Volume 591, Issue 5 p. 1179-1193
Free Access

Role of nocturnal rostral fluid shift in the pathogenesis of obstructive and central sleep apnoea

Laura H. White

Laura H. White

Sleep Research Laboratory and Department of Medicine of the University Health Network Toronto Rehabilitation Institute and Toronto General Hospital, and the Centre for Sleep Medicine and Circadian Biology of the University of Toronto, Toronto, Ontario, Canada

Search for more papers by this author
T. Douglas Bradley

T. Douglas Bradley

Sleep Research Laboratory and Department of Medicine of the University Health Network Toronto Rehabilitation Institute and Toronto General Hospital, and the Centre for Sleep Medicine and Circadian Biology of the University of Toronto, Toronto, Ontario, Canada

Search for more papers by this author
First published: 12 December 2012
Citations: 205
T. D. Bradley: University Health Network Toronto General Hospital, 9N-943, 200 Elizabeth Street, Toronto, Ontario, M5G 2C4, Canada. Email: [email protected]


Abstract Obstructive sleep apnoea (OSA) is common in the general population and increases the risk of motor vehicle accidents due to hypersomnolence from sleep disruption, and risk of cardiovascular diseases owing to repetitive hypoxia, sympathetic nervous system activation, and systemic inflammation. In contrast, central sleep apnoea (CSA) is rare in the general population. Although their pathogenesis is multifactorial, the prevalence of both OSA and CSA is increased in patients with fluid retaining states, especially heart failure, where they are associated with increased mortality risk. This observation suggests that fluid retention may contribute to the pathogenesis of both OSA and CSA. According to this hypothesis, during the day fluid accumulates in the intravascular and interstitial spaces of the legs due to gravity, and upon lying down at night redistributes rostrally, again owing to gravity. Some of this fluid may accumulate in the neck, increasing tissue pressure and causing the upper airway to narrow, thereby increasing its collapsibility and predisposing to OSA. In heart failure patients, with increased rostral fluid shift, fluid may additionally accumulate in the lungs, provoking hyperventilation and hypocapnia, driving inline image below the apnoea threshold, leading to CSA. This review article will explore mechanisms by which overnight rostral fluid shift, and its prevention, can contribute to the pathogenesis and therapy of sleep apnoea.

[ Laura White is a Clinical Research Fellow at the University Health Network, Toronto, investigating the role of fluid retention and redistribution in the pathogenesis and treatment of sleep apnoea. She obtained her medical degree at Edinburgh University and is a Specialist Trainee in Respiratory Medicine in the South West of England. T. Douglas Bradley is Professor of Medicine and Director of the Division of Respirology of the University of Toronto and of the Sleep Research Laboratories of the Toronto Rehabilitation Institute and Toronto General Hospital. He obtained his MD degree from the University of Alberta and his specialty and research training at the University of Toronto and McGill University. His research focuses on the causes and consequences of sleep apnoea and of its treatment on cardiovascular diseases, especially heart failure, hypertension and stroke. His work has been supported by the Canadian Institutes of Health Research and other agencies.]


  • AHI
  • apnoea–hypopnoea index
  • ANP
  • atrial natriuretic peptide
  • BMI
  • body mass index
  • CAPD
  • continuous ambulatory peritoneal dialysis
  • CPAP
  • continuous positive airway pressure
  • CSA
  • central sleep apnoea
  • ESRD
  • end-stage renal disease
  • IJV
  • internal jugular vein
  • LBPP
  • lower body positive pressure
  • LFV
  • leg fluid volume
  • MRI
  • magnetic resonance imaging
  • OSA
  • obstructive sleep apnoea
  • inline image
  • arterial partial pressure of oxygen
  • inline image
  • partial pressure of carbon dioxide
  • PCWP
  • pulmonary capillary wedge pressure
  • UA
  • upper airway
  • UA-XSA
  • upper airway cross-sectional area
  • UPPP
  • uvulopalatopharyngoplasty
  • Obstructive sleep apnoea (OSA) is a common condition, characterised by repetitive apnoeas due to collapse of the pharynx. These occur secondary to the normal withdrawal of pharyngeal dilator muscle tone at the onset of sleep, superimposed upon a narrowed or collapsible pharynx (AAMSTF, 1999). The prevalence of OSA increases with increasing body mass index and neck girth, probably due to fat deposition in the soft tissue surrounding the pharynx, which narrows the lumen and increases its collapsibility (Horner et al. 1989; Shelton et al. 1993). However, only approximately one-third of the variability in sleep apnoea severity is attributable to these two indices of obesity (Dempsey et al. 2002). Therefore, other factors must be playing a role in the pathogenesis of OSA.

    In contrast to OSA, central sleep apnoea (CSA) is characterised by partial or complete cessation of airflow due to a temporary reduction in, or cessation of, central respiratory drive. Central hypopnoeas and apnoeas usually occur following episodes of hyperventilation when inline image falls toward or below the apnoea threshold, respectively, during sleep. The tendency to hyperventilate is associated with augmented central and peripheral chemo-responsiveness (Solin et al. 2000). CSA, although rare in the general population, is common in patients with heart failure, where it can co-exist with OSA, raising the possibility of a common pathogenesis (Bixler et al. 1998; Tkacova et al. 2001; Tkacova et al. 2006; Yumino et al. 2009).

    The prevalence of sleep apnoea is much higher in patients with fluid-retaining states such as heart failure and end-stage renal disease (ESRD) than in the general population (Young et al. 1993; Jurado-Gamez et al. 2007; Yumino et al. 2009). This led to the hypothesis that fluid retention is involved in the pathogenesis of both OSA and CSA. According to this hypothesis, during the day, fluid accumulates in the intravascular and interstitial spaces of the legs due to gravity, and upon lying down at night redistributes rostrally, again due to gravity. Some of this fluid may accumulate in the neck, increasing tissue pressure and causing the upper airway (UA) to narrow, predisposing to OSA, or in the lungs, where it may provoke hyperventilation, predisposing to CSA (Fig. 1).

    Details are in the caption following the image

    The role of overnight rostral fluid shift in the pathogenesis of obstructive and central sleep apnoea (OSA and CSA)
    Overnight fluid shift from the legs to the neck can affect upper airway mechanics and lead to OSA, whereas fluid shift to the lungs can provoke hyperventilation, hypocapnia and CSA. CO, cardiac output; inline image, partial pressure of carbon dioxide; PCWP, pulmonary capillary wedge pressure; UA, upper airway; UA-XSA, upper airway cross-sectional area.

    There is now considerable evidence to support this hypothesis. Therefore, the primary objective of this article is to review this evidence and explore the mechanisms by which fluid retention during the day and its nocturnal redistribution might contribute to the pathogenesis of OSA and CSA.

    Prevalence of sleep apnoea in patients with fluid-retaining states

    Sleep apnoea is diagnosed by overnight polysomnography, during which apnoeas and hypopnoeas are detected by an absence or reduction in tidal volume for at least 10 s, respectively (Iber, 2007). During obstructive apnoeas, respiratory drive persists resulting in inspiratory efforts against the occluded airway, causing chest wall distortion. In contrast, during central apnoeas, respiratory drive is absent. The presence and severity of sleep apnoea is determined by the apnoea–hypopnoea index (AHI), which is the number of apnoeas and hypopnoeas per hour of sleep. A sleep apnoea disorder is present if the AHI is ≥5 and mild if the AHI is 5–15, moderate if 15–30 and severe if ≥30 (AAMSTF, 1999).

    In the general population, the prevalence of OSA is estimated at 3–14% in men and 4–9% in women, while the prevalence of CSA is <1% (Young et al. 1993; Bixler et al. 1998; Duran et al. 2001; Ip et al. 2001). In comparison, the prevalence of OSA is higher in patients with fluid-retaining conditions, and of CSA is higher in patients with heart failure (Table 1).

    Table 1. Prevalence of sleep apnoea in different populations
    Population Study N AHI OSA prevalence (%) CSA prevalence (%)
    Community Young et al. (1993) 602 ≥15 9 (M) 4 (F)
    Duran et al. (2001) 2148 ≥15 14 (M) 7 (F)
    Bixler et al. (1998)* 741 ≥20  6 0.5
    Hypertension Fletcher et al. (1985)* 46 ≥10 30
    Worsnop (1998) 34 ≥10 23
    Drug-resistant hypertension Logan et al. (2001) 41 ≥10 83
    Pratt-Ubunama (2007) 71 ≥10 65
    End-stage renal disease de Oliveira Rodrigues et al. (2005) 45 ≥5  31
    Jurado-Gamez et al. (2007) 32 ≥10 44
    Heart failure Javaheri et al. (2006)* 100 ≥15 12 37
    Yumino et al. (2009) 218 ≥15 26 21
    • AHI, apnoea–hypopnoea index; F, female; M, male; OSA, obstructive sleep apnoea; CSA, central sleep apnoea. *All men; †33/34 men.

    In contrast to the general population, severity of OSA is not associated with increasing BMI in heart failure patients, suggesting that factors other than obesity are more important in the pathogenesis of OSA in these patients (Arzt et al. 2006). In heart failure patients, those with sleep apnoea have greater sodium intake, which promotes fluid retention, than those without sleep apnoea, and the AHI correlates with sodium intake (Kasai et al. 2011).

    Mechanisms of fluid shift within the body and its role in the pathogenesis of OSA and CSA

    Starling's forces Overnight rostral fluid shift can contribute to the pathogenesis of OSA and CSA via Starling's forces. Isotonic fluid moves passively between the capillaries and the interstitial space according the balance of opposing forces: capillary hydrostatic versus colloid osmotic pressure (Starling, 1896). When moving from the lying to the upright position, hydrostatic pressure in the capillaries of the legs exceeds that in the interstitial compartment due to gravitational effects and this causes filtration of fluid from the capillaries into the interstitial space. This concept is key to understanding movement of fluid between the intravascular and extravascular spaces of the legs, lungs and neck between day and night.

    Leg fluid changes with posture Lying down and standing up are associated with significant alterations in body fluid distribution. (Thompson et al. 1928; Waterfield, 1931b; Youmans et al. 1934). Capillary pressure in the legs on standing (90–120 cmH2O) greatly exceeds the pressure required for movement into the interstitial space (just 15–20 cmH2O) and filtration rate is directly proportional to venous pressure (Krogh et al. 1932; Youmans et al. 1934; Levick & Michel, 1978). As a consequence, on standing, plasma volume decreases by 300–400 ml but leg volume increases by 100–300 ml, due to venous pooling and fluid filtration into the interstitial space, with the reverse occurring on lying down (Table 2).

    Table 2. Summary of studies examining postural fluid changes
    Study Subjects Outcome measured Method Manoeuvre Time course (minutes) Result
    Thompson et al. (1928) Normal Plasma volume Dye dilution Lying to standing  30 −290 ml
    Waterfield et al. (1931a) Normal Plasma volume Carbon monoxide dilution Lying to standing  40 −425 ml
    Waterfield et al. (1931b) Normal Leg volume Water displacement Calf muscle contraction Immediate −75 ml
    Lying to standing  40 +83 ml
    Youmans et al. (1934) Normal and low protein Leg volume Water displacement Lying to standing  60 +250 ml (normal) +330 ml (low protein)
    Fawcett & Wynn (1960) Normal and low protein/ oedematous Plasma volume Dye dilution Lying to standing  15 −11% (normal) −11% (low protein)
     60 −10% (normal) −16% (low protein)
    Plasma volume Dye dilution Standing to lying  15 +8% (normal) +14% (low protein)
     60 +14% (normal) +21% (low protein)
    Berg et al. (1993) Normal XSA of calf CT Standing to lying 120 −5.5%*
    Calf fluid volume Bioimpedance −8.7%†
    Baccelli et al. (1995) Normal Leg intravascular blood volume Blood-pool scintigraphy Lying to standing   2 +135%
    Lung blood volume −16%
    • *No further change after 60 min; †exponential decay in reduction in fluid volume; CT, computer tomography; XSA, cross-sectional area.

    Intravascular fluid loss on standing is quite rapid: plasma concentration increases over the first 20–40 min after which it plateaus. The rate of filtration from the capillaries to the interstitial space decreases over time due to rising capillary colloid osmotic and increasing tissue pressures, until a balance is achieved and filtration ceases (Krogh et al. 1932; Landis & Gibbon, 1933; Youmans et al. 1934). On changing from standing to lying, the reverse occurs, with most fluid reabsorbed within 30–60 min, with an initial rapid decrease followed by an exponential decay in fluid shift (Thompson et al. 1928; Waterfield, 1931a; Fawcett & Wynn, 1960; Berg et al. 1993). Therefore upon lying in bed at night, fluid rapidly shifts from the interstitial to the intravascular compartment of the legs.

    In fluid overloaded patients who have increased venous pressure, the interstitial space can accommodate large increases in volume, with oedema formation (Guyton, 1965). Hence on lying down, the greater the degree of leg oedema, the greater the volume of fluid filtered into the vascular space and the greater the decrease in leg fluid volume (Yumino et al. 2010). Similarly, in patients who are oedematous due to low protein states, decreased colloid osmotic pressure results in a greater movement of fluid from the vasculature to the interstitial space of the legs on standing, and vice versa (Youmans et al. 1934). Increases in microvascular permeability, as have been described in diabetic and heart failure patients and in conditions of acute and chronic inflammation, may increase the volume of fluid moving between the intravascular and interstitial spaces in response to changes in hydrostatic pressure (Jaap et al. 1993; Mahy et al. 1995; Kvietys & Granger, 2012). Theoretically, therefore, hypoalbuminaemia and increases in microvascular permeability could predispose to sleep apnoea, but no studies have addressed these possibilities.

    Rostral fluid shifts from the legs during recumbency Fluid re-entering the venous system on lying down moves rostrally to both the chest and the neck with the effects of gravity (Avasthey & Wood, 1974; Terada & Takeuchi, 1993). On lying down, leg blood volume decreases rapidly and fluid is redistributed to the thorax, neck and head (Hildebrandt et al. 1994; Baccelli et al. 1995). When fluid is shifted out of the legs during application of lower body positive pressure (LBPP) via inflation of anti-shock trousers, neck circumference increases within 1 min, indicating rapid movement of fluid into the neck (Chiu et al. 2006; Shiota et al. 2007; Su et al. 2008). Therefore, in oedematous patients, the increased volume of fluid re-entering the intravascular space at night would result in greater movement of fluid into the chest and the neck than in normovolaemic subjects.

    Obstructive sleep apnoea

    Complex anatomical, physiological and neural factors are involved in UA collapse in patients with OSA, as outlined in Fig. 2, and reviewed elsewhere (Ryan & Bradley, 2005). The UA is surrounded by bones and soft tissues. Due to mismatch between the space available within the bony cage and amount of soft tissue inside it, upper airway cross-sectional area (UA-XSA) is decreased in OSA patients (Schwab et al. 1995). Additionally, UA collapsibility is increased in OSA patients. The UA acts like a Starling resistor during sleep: under passive conditions it collapses when intraluminal pressure falls below extraluminal pressure. The intraluminal pressure at which the UA collapses is termed Pcrit. Factors influencing Pcrit include pharyngeal wall compliance, intra- and extraluminal pressures and pharyngeal dilator muscle activity (Fig. 2). In OSA patients, pharyngeal compliance is increased and Pcrit is frequently positive such that collapse can occur above atmospheric pressure during sleep, whereas in normal subjects Pcrit is negative (Brown et al. 1985; Gleadhill et al. 1991).

    Details are in the caption following the image

    Mechanisms influencing upper airway (UA) collapsibility in the pathogenesis of obstructive sleep apnoea (OSA)
    A number of dynamic factors, including fluid shift into the neck, determine the active (pharyngeal dilator muscle activity) and passive (intra- and extraluminal pressure and pharyngeal wall compliance) properties of the UA which can increase UA collapsibility and lead to UA obstruction and OSA. REM, rapid eye movement sleep.

    The pharyngeal dilator muscles maintain UA patency and their reduced activity during sleep predisposes to UA collapse. A number of pathways influence the activity of the UA dilator muscles (Fig. 2) (Ryan & Bradley, 2005). Compromise of one or more of these pathways, such as reduced UA mechanoreceptor and mucosal sensory sensitivity, can further predispose to UA collapse (Chadwick et al. 1991; Loewen et al. 2011). In addition, overnight fluid shift into the neck veins and pharyngeal mucosa may increase tissue pressure around the UA, decrease UA-XSA and increase UA collapsibility, predisposing to obstructive apnoeas.

    Cross-sectional area and pressure of the internal jugular vein (IJV) increase when moving from upright to supine (Lobato et al. 1998; Cirovic et al. 2003). Expansion of the IJVs is, however, limited by the mandible laterally and cervical spine posteriorly, so that increased pressure is most likely transmitted medially to the pharynx, narrowing its lumen. In addition, increased hydrostatic pressure is likely to cause transudation of fluid into the interstitial space surrounding the UA. In animals, increased extrapharyngeal tissue pressure, created by inflation of balloons within the peripharyngeal tissues, was transmitted to the pharynx and both decreased UA-XSA and increased UA resistance (Winter et al. 1995; Kairaitis et al. 2009).

    In histological specimens of patients with OSA obtained from uvulopalatopharyngoplasty (UPPP), mucosal oedema is common (Woodson et al. 1991; Sekosan et al. 1996; Anastassov & Trieger, 1998; Berger et al. 2002). In some studies, this was associated with inflammation, which could be due to vibratory injury from snoring rather than fluid accumulation (Woodson et al. 1991; Sekosan et al. 1996; Paulsen et al. 2002). However, inflammatory changes were also found in controls and non-apnoeic snorers and so were not specific to OSA. Specimens obtained at UPPP are from the uvula and soft palate and so may not be representative of changes throughout the pharyngeal walls, especially the tongue. Pharyngeal wall oedema has, however, been reported on MRI of the UA in OSA patients (Fig. 3) and pharyngeal mucosal water content decreased after continuous positive airway pressure (CPAP) therapy (Ryan et al. 1991; Elias et al. 2013). The relative importance of postural fluid accumulation and UA inflammation in the causation of UA mucosal oedema therefore remains to be seen.

    Details are in the caption following the image

    Magnetic resonance images of the upper airways from end-stage renal failure patients with (A) and without (B) obstructive sleep apnoea (OSA)
    Significantly more pharyngeal wall oedema (arrows) is observed in the OSA patient, as indicated by lighter shade and as quantified by arbitrary units. Reproduced with permission from Elias et al. 2013.

    Alterations in pharyngeal blood volume can also influence UA properties. In anaesthetised cats, vasodilatation induced by intravenous nitroprusside increased UA collapsibility in association with a 39% reduction in UA-XSA largely due to increased pharyngeal mucosal water (Wasicko et al. 1990). Conversely, topical application of phenylepinephrine, a vasoconstrictor, to the pharynx increased UA-XSA by 14% in normal human subjects (Wasicko et al. 1991).

    Shepard et al. (1996) first studied the possible effect of fluid shifts into and out of the neck on the UA of men with OSA. They used leg tourniquets and leg elevation to decrease and increase venous return from the legs, respectively, although neither fluid shifts nor neck vein distension were measured. At end-inspiration, mean UA-XSA decreased with leg elevation compared to leg tourniquets, although there was no difference at end-expiration, when UA collapse is most likely to occur. While not definitive, these data did suggest a tendency for UA narrowing in response to increased venous return in men with OSA.

    Application of LBPP, which displaces approximately 350 ml of fluid from the legs of healthy subjects, induces increases in neck circumference and UA resistance, and a decrease in UA-XSA within 1 min, probably reflecting an increase in IJV volume (Chiu et al. 2006; Shiota et al. 2007). However, after 5 min, although neck circumference did not increase any further, there was a further increase in UA resistance and decrease in UA-XSA which may have been due to transudation of fluid into the pharyngeal mucosa. Similarly, LBPP increased UA collapsibility, which correlated with the decrease in leg fluid volume and increase in neck circumference (Su et al. 2008). Taken together, these studies indicate that changes in intravascular and extravascular pharyngeal mucosal fluid content influence UA anatomy and physiology.

    Overnight rostral fluid shift from the legs to the neck has been measured in a number of patient populations. In non-obese men, severity of OSA was strongly related to the degree of overnight leg fluid volume (LFV) reduction and concomitant increase in neck circumference (Redolfi et al. 2009). Indeed, the reduction in LFV was the strongest correlate of the AHI and explained approximately 64% of its variability independently of other factors including BMI. The overnight increase in neck circumference correlated strongly with the overnight decrease in LFV supporting the concept that some leg fluid redistributed to the neck. Similar relationships were observed between overnight decrease in LFV and severity of OSA in patients with hypertension and men with heart failure (Fig. 4A) and ESRD (Friedman et al. 2010; Yumino et al. 2010; Elias et al. 2012).

    Details are in the caption following the image

    Relationship between overnight ΔLFV and AHI in men and women with heart failure
    A, correlation between overnight decrease in leg fluid volume (LFV) and apnoea–hypopnea index (AHI) in men with heart failure and either obstructive or central sleep apnoea. Central sleep apnoea was associated with greater overnight fluid shift. Reproduced with permission from Yumino et al. 2010. B, in women with heart failure, there is no correlation between overnight decrease in LFV and AHI. ΔLFV, overnight change in leg fluid volume. Reproduced with permission from Kasai et al. 2012.

    Overnight rostral fluid shift might not be a primary causal factor for OSA, but could be secondary to negative intrathoracic pressure generated by inspiratory efforts during obstructive apnoeas, drawing fluid from the legs into the neck. However, in heart failure patients with OSA, CPAP, while preventing obstructive apnoeas, did not reduce overnight fluid movement out of the legs, indicating that rostral fluid shift from the legs is a primary phenomenon (Yumino et al. 2010).

    Finally, it is possible that elevated systemic venous and pulmonary arterial pressures could prevent venous drainage from the head and neck thereby potentiating UA oedema, tissue pressure and collapsibility. As described earlier, OSA is common in heart failure patients, who have elevated pulmonary and central venous pressures. In patients with idiopathic or chronic thrombo-embolic pulmonary arterial hypertension, one study reported a very high prevalence of OSA of 89%, although the reason for this high prevalence was not determined (Jilwan et al. 2013). Additionally, in fluid-overloaded ESRD patients, there was a strong correlation between both IJV volume and UA mucosal water content, and AHI, although central venous pressure was not measured (Elias et al. 2013).

    Fluid balance in OSA Oedema is common in OSA patients, even in those without underlying fluid retaining states, suggesting that OSA itself may lead to oedema (Whyte & Douglas, 1991; Iftikhar et al. 2008). A minority of OSA patients may develop pulmonary hypertension either secondary to intermittent hypoxia-driven pulmonary vasoconstriction, or to elevated pulmonary capillary wedge pressure (PCWP) due to left ventricular diastolic dysfunction, both of which cause oedema due to raised systemic venous pressure (Bradley et al. 1985; Sajkov & McEvoy, 2009).

    The potential role of fluid-regulating hormones in oedema formation in OSA patients is unclear. Although OSA leads to sympathetic activation it remains uncertain whether this stimulates their rennin–angiotensin– aldosterone system (Somers et al. 1995). For example, studies have shown variable levels of these hormones in OSA patients compared to controls, or before and after CPAP treatment (Follenius et al. 1991; Rodenstein et al. 1992; Maillard et al. 1997; Moller et al. 2003). Atrial natriuretic peptide (ANP), which promotes diuresis and natriuresis, may be elevated in OSA patients at night (Genest, 1986; Maillard et al. 1997; Kita et al. 1998). This is probably due to the effects of negative intrathoracic pressure generation and intermittent apnoea-related hypoxia that lead to increased venous return, pulmonary vasoconstriction, left ventricular diastolic dysfunction and right atrial stretch, which stimulates ANP release (Yalkut et al. 1996). Increased ANP levels should counteract oedema formation. However, treatment of OSA by CPAP decreases ANP levels in association with reduced nocturia and oedema, probably by reducing right atrial stretch (Baruzzi et al. 1991; Kita et al. 1998; Blankfield et al. 2004). Thus the relative roles of the rennin–angiotensin–aldosterone system and ANP in fluid regulation in OSA remain uncertain.

    Since OSA can contribute to oedema formation, this could lead to a vicious cycle of increased overnight rostral fluid shift and more severe OSA. However, as treatment with CPAP for one night did not prevent fluid shift from the legs, this suggests that nocturnal rostral fluid shift is a primary mechanism in the pathogenesis of OSA (Yumino et al. 2010). However, over longer periods of time CPAP may reduce oedema, but it remains to be determined whether this can attenuate OSA via reduction in overnight rostral fluid shift since, on withdrawal of CPAP, AHI immediately increases (Blankfield et al. 2004; Kohler et al. 2011). However, changes in oedema and overnight rostral fluid shift were not measured in relation to CPAP withdrawal.

    Central sleep apnoea

    CSA is common in patients with heart failure and is often associated with Cheyne–Stokes respiration, during which there is a gradual waxing and waning pattern of tidal volumes followed by central apnoea (Yumino & Bradley, 2008). Although CSA is also seen at high altitude or may be idiopathic, this review is limited to CSA in heart failure. In heart failure patients, risk factors for CSA include male sex, older age, atrial fibrillation, increased left ventricular filling pressure and end-diastolic volume, increased peripheral and central chemosensitivity to CO2, and hypocapnia (Yumino & Bradley, 2008). Overnight rostral fluid shifts may also contribute to the pathogenesis of CSA.

    During sleep, ventilation is largely dependent upon metabolic CO2 production and ambient inline image. Central apnoea occurs when inline image falls below the threshold required to stimulate respiration (i.e. the apnoea threshold), which usually increases from wakefulness to sleep. However, heart failure patients with CSA chronically hyperventilate, causing hypocapnia with inline image closer to the apnoea threshold than normal. Thus, even slight perturbations that augment ventilation, such as arousals from sleep, can trigger a central apnoea by driving inline image below the apnoea threshold. In heart failure patients with CSA, AHI is inversely proportional to inline image (Naughton et al. 1993).

    Mechanisms by which hyperventilation occurs in patients with heart failure include increased peripheral and central chemosensitivity, and stimulation of pulmonary vagal irritant receptors by pulmonary congestion due to elevated PCWP (Roberts et al. 1986; Yu et al. 1998; Javaheri, 1999; Solin et al. 2000). PCWP is, in turn, determined by severity of heart failure and degree of intravascular fluid volume overload (Schober et al. 1985; Nawada et al. 1993; Yu et al. 2005). For example, in heart failure patients, inline image is inversely proportional to PCWP (Lorenzi-Filho et al. 2002). In addition, patients with heart failure and CSA have higher PCWP than those without CSA and intensive medical therapy decreases both PCWP and AHI (Solin et al. 1999).

    In fluid overloaded patients with heart failure, particularly those with leg oedema, lying down at night causes rostral fluid displacement out of the intravascular and interstitial compartments of the legs due to gravitational effects (Thompson et al. 1928). The resultant increased venous return to the heart and thorax can lead to increases in PCWP, intrathoracic fluid content and extracellular lung water (Larsen et al. 1986; van Lieshout et al. 2005; Drazner et al. 2010).

    These factors appear to play a role in the pathogenesis of CSA. For example, the AHI in men with heart failure and CSA strongly correlated with the degree of overnight decrease in LFV (Fig. 4A) (Yumino et al. 2010). In addition, there was a gradation in the degree of overnight LFV reduction from no or mild sleep apnoea (–98 ml), to OSA (–235 ml), to CSA (–404 ml). As patients with CSA had similar overnight increases in neck circumference to those with OSA, the extra fluid from the legs is likely to have been redistributed to the lungs, since the greater the overnight decrease in LFV, the lower the nocturnal inline image and the higher the AHI. Presumably, fluid movement into the lungs stimulated pulmonary irritant receptors which provoked hyperventilation and a fall in inline image below the apnoea threshold.

    Furthermore, in heart failure patients, severity of CSA is reduced by raising the head of the bed, presumably by reducing fluid shift into the lungs, thereby reducing pulmonary congestion and ventilation, and allowing inline image to rise (Altschule & Iglauer, 1958; Soll et al. 2009). Although in one study thoracic fluid content did not change with altered sleeping angle, this finding contrasted with other studies in which thoracic fluid content increased from the upright to supine position in subjects without heart failure (Larsen et al. 1986; van Lieshout et al. 2005). Accordingly, further studies are required to determine whether fluid accumulation in the lungs contributes to the pathogenesis of CSA in heart failure.

    Alterations in sleep apnoea type

    Tkacova et al. (2001) observed that in heart failure patients with both OSA and CSA, sleep apnoea type shifted from predominantly obstructive to predominantly central from the beginning to the end of the night. This was associated with increases in periodic breathing cycle duration and circulation time, and a decrease in inline image. Since the former two measurements are inversely proportion to cardiac output, the shift from obstructive to central events occurred in association with a fall in cardiac output, whereas the reduction in inline image suggested an increase in PCWP (Hall et al. 1996; Lorenzi-Filho et al. 2002). Furthermore, amongst heart failure patients who underwent two sleep studies several months apart, conversion from predominantly OSA to CSA was associated with a decrease in nocturnal inline image and an increase in periodic breathing cycle duration, and vice versa (Tkacova et al. 2006). These findings suggest that in HF patients with left ventricular systolic dysfunction, the adverse cardiovascular effects of OSA are potent enough, in some cases, to lower cardiac output and raise PCWP sufficiently to cause a shift to CSA.

    In heart failure patients cardiac function can alter with changes in medication use, sodium and fluid intake and cardiac ischaemia. Deterioration in cardiac function is often accompanied by increased fluid retention and leg oedema, and vice versa. Therefore, the above observations suggest that in some heart failure patients, OSA and CSA are part of a shifting spectrum of sleep-related breathing disorders whose predominant type is influenced by cardiac function and degree of nocturnal rostral fluid shift.

    Conversely, however, patients with OSA who do not have heart failure do not appear to develop central apnoeas over the course of the night. Although left ventricular diastolic dysfunction has been observed in OSA patients, it is quite mild and insufficient to cause a large enough increase in PCWP to give rise to a HF syndrome with pulmonary congestion (Fung et al. 2002; Arias et al. 2005; Shivalkar et al. 2006). Furthermore, in OSA patients both with and without heart failure, the maximum fluid shift from the legs is about 300 ml, whereas in heart failure patients with CSA maximum fluid shift is about 600 ml (Redolfi et al. 2009; Yumino et al. 2010). In OSA patients there is no correlation between overnight fluid shift and inline image, unlike in those with CSA, where there is an inverse correlation (Yumino et al. 2010). Thus the magnitude of fluid shift associated with OSA appears to be insufficient to cause pulmonary congestion, stimulate pulmonary vagal irritant receptors and provoke hyperventilation and a fall in inline image below the apnoea threshold.

    Factors influencing overnight rostral fluid shift and sleep apnoea

    As described earlier, when standing still or sitting, the calf muscles are relatively inactive, hydrostatic pressure exceeds osmotic pressure and fluid moves from the capillaries into the interstitial space, forming leg oedema. Calf muscle contraction compresses the leg veins, whose one-way valves only permit blood flow towards the heart, resulting in increased venous return (Arnoldi, 1966). Accordingly, contraction of the calf muscles prevents pooling of fluid in the legs; therefore walking and calf muscle contraction while sitting prevent leg oedema (Winkel & Jorgensen, 1986a,b; Noddeland & Winkel, 1988; Stick et al. 1992).

    Increasingly, modern life is associated with less physical activity, owing to an increase in sedentary occupations and hence greater time spent sitting. In addition, as people grow older, they lead an increasingly sedentary life (Murphy et al. 2011). This lack of physical activity increases fluid accumulation in the legs during the day. Thus, inactive individuals may be predisposed to greater rostral fluid displacement from the legs when lying down at night and therefore greater fluid movement into the chest and neck. In this way, physical inactivity and sitting could play a role in the pathogenesis of both OSA and CSA. Indeed, in non-obese men the AHI was proportional to overnight decrease in LFV which was in turn proportional to sitting time during the day (Redolfi et al. 2009). Similarly, it was shown that in heart failure patients with OSA or CSA, AHI and overnight decrease in LFV were related directly to sitting time and degree of leg oedema, and inversely to physical fitness (Yumino et al. 2010).

    Age may also be a factor affecting rostral fluid shifts. In elderly women, there is increased capillary filtration into the interstitial tissues of the legs in response to lower body negative pressure (which simulates increased venous pressure) as compared to young women, although such changes have not been observed in men (Lanne & Olsen, 1997; Lindenberger & Lanne, 2007). Similarly, dependent fluid accumulation in the legs is more likely to occur in the elderly due to reduced activity and compromised function of the venous valves of the legs, which allows gravitational fluid accumulation (Schirger & Kavanaugh, 1966). Indeed, in non-obese, otherwise healthy men with OSA, overnight decrease in leg fluid volume correlated with age independently of other factors, including sitting time (Redolfi et al. 2009). These observations support the concept that older age increases the tendency for fluid shifts into the interstitial tissues of the legs while standing and back into the vascular compartment while recumbent during sleep.

    Epidemiological studies demonstrate that higher levels of exercise are associated with reduced prevalence and incidence of OSA, independently of BMI (Peppard & Young, 2004; Quan et al. 2007; Awad et al. 2012). The observation that supervised exercise reduces the AHI modestly in patients with OSA, without change in body weight provides evidence that lack of exercise contributes to OSA causation (Giebelhaus et al. 2000; Sengul et al. 2011; Awad et al. 2012). Similarly, in heart failure patients, exercise training reduced severity of OSA or CSA modestly (Yamamoto et al. 2007; Ueno et al. 2009). One possible reason for this, not tested, was that increased activity reduced daytime leg fluid accumulation, thereby reducing rostral fluid shift into the neck or chest overnight. Regular exercise decreases sympathetic activity in heart failure patients and, in animal studies, reduces sodium and water retention, which could reduce fluid retention in the legs during the daytime (Zanesco & Antunes, 2007; Patel & Zheng, 2012).

    Effect of interventions directed at fluid shifts on sleep apnoea

    In different patient populations, interventions that reduce nocturnal fluid shift consistently attenuate sleep apnoea. For example, wearing compression stockings prevents daytime fluid accumulation in the legs by reducing fluid filtration from the intravascular to the interstitial space via an increase in tissue hydrostatic pressure. In otherwise healthy men with OSA, and in patients with OSA and chronic venous insufficiency, wearing compression stockings for 1 day or 1 week, respectively, reduced AHI by a third, in association with attenuations in the overnight decrease in LFV and increase in neck circumference (Redolfi et al. 2011a; Redolfi et al. 2011b).

    In patients with drug resistant hypertension, treatment with the aldosterone antagonist spironolactone for 8 weeks, decreased AHI by almost 50%, in association with diuresis and reduced blood pressure (Gaddam et al. 2010). Similarly, intensive diuretic therapy in patients with diastolic heart failure and severe OSA increased UA-XSA and decreased the AHI modestly (Bucca et al. 2007).

    In patients with ESRD and sleep apnoea, conversion from conventional to nightly nocturnal haemodialysis significantly reduced the AHI (Hanly & Pierratos, 2001). However, in another similar study, nocturnal haemodialysis had no effect on the AHI but did cause an increase in UA-XSA in those both with and without sleep apnoea (Beecroft et al. 2008). In patients undergoing nocturnal peritoneal dialysis, the AHI was significantly lower than when they converted to continuous ambulatory peritoneal dialysis (CAPD), in association with greater nocturnal fluid removal, lower total body water, greater UA-XSA and reduced tongue volume (Tang et al. 2009). Finally, in patients with nephrotic syndrome, leg oedema and OSA, treatment of the nephrotic syndrome with steroids resolved oedema, reduced total body water, and reduced the AHI by 50% (Tang et al. 2012). However, in none of these studies on ESRD patients was overnight fluid shift measured.

    The commonest treatment for OSA, CPAP, is thought to alleviate OSA by acting as a ‘pneumatic splint’ of the UA (Sullivan et al. 1981). However, it may have other mechanisms of action. For example, when patients with OSA were treated with 4–6 weeks of nasal CPAP, UA-XSA increased proportional to the reduction in UA mucosal water content measured by MRI (Ryan et al. 1991). In another study, UA-XSA increased after just 1 week of CPAP (Corda et al. 2009). In a study of patients with heart failure and OSA, CPAP reversed OSA in association with attenuation of the overnight increase in neck circumference (Yumino et al. 2010). Together these data provide evidence that one mechanism by which CPAP alleviates OSA is by counteracting overnight fluid accumulation in the neck and pharyngeal mucosa by increasing intraluminal UA pressure, which opposes intravascular hydrostatic pressure.

    Influence of sex on sleep apnoea

    The prevalence of OSA is considerably lower in women than in men, although reasons for this have not been fully elucidated (Young et al. 1993). Compared to men, pharyngeal length in women is shorter and reflex UA compensatory responses to UA obstruction are stronger, both of which would tend to stabilise the UA of women during sleep and protect them from developing OSA (Malhotra et al. 2002; Chin et al. 2012). Differences in UA collapsibility, genioglossus activity, neck fat distribution and hormonal status have also been investigated as potential explanations for the difference in prevalence of OSA between the sexes, but none fully explain this difference (Lin et al. 2008).

    Differences in overnight rostral fluid shift between men and women may be a mechanism contributing to the higher male prevalence of OSA. In men with heart failure, overnight increase in both neck circumference and AHI were proportional to the overnight decrease in LFV (Fig. 4A) (Yumino et al. 2010; Kasai et al. 2012). In women with heart failure, however, despite a similar overnight decrease in LFV, the overnight increase in neck circumference was much smaller than in men, with no relationship between AHI and overnight change in LFV (Fig. 4B). Similarly, in healthy men and women, application of LBPP caused a similar decrease in LFV, but UA collapsibility increased more in men than women (Su et al. 2009). These studies suggest that a greater proportion of the fluid shifting from the legs accumulates in the neck in men than in women, which may therefore result in increased peripharyngeal pressure, decreased UA-XSA and increased UA collapsibility, predisposing to OSA.

    Reasons for differing patterns of overnight fluid redistribution between men and women are not clear. Anatomical factors may play a role. For example, women have large venous plexi around the uterus, ovaries and vagina, with no equivalent in men (Kachlik et al. 2010). These plexi may sequester fluid that has shifted from the legs overnight, preventing accumulation in the neck. This possibility is supported by the observation that women had greater pooling of blood in the abdomen during application of lower body negative pressure than men (White & Montgomery, 1996). Further studies of postural fluid shifts with measurement of fluid in the abdomen and neck may help to determine if abdominal pooling of fluid in women accounts for lower overnight fluid accumulation in the neck than men.

    Conclusion and future directions

    There is now considerable evidence that sedentary living and fluid retention in the legs during the daytime and its overnight rostral shift contribute to the pathogenesis of both OSA and CSA, at least in men. Further work is required to better characterise patterns of fluid shifts in OSA and CSA. This will require direct measurements of fluid volumes of the legs, abdomen, chest and neck. In particular, it would be important to determine whether greater fluid sequestration in the abdomen and pelvis of women contributes to their lower prevalence of OSA than men.

    Although a number of studies have measured fluid displacement from the legs, characterisation of the time course of these overnight fluid shifts to the chest and neck would provide further insight into the pathogenesis of OSA and CSA. For example, overnight conversion from OSA to CSA in heart failure patients may be related to fluid initially accumulating in the neck, then subsequently in the chest in association with decreasing cardiac output and increasing PCWP.

    Finally, treatment of OSA and CSA by various forms of positive airway pressure, such as CPAP, is poorly tolerated by many. Hence, alternative effective therapies are required. In view of the above, opportunities now abound to examine prevention of fluid retention in the legs during the day and its overnight rostral shift as a novel therapeutic target for sleep apnoea in the setting of larger, longer-term randomised clinical trials. Potential interventions include diuretics and aldosterone antagonists, sodium restriction, compression stockings, elevating the head of the bed, exercise interventions and ultrafiltration.



    This work was supported by Canadian Institutes of Health Research operating grant MOP-82731. L. H. White is supported by an unrestricted fellowship from Sleep Country Canada.