CrossTalk proposal: The human upper airway does behave like a Starling resistor during sleep



[ Alan R. Schwartz and Philip L. Smith are Professors of Medicine at the Johns Hopkins School of Medicine. They have investigated mechanisms of upper airway obstruction during sleep in human and animal models, with particular emphasis on the pathogenesis of obstructive sleep apnea. Dr. Smith founded the Johns Hopkins Sleep Disorders Center and along with Dr. Schwartz has been directing its clinical and research activities for many years. Dr. Schwartz also codirects the Johns Hopkins Center for Interdisciplinary Sleep Research and Education, which provides core support for clinical research and education for sleep-related studies and scientists.]

The upper airway and collapsible biological conduits

The upper airway is composed of a wide array of soft tissues, muscles and bony structures that modulate its patency for respiratory, speech and alimentary functions. During natural or pharmacologically induced sleep, pharyngeal collapse leads to recurrent periods of upper airway obstruction with oxyhaemoglobin desaturations and arousals (Remmers et al. 1978), or to hypoventilation and death (Hillman et al. 2010). It is widely recognized that pharyngeal collapse is governed by a complex interplay of mechanical and neuromuscular factors.

In an effort to model the control of pharyngeal patency, the airway has been likened to a simple collapsible conduit or Starling resistor (Gold & Schwartz, 1996), as described for the intrathoracic airways and vasculature (Permutt et al. 1962; Permutt & Riley, 1963; Pride et al. 1967; West, 2012). In this model, a discrete collapsible segment is subject to a surrounding or critical pressure (Pcrit) that governs its collapsibility. This model incorporates several mechanical elements, representing upper airway structures (Fig. 1, top panel). A collapsible segment is infinitely compliant (i.e. elastically unstable) (Gold & Schwartz, 1996). Its patency is determined by its transmural pressure. The segment collapses whenever transmural pressure falls below zero (i.e. when intraluminal pressure falls below extraluminal surrounding or critical pressure). The collapsible segment is flanked by relatively rigid segments, corresponding to the upstream nasal and downstream tracheal airways, respectively. This model predicts the effects of transmural pressure on airflow dynamics and the severity of upper airway obstruction during sleep as follows.

Figure 1.

Mechanisms of upper airway obstruction and the effects of decreasing upstream and downstream pressures 
Pharyngeal collapsibility is determined by the surrounding pressure (Pcrit). As pressure within the chamber of the Starling resistor changes, so does the collapsibility of the tube. When upstream pressure (Pus) is lower than Pcrit, complete occlusion of the tube occurs (left panel). When downstream pressure (Pds) is decreased below Pcrit, inspiratory flow limitation ensues (right panel). Under these circumstances, the level of maximal inspiratory airflow (VImax) is determined by the upstream and critical pressures and upstream nasal resistance (Rus) as described in the equation at the lower right.

Effects of decreasing intraluminal pressures on upper airway obstruction

Decreases in downstream pressure induce inspiratory airflow limitation. The Starling resistor model predicts that the airway flow limits (Permutt & Riley, 1963; Pride et al. 1967) during inspiration when the downstream pressure falls below Pcrit (Fig. 1, lower right panel). Flow limitation is defined by an initial rise in airflow and subsequent plateau at a maximal level (VImax) despite further increases in inspiratory effort (Schwartz et al. 1989). During flow limitation, the pharynx collapses and Pcrit replaces tracheal pressure as the effective downstream pressure to inspiratory airflow. High frequency oscillations in flow (snoring) often occur, which are a hallmark of elastically unstable conduits. As VImax falls, obstructive hypopnoeas and ‘respiratory-related arousals’ ensue when marked reductions in VImax compromise ventilation, triggering recurrent desaturations and arousals (Hosselet et al. 2001). However, increasingly negative downstream ‘suction’ pressures do not cause the flow-limited airway to occlude (Schwartz et al. 1988, 1989; Smith et al. 1988) or contribute to the pathogenesis of obstructive apnoeas (Smith et al. 2011), as predicted by Starling resistors.

Decreases in upstream pressure induce pharyngeal occlusion. For the upper airway to occlude, the upstream nasal pressure must also fall below Pcrit (Fig. 1, lower left panel), as demonstrated experimentally in normal sleeping individuals (Schwartz et al. 1988; King et al. 2000). As upstream pressure falls to subatmospheric levels, VImax decreases progressively. This manoeuvre recapitulates the entire spectrum of pharyngeal airflow obstruction during sleep. Initially, modest decreases in VImax are associated with stable snoring. As nasal pressure is further decreased, periodic obstructive hypopnoeas ensue, and finally yield to obstructive apnoeas when upstream pressure falls below Pcrit (Schwartz et al. 1988; Smith et al. 1988, 2011; Gleadhill et al. 1991; Gold & Schwartz, 1996). At this level of nasal pressure, obstructive sleep apnoea with recurrent arousals, oxyhaemoglobin desaturations and daytime hypersomnolence ensue (King et al. 2000). Conversely, when patients with obstructive sleep apnoea breathe at atmospheric nasal pressure, a positive Pcrit causes the pharynx to occlude spontaneously. Pharyngeal occlusion is relieved by increasing nasal pressure (Smith et al. 1988). As nasal continuous positive airway pressure is applied, VImax rises progressively as repetitive obstructive hypopnoeas yield to non-periodic inspiratory airflow limitation (snoring) and ultimately normal (non-flow-limited) breathing (Smith et al. 1988). The Starling resistor model correctly predicted that the gradient between upstream and critical pressures determines the presence and severity of pharyngeal airflow obstruction during sleep.

Effects of increasing critical extraluminal pressures

Critical pressure predicts the degree of pharyngeal obstruction from health to disease. When individuals breathe at atmospheric nasal pressure, Pcrit determines pharyngeal patency during sleep (see equation in Fig. 1). In fact, quantitative differences in critical pressures distinguish among groups with varying degrees of pharyngeal obstruction from health (normal breathing) to disease (obstructive sleep apnoea) (Schwartz et al. 1988; Smith et al. 1988, 2011; Gleadhill et al. 1991; Gold & Schwartz, 1996). Thus, a negative critical pressure is required to maintain airway patency in normal individuals, while progressive elevations in Pcrit toward the positive range lead to greater degrees of airway obstruction in patients with stable snoring, obstructive hypopnoeas and apnoeas (Gleadhill et al. 1991). Conversely, sleep apnoea remits when therapeutic interventions drive the critical pressure below ∼−4 cmH2O, a threshold below which airflow (VImax) becomes sufficient to restore ventilation during sleep (Gold & Schwartz, 1996). The likelihood of achieving this goal will depend on the degree of elevation in Pcrit and the magnitude of the decrease with therapy (Gold & Schwartz, 1996; Boudewyns et al. 2000; Ng et al. 2003; Oliven et al. 2003).

Critical pressure models structural and neuromuscular control of pharyngeal patency.P crit integrates both structural and neuromuscular determinants of airway collapsibility. In humans and experimental animals, critical pressures rise markedly with manoeuvres that decrease pharyngeal neuromuscular tone (Seelagy et al. 1994; Rowley et al. 1997; Patil et al. 2007; McGinley et al. 2008; Polotsky et al. 2012), leading to spontaneous airway obstruction even when intraluminal negative pressure swings are absent (Dempsey et al. 2010). Under passive conditions (when pharyngeal neuromotor tone is suppressed), Pcrit reflects the influence of anatomic loads on extraluminal collapsing pressures, which promote airway obstruction during sleep (Patil et al. 2007; McGinley et al. 2008). Passive Pcrit is elevated in apnoeic patients compared to matched controls, suggesting an underlying structural predisposition to sleep apnoea (Patil et al. 2007). Further elevations in passive Pcrit have been associated with age, weight and male sex (Kirkness et al. 2008), providing evidence that anatomic loads mediate effects of recognized sleep apnoea risk factors on sleep apnoea susceptibility (Eastwood et al. 2002).

Structural loads can be mitigated by neuromuscular activity that restores pharyngeal patency during sleep (Dempsey et al. 2010). This activity is triggered by upper airway obstruction, which stimulates afferent reflexes that decrease airway collapsibility (Pcrit) (Seelagy et al. 1994; Patil et al. 2007; McGinley et al. 2008). These responses are severely impaired in patients with sleep apnoea compared to matched controls (Jordan et al. 2007; Patil et al. 2007; McGinley et al. 2008; Chin et al. 2012). These disturbances may be related to a loss of tonic expiratory (McGinley et al. 2008) and/or phasic inspiratory neuromuscular activity (Malhotra et al. 2000). These findings suggest that ‘two hits’ are required for sleep apnoea pathogenesis (Patil et al. 2007). Pathogenic increases in Pcrit result from elevations in passive mechanical loads (Mezzanotte et al. 1992) and a lack of compensatory neuromuscular responses (Schwartz et al. 2008).


We conclude that the upper airway exhibits pressure–flow dynamics during sleep that are comparable to those in a Starling resistor. Reductions in intraluminal pressures cause the pharynx to flow-limit or occlude when downstream and upstream pressures fall below Pcrit, respectively. Under these circumstances, the Pcrit surrounding the collapsible segment becomes the effective downstream pressure to flow through the airway. The gradient between upstream and critical pressures determines the severity of airflow obstruction, as reflected by the level of maximal inspiratory airflow.

The Starling resistor model has several distinct advantages. First, its application to the upper airway is completely consistent with flow dynamics observed in collapsible biological conduits including the intrathoracic airways and vasculature. Second, it accounts for variability in the degree of upper airway obstruction during sleep along the entire spectrum from health to disease. As upstream and critical pressures are altered for experimental or therapeutic purposes, airflow obstruction can be relieved completely whenever the upstream and downstream pressures remain greater than Pcrit (Pus, Pds > Pcrit). Alternatively, progressively greater degrees of airflow obstruction can be induced along with stable snoring (when PusPcrit > Pds) and periodic obstructive hypopnoeas as upstream pressure approaches Pcrit (when Pus > Pcrit > Pds). Finally, obstructive sleep apnoea ensues whenever Pcrit exceeds pressures upstream and downstream to the collapsible site (Pus, Pds < Pcrit). Third, patients can be stratified for specific therapies based on the degree to which Pcrit is elevated, and therapies can be combined to reduce Pcrit sufficiently and relieve sleep apnoea. Fourth, partitioning Pcrit into structural and neuromuscular components has disclosed that defects in both are required for sleep apnoea pathogenesis. Finally, the Starling resistor model provides a unified approach to the study of upper airway function in humans and animals, and a framework for probing specific pathogenic genetic, humoral and environmental mechanisms of disease pathogenesis and pharmacotherapy.

Call for comments

Readers are invited to give their views on this and the accompanying CrossTalk articles in this issue by submitting a brief comment. Comments may be posted up to 6 weeks after publication of the article, at which point the discussion will close and authors will be invited to submit a ‘final word’. To submit a comment, go to;591/9/2229