| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vol. 83, Issue 6, 1986-1997, December 1997
School of Behavioural Science, University of Melbourne, Parkville, Victoria 3052, Australia
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
FOOTNOTES
REFERENCES
ABSTRACT 
Trinder, John, Amanda Kay, Jan Kleiman, and Judith Dunai.
Gender differences in airway resistance during sleep.
J. Appl. Physiol. 83(6):
1986-1997, 1997.
At the onset of non-rapid-eye-movement (NREM)
sleep there is a fall in ventilation and an increase in upper airway
resistance (UAR). In healthy men there is a progressive increase in UAR
as NREM sleep deepens. This study compared the pattern of change in UAR
and ventilation in 14 men and 14 women (aged 18-25 yr) both during
sleep onset and over the NREM phase of a sleep cycle (from wakefulness
to slow-wave sleep). During sleep onset, fluctuations between
electroencephalographic alpha and theta activity were associated with
mean alterations in inspiratory minute ventilation and UAR of between 1 and 4.5 l/min and between 0.70 and 5.0 cmH2O · l
1 · s,
respectively, with no significant effect of gender on either change
(P > 0.05). During NREM sleep,
however, the increment in UAR was larger in men than in women
(P < 0.01), such that the mean
levels of UAR at peak flow reached during slow-wave sleep were ~25
and 10 cmH2O · l1 · s
in men and women, respectively. We speculate that the greater increase
in UAR in healthy young men may represent a gender-related susceptibility to sleep-disordered breathing that, in conjunction with
other predisposing factors, may contribute to the development of
obstructive sleep apnea.
respiration; respiratory instability; sleep onset; electroencephalogram
INTRODUCTION THE CHANGES IN RESPIRATORY ACTIVITY that occur from
wakefulness to non-rapid-eye-movement (NREM) sleep include an increase in upper airway resistance (UAR), an attenuation of the
neuromechanically mediated resistive load-compensating response, a
decrease in ventilation, and an increase in
PCO2 (4, 7, 8, 11, 12, 17, 18, 22,
26-28). Of particular interest to the present study are the
increase in UAR and fall in ventilation. In healthy young male subjects
both of these changes are initiated in synchrony very early during the
sleep onset process in association with the transition from alpha to
theta electroencephalographic (EEG) activity (9, 10). These
state-related changes become larger once the first signs of stage 2 NREM sleep emerge, and in healthy male subjects there is a progressive
increase in UAR as NREM sleep develops (11, 21, 26). These normal
sleep-induced respiratory changes render the system more susceptible to
disruption, and when exaggerated, may interact with other predisposing
factors to produce disorders of breathing during sleep such as
obstructive sleep apnea.
Some data indicate that the effect of sleep on respiratory activity may
differ between men and women. Although many studies have investigated
chemoresponsiveness during wakefulness and/or sleep in men vs.
women (for example, Refs. 1, 2, 5, and 24), results have been variable.
Studies investigating gender differences in respiratory mechanics are
relatively few. Brooks and Strohl (3) found that although the
pharyngeal airway was larger in men than in women, changes in
pharyngeal area in association with alterations in lung volume were
greater in men. This finding suggests that the airway may be more
compliant in men than in women. Some data indicate that UAR is higher
in men than women during wakefulness (25), although, in a later study
using younger subjects, Popovic and White (14) recorded similar waking
pharyngeal airway resistance in men and women. White (23) found that
"effective inspiratory impedance" increased during sleep more in
men than in women, although airway resistance was not measured.
Finally, Popovic and White (14) reported that waking genioglossus
electromyographic (EMG) activity in women tended to be higher during
the luteal phase of the menstrual cycle, lower in the follicular phase,
and lowest in postmenopausal women. It was concluded that decreasing hormone levels may be associated with reduced upper airway dilator muscle activity. The effect of sleep, however, was not assessed, and no
men were studied.
In summary, several findings suggest that respiratory mechanics may
differ between men and women. However, to our knowledge no study has
investigated the development of sleep-induced changes in ventilation
and UAR across the sleep period as a function of gender. This study was
therefore designed to determine, first, whether the synchronous and
reciprocal oscillations in ventilation and UAR that occur in
association with fluctuations in state during sleep onset (9, 10) are
larger in men than in women, and second, whether the progressive
increment in UAR observed in men as NREM sleep deepens (11, 21, 26) is
larger in men than in women.
METHODS Subjects and Design
The purpose of the study was to examine changes in ventilation and UAR both during sleep onset and over a full NREM sleep period from wakefulness to SWS. Not all subjects, however, attained SWS in the laboratory, and thus each subject was assigned to a subgroup on the basis of whether she/he attained SWS on the second experimental night. These groups will be referred to as the male and female SWS groups [M-SWS (n = 8) and F-SWS (n = 10), respectively] and the male and female no SWS groups [M-NSWS (n = 6) and F-NSWS (n = 4), respectively]. The data from one subject in the M-NSWS group were discarded (see RESULTS), and thus results presented for the M-NSWS group comprise data from five subjects. Anthropometric data for subjects in each group are shown in Table 1. With an average age of 20.31 yr (±2.18 yr), the men were just over 1 yr older than the women (mean age 19.14 ± 0.95 yr). The men also had a slightly higher body mass index (22.96 ± 2.23) than the women (20.33 ± 1.44) although, with the exception of one male subject, who had a body mass index of 27.36 (subject MA), all subjects were well within the normal range (see Table 1).
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
General Laboratory Procedures
Subjects were requested to refrain from consuming alcohol and caffeine for the entire day of each sleep study. After the subject arrived at the laboratory at ~2100, the recording equipment was attached as described in Ventilation Measurement and UAR Measurement, and the subject retired to bed at 2200-2230. Subjects were required to maintain a supine position during data collection.The purpose of the first experimental night was to obtain data over multiple sleep onset periods so that changes in ventilation and UAR could be examined as a function of the frequent fluctuations in state that characterize this period. Data were collected by using a multiple sleep onset technique, by which the subject was requested to remain alert for ~10 min after lights out before allowing herself/himself to fall asleep. Once asleep, the subject was left undisturbed until ~5-10 min of stage 2 NREM sleep had occurred. After this time, the subject was awakened and kept awake until alert. This procedure was repeated until ~0300-0400, after which time data collection was terminated.
The purpose of the second night was to obtain data not only from the sleep onset period but also from one full NREM sleep cycle from wakefulness to SWS. Rather than being awakened after ~5-10 min of stage 2 NREM sleep, subjects were permitted to sleep for an extended period of time, and if at least 5 min of SWS occurred within ~30 min after lights out, the subject was assigned to the SWS group and the data from that sleep period were used in the appropriate analyses. If no SWS had occurred within ~30 min, the subject was awakened for a brief interval and another sleep period was commenced. This procedure was repeated until SWS was attained, or until ~0300-0400. If at least 5 min of SWS occurred within any of the sleep periods from the second night, the subject was assigned to a SWS group. If not, the subject was assigned to a NSWS group and the sleep period with the longest recording time was used in subsequent data analyses. Thus for each subject the data from one extended sleep period, either with or without SWS, were used for subsequent analyses. Data from each sleep onset period from the second night, including the extended sleep period, were collated with data from the first night for use in the analyses examining the sleep onset period.
All sleep and respiratory measurements were recorded by using a 16-channel Grass polygraph (model 7D). Occipital EEG, airflow, and pressure measurements were also recorded on an IBM-compatible 486 PC via a 16-bit analog-to-digital converter sampling at 100 Hz.
Ventilation Measurement
For measurement of airflow, subjects wore a modified continuous positive airway pressure face mask (Vital Signs) that covered the nose and mouth. Route of breathing was not controlled. The upper port of the mask was sealed, and a heated pneumotachograph (Morgan) was attached directly to the lower port. The total dead space of the mask and pneumotachograph was ~155 ml, depending on facial configuration. The pneumotachograph was connected to a differential pressure transducer (Validyne DP45-14) and a carrier demodulator (Validyne CD15). Off-line analysis of the flow signal was used to calculate inspiratory minute ventilation (
I)
extrapolated from individual breaths.
UAR Measurement
UAR was calculated by using simultaneous recordings of airflow (see Ventilation Measurement), mask pressure, and epiglottal pressure. The airflow, mask pressure, and epiglottal pressure systems were phase tested, and the lengths of tubing on the mask pressure and airflow systems were adjusted so that the phase angle difference between signals was no more than 2° at 1 Hz. This phase difference was smaller than that which could be detected given that the sampling rate was 100 Hz. Mask pressure was recorded from a port in the center of the mask that was connected to a pressure transducer (Validyne DP45-28) and carrier demodulator (Validyne CD15). The other side of the pressure transducer was connected to an equal length of tubing left open to the atmosphere. Epiglottal pressure was recorded by using a transducer-tipped catheter (model MPC-500, Millar). After administration of a long-lasting nasal decongestant (oxymetazoline hydrochloride, 0.05%) and topical anesthesia of one nostril by using lidocaine gel (2%), the catheter was inserted transnasally, positioned visually ~1 cm below the tongue base, and secured at the nose with tape. Oxymetazoline hydrochloride was used to counteract any mucosal secretions due to catheter-induced irritation and to minimize any increase in UAR that may have occurred as a result of the airway space occupied by the catheter.The mask and epiglottal pressure signals were analyzed off-line by a computer algorithm that computed the pressure gradient across the upper airway segment (epiglottis to mask). Resistive pressure was automatically zeroed at points of zero airflow (end inspiration and end expiration) to minimize the effects of any baseline drift in the Millar catheter signal. In addition, sections of data characterized by rapid baseline drift were discarded. The computer algorithm calculated UAR at 10-ms intervals throughout each breath by dividing pressure gradient by airflow. Several measurements of UAR were extracted. Average inspiratory resistance (Ravg) was calculated by computing the mean of all samples within the inspiratory phase of each breath, and resistance was also calculated at several discrete points within the inspiratory phase of each breath: at a flow of 0.2 l/s, at peak airflow (Rafpeak), at peak pressure gradient, and at the point of maximum resistance. The measurement selected for presentation in this study was Rafpeak, a measurement from the relatively linear portion of the pressure-flow curve. It should be noted, however, that preliminary analyses showed that although absolute values varied between the various indexes, the particular measure selected did not affect the general pattern of results across the sleep period.
In addition to the calculation of UAR, pressure-flow loops were constructed by sampling resistive pressure and airflow at 5% intervals through inspiration and 5% intervals through expiration. Average loops were then constructed by averaging data over breaths at each of the 40 sampling points.
Sleep Measurements
Scoring of sleep state. Sleep state was monitored by using gold cup surface electrodes to record EEG (central: C3/A2; and occipital: O1/A2), electrooculogram (EOG; vertically displaced at the outer canthus of each eye), and EMG (submental) signals. Each sleep period was divided into "phases" according to criteria described previously (11). Phase 1 represented quiet wakefulness (alpha EEG activity), and phase 2 was the period of fluctuating alpha and theta EEG activity before the first appearance of sleep spindles and/or K complexes. Phase 3 was also characterized by fluctuations in state but was distinguished from phase 2 by the fact that the theta state could include sleep spindles and K complexes, and the aroused state was identified either by alpha EEG activity or by events such as bursts of EMG activity, eye movements, and/or a change in EEG frequency. Phase 4 represented sustained stage 2 sleep, and phase 5 represented stages 3 and 4 NREM sleep. The term "phase" rather than "stage" was used because the phases did not correspond exactly to the stages classified by Rechtschaffen and Kales (15). During phases 2 and 3, state could fluctuate between wakefulness and sleep. Within these phases state was scored on a breath-by-breath basis, such that the EEG epoch associated with individual breaths was classified as either "wakefulness" or "sleep." In phase 2 the EEG for each breath was classified as either alpha (wakefulness) or theta (sleep). In phase 3 the EEG for each breath was classified as alpha (wakefulness), other indications of arousal such as those described above (wakefulness), or theta with associated sleep spindles or K complexes (sleep). In phases 1, 4, and 5, state was, by definition, constant and therefore the EEG state did not fluctuate on a breath-by-breath basis. Thus the EEG was characterized as constant alpha in phase 1, theta with sleep spindles and K complexes in phase 4, and dominant delta activity in phase 5. To discriminate between alpha and theta EEG activity and thus to identify and score phases 1 and 2, an automated period analysis (peak to peak) of the occipital EEG recording was used as described previously (9, 10). The automated EEG analysis was also used to identify alpha EEG activity during phase 3, but in addition visual scoring was used to identify other criteria indicating arousal, as described above. The identification of sleep spindles and K complexes during phase 3 was performed by visual scoring according to standard criteria (15), except that rather than being scored on a 20- or 30-s epoch-by-epoch basis, data were scored on a breath-by-breath basis according to the features of the EEG recordings associated with each breath. Scoring of sustained theta activity with associated sleep spindles and K complexes (phase 4) and dominant delta activity (phase 5) was performed according to standard criteria (15). If an arousal or a shift to a lighter sleep stage occurred during phase 5, data were discarded from the point of the arousal until resumption of a "depth" of sleep approximately equal to that observed before the arousal. Sections of data contaminated by body movements, swallows, or invalid epiglottal pressure recordings were discarded. As described in General Laboratory Procedures, sleep periods obtained by using the multiple sleep onset technique were terminated after 5-10 min of early stage 2 sleep and therefore comprised only phases 1-3. Sleep periods in which subjects attained SWS comprised all five phases, and the "extended" sleep periods in which SWS was attempted but not obtained comprised phases 1, 2, and an extended phase 3 period. Frequency analysis of EEG. For the extended sleep periods a spectral (fast Fourier transform) analysis of the occipital EEG recording was performed as described previously (11). Briefly, for the EEG epoch corresponding to each breath, the power in each of four frequency bands was calculated: 0.4-3 (delta), 3-8 (theta), 8-12 (alpha), and 12-20 (beta) Hz. The power in each band was then expressed as the proportion of total power within that EEG breath epoch. This analysis was conducted to determine whether the frequency characteristics of sleep over the extended sleep period were similar in the male and female subjects.Data Reduction and Analysis
To examine both the sleep onset period and the full NREM sleep cycle, the data were represented in two different ways. Sleep onset analyses. To examine respiratory activity as a function of fluctuations in state within phases 2 and 3, data were averaged on a breath-by-breath basis over state transitions by using procedures described previously (9, 10, 22). Briefly, for each point at which EEG activity (scored on a breath-by-breath basis) changed from wakefulness to sleep, breaths on either side of the EEG change were labeled according to their position in relation to the change (
1 for the
last breath before the transition, +1 for the first breath after the
transition, and so forth) and respiratory data were averaged at each
breath position. Data were also averaged over changes that operated in
the opposite direction (sleep to wakefulness), and separate averages
were calculated for these transition types in phases
2 and 3. The criterion
for a state transition was that at least two breaths in one state were
followed by at least two breaths in the other state. Because the number
of breaths in each of the pre- and posttransition states varied over
transitions, from a minimum of two upward, the number of breaths
contributing to the mean at each breath position decreased as the
sampling points moved further from the point of EEG change. Data were
averaged within subjects and then over subjects, and mean data for each subject at any particular breath position were included in analyses only if the mean comprised at least 10 breaths. With reference to these
analyses, the terms "alpha" and "theta" will be used to
refer to the waking and sleeping states, respectively, but it should be
noted that, for activity during phase
3, alpha will refer to either
alpha EEG activity or other indications of arousal and theta will refer
to theta activity either with or without associated sleep spindles and
K complexes.
The values used in statistical analyses to examine changes over state
transitions within phases 2 and
3 were the mean of pretransition breaths 5 to
1, inclusively, and the mean of
posttransition breaths +1 to
+5, inclusively, for each type of
transition (alpha to theta and theta to alpha). Thus respiratory data
during each of the alpha and theta states were analyzed according to
whether they represented the pretransition state (for example,
alpha EEG activity before an alpha-to-theta transition) or the
posttransition state (for example, alpha activity after a
theta-to-alpha transition). Data were analyzed separately for
phases 2 and
3, and thus this procedure yielded
eight values for each subject for each variable: a pre- and
posttransition alpha value for each of phases
2 and 3 and a pre- and
posttransition theta value for each of phases 2 and 3. Preliminary
statistical analyses showed that during phases 2 and 3 there were no
differences between the SWS and NSWS groups and, therefore, in further
analyses concerned with state transitions, the data for all men were
compared with those for all women. The values described above were
entered into a four-way analysis of variance (ANOVA) with a
between-subjects factor, gender (male vs. female), and three
within-subject factors, i.e., state (alpha vs. theta), phase
(phase 2 vs. phase
3), and order (pre- vs. posttransition). The
variables analyzed were I and
Rafpeak.
Investigation of the development of changes over the extended sleep
period.
To investigate the development of changes in respiratory activity over
the extended sleep period, each phase in the extended sleep period was
divided into 10 consecutive sections comprising equal numbers of
breaths, as described previously (11). For example, if for a particular
subject phase 1 consisted of a total of 80 breaths, that phase would be divided into 10 sections, each comprising 8 breaths. A mean value for each of these sections was then
calculated for each variable, resulting in 10 values for each phase (50 values over a sleep period comprising phases 1-5). To confirm that changes occurring over
phases 1-5 were related to
"deepening" sleep rather than the passage of time, data for the
NSWS groups were included in this analysis. For subjects in the NSWS
groups phases 1 and
2 were divided into 10 equal sections as was the case for subjects in the SWS groups. Phase
3 in these subjects, however, was prolonged and covered
a period of time similar to that spent in phases
3-5 in the SWS groups. Thus this phase was divided
into 30 equal sections that represented an extended period of early or
"unsettled" stage 2 sleep characterized by intermittent arousals,
resulting in a total of 50 values per sleep period. These
"standardizing" procedures were conducted so that data from
phases of varying duration could be averaged over subjects. If a
movement, swallow, or other disruption occurred at any point, data from
all channels were discarded for each breath during the disruption and
the data on either side of the disruption were treated as continuous.
Two-trend analyses (ANOVAs with polynomial contrasts for the
within-subject factor "position") were conducted by using
standardized values for I and
Rafpeak. Position was
used to refer to the 50 mean values across the sleep period. The first
trend analysis was a two-way ANOVA that used data from the two SWS
groups to assess the effect of gender over phases
1-5. The second trend analysis was a three-way
ANOVA that used data from all four groups to assess the effects of both
group (SWS vs. NSWS) and gender (men vs. women) over the sleep period
(phases 1-5 for SWS groups and
phases 1-3 for NSWS groups, where
phase 3 was extended).
In addition to calculation of the 50 progressive values across the
sleep period, mean values for each phase were calculated for each
subject by averaging all breaths within each of phases 1-5. Mean pressure-flow loops for
phases 1-5 were also calculated by averaging data at each of the 40 sampling points for all breaths within each phase within subjects and then over subjects.
For all statistical analyses, results were accepted as statistically
significant when P < 0.01.
RESULTS
Statistical analyses showed that state-related changes in
I and UAR did not differ between men and
women during the sleep onset period but that, once NREM sleep
became established, there was a more marked and progressive
increase in UAR in men than women.
I was maintained at similar
levels in men and women during NREM sleep despite the marked difference
in UAR.
Sleep Onset Analyses
Mean values forI and
Rafpeak during the pre- and
posttransition states (the mean of breaths
1 to 5
and the mean of breaths +1 to
+5) for each type of transition
during phases 2 and
3 are shown in Table
2. Breath-by-breath changes plotted over
transitions, collapsed over SWS and NSWS groups, are shown in Fig.
1. As shown, I and UAR changed in a reciprocal and
synchronous fashion across these transitions in both men and women.
I decreased and UAR increased over
alpha-to-theta transitions, and the reverse pattern of change occurred
over theta-to-alpha transitions. The four-way (gender-by-state-by-phase-by-order) ANOVA revealed that the main effect
of gender was not significant for either
I
(F[1,22] = 1.12, P = 0.301) or
Rafpeak
(F[1,22] = 1.42, P = 0.246). Furthermore, none of the
interactions involving gender approached statistical significance for
Rafpeak, although as shown in Fig.
1D there was a suggestion of a larger
increment in UAR over consecutive theta breaths before
phase 3 arousals in men compared with
that in women. Two significant interaction effects involved gender for
I. First, although
I tended to be higher in posttransition
than pretransition states in both genders, the difference was greater
in women, causing a significant gender-by-order interaction
(F[1,22] = 12.91, P = 0.002). In addition, this effect
occurred predominantly during phase 3,
resulting in a significant gender-by-phase-by-order interaction effect
for I
(F[1,22] = 18.30, P < 0.001). Neither of these effects
appeared to be related to UAR differences between the groups.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
I; A
and C) and peak airflow resistance
(Rafpeak;
B and
D) plotted over alpha-to-theta and
theta-to-alpha transitions during phases 2 and 3 in male and
female subjects. Mean values were obtained by centering each transition
at the point of EEG change and averaging values at each breath position
before and after the transition. Data were averaged over all
transitions of each type within subjects and then over subjects. A
subject's data for a particular breath position were included in group
average only if mean at that breath position comprised at least 10 breaths. Data were collapsed over slow-wave sleep (SWS) and no SWS
(NSWS) groups. During phase 3 the term
"alpha" represents either alpha EEG activity or other indications
of arousal (see text), and the term "theta" represents theta EEG
activity either with or without associated sleep spindles and K
complexes. See text for definition of phases 2 and 3.
As expected on the basis of previous studies (9, 10), the main
effect of state was significant for both
I
(F[1,22] = 80.55, P < 0.001) and
Rafpeak
(F[1,22] = 51.64, P < 0.001), such that ventilation
was higher in association with alpha EEG activity than theta EEG
activity and UAR was lower during alpha than theta EEG activity. These
effects were more extreme during phase
3 than phase 2,
resulting in significant state-by-phase interactions for
I
(F[1,22] = 55.27, P < 0.001) and
Rafpeak (F[1,22] = 29.16, P < 0.001). Overall, UAR was higher
during phase 3 than
phase 2 and higher during
pretransition than posttransition states, as reflected in significant
main effects of phase
(F[1,22] = 31.91, P < 0.001) and order, respectively
(F[1,22] = 43.04, P < 0.001) for
Rafpeak. The latter effects were
not significant for I, although the
tendency for I to be higher in
posttransition than pretransition states approached statistical
significance (F[1,22] = 55.27, P = 0.013). Changes in
I and UAR over theta-to-alpha transitions
tended to be greater than those over alpha-to-theta transitions,
particularly during phase 3. Thus
there were significant state-by-order interactions for
I
(F[1,22] = 29.13, P < 0.001) and
Rafpeak
(F[1,22] = 31.30, P < 0.001) and significant
state-by-phase-by-order interactions for
I
(F[1,22] = 8.72, P = 0.007) and
Rafpeak
(F[1,22] = 24.25, P < 0.001). The phase-by-order
interaction was significant for
Rafpeak
(F[1,22] = 23.27, P < 0.001) but not for
I
(F[1,22] = 1.23, P = 0.279).
In summary, changes in UAR over state transitions during sleep onset
did not vary significantly as a function of gender. Fluctuations in
state during the sleep onset process were associated with synchronous and reciprocal changes in I and UAR in
both men and women, and in both sexes these effects were larger during
phase 3 than during phase 2. Changes at theta-to-alpha
transitions tended to be greater than those at alpha-to-theta
transitions, particularly during phase
3.
Investigation of Development of Changes Over Extended Sleep Period
Table 3 shows the mean amount of time subjects spent in each phase of the selected extended sleep periods for the SWS and NSWS groups. For the SWS groups the mean amount of time spent in each of phases 1, 2, 4, and 5 was similar in men and women. The M-SWS group, however, spent approximately twice as much time as the F-SWS group in phase 3. Thus the total sleep period was, on average, ~12 min longer in the M-SWS than the F-SWS group. For the NSWS groups the mean amount of time men spent in phase 3 was more than twice that in women. However, rather than reflecting any meaningful differences in phase duration, this effect was due to the fact that, when deciding to terminate sleep periods, the experimenter made an approximate attempt to match total sleep time in the sleep period for the SWS vs. NSWS groups within each gender. In male subjects this attempt resulted in an ~10-min difference between the M-SWS and M-NSWS groups, and in the women there was an ~7-min difference between the groups in the opposite direction.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Results of the spectral (fast Fourier transform) analysis of the EEG for each of the extended sleep periods revealed that the frequency characteristics of the EEG recordings were strikingly similar between the male and female groups, except that power in the 8-12 Hz (alpha) band was higher in women than in men during phase 1 and slightly higher in the F-SWS group than in the M-SWS group during phase 2. This difference was no longer apparent during phases 3-5. Of particular importance was the finding that power in the 0.4- to 3-Hz (delta) band was almost identical in the male and female groups. Thus, in terms of EEG-frequency characteristics, the depth of sleep did not vary between the male and female groups. It should also be noted that correlational analyses showed that the amount of time spent within a phase did not correlate with either maximum level of UAR within the phase or the level of UAR reached by the end of the phase.
Figure 2 shows the 50 standardized values
for I and
Rafpeak plotted across the
extended sleep period as a function of both gender and group. Each
point on the graphs in Fig. 2 is a mean value representing a variable
number of breaths and a 10% progression through the relevant phase
(or, for phase 3 in the NSWS groups, an ~3.3% progression through that phase).
I (A and
C) and
Rafpeak
(B and
D) across extended sleep period as a
function of gender and group. To average data over subjects, each phase
was divided into 10 equal sections comprising equal numbers of breaths,
and a mean value for each section was obtained by averaging data for
all breaths within the section. For NSWS groups extended
phase 3 period was divided into 30 sections. Thus each point is a mean value representing a variable
number of breaths and a 10% progression though the relevant phase (or
for phase 3 in the NSWS groups an
~3.3% progression through that phase). See text definitions of
phases 1-5.
The effect of gender within the SWS groups is shown in Fig. 2,
A and
B, and was assessed in the two-way
trend analysis. The main effect of gender was not significant for
I
(F[1,16] = 0.68, P = 0.422). As can be seen in Fig.
2A, I in
both men and women decreased across the sleep period.
I was higher in the M-SWS than in the
F-SWS group during phase 1 but was
very similar in the two groups during phases
3-5. These results were reflected in a significant
main effect of position for I
(F[49,784] = 14.38, P < 0.001) and a significant
gender-by-position interaction (F[49,784] = 2.98, P < 0.001). Post hoc contrasts
(polynomial) showed that for I both the
linear (F[1,16] = 85.64, P = 0.001) and quadratic
(F[1,16] = 9.18, P = 0.008) trends were significant, thus describing the linear fall in I over
phases 1-3 and the subsequent
plateau during phases 4 and
5. The linear trend varied between men
and women, as illustrated in Fig. 2A
and as reflected in a significant gender-by-position interaction effect
for the linear trend
(F[1,16] = 13.79, P = 0.002).
In both the male and female SWS groups UAR (Fig. 2B) increased over phases 1-5, resulting in a significant main effect of position for Rafpeak (F[49,785] = 25.82, P < 0.001). In both sexes UAR was relatively stable during phases 1 and 2 and then began to increase during phase 3, with further increases occurring during phases 4 and 5. This pattern of change resulted in significant linear (F[1,16] = 56.66, P < 0.001) and quadratic (F[1,16] = 30.25, P < 0.001) trends for Rafpeak. UAR was similar in men and women during phases 1 and 2, but the increment over phases 3-5 was greater and more progressive in men, causing a significant main effect of gender for Rafpeak (F[1,16] = 10.78, P = 0.005) and a significant gender-by-position interaction for this variable (F[49,785] = 4.76, P = 0.001). For Rafpeak there were also significant gender by position interaction effects for the linear (F[1,16] = 9.55, P = 0.007) and quadratic (F[1,16] = 9.11, P = 0.008) trends.
The effect of group (SWS vs. NSWS) can be observed by comparing Fig. 2,
A and
B, with Fig. 2,
C and
D, and was assessed in the three-way
(gender-by-group-by-position) ANOVA. As can be seen, the size of the
fall in I across the sleep
period was smaller in the NSWS than in the SWS groups, particularly
within male subjects. This effect was reflected in a significant
group-by-position interaction for I
(F[49,1127] = 1.74, P = 0.001), although neither the main effect of group (F[1,23] = 0.28, P = 0.600) nor the
group-by-gender interaction
(F[1,23] = 0.03, P = 0.874) was significant. As
expected on the basis of the results of the two-way ANOVA (above), the main effect of gender also failed to reach significance for
I (F[1,23] = 1.55, P = 0.226).
In female subjects the pattern of change in UAR across the sleep period was similar in the SWS and NSWS groups (see Fig. 2, B vs. D), except that UAR was slightly higher during phase 5 in the F-SWS group than it was during the latter part of the extended phase 3 period in the F-NSWS group. In men, however, there was a marked difference between the SWS and NSWS groups, such that the increase in UAR was greater in the M-SWS group than in the M-NSWS group. This variation in the pattern of results caused a significant main effect of group for Rafpeak (F[1,23] = 8.18, P = 0.009) and a significant gender-by-group interaction for this variable (F[1,23] = 8.48, P = 0.008). Rafpeak was higher in the M-SWS group than in the other groups only later during the sleep period, as shown in Fig. 2 and as reflected in significant group-by-position (F[49,1127] = 6.46, P < 0.001) and three-way (F[49,1127] = 2.92 P < 0.001) interactions.
Mean values for phases 1-5 for
each subject in the male and female SWS groups are shown in Table
4. These results are presented to
illustrate the degree of within- and between-subject variability in
I and UAR across phases. As can be seen,
UAR tended to be not only higher but also more variable in men than in
women during phases 4 and
5, and two male subjects
(MA and
PQ) exhibited particularly high
values for Ravg during
phase 5. The highest mean phase values for a female were 17.5 and 42.3 cmH2O · l1 · s
for Rafpeak and
Ravg, respectively
(subject LG), both of which are
lower than the corresponding overall means for the M-SWS group.
Although not presented in this study, statistical analyses performed by
using these values elicited results generally consistent with those
obtained in the trend analyses.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Figure 3 shows mean
pressure-flow loops for phases
1-5 in the male and female SWS groups. As shown,
during phases 1 and
2 pressure-flow relationships were
almost identical in men and women, but during phase
3 men exhibited a slightly greater tendency to develop
flow limitation, and this tendency became more evident during
phases 4 and
5. During phases
4 and 5 men generated
almost twice as much pressure as women to achieve similar peak flow
rates as those observed in women. This effect was evident during both the inspiratory and expiratory phases of the respiratory cycle.
In summary, changes in I and UAR from
wakefulness (phase 1) to early stage
2 NREM sleep (phase 3) were similar
in men and female subjects. Once NREM sleep became established and
developed to SWS, however, the pattern of change in UAR varied between
men and women. In men there was a progressive increment in UAR over the
sleep period, whereas in women UAR during SWS remained at a level
similar to, or only slightly above, that observed in subjects (either
male or female) who did not attain SWS but spent a prolonged period of
time in stage 2 NREM sleep interrupted by arousals. Men exhibited a
greater tendency than women to develop flow limitation during sleep. A
fall in I occurred across the sleep
period in all groups but was greatest in male subjects who attained
SWS.
DISCUSSION
It has been shown previously that in male subjects sleep-induced
changes in UAR and ventilation are initiated at sleep onset in
association with the transition from predominant alpha to predominant theta EEG activity (9, 10) and that later, as NREM sleep becomes
established and "deepens," there is a progressive increment in
UAR (11, 21, 26). The study reported here is the first to document
these changes across the sleep period as a function of gender. The
results showed that during the sleep onset period changes in
I and UAR were similar in men and women.
In both genders alpha-to-theta EEG transitions were associated with a decrease in I and an increase in UAR, and
arousals (theta-to-alpha transitions) were associated with an increase
in I and a decrease in UAR. In both
genders the alterations in I and UAR that
occurred in association with state fluctuations during sleep onset were initially small but became larger once the first signs of stage 2 sleep
emerged. Up to this point the magnitude of changes in I and UAR were similar in men and women.
During phase 3 UAR tended to increase
progressively over consecutive theta breaths, and although this
occurred in both men and women there was a suggestion that the
increment was more substantial in men so that UAR peaked at higher
levels before phase 3 arousals
(theta-to-alpha transitions). The latter effect, however, was not
significant. Once arousals ceased to occur and NREM sleep became
established, a marked gender difference began to emerge, such that in
men UAR increased progressively as NREM "deepened," whereas in
women it reached a plateau at lower levels. Thus in both genders an
increase in UAR occurred at sleep onset but, as NREM sleep developed,
UAR increased to a greater extent in men, and men exhibited a greater
tendency than women to develop flow limitation.
In both men and women the initial fall in
I at sleep onset was relatively large,
such that it represented the major portion of the total change from
wakefulness to NREM sleep. Further reductions in
I during established NREM sleep were
minimal. The reduction in I from
wakefulness to sleep was a consequence of alterations in both tidal
volume and cycle duration, and although results for the latter two
variables have not been reported in this study, neither of these
effects alone was statistically significant. A larger decrease in
I in the M-SWS group than in
the F-SWS group was a consequence of higher ventilation during
wakefulness rather than lower ventilation during sleep because
I was maintained at similar levels during
SWS in the male and female SWS groups. Within the men, however, a
larger decrease in I in the M-SWS group
than in the M-NSWS group appeared to be a consequence of both
slightly higher I during wakefulness and
lower I during sleep in the
M-SWS group. The latter effect was most likely a result of the higher
resistive loads observed during sleep in the M-SWS group. The finding
that men were able to maintain I at
levels similar to those observed in women during SWS despite the more
marked increases in UAR is of interest because it suggests that some
form of compensation was operating to maintain
I. Previous studies have shown that
immediate, neuromechanically mediated compensatory responses are
blunted or absent during NREM sleep (7, 8, 27), and it is
therefore likely that the maintenance of
I observed during SWS in this study was a
consequence of some delayed, chemically mediated compensation
and/or input from upper airway pressure-sensitive or stretch
receptors.
The results of the analyses of the extended sleep periods in this study indicate that the development of delta EEG activity is associated with large increases in UAR in men but not in women. The data from male subjects indicated that it was the depth of NREM sleep, as reflected in EEG-frequency characteristics, rather than sleep time that determined the level of UAR. Men who spent a prolonged period of time in phase 3 (stage 2 sleep interrupted by intermittent arousals) did not exhibit the progressive increase in UAR observed in men who achieved SWS within the same period of time. In women, however, there was little difference in UAR between subjects who achieved SWS and those who did not so that data from both groups of women were similar to data from the men who did not achieve SWS. EEG-frequency characteristics were strikingly similar in the male and female groups. Thus the development of delta EEG activity was associated with large increases in UAR in men but not in women.
There was large individual variability in the level of UAR recorded
during sleep in this study, particularly among the male subjects, two
of whom showed particularly high values. The gender difference,
however, was not due solely to these subjects. Removal of the data from
two male subjects who exhibited the most marked degree of flow
limitation did not affect the results for
Rafpeak. As might be
expected, however, removal of these subjects' data did reduce the size
of the gender difference for resistance measurements incorporating
sampling points from the relatively alinear portion of the
pressure-flow curve (such as Ravg
and UAR at peak resistive pressure). In the F-SWS group
the highest 10% standardized value for
Rafpeak for a subject during
sustained NREM sleep was 25 cmH2O · l1 · s,
and this level was exceeded by five of the eight M-SWS subjects. Informal examination of the individual sleep recordings suggested that
UAR fluctuated more, or was more unstable in men than women, and
although only one male subject from the NSWS group reported occasional
snoring, men appeared more likely to experience flow limitation.
Examination of the pressure-flow loops in men and women also suggested
a greater male susceptibility to flow limitation. Although the
relatively small sample sizes in this study do not allow us to predict
with any confidence the distribution of UAR occurring during SWS, there
was a suggestion from the data in our subjects that the distribution is
more positively skewed in men.
It is generally assumed that changes in upper airway muscle activity are a major factor contributing to sleep-induced changes in UAR. It has been proposed that the loss of wakefulness is associated with a reduction in activity of the upper airway-dilating muscles and that this effect is more marked in the muscles exhibiting predominantly tonic activity compared with those showing phasic, respiratory-related activity (20). Two recent studies examining the immediate effect of sleep on upper airway muscle function in men have shown that at sleep onset (alpha-to-theta EEG transition) there is a reduction in the activity of both the tonic (for example, the tensor palatini) and the phasic (for example, the genioglossus) upper airway muscles (13, 29). However, during sustained NREM sleep genioglossus muscle activity in men is either maintained at waking levels or augmented (16, 20, 21, 28), whereas tensor palatini activity remains attenuated (16, 20, 21). One explanation for the greater increase in UAR in men than women during NREM sleep is that sleep affects the respiratory musculature differently in men and women. This could occur in two ways. First, it is possible that there is a more marked attenuation of tonic upper airway muscle activity during sleep in men than in women, particularly after sleep onset as sleep deepens, thus causing a greater increase in UAR. The second possibility is that the effect of sleep on tonic upper airway muscle activity is similar in men and women, but in women the initial increase in UAR is prevented from developing further because the phasic upper airway muscles are more effectively recruited than in men. The latter explanation would be consistent with the finding of Popovic and White (14) that during wakefulness peak phasic genioglossus EMG activity increased more in women than in men in response to experimentally applied inspiratory resistive loads. Whether this gender difference in resistive load compensation continues in the sleeping state remains to be determined.
With regard to activity of the respiratory pump, it is assumed that in this study male subjects were able to maintain ventilation during established NREM sleep at a level similar to that observed in women via recruitment of the inspiratory pump muscles. Consistent
