Respiratory sinus arrhythmia (RSA) is classically described as a vagally mediated increase and decrease in heart rate concurrent with inspiration and expiration, respectively. However, although breathing frequency is known to alter this temporal relationship, the precise nature of this phase dependency and its relationship to blood pressure remains unclear. In 16 subjects we systematically examined the temporal relationships between respiration, RSA, and blood pressure by graphically portraying cardiac interval (R-R) and systolic blood pressure (SBP) variations as a function of the respiratory cycle (pattern analysis), during incremental stepwise paced breathing. The principal findings were 1) the time interval between R-R maximum and expiration onset remained the same (∼2.5–3.0 s) irrespective of breathing frequency (P = 0.10), whereas R-R minimum progressively shifted from expiratory onset into midinspiration with slower breathing (P < 0.0001); 2) there is a clear qualitative distinction between pre- versus postinspiratory cardiac acceleration during slow (0.10 Hz) but not fast (0.20 Hz) breathing; 3) the time interval from inspiration onset to SBP minimum (P = 0.16) and from expiration onset to SBP maximum (P = 0.26) remained unchanged across breathing frequencies; 4) SBP maximum and R-R maximum maintained an unchanged temporal alignment of ∼1.1 s irrespective of breathing frequency (P = 0.84), whereas the alignment between SBP minimum and R-R minimum was inconstant (P > 0.0001); and 5) β1-adrenergic blockade did not influence the respiration-RSA relationships or distinct RSA patterns observed during slow breathing, suggesting that temporal dependencies associated with alterations in breathing frequency are unrelated to cardiac sympathetic modulation. Collectively, these results illustrate nonlinear respiration-RSA-blood pressure relationships that may yield new insights to the fundamental mechanism of RSA in humans.
- autonomic nervous system
- blood pressure
the act of breathing modulates autonomic neural outflow from the brainstem (2, 3, 10, 19). One manifestation of this process is respiratory sinus arrhythmia (RSA), which is the fluctuation of cardiac cycle interval (R-R interval) synchronous with respiration. Although RSA is classically described as an alternation between inspiratory R-R interval shortening (cardiac acceleration) and expiratory R-R interval lengthening (cardiac deceleration), it is recognized that the respiration-RSA phase relationship is dependent, in part, on breathing frequency (1, 16). However, the precise nature of this relationship in supine resting subjects is debatable. Angelone and Coulter (1) first suggested that this dependency was characterized by R-R minimum shifting closer toward inspiratory onset with slower breathing frequencies. In contrast, Eckberg (16) subsequently found that R-R minimum was fixed to expiratory onset and that it was R-R maximum that shifted relative to inspiration with changes in breathing frequency. Several researchers have also described a fixed time delay of ∼1.0 s from the beginning of inspiration to R-R maximum irrespective of breathing frequency (12, 33), which is inconsistent with Eckberg's results.
The mechanisms underlying RSA are also unclear with some attributing it exclusively to fluctuations in efferent cardiac vagal activity (16), while others have suggested that the phase changes observed at slower breathing frequencies are caused by combined vagal and sympathetic heart rate modulation (42). Moreover, despite extensive research, it remains unclear as to whether RSA is driven by respiratory synchronous oscillations in blood pressure via the arterial baroreflex or whether RSA and blood pressure are independently related to respiration via nonbaroreflex mechanisms (4, 14, 17, 28, 41, 46). Clarification of these fundamental relationships is important for our understanding of how autonomic neural outflow is coupled with respiratory activity, especially given that RSA is considered a surrogate of cardiac parasympathetic modulation and is widely applied in the calculation of spontaneous baroreflex sensitivity (7, 24, 34, 36, 37, 39).
Most studies that have previously examined the temporal relationship between breathing, R-R interval, and blood pressure oscillations at the respiratory frequency have relied on techniques such as cross-correlation (33) and cross spectral analysis (39, 40, 42). These methods, however, assume that cardiovascular and respiratory oscillations are parallel fluctuations separated by a fixed phase difference. As a result, they do not allow for detailed study of temporal relationships taking place at different phases within the boundaries of the respiratory cycle. One approach that circumvents this limitation is the graphical portrayal of the pattern of R-R interval and blood pressure variations as a function of the respiratory cycle (pattern analysis) (21). In addition to providing amplitude information, pattern analysis also characterizes the nonlinear dynamics of R-R interval and blood pressure changes occurring throughout the respiratory cycle.
The objective of the current study was to apply pattern analysis to identify possible nonlinearities in the respiration-RSA-blood pressure relationship. This was achieved by examining the comparative temporal alignment of blood pressure and RSA pattern waveforms relative to the respiratory cycle across a range of breathing frequencies. Given that cardiac sympathetic responsiveness is known to be relatively low in the supine posture, we also tested the hypothesis that any nonlinear respiration-RSA-blood pressure phase shifts associated with changes in breathing frequencies would be mediated primarily by vagal mechanisms. This was achieved by examining cardiovascular patterns before and after β1-adrenergic blockade.
Sixteen healthy male subjects with mean age of 21 years (range, 20–25) were recruited. All subjects gave written informed consent and were fasted for at least 4 h before the study. Ethical approval was obtained from the New Zealand Central Regional Ethics Committee, and the experimental protocol was performed in accordance with the Declaration of Helsinki.
The ECG (Grass Instruments P511 amplifier), respiratory flow (Hans Rudolph heated pneumotach 8420A; Vacumed differential pressure transducer 4500), and noninvasive blood pressure via radial artery tonometry (Nellcor N-CAT N-500; Hayward, CA) were acquired at 500 Hz per channel via a 16-bit I/O data acquisition board (PCI-6023E series; National Instruments). Offline analysis was performed using custom written software in LabView 8.2.1 (National Instruments) on a Macintosh 2.66 GHz MacBook Pro computer.
Subjects were studied in the supine position in an air-conditioned and humidity-controlled laboratory. Target breathing frequencies were controlled in trained subjects by displaying a digital sequence of 20 light-emitting diodes (LEDs) coupled with an auditory tone generated from a computer (49). Subjects inhaled when the first of a sequence of 10 green LEDs was lit and exhaled without pause when the eleventh LED, which marked the beginning of a sequence of 10 red LEDs, was lit. Following an initial 5-min stabilization period, paced breathing was commenced at 0.20, 0.15, and 0.10 Hz in randomized order for 5 min each, with 2-min rests between trials. Thereafter, a subcohort (n = 10) received intravenous bolus injections of metoprolol tartrate (β1-adrenergic blocker, 0.2 mg/kg). Following a 10-min equilibration period, the breathing protocol was repeated. No restrictions were imposed on the tidal volume (Vt), although subjects were instructed to maintain a comfortable level of ventilation to maintain normal end-tidal CO2 levels via visual biofeedback from gas analyzer output.
Characterization of RSA and blood pressure pattern.
From the raw ECG, blood pressure, and respiratory flow signals, R-wave times, inspiratory and expiratory onset times, and systolic blood pressures (SBP) were extracted. RSA patterns were generated from 5-min artifact free recordings using previously described methods (21, 48). In brief, the occurrence of the preceding R-R interval of each heartbeat and SBP value of each pressure pulse were plotted to its relative position within the respiratory cycle, which was defined as a linear function of time between inspiratory onsets (Fig. 1A). Therefore, each data point is represented on the horizontal axis as a ratio (i.e., the phase) of the respiratory cycle. This was followed by cubic-spline interpolation of the R-R interval and SBP values in each respiratory period into n = 500 data points, giving the individual RSA and SBP pattern waveforms for each breath as a function of respiratory phase (Fig. 1B). This procedure was repeated for each of the m respiratory cycles. The spline interpolated curves from all respiratory cycles were then superimposed, and an averaged RSA pattern and SBP pattern expressed in the phase domain calculated as and (xij represents a sample of the ith interpolated point of the jth respiratory cycle, where i = 1,...n and j =1,...m; Fig. 1C). As subjects were fixed pace breathing, the time domain equivalent of both the RSA pattern and SBP pattern were taken as , where r is the predetermined respiratory period. For each averaged pattern we extracted times of the inflexion points (i.e., maximum and minimum) relative to inspiration and expiration onset and the differences in these between the RSA and SBP pattern (Fig. 2). RSA amplitude and SBP amplitude were calculated as the difference between maximum and minimum R-R interval and SBP, respectively.
Power spectral analysis.
Spectral analysis was performed on SBP and R-R interval time series of length 256 s. The R-R interval time series was manually checked, and spurious readings were corrected by linear interpolation. The resultant R-R interval and SBP time series were sampled at 4 Hz to provide 1,024 equidistant points, which were then high-pass filtered to remove fluctuations below 0.015 Hz and low-pass filtered to exclude components above 2 Hz. These series were passed through a Hanning window and subject to fast Fourier transform analysis. The spectral power was calculated as the integrated area under the power spectrum coincident with respiratory frequency and expressed in absolute units. Cross-spectral analysis was also performed to determine the coherence and phase difference between R-R interval and SBP.
All variables were assessed for normality, and those identified as nonparametric were log-transformed before statistical testing. All variables were reported as means ± SD. Differences between dependent variables across different breathing frequencies were assessed using one-way repeated-measures ANOVA. Where this was significant, paired comparisons were made with Student's paired t-test. Two-way repeated-measures ANOVA was used to assess within-subjects factors (β1-adrenergic blockade and breathing frequency). Because the main effect for interaction in all tests was not significant, these values have been omitted from the results. Where the two-way repeated-measures ANOVA indicated a significant main effect for β1-adrenergic blockade, pairwise comparisons were further made using Student's paired t-test. Statistical significance was set a priori at 0.05, and the Bonferroni correction was applied to control for type 1 error associated with multiple testing. Analyses were performed using SPSS 16.0.2 (SPSS, Chicago, IL).
The effects of breathing frequency on baseline variables are summarized in Table 1. RSA amplitude and SBP amplitude increased significantly with slower breathing. However, mean R-R interval, end-tidal CO2, and systolic, mean arterial, and diastolic blood pressure were not significantly affected by breathing frequency.
Relationship Between Breathing Frequency and RSA Pattern
Figure 3 shows representative RSA and SBP patterns for one subject during paced breathing at 0.20, 0.15, and 0.10 Hz. During 0.20-Hz breathing, RSA pattern resembles a symmetrical sinusoid with the R-R maximum and R-R minimum temporally aligned to inspiration and expiration onset, respectively. Therefore, during 0.20-Hz breathing, inspiration was associated with cardiac acceleration and expiration was associated with cardiac deceleration. In contrast, during 0.15-Hz breathing, R-R maximum commenced before inspiratory onset, whereas R-R minimum remained aligned to expiration. Finally, during 0.10-Hz breathing, the RSA pattern resembled a complex asymmetric sinusoid with cardiac acceleration occurring before inspiration and cardiac deceleration occurring before expiration.
Figure 4 summarizes the alignment relationships between RSA and respiration illustrated in Fig. 3 averaged for all subjects. R-R maximum was aligned, both in time and phase, to inspiratory onset during 0.20-Hz breathing but occurred before inspiration during 0.15- and 0.10-Hz breathing. R-R minimum occurred progressively later in relation to inspiratory onset with slow breathing when examined in the time domain (P < 0.0001). However, it occurred closer to inspiratory onset when examined in the phase domain (P < 0.0001).
R-R minimum shifted considerably with respect to expiration onset in both the time (P < 0.0001) and phase (P < 0.0001) domain, occurring with expiration onset at 0.20 and 0.15 Hz but before expiration during 0.10-Hz breathing. There was no fixed phase relationship between expiration onset and the R-R maximum (P < 0.0001). However, in the time domain, R-R maximum began ∼2.5–3.0 s following expiratory onset (P = 0.10).
In all subjects, cardiac acceleration commenced before inspiratory onset. However, detailed visual examination of the RSA patterns at 0.10-Hz breathing revealed two distinct components of cardiac acceleration in 13 of the 16 subjects. In these subjects, we observed a preinspiratory cardiac acceleration stage, which commenced before inspiratory onset, and an inspiratory cardiac acceleration stage, which occurred with inspiration onset. Moreover, the rate of inspiratory cardiac acceleration in these subjects was consistently greater than the rate of preinspiratory cardiac acceleration. Figure 5 shows data from four subjects demonstrating these two components of cardiac acceleration, and Fig. 3 shows an example of one subject where the two components were not apparent.
Relationship Between Breathing Frequency and Blood Pressure Pattern
Figure 6 shows SBP minimum maintained a constant temporal relationship to inspiratory onset (∼1.7 s) across different breathing frequencies in time (P = 0.16) but not in phase (P < 0.0001). There were no consistent time (P < 0.0001) or phase (P < 0.0001) differences between inspiratory onset and SBP maximum across different breathing frequencies. Similarly, across the breathing frequencies, SBP maximum occurred consistently ∼1.7 s after expiration (P = 0.28) and showed no fixed relationships in the phase domain (P > 0.0001). There were no constant time (P < 0.0001) or phase (P < 0.0001) alignments between expiration onset and SBP minimum.
Relationship Between RSA and Blood Pressure Pattern
Figure 7 shows that the cross-spectral phase angle between RSA and SBP became increasingly larger with slower breathing (P > 0.0001). A negative phase angle in this context refers to SBP preceding RSA. Coherence between the two signals was >0.5 for all subjects. The time lag between R-R minimum and SBP minimum showed the same trend of becoming more negative with slower breathing (P > 0.0001). However, the time lag of ∼1.1 between R-R maximum and SBP maximum was unchanged by breathing frequency (P = 0.84).
Effects of β1-Adrenergic Blockade
Given that RSA pattern analysis in the time domain yielded relationships that were not apparent in the phase domain, only the time domain results are presented in the following section.
Figure 8 shows that mean R-R interval increased after β1-adrenergic blockade [F (1, 9) = 36.95; P = 0.0002] but was unchanged across different breathing frequencies [F (1, 9) = 0.45; P = 0.65]. Both β1-adrenergic blockade [F (1, 9) = 8.25; P = 0.018] and reductions in breathing frequency [F (1, 9) = 21.61; P > 0.0001] enhanced RSA amplitude. The apparently large standard deviations were due to substantial interindividual variability in RSA amplitude rather than frequency-related variability within individuals.
β1-adrenergic blockade did not alter the timing of R-R maximum or R-R minimum relative to inspiration or expiration onset [Fs (1, 9) < 0.70; Ps > 0.42]. The breathing frequency-dependent relationships were the same as those under the control condition [Fs (1, 9) > 7.12; Ps > 0.0061].
The timing of SBP maximum to R-R maximum and SBP minimum to R-R minimum was unchanged following β1-adrenergic blockade [F (1, 9) = 0.63, P = 0.45; and F (1, 9) = 2.24, P = 0.17, respectively] but varied across breathing frequency in the same fashion as seen under the control condition [F (1, 9) = 1.90, P = 0.18; and F (1, 9) = 30.45, P < 0.001, respectively].
The present investigation is the first to apply a pattern analysis approach to characterize nonlinearities in the temporal relationships between respiration, RSA, and blood pressure within the boundaries of the respiratory cycle. Under the conditions of this study the five major findings are: 1) the time interval between R-R maximum and expiration onset remained the same irrespective of breathing frequency, whereas R-R minimum progressively shifted from expiratory onset into midinspiration with slower breathing; 2) two qualitatively distinct stages of cardiac acceleration during slow 0.10-Hz breathing were observed in most subjects; 3) both the time intervals between inspiration onset and SBP minimum and between expiration onset and SBP maximum were unchanged by breathing frequency; 4) SBP maximum and R-R maximum maintained a fixed temporal alignment irrespective of breathing frequency, whereas, in contrast, the alignment between SBP minimum and R-R minimum varied according to breathing frequency; and 5) β1-adrenergic blockade did not influence the respiration-RSA relationships or distinct RSA patterns observed during slow breathing. Before the physiological implications of these findings are discussed, several methodological aspects of the study warrant consideration.
Whereas previous studies have examined RSA patterns expressed only as a function of respiratory phase (21, 31, 32, 48, 49), we have characterized cardiovascular patterns in both the phase and time domain to ensure any time-based relationships were preserved in our analysis. The major advantage of pattern analysis is that R-R interval changes are tracked as a function of the respiratory cycle. Therefore, rather than output a single global phase angle value for a given recording (20, 33, 50), pattern analysis allows for more detailed assessment of temporal relationships occurring within different stages of the respiratory cycle (21, 31, 32, 48, 49). As an example of this potential utility, prior studies that altered breathing frequency without Vt control have shown that the SBP-to-R-R interval cross-spectral phase angle becomes increasingly negative as breathing is slowed from a fast (0.27 Hz) to a slow (0.10 Hz) frequency (39). With the use of pattern analysis it is possible to further ascertain whether the SBP-to-R-R interval lag time is the same across all stages of the respiratory cycle. However, to simplify our statistical analysis, only the time intervals between SBP and R-R interval at inflexion points were compared. Furthermore, since we sought to relate our findings to established mechanistic frameworks, such as the hypothesis that RSA is baroreflex mediated (13), these time intervals were interpreted on the a priori assumption that SBP lead R-R interval changes.
Breathing Frequency and Respiratory Sinus Arrhythmia
Across the different breathing frequencies we observed significant temporal shifts between respiration and RSA that were distinct from the classically described inspiratory cardiac acceleration and expiratory cardiac deceleration. R-R maximum was aligned to inspiratory onset only during 0.20-Hz breathing, whereas during 0.15- and 0.10-Hz breathing, R-R maximum began before inspiratory onset. Moreover, the interval between R-R maximum and expiration onset across the three breathing frequencies was relatively unchanged. Although R-R minimum was aligned to expiratory onset at 0.15- and 0.20-Hz breathing, it occurred in midinspiration during 0.10-Hz breathing (Figs. 3 and 4). These results are, therefore, in partial agreement with previous descriptions of respiration-RSA phase relationships in that R-R maximum shifts to a preinspiratory onset with slower breathing (16) and R-R minimum shifts with slower breathing (1, 30). However, the findings contrast reports of a 0.9–1.0-s time delay from inspiratory onset to R-R maximum irrespective of breathing frequency (12, 33). They are also inconsistent with the notion that across a wide range of breathing frequencies there is an obligatory alignment of R-R minimum to expiration (16, 25, 26); our data clearly indicate there is no fixed temporal alignment between R-R minimum and expiratory onset.
There are several possible explanations for the disparate observations between different studies. First is the lack of stringent breathing pattern control across the different breathing frequencies. A short inspiration relative to expiration is associated with enhanced RSA amplitude and more rapid cardiac acceleration compared with regular and relatively long inspiratory periods (43). Furthermore, both end-inspiration and end-expiration pauses alter the cardiac deceleration component of RSA (31). Such differences in inspiratory-to-expiratory period ratio control may have led to differing conclusions. Our study should be interpreted in cognizance of the 1:1 inspiration-to-expiration ratio without respiratory pauses. Another possible confounding factor is Vt control. Under fixed Vt conditions, Eckberg (16) observed a fixed temporal relationship between expiration and R-R minimum across a range of breathing frequencies. In contrast, without explicit Vt control we observed clear changes in the timing relationship between expiration and R-R minimum when breathing was slowed from 0.20 to 0.10 Hz. Prior studies have also reported significant phase shifts between respiration and RSA across a comparable range of breathing frequencies without Vt control (20, 50). However, the application of cross-spectral analysis in those studies preclude any detailed assessment of changes in respiration-RSA phase or time alignments that might be taking place within different stages of the respiratory cycle.
Beyond the classical description of RSA as inspiratory cardiac acceleration and expiratory cardiac deceleration, there are few qualitative details on how the R-R intervals are changing. At the slowest breathing frequency, we observed a clear qualitative distinction between preinspiratory and inspiratory cardiac acceleration; the rate of cardiac acceleration before inspiration was consistently less than the rate of cardiac acceleration following inspiratory onset in 13 of the 16 subjects (Fig. 5 shows 4 of these 13 subjects). This observation, which was possible because pattern analysis allowed the visualization of R-R interval dynamics taking place throughout the respiratory cycle, suggests that the cardiac acceleration component of RSA during slow breathing may be the product of complex processes. Although the mechanisms underlying these two stages of cardiac acceleration are unclear, we can draw some inferences from the data. First, as they clearly persist following β1-adrenergic blockade, it seems likely that neither are sympathetically mediated. Finally, although activation of pulmonary stretch afferents may be involved in RSA generation (44), this reflex mechanism is unlikely to cause preinspiratory cardiac acceleration, given that during slow breathing cardiac acceleration commences before inspiratory onset. However, it is possible that pulmonary stretch may contribute to the faster rate of cardiac acceleration following inspiratory onset.
Although commonly regarded as a vagal phenomenon, cardiac sympathetic efferent modulation may also influence RSA. For example, β1-adrenergic blockade augments RSA amplitude in humans across a broad range of breathing frequencies (45). Furthermore, using cross-spectral analysis, Saul et al. (42) showed that although the respiration-RSA phase relationship did not vary with breathing frequency in a pure parasympathetic state (i.e., supine position concurrent with β1-adrenergic blockade), slow breathing was associated with enhanced RSA amplitude and a lead of cardiac acceleration onset relative to inspiration in a pure sympathetic state (i.e., upright posture concurrent with cholinergic blockade). Collectively, these data indicate that, at least in the upright posture, cardiac sympathetic activity may modulate both RSA amplitude as well as the phase relationship between respiration and RSA. However, given that cardiac sympathetic tone is lower in the supine posture (51), we reasoned that the respiration-RSA phase relationships observed in supine humans would be unrelated to cardiac sympathetic modulation. Indeed, although β1-adrenergic blockade did augment RSA amplitude, the respiration-RSA phase relationships assessed using pattern analysis were unchanged following blockade. These findings, which differ from the conclusions of Saul et al. (42), suggest that at least in the supine posture cardiac sympathetic activity attenuates RSA amplitude but does not modulate frequency-dependent respiration-RSA phase relationships and changes in respiration-RSA relationships across different breathing frequencies primarily reflect changes in cardiac vagal activity due to central and/or baroreflex interactions (discussed below).
Breathing Frequency and Blood Pressure
Spontaneous blood pressure oscillations with respiration are generally attributed to mechanically induced changes in intrathoracic pressure, which in turn cause fluctuations in venous return and cardiac output (15, 22, 47, 52). In keeping with this notion, the inflexion points of the SBP waveform maintained constant alignment to inspiration and expiration onset, respectively, across all breathing frequencies in the time but not in the phase domain. The fixed time delay between respiration and the subsequent SBP may reflect the time required to transmit larger right ventricular stroke volumes associated with increased venous return through the pulmonary circulation and onward to the left ventricle. In contrast, for reasons that remain unclear, Laude et al. (33) found using cross correlation analysis that respiration and blood pressure were aligned with a fixed delay in phase rather than time. In keeping with prior studies, we observed significant augmentation of SBP fluctuations during slow 0.10-Hz breathing. This might be due to accentuation of the mechanical forces with elevated tidal volumes during slow breathing. However, it is also plausible that breathing within the range of the Mayer waves may lead to synergistic augmentation of 0.10-Hz blood pressure oscillations via direct central or baroreflex interactions with intrinsic mechanisms that generate blood pressure Mayer waves (27).
Relationship Between Blood Pressure and RSA
In keeping with previous studies (39), the cross-spectral phase angle between SBP and R-R interval fluctuations became increasingly negative as breathing slowed down (Fig. 7). However, although there was also an increasingly negative lag between SBP minimum and R-R interval minimum with pattern analysis, the interval between SBP maximum and R-R maximum stayed constant across the different breathing frequencies. Therefore, R-R interval and blood pressure fluctuations are not separated by a simple linear lag throughout the respiratory cycle.
Despite extensive research, the mechanism of respiratory-related R-R interval fluctuations (RSA) remains unclear. A central consideration has been whether blood pressure and RSA relationships are consistent with known properties of the human baroreflex. Eckberg (17) has recently argued that if RSA were purely baroreflex mediated arterial pressure should precede R-R interval changes by a time delay that is consistent with known baroreflex latencies and time lags should be well conserved within individuals independent of factors such as breathing frequency or body posture. In relation to the first assumption, the average lag between SBP maximum and R-R maximum across all breathing frequencies in this study was ∼1.1 s, whereas the time lag between SBP minimum and R-R minimum increased from 1.0 to 2.1 s at the slowest breathing frequency. Therefore, although these time lags broadly approximate the latency between an abrupt increase of carotid distending pressure and maximum R-R interval prolongation (∼1.5 s) (5) during fast breathing, they are consistent only for parts of the respiratory cycle during slow breathing. The second assumption is based on the premise that neither breathing frequency nor body posture should influence acetylcholine kinetics at the parasympathetic neuromuscular junction, which account for the vast majority (i.e., ∼72%) of the interval between baroreceptors sensing a blood pressure change and the final reflex modulation of sino-atrial nodal discharge rate (17). If this construct were correct, our data would suggest that although the present findings indicate that the baroreflex was indeed implicated in the generation of RSA, the baroreflex cannot be the sole generator of RSA across all stages of the respiratory cycle since the time lag between blood pressure and RSA was not conserved across all stages of the respiratory cycle with variations in breathing frequency.
We considered three possible explanations for our observations. First, as breathing frequency slows down and RSA amplitude increases, the shortest R-R interval within a respiratory cycle may become too brief to permit an in-phase cardiac response to the antecedent systolic pulse. Therefore, the time interval between SBP minimum and R-R minimum may have lengthened with slow breathing because the full effects of the pulse were seen in subsequent R-R intervals (38). This explanation, however, cannot fully account for our findings since SBP tends to alter the concurrent R-R interval when resting R-R interval is longer than 0.80 s (38); in this study the shortest R-R interval during slow breathing was longer than 0.8 s for all but two subjects.
The second explanation is that the phase changes may be mediated by cardiac sympathetic activity. Indeed, one study reported reductions in cross-spectral phase angle between SBP and RSA following chronic β1-adrenergic blockade with slow-acting metoprolol (39). However, the lag time between R-R interval minimum and maximum to SBP minimum and maximum did not change after acute β1-adrenergic blockade, suggesting that under the conditions of this study cardiac sympathetic activity does not account for the temporal relationships observed. The reason for this disparity is unclear but may relate to the chronicity of β1-adrenergic blockade.
The third possible explanation is that RSA may reflect the hybrid expression of both central and baroreflex processing. In humans during quiet breathing the respiratory gate manifests as a progressive decline in vagal motoneurone responsiveness throughout expiration, which does not begin to rise again until a point of minimum responsiveness around midinspiration has been reached (18). Accordingly, baroreflex-mediated excitation of cardiac vagal neurons may be relatively effective during early expiration but become relatively ineffectual during midinspiration as respiratory gating takes effect. Under this construct, the temporal evolution of RSA would be determined mainly by the level of vagal motoneurons excitation in the (relative) absence of respiratory gating and by the dynamics of the respiratory gating mechanism itself as the level of vagal motoneurone responsiveness declines to a relative minimum within the respiratory cycle. Speculatively, such dynamic variations in the efficacy of baroreceptor input to cardiac vagal neurons may explain why SBP maximum, which consistently occurred around early expiration, maintains a ∼1.1-s lag with R-R maximum, whereas SBP minimum, which consistently occurred around midinspiration, does not maintain a fixed lag with R-R minimum. Importantly, this hybrid model may also explain why the time interval between R-R maximum and expiratory onset (2.5–3.0 s), which was temporally fixed to SBP maximum, remained relatively unchanged across different breathing frequencies.
Finally, consistent with the established literature, we observed a clear relationship between breathing frequency and RSA amplitude (11, 23, 45, 49). In this study 0.10-Hz breathing was associated with a ∼1.8-fold higher RSA amplitude compared with 0.20-Hz breathing. This is most likely due to breathing frequency coinciding with, and thus significantly augmenting, low frequency R-R interval fluctuations. The mechanism(s) behind this resonance phenomenon is unclear, but one proposal is that the augmentation of cardiovascular oscillations associated with slow breathing is due to global enhancement of arterial baroreflex sensitivity (7).
The results of this study should be interpreted cognizant of two potential limitations. First, in addition to autonomic modulation, non-neural mechanisms may also mediate RSA. For example, RSA maintains a short and fixed time delay with respiration during strenuous exercise consistent with a direct intrinsic effect of respiration on R-R intervals (8, 9). Furthermore, heart transplant recipients without cardiac innervation retain a small degree of RSA at rest, suggesting that mechanical factors might also contribute to RSA (6, 44). However, such mechanical contributions are unlikely to confound our results given that RSA in conscious resting humans is mediated predominantly by neural mechanisms under resting conditions (53). Nevertheless, since we have not explicitly accounted for mechanical contributions to RSA, our findings should not be generalized to other conditions (e.g., intense exercise, alternate postures). Second, since we have not performed muscarinic cholinergic blockade in this study, our inference that changes in respiration-RSA relationships during paced breathing are vagally mediated is indirect. Finally, it is possible that incomplete blockade may have confounded some of our results. However, we consider this unlikely given that cardiac sympathetic activity is known to be suppressed in the supine position even without β1-adrenergic blockade (51), and the dose of metoprolol administered was comparable or greater than those previously given in studies that have demonstrated clear sympatholytic effects (29, 35).
In summary, this study revealed several previously undescribed nonlinearities in respiration-RSA-blood pressure relationships in conscious humans. In contrast to prior studies, we found R-R minimum was not temporally aligned to expiration, beginning in late inspiration with slower breathing. Similarly, the onset of R-R maximum was not fixed to inspiratory onset but occurred in late expiration at slower breathing frequiencies. We also found that R-R maximum consistently occured ∼2.5–3.0 s following expiratory onset irrespective of breathing frequency. We observed two qualitatively distinct stages of cardiac acceleration during slow 0.10-Hz breathing, whereby the rate of cardiac acceleration occurring before inspiration was consistently less than the rate of cardiac acceleration following inspiratory onset, which to the best of our knowledge has not previously been described. Since these temporal dependencies were unaltered by selective β1-adrenergic blockade, they are most likely due to vagally mediated mechanisms. Furthermore, SBP maximum and R-R maximum maintained a temporal alignment of ∼1.1 s irrespective of breathing frequency whereas the delay between SBP minimum and R-R minimum became longer with slower breathing. These results demonstrate that the application of pattern analysis to the study of heart rate and blood pressure variability has potential to yield new insights into fundamental relationships between breathing and autonomic regulation of cardiovascular function.
P. Y. W. Sin was supported by a University of Otago Postgraduate Scholarship. This research was supported by grants from the New Zealand Heart Foundation (Grant No. 1284) and from the New Zealand Health Research Council (09/186) awarded to Y. C. Tzeng.
No conflicts of interest, financial or otherwise, are declared by the author(s).
- Copyright © 2010 the American Physiological Society