Abstract
Possible mechanisms of arousal from respiratory stimuli include changes in Po 2, Pco 2, central respiratory drive, or respiratory mechanoreceptor activity. We sought to determine whether hypercapnia alone could induce arousal from sleep in four subjects with high ( ⩾ C3) neurologically complete spinal cord injuries while on constant positive pressure mechanical ventilation (hence, respiratory mechanoreceptor activity remained constant). Subjects were chronically hypocapnic (mean baseline Pet CO2 = 21 mm Hg; range, 13–30 mm Hg). On the first night, the baseline rate of spontaneous awakenings was determined by polysomnography. On night two, Fi CO2 was increased rapidly in stable NREM sleep. Awakenings occurred in 19 of 19 trials within 5 min, with each subject waking and complaining of shortness of breath (mean time to arousal, 115 s; range, 26–264 s). It is unlikely that these were spontaneous, as the times to awakening during hypercapnia were much higher than during baseline conditions (p < 0.05). During rapidly induced hypercapnia, Pet CO2 overestimates the Pco 2 at the central chemoreceptors. To determine more precisely the Pet CO2 arousal threshold, Pet CO2 was increased slowly (approximately 2 mm Hg/min); arousal occurred at a mean Pet CO2 of 37 mm Hg (range, 23–45 mm Hg; mean change from baseline, 15.8 mm Hg, range, 10–20 mm Hg). Hence, both rapid and slow increases in Pet CO2 can induce arousal in humans in the absence of changes in respiratory mechanoreceptor activity.
The repetitive arousals that fragment sleep in obstructive sleep apnea (OSA) have a variety of adverse consequences including daytime sleepiness (1). The mechanisms whereby obstructive apneas cause arousals are not completely clear. This is because during an obstructive event there are simultaneous changes in central respiratory drive, respiratory and oropharyngeal mechanoreceptor activity, arterial blood gases, and intrapleural pressure. It is, therefore, difficult to determine which of these changes contributes to the arousal response.
Prior work in healthy subjects has shown that arousal to a variety of respiratory stimuli (hypercapnia, hypoxemia, and resistive loading) occurs at a similar level of negative intrathoracic pressure within a particular subject (2). This has been interpreted as suggesting that afferents from respiratory mechanoreceptors or central respiratory drive play an important role in the arousal mechanism (3).
We sought to determine whether blood gas changes alone could induce arousal in the absence of changes in respiratory muscle mechanoreceptor afferent information (emanating from Golgi tendon receptors, muscle spindles of the diaphragm and chest wall, or from intrapulmonary stretch or pressure receptors). To this end, we studied the arousal response to hypercapnia in subjects who were incapable of changing their intrapleural pressures or the tension within their respiratory muscles during induced hypercapnia. Subjects had high (above the third cervical level) neurologically complete spinal cord injuries and received fixed positive pressure mechanical ventilation.
To reduce the confounding effect of spontaneous arousals, we chose the conservative approach of using full awakenings from sleep as our measure of arousal. This is because such awakenings are less likely to occur spontaneously than brief, subtle changes in electroencephalogram (EEG) that others have used to define arousal (4).
Four subjects with chronic neurologically complete spinal cord injuries above the C3 level were recruited. All subjects had chronic tracheostomies, respiratory muscle paralysis, and a complete loss of sensation below C3. Neurologic examinations confirmed the level and completeness of injury. Their lesions necessitated the use of either a phrenic nerve pacemaker (Subject 1) or positive pressure mechanical ventilation (Subjects 2–4) for ventilatory support. All subjects were hospitalized chronically but none had an acute respiratory problem at the time of the study. All procedures were approved by the appropriate institutional ethics review committee and informed consent was obtained from the subjects.
To verify respiratory muscle paralysis, maximal voluntary inspiratory pressures and vital capacities were determined (Table 1). Vital capacities were measured with a Wright spirometer in line with the tracheostomy tube.
|
| Subject | ||||||||
|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | |||||
| Age, yr | 48 | 42 | 78 | 31 | ||||
| Sex | M | F | F | M | ||||
| Reason for injury | Fall | Astrocytoma | Syringomyelia | Gunshot | ||||
| Medications | Diazepam (5 mg qid); baclofen (25 mg qid) | Propoxyphene N100 (1 tab tid); baclofen (15 mg qid) | Diazepam (2 mg bid) | Dantrolene (100 mg qid); baclofen (20 mg 5×/day); MS Contin (60 mg tid) | ||||
| Spontaneous Pi max, cm H2O | 0 | −1.3 | −2 | −9 | ||||
| Vital capacity, ml | 0 | 0 | 0 | 57 | ||||
| Usual tidal volume, ml | 1,200 | 900 | 1,000 | 900 | ||||
| Usual breathing rate, breaths/min | 14 | 12 | 10 | 10 | ||||
| Baseline bicarbonate concentration, mEq/L | 22 | 15 | 19 | 18 | ||||
| Baseline awake Pet co2, mm Hg | 30 | 13 | 21 | 20 | ||||
Definition of abbreviations: bid, tid, qid = two, three, four times daily, respectively; Pet CO2 = end-tidal carbon dioxide pressure; Pi max = maximal inspiratory pressure.
To reduce the likelihood of arousal arising from other than the experimental protocol, polysomnography was performed in the usual hospital bedroom of each subject rather than in the unfamiliar setting of a research laboratory. Ear plugs and headphones were used so subjects would not be awakened by environmental noise. For each subject (including the phrenic nerve-paced subject), polysomnograms were performed during fixed positive pressure mechanical ventilation with the cuff of the tracheostomy tube inflated to prevent gas leaks. For the polysomnogram, four pairs of electroencephalogram leads (C3–A2, C4–A1, FP1–A2, O2–A1), a pair of chin electromyogram (EMG) leads, two pairs of electrooculographic leads, and a pair of electrocardiogram leads were attached to each subject. In Subject 1, a pair of leads was also placed on the skin overlying the right sternocleidomastoid muscle (an accessory respiratory muscle innervated by a cranial nerve and thus spared by spinal cord lesions). Arterial oxygen saturation was measured continuously with a finger oximeter (Sat-Trak; Sensormedics, Yorba Linda, CA). A capnometer was placed in line with the tracheostomy tube to measure end-tidal carbon dioxide levels (Pet CO2 ) and the concentration of inhaled carbon dioxide. Chest wall movement associated with tidal inflation and deflation was monitored by inductance plethysmography (resp-EZ bands; EPM Systems, Midlothian, VA). All physiologic variables were digitized and recorded continuously (Somnotrac 4250; Sensormedics).
Night 1: baseline night. To ascertain the rate of spontaneous awakenings (defined below), a baseline polysomnogram was performed. Passive mechanical ventilation was instituted at the usual nocturnal rate and tidal volume of each subject (Table 1). All subjects utilized a passive moisture exchanger in the ventilator tubing to ensure comfortable inspiratory gas humidity. Supplemental oxygen was added to the inhaled gas mixture to keep the oxygen saturation greater than 95%. In Subject 1 it was necessary for additional carbon dioxide to be added to the inhaled gas mixture to achieve a Pet CO2 level that was the same as his usual nocturnal level while being ventilated with his diaphragm pacemakers. This was because of presumed differences in regional ventilation and perfusion between positive pressure and diaphragm-paced ventilation.
Night 2: addition of carbon dioxide. On a subsequent night, the procedures as outlined above were repeated. On this night, we also determined: (1) whether carbon dioxide induced arousals, and (2) at what level of Pet CO2 arousal occurred. To do this, carbon dioxide was mixed into the inspiratory line of the ventilator from a cylinder containing 10% carbon dioxide, 50% oxygen, and 40% nitrogen. During all the trials, tidal inflation volume and respiratory rate were kept constant.
During the first portion of the night, we assessed whether hypercapnia induces arousal. To do this, relatively large and rapid increases in Pet CO2 were used. After 2 min of consolidated nonrapid eye movement (NREM) sleep (defined as Stage 2, 3, or 4), the concentration of inhaled carbon dioxide was increased steadily over 1 min to achieve Pet CO2 levels roughly between 43 and 55 mm Hg (up to 30–35 mm Hg above baseline). After 5 min, or sooner if an awakening occurred, the concentration of inhaled CO2 was reduced to zero until the subject again experienced 2 min of consolidated NREM sleep and the hypercapnic trial was repeated. Two to eight trials of rapid increases in CO2 were performed with each subject. The time required to cause awakening from the onset of the change in Pet CO2 , and the Pet CO2 at which awakening occurred were recorded.
During rapidly induced hypercapnia, Pet CO2 overestimates the Pco 2 at the central chemoreceptors. Hence, once it had been determined that hypercapnia caused awakenings, to determine more precisely the Pet CO2 arousal threshold, in the second half of the night Pet CO2 was increased slowly (approximately 2 mm Hg/min) from baseline until arousal. After each arousal, the inhaled concentration of carbon dioxide was decreased by 5–6 mm Hg (i.e., slightly below the arousal threshold but above the baseline Pet CO2 ) to allow the subject to return to sleep. Once consolidated NREM sleep was reestablished, Pet CO2 was again increased at 2 mm Hg/min. Five to seven trials were performed with each subject.
Sleep was staged according to established criteria (5). Awakenings were defined as a change from an asleep to an awake alpha pattern EEG that lasted more than 15 s and was associated with a change in chin EMG tone. On Night 2, the lag time between the start of CO2 inhalation and awakening was determined. From Night 1, all awakenings occurring after more than 2 min of consolidated NREM sleep at the same time period after sleep onset were examined; the lag time between the end of 2 min of consolidated sleep and awakening was recorded. This was considered to represent the time to awakening under baseline conditions. To assess whether subjects on Night 2 aroused to carbon dioxide per se, the mean baseline time to awakening in each subject was compared with the mean awakening time under hypercapnic conditions, using a paired t test.
A characteristic polysomnogram tracing showing the awakening response to carbon dioxide is shown in Figure 1. In our study, hypercapnia consistently caused awakening in all subjects. During rapid increases of Pet CO2 , awakenings occurred within 5 min in 19/19 trials (mean time to awakening, 115 s; range, 26–264 s). In addition, on waking, each subject signaled being short of breath at least once during these hypercapnic challenges (prearranged distress signals included making clicking sounds with the tongue against the roof of the mouth, which was still possible in these ventilated subjects). The times to awakening during rapid increases in Fi CO2 (fraction of inspired CO2) were compared with those under baseline conditions (Figure 2). For each subject, the mean time to awakening during CO2 inhalation was much less than the mean time to spontaneous awakenings derived from Night 1 (p < 0.05).
Fig. 1. (A) Representative 4-min segment of a polysomnogram recording illustrating an awakening (asterisk) occurring after Pet CO2 was increased. The beginning of the hypercapnic trial is marked by a caret (^). (B) Enlargement of the polysomnogram to better show the awakening from sleep (i.e., the 30-s segment as marked in [A]). From top to bottom, the following physiologic information is shown: electroencephalogram (EEG) (C3/A2), left electrooculogram (EOG), right electrooculogram (EOG), carbon dioxide level (in mm Hg) of inspired and expired gas (CO2), oxygen saturation (O2), and electrocardiogram (ECG).
[More] [Minimize]
Fig. 2. The horizontal axis represents the subject number (1 through 4) as described in Table 1. For each subject, the time to each spontaneous awakening (derived from the first night) is illustrated on the left side (stippled columns). On the right side, for each hypercapnic trial the time between the start of CO2 inhalation and awakening is shown (derived from Night 2, open columns). The times to awakenings were much shorter during the trials of hypercapnia than under baseline conditions.
[More] [Minimize]The Pet CO2 arousal threshold was determined more precisely by slowly increasing Pet CO2 . In this part of the study, arousal occurred between 10 and 20 mm Hg above awake baseline values (mean change from baseline, 15.8 mm Hg) (Figure 3).
Fig. 3. The subject mean changes in Pet CO2 between awake baseline and CO2-induced awakenings during slow increases in Fi CO2 from Night 2 are shown. The mean change in Pet CO2 that caused awakening was 15.8 mm Hg.
[More] [Minimize]In Subject 1, surface sternocleidomastoid (SCM) EMG activity was measured. There were no progressive increases in SCM activity as Pet CO2 was increased although SCM activity did increase substantially after each awakening.
In these tetraplegic subjects, hypercapnia clearly and consistently induced arousal from sleep. These subjects were paralyzed, lacked central connections from chest wall mechanoreceptors, and were mechanically ventilated at constant tidal volumes. Therefore, arousal induced by hypercapnia in these subjects was most likely caused by central mechanisms such as changes in central respiratory drive projecting to arousal neurons or the direct effects of carbon dioxide on arousal neurons (discussed below).
There are a number of methodological issues that need to be considered when interpreting the results of the current study, including protocol design, small sample size, and subject characteristics (medications, chronic hypocapnia, possible accessory respiratory muscle activity above the level of the lesion). These concerns are discussed below.
It was not feasible to study more than four subjects because of the rarity of ventilated, medically stable tetraplegics with injury levels above C3. While this limitation increases the possibility of random chance affecting our interpretation, we believe this is unlikely because of the consistency of our results. For example, awakenings occurred within five min of rapidly induced hypercapnia in 19 of 19 trials, and there were striking differences in these awakening rates compared with spontaneous awakening rates.
Although we concluded that central mechanisms were responsible for inducing arousal in response to hypercapnia, it remains possible that mechanoreceptor input from the sternocleidomastoid or upper airway muscles could contribute to the arousal mechanism because these afferents are preserved in high-level tetraplegics. However, in the one subject in whom sternocleidomastoid EMG was measured, there was no detectable change in activity before arousal. This implies that the tension within this muscle (and presumably mechanoreceptor input to the central nervous system) before arousal was affected minimally by the changes in carbon dioxide. Because of the small size of these muscles and their high threshold for activation, it seems unlikely that increased accessory respiratory muscle mechanoreceptor activity could have been the principal mechanism underlying arousal induced by hypercapnia in this experiment.
The average rise in Pet CO2 associated with arousal in our subjects was 16 mm Hg during slow increases in Pet CO2 ; this value was similar to the change in Pet CO2 that caused arousal in a group of healthy adults (2). Because of this similarity, it may be speculated that the mechanisms by which hypercapnia induced arousal in our tetraplegic subjects were the same as in normal subjects (i.e., central mechanisms). However, this interpretation may be premature because there are important differences between the two experimental groups. First, the technique of inducing hypercapnia was different (slow increases in CO2 versus rebreathing). Second, these tetraplegic subjects had lower baseline Pet CO2 values and serum bicarbonate concentrations than normal individuals. Thus, a similar rise in Pet CO2 would cause greater acidemia in these subjects with tetraplegia compared with normal subjects. Third, these tetraplegic subjects used a variety of medications that could have blunted the arousal response to carbon dioxide (opiates, benzodiazepines) (6). Therefore, our subjects may have aroused at lower levels of Pet CO2 had these medications not been present. Fourth, it is possible that the mechanism of arousal in the tetraplegic subjects is not the same as in subjects without spinal lesions because the chronic lack of chest wall afferent activity could have resulted in remodeling of elements of the CNS responsible for arousal from respiratory stimuli.
A final concern is that an acclimatization night was not included in our experimental protocol. However, if a significant “first night effect” were present (i.e., an increase in the number of arousals on the first night of monitoring because the patient is unaccustomed to sleeping with the polysomnography equipment), a reduction in the arousal rate on the second night should have occurred. In our study, there was a substantial rise in the arousal rate on the second night (during hypercapnia) compared with the first; hence, any contribution of the first night effect would have made our results even more striking.
Carbon dioxide may cause arousal through a number of different mechanisms. First, there are multiple areas of the CNS that can be stimulated by carbon dioxide (7). These include areas directly involved in arousal (such as the locus coeruleus, the midline raphe, and the retrotrapezoid nucleus), suggesting that carbon dioxide may directly stimulate arousal centers (see Figure 4, Pathway 1) (8-10). Other areas of the brain (such as the ventrolateral medulla, and the nucleus tractus solitarius) will affect ventilation when stimulated by carbon dioxide but are not components of the arousal centers per se (7). However, it is possible that projections from these chemoreceptor sites may stimulate arousal centers (Pathway 2). Second, projections from the medullary respiratory center may result in arousal once a threshold of central respiratory drive is reached (2) (Pathway 3). Third, stimulation of vagal afferents (sensitive to changes in lung inflation) and chest wall afferents (located in the muscles and joints) can modulate both central respiratory drive and the perception of hypercapnia-induced dyspnea (11, 12). Thus, stimulation of these mechanoreceptors may modulate the arousal response to carbon dioxide either through direct connections with arousal centers or by affecting central respiratory drive (Pathways 4 and 5). Fourth, when our subjects were made hypercapnic, central respiratory drive increased but they could not increase ventilation. Therefore, in these subjects, a disparity between afferent feedback from chest wall and vagal mechanoreceptors and central respiratory drive occurred. It is possible that it was not the increased central respiratory drive per se that resulted in arousal, but rather the disparity between central drive and afferent mechanoreceptor activity. Some of these arousal mechanisms have been reviewed (3).
Fig. 4. Depicted are five possible mechanisms by which respiratory stimuli induce arousal from sleep. Patients with constant mechanical ventilation and C1–C2 lesions enable the separation of Pathways 1, 2, and 3 from Pathways 4 and 5. Not shown is the possibility that respiratory mechanoreceptor activity can modulate brainstem respiratory output (see Discussion).
[More] [Minimize]Because our tetraplegic subjects suffered from neurologically complete high-level lesions of the spinal cord, chest wall mechanoreceptor activity could not reach the CNS and tidal volume could not be increased in response to hypercapnia, hence negating Mechanisms 4 and 5 in the current experiment (see Figure 4). Despite this, arousal from hypercapnia did occur in these subjects, and must have been due to the direct effect of carbon dioxide, projections from chemoreceptors, or increased central respiratory drive (Mechanisms 1, 2, and 3; Figure 4).
Prior studies have suggested that either respiratory mechanoreceptor feedback or increased central respiratory drive may be the primary mechanism of arousal from respiratory stimuli (i.e., Pathways 3, 4, or 5). For instance, Gleeson and Zwillich demonstrated that arousal occurs consistently at a peak esophageal pressure of −15 to −17 cm H2O during hypoxia, hypercapnia, and flow-resistive loading despite differences in overall ventilation and blood gases during these stimuli (2). Similarly, the level of ventilation at arousal is the same during adenosine- and CO2-stimulated breathing (13). Berry and coworkers found that even though arousal during occluded breaths occurred sooner after occlusion during induced hypercapnia, arousal was consistently related to the absolute level of the peak subglottic pressure during occlusions (14). Furthermore, mechanically assisted ventilation increases the threshold for hypercapnic arousal compared with spontaneous breathing during REM sleep in dogs (although not during NREM sleep) (15). Until the present study, it has been difficult to distinguish arousal induced by increased central respiratory drive from arousal induced by mechanoreceptor stimuli. When our findings are viewed in the context of these studies, it seems likely that the increased central respiratory drive during hypercapnia was the arousal stimulus in our subjects. However, we cannot rule out contributions from chemoreceptor projections and the direct effect of carbon dioxide on arousal neurons (Pathways 1 and 2 in Figure 4).
In summary, we have shown conclusively that hypercapnia is sufficient to cause arousal in the absence of changes in mechanoreceptor input. This does not necessarily mean that input from mechanoreceptors is not important in the arousal response, but only that mechanoreceptor input is not necessary for arousal. Further studies need to be performed to establish the relative contribution of these five proposed mechanisms to respiratory arousals in intact humans.
Supported by a DVA Merit Review Grant, NIH Grant HL 62149, and the Department of Veterans Affairs.
| 1. | Epstein L. J., Weiss J. W.Clinical consequences of obstructive sleep apnea. Semin. Respir. Crit. Care Med.191998123132 |
| 2. | Gleeson K., Zwillich C. W., White D. P.The influence of increasing ventilatory effort on arousal from sleep. Am. Rev. Respir. Dis.1421990295300 |
| 3. | Berry R. B., Gleeson K.Respiratory arousal from sleep: mechanisms and significance. Sleep201997654675 |
| 4. | American Sleep Disorders AssociationEEG arousals: scoring rules and examples. Sleep151992174184 |
| 5. | Rechtschaffen, A., and A. Kales. 1968. A Manual of Standardized Terminology, Techniques, and Scoring System for Sleep Stages in Human Adults. Brain Information Service, UCLA, Los Angeles. |
| 6. | Berry R. B., Kouchi K., Bower J., Prosise G., Light R. W.Triazolam in patients with obstructive sleep apnea. Am. J. Respir. Crit. Care Med.1511995450454 |
| 7. | Coates E. L., Aihua L., Nattie E. E.Widespread sites of brain stem ventilatory chemoreceptors. J. Appl. Physiol.751993514 |
| 8. | Pineda J., Aghajanian G. K.Carbon dioxide regulates the tonic activity of locus coeruleus neurons by modulating a proton and polyamine sensitive inward rectifier potassium current. Neuroscience771997723743 |
| 9. | Bernard D. G., Li A., Nattie E. E.Evidence for central chemoreception in the midline raphe. J. Appl. Physiol.801996108115 |
| 10. | Aihua L., Randall M., Nattie E. E.CO2 microdialysis in retrotrapezoid nucleus of the rat increases breathing in wakefulness but not in sleep. J. Appl. Physiol.871999910919 |
| 11. | Manning H., Shea S. A., Schwartzstein R., Lansing R. W., Brown R., Banzett R. B.Decreased tidal volume increases air hunger at constant Pco2 in ventilated C1–C3 quadriplegics. Respir. Physiol.9019921930 |
| 12. | Harty H. R., Mummery C. J., Adams L., Banzett R. B., Wright I. G., Banner N. R., Yacoub M. H., Guz A.Ventilatory relief of the sensation of the urge to breathe in humans: are pulmonary receptors important? J. Physiol.4901996805815 |
| 13. | Gleeson K., Zwillich C. W.Adenosine stimulation, ventilation, and arousal from sleep. Am. Rev. Respir. Dis.1451992453457 |
| 14. | Berry R. B., Mahutte C. K., Light R. W.The effect of hypercapnia on the arousal response to airway occlusion during sleep in normal subjects. J. Appl. Physiol.74199322692275 |
| 15. | Kimoff R. J., Kozar L. F., Yasuma F., Bradley D., Phillipson E. A.Effect of inspiratory muscle unloading on arousal responses to CO2 and hypoxia in sleeping dogs. J. Appl. Physiol.74199313251336 |