Entrainment of Blood Pressure and Heart Rate ... - ATS Journals

Download PDF
4 downloads 112 Views 102KB Size Report
Comment
in blood pressure (BP) and heart rate (HR), which have been attributed to ... The magnitudes of systolic BP oscillations increased significantly from RB (14 5 mm.
Entrainment of Blood Pressure and Heart Rate Oscillations by Periodic Breathing GERALDO LORENZI-FILHO, HILMI R. DAJANI, RICHARD S. T. LEUNG, JOHN S. FLORAS, and T. DOUGLAS BRADLEY Sleep Research Laboratory, Mount Sinai Hospital; and Department of Medicine, The Toronto Hospital, University of Toronto, Toronto, Canada

Cheyne–Stokes respiration (CSR) is a form of periodic breathing associated with periodic oscillations in blood pressure (BP) and heart rate (HR), which have been attributed to hypoxia and arousals from sleep. We hypothesized that periodic alterations in ventilation alone would promote oscillations in BP and HR. Seven healthy, wakeful subjects breathed in three patterns, as follows: (1) regular breathing (RB); (2) periodic breathing with three (PB3: cycle frequency 5 0.035 Hz) augmented breaths alternating with 20-s apneas; and (3) periodic breathing with five (PB5: cycle frequency 5 0.030 Hz) augmented breaths alternating with 20-s apneas. SaO2 remained above 95% throughout. During periodic breathing, peaks in BP and HR occurred during the ventilatory period and troughs occurred during apnea. The magnitudes of systolic BP oscillations increased significantly from RB (14 6 5 mm Hg) to PB3 (20 6 4 mm Hg) and PB5 (25 6 7 mm Hg; p , 0.005). HR oscillations also increased from regular breathing (13 6 6.0 beats/min) to PB3 (20.2 6 2.3 beats/min) and PB5 (20.2 6 4.7 beats/ min; p , 0.01). Spectral analysis showed that during periodic breathing there were discrete peaks in the spectral power of ventilation, BP, and R-wave-to-R-wave interval at the periodic breathing cycle frequencies. We conclude that oscillations in ventilation occurring during periodic breathing can amplify and entrain oscillations in BP and HR in the absence of hypoxia or arousals from sleep. LorenziFilho G, Dajani HR, Leung RST, Floras JS, Bradley TD. Entrainment of blood pressure and heart rate oscillations by periodic breathing. AM J RESPIR CRIT CARE MED 1999;159:1147–1154.

Cheyne–Stokes respiration with central apneas (CSR) is a form of periodic breathing frequently observed in patients with congestive heart failure (CHF) (1, 2), in whom it is associated with increased risk of mortality (3, 4). One possible mechanism linking CSR with poor prognosis in patients with CHF is the increase in cardiac metabolic demands caused by periodic surges in blood pressure (BP) and heart rate (HR) (5, 6). These surges are similar to those observed in obstructive sleep apnea (OSA) (7, 8), in that peak increases in BP and HR occur during the hyperpnea that follows apnea. Several mechanisms have been proposed to account for these cyclic oscillations in BP and HR. These include chemostimulation by apnea-related hypoxia and CO2 retention (9, 10), and arousals from sleep (11), all of which can activate the sympathetic nervous system and increase BP and HR. However, these previ-

(Received in original form June 15, 1998 and in revised form September 21, 1998) Supported by operating grant MT 11607 from the Medical Research Council of Canada. Dr. Lorenzi-Filho was supported by research fellowships from the Conselho Nacional Pesquisa of Brazil and the Department of Medicine, University of Toronto. Dr. Floras is a Career Investigator of the Heart and Stroke Foundation of Ontario. Dr. Bradley is a Career Scientist of the Ontario Ministry of Health. Correspondence and requests for reprints should be addressed to T. Douglas Bradley, M.D., ES 12-421 Toronto Hospital (General Division), 200 Elizabeth St., Toronto, ON, M5G 2C4 Canada. E-mail: [email protected] Am J Respir Crit Care Med Vol 159. pp 1147–1154, 1999 Internet address: www.atsjournals.org

ously proposed pathways cannot fully account for cyclic increases in BP and HR observed during CSR. For example, the recent observation that alleviation of hypoxia does not affect oscillations of BP and HR during CSR (6) indicates that factors other than hypoxia play a role in their genesis. Furthermore, in contrast to OSA, CSR is not restricted to sleep, but can also occur during wakefulness. In this state, periodic increases in BP and HR are present (2, 5, 6, 12), but obviously cannot be due to arousals from sleep. Another possible explanation for cyclic increases in BP and HR, which has received little attention, is the resumption of ventilation at the termination of apneas. During normal tidal breathing, there are breath-to-breath, respiratory-related fluctuations in BP and HR, known as respiratory sinus arrhythmia (13). We have been struck by the observation that when CSR is present during wakefulness, there are also periodic increases in BP and HR closely linked to periodic increases in ventilation (5). We therefore hypothesized that: (1) periodic breathing with central apneas could provoke cyclic, respiratory-related oscillations in BP and HR, and entrain them at the periodic breathing frequency; and (2) the magnitudes of oscillations in BP and HR would be related to the magnitudes of increases in ventilation between apneas. To test these hypotheses, we examined the effects on BP and HR of periodic breathing at two different levels of hyperventilation between central apneas in healthy, wakeful subjects, in the absence of hypoxia and arousals from sleep. Spectral analysis was used to examine potential relationships between ventilation, BP, and HR during episodes of regular and periodic breathing.

1148

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

METHODS Subjects Seven healthy subjects (five males and two females), aged 27 to 37 yr and recruited from laboratory personnel and by advertisement, participated in the protocol. All subjects were normotensive (supine blood pressure , 140/90 mm Hg), were nonsmokers, were taking no medications, and had no history of respiratory or cardiovascular disease. The study protocol was approved by the local institutional ethics committee, and all subjects provided written informed consent prior to their participation in the study.

Experimental Setup Respiration was measured through respiratory inductance plethysmography (Respitrace; Ambulatory Monitoring Inc., White Plains, NY). Tidal volume (VT) was taken as the electrical sum of the rib cage and abdominal displacements, and was calibrated against spirometric values through the two-positions–simultaneous equations method (14). The VT signal was displayed on-line on a computer screen. Oxyhemoglobin saturation (SaO2) was continuously measured with a finger oximeter (Oxyshuttle; Sensormedics Corp., Anaheim, CA). Expired air from three subjects, on whom part of the protocol was repeated while they breathed a CO2-enriched gas (see the subsequent discussion), was sampled from cannulae inside the nares, from which the fraction of end-tidal CO2 (FETCO2) was measured with an infrared CO2 analyzer (Model LB-2; Beckman, Schiller Park, IL). Finger-arterial BP was measured on a beat-by-beat basis through digital photoplethysmography (Finapres 2300; Ohmeda, Englewood, CO). These measurements reliably reflect changes in intraarterial pressure during a variety of physiologic interventions (15), including vigorous respiratory maneuvers (16), and have been validated for

VOL 159

1999

power-spectral analysis of variability in arterial BP (17). HR was monitored continuously via an electrocardiogram recorded from a precordial lead.

Experimental Protocol Subjects were studied in a quiet room, while in the supine position and awake. Initially, subjects breathed spontaneously for 7 min, during which their mean VT was determined. The gain of the computer screen output was then adjusted so that two horizontal lines, 1 cm apart, represented each subject’s mean VT. Horizontal target lines were also drawn at 3, 5, and 7 cm above the reference line such that the target lines represented three, five, and seven times the baseline VT, respectively. The subjects then breathed in three patterns, lasting 7 min each, and separated by 15-min periods of spontaneous breathing, while visually monitoring their VT outputs on a computer screen as illustrated in Figure 1. The first breathing pattern was one of regular breathing (RB), during which the subjects were instructed to maintain, by visually monitoring their VT on the computer screen, a constant VT within the two horizontal lines separated by 1 cm as described earlier. This was followed, in random order, by two different periodic breathing patterns. Periodic breathing patterns were designed with the following considerations in mind. Because the subjects were healthy volunteers with normal cardiac function, we wished to maintain periodic breathing cycles of durations in keeping with normal cardiac outputs. Such cycle durations would correspond to periodic breathing cycles lasting approximately 28 to 35 s in patients with idiopathic central sleep apnea (18). In addition, because we wished to determine whether periodic breathing cycles of different durations would entrain BP and HR oscillations at their corresponding frequencies, we developed two periodic breathing patterns with different cycle lengths. Hall and col-

Figure 1. Schematic representation of the experimental setup. During experiments, the VT output from the respiratory inductance plethysmograph was displayed on a computer screen. Horizontal lines on the screen represent, from bottom to top, baseline V T during regular breathing, and three, five, and seven times baseline VT. By using visual feedback of their VT from the computer screen, subjects were able to follow a given breathing pattern. Note that the maximal V T and the apnea time were identical during periodic breathing with either three or five breaths. However, the cycle length was different, owing to the different number of breaths in the hyperpneic phase.

Lorenzi-Filho, Dajani, Leung, et al.: Periodic Breathing Entrains Blood Pressure Oscillations leagues (18) previously demonstrated that the durations of periodic breathing cycles in patients with and without heart failure differ according to variations in the duration of hyperpneas rather than in the duration of apneas (which are relatively consistent, at approximately 20 s). Accordingly, we varied the durations of periodic cycles of breathing by varying the number of breaths during hyperpneas between apneas, while holding the durations of apneas constant. Additionally, to test the hypothesis that the magnitudes of increases in BP and HR oscillations during periodic breathing would be related to the magnitudes of augmentation of ventilation, we generated two different degrees of augmented ventilation between apneas. Periodic breathing pattern three (PB3) consisted of repeated series of three consecutive augmented breaths with VT values of three, seven, and three times the baseline VT, alternating with 20-s periods of apnea, with a cycle length of approximately 29 s. Periodic breathing pattern five (PB5) consisted of repeated series of five consecutive augmented breaths with VT values of three, five, seven, five, and three times the baseline VT, also alternating with 20-s periods of apnea, with a cycle length of approximately 33 s. Accordingly, the cycle lengths of PB3 and PB5 patterns corresponded to what one would expect during periodic breathing in someone with normal cardiac function (18). To allow subjects to become comfortable, and to minimize any potential anxiety associated with performing these breathing patterns, subjects practiced several periodic breathing cycles before the actual protocol. To simulate the sequence of events observed in patients with CSR (19), the PB3 and PB5 patterns were initiated by hyperventilation (i.e., augmented breaths) followed by apneas and then by hyperventilation, and so on. Subjects were instructed to maintain a constant respiratory frequency. Because there was a tendency for FETCO2 to decline over the course of the PB3 and PB5 patterns, three subjects repeated the PB5 protocol while breathing through a tight-fitting face mask with separate inspiratory and expiratory valves. The inspiratory valve was connected by wide-bore tubing to a three-way stopcock, both valves of which were connected to 60-L Douglas bags. One Douglas bag contained air from a compressed air tank to which it was connected by tubing. The second Douglas bag was connected to compressed gas tanks, one containing a 3% CO2, 21%O2, and 76% N2 mixture, and the other containing compressed air. To prevent progressive hypocapnia, these two gases were mixed in this Douglas bag, and the fraction of inspired CO2 (FIO2) inhaled from the bag was adjusted by manually controlling the inflow rates of the two gas streams. This technique allowed the administration of different CO2 concentrations while maintaining a constant 21% inspired O2 concentration. Subjects breathed, in random order, either room air or the CO2-enriched gas.

Data Analysis All data were acquired and analyzed by a customized computer program (LabVIEW; National Instruments, Austin, TX). All signals were sampled at 200 Hz except for the electrocardiograph signal, which was sampled at 1000 Hz. The minimum SaO2 during RB, PB3, and PB5 was determined for each subject. To express the variability in BP, the maximum and minimum systolic, diastolic, and mean BPs were determined for RB, PB3, and PB5. The differences between the maximum and minimum systolic, diastolic, and mean BPs were then calculated for three cycles of PB3 and PB5, and during a similar period during RB. HR was analyzed in the same way. The same analyses were done for the three subjects who inhaled room air and the CO2-enriched gas mixture. In these subjects, the lowest FETCO2 during the first and last hyperpneas during inhalation of room air and inhalation of CO2 was also determined. Oscillations in BP and HR during RB, PB3, and PB5 were compared through one-way analysis of variance (ANOVA) for repeated measures, with post hoc testing by the Student–Newman– Keuls method. Minimum SaO2 values for the three breathing patterns were also compared through this method. For the three subjects who breathed air and the CO2-enriched gas mixture, comparisons of the lowest FETCO2 values during the first and last hyperpneas, and of mean oscillations in BP and HR with PB5 for air and for CO2 inhalation, respectively, were made with paired t tests. The BP, HR (expressed as its inverse; R-wave-to-R-wave, or RR interval), and ventilatory signals were then analyzed in the frequency domain, using Fourier transforms. This allowed study of the signals

1149

and of their interactions at separate frequencies. Once data were acquired, they were analyzed over 7-min intervals in two stages. The first stage (preprocessing) involved extracting the RR intervals. These were interpolated to generate continuous curves. These and all the other signals were then passed through a low-pass digital filter (5th order, cutoff frequency 5 0.8 Hz) and resampled at 10 Hz. In addition to allowing reduction of the data set, the low-pass filter suppressed the spectral power at the heartbeat frequency. The second stage of processing involved subdividing the 7-min data set, which comprised 4,096 points, into three segments of 2,048 points each, with half-overlapping of each segment. The mean and linear trend were subtracted from the data in each segment. Because breathing patterns during RB, PB3, and PB5 were controlled and were consistent throughout, all signals remained stationary during each of the 7-min intervals. This was confirmed by visual inspection. A Blackman–Harris window was applied to minimize spectral leakage. The cross-power spectrum and autopower spectrum were then ensemble-averaged over the three segments to reduce spectral variance. Power spectra were obtained over a frequency range of 0.0049 to 0.5 Hz. With instantaneous lung volume derived from the VT signal taken as the input variable, and BP and the RR-interval taken as the two output variables, coherence, transfer-function magnitude, and phase angle were calculated at the two periodic breathing frequencies. Coherence is a measure (from 0 to 1) of the linear correlation between input and output signals. A coherence . 0.5 at a given frequency implies a strong relationship of the output to the input signal (20). Transfer-function magnitude (or gain) relates the magnitude of the output (e.g., BP or RR interval) to that of the input (e.g., instantaneous lung volume) signal at a given frequency, and is calculated by dividing the power in the cross-spectra of the output and input variables by power in the autospectra of the input variable (20). Transfer-function phase angle relates the timing of the output to that of the input signal. At a known frequency, a negative phase angle occurs when input precedes output. Both transfer-function magnitude and phase are considered reliable only at those frequencies at which coherence is . 0.5. Coherence, transfer-function magnitude, and phase angles were determined and averaged at the discrete power-spectral frequency peaks of instantaneous lung volume corresponding to the periodic breathing frequency, its prominent harmonics, and the highest peak in the respiratory frequency range. Because respiratory frequencies were not completely restrained, some minor variability in periodic breathing and respiratory frequencies was anticipated. This could give rise to more than one power peak in the periodic breathing and respiratory frequency ranges. To take this possibility into account, we quantified the spectral power of a given variable within three standard frequency bandwidths used for the analysis of HR variability (21): the very low frequency range (0.0049 to 0.05 Hz), which would encompass cycle frequencies of periodic breathing; the low-frequency range (0.05 to 0.15 Hz); and the high-frequency range (0.15 to 0.5 Hz), which would encompass breath-to-breath respiratory cycle frequencies. The power within a frequency band equals the area under the power-spectral density curve, and corresponds to the variance of a signal in that frequency range. These analyses quantify the power within a given frequency band during each of the three respiratory patterns. The total power within the periodic breathing, low-frequency, and respiratory frequency ranges was compared among the three breathing patterns through Friedman’s repeated measures ANOVA on ranks, with post hoc testing with the Student–Newman–Keuls method. All data are presented as mean 6 SD. A value of p , 0.05 was considered statistically significant.

RESULTS SaO2 did not fall below 95% at any time during RB, PB3, or PB5 during breathing of either room air or the CO2-enriched gas. In addition, there were no significant differences in minimal SaO2 during RB (96 6 2%), PB3 (97 6 2%), or PB5 during breathing either of room air (98 6 1%) or the CO2enriched gas (97 6 1%). This was because the first 20-s period of apnea followed the initial hyperpnea (18), and because the O2 concentration in the CO2 gas mixture was the same as in air (i.e., 21%). The lowest FETCO2 decreased from the first (3.0 6

1150

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

0.2%) to the last period of hyperpnea (2.0 6 0.2%, p 5 0.009) during breathing of room air, whereas it did not change significantly from the first (3.8 6 1.1%) to the last hyperpnea (3.7 6 0.9%) during CO2 inhalation. Effects of Periodic Breathing on Oscillations in BP and HR

Figure 2 shows tracings of the respiratory pattern, BP, and RR interval from one representative subject during RB, PB3, and PB5. During PB3 and PB5, oscillations in BP and in the RR interval were entrained by ventilatory oscillations such that peaks in BP and HR (i.e., troughs in RR intervals) consistently occurred during hyperpnea and troughs during apnea. This was also the pattern observed during CSR (6) and OSA (11). In Figure 3, average data for all seven subjects show that the peak-to-trough oscillations in systolic, diastolic, and mean BP were significantly greater during PB3 and PB5 than during RB (p , 0.005). Also, these oscillations were significantly greater during PB5 than during PB3 (p , 0.05). Additionally the peak-to-trough oscillations in HR during both PB3 (20.2 6 2.3 beats/min) and PB5 (20.2 6 4.7 beats/min) were significantly greater than during RB (13 6 6.0 beats/min, p , 0.01). In the three subjects who inhaled both air and CO2 during PB5, the oscillations in systolic, diastolic, and mean BP and in HR during CO2 inhalation (22.8 6 2.1 mm Hg, 17.3 6 4.2 mm Hg, 17.5 6 4.8 mm Hg, and 19.2 6 5.8 beats/min, respectively) were no different than during air breathing (24.6 6 5.1 mm Hg, p 5 0.61; 16.7 6 1.4 mm Hg, p 5 0.86; 17.2 6 2.8 mm Hg, p 5 0.93; and 21.7 6 2.9 beats/min, p 5 0.48, respectively). Effects of Periodic Breathing on Power-Spectral Density

Data for the power-spectral density of instantaneous lung volume, BP, and the RR interval during RB, PB3, and PB5 are presented in Figures 4, 5, and 6, respectively. During RB (Figure 4), the power-spectral density of instantaneous lung vol-

VOL 159

1999

ume was concentrated within the breath-to-breath respiratory frequency range (0.15 to 0.5 Hz). The highest peak was at 0.210 6 0.03 Hz. The corresponding peaks in the power spectral densities for BP and the RR interval observed at a similar frequency were consistent with a breath-to-breath respiratory influence (also known as respiratory sinus arrhythmia). At this point the coherence, transfer function, and phase angle between instantaneous lung volume and BP were 0.94 6 0.05, 6.8 6 5.5 mm Hg/L, and 2122 6 41 degrees, respectively, and the coherence, transfer function, and phase angle between the instantaneous lung volume and RR interval were 0.88 6 0.22, 240 6 235 ms/L and 2202 6 27 degrees. The relatively large power for BP and somewhat smaller power for the RR interval at , 0.1 Hz were not associated with corresponding peaks in instantaneous lung volume in this frequency range. This is consistent with the concept that during RB, oscillations are present in BP and in the RR interval at frequencies below 0.15 Hz, but are not related to respiratory oscillations. During PB3 and PB5, discrete peaks in the power spectrum for instantaneous lung volume were concentrated within the periodic cycle frequency range (0.0049 to 0.05 Hz), and were an order of magnitude higher than at the respiratory cycle frequencies (Figures 5 and 6). The fundamental frequencies of instantaneous lung volume peaks were 0.035 6 0.002 Hz for PB3 and 0.030 6 0.019 Hz for PB5. These corresponded to the respective mean periodic breathing cycle lengths of 28.6 s and 33.3 s. The power-density peaks for BP and the RR interval also shifted to these lower frequencies, and corresponded precisely to the power density peaks for instantaneous lung volume at the periodic breathing frequencies (Figures 4 and 5). At this frequency during PB3, the coherence, transfer function, and phase angle between instantaneous lung volume and BP were 0.97 6 0.01, 8.3 6 2.9 mm Hg/L, and 232 6 16 degrees, and the coherence transfer function, and phase angle between instantaneous lung volume and the RR interval were 0.91 6 0.10, 84 6 77 ms/L, and 2129 6 24 degrees. During

Figure 2. Representative recordings of VT, BP, the RR interval, and SaO2 in one subject during RB, PB3, and PB5. Oscillations in BP and the RR interval were present during RB, but they became more prominent and regular, and were entrained at the periodic breathing cycle frequency, during PB3 and PB5. Sa O2 did not decrease during PB3 or PB5 because overall, the subjects hyperventilated.

Lorenzi-Filho, Dajani, Leung, et al.: Periodic Breathing Entrains Blood Pressure Oscillations

1151

Figure 3. Comparison of peak-to-trough oscillations in systolic, diastolic, and mean BP averaged over RB, PB3, and PB5. The magnitudes of oscillations in systolic, diastolic, and mean BP were significantly greater during PB3 and PB5 than during RB. In addition, the oscillations in systolic, diastolic, and mean BP were greater during PB5 than during PB3. * p , 0.05.

PB5, the corresponding values between instantaneous lung volume and BP were 0.91 6 0.10, 8.1 6 3.7 mm Hg/L, and 218 6 16 degrees, and those between instantaneous lung volume and the RR interval were 0.85 6 0.14, 85 6 55 ms/L, and 2118 6 92 degrees. The transfer magnitudes between instantaneous lung volume and both BP and the RR interval at the periodic breathing cycle frequencies were similar for PB3 and PB5. It is important to note that although the largest values of VT during PB3 and PB5 were the same, the power peak for instantaneous lung volume was higher for PB5. This is because these power peaks do not represent peak VT alone, but rather an integration of total VT during hyperpnea, which by design was greater during PB5 than during PB3. In addition, the presence of two prominent harmonics of the fundamental frequency of instantaneous lung volume during PB3 (Figure 5), and of a single prominent harmonic during PB5 (Figure 6), is noteworthy. The prominent harmonics of instantaneous lung volume occurred at 0.072 6 0.002 Hz and 0.105 6 0.006 Hz for PB3, and at 0.061 6 0.004 Hz for PB5. Power-density peaks in BP and the RR interval were also evident at these frequencies. During PB3 the coherence, transfer function, and phase angle between instantaneous lung volume and BP for the first of the two prominent harmonics had values of 0.78 6 0.24, 3.5 6 1.8 mm Hg/L, and 2101 6 90 degrees, respectively, and for the second harmonic had values of 0.66 6 0.20, 3.6 6 1.8 mm Hg/L, and 294 6 94 degrees, respectively. The corresponding values between instantaneous lung volume and the RR interval during PB3 were 0.88 6 0.11, 80 6 34 ms/L, and 2173 6 31 degrees, respectively. During PB5, the values between instantaneous lung volume and BP for coherence, transfer function, and phase angle at the single prominent harmonic were 0.65 6 0.34, 4.1 6 2.8 mm Hg/L, and 2116 6 83 degrees, respectively. At this harmonic, the corresponding values between instantaneous lung volume and the RR interval were 0.90 6 0.07, 97 6 41 ms/L, and 2178 6 30 degrees, respectively.

Figure 4. Mean (solid lines) plus SD (dotted lines) of power-spectral densities for instantaneous lung volume (ILV) (respiration), BP, and the RR interval during regular breathing for all seven subjects. Note the discrete peaks in spectral power for BP and the RR interval within the respiratory frequency range. At this frequency the coherences between ILV and BP and between ILV and the RR interval were very high (0.94 and 0.88, respectively).

A quantitative transformation of the power-spectral density plots, shown in Figures 4 through 6, into spectral power under the curve within the periodic breathing (very low), low frequency breathing, and respiratory (high) frequency breathing ranges is presented for statistical comparison in Table 1. Within the periodic breathing frequency band, there were progressive and significant increases in the power spectra for instantaneous lung volume, BP, and the RR interval from RB to PB3 and PB5, as consistent with the power-density plots in Figures 4 to 6. There was also a progressive and significant increase in the power for instantaneous lung volume within the respiratory frequency range from RB to PB3 and PB5. This probably occurred because subjects increased their overall ventilation from RB to PB3 and PB5. However, there were no significant increases in power for BP or for the RR interval in the respiratory frequency range from RB to PB3 and PB5. The increase in the low frequency spectral power for instantaneous lung volume from RB to PB3 and PB5 was due to a contribution by power from harmonics of the fundamental frequency of the periodic breathing cycles.

DISCUSSION We have shown that, in healthy awake subjects, periodic breathing with central apneas entrains BP and HR, and increases the magnitude of their oscillations. Note that throughout the discussion, the term HR is used when referring to time-domain data, whereas its inverse, RR interval, is used when referring to frequency-domain data. Oscillations in BP

1152

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

Figure 5. Average data for all seven subjects during PB3, displayed in the same format as for Figure 3. Note the increase in the vertical scales for the power densities of instantaneous lung volume (ILV), BP, and the RR interval over those for RB in Figure 4. Power density peaks for ILV, BP, and the RR interval in the respiratory frequency range were of the same order of magnitude as during RB. They appear much smaller because of the change in the vertical scale from that in Figure 4. Discrete power-density peaks for ILV, BP, and the RR interval, not present during RB, are now observed at precisely the PB3 cycle frequency (0.035 Hz). These power peaks are an order of magnitude greater than those at the respiratory frequency during both RB (Figure 4) and PB3, indicating a powerful entrainment of variability in BP and the RR interval by the PB3 respiratory pattern. At this frequency, coherences between ILV and BP and between ILV and the RR interval were very high (0.97 and 0.91, respectively).

and HR intervals were proportional to oscillations in respiration at the periodic breathing frequencies. These were evidenced by discrete peaks in spectral power for BP and the RR interval that corresponded precisely, and with remarkably high levels of coherence, to the discrete peaks in respiratory power observed during the PB3 and PB5 patterns. Magnitudes of the transfer function between respiration and both BP and the RR interval at these frequencies were similar during PB3 and PB5. The negative phase angles between peaks in the spectral power of ventilation during PB3 and PB5 and peaks in the spectral power of BP and the RR interval indicate that changes in ventilation preceded changes in BP and the RR interval. Therefore, our data are novel in showing that even in the absence of hypoxia and arousals from sleep, periodic breathing produces an overall pattern of cyclic increases in BP and HR at the resumption of ventilation following apnea. This pattern closely resembles that observed in CSR and OSA. During CSR, other physiologic variables than ventilation also fluctuate in concert with the periodic breathing cycle, and

VOL 159

1999

Figure 6. Average data for all seven subjects during PB5, displayed in the same format as for Figures 4 and 5. Note that as compared with PB3 (Figure 5), the scales for the spectral power of ILV, BP, and the RR interval are increased. Findings are similar to those for PB3, except that the discrete power-spectral peaks for ILV, BP, and the RR interval in the periodic breathing frequency range are much higher and correspond precisely to the PB5 cycle frequency (0.030 Hz). At these spikes, coherences between ILV and BP and between ILV and the RR interval were very high (0.91 and 0.85, respectively).

could precipitate oscillations in BP and HR. These variables and their fluctuations include dips in SaO2, changes in PaCO2, arousals from sleep, and oscillations in cardiac output (CO). Periodic arterial O2 desaturation can occur in association with ventilatory periodicity in disease states, and could stimulate chemoreceptor reflexes and thus modulate BP and HR via autonomic afferents (9, 22). Aardweg and colleagues (23) examined the effects of hypoxia on BP and HR oscillation in healthy awake subjects who performed voluntary periodic breathing with 20-s central apneas. This pattern was similar to that used in the present study, except that the ventilatory pattern between apneas was unrestrained. Aardweg and colleagues found that periodic surges in BP observed under hypoxic conditions were damped, but not eliminated by hyperoxia, and concluded that hypoxia was the primary stimulus to increases in BP. However, they were unable to completely separate the effects on BP of hypoxia from those of ventilation, because subjects hyperventilated more after central apneas under hypoxic conditions. In our study hypoxia did not occur, since SaO2 was maintained above 95% and was as high or higher than during RB throughout; nevertheless marked oscillations in BP were present. Our findings are therefore more in keeping with those of Franklin and coworkers and Ringler and associates (11), who found that supplemental O2

1153

Lorenzi-Filho, Dajani, Leung, et al.: Periodic Breathing Entrains Blood Pressure Oscillations TABLE 1 DISTRIBUTION OF POWER-SPECTRAL DENSITY IN BANDS Periodic Breathing Frequency Band (VLF) (0.0049–0.05 Hz)

RB PB3 PB5

Low-Frequency Band (LF) (0.05–0.15 Hz)

Breath-to-Breath Respiratory Frequency Band (HF) (0.15–0.5 Hz)

ILV (L2 )

BP (mm Hg2 )

RR (ms2 )

ILV (L2 )

BP (mm Hg2 )

RR (ms2 )

ILV (L2 )

BP (mm Hg2 )

RR (ms2 )

0.001 6 0.001 0.186* 6 0.079 0.281* 6 0.096

4.5 6 2.6 14.7* 6 4.8 23.3* 6 14.9

567 6 234 2,444* 6 3,405 3,466*† 6 4,010

0.001 6 0.001 0.097* 6 0.033 0.066*† 6 0.029

2.3 6 1.6 3.2 6 2.2 3.7 6 2.7

613 6 306 1,362 6 1,004 1,228 6 792

0.021 6 0.008 0.176* 6 0.126 0.239* 6 0.178

1.4 6 1.1 1.9 6 0.5 2.2 6 0.8

1,698 6 1,919 1,421 6 1,091 1,505 6 1,445

Definition of abbreviations: BP 5 blood pressure; HF 5 high frequency; ILV 5 instantaneous lung volume; LF 5 low frequency; PB3 5 periodic breathing pattern three; PB5 5 periodic breathing pattern five; RB 5 regular breathing; RR 5 RR interval; VLF 5 very low frequency. * p , 0.005 as compared with RB. † p , 0.05 as compared with PB3.

administration did not affect the magnitude of postapneic surges in BP during CSR or OSA, respectively. Hypocapnia and cyclic oscillations in PaCO2 were present during periodic breathing in our protocol. These mimic the hypocapnia and fluctuations in PaCO2 fluctuations that trigger central apneas in patients with CSR (18, 24). Although it is unlikely that hypocapnia would provoke increases in BP and HR during hyperpnea, this possibility was nevertheless ruled out by demonstrating that raising FETCO2 by inhalation of a CO2enriched gas did not affect the timing or magnitude of oscillations in BP and HR related to periodic breathing. In addition, we (25) and others (26) have shown that by increasing lung CO2 stores, CO2 inhalation has a buffering effect that damps breath-by-breath and cycle-by-cycle oscillations in PaCO2. Therefore, in our experiment, it is very unlikely that hypocapnia or oscillations in PaCO2 played a major role in provoking periodic oscillations in BP and HR. Although arousals are thought to play an important role in triggering increases in BP at the termination of both central and obstructive apneas during sleep (6, 11, 27), periodic increases in BP and HR have been observed in association with CSR during wakefulness (2, 5, 6). The present data provide further evidence that periodic oscillations in BP and HR can be generated by periodic breathing without arousals from sleep (28). It is possible that fluctuations in BP are, at least in part related to fluctuations in CO caused by periodic breathing. Indirect evidence for this comes from the observation that both cerebral blood flow and BP increase during hyperpnea and decrease during apnea in patients with CSR (6). However, CO was not measured in the present study. Increases in BP could modulate HR via arterial baroreceptor-mediated reflexes (22). Although in our study such modulation was apparent on a breath-to-breath basis during hyperpneas between apneas, the overall pattern during hyperpneas was for both BP and HR to increase (Figure 2). Stimulation of arterial baroreceptors by increases in BP should have caused reflex reductions in HR rather than the observed increases in HR. Our findings are therefore not consistent with a primary baroreceptor-mediated reflex as the cause of peaks in HR during the ventilatory phase. Periodic oscillations in BP and HR during CSR appear to be respiratory-related, and resemble an exaggerated form of sinus arrhythmia, but at the lower frequency of periodic breathing (5, 6, 29). Nevertheless, the possibility that ventilation could contribute to these oscillations has received little attention. In one study, Ringler and colleagues (27) tested the hypothesis that resumption of ventilation at the termination of obstructive apneas during sleep might contribute to surges in BP (27). Airway occlusions begun during sleep and main-

tained for 6 to 12 s after arousal were associated with increases in BP coincident with arousal, not with resumption of ventilation. Ringler and colleagues concluded that resumption of ventilation was unlikely to play a major role in provoking surges in BP following apnea. However, they acknowledged that reflex pathways elicited by the distressing sensation of asphyxia prolonged into the period of full wakefulness, not normally present in OSA, could explain their findings. The present data indicate that oscillations in ventilation themselves are sufficient to precipitate and entrain oscillations in BP and RR intervals during periodic breathing in healthy subjects. The degree of entrainment is emphasized by two further observations. First, the magnitudes of oscillations in BP and the RR interval increased from PB3 to PB5, and were proportional to the associated increases in ventilation and respiratory spectral power. Second, both PB3 and PB5 produced harmonics in the power spectra for BP and the RR interval. The prominent harmonics of the fundamental frequencies of periodic breathing all fell within what is known as the low-frequency range, where respiration is thought to have little influence on cardiovascular oscillations (2, 12). Therefore, the appearance of discrete peaks in the spectra of BP and the RR interval at these frequencies, and with relatively high coherence, is further evidence of the extent to which periodic breathing entrains these two variables. Moreover, for both PB3 and PB5, the transfer function between respiration and the two cardiovascular variables at the frequencies of the harmonics was of the same order of magnitude as the transfer function at the fundamental frequency of periodic breathing. This suggests that respiration is equally efficient in entraining BP and HR oscillations whether they occur in the low-frequency range or at the fundamental frequency of periodic breathing in the very-low–frequency range. Although our study was not designed to identify the precise mechanisms by which oscillations in ventilation produce oscillations in BP and the RR interval, our findings indicate that oscillations in ventilation are relevant to the pathophysiology of fluctuations in BP and the RR interval in various forms of periodic breathing, including OSA, high-altitude periodic breathing, and CSR (5, 6, 8, 30). Intermittent reductions in PaO2, increases in PaCO2, or arousals from sleep are additional factors that could contribute to oscillations in BP and the RR interval in these disorders (11, 23, 27, 28). However, the extent to which each of these factors contributes to oscillations in BP and the RR interval probably varies among individuals. Another mechanism for the oscillations in BP and the RR interval during periodic breathing might be a primary, phaselinked oscillation in central nervous system outflow that simultaneously influences ventilation, BP, and the RR interval (6). This could be related to an overall increase in central

1154

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

drive from the respiratory and sympathetic cardiovascular neurons in the brain stem (31), which could modulate all these responses. According to this perspective, a number of stimuli, including hypoxia, CO2, arousals from sleep, or voluntary cortical influences, could contribute to periodic cardiorespiratory outflow. Further studies, involving techniques such as sympathetic neural recording and ganglionic blockade, will be required to test this hypothesis. Voluntary override of the spontaneous breathing pattern could lead to reduced parasympathetic control of HR, and could have influenced RR intervals and BP during the ventilatory phase of PB3 and PB5 in our study (32). However, this appears very unlikely, because the spectral power of the RR interval at the respiratory frequency, which should have decreased in the face of parasympathetic withdrawal, did not change during periodic breathing (Table 1). In summary, the present work shows that in healthy subjects, periodic breathing with central apneas increases the amplitude of oscillations in BP and HR, and entrains these oscillations at the frequency of the periodic breathing. Hypoxia, CO2 retention, and arousals from sleep are not involved in this process. Furthermore, the magnitude of the oscillations is proportional to oscillations in ventilation. Therefore, oscillations in ventilation itself should be taken into account when investigating the pathophysiology of fluctuations in BP and HR in various forms of periodic breathing, including CSR and OSA. Whether these respiratory-related oscillations involve mechanisms similar to those that mediate respiratory-related, breath-to-breath changes in BP and HR will require further investigation.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

References 1. Bradley, T. D., and J. S. Floras. 1996. Pathophysiologic and therapeutic implications of sleep apnea in congestive heart failure. J. Card. Fail. 2:223–240. 2. Mortara, A., P. Sleight, G. D. Pinna, R. Maestri, A. Prpa, M. T. La Rovere, F. Cobelli, and L. Tavazzi. 1997. Abnormal awake respiratory patterns are common in chronic heart failure and may prevent evaluation of autonomic tone by measures of heart rate variability. Circulation 96:246–252. 3. Hanly, P. J., and N. S. Zuberi-Khokhar. 1996. Increased mortality associated with Cheyne-Stokes respiration in patients with congestive heart failure. Am. J. Respir. Crit. Care Med. 153:272–276. 4. Andreas, S., G. Hagenah, C. Moller, G. S. Werner, and H. Kreuzer. 1996. Cheyne-Stokes respiration and prognosis in congestive heart failure. Am. J. Cardiol. 78:1260–1264. 5. Trinder, J., R. Merson, J. Rosenberg, F. Fitzgerald, J. Kleiman, and T. D. Bradley. 1997. Effects of arousal on blood pressure fluctuations in patients with Cheyne-Stokes respiration and heart failure (abstract). Am. J. Respir. Crit. Care Med. 155:A776. 6. Franklin, K. A., E. Sandstrom, G. Johansson, and E. M. Balfors. 1997. Hemodynamics, cerebral circulation, and oxygen saturation in Cheyne-Stokes respiration. J. Appl. Physiol. 83:1184–1191. 7. Guilleminault, C., S. Connolly, R. Winkle, K. Melvin, and A. Tilkian. 1984. Cyclical variation of the heart rate in sleep apnoea syndrome: mechanisms, and usefulness of 24 h electrocardiography as a screening technique. Lancet 1:126–131. 8. Tilkian, A. G., C. Guilleminault, J. S. Schroeder, K. L. Lehrman, F. B. Simmons, and W. C. Dement. 1976. Hemodynamics in sleep-induced apnea: studies during wakefulness and sleep. Ann. Intern. Med. 85: 714–719. 9. Leuenberger, U., E. Jacob, L. Sweer, N. Waravdekar, C. Zwillich, and L. Sinoway. 1995. Surges in muscle sympathetic activity during obstructive apnea are linked to hypoxemia. J. Appl. Physiol. 79:581–599. 10. Morgan, B. J., T. Denahan, and T. J. Ebert. 1993. Neurocirculatory consequences of negative intrathoracic pressure vs. asphyxia during voluntary apnea. J. Appl. Physiol. 74:2969–2975. 11. Ringler, J., R. C. Basner, R. Shannon, R. Schwartzsein, H. Manning, S. E. Weinberger, and J. W. Weiss. 1990. Hypoxemia alone does not

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

VOL 159

1999

explain blood pressure elevations after obstructive apneas. J. Appl. Physiol. 69:2143–2148. Ponikowski, P., T. P. Chua, A. A. Amadi, M. Piepoli, D. Harrington, M. Volterrani, R. Colombo, G. Mazzuero, A. Giordano, and A. J. S. Coats. 1996. Detection and significance of a discrete very low frequency rhythm in RR interval variability in chronic congestive heart failure. Am. J. Cardiol. 77:1320–1326. Daly, M. D. B. 1986. Interactions between respiration and circulation. In N. S. Cherniack and J. G. Widdicombe, editors. Handbook of Physiology, 3rd ed. American Physiological Society, Bethesda, MD. 529–594. Chadra, T. S., H. Watson, S. Birch, G. A. Jenouri, A. W. Schneider, M. A. Cohn, and M. A. Sackner. 1982. Validation of respiratory inductive plethysmography using different calibration procedures. Am. Rev. Respir. Dis. 125:644–649. Parati, G., R. Casadei, A. Groppelli, M. Di Rienzo, and G. Mancia. 1989. Comparison of finger and intra-arterial blood pressure monitoring at rest and during laboratory testing. Hypertension 13:647–655. Virolainen, J. 1992. Use of non-invasive finger blood pressure monitoring in the estimation of aortic pressure at rest and during the Mueller manoeuvre. Clin. Physiol. 12:619–628. Omboni, S., G. Parati, A. Frattola, E. Mutti, M. Di Rienzo, P. Castiglioni, and G. Mancia. 1993. Spectral and sequence analysis of finger blood pressure variability: comparison with analysis of intra-arterial recordings. Hypertension 22:26–33. Hall, M. J., A. Xie, R. Rutherford, S. I. Ando, J. S. Floras, and T. D. Bradley. 1996. Cycle length of periodic breathing in patients with and without heart failure. Am. J. Respir. Crit. Care Med. 154:376–381. Naughton, M., D. Benard, A. Tam, R. Rutherford, and T. D. Bradley. 1993. Role of hyperventilation in the pathogenesis of central sleep apneas in patients with congestive heart failure. Am. Rev. Respir. Dis. 148:330–338. Berger, R. D., J. P. Saul, and R. J. Cohen. 1989. Transfer function analysis of autonomic regulation: I. Canine atrial rate response. Am. J. Physiol. 256:H142–H152. Butler, G. C., S. Ando, and J. S. Floras. 1997. Fractal component of variability of heart rate and systolic blood pressure in congestive heart failure. Clin. Sci. 92:543–550. Goldberger, A. L., L. J. Findley, M. R. Blackburn, and A. J. Mandell. 1984. Nonlinear dynamics in heart failure: implications of long-wavelength cardiopulmonary oscillations. Am. Heart J. 107:612–615. Aardweg, J. G. V., and J. M. Karemaker. 1992. Repetitive apneas induce periodic hypertension in normal subjects through hypoxia. J. Appl. Physiol. 72:821–827. Xie, A., B. Wong, E. A. Phillipson, A. S. Slutsky, and T. D. Bradley. 1994. Interaction of hyperventilation and arousal in the pathogenesis of idiopathic central sleep apnea. Am. J. Respir. Crit. Care Med. 150: 489–495. Xie, A., F. Rankin, R. Rutherford, and T. D. Bradley. 1997. Effects of inhaled CO2 and added dead space on idiopathic central sleep apnea. J. Appl. Physiol. 82:918–926. Band, D. M., I. R. Cameron, and S. J. G. Semple. 1969. Effect of different methods of CO2 administration on oscillations of arterial pH in the cat. J. Appl. Physiol. 26:268–273. Ringler, J., E. Garpestad, R. C. Basner, and J. W. Weiss. 1994. Systemic blood pressure elevation after airway occlusion during NREM sleep. Am. J. Respir. Crit. Care Med. 150:1062–1066. O’Donnell, C. P., T. Ayuse, E. D. King, A. R. Schwartz, P. L. Smith, and J. L. Robotham. 1996. Airway obstruction during sleep increases blood pressure without arousal. J. Appl. Physiol. 80:773–781. Tilkian, A. G., C. Guilleminault, J. S. Schroeder, K. L. Lehrman, F. B. Simmons, and W. C. Dement. 1977. Sleep-induced apnea syndrome: prevalence of cardiac arrhythmias and their reversal after tracheostomy. Am. J. Med. 63:348–358. Insalaco, G., S. Romano, A. Salvaggio, O. Marrone, M. R. Bonsignore, A. Braghiroli, P. Lanfrachi, and V. Patruno. 1997. Blood pressure behaviour during sleep periodic breathing at high altitude (abstract). Am. J. Respir. Crit. Care Med. 155:A852. Guyenet, P. G., N. Koshiya, D. Huangfu, A. J. Verberne, and T. A. Riley. 1993. Central respiratory control of A5 and A6 pontine noradrenergic neurons. Am. J. Physiol. 264:R1035–R1044. Patwarhan, A. R., S. Vallurupalli, J. M. Evans, E. N. Bruce, and C. F. Knapp. 1995. Override of spontaneous respiratory pattern generator reduces cardiovascular parasympathetic influence. J. Appl. Physiol. 79:1048–1054.