Respiratory Short-term Poststimulus Potentiation (After-Discharge) in Patients with Obstructive Sleep Apnea 1- 3
D. GEORGOPOUWS, E. GIANNOULI, V. TSARA, P. ARGIROPOULOU, D. PATAKAS, and N. R. ANTHONISEN
Introduction
It has been shown that after abrupt cessation of a primary respiratory stimulus, respiratory output, expressed as phrenic nerve activity or minute ventilation, declines gradually to prestimulus levels (1-9). This phenomenon is referred to as short-term poststimulus potentiation (STP) or after-discharge, and is attributed to activation of a brainstem mechanism with slowdynamics that drives ventilation for some time independent of both peripheral and central chemoreceptor inputs (3). The development of STP is not stimulus-specific, having been observed after electrical stimulation of carotid sinus nerve (2), mechanical stimulation of calf muscles (2), brief hypoxic stimulation of carotid bodies (7-9), and voluntary hyperventilation (5, 6). On the other hand, the mechanism is not activated by passive increases in ventilation, which are usually followed by apnea when Paco2 is not controlled (1). It has been proposed that STP is an important factor promoting ventilatory stability. According to this hypothesis, the activation of STP during active hyperventilation would preclude an appreciable ventilatory undershoot later and tend to prevent cyclicbreathing (10, 11). Conversely, cyclic breathing would be promoted if STP were absent. Sleep affects almost all factors that influence ventilatory stability. It decreases the lung gas stores due to recumbency (12), impairs the ventilatory response to added mechanical loads (13), depresses ventilation (14, 15), raises the apneic threshold (13), and decreases the cardiac output (14). Despite these destabilizing influences (10, 11, 15), periodic breathing seldom occurs during sleep, at least in young normal subjects (16), perhaps because of the STP mechanism (11). On the other hand, periodic breathing may playa key role in the development of up1250
SUMMARY In conscious normal humans after a brief hypoxic ventilatory stimulus, ventilation slowly decays to baseline and does not undershoot though the subjects are hyperoxlc and hypocapnlc. This phenomenon Is attributed to short·term poststlmulus potentiation (STP), which may be an important factor promoting ventilatory stability by preventing periodic breathing. It has been pro· posed that obstructive sleep apnea (OSA) Is a variant of periodic breathing, with obstruction oeeurring when ventilatory drive Is low. If this were the case, patients with OSA might have reduced STP. Totest this, seven normal adults and 12patients with OSA(mean apnea Index, 52.4 ± 6.9SE events/h) were studied. ventilation (VI)was measured In awake seated subjects during 30 to 45 s of exposure to hypoxia (end-tidal O2 : 50 mm Hg) followed by hyperoxla. A total of 57 hypoxlc-hyperoxlc runs were analyzed (36 In the patients and 21 In the normal subjects). During hypoxia VI Increased and end ·tldal CO2 decreased by similar amounts In both groups. In normal subjects after hypoxia there was a gradual decay In VI to prehypoxlc levels without an undershoot. In patients, there was on average a ventilatory undershoot at 35 s of hyperoxla, with a mean VI of 83% of baseline. The undershoot was due mainly to a decrease In tidal VOlume, which was significantly lower than that of the normal subJacts for several seconds. These changes were partiCUlarly prominent In seven patients who were not different from the others In terms of baseline characteristics, hypoxic reo sponses, and OSA severity. Our results Indicate that a significant number of patients with OSA have reduced STP after brief hypoxia, a finding that may be of relevance to the pathogenesis of OSA. AM REV RESPIR DIS 1992; 148:1250-1255
per airwaysocclusion in patients with obstructive sleep apnea (OSA) (17-19). In patients with anatomic and functional abnormalities of the upper airways, when the respiratory drive is low,upper airways resistance increases markedly and may lead to complete obstruction (19),which increases ventilatory drive and sets up a potential cycle between high levels of drive with hyperventilation and arousal and low levels of drive with airway occlusion (19). Activation of STP during the hyperventilation phase would tend to prevent hypoventilation and occlusion when drive was low, and failure of STP might contribute to OSA. Recently, we have shown in conscious normal young (7) and old adults (8) that STP can be activated by hypoxia; when a brief (30to 45-s)period of hypoxia was terminated by hyperoxia, ventilation declined slowly without a significant undershoot even though the subjects were hypocapnic. Toour knowledge, hypoxichyperoxic ventilatory responses of this kind have been performed only in normal humans, and the response pattern
is not known in patients with unstable breathing such as those with OSA. The primary goal of the present work was to study STP in awake patients with OSA, measuring ventilatory responses to acute hyperoxia after brief (30 to 45-s) exposure to hypoxia. In addition, similar hypoxic-hyperoxic exposures were performed in a control group of normal subjects. Methods Twelvemale patients with GSA were selected randomly from a larger population of patients (Received in original form January 21, 1992 and in revised form May 15, 1992) 1 From the Respiratory Failure Unit, University ofThessaloniki, Thessaloniki, Greece, and the Respiratory Investigation Unit, Universityof Manitoba, Manitoba, Canada. 2 Supported by the Medical Research Council of Canada. 3 Correspondence and requests for reprints should be addressed to N. R. Anthonisen, M.D., Office of the Dean, Faculty of Medicine, 753 McDermot Avenue, Winnipeg, Manitoba, Canada R3E OW3.
1251
OBSTRUCTIVE SLEEP APNEA AND SHORT·TERM POSTSTIMUWS POTENTIATION
with OSA attending our sleep disorder clinic. Criteria for selection included sleep apnea documented by nocturnal polysomnography and a complaint of daytime hypersomnolence. Patients with hypertension, coronary artery disease, or airway obstruction, plus those using theophylline or CPAP, were excluded. The study was approved by the Hospital Ethics Committee, and all patients gave their written informed consent prior to study entry.
Sleep Studies Patients were monitored in the sleep laboratory between the hours of 10:00 P.M. and 7:00 A.M. Electrooculographic and electroencephalographic measurements were recorded, abdominal and rib-cage movements were measured using respiratory inductance plethysmography (Respitrace'"; Ambulatory Monitoring Inc., White Plains, NY), air flow at the nose and the mouth was measured using a CO 2 gas analyzer (Beckman Instruments, Fullerton, CAl, and arterial oxygen saturation was documented using an ear oximeter (Criticon, Markham, Ontario, Canada). All sleep data were recorded continuously on a multichannel polygraph. Apnea was defined as the cessation of airflow at the nose and mouth for 10 s or more and classified as central, mixed, or obstructive according to standard criteria (20). The apnea index was calculated as the mean number of apneas per hour of sleep, and the mean duration of apnea was computed. The mean Sa02 during wakefulness, minimal Sa02 during sleep, and the percentage of sleep time spent below saturations of 90 and 800/0 were calculated. Hypoxic-Hyperoxic Responses At least 24 h after the sleep study, the patients came to the laboratory where hypoxic-hyperoxic ventilatory responses were documented. The patients wereinstructed not to smoke and to have no drinks with caffeine for at least 8 h before the study. The methods used for these ventilatory responses has been fully described in a previous publication (7). During the experiment the patients were seated in a comfortable chair, distracted with nonrhythmic music, and observed closely to ensure wakefulness. They were instrumented with electrocardiogram leads and an ear oximeter to monitor heart rate and O 2 saturation, respectively.Wearing noseclips, they breathed through a low resistance unidirectional valve, the inspiratory side of which was connected to a pneumotachograph and then, through a manifold, to four tubes 3.2 em in diameter. Three ofthe tubes were connected to balloons containing nitrogen, oxygen, and 8.5% oxygen in nitrogen, and the fourth tube was open to room air. Each tube could be occluded by inflating a small rubber balloon placed near the manifold; this could be done without auditory clues and did not alert the subjects. The dead space of the device from mouth to tube entrance was approximately 150ml. Partial pressures of end-tidal O 2 (PET02) and CO 2 (PETC02) were measured by a fuel cell O 2me-
ter (E. Jaeger, Wiirzburg, Germany) and an infrared CO 2meter (Beckman) that sampled the expiratory side of the valve. The pneumotachograph output was integrated to give tidal volume (VT). VT, Sao., PET02 and PETC02 were recorded on a strip chart recorder. The patients breathed room air until PETC02 and ventilation stabilized, usually within 3 to 5 min. Then the inspirate was changed to N 2 for two to three breaths, during which PET02 decreased rapidly to about 55 mm Hg. The inspirate was then switched to 8.5% O 2, which maintained PET02 at approximately 50 mm Hg. After 30 to 45 s, hypoxia was terminated abruptly by switching to 100% O 2, so that PET02 of the first hyperoxic breath was at or above the normoxic baseline, whereas that of the second was above 150mm Hg. Hyperoxia was sustained for 2 min and terminated by switching the inspirate to room air. When PET02 returned to normoxic baseline, a stabilization period of about 2 min of room air breathing followed, and then the same hypoxic-hyperoxic procedure was repeated. Three to four such runs were performed in each patient. Runs associated with swallowing or coughing during the critical period of transition to O 2were excluded from the analysis. In addition, in nine patients the effect of switching the inspirate from room air to 100% O 2, was tested. The control group was seven normal nonsnoring adults (six male and one female). Their mean age was 42 ± 3 (SEM) yr, similar to that of the patients. None was obese or complained of sleep disorders. Three hypoxichyperoxic runs were performed in each subject, using the same method as in patients. Minute ventilation (VI), VT, inspiratory time (TI), expiratory time (Ts), and total breath duration (Ttot) were measured breath by breath for 30 s prior to hypoxia (baseline), during hypoxia, and for 2 min after switching to 100% O 2. All breaths during hypoxia and hyperoxia were expressed as percentages of the mean baseline value. The results for all tests in a given subject were averaged at 5-s intervals, the individual values being obtained by interpolation between the adjoining breaths.
To assess STP in each subject, the average hyperoxic ventilation was plotted logarithmically as a function of time, assuming zero time to be 5 s after institution of hyperoxia, and half time decays to control values (T J/z) were calculated. In addition, the lowest hyperoxic ventilation was noted. Data were analyzed by an unpaired t test and the nonparametric Mann-Whitney test on two-way analysis of variance (ANOVA). TJ/z and the lowest hyperoxic VI were related to other variables by linear regression using the least-squares method; p < 0.05 was considered statistically significant. Values were expressed as mean ± SE.
Results
Patient characteristics are shown in table 1.On average, the patients wereobese, but they had normal spirometric and blood gas determinations. Two patients (Patients 8 and 9) had borderline high values of Paco2' Only one (Patient 9) had Pan, of less than 70 mm Hg.
Sleep Studies All patients displayed repeated episodes of upper airway occlusion during sleep. Invariably, obstruction occurred when respiratory effects, as detected by inductance plethysmography, were low and in some cases absent (mixed apnea). In addition, during sleep, all patients demonstrated periodic breathing. For the group the mean apnea index was 52.4 ± 6.9 events/h, and the mean apnea duration was 22.9 ± 2.1 s. Significant desaturation was observed in all patients, and in 10 the minimal Sao, was less than 80070. Nine patients spent more than 10070 of their total sleep time with Sao, less than 90%. Individual sleep-related parameters are shown in table 2. Analysis of apneas (table 3) showed that five patients (Patients 2, 4, 8, 9, and 12) exhibited some central apneas. Patient 7 had only obstructive apneas.
TABLE 1 PATIENT CHARACTERISTICS Patient No.
Age
Weight
Height (em)
Pacoz (mmHg)
FVC
(kg)
Pao z (mmHg)
FEV1
(yr)
(% pred)
(% pred)
1 2 3 4 5 6 7 8 9 10 11 12 Mean ± SE
32 54 50 55 53 35 43 47 60 51 47 37 47 2.5
110 91 99 100 99 130 97 90 106 93 93 100 100.7 3.1
172 175 177 174 172 173 167 166 168 178 171 166 172 1
84 76 70 80 85 80 75 70 68 88 72 87 77.9 2.1
35 37 43 40 40 35 40 46 47 41 37 40 40.1 1.1
84 117 65 80 95 98 106 72 90 92 94 93 91.5 4.1
87 118 76 79 96 95 99 75 88 95 99 94 91.8 3.4
1252
GEORGOPOULOS, GIANNOULI, TSARA, ARGIROPOULOU, PATAKAS, AND ANTHONISEN
TABLE 2 SLEEP STUDY PARAMETERS
Patient No.
1 2 3 4 5 6 7 8 9 10 11 12 Mean ± SE
AI
TST
REM
AI
AD
206 191 325 250 293 315 328 319 300 327 302 180 278 16
14.2 14.4 31.8 17.8 10.1 19.2 23.6 25.3 21.1 15.4 6.7 14.9 17.9 2.0
79.5 35.3 42.2 55.0 56.3 78.3 41.2 87.7 63.2 14.0 13.6 63.0 52.4 6.9
17.7 28.9 32.3 24.3 32.1 23.7 32.8 20.1 14.2 15.0 12.8 20.3 22.9 2.1
TST with Sao 2 < 90% (%)
27.8 4.2 24.1 25.2 16.0 28.1 11.4 24.2 26.0 2.4 33.7 8.0 19.3 3.0
79 82 50 50 70 54 54 65
n
85 62 72 67 4
Definition of abbreviations: TST = total sleep time (min); REM .. rapid eye movement (% of = apnea index (events/h); AD = mean apnea duration (s).
Hypoxic-Hyperoxic Ventilatory Responses Baseline ventilation and PETC0 2 are shown in table 4. There was no significant difference between the patients and the control subjects with regard to ventilation, tidal volume, frequency, or PETC02•
A total of 57 hypoxic-hyperoxic runs were analyzed, three in each subject. The durations of hypoxic exposure were simiTABLE 3 APNEA CLASSIFICATION* Patient No.
Minimal Sa0 2
Obstructive (%)
Central (%)
Mixed (%)
81.2 10.0 91.3 80.7 86.9 95.4 100.0 68.0 16.8 91.2 84.2 52.8
0 4.3 0 4.1 0 0 0 7.9 31.9 0 0 3.5
18.8 85.7 8.7 15.2 13.1 4.6 0 24.1 51.3 8.8 15.8 43.7
1 2 3 4 5 6 7 8 9 10 11 12
* Percentage of total number of apneas.
lar in the two groups, averaging 38.8 ± 1.3 s in the patients, and 39.3 ± 2.0 in the control subjects. During hypoxia PET02 decreased exponentially and reached a plateau value of about 50 mm Hg approximately 30 to 45 s after institution of hypoxia. Hypoxic changes in VI and PETC 02 are shown in figure 1, and they did not differ significantly between normal subjects and patients. There was, on average, little change in VI in the first 10 s of hypoxia. Thereafter, VI increased significantly, and it may not have reached a steady state before the end of hypoxia; VI continued to increase during the first 5 s of hyperoxia, amounting to 165 and 152070 of baseline in patients and in normal subjects, respectively. The increase in hypoxic ventilation was mainly due to increase in VT. In patients, VT increased 49070 and frequency increased 11070, whereas the corresponding values in normal subjects were 61 and 5 070. Changes in PETC02 reflected the changes in VI, the
c 0
i
TABLE 4 BASELINE (ROOM AIR) VALUES OF VENTILATION AND PETCOz IN NORMAL SUBJECTS AND PATIENTS VT (L) Normal subjects ± SEM Patients ± SEM
0.55 1.1 0.59 0.04
VL flmin
(Llmin)
PETCO z (mmHg)
16.1 0.70 14.4 1.0
8.72 0.70 8.50 0.65
36.8 1.2 38.4 1.1
Definition of abbreviations: VT .. tidal volume; f - breathing frequency; VI - minute ventilation; PETC02 .. partial pressure of end-tidal CO2,
TST);
~CD >
maximal decline averaging 3.03 ± 0.64 mm Hg (range, 8.4 to 0.74) in patients and 2.90 ± 1.04 mm Hg (range, 8.2 to 0.1) in normal subjects. Breath-by-breath changes in Sao, are not reported because of the slow response of the ear oximeter. In all subjects the lowest value of Sao, was observed several seconds after switching the inspirate to hyperoxia. In the normal subjects, abrupt termination of hypoxia with hyperoxia was associated with a slow ventilatory decay to baseline (figure 2). Mean ventilation decreased to baseline several seconds after institution of hyperoxia, and, despite the presence of hypocapnia and hyperoxia, no apparent undershoot in VI could be detected. Mean hyperoxic VI followed an exponential course, with r = 0.94 and TYl = 9.7 s. In the patients, the hyperoxic ventilatory response pattern was different from that in the normal subjects. Mean hyperoxic ventilation also declined exponentially (r = 0.98), with a TYz of 4.8 s, and it reached a value of 810J0 of baseline at 45 s of hyperoxia, remaining lower than the baseline for several seconds (figure 2). The lowest VI attained during hyperoxia was 65.9 ± 4.7070 of the control value in the patients, which was significantly less (p < 0.01, nonparametric MannWhitney test) than the nadir value of 85.0 ± 4.9070 observed in the normal control subjects. In the patients, the relative hypoventilation was due mainly to a decrease in tidal volume (figure 3), which was significantly lower than that of the normal subjects at 30,35,40, and 45 s of hyperoxia (P < 0.05, two-way ANOVA).Although during hyperoxia breathing frequency tended to be lower than that of the normal subjects (figure 4), largely because
160 150 140 130 120 110 100
6
3
0
10
20
30
40
50
100 N
0
...0w
95
D.
90
50
0 Time (sec)
Fig. 1. Average minute ventilation and end-tidal Poo, (PETCOz) as a function of time during hypoxia. Both are expressed as percentages of mean prehYPOxic values. Bars are ± SE. Each circle is the average for 12 patients (closed circles) and seven normal subjects (open circles), except at 40 and 45 s where numbers are smaller as indicatad in the upper panel. This was because the length of hypoxic exposure varied slightly from subject to subject.
1253
OBSTRUCTIVE SLEEP APNEA AND SHORT-TERM POSTSTIMUWS POTENTIATION
120 160
G
.5
115
'ii
at
140
tV
.Q
G
'0
'ii
C
.5 at
•
>-
120
C>
.Q
C
C
. C
~
~
105
CD :::l
'0
a
110
c:r CD
~
100
.
100
01 C
£
CD
95
cD
80
90'
0
20
40
60
80
100
120
Fig. 2. Average minute ventilation during hyperoxia after hypoxia. Ventilation is expressed as a percent of mean prehypoxic values. Open circles indicate normal subjects; closed circles indicate patients. SE bars shown at selected points are representative of variability in responses.
of prolongation of Ts, neither frequency nor Th differed significantly between patients and control subjects. Although the patients as a group differed from the normal control subjects, there were substantial differences among patients and we subdivided them into two groups. Seven patients (Group A: Patients 2,4, 6, 8, 9, 10, and 12) exhibited T Y2 less than 4 s and a nadir hyperoxic VI that was less than 720/0 of baseline. The remaining five patients (Group B: Patients 1, 3, 5, 7, and 11) had TY2 greater than 4.5 s, and in all but one (Patient 11) the lowest hyperoxic VI was greater than 78070 of baseline. In Group A mean hyperoxic VI declined rapidly, with a mean T Y2 of 3.0 s. Mean hyperoxic ventilation reached a value of 72% of baseline at 30 s of hyperoxia, and it did not return to baseline for the next 60 s (figure 5). The relative hypoventilation was due to a decrease in both tidal volume
.
· ·
160
140
'ii
.Q
'0
t.
E :::l '0
> 'ii
120
100
'U
i= 80 0
20
60
20
40
60
80
100
120
TIme (sec)
Time (sec)
.5
0
80
100
120
TIme (sec)
Fig. 3. Average tidal volume, expressed as a percent of mean prehypoxic values, during hyperoxia after hypoxia. Symbols as in figure 2. Asterisks indicate significant difference from normal subjects (p < 0.05, two-way ANOVA).
Fig. 4. Average breathing frequency, expressed as a percent of mean prehypoxic values, during hyperoxia after hypoxia. Symbols as in figure 2.
1~~
115 105
110 100
00 0
and frequency (figure 5), with frequency being low largely because Th was prolonged. 11did not differ between groups, whereas in Group A, Ts was significantly greater than in Group B at 25, 30, and 35 s of hyperoxia. Six of the seven patients in Group A had breaths with Th > 5 s, and three (Patients 4, 8, and 9) had breaths with Th > 10s, meeting some definitions of apnea. On the other hand, in patients in Group B, hyperoxia was associated with a slower ventilatory decay to baseline, with a mean T Y2 of 6.2 s (figure 5). Mean hyperoxic ventilation reached 85% of baseline at 45 s of hyperoxia and returned to control at 55 s. None of these patients exhibited Th > 5 s during hyperoxia. The patients in Group B did not differ from those in Group A in terms of baseline characteristics, severity of sleep apnea syndrome, hypoxic ventilatory responses, or changes in PETC02. However, the five patients who exhibited pure central apneas during sleep (table 3) were all included in Group A. Among the patients, TY2 did not correlate significantly with the duration of hypoxic exposure, the magnitude of hypoxic change in VI, VT, frequency, or PETC02 or baseline Pa02. In addition, there was no relationship between T Y2 and sleep-related parameters such as apnea index, minimal Sa02, mean duration of apnea, and percent of total sleep time with Sa02less than 90 and 80%, as well as with apnea characteristics. Among the above parameters, the lowest hyperoxic VIwas correlated significantly only with the hypoxic changes in PETC02 (r = - 0.58). In the normal subjects none of the above relationships was significant. In individual subjects the highest and lowest hyperoxic VIand hypoxic changes of PETC02 in individual runs did not differ significantly (one-way ANOVA), in-
.....
......
**~...
.
.
95
~
40
~
80
100
1~
Time (sec)
Fig. 5. Ventilation, tidal volume, and breathing frequency, expressed as a percent of mean prehypoxic values, during hyperoxia after hypoxia in patients of Group A (closed circles) and of Group B (open circles). Asterisks indicate significant difference from Group B (two-way ANOVA).
dicating reproducibility of the hypoxichyperoxic responses. In nine patients (five from Group A and four from Group B), hyperoxia after room air breathing without an intervening period of hypoxia resulted in little decrease in VI. Average VIat 5-s intervals over the first minute of O 2 breathing varied from 91 to 108% of control values in a nonsystematic way. This hyperoxic response did not differ between the two groups of patients. Discussion The patients studied had the typical symptoms and signs of OSA. All were obese and had daytime hypersomnolence and normal or near normal lung function. During sleep all showed a significant number of obstructive apneas, with additional mixed and central apneas in all but one. Apneas were associated with significant desaturation, and most patients spent an appreciable percentage of their sleep time with Saoj less than 90%. The diagnosis of OSA in this group of patients was well established. None of the subjects were aware of the physiologic purpose of the hypoxichyperoxic experiments, or what changes in ventilation were expected. Furthermore, at the end of each experiment careful questioning revealed that none of them could define precisely the time of transition from hypoxia to hyperoxia, and most were not able to recall the number of hypoxic exposures. Indeed, six patients and three normal subjects were un-
1254
aware of any different sensations during hypoxia. Therefore, we can assume with confidence that the breathing patterns observed during hypoxia and hyperoxia were not due to exogenous stimuli. Our apparatus was designed to achieve rapid changes in alveolar P0 2. In all subjects PET02 at the end of the first hyperoxic breath was at or above the normoxic baseline, and that of the second was above 150 mm Hg. In previous studies of the onset of hypoxia in normal subjects (7, 8), we found that ventilation increased about 5 s after PET02 reached 75 mm Hg, so that it is reasonable to assume that the hypoxic stimulus was withdrawn some 5 s after the first hyperoxic breath in these experiments (21) while PETC02 was lower than baseline levels. Despite the hyperoxia and hypocapnia, VI remained above baseline values in the normal subjects for some time, declining slowly to baseline without a significant undershoot (figure 2). These results are similar to our previous findings in normal young and elderly adults (7, 8). We have interpreted this gradual decline in ventilation as indicative of STP or after-discharge. On the other hand, in patients with OSA, VI exhibited a rapid drop below baseline, averaging 81% of baseline after 45 s of hyperoxia and returning to baseline only after 85 s of hyperoxia (figure 2). This ventilatory depression was mainly due to a tidal volume decrease, which was significantly lower than that of the normal subjects at 30, 35, 40, and 45 s of hyperoxia. In addition, the lowest hyperoxic ventilation was also significantly lower than that of the normal subjects. The hyperoxic ventilatory depression was particularly prominent in seven patients (Group A) whose ventilation during hypoxia was consistent with absent or greatly diminished STP (figure 5). In the other five patients (Group B), an appreciable ventilatory STP was evident (figure 5). These results suggest that a significant number of patients with OSA have decreased STP. It should be noted that our test for STP was an artificial one in that hypoxia was induced by changing the inspirate. In OSA, hypoxia is largely the result of hypoventilation, which is attended by an increase in CO 2, not by a decrease, as in our tests. A combined hypoxic-hypercapnic stimulus would be inappropriate for the detection of STP, however, for although PET02 is a reasonable indicator of hypoxic stimulus during rapid transients, this is not true for PETC02 due to the relatively slow dynamics of change
GEORGOPOULOS, GIANNOULI, TSARA, ARGIROPOULOU, PATAKAS, AND ANTHONISEN
in CO 2 at the level of the central chemoceptor (22-24), which can be accentuated by changes in cerebral blood flow. Indeed, had we employed an hypoxic-hypercapnic stimulus, it is likely that we would have observed considerable poststimulus hyperventilation because of residual activation of central chemoceptors by CO 2, It is possible that these factors influenced our results, but they would be unlikely to explain the differences we observed between groups. The hypoxic-hyperoxic exposures we used were designed to demonstrate STP as unequivocally as possible by avoiding hypercapnia, and our results indicate that this mechanism appears to be deficient in at least some subjects with OSA. It is conceivable that the differences we observed were because the patients with OSA were asleep during testing and the normal subjects were not. There is evidence that sleep depresses posthypoxic hyperoxic ventilation in goats (27) and in humans (28) in the presence of hypocapnia. We think it unlikely that our patients with OSA fell asleep during STP testing. We observed them closely and they did not give behavioral evidence (14), indicating low levels of wakefulness during the approximately 20 min it took to complete the tests. Further, the patients with OSA did not demonstrate the irregular breathing that characterized their sleep studies, except transiently immediately after the institution of hyperoxia. It has been proposed that OSA is closely related to periodic breathing (11, 17-19), with patients cycling between airway occlusion when ventilatory drive is low and arousal when drive is high. There is evidence that inspiratory effort, as assessed by measurement of diaphragmatic EMG and esophageal pressure, waxes and wanes in OSA, with airway occlusion occurring when efforts reach a nadir. This is best exemplified by patients who develop obstruction during central apneas. There are reasons to believe that periodic breathing and OSA may be mutually reinforcing phenomena. In both normal and abnormal subjects, periodic breathing is more common during sleep than during wakefulness. With sleep there is hypoventilation, a reduction in cardiac output, and an increase in the apneic threshold (13, 14), all of which tend to promote periodic breathing (10, 11, 15), as does the reduced lung gas stores consequent to recumbency (12). In patients with OSA, upper airways resistance is high and varies greatly and inversely with inspiratory drive (19). Further, sleep blunts respiratory load com-
pensation (13), so that increases in resistance produce larger decreases in ventilation than in the awake state. Episodic airway occlusion, if it occurs when respiratory drive is low, is likely to give rise to cyclic breathing since it produces episodic ventilatory stimulation and swings in ventilation and ventilatory drive. During sleep our patients with OSA all showed unstable breathing patterns even when airway occlusion did not occur, and in many instances obstruction occurred after apnea or near apnea. Other investigators have made the same observations (17-19, 25, 26). In this setting, respiratory STP would tend to stabilize ventilation. If hypoxia consequent to airway occlusion activated STP, then the hypoxic responsehyperventilation and arousal- would be followed by a gradual decrease in ventilation. In the absence of STP, the hyperventilation of arousal would be followed by a more abrupt decrease in ventilatory drive and ventilation, rendering the subject more vulnerable to another episode of airway occlusion. This general phenomenon is illustrated by the seven patients of Group A. With relief of hypoxia, ventilation fell precipitously to levels below control values, and episodes of apnea were observed, though the patients were awake (figure 5). In these 'subjects the absence or reduction of STP may have played a part in the pathogenesis of their disease. On the other hand, the other five patients with OSA clearly demonstrated that the syndrome can develop in subjects with STP that is not clearly abnormal according to our measurements. We assessed STP in awake patients after an hypoxic stimulus. We believe that even in the absence of hypercapnia, hypoxia is the most physiologically relevant stimulus in patients with OSA, but we are uncertain as to the applicability of our measurement of STP to the sleeping state. In animals STP is not affected by anesthetics, decerebration, or decerebellation (3), so it would seem likely that respiratory STP is not obliterated by sleep. However, there is evidence that STP is reduced in slow-wave sleep in goats (27), and Badr and coworkers (28) observed ventilatory inhibition with hyperoxia after brief hypoxic exposures in sleeping humans. Thus, it may well be that waking measurements overestimate STP during sleep. It is possible that our Group B patients would have demonstrated a defect in STP had they been examined during sleep; the fact that their tidal volume fell below control levels during
OBSTRUCTIVE SLEEP APNEA AND SHORT·TERM POSTSTIMUWS POTENTIATION
hyperoxia suggests that this might have been the case. It is also possible that our normal subjects would have shown little evidence of STP had they been studied during sleep. On the other hand, STP in OSA would be most significant during and immediately after arousal, so that waking measurements may have been appropriate. Although it is relatively easy to postulate how the presence or absence of STP might influence OSA, it is not clear why seven out of 12 patients with OSA had clear-cut defects in STP, a much larger number than are encountered in series of normal subjects (7, 8). We have shown that modest hypoxia interferes with STP in normal subjects; this effect certainly occurs after 20 to 30 min at an arterial saturation of 80%, and it probably occurs after as little as 5 min (7, 29-31). Our patients with OSA were not hypoxic while awake, and oxygenation did not differ between our two patient groups, so hypoxia per se probably did not explain the defect in STP in seven of our patients. On the other hand, we did employ hypoxia in our test for STP, and it is possible that some of our patients were abnormally sensitive to the "hypoxic depression" that accompanies loss of STP. This was not evident in terms of hypoxic ventilatory responses, which did not differ between patient groups. Abnormal levels of endogenous opiates have been found in the cerebrospinal fluid of patients with OSA (32), and it is possible that these substances reduced STP in our patients. It is not clear, however, why endogenous opiates should be increased more in one group of subjects than in another when the degree of obstruction and sleep disorder did not differ. It is perhaps more attractive to postulate that a certain number of otherwise normal people have greatly reduced STP as we measure it; we have observed several such persons (8). If this were true, people with defective STP who became obese would be at high risk for developing OSA, and unselected groups of such patients would show a high incidence of defective STP. We have restudied one of our Group A patients (Patient 8) after his apneic episodes were treated successfully for 2 months with CPAP. Results were the same as those of the initial study, suggesting that his defective STP was not a consequence of the nocturnal obstructive episodes. As noted earlier, the fact that five of our patients with OSA demonstrated normal or near normal STP suggests that the syndrome can develop in the absence
of a defect in STP as we measure it. Indeed, our measurements of STP did not relate to any of the indices of OSA severity noted in table 2. This is not surprising since the severity of OSA must be more dependent upon the morphology and function of the upper airways than any other factors, and these vary among patients. Indeed, it is possible that the presence or absence of STP is reflected in contraction of the muscles controlling the upper airways. Genioglossal contraction, which tends to maintain upper airway patency (33), also tends to reflect ventilatory drive, so it is possible that the absence of STP also causes unusual poststimulus instability of the upper airways. It is also possible that STP as measured in terms of ventilation does not apply or applies in a variable way to upper airway dilators. It is worth noting, however, that the preservation of STP appeared to predict one sleep-related event in these patients: all patients with central apneas were included in the group with defective STP. This again suggests that reduced STP may contribute to respiratory arrhythmias with or without airway obstruction. References 1. Eldridge FL. Postthyperventilation breathing: different effects of active and passive hyperventilation. J Appl Physiol 1973; 34:422-30. 2. Eldridge FL. Central neural respiratory stimulatory effect of activerespiration.J Appl Physiol1974; 37:723-35. 3. Eldridge FL. Maintenance of respiration by central neural feedback mechanisms. Fed Proc 1977; 36:2400-4. 4. Eldridge FL, Gill-Kumar P. Central neural respiratory drive and after-discharge. Respir Physiol 1980; 40:49-63. 5. Swanson GD, Ward DS, Bellville lW. Posthyperventilation isocapnic hyperpnea. 1 Appl Physiol 1976; 40:49-63. 6. Tawadrous FD, Eldridge FL. Posthyperventilation breathing patterns after active hyperventilation in man. J Appl Physiol 1974; 37:353-6. 7. Georgopoulos D, Bshouty Z, Younes M, Anthonisen NR. Hypoxic exposure and activation of the after-discharge mechanism in conscious humans. J Appl Physiol 1990; 69:1159-64. 8. Ahmed M, Giesbrecht GG, Serrette C, Anthonisen NR. Ventilatory after-discharge in elderly humans (abstract). Am Rev Respir Dis 1990; 141:A379. 9. Fregosi RE. Short term potentiation of breathing in humans. 1 Appl Physiol 1991; 71:892-9. 10. Cherniack NS, Lougobordo GS. Abnormalities in respiratory rhythm. In: Cherniack NS, Widdicombe IG, eds. Handbook of Physiology. Vol II, part 2: The respiratory system. Bethesda, MD: American Physiological Society, 1986; 729-49. 11. Younes M. The physiological basis of central apnea and periodic breathing. Curr Pulmonol1989; 10:265-326. 12. Agostoni E, Hyatt RE. Static behaviour ofthe respiratory system. In: MacKlem PT, Mead 1, eds. Handbook of physiology. Vol 3: The respiratory
1255 system.Bethesda, MD: American PhysiologicalSociety, 1986; 113-30. 13. Dempsey JA, Skatrud JB. Fundamental effects of sleep state on breathing. Curr Pulmonol 1988; 9:267-304. 14. Phillipson EA, Bowes G. Control of breathing during sleep. In: Cherniack NS, Widdicombe JG, eds. Handbook of Physiology. Vol 2: The respiratory system. Bethesda, MD: American Physiological Society, 1986; 649-89. 15. Khoo MCK, Kronauer RE, Strohl KP, et al. Factors inducing periodic breathing in humans: a general model. J Appl Physiol 1982; 53:644-59. 16. Webb P. Periodic breathing during sleep. J Appl Physiol 1974; 37:899-903. 17. Onal E, Lopata M, O'Connor TD. Pathogenesisof apnea in hypersomnia-sleepapnea syndrome. Am Rev Respir Dis 1982; 125:167-74. 18. Onal E, Lopata M. Periodic breathing and the pathogenesis of occlusive sleep apneas. Am Rev Respir Dis 1982; 126:676-80. 19. Martin Rl, Pennock BE, Orr WC, et al. Respiratory mechanics and timing during sleep in occlusivesleep apnea. J Appl Physiol1980; 48:432-7. 20. Guilleminault C, Tilkian A, Dement WC. The sleep apnea syndromes. Annu Rev Med 1976; 27:465-84. 21. Lange RL, Horgan JD, Botticelli IT, Tsagaris T, Carlisle RP, Kuida H. Pulmonary to arterial circulatory transfer function: importance in respiratory control. J Appl Physiol 1966; 21:1281-91. 22. Fujishima MP, Scheinberg P, Busto P, Reimuth O. The relation between cerebral oxygenconsumption and cerebral vascular reactivity to carbon dioxide. Stroke 1971; 2:251-7. 23. Neubauer JA, Santiago TV,Posner MA, Edelman NH. Ventral medullary pH and ventilatory response to hyperperfusion and hypoxia. J Appl Physiol 1985; 58:1659-68. 24. Lambertsen CJ, Kough RH, Cooper DY, Emmel GL, Loeschcke HH, Schmidt CF. Oxygen toxicity. Effect in man of oxygen inhalation at 1 and 3.5 atmospheres upon blood gas transport, cerebral circulation, and cerebral metabolism. J Appl Physiol 1953; 5:471-86. 25. Strohl K. Control of upper airwayduring sleep. In: Edelman NH, Santiago TV, eds. Breathing disorders of sleep. Vol5. New York: Churchill Livingstone, 1986; 115-37. 26. Skatrud JB, Dempsey JA. Airway resistance and respiratory muscle function in snorers during NREM Sleep. J Appl Physiol 1985; 59:328-35. 27. Daristotle L, Petrozzino 11, Santiago TV. Effect of sleep-wake state on ventilatory afterdischarge (abstract). FASEB J 1991; 5:AI478. 28. Badr S, Skatrud JB, Dempsey JA. Ventilatory inhibition following active hyperventilation during NREM sleep (abstract). Am Rev Respir Dis 1990; 141:AI25. 29. Holtby SG, BerezanskiD,Anthonisen NR. The effect of 100070 oxygen on hypoxic eucapnic ventilation. J Appl Physiol 1988; 65:1157-62. 30. Georgopoulos D, Holtby SG, Berezanski D, Anthonisen NR. Aminophylline effects on ventilatory response to hypoxia and hyperoxia in normal adults. J Appl Physiol 1989; 67:1150-6. 31. Georgopoulos D, Walker S, Anthonisen NR. Increased chemoreceptor output and ventilatory response to sustained hypoxia. J Appl Physiol1989; 67:1157-63. 32. Gislason T, Almqvist M, Boman G, Lindholm CE, Terenius L. Increased CSF opioid activity in sleep apnea syndrome. Regression after successful treatment. Chest 1989; 96:250-4. 33. Sullivan CE, Issa FG. Obstructive sleep apnea. Clin Chest Med 1985; 6:633-50.
D. GEORGOPOUWS, E. GIANNOULI, V. TSARA, P. ARGIROPOULOU, D. PATAKAS, and N. R. ANTHONISEN
Introduction
It has been shown that after abrupt cessation of a primary respiratory stimulus, respiratory output, expressed as phrenic nerve activity or minute ventilation, declines gradually to prestimulus levels (1-9). This phenomenon is referred to as short-term poststimulus potentiation (STP) or after-discharge, and is attributed to activation of a brainstem mechanism with slowdynamics that drives ventilation for some time independent of both peripheral and central chemoreceptor inputs (3). The development of STP is not stimulus-specific, having been observed after electrical stimulation of carotid sinus nerve (2), mechanical stimulation of calf muscles (2), brief hypoxic stimulation of carotid bodies (7-9), and voluntary hyperventilation (5, 6). On the other hand, the mechanism is not activated by passive increases in ventilation, which are usually followed by apnea when Paco2 is not controlled (1). It has been proposed that STP is an important factor promoting ventilatory stability. According to this hypothesis, the activation of STP during active hyperventilation would preclude an appreciable ventilatory undershoot later and tend to prevent cyclicbreathing (10, 11). Conversely, cyclic breathing would be promoted if STP were absent. Sleep affects almost all factors that influence ventilatory stability. It decreases the lung gas stores due to recumbency (12), impairs the ventilatory response to added mechanical loads (13), depresses ventilation (14, 15), raises the apneic threshold (13), and decreases the cardiac output (14). Despite these destabilizing influences (10, 11, 15), periodic breathing seldom occurs during sleep, at least in young normal subjects (16), perhaps because of the STP mechanism (11). On the other hand, periodic breathing may playa key role in the development of up1250
SUMMARY In conscious normal humans after a brief hypoxic ventilatory stimulus, ventilation slowly decays to baseline and does not undershoot though the subjects are hyperoxlc and hypocapnlc. This phenomenon Is attributed to short·term poststlmulus potentiation (STP), which may be an important factor promoting ventilatory stability by preventing periodic breathing. It has been pro· posed that obstructive sleep apnea (OSA) Is a variant of periodic breathing, with obstruction oeeurring when ventilatory drive Is low. If this were the case, patients with OSA might have reduced STP. Totest this, seven normal adults and 12patients with OSA(mean apnea Index, 52.4 ± 6.9SE events/h) were studied. ventilation (VI)was measured In awake seated subjects during 30 to 45 s of exposure to hypoxia (end-tidal O2 : 50 mm Hg) followed by hyperoxla. A total of 57 hypoxlc-hyperoxlc runs were analyzed (36 In the patients and 21 In the normal subjects). During hypoxia VI Increased and end ·tldal CO2 decreased by similar amounts In both groups. In normal subjects after hypoxia there was a gradual decay In VI to prehypoxlc levels without an undershoot. In patients, there was on average a ventilatory undershoot at 35 s of hyperoxla, with a mean VI of 83% of baseline. The undershoot was due mainly to a decrease In tidal VOlume, which was significantly lower than that of the normal subJacts for several seconds. These changes were partiCUlarly prominent In seven patients who were not different from the others In terms of baseline characteristics, hypoxic reo sponses, and OSA severity. Our results Indicate that a significant number of patients with OSA have reduced STP after brief hypoxia, a finding that may be of relevance to the pathogenesis of OSA. AM REV RESPIR DIS 1992; 148:1250-1255
per airwaysocclusion in patients with obstructive sleep apnea (OSA) (17-19). In patients with anatomic and functional abnormalities of the upper airways, when the respiratory drive is low,upper airways resistance increases markedly and may lead to complete obstruction (19),which increases ventilatory drive and sets up a potential cycle between high levels of drive with hyperventilation and arousal and low levels of drive with airway occlusion (19). Activation of STP during the hyperventilation phase would tend to prevent hypoventilation and occlusion when drive was low, and failure of STP might contribute to OSA. Recently, we have shown in conscious normal young (7) and old adults (8) that STP can be activated by hypoxia; when a brief (30to 45-s)period of hypoxia was terminated by hyperoxia, ventilation declined slowly without a significant undershoot even though the subjects were hypocapnic. Toour knowledge, hypoxichyperoxic ventilatory responses of this kind have been performed only in normal humans, and the response pattern
is not known in patients with unstable breathing such as those with OSA. The primary goal of the present work was to study STP in awake patients with OSA, measuring ventilatory responses to acute hyperoxia after brief (30 to 45-s) exposure to hypoxia. In addition, similar hypoxic-hyperoxic exposures were performed in a control group of normal subjects. Methods Twelvemale patients with GSA were selected randomly from a larger population of patients (Received in original form January 21, 1992 and in revised form May 15, 1992) 1 From the Respiratory Failure Unit, University ofThessaloniki, Thessaloniki, Greece, and the Respiratory Investigation Unit, Universityof Manitoba, Manitoba, Canada. 2 Supported by the Medical Research Council of Canada. 3 Correspondence and requests for reprints should be addressed to N. R. Anthonisen, M.D., Office of the Dean, Faculty of Medicine, 753 McDermot Avenue, Winnipeg, Manitoba, Canada R3E OW3.
1251
OBSTRUCTIVE SLEEP APNEA AND SHORT·TERM POSTSTIMUWS POTENTIATION
with OSA attending our sleep disorder clinic. Criteria for selection included sleep apnea documented by nocturnal polysomnography and a complaint of daytime hypersomnolence. Patients with hypertension, coronary artery disease, or airway obstruction, plus those using theophylline or CPAP, were excluded. The study was approved by the Hospital Ethics Committee, and all patients gave their written informed consent prior to study entry.
Sleep Studies Patients were monitored in the sleep laboratory between the hours of 10:00 P.M. and 7:00 A.M. Electrooculographic and electroencephalographic measurements were recorded, abdominal and rib-cage movements were measured using respiratory inductance plethysmography (Respitrace'"; Ambulatory Monitoring Inc., White Plains, NY), air flow at the nose and the mouth was measured using a CO 2 gas analyzer (Beckman Instruments, Fullerton, CAl, and arterial oxygen saturation was documented using an ear oximeter (Criticon, Markham, Ontario, Canada). All sleep data were recorded continuously on a multichannel polygraph. Apnea was defined as the cessation of airflow at the nose and mouth for 10 s or more and classified as central, mixed, or obstructive according to standard criteria (20). The apnea index was calculated as the mean number of apneas per hour of sleep, and the mean duration of apnea was computed. The mean Sa02 during wakefulness, minimal Sa02 during sleep, and the percentage of sleep time spent below saturations of 90 and 800/0 were calculated. Hypoxic-Hyperoxic Responses At least 24 h after the sleep study, the patients came to the laboratory where hypoxic-hyperoxic ventilatory responses were documented. The patients wereinstructed not to smoke and to have no drinks with caffeine for at least 8 h before the study. The methods used for these ventilatory responses has been fully described in a previous publication (7). During the experiment the patients were seated in a comfortable chair, distracted with nonrhythmic music, and observed closely to ensure wakefulness. They were instrumented with electrocardiogram leads and an ear oximeter to monitor heart rate and O 2 saturation, respectively.Wearing noseclips, they breathed through a low resistance unidirectional valve, the inspiratory side of which was connected to a pneumotachograph and then, through a manifold, to four tubes 3.2 em in diameter. Three ofthe tubes were connected to balloons containing nitrogen, oxygen, and 8.5% oxygen in nitrogen, and the fourth tube was open to room air. Each tube could be occluded by inflating a small rubber balloon placed near the manifold; this could be done without auditory clues and did not alert the subjects. The dead space of the device from mouth to tube entrance was approximately 150ml. Partial pressures of end-tidal O 2 (PET02) and CO 2 (PETC02) were measured by a fuel cell O 2me-
ter (E. Jaeger, Wiirzburg, Germany) and an infrared CO 2meter (Beckman) that sampled the expiratory side of the valve. The pneumotachograph output was integrated to give tidal volume (VT). VT, Sao., PET02 and PETC02 were recorded on a strip chart recorder. The patients breathed room air until PETC02 and ventilation stabilized, usually within 3 to 5 min. Then the inspirate was changed to N 2 for two to three breaths, during which PET02 decreased rapidly to about 55 mm Hg. The inspirate was then switched to 8.5% O 2, which maintained PET02 at approximately 50 mm Hg. After 30 to 45 s, hypoxia was terminated abruptly by switching to 100% O 2, so that PET02 of the first hyperoxic breath was at or above the normoxic baseline, whereas that of the second was above 150mm Hg. Hyperoxia was sustained for 2 min and terminated by switching the inspirate to room air. When PET02 returned to normoxic baseline, a stabilization period of about 2 min of room air breathing followed, and then the same hypoxic-hyperoxic procedure was repeated. Three to four such runs were performed in each patient. Runs associated with swallowing or coughing during the critical period of transition to O 2were excluded from the analysis. In addition, in nine patients the effect of switching the inspirate from room air to 100% O 2, was tested. The control group was seven normal nonsnoring adults (six male and one female). Their mean age was 42 ± 3 (SEM) yr, similar to that of the patients. None was obese or complained of sleep disorders. Three hypoxichyperoxic runs were performed in each subject, using the same method as in patients. Minute ventilation (VI), VT, inspiratory time (TI), expiratory time (Ts), and total breath duration (Ttot) were measured breath by breath for 30 s prior to hypoxia (baseline), during hypoxia, and for 2 min after switching to 100% O 2. All breaths during hypoxia and hyperoxia were expressed as percentages of the mean baseline value. The results for all tests in a given subject were averaged at 5-s intervals, the individual values being obtained by interpolation between the adjoining breaths.
To assess STP in each subject, the average hyperoxic ventilation was plotted logarithmically as a function of time, assuming zero time to be 5 s after institution of hyperoxia, and half time decays to control values (T J/z) were calculated. In addition, the lowest hyperoxic ventilation was noted. Data were analyzed by an unpaired t test and the nonparametric Mann-Whitney test on two-way analysis of variance (ANOVA). TJ/z and the lowest hyperoxic VI were related to other variables by linear regression using the least-squares method; p < 0.05 was considered statistically significant. Values were expressed as mean ± SE.
Results
Patient characteristics are shown in table 1.On average, the patients wereobese, but they had normal spirometric and blood gas determinations. Two patients (Patients 8 and 9) had borderline high values of Paco2' Only one (Patient 9) had Pan, of less than 70 mm Hg.
Sleep Studies All patients displayed repeated episodes of upper airway occlusion during sleep. Invariably, obstruction occurred when respiratory effects, as detected by inductance plethysmography, were low and in some cases absent (mixed apnea). In addition, during sleep, all patients demonstrated periodic breathing. For the group the mean apnea index was 52.4 ± 6.9 events/h, and the mean apnea duration was 22.9 ± 2.1 s. Significant desaturation was observed in all patients, and in 10 the minimal Sao, was less than 80070. Nine patients spent more than 10070 of their total sleep time with Sao, less than 90%. Individual sleep-related parameters are shown in table 2. Analysis of apneas (table 3) showed that five patients (Patients 2, 4, 8, 9, and 12) exhibited some central apneas. Patient 7 had only obstructive apneas.
TABLE 1 PATIENT CHARACTERISTICS Patient No.
Age
Weight
Height (em)
Pacoz (mmHg)
FVC
(kg)
Pao z (mmHg)
FEV1
(yr)
(% pred)
(% pred)
1 2 3 4 5 6 7 8 9 10 11 12 Mean ± SE
32 54 50 55 53 35 43 47 60 51 47 37 47 2.5
110 91 99 100 99 130 97 90 106 93 93 100 100.7 3.1
172 175 177 174 172 173 167 166 168 178 171 166 172 1
84 76 70 80 85 80 75 70 68 88 72 87 77.9 2.1
35 37 43 40 40 35 40 46 47 41 37 40 40.1 1.1
84 117 65 80 95 98 106 72 90 92 94 93 91.5 4.1
87 118 76 79 96 95 99 75 88 95 99 94 91.8 3.4
1252
GEORGOPOULOS, GIANNOULI, TSARA, ARGIROPOULOU, PATAKAS, AND ANTHONISEN
TABLE 2 SLEEP STUDY PARAMETERS
Patient No.
1 2 3 4 5 6 7 8 9 10 11 12 Mean ± SE
AI
TST
REM
AI
AD
206 191 325 250 293 315 328 319 300 327 302 180 278 16
14.2 14.4 31.8 17.8 10.1 19.2 23.6 25.3 21.1 15.4 6.7 14.9 17.9 2.0
79.5 35.3 42.2 55.0 56.3 78.3 41.2 87.7 63.2 14.0 13.6 63.0 52.4 6.9
17.7 28.9 32.3 24.3 32.1 23.7 32.8 20.1 14.2 15.0 12.8 20.3 22.9 2.1
TST with Sao 2 < 90% (%)
27.8 4.2 24.1 25.2 16.0 28.1 11.4 24.2 26.0 2.4 33.7 8.0 19.3 3.0
79 82 50 50 70 54 54 65
n
85 62 72 67 4
Definition of abbreviations: TST = total sleep time (min); REM .. rapid eye movement (% of = apnea index (events/h); AD = mean apnea duration (s).
Hypoxic-Hyperoxic Ventilatory Responses Baseline ventilation and PETC0 2 are shown in table 4. There was no significant difference between the patients and the control subjects with regard to ventilation, tidal volume, frequency, or PETC02•
A total of 57 hypoxic-hyperoxic runs were analyzed, three in each subject. The durations of hypoxic exposure were simiTABLE 3 APNEA CLASSIFICATION* Patient No.
Minimal Sa0 2
Obstructive (%)
Central (%)
Mixed (%)
81.2 10.0 91.3 80.7 86.9 95.4 100.0 68.0 16.8 91.2 84.2 52.8
0 4.3 0 4.1 0 0 0 7.9 31.9 0 0 3.5
18.8 85.7 8.7 15.2 13.1 4.6 0 24.1 51.3 8.8 15.8 43.7
1 2 3 4 5 6 7 8 9 10 11 12
* Percentage of total number of apneas.
lar in the two groups, averaging 38.8 ± 1.3 s in the patients, and 39.3 ± 2.0 in the control subjects. During hypoxia PET02 decreased exponentially and reached a plateau value of about 50 mm Hg approximately 30 to 45 s after institution of hypoxia. Hypoxic changes in VI and PETC 02 are shown in figure 1, and they did not differ significantly between normal subjects and patients. There was, on average, little change in VI in the first 10 s of hypoxia. Thereafter, VI increased significantly, and it may not have reached a steady state before the end of hypoxia; VI continued to increase during the first 5 s of hyperoxia, amounting to 165 and 152070 of baseline in patients and in normal subjects, respectively. The increase in hypoxic ventilation was mainly due to increase in VT. In patients, VT increased 49070 and frequency increased 11070, whereas the corresponding values in normal subjects were 61 and 5 070. Changes in PETC02 reflected the changes in VI, the
c 0
i
TABLE 4 BASELINE (ROOM AIR) VALUES OF VENTILATION AND PETCOz IN NORMAL SUBJECTS AND PATIENTS VT (L) Normal subjects ± SEM Patients ± SEM
0.55 1.1 0.59 0.04
VL flmin
(Llmin)
PETCO z (mmHg)
16.1 0.70 14.4 1.0
8.72 0.70 8.50 0.65
36.8 1.2 38.4 1.1
Definition of abbreviations: VT .. tidal volume; f - breathing frequency; VI - minute ventilation; PETC02 .. partial pressure of end-tidal CO2,
TST);
~CD >
maximal decline averaging 3.03 ± 0.64 mm Hg (range, 8.4 to 0.74) in patients and 2.90 ± 1.04 mm Hg (range, 8.2 to 0.1) in normal subjects. Breath-by-breath changes in Sao, are not reported because of the slow response of the ear oximeter. In all subjects the lowest value of Sao, was observed several seconds after switching the inspirate to hyperoxia. In the normal subjects, abrupt termination of hypoxia with hyperoxia was associated with a slow ventilatory decay to baseline (figure 2). Mean ventilation decreased to baseline several seconds after institution of hyperoxia, and, despite the presence of hypocapnia and hyperoxia, no apparent undershoot in VI could be detected. Mean hyperoxic VI followed an exponential course, with r = 0.94 and TYl = 9.7 s. In the patients, the hyperoxic ventilatory response pattern was different from that in the normal subjects. Mean hyperoxic ventilation also declined exponentially (r = 0.98), with a TYz of 4.8 s, and it reached a value of 810J0 of baseline at 45 s of hyperoxia, remaining lower than the baseline for several seconds (figure 2). The lowest VI attained during hyperoxia was 65.9 ± 4.7070 of the control value in the patients, which was significantly less (p < 0.01, nonparametric MannWhitney test) than the nadir value of 85.0 ± 4.9070 observed in the normal control subjects. In the patients, the relative hypoventilation was due mainly to a decrease in tidal volume (figure 3), which was significantly lower than that of the normal subjects at 30,35,40, and 45 s of hyperoxia (P < 0.05, two-way ANOVA).Although during hyperoxia breathing frequency tended to be lower than that of the normal subjects (figure 4), largely because
160 150 140 130 120 110 100
6
3
0
10
20
30
40
50
100 N
0
...0w
95
D.
90
50
0 Time (sec)
Fig. 1. Average minute ventilation and end-tidal Poo, (PETCOz) as a function of time during hypoxia. Both are expressed as percentages of mean prehYPOxic values. Bars are ± SE. Each circle is the average for 12 patients (closed circles) and seven normal subjects (open circles), except at 40 and 45 s where numbers are smaller as indicatad in the upper panel. This was because the length of hypoxic exposure varied slightly from subject to subject.
1253
OBSTRUCTIVE SLEEP APNEA AND SHORT-TERM POSTSTIMUWS POTENTIATION
120 160
G
.5
115
'ii
at
140
tV
.Q
G
'0
'ii
C
.5 at
•
>-
120
C>
.Q
C
C
. C
~
~
105
CD :::l
'0
a
110
c:r CD
~
100
.
100
01 C
£
CD
95
cD
80
90'
0
20
40
60
80
100
120
Fig. 2. Average minute ventilation during hyperoxia after hypoxia. Ventilation is expressed as a percent of mean prehypoxic values. Open circles indicate normal subjects; closed circles indicate patients. SE bars shown at selected points are representative of variability in responses.
of prolongation of Ts, neither frequency nor Th differed significantly between patients and control subjects. Although the patients as a group differed from the normal control subjects, there were substantial differences among patients and we subdivided them into two groups. Seven patients (Group A: Patients 2,4, 6, 8, 9, 10, and 12) exhibited T Y2 less than 4 s and a nadir hyperoxic VI that was less than 720/0 of baseline. The remaining five patients (Group B: Patients 1, 3, 5, 7, and 11) had TY2 greater than 4.5 s, and in all but one (Patient 11) the lowest hyperoxic VI was greater than 78070 of baseline. In Group A mean hyperoxic VI declined rapidly, with a mean T Y2 of 3.0 s. Mean hyperoxic ventilation reached a value of 72% of baseline at 30 s of hyperoxia, and it did not return to baseline for the next 60 s (figure 5). The relative hypoventilation was due to a decrease in both tidal volume
.
· ·
160
140
'ii
.Q
'0
t.
E :::l '0
> 'ii
120
100
'U
i= 80 0
20
60
20
40
60
80
100
120
TIme (sec)
Time (sec)
.5
0
80
100
120
TIme (sec)
Fig. 3. Average tidal volume, expressed as a percent of mean prehypoxic values, during hyperoxia after hypoxia. Symbols as in figure 2. Asterisks indicate significant difference from normal subjects (p < 0.05, two-way ANOVA).
Fig. 4. Average breathing frequency, expressed as a percent of mean prehypoxic values, during hyperoxia after hypoxia. Symbols as in figure 2.
1~~
115 105
110 100
00 0
and frequency (figure 5), with frequency being low largely because Th was prolonged. 11did not differ between groups, whereas in Group A, Ts was significantly greater than in Group B at 25, 30, and 35 s of hyperoxia. Six of the seven patients in Group A had breaths with Th > 5 s, and three (Patients 4, 8, and 9) had breaths with Th > 10s, meeting some definitions of apnea. On the other hand, in patients in Group B, hyperoxia was associated with a slower ventilatory decay to baseline, with a mean T Y2 of 6.2 s (figure 5). Mean hyperoxic ventilation reached 85% of baseline at 45 s of hyperoxia and returned to control at 55 s. None of these patients exhibited Th > 5 s during hyperoxia. The patients in Group B did not differ from those in Group A in terms of baseline characteristics, severity of sleep apnea syndrome, hypoxic ventilatory responses, or changes in PETC02. However, the five patients who exhibited pure central apneas during sleep (table 3) were all included in Group A. Among the patients, TY2 did not correlate significantly with the duration of hypoxic exposure, the magnitude of hypoxic change in VI, VT, frequency, or PETC02 or baseline Pa02. In addition, there was no relationship between T Y2 and sleep-related parameters such as apnea index, minimal Sa02, mean duration of apnea, and percent of total sleep time with Sa02less than 90 and 80%, as well as with apnea characteristics. Among the above parameters, the lowest hyperoxic VIwas correlated significantly only with the hypoxic changes in PETC02 (r = - 0.58). In the normal subjects none of the above relationships was significant. In individual subjects the highest and lowest hyperoxic VIand hypoxic changes of PETC02 in individual runs did not differ significantly (one-way ANOVA), in-
.....
......
**~...
.
.
95
~
40
~
80
100
1~
Time (sec)
Fig. 5. Ventilation, tidal volume, and breathing frequency, expressed as a percent of mean prehypoxic values, during hyperoxia after hypoxia in patients of Group A (closed circles) and of Group B (open circles). Asterisks indicate significant difference from Group B (two-way ANOVA).
dicating reproducibility of the hypoxichyperoxic responses. In nine patients (five from Group A and four from Group B), hyperoxia after room air breathing without an intervening period of hypoxia resulted in little decrease in VI. Average VIat 5-s intervals over the first minute of O 2 breathing varied from 91 to 108% of control values in a nonsystematic way. This hyperoxic response did not differ between the two groups of patients. Discussion The patients studied had the typical symptoms and signs of OSA. All were obese and had daytime hypersomnolence and normal or near normal lung function. During sleep all showed a significant number of obstructive apneas, with additional mixed and central apneas in all but one. Apneas were associated with significant desaturation, and most patients spent an appreciable percentage of their sleep time with Saoj less than 90%. The diagnosis of OSA in this group of patients was well established. None of the subjects were aware of the physiologic purpose of the hypoxichyperoxic experiments, or what changes in ventilation were expected. Furthermore, at the end of each experiment careful questioning revealed that none of them could define precisely the time of transition from hypoxia to hyperoxia, and most were not able to recall the number of hypoxic exposures. Indeed, six patients and three normal subjects were un-
1254
aware of any different sensations during hypoxia. Therefore, we can assume with confidence that the breathing patterns observed during hypoxia and hyperoxia were not due to exogenous stimuli. Our apparatus was designed to achieve rapid changes in alveolar P0 2. In all subjects PET02 at the end of the first hyperoxic breath was at or above the normoxic baseline, and that of the second was above 150 mm Hg. In previous studies of the onset of hypoxia in normal subjects (7, 8), we found that ventilation increased about 5 s after PET02 reached 75 mm Hg, so that it is reasonable to assume that the hypoxic stimulus was withdrawn some 5 s after the first hyperoxic breath in these experiments (21) while PETC02 was lower than baseline levels. Despite the hyperoxia and hypocapnia, VI remained above baseline values in the normal subjects for some time, declining slowly to baseline without a significant undershoot (figure 2). These results are similar to our previous findings in normal young and elderly adults (7, 8). We have interpreted this gradual decline in ventilation as indicative of STP or after-discharge. On the other hand, in patients with OSA, VI exhibited a rapid drop below baseline, averaging 81% of baseline after 45 s of hyperoxia and returning to baseline only after 85 s of hyperoxia (figure 2). This ventilatory depression was mainly due to a tidal volume decrease, which was significantly lower than that of the normal subjects at 30, 35, 40, and 45 s of hyperoxia. In addition, the lowest hyperoxic ventilation was also significantly lower than that of the normal subjects. The hyperoxic ventilatory depression was particularly prominent in seven patients (Group A) whose ventilation during hypoxia was consistent with absent or greatly diminished STP (figure 5). In the other five patients (Group B), an appreciable ventilatory STP was evident (figure 5). These results suggest that a significant number of patients with OSA have decreased STP. It should be noted that our test for STP was an artificial one in that hypoxia was induced by changing the inspirate. In OSA, hypoxia is largely the result of hypoventilation, which is attended by an increase in CO 2, not by a decrease, as in our tests. A combined hypoxic-hypercapnic stimulus would be inappropriate for the detection of STP, however, for although PET02 is a reasonable indicator of hypoxic stimulus during rapid transients, this is not true for PETC02 due to the relatively slow dynamics of change
GEORGOPOULOS, GIANNOULI, TSARA, ARGIROPOULOU, PATAKAS, AND ANTHONISEN
in CO 2 at the level of the central chemoceptor (22-24), which can be accentuated by changes in cerebral blood flow. Indeed, had we employed an hypoxic-hypercapnic stimulus, it is likely that we would have observed considerable poststimulus hyperventilation because of residual activation of central chemoceptors by CO 2, It is possible that these factors influenced our results, but they would be unlikely to explain the differences we observed between groups. The hypoxic-hyperoxic exposures we used were designed to demonstrate STP as unequivocally as possible by avoiding hypercapnia, and our results indicate that this mechanism appears to be deficient in at least some subjects with OSA. It is conceivable that the differences we observed were because the patients with OSA were asleep during testing and the normal subjects were not. There is evidence that sleep depresses posthypoxic hyperoxic ventilation in goats (27) and in humans (28) in the presence of hypocapnia. We think it unlikely that our patients with OSA fell asleep during STP testing. We observed them closely and they did not give behavioral evidence (14), indicating low levels of wakefulness during the approximately 20 min it took to complete the tests. Further, the patients with OSA did not demonstrate the irregular breathing that characterized their sleep studies, except transiently immediately after the institution of hyperoxia. It has been proposed that OSA is closely related to periodic breathing (11, 17-19), with patients cycling between airway occlusion when ventilatory drive is low and arousal when drive is high. There is evidence that inspiratory effort, as assessed by measurement of diaphragmatic EMG and esophageal pressure, waxes and wanes in OSA, with airway occlusion occurring when efforts reach a nadir. This is best exemplified by patients who develop obstruction during central apneas. There are reasons to believe that periodic breathing and OSA may be mutually reinforcing phenomena. In both normal and abnormal subjects, periodic breathing is more common during sleep than during wakefulness. With sleep there is hypoventilation, a reduction in cardiac output, and an increase in the apneic threshold (13, 14), all of which tend to promote periodic breathing (10, 11, 15), as does the reduced lung gas stores consequent to recumbency (12). In patients with OSA, upper airways resistance is high and varies greatly and inversely with inspiratory drive (19). Further, sleep blunts respiratory load com-
pensation (13), so that increases in resistance produce larger decreases in ventilation than in the awake state. Episodic airway occlusion, if it occurs when respiratory drive is low, is likely to give rise to cyclic breathing since it produces episodic ventilatory stimulation and swings in ventilation and ventilatory drive. During sleep our patients with OSA all showed unstable breathing patterns even when airway occlusion did not occur, and in many instances obstruction occurred after apnea or near apnea. Other investigators have made the same observations (17-19, 25, 26). In this setting, respiratory STP would tend to stabilize ventilation. If hypoxia consequent to airway occlusion activated STP, then the hypoxic responsehyperventilation and arousal- would be followed by a gradual decrease in ventilation. In the absence of STP, the hyperventilation of arousal would be followed by a more abrupt decrease in ventilatory drive and ventilation, rendering the subject more vulnerable to another episode of airway occlusion. This general phenomenon is illustrated by the seven patients of Group A. With relief of hypoxia, ventilation fell precipitously to levels below control values, and episodes of apnea were observed, though the patients were awake (figure 5). In these 'subjects the absence or reduction of STP may have played a part in the pathogenesis of their disease. On the other hand, the other five patients with OSA clearly demonstrated that the syndrome can develop in subjects with STP that is not clearly abnormal according to our measurements. We assessed STP in awake patients after an hypoxic stimulus. We believe that even in the absence of hypercapnia, hypoxia is the most physiologically relevant stimulus in patients with OSA, but we are uncertain as to the applicability of our measurement of STP to the sleeping state. In animals STP is not affected by anesthetics, decerebration, or decerebellation (3), so it would seem likely that respiratory STP is not obliterated by sleep. However, there is evidence that STP is reduced in slow-wave sleep in goats (27), and Badr and coworkers (28) observed ventilatory inhibition with hyperoxia after brief hypoxic exposures in sleeping humans. Thus, it may well be that waking measurements overestimate STP during sleep. It is possible that our Group B patients would have demonstrated a defect in STP had they been examined during sleep; the fact that their tidal volume fell below control levels during
OBSTRUCTIVE SLEEP APNEA AND SHORT·TERM POSTSTIMUWS POTENTIATION
hyperoxia suggests that this might have been the case. It is also possible that our normal subjects would have shown little evidence of STP had they been studied during sleep. On the other hand, STP in OSA would be most significant during and immediately after arousal, so that waking measurements may have been appropriate. Although it is relatively easy to postulate how the presence or absence of STP might influence OSA, it is not clear why seven out of 12 patients with OSA had clear-cut defects in STP, a much larger number than are encountered in series of normal subjects (7, 8). We have shown that modest hypoxia interferes with STP in normal subjects; this effect certainly occurs after 20 to 30 min at an arterial saturation of 80%, and it probably occurs after as little as 5 min (7, 29-31). Our patients with OSA were not hypoxic while awake, and oxygenation did not differ between our two patient groups, so hypoxia per se probably did not explain the defect in STP in seven of our patients. On the other hand, we did employ hypoxia in our test for STP, and it is possible that some of our patients were abnormally sensitive to the "hypoxic depression" that accompanies loss of STP. This was not evident in terms of hypoxic ventilatory responses, which did not differ between patient groups. Abnormal levels of endogenous opiates have been found in the cerebrospinal fluid of patients with OSA (32), and it is possible that these substances reduced STP in our patients. It is not clear, however, why endogenous opiates should be increased more in one group of subjects than in another when the degree of obstruction and sleep disorder did not differ. It is perhaps more attractive to postulate that a certain number of otherwise normal people have greatly reduced STP as we measure it; we have observed several such persons (8). If this were true, people with defective STP who became obese would be at high risk for developing OSA, and unselected groups of such patients would show a high incidence of defective STP. We have restudied one of our Group A patients (Patient 8) after his apneic episodes were treated successfully for 2 months with CPAP. Results were the same as those of the initial study, suggesting that his defective STP was not a consequence of the nocturnal obstructive episodes. As noted earlier, the fact that five of our patients with OSA demonstrated normal or near normal STP suggests that the syndrome can develop in the absence
of a defect in STP as we measure it. Indeed, our measurements of STP did not relate to any of the indices of OSA severity noted in table 2. This is not surprising since the severity of OSA must be more dependent upon the morphology and function of the upper airways than any other factors, and these vary among patients. Indeed, it is possible that the presence or absence of STP is reflected in contraction of the muscles controlling the upper airways. Genioglossal contraction, which tends to maintain upper airway patency (33), also tends to reflect ventilatory drive, so it is possible that the absence of STP also causes unusual poststimulus instability of the upper airways. It is also possible that STP as measured in terms of ventilation does not apply or applies in a variable way to upper airway dilators. It is worth noting, however, that the preservation of STP appeared to predict one sleep-related event in these patients: all patients with central apneas were included in the group with defective STP. This again suggests that reduced STP may contribute to respiratory arrhythmias with or without airway obstruction. References 1. Eldridge FL. Postthyperventilation breathing: different effects of active and passive hyperventilation. J Appl Physiol 1973; 34:422-30. 2. Eldridge FL. Central neural respiratory stimulatory effect of activerespiration.J Appl Physiol1974; 37:723-35. 3. Eldridge FL. Maintenance of respiration by central neural feedback mechanisms. Fed Proc 1977; 36:2400-4. 4. Eldridge FL, Gill-Kumar P. Central neural respiratory drive and after-discharge. Respir Physiol 1980; 40:49-63. 5. Swanson GD, Ward DS, Bellville lW. Posthyperventilation isocapnic hyperpnea. 1 Appl Physiol 1976; 40:49-63. 6. Tawadrous FD, Eldridge FL. Posthyperventilation breathing patterns after active hyperventilation in man. J Appl Physiol 1974; 37:353-6. 7. Georgopoulos D, Bshouty Z, Younes M, Anthonisen NR. Hypoxic exposure and activation of the after-discharge mechanism in conscious humans. J Appl Physiol 1990; 69:1159-64. 8. Ahmed M, Giesbrecht GG, Serrette C, Anthonisen NR. Ventilatory after-discharge in elderly humans (abstract). Am Rev Respir Dis 1990; 141:A379. 9. Fregosi RE. Short term potentiation of breathing in humans. 1 Appl Physiol 1991; 71:892-9. 10. Cherniack NS, Lougobordo GS. Abnormalities in respiratory rhythm. In: Cherniack NS, Widdicombe IG, eds. Handbook of Physiology. Vol II, part 2: The respiratory system. Bethesda, MD: American Physiological Society, 1986; 729-49. 11. Younes M. The physiological basis of central apnea and periodic breathing. Curr Pulmonol1989; 10:265-326. 12. Agostoni E, Hyatt RE. Static behaviour ofthe respiratory system. In: MacKlem PT, Mead 1, eds. Handbook of physiology. Vol 3: The respiratory
1255 system.Bethesda, MD: American PhysiologicalSociety, 1986; 113-30. 13. Dempsey JA, Skatrud JB. Fundamental effects of sleep state on breathing. Curr Pulmonol 1988; 9:267-304. 14. Phillipson EA, Bowes G. Control of breathing during sleep. In: Cherniack NS, Widdicombe JG, eds. Handbook of Physiology. Vol 2: The respiratory system. Bethesda, MD: American Physiological Society, 1986; 649-89. 15. Khoo MCK, Kronauer RE, Strohl KP, et al. Factors inducing periodic breathing in humans: a general model. J Appl Physiol 1982; 53:644-59. 16. Webb P. Periodic breathing during sleep. J Appl Physiol 1974; 37:899-903. 17. Onal E, Lopata M, O'Connor TD. Pathogenesisof apnea in hypersomnia-sleepapnea syndrome. Am Rev Respir Dis 1982; 125:167-74. 18. Onal E, Lopata M. Periodic breathing and the pathogenesis of occlusive sleep apneas. Am Rev Respir Dis 1982; 126:676-80. 19. Martin Rl, Pennock BE, Orr WC, et al. Respiratory mechanics and timing during sleep in occlusivesleep apnea. J Appl Physiol1980; 48:432-7. 20. Guilleminault C, Tilkian A, Dement WC. The sleep apnea syndromes. Annu Rev Med 1976; 27:465-84. 21. Lange RL, Horgan JD, Botticelli IT, Tsagaris T, Carlisle RP, Kuida H. Pulmonary to arterial circulatory transfer function: importance in respiratory control. J Appl Physiol 1966; 21:1281-91. 22. Fujishima MP, Scheinberg P, Busto P, Reimuth O. The relation between cerebral oxygenconsumption and cerebral vascular reactivity to carbon dioxide. Stroke 1971; 2:251-7. 23. Neubauer JA, Santiago TV,Posner MA, Edelman NH. Ventral medullary pH and ventilatory response to hyperperfusion and hypoxia. J Appl Physiol 1985; 58:1659-68. 24. Lambertsen CJ, Kough RH, Cooper DY, Emmel GL, Loeschcke HH, Schmidt CF. Oxygen toxicity. Effect in man of oxygen inhalation at 1 and 3.5 atmospheres upon blood gas transport, cerebral circulation, and cerebral metabolism. J Appl Physiol 1953; 5:471-86. 25. Strohl K. Control of upper airwayduring sleep. In: Edelman NH, Santiago TV, eds. Breathing disorders of sleep. Vol5. New York: Churchill Livingstone, 1986; 115-37. 26. Skatrud JB, Dempsey JA. Airway resistance and respiratory muscle function in snorers during NREM Sleep. J Appl Physiol 1985; 59:328-35. 27. Daristotle L, Petrozzino 11, Santiago TV. Effect of sleep-wake state on ventilatory afterdischarge (abstract). FASEB J 1991; 5:AI478. 28. Badr S, Skatrud JB, Dempsey JA. Ventilatory inhibition following active hyperventilation during NREM sleep (abstract). Am Rev Respir Dis 1990; 141:AI25. 29. Holtby SG, BerezanskiD,Anthonisen NR. The effect of 100070 oxygen on hypoxic eucapnic ventilation. J Appl Physiol 1988; 65:1157-62. 30. Georgopoulos D, Holtby SG, Berezanski D, Anthonisen NR. Aminophylline effects on ventilatory response to hypoxia and hyperoxia in normal adults. J Appl Physiol 1989; 67:1150-6. 31. Georgopoulos D, Walker S, Anthonisen NR. Increased chemoreceptor output and ventilatory response to sustained hypoxia. J Appl Physiol1989; 67:1157-63. 32. Gislason T, Almqvist M, Boman G, Lindholm CE, Terenius L. Increased CSF opioid activity in sleep apnea syndrome. Regression after successful treatment. Chest 1989; 96:250-4. 33. Sullivan CE, Issa FG. Obstructive sleep apnea. Clin Chest Med 1985; 6:633-50.
