Enhanced chemo-responsiveness in patients with sleep apnoea and end-stage renal disease

Sleep apnoea has been reported in up to 50–70% of patients with end-stage renal disease (ESRD) 1, which is at least 10 times higher than the prevalence reported in the general population 2. The pathogenesis of sleep apnoea in patients with ESRD remains unclear. Although sleep apnoea is not corrected by conventional haemodialysis or peritoneal dialysis 3–5, it has been reversed both by nocturnal haemodialysis and kidney transplantation 6–8, indicating that its pathophysiology is uniquely associated with the development of chronic renal failure. Previous investigators have observed features of both central and obstructive sleep apnoea in patients with ESRD 3–9, which suggests that its pathogenesis is related to both destabilisation of central respiratory control and upper airway occlusion.

During nonrapid eye movement (NREM) sleep, respiratory drive operates predominantly under chemoreflex control 10. Changes in chemoreflex responsiveness can promote periodic breathing during sleep by destabilising central respiratory control 11. The chemoreflex response depends on the ventilatory recruitment threshold, which is the level of carbon dioxide tension (P_CO2) at which ventilation begins to respond to CO₂, as well as the sensitivity or slope of the ventilatory response above the threshold. Increased chemoreflex sensitivity can destabilise respiratory control by increasing the likelihood of lowering P_CO2below the “apnoeic threshold” 12. Increased threshold can destabilise respiratory control by increasing plant gain 11. Due to the metabolic hyperbola that models the control of P_CO2by pulmonary gas exchange, a high ventilatory threshold destabilises respiratory control as small increases in ventilation result in large decreases in P_CO2, which increases the likelihood of ventilatory overshoot. Periodic breathing can be associated with upper airway occlusion during sleep, possibly because of disproportionately greater drive to the inspiratory pump muscles than the upper airway dilators 13, 14. More recently, ventilatory instability has been reported in patients with severe obstructive sleep apnoea (OSA) reflected by a high “loop gain” 15, suggesting a role for central destabilisation in the pathogenesis of OSA. Chemoreflex responsiveness may be altered by the metabolic changes that accompany renal failure, such as metabolic acidosis 16, and possibly by uraemia or the accumulation of so-called “middle molecules”; therefore, it is likely that it contributes to the development of central or OSA or both. While altered chemoreflex responsiveness has been described in patients with ESRD 17–19, the association between chemoreflex responsiveness and sleep apnoea in such patients has not been determined.

Chemoreflex responsiveness changes throughout the 24-h cycle among healthy subjects 20, 21 and recent data from the current authors’ laboratory 22 suggest that chemoreflex responsiveness may change significantly overnight in patients with OSA, raising the possibility that changes in chemoreflex responsiveness are related to the intermittent hypoxaemia and/or sleep disruption that such patients experience. It is not known whether overnight changes in chemoreflex responsiveness are modified by the development of ESRD or the initiation of chronic haemodialysis or whether they differ between those who have sleep apnoea and those who do not.

The objectives of the present study were to measure chemoreflex responsiveness in patients with ESRD in a comprehensive and standardised manner, to assess whether it changes overnight and to determine whether responsiveness is altered in a manner that could contribute to the development of sleep apnoea. These questions were addressed by comparing measurements of chemoreflex responsiveness before and after overnight polysomnography in a large group of patients with ESRD. The measurements used included ventilatory sensitivity, which is a major determinant of respiratory control system stability, and the ventilatory recruitment threshold, which is a reflection of the acid–base status 23 and also determines plant gain. In addition, since the subjects were awake while these measurements were made, the sub-threshold ventilation was also measured. This basal ventilation or nonchemoreflex drive to breathe is sometimes called the wakefulness drive to breathe because it is absent during sleep 24. Therefore, basal ventilation reflects the state of awareness or arousal 25.

METHODS

Chemoreflex responsiveness

Chemoreflex responsiveness was measured using a modified 29–31 Read rebreathing technique 32. The modifications were as follows. 1) Hyperventilation for 5 min at the beginning of the test. This modification facilitated sub- and supra-threshold measures of ventilation, enabling the response threshold to CO₂ to be measured directly rather than by extrapolation. 2) Maintenance of isoxia throughout the test. This provided an estimation of the individual contributions of the central and peripheral chemoreflexes.

Measurements were obtained during isoxic hypoxia (partial pressure of oxygen (P_O2) 6.65 kPa) and isoxic hyperoxia (P_O2 19.95 kPa), and changes in ventilation were measured as the level of P_CO2rose. The main test parameters obtained were: sub-threshold or basal ventilation (V_Eb); ventilatory recruitment threshold for carbon dioxide (T); and supra-threshold slope, termed ventilatory sensitivity (S; fig. 1⇓). When the test is performed during hypoxia, the ventilatory response reflects the activity of the central and peripheral chemoreflexes, but when it is performed during hyperoxia, the ventilatory response predominantly reflects the activity of the central chemoreflex as hyperoxia attenuates the activity of the peripheral chemoreceptors 31. Since the contributions of the central and peripheral chemoreflexes are assumed to be additive, the sensitivity of the peripheral chemoreflex can be estimated by calculating the difference between the slopes of the ventilatory responses during hypoxic and hyperoxic conditions 33. All measures of chemoreflex responsiveness were performed during wakefulness, immediately before and after overnight polysomnography. In addition, blood was drawn from the radial artery the morning after polysomnography to determine arterial blood gases. A complete description of the rebreathing collection and analysis protocol has been previously published 33.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1—

Example of a typical chemoreflex response. The main test parameters obtained were: sub-threshold or basal ventilation (grey arrow); ventilatory recruitment threshold for carbon dioxide (black arrow); and supra-threshold slope termed ventilatory sensitivity (open arrow). P_CO2: carbon dioxide tension. 1 mmHg = 0.133 kPa.

RESULTS

Chemoreflex responsiveness

The results of the rebreathing tests are shown in table 5⇓. Within both groups, T was significantly lower while the S to P_CO2 ratio was significantly higher during hypoxic as compared with hyperoxic rebreathing tests (p<0.001). The current authors also noted that the rate of rise of carbon dioxide (rrCO₂) in the rebreathing bag, which is a function of resting metabolic rate, was significantly higher during hypoxic as compared with hyperoxic conditions (p<0.001) along with measures of V_Eb (p = 0.008). There was a significant overnight decrease in the rate of rrCO₂ and ventilatory recruitment threshold and a trend towards an overnight increase in ventilatory sensitivity, which did not reach statistical significance (p = 0.073). These changes were similar between apnoeic and nonapnoeic groups. There were no inter-group differences in V_Eb, ventilatory recruitment threshold or the rrCO₂. However, the ventilatory sensitivity to P_CO2 was significantly higher in the apnoeic group for both hypoxic and hyperoxic rebreathing tests. In addition, the difference in ventilatory sensitivity between hypoxic and hyperoxic rebreathing tests was significantly greater in the apnoeic group. Figure 2⇓ provides a graphical representation of these findings.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2—

Mean chemoreflex responses in the apnoeic and nonapnoeic groups under a) hyperoxic and b) hypoxic conditions. Sub-threshold basal ventilation and ventilatory recruitment threshold are similar, whereas the supra-threshold slope (ventilatory sensitivity) is greater among apnoeic patients under both conditions. Ventilatory recruitment threshold decreases overnight in both groups under both conditions. P_CO2: carbon dioxide tension. · · ·: apnoeic evening; –––: apnoeic morning; –– - ––: nonapnoeic evening; - - -: nonapnoeic morning. 1 mmHg = 0.133 kPa.

View this table:

• View inline
• View popup

Table 5—

Chemoreflex responsiveness to arterial CO₂ tension in apnoeic and nonapnoeic groups during isoxic hypoxia and hyperoxia

There was a significant positive correlation between hypoxic and hyperoxic ventilatory sensitivity and the severity of sleep apnoea, reflected by the AHI, and between hyperoxic ventilatory sensitivity and BMI (table 6⇓). Further analysis of these variables by multiple linear regression identified AHI and excluded BMI as significant predictors of ventilatory sensitivity (evening hypoxic: model r² = 0.104, AHI β = 0.322, p<0.05, BMI excluded; morning hypoxic: model r² = 0.205, AHI β = 0.453, p<0.005, BMI excluded; evening hyperoxic: model r² = 0.400, AHI β = 0.545, p<0.001, BMI β = 0.193, p = 0.096; morning hyperoxic: model r² = 0.275, AHI β = 0.437, p<0.005, BMI β = 0.168, p = 0.229). There was no significant correlation between age and ventilatory sensitivity. Independent of apnoea designation, ventilatory sensitivity did not differ between sexes. In apnoeic patients, the change in ventilatory sensitivity from evening to morning did not correlate with AHI (hypoxic: r = 0.156, p = 0.456; hyperoxic r = -0.026, p = 0.893). However, in nonapnoeic patients, the evening to morning change in hypoxic ventilatory sensitivity correlated with the AHI (r = 0.640, p = 0.014), whereas the change in hyperoxic sensitivity did not (r = -0.069, p = 0.793).

View this table:

• View inline
• View popup

Table 6—

Pearson correlations

DISCUSSION

A high prevalence of sleep apnoea was found in patients with ESRD, which is consistent with previous reports in this patient population 3–6, 9. The majority of the apnoeas and hypopnoeas observed were obstructive events, which contrasts with many previous studies that reported a more heterogeneous apnoea profile consisting of central, mixed and obstructive events 4–6, 8. The current study patients with sleep apnoea were heavier, older and were predominantly male, which may explain the predominance of obstructive events. However, these differences were small and neither BMI nor age were predictors of apnoea severity when included in regression analysis models. Nevertheless, these findings support the notion that sleep apnoea is common in patients with ESRD and that its pathogenesis is influenced by factors that are uniquely associated with ESRD over and above the mechanisms that cause sleep apnoea in the general population.

One of these potential mechanisms is altered respiratory chemoreflex responsiveness. During NREM sleep, respiratory drive operates predominantly under chemoreflex control 10. Increased dependence on chemical control makes the system more vulnerable to changes in the level of arterial P_CO2 in response to ventilatory stimuli. Increased chemoreflex threshold and/or sensitivity can lead to ventilatory instability and oscillations in respiratory output by increasing the likelihood of lowering P_CO2below the apnoeic threshold 12. High ventilatory threshold destabilises respiratory control because a small increase in ventilation causes a large decrease in P_CO2 11. High ventilatory sensitivity destabilises respiratory control by facilitating an increased ventilatory response to sufficient ventilatory stimuli 11. In other words, increased ventilatory threshold and/or sensitivity lead to ventilatory instability by increasing loop gain 11.

In the present study, apnoeic patients were found to have a higher ventilatory sensitivity to hypercapnia during both hypoxic and hyperoxic rebreathing tests. The ventilatory sensitivity also showed a significant positive correlation with apnoea severity. These data indicate that the sensitivity of both the central and peripheral chemoreceptors is increased in patients with sleep apnoea and ESRD, which the current authors anticipate destabilises respiratory control by increasing feedback and, therefore, loop gain 11. Altered chemoreflex responsiveness has been well described in the pathogenesis of central sleep apnoea, including idiopathic central sleep apnoea 34, apnoea at high altitude 35 and Cheyne–Stokes respiration in patients with congestive heart failure 36. Most of the apnoeas and hypopnoeas in the study patients were classified as obstructive events according to conventional polysomnographic criteria. However, there is abundant evidence that oscillation in respiratory control does occur in patients with OSA. Periodic breathing was observed over 20 yrs ago following tracheostomy for the treatment of severe OSA 37, 38 and, more recently, ventilatory instability reflected by a high loop gain, has been reported in patients with severe OSA 15. Furthermore, there is evidence that upper airway closure can occur during central sleep apnoea. Airway narrowing and closure was identified by bronchoscopic evaluation in healthy individuals with central apnoea that occurred spontaneously or was induced experimentally by transient hyperventilation 39. Earlier studies attributed this phenomenon to disproportionate activation of the inspiratory muscles, which was greater or earlier than activation of the upper airway dilator muscles 14. Others studies have reported that upper airway closure is more likely to occur during central apnoea in individuals whose upper airway is functionally or anatomically predisposed to collapse during sleep 40. Consequently, the increased ventilatory sensitivity that was observed during wakefulness in the present study’s patients may have led to ventilatory instability and obstructive apnoea during sleep.

The mechanisms responsible for increased ventilatory sensitivity among the apnoeic study patients are unclear. Chemoreflex responsiveness can be increased by metabolic acidosis; acetazolamide-induced metabolic acidosis enhances central chemosensitivity and reduces the ventilatory threshold to hypercapnia 41, 42. However, the current blood gas data did not indicate acidosis, at least in the first 24 h after dialysis. Moreover, there was no difference in pH between apnoeic and nonapnoeic patients (table 4⇑), nor P_t,CO2measured during sleep (table 2⇑). Furthermore, the ventilatory threshold to hypercapnia during isoxic hypoxia and hyperoxia did not differ between groups.

A single previous study reported that dialysis and uraemia may be associated with increased chemoreflex responsiveness 18. It was suggested that removal of urea from the blood causes an influx of water into the brain where clearance of urea is slower, which may cause a “dilutional acidosis” at the central chemoreceptor, thereby increasing central chemoreflex sensitivity and threshold. However, no difference was found in either the severity of uraemia nor in the adequacy of dialysis in apnoeic patients compared with those without apnoea (table 1⇑), suggesting that augmented chemoreflex responsiveness is not related to dialysis or uraemia in the study patients. In addition, there were no significant differences in chemoreflex responsiveness between patients on CHD and PD, which suggests that increased ventilatory sensitivity is not due to changes in brain osmolality since PD does not lower urea like conventional haemodialysis does. The use of acetate buffer has been reported to augment apnoea severity in haemodialysis patients and has been attributed to alteration of ventilatory control 43. However, all of the patients received bicarbonate buffer dialysis.

In the nonrenal failure population, OSA is associated with altered chemoreflex responsiveness 44. This literature found increased ventilatory threshold and sub-threshold ventilation among apnoeic patients with no difference in ventilatory sensitivity, which differs from the current findings. While it is possible that the present findings were due to inherent differences between the apnoeic and nonapnoeic groups that were not related to renal failure, this seems unlikely, given the high prevalence of sleep apnoea in ESRD. It is also possible that increased ventilatory sensitivity is an adaptive response to the frequent bouts of hypoxia 45, hypercapnia and/or arousals associated with OSA. If this were true, greater overnight changes in chemoreflex measures in the apnoeic patients would have been anticipated, but this was not the case. Consequently, the current authors believe that the differences in chemoreflex responsiveness that were observed between apnoeics and nonapnoeics were due to the presence of renal failure. It is possible that increased ventilatory sensitivity is a consequence of factors that occur in chronic renal failure which were not measured in the present study, such as changes in autonomic tone 19 and/or the clearance of medium-sized molecules.

Significant evening to morning changes were also found in the chemoreflex responsiveness measurements. Specifically, the current authors found a decrease in the rrCO₂ and ventilatory threshold overnight and a trend towards an overnight increase in ventilatory sensitivity that did not reach statistical significance, indicating an overnight increase in chemoreceptor activity. However, these changes were not specific to either the hypoxic or hyperoxic rebreathing test responses nor were they specific to apnoeic or nonapnoeic patients. Although the decline in the rrCO₂ during the rebreathing tests and the overnight increase in ventilatory sensitivity would tend to destabilise respiratory control, these effects would be opposed by the overnight decrease in ventilatory threshold, which tends to stabilise the system. Previous reports from the current authors’ laboratory in the nonrenal failure population 22 demonstrated an overnight decrease in ventilatory threshold that was confined to nonapnoeic patients and an overnight increase in ventilatory sensitivity that was confined to patients with OSA. The current findings are consistent with some previous studies', assessing the effect of circadian rhythm on chemoreflex control, which showed an evening to morning decline in metabolic rate 46 and an increase in chemoreflex sensitivity 20, but in contrast to others showing an evening to morning increase in ventilatory threshold 21. It is not possible to determine how much of the overnight changes in chemoreflex responsiveness in the present study were due to renal failure, sleep apnoea or circadian influences, although the positive correlation (r = 0.640, p = 0.014) between AHI and hypoxic ventilatory sensitivity in those with AHI <10 events·h^-1 (nonapnoeic group) suggests that sleep apnoea may have some impact in patients with mild apnoea.

The study has some limitations that should be addressed. First, measurements of chemoreflex responsiveness were carried out during wakefulness and correlated with other respiratory measurements that were performed during sleep. The current authors acknowledge that the interaction between chemoreflex responsiveness and sleep onset plays a pivotal role in the development of sleep apnoea. However, it is believed that the manifestations of altered respiratory control observed during wakefulness continued during sleep and, therefore, contributed to the pathogenesis of sleep apnoea. Secondly, chemoreflex responsiveness varies significantly between individuals, which can make it difficult to find significant differences between groups of subjects. The fact that the present authors were able to find significant differences between apnoeics and nonapnoeics, despite this inherent variability in the measurement, makes the current findings more robust. Finally, the current study design was cross-sectional and consequently cannot infer causality between augmented chemoreflex responsiveness and sleep apnoea. Nevertheless, as outlined in the previous paragraphs, it is believed that the changes in chemoreflex responsiveness were not a consequence of OSA and that, more likely, they contributed to its pathogenesis.

In conclusion, sleep apnoea in end-stage renal disease is associated with an increase in respiratory chemoreflex sensitivity to hypercapnia during both hypoxic and hyperoxic conditions. These data indicate an increase in the sensitivity of both the central and peripheral chemoreflex responses. Increased chemoreflex sensitivity destabilises respiratory control, and so may contribute to the pathogenesis of sleep apnoea in end-stage renal disease. Further studies are required to confirm a causal relationship between increased chemoreflex sensitivity and obstructive sleep apnoea in this patient population, which the current authors anticipate will provide the basis for new therapeutic options.