The difference in arterial baroreflex sensitivity between the supine and standing positions in healthy subjects

Abstract

The spectral analysis of heart rate variability (HRV) has long been considered a practical, noninvasive tool to assess autonomic nervous functions that regulate the cardiovascular system. The conventional method, however, has limitations in characterizing transient changes of HRV. We have overcome this problem by adopting the time-frequency analysis method. Using this method, we attempted to clarify differences in the arterial baroreflex (ABR) activity during a transient change in the heart rate between the supine and standing positions in healthy subjects. We found that the ABR gain was significantly greater in the supine position compared to standing, and this gain increase was due to transient increases in 0.15 to 0.20 Hz components of HRV spectral powers caused by enhanced cardiac vagal outflow. Based on these findings, we conclude that the orthostatic stress induced by the postural change from supine to standing markedly reduces the baroreflex gain by suppressing high-frequency cardiac vagal outflow.

Highlights

• •Transient increases in spectral powers of heart rate fluctuation occasionally occur.
• •These transient power increases are generated by arterial baroreflex response.
• •Arterial baroreflex gain is greater in the supine position than during standing.
• •This supine gain increase is linked to increased 0.15-0.20 Hz spectral powers.
• •Orthostatic stress reduces baroreflex gain by suppressing cardiac vagal outflow.

Introduction

Spectral analysis of heart rate variability (HRV) is a practical, noninvasive method to assess cardiovascular autonomic functions [1], [2], [3], [4], [5]. In the power spectral density (PSD) calculated from a time series of the R-R interval (RRI), three main components are distinguished: very-low-frequency (VLF; 0.003–0.04 Hz), low-frequency (LF; 0.04–0.15 Hz), and high-frequency (HF; 0.15–0.4 Hz) components [5]. However, the physiological correlates of these frequency components are yet to be determined.

To understand the physiological and pathophysiological significance of HRV, in-depth assessments of RRI sequence dynamics and transient changes are inevitable. Previously, we conducted spectral analyses of data segments of various lengths of the R-R interval (RRI) time series [6], [7]. These studies adopted the time-frequency analysis method based on the short-time Fourier transform (STFT), widely used for time-frequency representations of real-world nonstationary signals [8], [9], [10], [11]. We found that the RRI data segment analysis shorter than 30 s was required to identify the temporal relationships between sporadically occurring individual transient RRI changes and the resulting spectral power increases, because of the trade-off relationship between time and frequency resolutions in signal processing using Fourier transform. In these pilot studies, we also found that transient HRV power increases during standing were generated by arterial baroreflex (ABR) activity [7].

In this study, we aimed to find, in the supine position, the counterpart of the previously discovered ABR-generated HRV changes during standing and to clarify the difference in the ABR response between the two postural positions. We found that in the supine position with the metronome respiration at 0.25 Hz, the transient ABR activity generated significantly greater increases in the HF (0.15–0.20 Hz) powers compared to during standing. We then performed a cross-spectral analysis of heart rate and mean arterial pressure (MAP) variabilities to quantify their coherence and transfer function gain (TFG). In the supine position, a marked increase in TFG was found in the 0.15–0.20 Hz frequency range during the transient ABR response, resulting in the baroreflex with faster kinetics than during standing.

Materials and methods

Participants

Ten volunteers (four males and six females, 20–25 years of age) participated in this study. They were in good health without cardiovascular and respiratory diseases and were not taking subscription medications regularly. They provided their written informed consent to participate in the study protocol, which the Ethical Committee of Takasaki University of Health and Welfare approved (Approval number: 2350).

Experimental protocols

The experiment was performed between 9:00 and 17:00 in a quiet, light-attenuated room with an ambient temperature of 20 to 25 ℃. Subjects, wearing measuring devices, were requested to adopt a supine position followed by active standing for 12 min each. During the data acquisition, participants were asked to breathe quietly with a steady breathing volume at a rate of 15 breaths/min (0.25 Hz) with timing indicated by the metronome.

We previously compared LF power amplitudes between the supine and standing positions. When the standing: supine ratio of the LF power was $\le$1, we referred to the response pattern to standing as type 1 and as type 2 when the ratio was > 1 [7]. As type 1 cases were most frequently observed, the present analysis was limited to type 1 cases.

Instrumentation

The HRV was measured with a wearable electrocardiogram sensing device (myBeat; Union Tool, Co., Tokyo, Japan). The device was worn around the epigastrium on the subject's chest, and the peak of the QRS complex was used to obtain an RRI time series. This device incorporates a three-axis accelerometer, allowing for the detection of the time points of body movement and postural change.

The regularity of respiratory frequency and depth during HRV measurement was evaluated with a respiration monitor (ResMo; Aimedic MMT, Tokyo, Japan) that detects thorax volume changes through a thoracic belt incorporating a pair of stretchable strain sensors. As this device accurately senses the small thorax volume changes associated with postural change, these signals were used to synchronize the respiratory movement time series with the RRI time series obtained with the electrocardiogram sensing device incorporating a three-axis accelerometer which enabled the detection of the time points of postural change.

We measured beat-to-beat arterial blood pressure with a finger arterial blood pressure monitoring device (viewphii-CBP; Socionext, Inc., Yokohama, Japan) during HRV measurement [12], [13]. Using this device, we could simultaneously measure changes in MAP and the interbeat interval (IBI). We previously reported the similarity of IBI time series to the RRI time series, which was simultaneously recorded with the electrocardiogram sensing device [7]. The current study confirmed the similarity between the IBI and RRI time series with no fundamental difference in the results of their spectral analyses. We used IBI for the heart rate signals in the cross-spectral analysis between heart rate and blood pressure variabilities.

Data processing and analysis

Analysis was performed on 300–360 s RRI, IBI, and MAP time series in the supine and standing positions, each covering 3–4 baroreflex responses.

We developed a Python program to track temporal relationships among the transient RRI (IBI) changes and changes in spectral power of various frequency ranges whose central frequencies were 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, and 0.35 Hz. With this program, time-frequency analysis of nonstationary signals was performed with the window length set to 20 s. Because the frequency resolution of this analysis is inversely equal to the time window length, namely 0.05 Hz (1/20 s), it enables the detection of power changes at 0.05 Hz intervals. We consecutively performed the STFT computation on the 20-s window with a 1-s shift, and detected changes in power amplitudes related to the baroreflex response in 0.05–0.35 Hz frequency ranges throughout the RRI (IBI) time series.

We also developed a new Python program to perform a cross-spectral analysis of the IBI and MAP time series to quantify frequency-related squared coherence (Coh², hereafter referred to as coherence) and TFG between them. This program calculates coherence and TFG using Welch’s method [14]. Coherence estimates the proportion of non-random variance common to both variables at a given frequency. The TFG quantifies the slope of the linear relationship between IBI and MAP. The computational algorithms in the program developed for calculating coherence and TFG are essentially the same as those for calculating modulus or α − index, which has been used to estimate baroreflex sensitivity when the coherence is higher than the conventionally accepted value of 0.5 [15], [16].

To calculate coherence and TFG values, we set a 30-s segment of IBI and MAP time series to be analyzed and made 11 20-s windows with a 1-s shift from the segment's starting point. Next, cross-correlation functions between spline-interpolated continuous MAP and IBI time series at a 0.1-s interval were obtained and cross-power spectra were computed for the 11 windows. The spectral powers of MAP and IBI were also obtained separately in these windows. The 11 cross-power spectra and MAP and IBI spectral powers were then averaged. Coherence was calculated by dividing the square of the averaged cross-spectral power by the product of the averaged MAP and IBI powers at each frequency range. TFG was estimated as the ratio of the averaged MAP-IBI cross-spectral powers to the averaged MAP powers with the unit of ms/mmHg.

Statistical analysis

Data are presented with the median and interquartile range (IQR). Statistical comparison was conducted with the Wilcoxon signed-rank test for paired data sets using the SPSS software (version 29.0 for Windows; IBM Corp., Armonk, NY, USA).

Results

Occasionally occurring sequence of a transient RRI decrease followed by an abrupt reversal to increase in the supine position

In the previous study, we found that a transient rise in arterial blood pressure occasionally occurred during standing in healthy subjects [7]. This blood pressure rise was associated with a transient RRI decrease followed by a reversal to increase, which in turn caused the increased blood pressure to fall. During this sequence of RRI changes, simultaneous increases in the VLF, LF, and HF powers were detected, although the HF power increase was slight. Based on those findings, we concluded that the sequences of RRI changes associated with HRV power increases in the wide frequency range reflect ABR activity during standing [7]. This conclusion agreed with the previous reports that the increase in LF powers of HRV fluctuations of RRI reflects baroreflex activity except that we detected a modest increase of HF power [17], [18], [19], [20].

To find out the counterpart of the transient ABR response for the supine position, we attempted to detect transient RRI changes independent of the respiratory sinus arrhythmia (RSA)-related RRI oscillations. For this purpose, subjects were asked to breathe quietly at 0.25 Hz with a steady breathing volume to identify RRI power changes unrelated to RSA. Fig. 1A shows typical results obtained for the supine position in such trials. In this case, the 320-s RRI time series had three sequences of RRI changes, a transient RRI decrease followed by an abrupt reversal to rapid increase, which differed from regularly occurring 0.25 Hz RRI oscillations due to RSA. These events occurred under conditions of constant respiratory frequency and volume.

Fig. 1.

Sequences of R-R interval (RRI) changes unrelated to respiratory movement in the supine position. (A) Top panel: Respiratory movement-induced thorax volume changes monitored with a thoracic belt incorporating a pair of stretchable strain sensors in a subject breathing quietly with a steady breathing volume of 15 breaths/min (0.25 Hz) with the aid of a metronome. The y-axis value indicates the output voltage of the strain sensor. Bottom panel: 320-s RRI time series with 0.25 Hz oscillation synchronized to respiratory movement. This time series also contains three sequences of large RRI changes, initial decreases followed by abrupt reversals to increases not associated with respiratory frequency and volume changes. (B) Detailed time courses of the three sequences of RRI changes. The rapid RRI increase from nadir to peak occurred over 2 s in all three cases. The case indicated by a red arrow was used for cross-spectral analysis whose results are shown in Fig. 4. (C) Power spectral densities (PSDs) of 100-s RRI time sequences containing the two sequences of RRI changes (a) and no such event (b) indicated by the arrows in the bottom panel in (A). In the PSD (a), in addition to the prominent 0.25 Hz power due to RSA, small but significant powers in wide frequency bands lower than 0.25 Hz were detected.

Fig. 1B shows the detailed time courses of these three events. The distinctive feature of these events was the rapid recovery of the RRI value from the nadir to the level before the event. In most cases, the recovery time was short and constant, 2 s, only rarely 3 s.

Fig. 1C shows the power spectral densities (PSDs) of the 100-s RRI time series containing two events and no event indicated by arrows (a) and (b), respectively, in the lower panel of Fig. 1A. Only the PSD (a) of the RRI segment containing the two sequences of RRI changes detected small but significant powers in wide frequency ranges from VLF to HF bands lower than 0.25 Hz, in addition to the prominent 0.25 Hz power attributed to RSA. We focused on the functional significance of these power increases, particularly those in higher frequency ranges, as shown by the arrow in PSD (a), which may be closely related to ABR activity in the supine position.

Temporal relationship between the sequence of transient RRI changes and RRI power increases in the supine position

Next, the temporal relationship between the sequence of RRI changes and RRI power increases in the supine position was examined. We performed the time-frequency analysis based on STFT to elucidate the temporal association of dynamic changes in the RRI spectral powers with the sequences of RRI changes. In this analysis, we computed RRI spectral powers at 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, and 0.35 Hz in the 20-s window of the RRI time series consecutively with a 1-s shift. The time series of RRI spectral powers obtained at 1-s intervals enabled us to detect the temporal localization of RRI power changes and to clarify their temporal relationship with the sequence of RRI changes. Fig. 2A shows the result of this analysis for the RRI time series presented in Fig. 1A. The second panel in this figure shows the temporal relationship between the sequence of RRI changes and spectral power changes at 0.05, 0.10, 0.15, and 0.20 Hz, and the third panel at 0.20, 0.25, 0.30, and 0.35 Hz. The 0.20 Hz power change is presented in both panels to facilitate power amplitude comparison among these seven frequency ranges.

Fig. 2.

Temporal relationships of changes in 0.05–0.35 Hz frequency powers with the sequence of R-R interval (RRI) and interbeat interval (IBI) changes in the supine position. (A) Temporal relationships of RRI spectral power changes over 0.05–0.35 frequency ranges with the sequence of RRI changes in the RRI time series shown in Fig. 1A (top panel). The RRI spectral powers at 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, and 0.35 Hz were calculated consecutively in the 20-s window of the RRI time series with a 1-s shift. The time series of RRI spectral power amplitudes thus obtained were plotted at 1-s intervals in the second and third panels. The second panel shows the results for 0.05, 0.10, 0.15, and 0.20 Hz powers, and the third panel results for 0.20, 0.25, 0.30, and 0.35 Hz powers. The 0.20 Hz power changes are presented in both panels to facilitate power amplitude comparison over the seven frequency ranges. (B) The same analyses as (A) for the IBI time series. No substantial difference was observed between the results of spectral analyses of RRI and IBI time series.

Panel 2 of Fig. 2A indicates the temporal association of the sequence with increased 0.05, 0.10, and 0.15 Hz powers. The sequence of RRI changes also generated a clear increase in 0.20 Hz power, but a slow oscillation was superimposed on its time series. Panel 3 shows the time series of spectral powers in the HF range at 0.20, 0.25, 0.30, and 0.35 Hz. A large oscillation of 0.25 Hz powers due to RSA under the 0.25 Hz metronome respiration was observed throughout the time series. The main component of this 0.25 Hz oscillation was unrelated to the sequence of RRI changes, although a slight increase seemed to synchronize occasionally. In the time series of the power at 0.30 Hz, a slow oscillation similar to that seen at 0.20 Hz was present, but a power increase related to the sequence of RRI changes was not detected. The slow oscillations seen at 0.20 Hz and 0.30 Hz, which were similar, are likely the overspills from the potent 0.25 Hz powers. These oscillations were no longer visible at 0.15 Hz and 0.35 Hz, each distanced 0.10 Hz from 0.25 Hz.

In Fig. 2B, we confirmed the similarity between RRI and IBI time series and no substantial difference between the results of spectral analyses of both time series. For the cross-spectral analysis of heart rate and blood pressure variabilities, using the IBI time series obtained with the finger arterial blood pressure monitoring device was more convenient. Therefore, we used IBI instead of RRI for HRV signals in the following experiments.

Comparison of baroreflex-related spectral power increases between the supine and standing positions

We compared spectral power changes linked to the sequence of IBI changes in the supine position with those during standing. Fig. 3 shows the IBI time series and associated changes in IBI spectral powers in the 0.05–0.25 Hz range during standing taken from the subject whose corresponding data in the supine position before the postural change are shown in Fig. 2B. During standing, the recovery time of IBI from the nadir to the original level was prolonged, and the 0.15–0.25 Hz power increases were markedly suppressed, whereas substantial portions of 0.05 and 0.10 Hz powers remained. Although the 0.15–0.25 Hz power increases were markedly suppressed, on closer inspection slight 0.15 and 0.20 Hz power increases linked to IBI recovery were detected, as indicated by three arrows in the bottom panel. RSA-generated 0.25 Hz power was similarly suppressed, but the temporal association of the reduced power with IBI recovery was not so tight as 0.15 and 0.20 Hz power increases.

Fig. 3.

Temporal relationships of changes in 0.05–0.25 Hz frequency powers with the sequence of interbeat interval (IBI) changes during standing. Example of the IBI time series and the sequence of IBI changes in spectral powers at 0.05- 0.25 Hz during standing. Data was obtained from the subject, whose RRI time series in the supine position is shown in Fig. 2, after postural change from the supine position. The top panel shows the IBI time series containing three sequences of IBI changes, and the bottom panels show spectral power increases associated with the IBI changes. The vertical lines through the top and bottom panels indicate the temporal relationship between the sequence of transient IBI changes and spectral power increases. Postural change from supine to standing markedly reduced the sequence-related power increases in the frequency range higher than 0.15 Hz, although slight 0.15–0.20 Hz components remained as indicated by arrows. Note that standing also markedly reduced the RSA-generated 0.25 Hz component. The second sequence of IBI changes indicated by a red arrow in the top panel was used for cross-spectral analysis, whose results are shown in Fig. 4.

Evaluation of coherence and transfer function gain between heart rate and blood pressure variabilities during the transient arterial baroreflex response

Next, we attempted to examine the relationship between heart rate and blood pressure variabilities during the sequence of transient IBI decrease followed by a rapid reversal to increase, and compared the difference between the supine position and standing. For this purpose, we adopted a cross-spectral analysis of IBI and MAP variabilities to obtain coherence and TFG between the two signals.

Supine position

Fig. 4A shows results obtained for the IBI and MAP variabilities in the second sequence of rapid RRI changes in the supine position indicated by a red arrow in Fig. 1A. As shown in Fig. 4Aa, during the sequence of events a transient MAP increase associated with a decrease in IBI was detected, and after several seconds, the IBI decrease abruptly reversed to a rapid increase with the MAP simultaneously returning to the initial level. This sequence of events was similar to those during standing, which we have previously reported [7], except that in the supine position, the rate of the reversed IBI increase was much faster. We considered that this sequence of events was generated by an occasionally occurring ABR activity and defined it as the transient ABR response.

Fig. 4.

Example of spectral and cross-spectral analysis of time series of heart rate and blood pressure variabilities taken from a subject in the supine and standing positions. (A) Results of analyses in the supine position. Thirty-second IBI and mean arterial pressure (MAP) data segments containing the transient arterial baroreflex (ABR) response enclosed by a blue rectangle (a. Baroreflex) and immediately before it (b. Control) were prepared from the data indicated by a red arrow in Fig. 2A. The results of analyses of (a) and (b) are plotted in graphs (c), (d), (e), and (f) in red and black dots, respectively. Graphs A(c) and A(d) show the averaged power spectra of MAP and IBI variabilities, respectively. Graph A(e) shows high coherences (>0.7) between IBI and MAP variabilities in all frequency ranges between 0.05 and 0.30 Hz both during and before the ABR response in this case. Graph A(f) shows that ABR caused a modest increase in transfer function gain (TFG) at 0.05, and 0.10 Hz, and the increase became marked at 0.15 and 0.20 Hz. In contrast, before the ABR response, an increase in TFG was detected at frequencies higher than 0.20 Hz which reached the level during the ABR response at 0.30 Hz. (B) Results of analyses during standing. The same analysis was performed for 30-s IBI and MAP data segments containing the ABR response enclosed by a blue rectangle (a. Baroreflex) and immediately before it (b. Control), which were prepared from the data indicated by a red arrow in Fig. 3 taken from the subject during standing after postural change from the supine position. The results of analyses of (a) and (b) are plotted in graphs (c), (d), (e), and (f) in red and black dots, respectively. Graphs B(c) and B(d) show the averaged power spectra of MAP and IBI variabilities, respectively. Graph B(e) shows that coherences between MAP and IBI variabilities are greater than 0.5 in all frequency ranges between 0.05 and 0.30 Hz during and before ABR response. Graph B(f) shows that ABR caused a modest increase in TFG at 0.05 and 0.10 Hz, similar to the supine position. However, no further ABR-induced increase in TFGs at 0.15 and 0.20 Hz was detected during standing. Note that the y-axis in graph B(f) is magnified fourfold compared to graph A(f).

To analyze the relationship between IBI and MAP variabilities during the transient ABR response, a 30-s data segment was selected (enclosed by the blue rectangle), in which the rise of MAP started 10 s after the onset of the data segment, indicated by a vertical line in Fig. 4Aa. Then, 11 20-s windows with a 1-s shift from the segment's starting point were made. Each of these 11 windows overlapped in the middle 10-s section of the 30-s data segment. This makes most of the IBI and MAP changes caused by the ABR response being contained in all 11 windows, thus preserving the frequency components related to the IBI and MAP changes after averaging. The length of the single window was set to 20 s to ensure a consistent frequency resolution with the frequency analysis as explained in the previous section.

Results obtained from the IBI and MAP variation data in Fig. 4Aa are presented as red dots and lines in the graphs Ac, Ad, Ae, and Af. Graph Ac is the averaged power spectrum showing two MAP power peaks, a dominant peak at 0.05–0.10 Hz and a smaller peak at 0.25 Hz. The averaged power spectrum in graph Ad shows increasing IBI power occurring in both LF and HF ranges. Graph Ae shows that in this 30-s data segment, the coherence between IBI and MAP variations was high (>0.8) in all frequency ranges between 0.05 and 0.30 Hz, indicating the close correlation between IBI and MAP signals. Graph Af shows that TFG increased sharply at 0.15 Hz, remaining high until 0.30 Hz.

Concerning TFG increase detected at a frequency range higher than 0.20 Hz in this experiment, it may be due, at least partly, to heart rate and blood pressure oscillations caused by RSA. During RSA, blood pressure oscillations occur mainly in lower frequency ranges, whereas heart rate oscillations are driven by HF vagal activity. Therefore, RSA may induce the TFG increase in the HF range, independently of baroreflex activity. To examine this possibility, we attempted to obtain the TFG in the 30-s IBI and MAP data segment immediately before the baroreflex response. In 13 out of 30 cases, including the case shown in Fig. 4A, a data segment was present in which coherence between IBI and MAP variations exceeded 0.5 in all frequency ranges from 0.05 to 0.30 Hz (Fig. 4Ab Control). This data segment, whose example is shown in Fig. 4Ab, is referred to as the control segment. The results of analyses of the 30-s IBI and MAP data shown in Fig. 4Ab are superimposed as black dots and lines in the graphs Ac, Ad, Ae, and Af. The LF MAP power was modest for the control condition before the baroreflex response (Graph Ac). IBI powers increased from 0.20 Hz, with an amplitude similar to that detected during the ABR at 0.30 Hz (Graph Ad). Graph Ae confirmed the high coherence between the two signals. Graph Af shows the calculated TFG, which gradually increased from 0.15 Hz to 0.30 Hz and reached a similar level to that during the baroreflex response at 0.30 Hz. Comparison of the TFG values before and during the transient ABR response indicates: 1) At 0.05 and 0.10 Hz, the ABR slightly increased TFG. 2) At 0.15 Hz, ABR markedly increased TFG. 3) The TFG increase during the baroreflex response appeared to be retained until 0.30 Hz. However, a significant component of TFG increase higher than 0.20 Hz was related to RSA-driven high-frequency IBI activity. Thus, the pure ABR-generated TFG increase in the supine position mainly occurred at 0.15 and 0.20 Hz as shown in Graph Af.

Standing

Next, we performed the same analyses during standing in 13 cases, where the coherence between IBI and MAP variabilities exceeded 0.5 in all frequency ranges between 0.05 and 0.30 Hz, both before and during the transient baroreflex response. Fig. 4Ba and 4Bb show examples of the 30-s MAP and IBI data segments during ABR response and the control period before it, respectively, during standing. These data were obtained from the same subject who provided the data shown in Fig. 4A. The results of the analyses were shown in red dots and lines, and in black dots and lines for the data during and before the baroreflex response, respectively, in graphs Bc, Bd, Be, and Bf.

Graph Bc is the power spectrum showing two peaks of MAP powers at 0.05–0.10 Hz and 0.25 Hz; the former was dominant. The power distribution shape was similar to that in Fig. 4Ac. In contrast, IBI powers during the transient ABR response shown in Graph Bd were reduced compared to those in the supine position shown in Graph Ad in all frequency ranges, especially markedly in HF ranges. In 13 cases, the medians (IQRs) were 1483.0 (1162.6), 1220.0 (964.3), 861.8 (816.6), and 688.9 (950.4) ms², at 0.05, 0.10, 0.15, and 0.20 Hz, respectively, in the supine position. Postural change to standing significantly reduced these values to 624.0 (1029.5, p < 0.01), 398.7 (522.0, p < 0.01), 74.4 (184.2, p < 0.002), and 96.9 (18.2, p < 0.001) ms², at 0.05, 0.10, 0.15, and 0.20 Hz, respectively. The ratios of median values during standing relative to the supine position were 42.1 %, 32.8 %, 8.6 %, and 14.1 % at 0.05, 0.10, 0.15, and 0.20 Hz, respectively. Thus, suppression of IBI power during the transient baroreflex response in the standing position was more marked in HF ranges.

Graph Be shows that in these 30-s data segments, the coherence between IBI and MAP variations exceeded 0.5 in all frequency ranges between 0.05 and 0.30 Hz. In graph Bf, the TFG during the ABR response was similarly higher at 0.05 and 0.10 Hz than before it, as was in the supine position. However, the prominent TFG increase at 0.15–0.20 Hz seen in the supine position was not detected. In the period before the ABR response, a much smaller increase in TFG was observed in the range higher than 0.20 Hz compared to the supine position, reflecting the strong suppression of RSA by orthostatic stress. Thus, during standing, the TFG was markedly suppressed except for the 0.05 and 0.10 frequency ranges during the ABR response. Note that the y-axis is magnified fourfold in graph Bf compared to that in graph Af in the supine position.

The box plots in Figs. 5A and 5B depict TFGs in the supine (A) and standing (B) positions in 13 cases obtained in the present study. TFGs during and before the transient ABR response are compared in 0.05–0.30 frequency ranges. Note that the y-axis in Fig. 5B during standing is magnified fourfold. The results shown in these box plots are summarized as follows: 1) Both in the supine and standing positions, at 0.05 and 0.10 Hz, the ABR activity significantly increased TFG (p < 0.01–0.001). 2) In the supine position, the median of TFG at 0.10 Hz and 0.15 Hz were 20.2 (IQR 11.3) and 40.2 (IQR 30.5), respectively. Thus the baroreflex activity markedly increased TFG at 0.15 Hz compared to 0.10 Hz (p < 0.001). In contrast, during standing, the corresponding values were 9.7 (IQR 5.7) and 11.6 (IQR 8.1) at 0.10 Hz and 0.15 Hz, respectively, with no statistical difference between the two frequencies (p = 0.50). 3) In the supine position, at 0.20 Hz, the ABR-induced TFG increase remained high, and was sustained at 0.25 Hz. 4) In the supine position, at 0.30 Hz, no statistical difference was observed in the TFG value before and during the ABR response, although both were high. 5) During standing, the ABR-induced TFG remained low between 0.15 and 0.30 Hz (median <15). The value before the ABR response showed a slight increase in the range higher than 0.20 Hz, but the median was less than 10 in all frequency ranges.

Fig. 5.

Summary of transfer function gains (TFGs) between heart rate and blood pressure variabilities during and before the transient arterial baroreflex response in the supine and standing positions. TFGs during and before the transient baroreflex response are shown in red and black, respectively. The box indicates the 25th and 75th percentiles, the solid horizontal line indicates the median, the x indicates the mean, and the whisker indicates the range of data. The Wilcoxon signed-rank test statistically compared TFGs during and before baroreflex. n = 13 both in the supine (A) and standing (B) positions. Note that the y-axis in graph (B) is magnified fourfold.

Discussion

Difference in arterial baroreflex between the supine position and standing

In this study, we observed the occasional occurrence of an ABR sequence of heart rate and blood pressure changes, namely a transient MAP rise accompanied by IBI decrease followed by a reversal to IBI increase and MAP decrease, both in the supine position and during standing. Concerning the mechanism underlying this sequence of ABR response, the following is suggested: Transient sympathetic activation due to naturally occurring physiological perturbations causes blood pressure rise by peripheral blood vessel constriction through α1-adrenergic receptor activation with a simultaneous heart rate increase through activation of β1-adrenergic receptor mainly of the sinoatrial node [20]. This blood pressure increases and IBI decreases then reverse due to ABR negative feedback regulation through cardiac vagal nerve activation and sympathetic withdrawal. Although this response sequence was similar in both the supine and standing positions, spectral and cross-spectral analyses revealed a large difference in the negative feedback baroreflex sensitivity between them. The results of the analyses showed that the TFG during the sequence of the ABR responses was frequency- and posture-dependent: at 0.05 and 0.10 Hz, a small but significant increase in TFG was observed in both the supine position and standing, whereas only in the supine position was a marked TFG increase at 0.15 and 0.20 Hz detected. The results also highlight substantial ABR-related IBI power increases at 0.05, 0.10, 0.15, and 0.20 Hz in the supine position, whereas during standing, the 0.15 and 0.20 Hz power increases were markedly suppressed. Since the HRV spectral power increase in the HF range (>0.15 Hz) is exclusively caused by fluctuations of cardiac vagal nerve activity [1], [2], [3], [4], [5], the present results indicate that the increase in vagal efferent output generating 0.15 and 0.20 Hz IBI powers is crucial in determining the negative feedback baroreflex sensitivity. The vagal activity in this frequency range is suppressed during standing, which reduces the baroreflex action, delaying the recovery of decreased IBI to the original level.

The mechanism underlying the tonic suppression of cardiac vagal activity in the upright posture has been considered primarily to result from sustained unloading of arterial baroreceptors [21]. However, we discuss the possibility that another mechanism may be involved in orthostatic stress-induced suppression of cardiac vagal outflow in the next section.

Interaction between cardiac vagal and sympathetic activities

The prominent decrease in the transient ABR sensitivity with the HF HRV decrease caused by standing may be due partly to the sympathetic restraint of cardiac vagal nerve outflow to the heart [22], [23], [24], [25], [26]. Hedman et al. [22] reported in canines that sympathetic stimulation reduced the magnitude of the RSA-like HF HRV produced by artificial electrical stimulation of the cardiac vagal nerve, suggesting that although the HF component of HRV is mainly under parasympathetic regulation, it may also be under the indirect control of the sympathetic nervous system. Taylor et al. [23] showed in healthy subjects that cardiac sympathetic activation produced by head-up tilt reduced RSA and heart rate oscillations at all frequency ranges, and β-adrenergic blockade abolished this sympathetic activation effect. Pitzalis et al. [24], [25] reported that β-adrenergic blockade increased RSA, HF vagal outflow, and ABR sensitivity. They also concluded that the action site of the β-adrenergic blockade is located in the periphery since the hydrophilic β-blocker, which does not cross the blood-brain barrier, similarly enhanced HF vagal outflow and ABR sensitivity as the lipophilic β-blocker. Furthermore, Cook et al. [26] showed that upright posture enhanced muscle sympathetic nerve activities, correlating with the tilt angle from supine to 90 deg. They also reported a significant ABR sensitivity decrease at 60, 70, and 80 deg compared with supine, 20, and 40 deg tilt levels.

The above findings suggest that the sympathetic nervous system activity, which is too slow to mediate HF heart rate changes, indirectly exerts its tonic action through inhibitory control of vagal cardiac activity [27]. At first glance, this notion contradicts the phenomenon of “accentuated antagonism” that vagal chronotropic effects are augmented with increased sympathetic background activities [28]. However, sympathovagal interactions to govern heart rate are complex and there is much debate about “accentuated antagonism” phenomenon and its underlying mechanisms involving interactions within the cardiac ganglia, at the end terminals of sympathetic and vagal nerves, and through second-messenger systems in the innervated sinoatrial cells. Salata et al. [29] examined the effects of stellate ganglia stimulation on the phase-dependent chronotropic and dromotropic responses to brief vagal stimulation in canines. They found no significant sympathetic-parasympathetic interaction for heart rate, indicating that accentuated antagonism did not occur with brief vagal bursts. In contrast, stellate stimulation decreased the phasic vagal effects on atrioventricular nodal conduction. Stuesse et al. [30] reported that sympathetic stimulation shifted the temporal relationship between vagal stimulation and pacemaker response, and small changes in sympathetic tone greatly altered the cardiac response to phasic vagal stimulation when the vagal stimulus was given at certain times in the cardiac cycle. More work is needed to elucidate detailed mechanisms underlying complex sympathovagal interactions in regulating cardiac functions.

Short-term spectral analysis for estimating the frequency-dependent transfer function gain of an individual ABR response

After Robbe et al. [15] quantified baroreflex sensitivity as modulus (or baroreceptor reflex gain) using spectral and cross-spectral analysis, this figure became widely used as an index of autonomic regulation of the cardiovascular system. The technique based on spectral analysis is simpler, noninvasive, and less costly than the classical phenylephrine method. It allows detailed assessment of the relationship between baroreflex function and cardiovascular modulation during daily activities. TFG decreases with exercise and mental stresses, but increases with physical training to prevent hypertension [15], [16], [31]. Moreover, many clinical studies demonstrated TFG as an early index of dysfunction of the cardiovascular system's autonomic control for a variety of cardiovascular diseases, including myocardial infarction and heart failure [31], [32], [33], [34].

There are some differences in the computational algorithms of spectral indexes in the calculation of TFG. According to the transfer function method proposed by Robbe et al. [15], ABR sensitivity is computed as the average value of modulus between systolic pressure and RRI in the frequency range 0.07–0.14 Hz. Pagani et al. computed two TFGs, one in the LF and the other in the HF bands, as the square root of the ratio between RRI and systolic pressure spectral components, referring them to as α-LF and α-HF, respectively [16]. Pinna et al. [33] considered that LF oscillations of IBI mainly represent the vagal transmission of blood pressure oscillations due to the baroreflex and computed the average transfer function between IBI and arterial pressure variabilities in the LF band (0.04–0.15 Hz). In these three groups, the data segment length of heart rate and blood pressure variabilities to be analyzed was longer than 3 min [15], [16], [33].

The present method of calculating TFG differs from the above methods in the following two respects: Firstly, we analyzed a 30-s data segment that contained only one transient baroreflex response sequence, which sporadically occurred and was completed within 10 s in the supine position. Secondly, TFGs for the LF or HF-frequency bands were not averaged but estimated separately for each frequency. This highlighted the increase in HRV power at 0.15 and 0.20 Hz as most responsible for the marked increase in TFG during the transient baroreflex response in the supine position. Moreover, it is conceivable that in the previous methods, with fewer baroreflex events per data segment, TFG values would decrease accordingly. This implies that the calculated baroreflex gain over a specific time period varies, depending both on the individual gain amplitude and on the number of baroreflex responses contained in that period. In contrast, the present analysis clearly indicates that the decrease in TFG under orthostatic stress is due to a reduction in individual baroreflex gain.

Conclusion

Adopting the time-frequency analysis, which enabled the identification of the temporal relationship between sporadically occurring sequences of transient heart rate changes and resulting spectral power increases, we attempted to clarify differences in the ABR response during these sequences of heart rate and spectral power changes between the supine and standing positions. We found that the ABR gain was significantly greater in the supine position compared to standing, and this gain increase in the supine position was linked to increases in 0.15 to 0.20 Hz components of HRV spectral powers due to activation of the cardiac vagal outflow. Based on these findings, we conclude that the orthostatic stress induced by the postural change from supine to standing markedly reduces the baroreflex gain by suppressing high-frequency vagal outflow.

Funding sources

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CRediT authorship contribution statement

Sakamoto Teruhiko: Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Conceptualization. Mitsuyama Satoshi: Writing – original draft, Validation, Software, Methodology, Investigation, Formal analysis, Conceptualization. Nagasawa Toru: Writing – review & editing, Software, Formal analysis. Kario Kazuomi: Writing – review & editing, Investigation. Ozawa Seiji: Writing – original draft, Supervision, Project administration, Methodology, Investigation, Formal analysis, Conceptualization.

Author contributions

S.O., T.S., and S.M designed the study. T.S., S.M., and S.O. performed the experiments. T.S., S.M., S.O., and T.N. analyzed the data. S.O., T.S., and S.M. drafted the manuscript. T.N., and K.K. edited and revised the manuscript.

Declaration of Competing Interest

The authors declare no conflicts of interest associated with this study.

Data availability

Data will be made available on reasonable request.