There are three major types of sleep-related breathing disorders (SRBD): 1) obstructive sleep apnea (OSA), which is caused by airway obstruction; 2) central sleep apnea (CSA), which is caused by a decrease or cessation of stimulation from the respiratory center in the medulla oblongata; and 3) mixed sleep apnea (MSA), which involves the 2 types, with the central pattern preceding the obstructive pattern. OSA accounts for the majority of SRBDs and its onset had been proposed to be affected by the following four factors, according to White et al. [1]: 1) anatomical factor, which is caused by narrowing of the pharynx; 2) instability of the respiratory center; 3) decreased pharyngeal reflex in sleep or aging; and 4) low arousal threshold, which results in a greater likelihood of waking up [1-5]. The regulation of respiratory function comprises 3 components, including chemical, neural, and behavioral controls; the effects of the chemical control are known to relatively increase with the induction of sleep [6]. In chemical control, the central chemoreceptors in the medulla oblongata detect the partial pressure of carbon dioxide (PaCO₂) in the arterial blood to control breathing. When the PaCO₂ increases above the normal level of 40 mm Hg, the respiratory center sends signals to increase minute ventilation, in order to eliminate the carbon dioxide (CO₂). Conversely, when the PaCO₂ drops below 40 mm Hg, the stimulation from the respiratory center stops, and CSA is induced [7,8]. In MSA, respiration attempts to resume when the PaCO₂ increases due to CSA, but the reflex activity of the upper airway dilator muscles is inhibited or decreases, resulting in obstruction and constriction of the upper airway, followed by the appearance of OSA [9].

Polysomnography (PSG) is a technique to diagnose and assess the severity of SRBD. However, it cannot assess the causative factors, such as the shape or collapsibility of the upper airway, or quantitatively evaluate the function of respiratory regulation. The maintenance of normal partial pressure of oxygen (PaO₂) in arterial blood and PaCO₂ is controlled by the respiratory regulation function, which can be assessed through ventilatory response examinations. Hypoxia and hypercapnia are the two types of ventilatory stimulation [10-12]. Kronenberg et al. [13] reported that the ventilatory response to hypoxia was weaker in older individuals than in younger individuals. They also investigated the ventilatory response to hypercapnia by maintaining PaO₂ at ≥ 200 mm Hg through oxygen inhalation, in order to eliminate the effects of hypoxia. As a result, they found that the ventilatory response to hypercapnia, similar to hypoxia, as less in older individuals than in younger individuals.

Due to the fact that PaCO₂ is responsible for the chemical control of respiration, evaluation of the ventilatory response to CO₂ (hypercapnic ventilatory response [HCVR]) is important [14,15]. When the changes in HCVR during the transition from the awake to the sleep states are large, CSA occurs due to diminished PaCO₂. Small changes in HCVR are accompanied by periodic breathing, in which there is repeated fluctuation in tidal volume [16,17]. In recent years, reduction in HCVR had been implicated in the pathophysiology of pediatric and obese adolescent (age, 12–16 years) OSA [18-20]. However, in adult patients with OSA, awake HCVR had been reported to be augmented, decreased, or similar, compared with that in healthy adults, indicating the absence of a consensus [19,21-24].

Based on the above reports, we hypothesized that HCVR could be an index of respiratory response and that worsening of the HCVR increases the appearance of CSA and MSA, resulting in the exacerbation of OSA. The purpose of this study was to investigate awake HCVR in order to elucidate the association of HCVR in SRBD with the severity and pattern of respiratory impairment in adult OSA.

In order to compare the effect of patient characteristics on the CO₂ ventilatory response, awake HCVR was assessed in terms of AHI, ODI, age, and BMI. Patients were stratified into four OSA groups in terms of AHI: mild (n = 100), moderate (n = 182), severe (n = 174), and very severe (n = 196) (Tables 1 and 2). Patients were stratified into four ODI groups as follows: mild (ODI < 15/hr), moderate (15/hr ≤ ODI < 30/hr), severe (30/hr ≤ ODI < 50/hr), and very severe (ODI ≥ 50/hr) (Table 2). The four age groups were 20–39 years (n = 81), 40–59 years (n = 273), 60–79 years (n = 272), and ≥ 80 years (n = 26). The four BMI groups were < 20 kg/m², ≥ 20–< 25 kg/m², ≥ 25–< 30 kg/m², and ≥ 30 kg/m². In order to compare the percentages of apnea events among the central, mixed and obstructive patterns, the patients were divided into three groups according to AHI (mild, n = 100; moderate, n = 182; and severe, n = 370) and into three groups based on HCVR, including the low (< 0.5 L/min/mm Hg, n = 24), normal (≥ 0.5–< 2.7 L/min/mm Hg, n = 498), and augmented groups (≥ 2.7 L/min/mm Hg, n = 130). The study population had a mean age of 58 ± 14 years (range, 20–86 years), and the mean values were 26.7 ± 5.1 kg/m² (range, 18.07–62.94 kg/m²) for BMI; 38.7 ± 23.8 for AHI; and 1.9 ± 1.2 L/min/mm Hg for HCVR slope. The number of patients by OSA severity was 100 in the mild group, 182 in the moderate group, 174 in the severe group, and 196 in the very severe group. There was no significant difference in age among the groups. With increasing OSA severity, the BMI increased, the longest duration of apnea increased, and the minimum SpO₂ decreased (Table 1). In the analysis by apnea severity, very severe OSA had augmented awake HCVR, compared with that in mild or severe OSA. Awake HCVR was augmented in the ODI > 50/hr group, compared with that in the ODI < 49/hr group (Table 2). Awake HCVR did not significantly differ by age, but it was significantly augmented in the morbidly obese OSA group than in the thin OSA group (Table 2). The frequency of CSA was significantly less in the severe and very severe OSA groups than in the moderate OSA groups (Table 3). The frequency of MSA was significantly higher in the very severe OSA group than in the mild, moderate, and severe OSA groups and was significantly higher in the severe OSA group than in the mild OSA group (Table 3). However, regardless of the severity, the frequency of OSA was not significantly different among the groups (Table 3). The percentages of CSA and OSA were not significantly different among the low, normal, and augmented awake HCVR groups, but the percentage of MSA was significantly higher in the augmented awake HCVR group than in the low awake HCVR group (Table 4). In order to evaluate the effect of age, BMI, AHI and ODI on HCVR, the patients were divided into two groups using cutoff values of 2.7 L/min/mm Hg for HCVR, 60 years old for age, 25 kg/m² for BMI, 60/hr for AHI, and 30/hr for ODI, and univariable and multivariable analysis of HCVR were performed in terms of these risk factors. BMI had a significant effect on HCVR in both univariable and multivariable analysis. The ODI tended to affect HCVR only in univariable analysis (Table 5). Furthermore, when we divided the patients into three groups using HCVR cut-off values of 0.5 and 2.7 L/min/mm Hg (normal HCVR versus low HCVR group, normal HCVR versus augmented HCVR group), and performed univariable and multivariable analysis of HCVR in a similar way, BMI was seen to affect HCVR in both univariable and multivariable analysis in the comparison between the normal HCVR group and the low HCVR group, and the normal group and the augmented HCVR group (Table 5).

In this study, we found that awake HCVR was augmented in very severe OSA, compared with that in mild and severe OSA, and that it was augmented in those with ODI > 50/hr than in those with ODI < 49/hr. A higher AHI resulted in greater reduction in CSA and greater increase in MSA. The percentage of OSA appearance was not different, regardless of the AHI. On the other hand, the percentage of MSA appearance was increased in the augmented HCVR group; this was thought to be related with the changes in the clinical state of OSA. Multivariable analysis of HCVR showed that BMI was an independent risk factor affecting awake HCVR.

Patients with untreated very severe OSA have abnormal HCVR because of the recurring apnea events at night and the associated elevation in PaCO₂, as well as the decrease in PaCO₂ with the resumption of respiration after the apnea. Based on our results (Table 2), a greater severity of OSA could lead to repeated ventilation abnormalities in response to changes in PaCO₂ during sleep, thereby, inducing unstable chemical control of respiration and, ultimately, the augmentation of awake HCVR [29].

Awake HCVR was significantly greater in OSA patients with BMI ≥ 30 kg/m² than in those with BMI < 20 kg/m² (Table 2), and the group with high BMI had high awake HCVR (Table 5); this implied augmentation in awake HCVR due to obesity. This finding was in line with the results of the study by Narkiewicz et al. [30], who showed that HCVR was enhanced in the obese group [30-32]. This augmentation in HCVR may be explained by the synergistic action between increased PaCO₂ and decreased PaO₂ in the carotid bodies during HCVR measurement, because PaO₂ declines further in obese patients [33]. Moreover, the leptin called “obesity gene,” which is a powerful appetite suppressant, increases with an increasing degree of obesity [34]. Leptin regulates the central chemical reflex and its plasma concentration is known to be positively correlated with HCVR [22]. Because increased leptin is observed with sympathetic nerve activation in OSA [35], it is possible the leptin elevation that accompanies the increase in BMI may also be associated with the augmentation of HCVR.

Kronenberg et al. [13] reported that the ventilatory response to hypoxia and hypercapnia was decreased in older individuals. The present study did not show aging-related reduction in HCVR; older adults maintained a level of HCVR that was similar to that in young adults (Table 2). Binary logistic analysis of HCVR with age, BMI, AHI and ODI showed that only BMI affected HCVR (Table 5). Because HCVR was augmented in obese individuals, compared with that in non-obese individuals, the high BMI in our study patients was presumed to have suppressed the effects of aging on diminishing the HCVR [33]. We postulated that in older obese patients with OSA, the awake HCVR did not decline, the sleep HCVR was maintained or even augmented, the respiratory response was augmented by a slight change in CO₂, the respiration became unstable, and the MSA appeared more frequently. On the other hand, age-associated decreases in the ventilatory response are brought about by a decline in respiratory muscle strength and hardening of the rib cage; consequently, these do not simply reflect respiratory center activity. To correct for the reduction in respiratory muscle strength in the elderly, Peterson et al. measured airway occlusion pressure, P0.1, which is determined as the change in mouth pressure within 0.1 second from the start of inspiration. As a result, they found that the ventilatory responses to hypoxia and hypercapnia were significantly reduced in older individuals, even after correcting for the decreased respiratory muscle strength [36,37]. Khoo [38] described the response to ventilatory changes as the “loop gain (LG) theory”; specifically, the respiratory response to changes in ventilation is large and unstable with a large LG, but is small and stable with a small LG. In OSA with a high LG, a hyper-responsive state arises when breathing resumes after apnea, and patients become prone to recurring cycles of hyperventilation and hypoventilation, such as in CSA or Cheyne-Stokes respiration. Dernaika et al. [39] reported that patients with untreated OSA have a low apnea threshold for PaCO₂ and unstable respiration in response to changes in PaCO₂, although the condition improved after about 1 to 3 months of continuous positive airway pressure (CPAP) [39-41].

In the present study, we did not observe a significant difference in the percentage of OSA, regardless of apnea severity. On the other hand, CSA appeared less frequently and MSA appeared more frequently in severe and very severe OSA than in mild and moderate OSA (Table 3). To our knowledge, there had been no previous reports that investigated the frequency of CSA or MSA in patients with OSA. HCVR is known to exacerbate with the worsening of OSA [39-41] and that HCVR exacerbation contributes to the onset of CSA [16,17]. These were contradictory to our findings. However, the BMI was high in the severe and very severe groups (Table 1). This difference in body frame may have affected the frequency of OSA in the severe and very severe groups [5], thereby, inducing the changes in the frequency of MSA.

Compared with the low HCVR group, the augmented HCVR group had a significantly greater frequency of MSA appearance. The augmented HCVR group also had more frequent CSA and fewer OSA compared with the low HCVR group, albeit without statistical significance (Table 4). According to a previous report, OSA can appear with the recovery of PaCO₂, potentially resulting in MSA [42]. Moreover, in patients with OSA, the difference in PCO₂ levels between resting respiration and apnea threshold had been reported to be chronically reduced and that there was augmented ventilatory response by the chemoreceptor to changes in PCO₂ [43]. These above reports suggested that the transition to the CSA was associated with the elevation in HCVR. However, there had been no previous reports on HCVR elevation and the accompanying transition from OSA to MSA. Indeed, the results of the present study implicate the values of HCVR measurements.

There were several limitations to this study. First, a large proportion of the study patients were middle-aged or older (Table 1). Since aging is known to modulate sleep respiration, the HCVR in this study may have been affected by the age of the patients. Future studies that include HCVR measurements in young adults are necessary. Second, although the HCVR measurements in this study were performed after confirming that there was no decrease in SpO₂ in room air, but unlike Kronenberg et al. [13], PaO₂ was not maintained at 200 mm Hg or higher. The slight changes in PaO₂ due to changes in ventilation volume during measurement may have affected the outcomes. Third, ventilatory impairment may have induced changes in the ventilatory response to hypercapnia. Finally, patients with cardiovascular or cerebrovascular disease, which may affect the ventilatory response, were not completely excluded. Therefore, investigations in OSA patients without underlying diseases are necessary.

In conclusion, this study suggested that measurements of awake HCVR could become a predictor of the transition to MSA and CSA in patients with OSA. Awake HCVR is a potential index of OSA progression and a convenient tool to assess the effects of treatment, such as outpatient CPAP.