| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Physiology, Dartmouth Medical School, Lebanon, New Hampshire 03756
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
The quantification of respiratory variability may provide insight into the integrative control of breathing. To test the hypothesis that sleep and/or increased chemical drive modifies respiratory variability, six male adult Sprague-Dawley rats were instrumented with diaphragm electromyographic (EMG) electrodes and exposed to 0, 2.5, and 5.0% CO2 with a balance of room air during wakefulness and behaviorally determined sleep. Respiratory interval (Ttot), peak diaphragm EMG, and ventilation index (peak diaphragm EMG/Ttot) were calculated for 1,024 sequential breaths. The variability of breathing was quantified with a measurement of signal complexity, the approximate entropy, and two autocorrelation measurements, the autoregressive power spectrum slope and the detrended fluctuation analysis slope. Elevated chemical drive and/or sleep significantly modulated the variability of ventilation index and Ttot. There were also significant interactions between state and CO2 drive in all respiratory parameters. We conclude that state (sleep or wakefulness) and increased chemical drive affect respiratory variability differentially.
nonlinear analysis; control of breathing; sleep
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
THE CONTROL OF BREATHING IS an integrative process, such that the characteristics of any current breath are correlated with the characteristics of previous breaths, and these correlations include information from short- and long-term time scales (2). This property imparts a dynamic quality to the time-dependent variability of respiration, and thus the quantification of the variability of breathing may provide insight into the integrative nature of the control of breathing. Respiratory feedback originates from the neurons or networks that drive breathing and is modulated further by chemosensors, peripheral respiratory receptors, and nonrespiratory centers in the brain. Thus, even during a "steady-state" condition, the variability of respiration is encoded with information pertaining to previous breaths and to the integrative nature of breathing.
Bruce (4) showed that peripheral chemoreflex feedback modifies respiratory variability in anesthetized adult rats breathing either room air or 100% O2. Comparisons of breath-to-breath variability were made with classic [standard deviation (SD) and coefficient of variation (CV)] and nonlinear (spectral analysis) techniques. Spectral analyses allow for the dissolution of changes in variability on different temporal or correlation scales (i.e., low-frequency, long-term correlation; high-frequency, short-term correlation). Although no significant differences in the classic measurements of breath-to-breath variability were detected between rats breathing room air or 100% O2, a comparison of variability as measured by spectral analysis indicated a significant difference. These findings suggest that feedback from peripheral chemosensors modulates the temporal pattern of the variability of breathing and that standard statistical measurements may not detect the change in variability.
The integration of central drive and mechanoreceptor feedback imparts long-term correlations in the variability of end-expiratory lung volume (EEV). In the anesthetized rat, negative or positive transrespiratory pressure resulted in subsequent increases in the variability of the long-term correlations compared with randomly shuffled data, and sectioning of the vagi eliminated these responses (25). These data suggest that pulmonary stretch receptor feedback also contributes to the modulation of the temporal pattern of the variability of respiration.
A more recent study indicated that nonrespiratory cortical activity also modified the variability of breathing. Nonlinear measurements of the variability of breathing were increased significantly during a visuomotor reaction task compared with a control condition; however, the SD was unaffected (26). These data suggest that cortical motor activity modulates breathing, such that increased cortical motor activity results in an increase in nonlinear respiratory variability, without concomitant effects on classic measurements of variability.
In this investigation, respiratory variability was quantified with classic statistical measurements (i.e., SD and CV), a measure of complexity, and the relationship between short- and long-term autocorrelation. The objective of this study was to characterize the change in respiratory variability during increased chemical drive and/or during sleep and wakefulness. We hypothesized that respiration in the awake and unrestrained rat breathing room air would exhibit significantly different variability characteristics compared with randomly shuffled data, suggesting long-term correlation in the pattern of breathing. We also hypothesized that, compared with wakefulness with room air, increased inspired CO2 and/or sleep would result in a significant decrease in the variability of respiration.
| |
METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
These studies had the approval of the Institutional Animal Care and Use Committee at Dartmouth College and Dartmouth-Hitchcock Medical Center, Lebanon, NH.
Surgical preparation. Six male Sprague-Dawley rats (298.7 ± 71.09 g) were anesthetized with ketamine (60 mg/kg im) and xylazine (3.75 mg/kg ip) for surgical instrumentation under sterile conditions. Ketamine anesthesia was supplemented throughout the surgical procedure as necessary (30-60 mg/kg). Electromyographic (EMG) activity of the diaphragm muscle (Dia EMG) was obtained by using a technique described by Remmers et al. (21). Bipolar electrodes (multistranded stainless steel wire, 0.002 in. bare, 0.009 in. Teflon coated; A-M Systems) were stitched into the costal diaphragm ~5 mm apart. A grounding electrode (Plastics One, E363/76) was stitched onto the chest wall musculature. Wires were tunneled subcutaneously, exteriorized in the shoulder blade area, and connected to a six-pin electrode socket pedestal (Plastics One, MS363).
After completion of the surgical instrumentation, animals were injected with an analgesic to minimize discomfort (Buprenex, 0.05 mg/kg sc) and returned to the vivarium, where they were kept on a 12:12-h dark-light cycle. A surgical recovery period of 4 days was allowed before the experimental protocol. Animals were treated with an antibiotic (amikacin sulfate, 2.5 mg/kg im) daily, for 7 days, to minimize the risk of infection.Measurements. The experimental setup consisted of a Plexiglas chamber in which the rats were placed. The rat's electrode socket pedestal was connected via a wire-covered cable to a swivel commutator fixed to the center of the chamber's lid. The EMG signal was band-pass filtered (30-10,000 Hz), amplified, and moving-time averaged (time constant, 25 ms). EMG was recorded for periods of 10-20 min while the rats were awake (upright or lying down, head up, eyes open) or asleep (lying down, head down, motionless, eyes closed), while exposed to the following inspired gas mixtures: room air, 2.5% CO2 in room air, and 5% CO2 in room air. The inspired gas mixtures were continuously flushed through the chamber's inlet and outlet valves at a rate of 2 l/min. The CO2 level was continuously monitored at the outlet valve. Each rat was studied on 4-5 days for up to 2 wk and was then euthanized.
Protocol. On each study day, the rat was initially allowed to acclimate to the chamber for ~30 min. The rat was then exposed to each of the three inspired gas mixtures during both wakefulness and behaviorally determined sleep, for periods of 10-20 min. Sleep was defined as the animal lying down, motionless, and with eyes closed. The time was chosen to collect ~1,100 breaths during each trial; analyses were performed by using 1,024 sequential breaths. The order of the six trials was randomized. After a change of inspired gas mixtures, at least 10 min were allowed between trials. If the rat awoke during a sleeping trial or became very active during a waking trial, the trial was repeated. After completion of six satisfactory trials, the rat was disconnected from the swivel commutator, placed in its cage, and returned to the vivarium. The complete protocol was repeated four times, on 4 different days, for each rat.
Data analysis. Moving-time-averaged Dia EMG was recorded on videotape (Vetter 400A) and on a microcomputer sampled with data-acquisition hardware and software (Windaq, DATAQ) at 10,000 Hz. Respiratory cycle duration (Ttot), peak moving-time-averaged Dia EMG amplitude above tonic levels for each respiratory cycle (Peak EMG), and the product of Peak EMG and instantaneous breathing frequency [ventilation index (VI)] were calculated for 1,024 sequential breaths from each trial of each study day. The variability of these parameters was assessed by calculation of classic measures of variability (SD and CV) from each trial of each study day. These measures of variability per trial were averaged for multiple study dates in each rat. Finally, an averaged response for each measure of variability per trial from all rats studied was computed, weighting the rats equally.
To further characterize the breath-to-breath variability, the signal's complexity [approximate entropy (ApEn)] and two scaling indexes were calculated for Ttot, Peak EMG, and VI data. A scaling index is a measurement that quantifies the relationship between short- and long-term correlation in a signal. The value of the scaling index indicates whether the signal has white noiselike behavior, fractal-like behavior, or simple long-term correlations. A signal that is fractal-like exhibits self-similar behavior, such that the signal's variability appears unchanged when viewed at different time scales. Two previously validated methods to determine the scaling index were chosen: the negative slope of the autoregressive spectrum (
)
(17) and the slope of the detrended fluctuation analysis
(DFA) for short- (DFAs1) and long-term correlations (DFAs2) (9,
10, 16). These measures of variability were computed for each
trial during each day of the study, and then each measurement was
averaged for multiple study dates in each rat. Finally, an averaged
response for each measure of variability from all rats was computed,
weighting the rats equally.
ApEn.
Pincus (19) first described the calculation of ApEn
(m, r, N). ApEn has since been
used to quantify the complexity of physiological signals
(20). Whereas the SD quantifies the variability of a set
of individual data points to the mean value, the ApEn is a measure of
the likelihood that all runs of data with a specific length
(m) and variability (r, a percentage of the SD)
are similar to all other runs of data of the same length. As described
by Pincus and Goldberger (20), before the calculation of
ApEn, two input parameters, m (the number of data points in
a run to be compared) and r (a filtering coefficient) must
be fixed, and the value N {the total length of the data
[u(i)]} must be selected. To calculate ApEn,
the vector sequences x(1) through
x(N 
m + 1) are formed,
where x(i) = [u(i),..., u(i + m 1)]. These vectors represent m
consecutive u values, commencing with the ith
point. The maximum distance (d) between the scalar
components of the vectors x(i) and
x(j) is defined as
d[x(i),x(j)]. The sequence x(1),...,
x(N m + 1) is used to
construct, for each i 
N m
+ 1, C
r}/(N m + 1). The values
of C
m(r) is defined
as the natural logarithm of the average of
Cm(r) m+1(r). The ApEn
measures the logarithmic likelihood that a run of data with a specific
variability will be similar to another run of data with the same
length. A greater likelihood of remaining close (regularity) produces a
smaller ApEn value (20). Previous implementation of this
test with clinical data using N = 1,000 suggests that,
for m = 2, values of r of 0.15 and 0.20 produce valid statistical measurements of ApEn (19). For
our study, we chose m = 2, r = 0.15, and N = 1,024.
Slope of the autoregressive power spectrum.
The of the linear regression fit to the log-log plot of the
autoregressive power spectrum (AR) provides a measurement of the
scaling behavior of a time series signal. A equal to 0 suggests that the signal consists primarily of white noise, indicating a lack of
short- or long-term correlation. A equal to 1 suggests fractal-like
behavior, and a equal to 2 indicates that the data are dominated by
long-term correlations. The AR differs from the classic fast-Fourier
transform approach in that it provides a best fit model that is used
for the construction of the power spectrum. This enhances the ability
of the AR approach to correctly model the frequency characteristics of
short data segments (17).
DFA.
DFA quantifies the relative amount of correlation in time series data,
while being reasonably insensitive to the effect of noise in a signal
(16). Briefly, the time series data are integrated to form
an accumulated sum. The integrated time series is then divided into
sets of equal length (n). For each set of data, a least
squares line is fit (representing the local trend). The integrated time
series is "detrended" by subtracting the local trend of each data
set. The root-mean-square fluctuations [F(n)] of the
integrated and detrended data are calculated, and this process is
repeated over all time scales (n, data set lengths) to
provide a relationship F(n) and data set length
(n, i.e., number of breath-to-breath intervals). A total
slope (DFA) can be extracted from the plot of log F(n) vs.
log n. A DFA 
0.5 indicates that the signal (i.e.,
breath-to-breath variability) is random or white noiselike, DFA 1.0 indicates fractal-like scaling, and DFA 1.5 suggests a simple
long-term correlation.
5 (where n is the number of points used
to construct the slope) was constructed with each sum of squared error
corresponding to the location of the center point of each five-point
line segment. The point within the vector at which the error
calculation was the greatest was defined as the breakpoint or the point
at which two different slopes would diverge. The line-fitting algorithm
was designed to fit linear slopes to short-term correlations and to
long-term correlations. Breakpoints were calculated for every sequence,
and the DFA calculations were forced to use the average breakpoint,
which allowed for consistency in the model. The technique separates
short-term correlations (by using a slope for breaths up to 25 s
apart, DFAs1) from long-term correlations (by using another slope for
breaths 25-140 s apart, DFAs2). This technique allows for the
interpretation of data that may behave like white noise for the short
term but not for the long term and vice versa.
Surrogate data testing.
Surrogate data were constructed from the original data to determine
whether the variability of the original time series data was
significantly different from a random process with no long-term correlation (i.e., white noise). From each study during wakefulness with room air as the inspired gas, a surrogate time series was constructed by randomly shuffling the original data (VI, Ttot, and Peak
EMG). The surrogate data for each time series had the same mean, SD,
and CV as the original data (an example of surrogate data testing is
presented in Fig. 1).
|
Statistics. The same statistical analyses were performed on each time series (Ttot, Peak EMG, and VI). Comparisons of surrogate and original data were made with a repeated-measures analysis. To determine whether increased CO2 or state (wakefulness vs. sleep) affected the classic measurements (mean, SD, or CV) or the nonlinear measurements (ApEn, ARs, DFAs1, or DFAs2), a repeated-measures analysis was performed. The Bonferroni adjustment was used to compensate for multiple comparisons. Differences were considered significant when P < 0.050.
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
During each day of the experiment, we attempted to record one run of data from each CO2 level and wakefulness state. Because not all rats were able to sleep for a sufficient amount of time during each day of experiments, it was not possible to collect 1,024 sequential breaths from certain trials. We were unable to collect one trial at 5.0% CO2 from one rat, one trial in room air and two trials at 5.0% CO2 from another rat, and one trial in room air from a third rat. All other data from all other trials from these three rats and all data from all trials in all other rats were collected successfully. To balance the study and have four trials (of each CO2 level and state) for each animal, additional experiments were conducted to record the missing trials.
Increased inspired CO2 resulted in a significant increase
in mean VI (P < 0.001) and a significant decrease in
mean Ttot (P < 0.001). Multiple-comparison analysis
indicated that each level of increased inspired CO2
significantly changed (P < 0.013) VI and Ttot.
Relative to wakefulness, behaviorally determined sleep resulted in a
significant decrease in mean VI (P = 0.038) and a
significant increase in mean Ttot (P = 0.014). For each
rat, Peak EMG was normalized to the value during wakefulness with room air; thus it was not possible to analyze these data statistically (see
Table 1).
|
Surrogate data.
The randomization of the order of the VI, Ttot, or Peak EMG data did
not change the values for the mean, SD, or CV. Randomization of the
order of the data resulted in a significant increase in the ApEn
(P < 0.050), a significant decrease (from a negative value toward zero) in the (P < 0.050), and
significant reductions in DFAs1 and DFAs2 (P < 0.050)
of the VI, Ttot, and Peak EMG data (Fig.
2).
|
SD and the CV.
Increased inspired CO2 had no significant effect on the SD
of VI or Peak EMG but resulted in a significant decrease in the CV of
both variables (P = 0.011 and P = 0.001, respectively). Multiple-comparison analysis indicated a
significant difference between the CV of VI and Peak EMG at 2.5 and
5.0% CO2 (P < 0.017 and P < 0.027), respectively. The SD and CV of Ttot were reduced significantly with increased inspired CO2
(P < 0.001 and P = 0.002, respectively). Multiple-comparison analysis indicated a significant difference between the SD of Ttot at 0.0 and 5.0 and at 2.5 and 5.0%
CO2 (P < 0.012) and for the CV at 2.5 and
5.0% CO2 (P < 0.002). Increased inspired
CO2 resulted in no statistically significant changes in the
SD of Peak EMG (see Fig. 3).
|
Complexity and autocorrelation.
Increased inspired CO2 resulted in a significant increase
in the ApEn of VI (P = 0.016; see Fig.
4) and a significant change in Ttot
(P = 0.001). Multiple-comparison analysis indicated a significant difference between the ApEn at 2.5 and 5.0%
CO2 (P < 0.012) of VI, and between 0.0 and
2.5, and 2.5 and 5.0% CO2 of the Ttot data
(P < 0.028). These data suggest that, with increased inspired CO2, breathing, as measured by VI, became more
variable. There was also a significant interaction between the
responses to increased inspired CO2 and sleep with the ApEn
of Ttot (P < 0.047).
|
|
|
|
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
The SD and CV quantify how well the output of the respiratory controller sustains a mean value; however, the measurement of complexity (ApEn) quantifies the difference between breaths, and the autocorrelation measurements (ARs, DFAs1, and DFAs2) quantify the relationship between short- and long-term correlations. Therefore, the ApEn and the relationship between the short- and long-term correlation provide techniques to quantify the level of integration of respiratory inputs and the variability of breathing. As expected, random shuffling of the order of the respiratory data resulted in an increase in ApEn and ARs and a decrease in DFAs1 and DFAs2. These results suggest that, after randomization, the variability of the data became more like white noise, such that each breath was independent from all other breaths in the sequence. Thus we conclude that there are short- and long-term correlations in the variability of breathing, inasmuch that the inherent variability of these data is significantly different from white noise. Although increased inspired CO2 resulted in a decrease in the CV of all of the respiratory parameters, the ApEn of VI increased, suggesting increased breath-to-breath variability. There was a significant effect of increased inspired CO2 on the ApEn of Ttot and a significant reduction in DFAs1. Sleep, compared with wakefulness, resulted in a small but significant reduction in DFAs2 of VI, indicating a possible increase in the white noiselike behavior of long-term correlation. Autocorrelation and ApEn measurements of variability of Ttot were significantly reduced during sleep, compared with wakefulness. There were significant interactions between the response to increased CO2 and sleep in the ApEn and autocorrelation measurements of VI, the ApEn of Ttot, and the ARs and DFAs1 of Peak EMG. The interactions of the ApEn data suggest that increased variability as a result of increased inspired CO2 was more robust during sleep than wakefulness. The interactions with the ARs, DFAs1, and DFAs2 suggest that, during wakefulness, increased inspired CO2 results in the variability of VI becoming more correlated. During sleep, however, the effect is opposite: VI became more uncorrelated, or more like white noise.
The inherent variability measured in breath-to-breath characteristics may arise from the effect of primary regulators, modulation by secondary systems, and/or by the interactions of all of the controlling systems. The variability ranges from the regularity of periodic breathing, through nonexistence of correlation or white noiselike behavior (3). The data from our investigation support these findings inasmuch that we quantified the changes in nonlinear variability, which resulted from increased inspired CO2 and/or sleep.
How do the observed changes in variability relate to respiratory stability? An output similar to white noise suggests that the respiratory controller is not relying on any past information to produce the current breath, thus leaving the system vulnerable to produce a relatively large response to a transient perturbation. On the other hand, a system relying heavily on feedback and past influences may disregard an important transient perturbation. Normal respiratory control operates at a midpoint between white noiselike behavior and periodic breathing (3).
The variability of respiration is modulated by changes in EEV (24). Measurements from rats breathing through a tracheotomy and housed in a head-out plethysmograph showed that the fractal scaling of EEV was modulated by positive or negative transrespiratory pressures, and a measure of the long-term correlation of the EEV was significantly different from those obtained after randomly shuffling the original data. Our results agree with these data in that respiration in the awake, unrestrained rat has significant short- and long-term correlations, which may indicate the presence of fractal scaling in the variability of breathing.
Nattie (14, 15) proposed that there are multiple sites of central chemoreception throughout the brain stem. Different chemosensors may be more or less active during different levels of CO2 drive or sleep and wakefulness states. In our studies, the ApEn of VI increased with elevated inspired CO2, suggesting that respiratory output became more dynamic. The increase in variability may result from additional chemoreceptor sites "turning on" as the level of CO2 in the blood increased, thus driving the respiratory centers with a greater number of inputs and resulting in a greater dynamic behavior in the output. During sleep, compared with wakefulness, the ApEn of Ttot was reduced. This change may be reflective of the reduction in inputs (i.e., cortical inputs) that the respiratory control center receives during sleep. Inasmuch as the ApEn increased during the elevated CO2 as a result of a recruitment of additional chemosensors, a reduction in feedback and or inputs to the respiratory center would decrease the variability.
Respiratory output is the result of an integrative process, and the characteristics of any current breath are related to the characteristics of previous breaths (2). According to test data presented in Fig. 1, a decrease in ARs or an increase in DFAs1 or DFAs2 could be the result of an increase in the integration (or memory) within the central respiratory controller. Thus an increase in ARs or a decrease in DFAs1 or DFAs2 may result from a decrease in the central integration of the respiratory output. After the removal of an excitatory stimulus of respiration, an increase in the correlation between breaths occurs for a short duration and is classified as short-term potentiation or memory (5). Hyperoxic termination of short applications of isocapnic hypoxia during non-rapid eye movement (NREM) resulted in short-term potentiation of respiration (1). Although a decrease in posthyperoxic termination of hypoxic ventilation during NREM sleep has been reported (8), CO2 was not monitored and was inevitably reduced in these subjects. This reduction in CO2 could be the cause for the discrepancy in the data from NREM sleep. Sleep also removes a tonic stimulus to respiration (6), presently termed the wakefulness stimuli. During increased inspired CO2, the DFAs1 of the Ttot was reduced. These data suggest decreased correlation in Ttot during the short term or a decrease in the memory or integration of Ttot. During sleep relative to wakefulness, DFAs2 of VI and DFAs1 and DFAs2 of Ttot decreased, and ARs of Ttot increased, all suggesting a decrease in the memory or integration of respiration.
An increase in the gain of the chemosensors or an increase in the CO2 within the blood would impart greater chemical feedback on breathing and result in a more correlated breathing pattern (12). Compared with wakefulness, the ventilatory response to increased inspired CO2 is depressed during NREM sleep and further during rapid eye movement (REM) sleep (22). According to these studies, during sleep, when the chemosensors have decreased gain, there should be decreased variability compared with wakefulness, although transient REM bursts can result in the fractionation of the diaphragm output, resulting in an apparent reduction in CO2 gain. Our data partially agree with this statement, in that, during wakefulness, increased CO2 results in more correlated variability in VI and Peak EMG; however, during sleep, our data disagree, in that increasing CO2 resulted in the variability of VI and Peak EMG becoming more uncorrelated. The discrepancy may be the result of a decrease in the integration or memory of the central respiratory controller during sleep.
Technical limitations. The sleep or wakefulness state of the rat during each protocol was determined by using a behavioral analysis; sleep was defined when the rats were lying down, head down, and motionless with eyes closed. This technique of state identification was used for two reasons: it was necessary to record data from time periods greater than the amount of time the animal would spend in one continuous sleep state (NREM or REM), and we wanted to limit the invasiveness of the experimental design. It is also noted that the rats frequently slept for time periods that yielded many more breaths than necessary. When this occurred, the last 1,024 breaths before the arousal of the animal were used to limit the amount of breaths taken from the onset of behaviorally determined sleep, which may contain data from a drowsy state and/or wakefulness-to-sleep transition discontinuities.
Through the use of clinical and theoretical data, the ApEn calculation produced statistically valid results with data sets of 1,000 points (20), and, therefore, our data sets contained >1,000 points. The DFA and ARs routines also produced consistent and accurate results with data sets >1,000 points (unpublished observations). The average resting respiratory rate of an adult rat is2 Hz, and thus
is was necessary to record breathing for >10 min to ensure the
acquisition of enough data. Human and feline data suggest that the
variability of breathing (as measured by the CV) during REM sleep is
greater than during NREM sleep (13, 21). The NREM-REM
sleep cycle in adult rats has an average length of ~13 min
(23), and, during the cycle, almost 80% of the time is
spent in NREM sleep (7). Because it was necessary for the analyses used in this study to include both REM and NREM sleep, to
account for any changes in the proportion of NREM and REM sleep during
each sleep episode, each rat was tested on 4 separate days, and the
results reflect the average response.
Increased inspired CO2 alters the sleeping pattern in adult
rats. Exposure to 6.0-8.0% CO2 resulted in increased
sleep onset latency and modified the pattern of REM sleep, such that
there were fewer, but longer, REM episodes; however, the percentage of
total sleep time spent in REM was not altered (11). Based on those findings, during our experiments with 0.0, 2.5, and 5.0% inspired CO2, we do not expect differences in the
percentages of NREM and REM recorded during the visually defined sleep periods.
In conclusion, random shuffling of the time series data resulted in an
increase in the ApEn and a decrease in the measurements of ARs, DFAs1,
and DFAs2. These data support the hypothesis that there are long-term
correlations of breathing in awake animals breathing room air. Elevated
chemical drive and/or sleep significantly modulated the variability of
a measure of ventilatory output and Ttot. Furthermore, there were
significant interactions between state (sleep vs. wakefulness) and
CO2 drive in the variability measurements of all
respiratory parameters. During wakefulness, increased inspired
CO2 resulted in an increase in the long-term correlation of
breathing. However, during sleep, the effect was the opposite:
breathing became more uncorrelated or more similar to white noise.
These data suggest that, during sleep, as inspired CO2 is
increased, there is a decrease in the integration of breathing, such
that the respiratory system is relying less on information from
previous breaths to produce the current breath. Although these data
disagree with our hypothesis stating that respiratory variability would
increase with increased CO2 drive during sleep, the results
indicate a potential weakness in the integration of breathing during
sleep. If the increase in variability is manifested by a reduction in
central integration of breathing during sleep, this phenomenon may
decrease the stability of breathing by allowing the respiratory centers
to generate a relatively large response to a transient perturbation.
| |
ACKNOWLEDGEMENTS |
|---|
We express our gratitude to Dr. Donald Bartlett, Jr. for providing us with the opportunity to pursue our own research interests, which evolved into the development of this study. We are grateful for his insightful comments and review of this manuscript, as well as for his continued mentorship.
| |
FOOTNOTES |
|---|
This work was supported by National Heart, Lung, and Blood Institute Grant HL-07449.
Address for reprint requests and other correspondence: B. F. BuSha, Dept. of Physiology, One Medical Center Drive, Dartmouth Medical School, Lebanon, NH 03756 (E-mail: brett.bu.sha{at}dartmouth.edu).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
April 19, 2002;10.1152/japplphysiol.00951.2001
Received 14 September 2001; accepted in final form 16 April 2002.
| |
REFERENCES |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
1.
Badr, MS,
Skatrud JB,
and
Dempsey JA.
Determinants of poststimulus potentiation in humans during NREM sleep.
J Appl Physiol
73:
1958-1971,
1992.
2.
Bruce, EN,
and
Daubenspeck JA.
Mechanisms and analysis of ventilatory stability.
In: Regulation of Breathing (2nd ed.), edited by Dempsey JA,
and Pack AI.. New York: Dekker, 1995, p. 285-314.
3.
Bruce, EN.
Temporal variations in the pattern of breathing.
J Appl Physiol
80:
1079-1087,
1996.
4.
Bruce, EN.
Chemoreflex and vagal afferent mechanisms enhance breath to breath variability of breathing.
Respir Physiol
110:
237-244,
1997.
5.
Eldridge, FL.
Central neural respiratory stimulatory effect of active respiration.
J Appl Physiol
37:
723-735,
1974.
6.
Fink, BR.
Influence of cerebral activity in wakefulness on regulation of breathing.
J Appl Physiol
16:
15-20,
1961.
7.
Frank, MG,
and
Heller HC.
Development of REM and slow wave sleep in the rat.
Am J Physiol Regul Integr Comp Physiol
272:
R1792-R1799,
1997.
8.
Gleeson, K,
and
Sweer LW.
Ventilatory pattern after hypoxic stimulation during wakefulness and NREM sleep.
J Appl Physiol
75:
397-404,
1993.
9.
Hausdorff, JM,
Peng CK,
Ladin Z,
Wei JY,
and
Goldberger AL.
Is walking a random walk? Evidence for long-range correlations in stride interval of human gait.
J Appl Physiol
78:
349-358,
1995.
10.
Hausdorff, JM,
Purdon PL,
Peng CK,
Ladin Z,
Wei JY,
and
Goldberger AL.
Fractal dynamics of human gait: stability of long-range correlations in stride interval fluctuations.
J Appl Physiol
80:
1448-1457,
1996.
11.
Ioffe, S,
Jansen AH,
and
Chernick V.
Hypercapnia alters sleep state pattern.
Sleep
7:
219-222,
1984.
12.
Khoo, CK.
Determinants of ventilatory instability and variability.
Respir Physiol
122:
167-182,
2000.
13.
Mador, MJ,
and
Tobin MJ.
Effect of alterations in mental activity on the breathing pattern in healthy subjects.
Am Rev Respir Dis
144:
481-487,
1991.
14.
Nattie, EE.
CO2, brainstem chemoreceptors and breathing.
Prog Neurobiol
59:
299-331,
1999.
15.
Nattie, EE.
Multiple sites for central chemoreception: their roles in response sensitivity and in sleep and wakefulness.
Respir Physiol
111:
223-235,
2000.
16.
Peng, CK,
Havlin S,
Stanley HE,
and
Goldberger AL.
Quantification of scaling exponents and crossover phenomena in nonstationary heartbeat time series.
Chaos
6:
82-87,
1995.
17.
Persson, PB.
Spectrum analysis of cardiovascular time series.
Am J Physiol Regul Integr Comp Physiol
273:
R1201-R1210,
1997.
19.
Pincus, SM.
Approximate entropy as a measure of system complexity.
Proc Natl Acad Sci USA
88:
2297-2301,
1991.
20.
Pincus, SM,
and
Goldberger AL.
Physiological time-series analysis: what does regularity quantify?
Am J Physiol Heart Circ Physiol
266:
H1643-H1656,
1994.
21.
Remmers, JE,
Bartlett D, Jr,
and
Putnam MD.
Changes in the respiratory cycle associated with sleep.
Respir Physiol
28:
227-238,
1976.
22.
Sullivan, CE,
Murphy E,
Kozar LF,
and
Phillipson EA.
Ventilatory responses to CO2 and lung inflation in tonic versus phasic REM sleep.
J Appl Physiol
47:
1305-1310,
1979.
23.
Trachsel, L,
Tobler I,
Achermann P,
and
Borbely AA.
Sleep continuity and the REM-nonREM cycle in the rat under baseline conditions and after sleep deprivation.
Physiol Behav
49:
575-580,
1991.
24.
Zhang, X,
and
Bruce EN.
Response of breathing pattern to flow and pressure in the upper airway of rats.
Respir Physiol
113:
191-200,
1998.
25.
Zhang, X,
and
Bruce EN.
Fractal characteristics of end-expiratory lung volume in anesthetized rats.
Ann Biomed Eng
28:
94-101,
2000.
26.
Zhang, T,
and
Turner D.
A visuomotor reaction time task increases the irregularity and complexity of inspiratory airflow pattern in man.
Neurosci Lett
297:
41-44,
2001.
This article has been cited by other articles:
|
|
 |
|
  G. D. Mitsis, R. J. M. Governo, R. Rogers, and K. T. S. Pattinson The effect of remifentanil on respiratory variability, evaluated with dynamic modeling J Appl Physiol, April 1, 2009; 106(4): 1038 - 1049. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M.-N. Fiamma, C. Straus, S. Thibault, M. Wysocki, P. Baconnier, and T. Similowski Effects of hypercapnia and hypocapnia on ventilatory variability and the chaotic dynamics of ventilatory flow in humans Am J Physiol Regulatory Integrative Comp Physiol, May 1, 2007; 292(5): R1985 - R1993. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Li and E. Nattie Catecholamine neurones in rats modulate sleep, breathing, central chemoreception and breathing variability J. Physiol., January 15, 2006; 570(2): 385 - 396. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. R. Price, F. Han, T. E. Dick, and K. P. Strohl 7-Nitroindazole and posthypoxic ventilatory behavior in the A/J and C57BL/6J mouse strains J Appl Physiol, September 1, 2003; 95(3): 1097 - 1104. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
, Combined awake and sleep data; 
, data
recorded only during wakefulness; 
, data recorded only
during behaviorally defined sleep. * Significant effect with a
repeated-measures analysis of variance, P < 0.050. # Significant difference with a multiple-comparison analysis,
P < 0.050.
