Tilt-evoked, breathing-driven blood pressure oscillations: Independence from baroreflex-sympathoneural function

All the participants in this study gave written informed consent before any research procedures were conducted. The protocol was approved by the institutional review board of the National Institute of Neurological Disorders and Stroke (NINDS) or the National Institutes of Health (NIH). All the patients had been referred for evaluation by the Autonomic Medicine Section (formerly Clinical Neurocardiology Section) of the Division of Intramural Research of the NINDS at the NIH Clinical Center.

All the patients had past medical records reviewed for signs and symptoms of autonomic dysfunction or a central movement disorder, underwent screening autonomic function testing at the NIH Clinical Center that confirmed the referral diagnosis [11, 29], and had additional clinical laboratory testing that supported and refined the findings from the screening examination. ¹⁸F-DOPA positron emission tomography (PET) was used to identify striatal dopamine deficiency [16], as typically occurs in patients with PD (with or without OH) and patients with MSA-P [16] but not in patients with PAF [18]. MSA-P was distinguished from PD+OH by ¹⁸F-dopamine PET [31]. In the VHL cohort, patients were recruited through consultation after neurosurgery for brainstem hemangioblastomas. In the control cohort the HVs had unremarkable medical histories and physical examinations and results of screening clinical laboratory tests done at the NIH Clinical Center within 6 months of research participation.

All the subjects in the neurogenic OH cohort satisfied criteria for vascular-sympathetic baroreflex failure on the basis of an algorithm published previously by our group [22]. Briefly, the patients had persistent, consistent OH without a known secondary cause such as medications or diabetes mellitus; had abnormal BP responses in both Phase II and Phase III/IV of the Valsalva maneuver [21, 23, 34]; had prolonged pressure recovery times after release of the maneuver [12, 34]; and in most cases had attenuated plasma norepinephrine responses to head-up tilt [19].

The autonomic function testing was done in a dedicated Patient Observation Room operated by the Autonomic Medicine Section at the NIH Clinical Center. After urinating to empty the bladder, the participant lay supine with head on pillow on a motorized tilt table. An intravenous (IV) catheter was placed in an arm vein, usually the left antecubital vein, and attached by a 3-way stopcock and connection tubing to a plastic bag containing normal saline that was infused continuously at a slow rate to keep the vein open.

Physiological data were recorded using LabChart Pro 8 running a 16-channel PowerLab electronic physiological recorder (ADInstruments, Colorado Springs, CO). Electrocardiographic leads were attached to the skin and connected by a harness to the PowerLab. A noninvasive automated finger cuff system was used for continuous BP recording [BMEYE Nexfin, BMEYE B.V., Amsterdam, The Netherlands) or Finapres Nova (Finapres B.V., Amsterdam, the Netherlands)]. Respiration was monitored using a Respitrace plethysmography device placed around the upper abdomen or lower chest and connected to the PowerLab (ADInstruments, Colorado Springs, CO).

Physiological tests included slow, deep breathing at about 6 breaths per min, the Valsalva maneuver (30 mmHg for 12 s, ≥ 3 times until a technically adequate tracing was obtained [12, 21], with 20-degree head-up tilt in the event of a “flat top” BP pattern), and up to 5′ of head-up tilt at 90 degrees from horizontal. In the HVs head-up tilt could also be done at 70 degrees from horizontal for up to 40 min [3, 15].

Blood was drawn through the indwelling IV catheter into heparinized sample tubes during supine rest and at 5′ of head-up tilt (less than 5′ if the tilting was ended sooner for patient safety reasons). The plasma was separated by refrigerated centrifugation and transferred to plastic cryotubes for storage in an ultra-low temperature freezer. Freshly thawed samples were assayed for catechols by batch alumina extraction followed by liquid chromatography with series electrochemical detection as described previously [24].

Systolic BP data in the control, tilt, and recovery periods (Pre-tilt, Tilt, and Recovery) were analyzed using a cubic spline. Each segment was then linearly detrended. Power spectral densities of systolic BP were estimated using Welch’s method of averaged periodograms (300-point Hamming windows with 150-point overlap), using Matlab (MathWorks, Natick, MA). Spectral powers in the low-frequency (LF; 0.04–0.15 Hz) and high-frequency (HF; 0.15–0.4 Hz) bins were obtained using trapezoidal integration over the specified frequency range.

Physiological recordings were shortened during processing to undergo spectral analysis, and data were selected for the periods 5′ prior to tilt (Pre-tilt), 5′ during tilt (Tilt) if the patient withstood being upright for that long, and 5′ immediately upon return to horizontal (Recovery). In the few cases where patients did not tolerate 5′ of 90-degree tilt, a minimum of 1.5′ was used for power spectral analysis of the data during the tilt phase.

GraphPad Prism 9 (GraphPad Software, LLC) was used for statistical analyses and graphics.

Cohorts were grouped on the basis of the diagnosis assigned for their disorder after comprehensive testing and the presence or absence of OH. Because of substantial interindividual variability that depended on the mean values, the power data were log transformed for statistical testing. Absolute differences in log-transformed data represent proportionate changes.

Cardio-vagal baroreflex gain was calculated from the slope of the relationship between the cardiac interbeat interval (with 1-beat delay) and systolic BP between the highest BP value in Phase I and the lowest BP value during Phase II of the Valsalva maneuver [21].

The standard deviations (SD) of systolic BP in the Baseline, Tilt, and Recovery periods were used as measures of BP variability in the time domain.

Two-way analyses of variance (ANOVAs) were performed to compare groups across the Baseline, Tilt, and Recovery conditions using the tilt condition as the grouping factor for each cohort. Logarithmic transformations allowed for the data to meet necessary assumptions about homogeneity of variance for ANOVA testing. Repeated measures ANOVAs were used to examine differences among tilt conditions across individual groups. F-values were expressed as F (between groups df, within groups df), and effect sizes as partial eta squared. Dunnett’s post-hoc test was used for multiple comparisons of other cohorts vs. the HV group or Tilt vs. Baseline as appropriate. Mixed-effects analyses of power spectral data were carried out when there were missing data points using the same grouping factors as the two-way ANOVA.

To compare the control vs. neurogenic OH groups, independent means t-tests were used. To analyze changes from Pre-tilt to Tilt, pairwise comparison (dependent means) t-tests were used. A p-value less than 0.05 defined statistical significance.

Across all subjects, mixed-effects analysis showed that HF power varied across tilting conditions [F (1.552, 111.0) = 11.45, p < 0.001, partial eta squared = 0.1404, df = 72 for Baseline versus Tilt, and df = 71 for Baseline versus Recovery]. Based on Dunnett’s multiple comparisons test vs. the baseline condition, HF power increased between the Baseline and Tilt conditions (p = 0.002), while Recovery did not differ from Baseline (p = 0.180). Normalized HF power (HFnu) also varied as a function of tilting condition [F (1.960, 140.1) = 4.5, p = 0.010, partial eta squared = 0.0551, df = 72 for Baseline versus Tilt, and df = 71 for Baseline vs. Recovery], with both Tilt and Recovery having increased power from baseline (p = 0.020, p = 0.010). The patient groups did not differ from the HV group in HF power during any of the 3 conditions.

Comparing Tilt vs. Baseline by pairwise t-tests, head-up tilting increased the log of HF power across all subjects (p = 0.001, partial eta squared = 0.1457; Fig. 1A). Within all groups with the exception of PD No OH (p = 0.030, partial eta squared = 0.2934), there were no significant increases in the log of HF power between the Baseline and Tilt conditions (Fig. 1B–F).

Across all subjects, LF power varied as a function of tilting condition [F (1.668, 119.3) = 4.6, p = 0.020, partial eta squared = 0.0618, df = 72 for Baseline vs. Tilt, and df = 71 for Baseline vs. Recovery]. Based on Dunnett’s post-hoc test, LF power during Tilt did not differ from that during Pre-tilt. Among HVs, LF varied with tilting condition [F (1.131, 16.96) = 4.1, p = 0.050], but neither Tilt nor Recovery differed from baseline. In the control cohort without OH (HV and PD No OH) mean LF power increased during tilt (p = 0.039), whereas patient groups with OH showed no significant changes.

Upon tilt, the HV group had significantly greater LF power than did the patient groups, which had highly variable responses and no significant overall changes. The HV group was statistically significantly different from all the patient groups (p = 0.002) and had higher LF power. On the basis of Dunnett’s post-hoc test comparing data in the patient groups with the HV group, statistically significant differences were found between HV and PAF (p = 0.002), PD+OH (p = 0.017), PD No OH (p = 0.016), MSA-P (p = 0.018), and VHL (p = 0.015).

Comparing Tilt vs. Baseline by pairwise t-tests, head-up tilting did not increase the log of LF power across all subjects (Fig. 2). The only group with a significant increase in the log of LF power during Tilt was the HV group (Fig. 2B).

In both the control and nOH cohorts the log of HF power increased between the Pre-tilt and Tilt conditions (Fig. 3A, B). The two cohorts did not differ in either the mean log of HF power during Tilt or the mean increment in the log of HF power from Pre-tilt to Tilt (Fig. 3C, D).

The log of LF power increased between the Pre-tilt and Tilt conditions in the control but not in the nOH cohort (Fig. 3E, F). The control cohort had a higher mean log of LF power during Tilt and a larger mean increment in the log of LF power from Pre-tilt to Tilt than did the nOH cohort (Fig. 3G, H).

The nOH cohort had lower mean cardio-vagal baroreflex gain (1.39 ms/mmHg) than did the control cohort (4.65 ms/mmHg, t = 5.658, p < 0.001, df = 59). Individual values for the log of HF power during Tilt were unrelated to the cardio-vagal baroreflex gain across all subjects (r = −0.01), in the nOH cohort (r = −0.13), and in the control cohort (r = 0.27).

In both the nOH and control cohorts SD BP increased between Pre-tilt and Tilt (for nOH, t = 3.26, p = 0.002; for control, t = 2.96, p = 0.006). Across all subjects, individual values for SD BP during Tilt tended to be negatively correlated with the log of the cardio-vagal baroreflex gain (r = −0.2337, p = 0.070). Pre-tilt SD BP was unrelated to the log of the cardio-vagal baroreflex gain.

Figure 4 shows a recording from a HV, and Figs. 5 and 6 show recordings from patients with nOH. In Fig. 4 the HV shows no evidence of breathing-driven BP oscillations, while low-frequency Mayer waves are evident. Figure 5 is a recording from a patient with PAF evolving to PD+OH, and Fig. 6 is from a patient with MSA-P. Both nOH patients have tilt-evoked, breathing-driven BP oscillations. It should be noted that such clear evidence was unusual, and many nOH patients did not have obvious increases in power of BP variability at the respiratory frequency during Tilt.

There was marked individual variability in responses of both HF and LF power to tilting in all the groups in this study, and the durations of recording periods were relatively short, further limiting data reliability.

If tilt-evoked Traube–Hering waves were purely mechanical, respiratory amplitude and venous return to the heart would be expected to play important roles in the BP oscillations, but neither variable was tracked. Individuals probably vary in breathing responses to tilting, yet neither the rate nor depth of respiration was controlled. It was evident that some patients had slower breathing during tilting than at Baseline or during Recovery, meaning that part of the breathing-driven BP oscillations during tilting were in the LF range. Effects of orthostatic changes on respiratory frequency would have influenced the variability of power data calculated from within predefined frequency bins. In future studies on this topic individual variability might be reduced by having subjects perform paced breathing during orthostasis. There may also have been large individual differences in the effects of orthostasis on venous return to the heart and consequently on cardiac stroke volume and pulse pressure. Although the automated finger cuff systems we used for tracking BP continuously came with software applications that reported estimated values for beat-to-beat stroke volume, to our knowledge the algorithms have not been validated in nOH. Another possible complicating factor is the chemoreflex [32], but capnography was not performed.