widely used therapy for moderate and severe OSA in the foreseeable future, alternative therapies targeting specific endotypes, especially pharmacotherapy, are likely to play an increasing role in less severe cases in the medium term, and in patients who do not comply with CPAP.

1. Introduction

Obstructive sleep apnoea (OSA) is highly prevalent with population estimates of close to one billion adults globally having at least mild OSA based on an apnoea‐hypopnoea frequency per hour (AHI) greater than 5 [1]. However, only a minority of these have a clinically significant syndrome where an elevated AHI is linked to relevant clinical features and active treatment is indicated [2]. The most common treatment for symptomatic OSA is nasal continuous positive airway pressure (CPAP), which is highly effective in preventing upper airway obstruction in such patients, but is cumbersome and often poorly tolerated with consequent poor adherence [3]. These considerations have stimulated interest in non‐CPAP therapies for OSA, which cover a broad range of options including alternative devices such as mandibular advancement devices, surgical options, and more recently, a growing interest in pharmacological therapy [4].

As alternatives to CPAP become broader based, there is growing recognition of the importance of patient‐related traits that help identify the optimum treatment for an individual patient. These include clinical phenotypes such as obesity and excessive daytime sleepiness (EDS) and pathophysiological endotypes such as upper airway narrowing and disordered ventilatory control [5, 6]. These developments have resulted in a growing number of reports about the identification of different endotypes from sleep studies, which is important to the application of these considerations to clinical practice. While CPAP remains the most important treatment for patients whose principal pathophysiological mechanism relates to upper airway narrowing, a growing number of pharmacological approaches have been reported to manage OSA from the perspective of respiratory control and upper airway muscle function [7]. This review will provide an overview of the pathophysiological mechanisms underpinning different OSA endotypes, strategies to identify individual endotypes from sleep studies with a special emphasis on applicability to clinical practice, in addition to a review of management strategies, especially developments in pharmacotherapy targeting different endotypes.

2. Pathophysiology of OSA Relating to Endotype Characteristics

The fundamental pathophysiology of OSA relates to an imbalance between the upper airway collapsing forces during inspiration and the counteracting forces of upper airway dilating muscles that act to keep the upper airway open [8]. A narrowed oropharynx is very common in patients with OSA, which increases the collapsing pressure during inspiration. Non‐anatomical factors that contribute to upper airway collapsibility during sleep include diminished oropharyngeal muscle responsiveness, heightened sensitivity to arousals, and a high ventilatory control system gain, called loop gain. Recognition of these individual anatomical and non‐anatomical factors represents a key objective in selecting the optimum choice of therapy for each individual patient.

2.1. Upper Airway Narrowing

Most OSA patients demonstrate narrowing of the oropharyngeal airway, which can be clinically assessed by the Mallampati score [9], and is at least partly a result of genetic factors [10]. Contributing bony factors include micro/retrognathia [11], which predispose to retro positioning of the tongue into the oropharyngeal space. Additionally, fat accumulation in the neck due to obesity, or inside the oropharyngeal lumen in the case of adenotonsillar hypertrophy, further narrow the oropharyngeal airway and increase collapsibility [8]. Heart failure and end‐stage renal failure are associated with fluid retention, which tends to gravitate towards the neck region in the supine position during sleep and may further contribute to upper airway narrowing and increase collapsibility [12, 13]. Upper airway dimensions may also be reduced by gravitational forces in the supine position [14], which is an important factor in the common observation that AHI is often highest in this position. Although commonly considered an important contributing factor by patients with snoring and OSA, nasal obstruction is a relatively unimportant factor in OSA pathophysiology. However, variable nasal obstruction, such as seen in rhinitis, may be a more significant factor [15], and a significant reduction in snoring and AHI has been reported in patients with OSA and rhinitis treated with intranasal corticosteroid sprays [16].

2.2. Upper Airway Dilator Muscle Responsiveness

A narrowed upper airway in OSA results in increased collapsing forces during inspiration, which require increased contraction of the pharyngeal dilator muscles, especially the genioglossus, which stiffen the collapsible segment of the upper airway during inspiration to resist occlusion [8]. The phasic contraction of these muscles is coordinated with inspiration and precedes contraction of the diaphragm by milliseconds [17]. Contractility of these dilating muscles is influenced by numerous factors that include vagal inputs, chemical stimuli, and baroreceptor activity [18]. Oropharyngeal dilator muscle contraction in patients with OSA is stronger than in normal subjects while awake, but reduces more during sleep, thus predisposing to occlusion [19], especially during REM sleep [20]. These findings indicate that an insufficient response by these muscles to the increased upper airway negative pressure during inspiration while asleep represents the key factor predisposing to occlusion rather than a primary abnormality in muscle function. A further contributing factor is that the upper airway muscles are skeletal muscles, which show greater reductions in contractility during REM sleep compared to the diaphragm, thus further predisposing to oropharyngeal occlusion during this sleep stage.

2.3. Loop Gain

Oropharyngeal dilator muscle contractility may be adversely affected by disturbances in respiratory control. A common observation in patients with OSA is that apnoeas/hypopnoeas occur in clusters, which likely reflects an instability in respiratory control where the hyperventilation that typically follows apnoea termination is followed by a progressive reduction in ventilation and muscle contractility towards an apnoea threshold that predisposes to further oropharyngeal occlusion [8, 21]. Loop gain, which measures the sensitivity of the feedback loop modifying ventilatory responses to respiratory disturbances, represents an important predisposing factor to respiratory control instability. Loop gain predisposes to obstructive apnoea/hypopnoea when a high loop gain, which is associated with a larger increase in ventilation following apnoea termination, increases the degree of ventilatory instability with a greater subsequent reduction in ventilation and dilating muscle contraction, thus increasing the susceptibility to recurring oropharyngeal collapse and associated obstructive apnoea [21].

2.4. Arousal

The termination of an obstructive apnoea/hypopnoea is typically associated with cortical mini arousal, which represents a critically important factor in the restoration of breathing. However, this arousal may also contribute to hyperventilation following apnoea termination, thus increasing ventilatory instability and predisposing to further obstructive apnoea [22, 23, 24], and thereby representing a distinct, potentially treatable, pathophysiologic feature. Apnoea‐associated cortical arousals vary in degree [25], which may be quantified by the arousal threshold and can be assessed noninvasively by polysomnography (PSG) [26]. A low arousal threshold may contribute to recurring apnoeas and represents a potential therapeutic target in selected patients [27].

2.5. Integrated Pathophysiology and Implications for Treatment

The recognition of an important role for non‐anatomical factors in the pathophysiology of OSA has expanded the perspectives for precision therapy of the disorder [28]. While individual endotypes may be dominant, many patients with OSA demonstrate a combination of multiple traits [29], which implies that a single treatment choice may not be optimal. Although the most important manifestation of increased UA collapsibility in OSA is reversed by CPAP therapy, a more detailed insight into the underlying pathophysiology may offer additional treatment options for selected patients (Figure 1) [30]. Identification of individual pathophysiological traits may offer the additional treatment options of targeting inadequate oropharyngeal muscle compensation by drug therapy or hypoglossal nerve stimulation [31, 32], a high loop gain by carbonic anhydrase inhibition [33], and/or a low arousal threshold by selected sedative/hypnotic medication [34]. These additional therapies may have a role in selected patients where such factors are found to be contributing factors, alone or in combination, in the integrated pathophysiology of the disorder.

FIGURE 1.

Balance of forces contributing to upper airway obstruction in OSA and the potential for endotype‐based treatment. Red‐shaded boxes contain factors contributing to obstruction and green‐shaded boxes contain therapeutic possibilities directed at specific endotypes. CPAP, continuous positive airway pressure; MAD, mandibular advancement device; UA, upper airway.

3. Methods for Pathophysiological Endotyping

Pathophysiological traits, including upper airway muscle compensation, loop gain, and arousal threshold, can be measured using gold standard techniques, clinical standard techniques, or noninvasive measurement techniques (Table 1). Assessing the different endotypes can help define an individual patient's OSA profile. Here we will further describe these techniques. Gold‐standard measurements of endotypic traits are based on manipulation of CPAP during sleep to assess its effects on the upper airway and breathing. These procedures are mainly performed in academic physiology laboratories as widespread use of these techniques is hampered by the requirement of specialized equipment, invasiveness of the technique, and the need for skilled personnel overnight. More simplified approaches are based on the premise that the naturally occurring apnoeas and hypopnoeas in patients with obstructive sleep apnoea allow a similar approach.

TABLE 1.

Gold standard technique, clinical standard measurement techniques and non‐invasive measurement techniques for endotype evaluation.

	Upper airway collapsibility	Ventilatory control stability (loop gain)	Pharyngeal muscle responsiveness	Arousal threshold
Gold standard technique	Pcrit Awake upper airway collapsibility index determination (UACI)	Airflow pressure drops	Airflow pressure drops	Airflow pressure drops
Clinical measurement techniques	Clinical score combining NREM‐OAI/AHI, waist circumference, obstructive apnoea duration and REM‐AHI (allows to predict a Pcrit > 2.5 cm H2O). Therapeutic CPAP level as a surrogate for collapsibility Airflow pressure drops	Breath‐hold duration	No clinical standard.	Combination of AHI, oxygen saturation and the fraction of hypopnoeas (allows estimation of the chance of having a low arousal threshold)
Non‐invasive measurement techniques	PSG Airflow analysis	PSG Airflow analysis	PSG Airflow analysis	Baseline PSG analysis

3.1. Endotyping Using Gold Standard Techniques

The gold‐standard method to determine the different (patho)physiological traits uses abrupt or gradual lowering of the air pressure in the upper airways. In practice, this is performed with a modified CPAP device that allows quick switching between various air pressures. Endotypes are measured by lowering the air pressure from the holding pressure, defined as the CPAP level at which the upper airways remain open during sleep [35].

3.1.1. Upper Airway Collapsibility

Using the gold standard technique, upper airway collapsibility is expressed by the critical closing pressure (P_crit). In humans, P_crit can be assessed by studying flow/pressure changes in the upper airways (Table 1). Changes in maximal inspiratory flow are compared against the pressure at the level of the nose (PN), controlled by using a modified CPAP device (which can generate negative pressure) (Figure 2). The pressure is first set to the holding pressure, which is defined as the pressure at which all apnoeas, hypopnoeas and respiratory effort related arousals (RERAs) are abolished. This equals the “effective CPAP level” or “upper airway opening pressure” [36, 37]. Next, the pressure at which to create a reproducible apnoea (Flow = 0) is determined [38]. By comparing a range of flow measurements against the corresponding pressure in the mask, P_crit can be determined as the zero‐flow intercept on the X‐axis using regression analysis. Active as well as passive P_crit can be measured. Active P_crit is the critical closing pressure at which muscular response can be observed. Passive P_crit is the critical closing pressure in the absence of muscular activation. To determine the active P_crit, the pressure in the upper airway is gradually decreased, allowing activation of the muscles in the upper airways. The passive P_crit is determined by an abrupt lowering of the pressure, which prevents activation of the upper airway muscles [35, 39]. At each pressure level, flow is determined. After each pressure drop, the pressure is set back to the initial holding pressure. These steps are repeated until no stable breathing can be achieved [35, 39].

FIGURE 2.

P_crit determination by the airflow pressure drop method. A final P_crit value of—4 cmH₂O was obtained. P_crit, critical closing pressure; PNINSP, pressure at the level of the nose; ViMAX, maximal inspiratory flow.

Alternatively, upper airway collapsibility can be assessed using ventilatory parameters, assessing ventilation as a function of ventilatory drive. The minimal ventilation at normal ventilatory drive is called passive ventilation (V_passive). V_passive is defined as the ventilation at which no additional muscles are recruited, and thus reflects the inherent collapsibility of the upper airway during sleep. On the other hand, active ventilation (V_active) is defined as ventilation at maximal ventilatory drive just preceding arousal, and thus takes into account upper airway muscle activity [39, 40]. V_passive is determined by lowering the pressure to 0 cm H₂O during 5 breaths. V_active is measured at the minimal CPAP level at which no arousals occur [39].

3.1.2. Loop Gain

Using the gold standard technique, loop gain is defined by assessing the ratio between the ventilatory disturbance and ventilatory response as a response to airflow pressure drops (Table 1). When the CPAP mask level in the upper airways is abruptly lowered from the holding pressure, the breathing rate will also drop abruptly. As a response, the upper airways will stiffen and the respiratory effort will increase, reaching a new level of steady‐state breathing. After a small delay, the upper airway muscles will be recruited to increase airflow in order to restore ventilation. During this sub‐optimal new steady‐state, CO₂ will accumulate, which will finally result in an increase in ventilatory drive. Ventilatory disturbance is defined as the difference in steady‐state breathing between optimal and lower CPAP level. If the CPAP level is subsequently increased back to the holding pressure, a ventilatory response occurs which exceeds the initial breathing level. The ratio of this overshoot in ventilation and the ventilatory disturbance is called the loop gain [35].

3.1.3. Muscle Responsiveness

The gold standard technique for measuring pharyngeal muscle responsiveness involves the use of intramuscular EMG electrodes during CPAP level drops, starting from the holding pressure to suboptimal pressures (Table 1). This results in a drop in ventilation, causing an increase in ventilatory drive (resulting from an increase in pharyngeal muscle activity). Muscle responsiveness can be defined as the response to the intraluminal pressure associated with lowering the applied pressure from the holding pressure [41].

An alternative measurement method consists of using ventilation measures during airflow pressure drops, where muscle responsiveness is measured via the resulting compensation. Compensation is defined as increased ventilation caused by a stiffening of the upper airway. Compensation can be determined noninvasively by lowering the CPAP level. When an abrupt drop in pressure is imposed, the breathing rate will also drop abruptly, after which a lower steady‐state breathing will be attained. The difference between ventilation immediately after the abrupt lowering of the pressure and the new, lower steady‐state ventilation is what is called compensation. The higher this compensation, the better a patient can ‘eliminate’ an imposed disturbance and the smaller the difference between the new steady‐state ventilation under lower pressure and ventilation under optimal CPAP [40].

3.1.4. Arousal Threshold

Similar to the techniques described earlier, the gold standard technique to determine the arousal threshold includes pressure drops imposed on the patient (Table 1). To assess the arousal threshold, adequate ventilatory drive measurements are mandatory. This can be obtained by the assessment of epiglottic pressure, diaphragm EMG or oesophageal pressures that reflect increasing or decreasing ventilatory drive. Another approach is to mathematically model respiratory effort (drive) using parameters derived from the individual ventilation signal. When an abrupt, temporary CPAP level drop is imposed on the patient, the loop gain can be determined using this method. After the overshooting ventilatory response, ventilation will gradually restore to the normal level as it was before the pressure was lowered. The delay and time constant associated with this recovery allow mathematical modelling and calculation of ventilatory drive before arousal, defined as the arousal threshold [35].

3.2. Clinically Available Measurement Techniques

Recent research shows promising results regarding potential clinical parameters to measure upper airway collapsibility, loop gain and the arousal threshold. Currently, no clinically available techniques to measure muscle responsiveness are available.

3.2.1. Upper Airway Collapsibility

Upper airway collapsibility can also be assessed during wakefulness with the upper airway collapsibility index (UACI) (Table 1). The UACI is obtained by applying brief (250 ms) pulses of negative airway pressure (−12 cm H₂O) during early inspiration, and is then calculated as the pressure difference between the choanae and the epiglottis during the pulse [42, 43]. Significant correlations were found with P_crit [43]. While this technique shows great potential for application in clinical routine, the invasiveness (including catheter insertion) hampers its applicability. Another measure with great promise is the therapeutic CPAP level as a surrogate for collapsibility. Specifically, a therapeutic CPAP level of ≤ 8 cm H₂O predicts a mild collapsibility of the upper airway [44]. Genta et al. developed a clinical score to distinguish between male patients with high and low collapsibility. In this method the ratio of the obstructive apnoea index during NREM sleep over AHI (NREM‐OAI/AHI), waist circumference, obstructive apnoea duration and REM‐AHI are combined, enabling to predict a P_crit > 2.5 cm H₂O [45], a level associated with OSA where major anatomic or mechanical intervention, like CPAP therapy, is likely required [29].

3.2.2. Loop Gain

Currently, loop gain is not measured in clinical practice. However, a breath‐holding technique, developed by Messineo et al. shows promise for future clinical application (Table 1). In this research, breath‐holding manoeuvres during wakefulness were associated with loop gain [46]. Patients with a higher loop gain showed a shorter maximal breath‐hold duration and a larger ventilatory response to 20 s breath‐holds.

3.2.3. Arousal Threshold

Edwards et al. demonstrated that the combination of AHI, oxygen saturation and the fraction of hypopnoeas could estimate the chance of a patient presenting with a low arousal threshold (< −15 cmH₂O epiglottic pressure) [47]. However, this technique is not precise (Table 1).

3.3. Endotyping Using Diagnostic Polysomnography Signals

To make the endotyping accessible in general clinical practice, an algorithm was recently developed where the signals, as measured during a standard PSG, can be used to mathematically calculate the pathophysiological traits [40, 48]. This approach is also called the endo‐Phenotyping Using Polysomnography (PUP) method. It does not require additional, invasive measurements and relies on the premise that the naturally occurring respiratory events will cause similar disturbances as the imposed disturbances using CPAP level drops.

3.3.1. Upper Airway Collapsibility

Ventilation parameters obtained during routine PSG can also be used to define collapsibility (Table 1). Recently, several respiratory parameters were evaluated by Vena et al. [49]. The most commonly used are V_passive and V_active [35, 40]. V_passive (passive collapsibility) is defined as the minimal ventilation at normal respiratory effort. Normal respiratory effort or ventilatory drive is defined as the period at which no additional upper airway muscles are activated. V_active can be analogously defined as the breathing with maximal respiratory effort. Maximal respiratory effort or ventilatory drive is achieved when the upper airway muscle are maximally activated. This happens at the end of a respiratory event at the waking threshold, just prior to an arousal. Furthermore, V_min (ventilation at nadir drive) is strongly correlated to P_crit measurements.

3.3.2. Loop Gain

Loop gain is estimated by creating a mathematical model for the ventilatory control system (Table 1). Here, respiratory effort is defined as the sum of the chemical respiratory efforts, the respiratory effort associated with the chemical control system, and the respiratory effort associated with arousal [48].

An obstruction of the upper airway causes a ventilatory disturbance and consequently lowers ventilation. The ventilatory disturbance causes a disturbance in the chemical control system, through a drop in PaO₂. This chemical disturbance causes the respiratory effort to increase, according to a mathematical comparison that is defined by the delay and time constant as determined by the decrease in ventilation after a sudden rise caused by the reopening of the upper airways. Loop gain is defined as the rise in chemical respiratory effort divided by the drop in ventilation [48].

3.3.3. Muscle Responsiveness

Using polysomnography parameters, muscle responsiveness is defined as the difference between passive ventilation (V_passive) and ventilation under active conditions (V_active) (Table 1).

3.3.4. Arousal Threshold

Finally, the arousal threshold is measured by determining the respiratory effort shortly before a scored arousal, as visible in the EEG signal (Table 1). The respiratory effort is mathematically modelled as discussed earlier [26, 48].

3.3.5. Endogram

The concomitant assessment of V_passive, V_active, muscle responsiveness and arousal threshold can be visualized on a patient endogram (Figure 3), in which the ventilatory drive is plotted against actual ventilation during sleep for the whole night. This endogram shows the interplay between the different endotypes and the potential effect of treatment and was first established by Wellman et al. [35, 39].

FIGURE 3.

Endogram where the ventilatory drive is plotted against actual ventilation during sleep for the whole night showing the interplay between the different endotypes and the potential effect of treatment. Black dots and lines represent baseline values. Grey dots and lines represent treatment effect. The slope of the line connecting Veupnoea and Varousal is 1/loop gain. The slope of the line connecting V_passive and V_active is defined as the upper airway gain. Stable breathing occurs at the intersection between both lines. However, this can only occur if this point is achieved prior to (at the left of) the arousal threshold. OSA treatment (grey) aims to modify the different endotypes in order to achieve stable breathing at a lower ventilatory drive than (to the left of) the arousal threshold. Based on Wellman et al. [35, 39]. V_active, active ventilation, inversely related to active collapsibility; Varousal: ventilation tolerated without arousal; Veupnoea: ventilation at eupnoea; V_passive: passive ventilation, inversely related to passive collapsibility.

Eupnoea or resting breathing occurs when the upper airway is patent. At this point, ventilation matches ventilatory drive. Arousal threshold (dotted line) is the ventilatory drive at which the patient arouses. Varousal is the minimal ventilation that can be tolerated without arousal. Loop gain is defined as the reciprocal of the slope connecting Veupnoea and Varousal. V_passive (passive collapsibility) is the minimal ventilation that can be achieved without muscle activation. V_active (active collapsibility), is defined as the ventilation at maximal ventilatory drive that can be achieved through muscle activation (upper airway gain). Stable breathing is achieved if both lines intersect at a ventilatory drive which is lower than (to the left of) the arousal threshold.

OSA treatment (grey) aims to modify one or more of the traits (loop gain, upper airway muscle compensation, collapsibility) to allow stable breathing at a ventilatory drive below the arousal threshold [35, 39].

Such an endogram, has the potential in the near future to become a useful tool for Somnologists to assess OSA patients and support therapeutic decisions.

4. Management Considerations Based on Endotyping

4.1. The Endotypic Paradigm in OSA Management

The 2021 ERS guideline on non‐CPAP therapies [4] firmly established that OSA should no longer be approached as a uniform disorder corrected by a single mechanical therapy, that is, CPAP. Rather, its management should align with the underlying pathophysiological endotypes, defined as distinct, measurable mechanisms that predispose the upper airway to collapse during sleep. CPAP remains the reference treatment when anatomical airway narrowing predominates, but non‐anatomical factors such as impaired pharyngeal muscle responsiveness, unstable ventilatory control, and a low arousal threshold, frequently coexist and may dominate in many patients [5, 29]. These traits, individually or in combination, explain the heterogeneity of the clinical expression and the therapeutic response that frustrates conventional algorithms.

In the ERS guideline framework, management becomes a process of matching treatment to trait. When the structural load is overwhelming, mechanical or surgical correction is rational; when neurochemical or ventilatory control abnormalities are key, targeted pharmacologic or neurostimulatory modulation appears more appropriate. This approach moves beyond the binary choice of “CPAP user versus non‐user” towards precision physiology, in which each therapy is selected because it corrects the dominant failure mode underlying obstruction [5, 7].

4.2. Anatomical and Neuromuscular Endotypes

Anatomical predisposition remains central in many patients, but endotyping helps refine management. CPAP and mandibular advancement devices (MADs) [4, 5, 6] remain first‐line therapy when collapse is purely structural. However, recent work has demonstrated that the efficacy of MAD therapy is strongly modulated by non‐anatomical physiological traits. Bamagoos and colleagues [50] demonstrated that polysomnography‐derived endotyping can identify a subgroup of patients with greater MAD efficacy, characterized by lower loop gain, higher arousal threshold, lower ventilatory response to arousal, and moderate (non‐severe, non‐mild) pharyngeal collapsibility, independent of baseline AHI or body mass index. In addition, Op de Beeck and colleagues [51] further showed that elevated loop gain, reflecting increased ventilatory control instability, is an independent predictor of nonresponse to MAD therapy, even after adjustment for baseline AHI, body mass index, and collapsibility. Conversely, individuals with moderate collapsibility and stable ventilatory control (low loop gain) experience the most pronounced reductions in AHI. These findings place MAD therapy within an “anatomical‐dominant but physiology‐sensitive” framework: its mechanical protrusive action reduces pharyngeal collapsibility, yet a successful response depends on the integrity of ventilatory control mechanisms. In this regard, MAD represents a transitional intervention bridging purely structural correction and functional modulation, offering a precision‐guided alternative to CPAP in selected endotypes.

Similarly, positional therapy (PT) provides a simple yet mechanistically rational intervention for patients with supine‐predominant OSA. Compared with non‐positional OSA, the supine‐predominant phenotype appears characterized by lower collapsibility but impaired upper‐airway muscle compensation and a reduced arousal threshold when sleeping supine [52]. Inadequate recruitment of the genioglossus and related dilator muscles thus emerges as the key endotypic deficit in this group. By preventing the supine posture, PT effectively eliminates the mechanical trigger for instability and reduces AHI by 50%–60% in responders, with adherence rates comparable to CPAP [53]. Integration of PT within an endotype‐guided framework allows identification of patients most likely to benefit—those with mild‐to‐moderate collapsibility, low loop gain, and supine‐dependent events.

Surgical options further address anatomical predisposition. Palatal and multilevel upper‐airway surgeries (e.g., uvulopalatopharyngoplasty, expansion sphincter pharyngoplasty, tonsillectomy) reduce localized collapsibility, whereas maxillomandibular advancement expands the skeletal framework and lowers critical closing pressure. These procedures primarily target the extreme anatomical endotype, where structural compromise predominates [54].

Hypoglossal nerve stimulation (HNS), on the other hand, is an example of the anatomical‐neuromuscular continuum. By synchronously activating the genioglossus, HNS restores physiologic upper‐airway muscle tone during inspiration. Clinical observations indicate that HNS is most effective in patients with moderate airway collapsibility and preserved neuromuscular function, whereas high collapsibility or limited muscle responsiveness are associated with reduced benefit [55, 56]. In a large endotypic analysis of the STAR trial, Op de Beeck and colleagues demonstrated that favourable HNS outcomes are independently associated with non‐anatomical physiological traits, with higher arousal threshold emerging as the principal determinant after correction for other traits, together with greater pharyngeal muscle compensation and lower loop gain [55]. Thus, endotyping guides patient selection for neurostimulation and clarifies non‐response mechanisms. Nevertheless, the clinical application of HNS remains constrained by several factors. Although eligibility criteria have expanded in recent years, HNS is still primarily offered to carefully selected patients, typically those with moderate‐to‐severe OSA, emphasizing appropriate upper‐airway anatomy and the absence of unfavourable collapse patterns on drug‐induced sleep endoscopy [56]. The procedure requires surgical implantation and subsequent device titration, entailing substantial cost, specialized expertise, and the potential for procedural complications. Moreover, stimulation is unilateral in most systems, and treatment response remains variable despite phenotypic selection. Long‐term outcomes are encouraging but the absence of standardized physiological predictors or titration algorithms limits widespread clinical adoption.

Progress in the pharmacologic activation of the same motor pathways as in HNS has transformed the field. The work of Taranto‐Montemurro and colleagues [31, 57, 58] demonstrated that the combination of the noradrenergic reuptake inhibitor atomoxetine and the antimuscarinic oxybutynin can restore airway patency by enhancing hypoglossal motoneuron excitability. Average AHI reductions of 60%–70% were achieved in proof‐of‐concept studies [31], effects reproduced with alternative combinations such as reboxetine plus oxybutynin [59]. These dual agents act by augmenting the wake‐related noradrenergic drive while blocking REM‐related cholinergic inhibition, effectively reinstating the neural tone normally withdrawn in sleep. Cardiovascular monitoring confirmed improved baroreflex sensitivity and no sympathetic overstimulation [59]. Furthermore, differential responses aligned with physiological expectations in that patients with moderate collapsibility and intact ventilatory control derived the greatest benefit, whereas those with very high loop gain or severe structural compromise did not [60, 61].

Collectively, these findings validate the concept that impaired upper‐airway muscle responsiveness is a treatable trait. Pharmacologic or neurostimulatory activation of dilator muscles now represents a legitimate alternative to CPAP for a defined endotype rather than a general substitute for non‐adherence [5, 7].

4.3. Ventilatory Control and Arousal Endotypes

A second group of patients exhibits OSA primarily driven by ventilatory instability. High loop gain, an exaggerated sensitivity of ventilatory feedback to CO₂ fluctuations, causes cyclical over‐ and under‐ventilation, predisposing to recurrent collapse [29]. In many individuals, elevated loop gain coexists with a low arousal threshold, creating a self‐perpetuating cycle of respiratory overshoot, arousal, and renewed instability. Pharmacologic stabilization of loop gain has become a credible strategy. Carbonic anhydrase inhibitors such as acetazolamide reduce loop gain by inducing mild metabolic acidosis, increasing baseline drive and widening the CO₂ reserve. Trials summarized by Hedner and Zou [62] and Baillieul et al. [63] demonstrate consistent AHI reductions of approximately 30%–40%, though chronic tolerance and paresthesia limit long‐term use.

Supplemental oxygen represents an alternative physiological means to dampen ventilatory instability. By reducing hypoxic chemosensitivity and the magnitude of CO₂ oscillations, oxygen lowers loop gain and stabilizes breathing. Supplemental nocturnal oxygen has been shown to reduce ventilatory instability by lowering loop gain, thereby improving obstructive events in selected patients. Wellman et al. [64] demonstrated that oxygen supplementation significantly decreases loop gain and reduces the propensity for periodic breathing in OSA, while subsequent analyses and systematic reviews have confirmed modest reductions in AHI in specific physiological phenotypes [65]. Combination therapy, oxygen with acetazolamide or sedatives, has been proposed to achieve additive stabilization of ventilatory control [63]. However, oxygen therapy does not alleviate anatomical obstruction and its effects on AHI and symptoms are modest and phenotype‐dependent. Moreover, prolonged use may blunt chemoreflex drive or cause CO₂ retention in susceptible individuals, and no consistent cardiovascular benefit has been demonstrated [66].

A newer compound, sultiame, provides a more selective and tolerable means of carbonic anhydrase inhibition. A large multicentre trial [67] reported placebo‐adjusted AHI reductions up to 35% over 15 weeks, with proportional improvements in oxygen desaturation and minimal side effects. Importantly, benefit correlated with baseline loop‐gain estimates, confirming endotype specificity. Physiological investigations suggest that both acetazolamide and sultiame not only stabilize ventilation, but also enhance pharyngeal motor recruitment by maintaining CO₂‐driven excitatory input [68]. Specifically, the FLOW trial [67], the largest double‐blind, placebo‐controlled, dose‐finding study in moderate‐to‐severe OSA demonstrated that sultiame produced dose‐dependent reductions of AHI (placebo‐adjusted −30% to −40% at 200–300 mg) and significant improvements in oxygen desaturation, sleep continuity, and excessive daytime sleepiness. In addition, sultiame reduced the arousal index and slightly lowered systolic blood pressure without cardiovascular side‐effects. These data provide robust clinical evidence that carbonic anhydrase inhibition can modify multiple pathophysiological traits, that is, loop gain, hypoxic burden, and sleep fragmentation, thus representing a truly mechanism‐targeted pharmacological approach in OSA.

Closely related is the low arousal threshold endotype, in which premature awakenings interrupt respiratory events before upper airway dilator muscles can come into action. Contrary to early safety concerns, controlled sedation can be advantageous in these patients. Studies reviewed by Carter and Eckert [69] show that eszopiclone and zolpidem increase arousal threshold without worsening obstruction, improving sleep efficiency and ventilatory stability. When combined with agents that enhance pharyngeal tone, such as atomoxetine‐oxybutynin, sedatives may exert synergistic effects [63, 70]. In this setting, hypnotics cease to be contraindicated and become mechanistically rational tools within an endotype‐guided plan.

4.4. Metabolic and Composite Endotypes

Many patients present with overlapping traits in which obesity, ventilatory control, and neuromuscular responsiveness interact. Obesity increases anatomical load but also alters chemosensitivity and arousability through inflammatory and neuroendocrine mechanisms, generating a mixed metabolic–anatomical endotype [29, 71]. Adipose tissue distribution and leptin resistance modulate ventilatory drive and upper‐airway muscle tone, explaining why obese OSA often display combined metabolic, ventilatory, and neuromuscular traits rather than a single dominant mechanism.

The growing evidence that weight loss itself can improve upper‐airway mechanics has now been extended by pharmacologic innovation. The dual GIP/GLP‐1 receptor agonist tirzepatide produced a 20% mean reduction in body weight and a 55% reduction in AHI in the SURMOUNT‐OSA trials [72]. These findings establish weight‐loss pharmacotherapy as an effective disease‐modifying intervention for obese OSA, yet the persistence of residual events in roughly half of participants indicates that isolated anatomical relief is insufficient. Beyond mechanical unloading, metabolic therapy attenuates chemosensitivity and loop gain, thereby improving ventilatory stability and reducing hypoxic burden [33, 73]. Future strategies are expected to combine GLP‐1 agonists with agents targeting neuromuscular or ventilatory traits [74], reinforcing the multidimensional logic of endotypic therapy.

A subset of patients manifests a cardiometabolic endotype, characterized by OSA coexisting with systemic metabolic dysfunction, that is, insulin resistance, hepatic steatosis, and hypertension. These individuals often exhibit partial response to CPAP and require multimodal management incorporating anti‐inflammatory and metabolic pharmacotherapy, including GLP‐1/GIP agonists such as tirzepatide, sodium‐glucose cotransporter‐2 (SGLT2) inhibitors, and peroxisome‐proliferator‐activated receptor‐δ (PPAR‐δ) agonists [73]. Such agents may simultaneously improve weight, glycemic control, endothelial function, and ventilatory stability, addressing both systemic and airway pathophysiology.

Sex and age further modulate the metabolic endotype. Hormonal changes after menopause, as well as sarcopenia and fat redistribution in older adults, influence chemoreflex sensitivity and pharyngeal muscle responsiveness, thereby altering both disease expression and treatment response [75, 76].

The same integrative reasoning applies to positional and phenotype‐specific OSA. Wang et al. [77] showed that physiological traits differ between positional and non‐positional disease, explaining variable responses to PT and highlighting how trait identification can refine non‐CPAP interventions. Multimodal treatment, such as hypoglossal stimulation combined with acetazolamide in patients with dual deficits in muscle responsiveness and control stability, or tirzepatide plus atomoxetine‐oxybutynin in those with combined metabolic and neuromuscular impairment, illustrates the maturity of this mechanistic approach. In clinical terms, the physician will have to orchestrate combinations of therapies according to the patient's dominant endotypic architecture, not merely choosing therapy based on clinical and polysomnographic‐derived disease severity.

4.5. Current Challenges and Unmet Needs in Endotype‐Based Management of OSA

Endotype‐based management reframes OSA as a network disorder of structure, neurochemistry, and ventilatory control, each node being a target for intervention. Some comprehensive reviews [5, 7] emphasize that these insights have turned pharmacologic therapy from speculative to realistic. The most effective combinations, adrenergic‐antimuscarinic agents for muscle activation, carbonic anhydrase inhibitors for ventilatory stabilization, hypnotics for arousal modulation, and GLP‐1 analogues for weight reduction, collectively address the multidimensional nature of OSA pathogenesis. Their integration into practice, as envisioned by the ERS guideline [4] and expanded by the ATS roadmap [78], marks the transition from device‐dependent palliation to physiological normalization.

Yet the translation of endotyping into routine care remains incomplete. Despite impressive progress, trait quantification is still semi‐empirical, and physiological abnormalities cannot always be measured or reversed with precision [6, 29]. The FLOW study and accompanying commentary highlight that even mechanism‐specific drugs such as sultiame yield variable outcomes depending on baseline endotype and trait interaction [67, 74].

When traits coexist, such as moderate collapsibility combined with high loop gain or low arousal threshold, therapeutic rules become less predictable, and single‐axis interventions rarely suffice [29]. True precision management will therefore rely on multimodal strategies that align pharmacologic, mechanical, and neurostimulatory approaches (e.g., MAD, positional therapy, HNS, acetazolamide, or tirzepatide) with each patient's dominant physiological pattern [6, 52, 54, 72].

In this context, therapeutic success should no longer be measured solely by reductions in AHI or CPAP adherence, but by restoration of stable sleep, preserved oxygenation, and reduced neurocognitive and cardiovascular burden [79]. However, predicting who will achieve such multidimensional recovery remains challenging. Current endotyping models are largely static, whereas physiological traits evolve with ageing, weight gain, and metabolic change, altering responsiveness to therapy [5, 29, 71, 73]. Importantly, emerging evidence indicates that endotypes also influence treatment adherence itself. Zinchuk and colleagues [80] showed that a low arousal threshold is independently associated with poorer long‐term CPAP adherence, even after adjustment for OSA severity, demographic factors, and comorbidities, highlighting sleep instability as a key barrier to sustained therapy use. These findings further support the concept that physiological traits shape not only treatment efficacy but also tolerability and long‐term engagement, reinforcing the need for dynamic and longitudinal endotyping frameworks.

Moreover, most existing algorithms have been developed from limited cohorts and require prospective validation across sex, ethnicity, and comorbidity profiles [5]. The field also lacks standardized physiological metrics to define true “normalization,” making it difficult to quantify when a trait has been successfully corrected [78].

Thus, while OSA management has entered the era of mechanism‐based precision, the fine‐tuning of treatment remains an open frontier. Figure 4 summarizes the process leading from different clinical presentations of OSA to their combination with treatable traits to the actual choice of treatment in clinical practice. Future research must integrate longitudinal phenotyping, adaptive combination therapy, and digital biomarkers to achieve genuine personalization, where therapy evolves dynamically with the patient's physiology rather than against it.

FIGURE 4.

The clinical presentation of OSA is heterogeneous, and the complexity increases when treatable traits are added to the clinical picture of each patient. The therapeutic approaches are limited, and the choice of treatment has to be based on the combination of clinical and endotypic traits, and according to the patient's preferences.

5. Conclusion

The understanding of OSA pathophysiology has evolved over the years from the basic principle of the inability of oropharyngeal dilating muscle to withstand the increased collapsing forces of a narrowed upper airway to a greater appreciation of the potential contribution from endotypes associated with disturbances in respiratory control, both during sleep and in association with apnoea. These latter aspects offer the opportunity to develop novel therapies, which have been a major focus of research in recent years. However, key to the provision of endotype‐based therapy will be the ability to measure individual endotypes in clinical practice. While pharmacological agents have not yet been developed to the point where they are likely to replace CPAP as the first‐choice therapy in patients with moderate or severe OSA, they show promise as viable alternatives for patients with mild OSA, or those who are intolerant to CPAP. The pace of development for endotype‐focused pharmacological agents has been rapid over recent years and offers the prospect for more precision management of OSA, based on the identification of specific endotypes in individual patients.

Funding

The authors have nothing to report.

Conflicts of Interest

Walter T. McNicholas reports speaker fees from Bioprojet but no conflicts of interest relating to this manuscript. Sara Op de Beeck reports no conflict of interest. Pasquale Tondo reports a speaker fee from MedicAir but no conflicts of interest relevant to the manuscript. Johan Verbraecken received institutional fees and educational grants from AirLiquide, Bioprojet, Löwenstein Medical, Inspire, Mediq, Nyxoah, OSG Natus, Philips, ProSomnus, ResMed, Sefam, SomnoMed, SOS Oxygène, Tilman, Total Care, Vivisol, Vlaamse Gemeenschap, VLAIO, Westfalen Medical, Zoll Itamar; consultancy fees from Aurobindo, Bioprojet, Epilog, Withings; speaker fees from Azelis, Bioprojet, DEME, Pfizer, Procter & Gamble, SD Worx, Total Care, Vlaamse Gemeenschap, outside the submitted work. Maria R. Bonsignore reports consultancy fees from Takeda, Bioprojet, and ResMed.

Biographies

Walter T. McNicholas, FRCPI, FERS is Newman Professor at University College Dublin and Consultant in Respiratory/Sleep Medicine at St. Vincent's Hospital, Dublin, Ireland. He is Past President of the European Respiratory Society (ERS) and the European Sleep Research Society (ESRS). Walter is Associate Editor of ERJ Open Research, Joint Chief Editor of the ESRS Sleep Medicine Textbook (now preparing its Third Edition), and has over 300 PubMed‐listed publications in International Scientific Journals, many on OSA pathophysiology (h‐index: 88). Walter was Chair of the ERS‐ESRS Sleep and Breathing Conference (2011–2021) and received an ERS Lifetime Achievement Award in 2022.

Sara Op de Beeck is a postdoctoral researcher at the University of Antwerp. Her main research interests include non‐invasive pathophysiological endotyping and non‐invasive site of collapse detection to come to precision medicine for obstructive sleep apnea patients. She has authored over 30 peer‐reviewed articles and serves as reviewer for several scientific journals. She has received multiple national and international awards recognizing her research.

Pasquale Tondo, MD, PhD, is a postdoctoral researcher at the University of Foggia and consultant physician at the University Hospital of Foggia, Italy. He has authored over 70 peer‐reviewed international publications and serves as editor and reviewer for several scientific journals. He chairs the Respiratory Sleep Disorders Study Group of the Italian Respiratory Society (SIP/IRS). His research focuses on sleep‐disordered breathing, spanning molecular biomarkers such as miRNAs and the application of machine‐learning models for phenotyping and outcome prediction.

Johan Verbraecken is medical coordinator of the Multidisciplinary Sleep Disorders Centre of the Antwerp University Hospital (UZA). He is also Associate Professor and Training Director for Pulmonology at the University of Antwerp (Belgium). He holds leadership roles in national and international scientific societies and networks (ERS, ESRS, ESADA, ANSS), and is current Chair of the ESRS Sleep Medicine committee. He has authored over 300 scientific papers in journals and books, including some European Guidelines as chair or co‐author. Part of his research is dedicated to the pathophysiology of obstructive sleep apnea.

Maria R. Bonsignore, MD, FERS is Professor in Respiratory Medicine, University of Palermo, and Research Consultant for Istituti Clinici Maugeri, Pavia, Italy. Her main research interest is sleep‐disordered breathing. She was Head of ERS Assembly 4 (2014–2017), ERS Congress Co‐Chair (2023), ERS Sadoul Lecturer (2020), and is current Chair of the ERS‐ESRS Sleep and Breathing Conference. She is a Steering Committee member of the AASM Task Force: Re‐Envisioning OSA Characterization. She Chaired ERS‐ESRS Task Forces on fitness to drive in OSA patients, and Advanced Telemedicine for OSA. She received an ERS Lifetime Achievement Award for Sleep Disordered Breathing in 2025.

McNicholas W. T., Op de Beeck S., Tondo P., Verbraecken J., and Bonsignore M. R., “ OSA Endotyping as Targets for Treatment. How Far Have We Come?,” Respirology 31, no. 4 (2026): 339–351, 10.1002/resp.70233.