The effect of oscillatory hemodynamics on the cardiovascular responses to simulated hemorrhage during isocapnia

Abstract

During cerebral hypoperfusion induced by lower body negative pressure (LBNP), cerebral tissue oxygenation is protected with oscillatory arterial pressure and cerebral blood flow at low frequencies (0.1 Hz and 0.05 Hz), despite no protection of cerebral blood flow or oxygen delivery. However, hypocapnia induced by LBNP contributes to cerebral blood flow reductions, and may mask potential protective effects of hemodynamic oscillations on cerebral blood flow. We hypothesized that under isocapnic conditions, forced oscillations of arterial pressure and blood flow at 0.1 Hz and 0.05 Hz would attenuate reductions in extra- and intracranial blood flow during simulated hemorrhage using LBNP. Eleven human participants underwent three LBNP profiles: a nonoscillatory condition (0 Hz) and two oscillatory conditions (0.1 Hz and 0.05 Hz). End-tidal (et) CO₂ and etO₂ were clamped at baseline values using dynamic end-tidal forcing. Cerebral tissue oxygenation (ScO₂), internal carotid artery (ICA) blood flow, and middle cerebral artery velocity (MCAv) were measured. With clamped etCO₂, neither ICA blood flow (ANOVA P = 0.93) nor MCAv (ANOVA P = 0.36) decreased with LBNP, and these responses did not differ between the three profiles (ICA blood flow: 0 Hz: 2.2 ± 5.4%, 0.1 Hz: −0.4 ± 6.6%, 0.05 Hz: 0.2 ± 4.8%; P = 0.56; MCAv: 0 Hz: −2.3 ± 7.8%, 0.1 Hz: −1.3 ± 6.1%, 0.05 Hz: −3.1 ± 5.0%; P = 0.87). Similarly, ScO₂ did not decrease with LBNP (ANOVA P = 0.21) nor differ between the three profiles (0 Hz: −2.6 ± 3.3%, 0.1 Hz: −1.6 ± 1.5%, 0.05 Hz: −0.2 ± 2.8%; P = 0.13). Contrary to our hypothesis, cerebral blood flow and tissue oxygenation were protected during LBNP with isocapnia, regardless of whether hemodynamic oscillations were induced.

NEW & NOTEWORTHY We examined the role of forcing oscillations in arterial pressure and blood flow at 0.1 Hz and 0.05 Hz on extra- and intracranial blood flow and cerebral tissue oxygenation during simulated hemorrhage (using lower body negative pressure, LBNP) under isocapnic conditions. Contrary to our hypothesis, both cerebral blood flow and cerebral tissue oxygenation were completely protected during simulated hemorrhage with isocapnia, regardless of whether oscillations in arterial pressure and cerebral blood flow were induced. These findings highlight the protective effect of preventing hypocapnia on cerebral blood flow under simulated hemorrhage conditions.

INTRODUCTION

Hemorrhage, or massive blood loss, challenges vital organ perfusion and oxygenation, and can result in organ damage or death if left untreated. Hemorrhage is the leading cause of preventable deaths in the battlefield (1), and accounts for over 61,000 deaths per year in the United States of America, and 1.9 million deaths per year worldwide (2). Because hemorrhage is common in younger populations, it accounts for over 85.6 million years of life lost per year worldwide (2). These statistics highlight the need for improved therapies to treat hemorrhagic injuries.

Lower body negative pressure (LBNP) has been validated as a model of actual blood loss in animals (3) and humans (4, 5), and provides the benefit of inducing the compensatory cardiovascular responses to hemorrhage without removal of blood volume. Using this technique, physiological mechanisms that affect tolerance to blood loss have been explored (6–8). In 2011, Rickards et al. (9) highlighted low-frequency oscillations (around 0.1 Hz) in arterial pressure and cerebral blood flow as a potential mechanism influencing tolerance to simulated hemorrhage induced by LBNP.

Following these initial observations, our laboratory examined the effects of forcing low-frequency oscillations in arterial pressure and cerebral blood flow on both tolerance to simulated hemorrhage (via LBNP), and on cerebral tissue oxygenation under conditions of both normoxia (10) and hypoxia (11). We demonstrated that oscillating arterial pressure and cerebral blood flow at both 0.1 Hz and 0.05 Hz protected against reductions in cerebral tissue oxygenation during LBNP, and improved tolerance to this stress (10, 11). This effect also occurred without a protection of cerebral blood flow [via middle cerebral blood velocity (MCAv) or internal carotid artery (ICA) blood flow].

Cerebral blood flow during LBNP is primarily regulated by arterial pressure and arterial Pco₂. In our previous studies exploring forced hemodynamic oscillations, arterial pressure and end-tidal CO₂ (etCO₂; an index of arterial Pco₂) decreased under all conditions, with no differences in the magnitude of decrease between conditions (10, 11). Of these parameters, etCO₂ is arguably the most potent regulator of cerebral blood flow (e.g., see Ref. 12), and can also be experimentally controlled during LBNP. In a study examining the effect of clamping etCO₂ at baseline values (isocapnia) on the cerebral blood flow response to presyncopal LBNP (13), cerebral blood flow decreased during the isocapnic trial, but to a lesser magnitude than during the poikilocapnic (hypocapnia with free breathing) trial; there was no difference in LBNP tolerance between trials. As reductions in $\text{P}{\text{a}}_{\text{C}{\text{O}}_2}$ contribute to the reduction in cerebral blood flow during LBNP, the aim of the current study was to control this hypocapnic response to assess whether hemodynamic oscillations protect cerebral blood flow without this additional stimulus. Interestingly, etCO₂ also oscillated at the same frequency as oscillatory LBNP in our previous study (10), which likely contributed to the magnitude of oscillations in cerebral blood flow. By clamping etCO₂ at baseline values, this effect can also be removed and improve our ability to assess the direct effect of hemodynamic oscillations on cardiovascular and cerebrovascular responses to simulated hemorrhage.

In addition, as hemodynamic oscillations may increase and decrease shear stress, the production of endothelial-derived vasoactive mediators (such as nitric oxide and endothelin-1) could be altered, and play a role in the vascular response to these oscillations. LBNP itself could also induce release of these mediators; the central hypovolemic stimulus of head-up tilt increases circulating endothelin-1 in healthy participants (14). Exploring the balance between vasodilatory (via nitric oxide) and vasoconstrictive (via endothelin-1) mediator release may aid in understanding the cardiovascular responses to simulated hemorrhage with and without forced oscillations in arterial pressure and blood flow.

In the present study, we used end-tidal forcing (15) to clamp etCO₂ at baseline resting values during LBNP with and without forced hemodynamic oscillations. We hypothesized that forcing oscillations in arterial pressure and cerebral blood flow during simulated hemorrhage, while also clamping etCO₂ at eucapnic baseline values, would increase cerebral blood flow compared with the nonoscillatory condition. We also hypothesized that circulating concentrations of nitric oxide would increase and/or endothelin-1 would decrease during simulated hemorrhage with forced oscillations in arterial pressure and blood flow, further contributing to the protection of cerebral blood flow.

METHODS

Ethical Approval

Ethical approval for this study was received from the North Texas Regional Institutional Review Board (Protocol No. 2019-046), and all experimentation was performed at the University of North Texas Health Science Center.

Participants

Healthy human participants between ages 18 and 40 yr were recruited for this study. Participants were briefed on the purpose, design, and risks of the study, and provided written, informed consent before inclusion. Prior to experimentation, participants completed a medical screening and familiarization session, and were excluded if they had any history of cardiovascular, respiratory, or neurological disease. Female participants completed a urine pregnancy test before each experiment to ensure they were not pregnant. Female participants were also tested during days 1–4 of their menstrual cycle (self-report) to account for hormonal fluctuations.

Due to potential confounding effects on cardiovascular regulation, participants abstained from alcohol, caffeine, exercise, dietary supplements, and medication for 24 h, and from food for at least 2 h before each experiment. All experiments were conducted in a temperature-controlled laboratory (22–24°C), and in the morning to control for the effect of circadian rhythm on cardiovascular responses.

Instrumentation

Upon arrival for each experiment, the height, weight, and age of participants were recorded. Participants were positioned in the LBNP chamber in the supine posture with their iliac crest in line with the chamber opening. A heavy-duty plastic liner and neoprene waist wrap were then used to create an air-tight seal between the participant and the chamber. ECG electrodes were placed in a lead-II configuration for continuous recording and subsequent R-wave detection (shielded leads, cable, and amplifier, AD Instruments, Bella Vista, NSW, Australia). Finger photoplethysmography was used for continuous recording of arterial pressure (Finometer, Finapres Medical Systems, Amsterdam, The Netherlands). ModelFlow estimates of stroke volume were corrected by measuring aortic diameter at rest (average of two measures; 4 MHz 3Sc probe, GE Vivid T8, Chicago, IL). Cerebral blood velocities in a middle cerebral artery (MCAv) and posterior cerebral artery (PCAv) were measured through the right and left temporal windows using 2 MHz probes fastened to an adjustable head frame (ST3, Spencer Technologies, Seattle, WA). Oxy-hemoglobin (Hb) and deoxy-Hb were measured, and cerebral tissue oxygenation was calculated (ScO₂; [oxy-Hb/(oxy-Hb + deoxy-Hb) × 100]) using a near-infrared spectroscopy probe (NIRS; OxiplexTS, ISS, Champaign-Urbana, IL) that was placed over the right frontal lobe. For this device, the emitters are spaced ∼20, 25, 30, and 35 mm from the detector, and the sampling frequency is 50 Hz. ICA velocity and diameter measurements were made using a 11 MHz linear array probe with an insonation angle of 60° (Phillips iE33, Andover, MA). Video recordings were captured at 30 Hz using an Epiphan capture card (DVI2USB 3.0, Epiphan Systems, Inc., Palo Alto, CA) and screen recording software (Camtasia, TechSmith Corporation, Okemos, MI) for later analysis. Respiratory gases (etCO₂; end-tidal O₂, etO₂) were captured using a mouthpiece and a gas-sampling line connected to a gas analyzer (ML206 Gas Analyzer, ADInstruments, Bella Vista, NSW, Australia). Respiratory volumes were also measured using a pneumotachograph (Model 4830, 0-400 LPM, Hans Rudolph, Shawnee, KS) connected to a bacteriological filter and a two-way nonrebreathing valve (model 7900 series, Hans Rudolph, Shawnee, KS). The respiratory gas and volume signals were subsequently routed into a dynamic end-tidal forcing system for the clamping of etCO₂ and etO₂. This system controls breath-by-breath inspired fractions of O₂ and CO₂ to control end-tidal gas values independent of ventilation (15). Finally, a 20-gauge flexible venous catheter was inserted in a vein within the antecubital fossa of the right arm for repeated venous blood sampling.

Protocol

After instrumentation, a 5-min baseline period was completed to measure resting cardiovascular status, and to collect resting measures of etCO₂ and etO₂, which were subsequently entered into the dynamic end-tidal forcing software. The end-tidal forcing system was then connected to the inhalation port of the two-way nonrebreathing valve to clamp each individual participant’s end-tidal gases to their resting values before the initiation of LBNP. Once clamping was achieved (within 1–2 min), participants then underwent one of three randomized LBNP profiles, each separated by at least 48 h: 1) a 0 Hz condition where chamber pressure was gradually reduced to −60 mmHg over 30 s and subsequently held constant for an additional 10.5 min, 2) a 0.1 Hz condition where chamber pressure was gradually reduced to −60 mmHg over 30 s, held for another 30 s and then oscillated between −30 and −90 mmHg for 5 s at each pressure over 10 min, and 3) a 0.05 Hz condition where chamber pressure was gradually reduced to −60 mmHg over 30 s, held for another 30 s and then oscillated between −30 and −90 mmHg for 10 s at each pressure over 10 min. See Fig. 1 for representative tracings of each profile. These profiles were terminated early if participants reached presyncope, which we defined as a systolic arterial pressure <80 mmHg, or onset of subjective symptoms such as nausea, lightheadedness, dizziness, or visual disturbances (such as gray out or tunnel vision). After release of chamber pressure, the dynamic end-tidal forcing system was removed, and participants rested for 10 min of recovery.

Figure 1.

Representative hemodynamic responses to each lower body negative pressure (LBNP) protocol. Mean arterial pressure (MAP) and mean middle cerebral artery velocity (MCAv) responses to 0 Hz LBNP (left), 0.1 Hz oscillatory LBNP (middle), and 0.05 Hz oscillatory LBNP (right).

Blood Sample Collection and Analysis

For all three LBNP profiles, 5 mL venous blood samples were drawn at rest and at the end of LBNP. Blood samples were placed in a 6-mL EDTA collection tube and centrifuged at 2,000 g for 15 min at 4°C. Plasma was immediately extracted into microcentrifuge tubes, snap-frozen in liquid nitrogen, then placed in a −80°C freezer for storage. Plasma nitrite was analyzed at the UNTHSC Pharmaceutical Analysis Core Laboratory using liquid chromatography with tandem mass spectrometry following a sample preparation optimized for human samples (16). Plasma concentrations of endothelin-1 were analyzed in our laboratory using a commercially available enzyme-linked immunosorbent assay (ELISA) in accordance with the manufacturer’s instructions (DET-100, R&D Systems, Minneapolis, MN).

Data Analysis

Continuous recordings of ECG, arterial pressure, stroke volume, LBNP chamber pressure, MCAv, PCAv, and respiratory gases and volumes were collected at 1,000 Hz using PowerLab and LabChart 8 (AD Instruments, Bella Vista, NSW, Australia). These data were subsequently imported into and analyzed using a commercially available analysis package (Ensemble, Elucimed, Wellington, New Zealand). R-wave detection was used to identify cardiac cycles, facilitating calculation of beat-to-beat metrics of mean arterial pressure, mean MCAv, and mean PCAv. End-tidal values for CO₂ and O₂ were extracted from the continuous breath-to-breath waveforms of Pco₂ and Po₂. The last 3 min of baseline and last 1 min of LBNP were used to calculate time domain averages of these signals. Power spectral density for mean arterial pressure, MCAv, and PCAv was calculated following spline interpolation of beat-to-beat measures and resampling at 4 Hz. The Welch algorithm and a Hanning window were used for fast Fourier transform. Power spectral density within the low-frequency range (LF, 0.07–0.15 Hz), the very-low-frequency range (VLF, 0.02–0.07 Hz), and at exact frequencies of 0.1 Hz and 0.05 Hz were examined for the entire 5 min of baseline and the last 5 min of LBNP.

ICA diameter, velocity, and shear rate measures were extracted and down sampled to 1-s averages using specialized wall tracking software (Cardiovascular Suite, Quipu, Pisa, Italy). Current standards for measurement of ICA velocity and diameter were followed, including matching anatomical location between experiments, measuring velocities 1.5–2 cm distal from the bifurcation, and collecting a minimum of 10 cardiac cycles of data for averaging (17). ICA blood flow was calculated as $\text{π}{\left(\frac{\text{diameter}}{2}\right)}^2\times \left(\frac{\text{peak\hspace{0.17em}velocity}}{2}\right)\times 60.$ MCA conductance index and ICA conductance were calculated by dividing MCAv or ICA blood flow by mean arterial pressure. MCA resistance index and ICA resistance were calculated by dividing mean arterial pressure by MCAv or ICA blood flow.

Statistical Analysis

Both absolute and percent change from baseline data are presented. A two-way linear mixed model for repeated measures was used to analyze the absolute data, with LBNP profile (0 Hz, 0.1 Hz, 0.05 Hz) and time (baseline, LBNP) as the two factors. Post hoc tests were chosen a priori to compare LBNP profiles at both baseline and during LBNP, and between baseline and LBNP within profile. These pair-wise comparisons were corrected for multiple comparisons with the Holm correction. A one-way linear mixed model was used to analyze the percent change from baseline across LBNP profiles (0 Hz, 0.1 Hz, 0.05 Hz), followed by Tukey post hoc tests. All statistical analysis was performed in R (R Core Team, 2021) with lme4 package (18) for linear mixed model analysis.

RESULTS

Participants

Eleven participants (9 males, 2 females; 29 ± 5 yr, 172 ± 9 cm, 78 ± 18 kg) completed the 0 Hz and 0.1 Hz profiles, and 9 participants (8 males, 1 female) completed the 0.05 Hz profile. Due to a signal recording error, one stroke volume recording was lost during the 0.1 Hz profile (n = 10) and two recordings were lost during the 0.05 Hz profile (n = 7). Difficulties in signal acquisition (due to the dynamic movement of the participants during oscillatory LBNP, and/or respiratory-mediated vessel movement) also resulted in a reduced number of recordings for ICA blood flow during the 0.05 Hz profile (n = 7); MCAv during the 0 Hz (n = 9), 0.1 Hz (n = 9), and 0.05 Hz (n = 6) profiles; and PCAv during the 0 Hz (n = 8), 0.1 Hz (n = 5), and 0.05 Hz (n = 6) profiles.

Experimental Design

LBNP.

The average LBNP chamber pressure was similar (by design) during each of the three profiles (0 Hz, −62.3 ± 0.5 mmHg; 0.1 Hz, −58.3 ± 0.7 mmHg; 0.05 Hz, −59.2 ± 0.6 mmHg). Due to onset of presyncopal symptoms, 4 of the 11 (36%) participants did not complete the entire 0 Hz profile, whereas all 11 participants completed the 0.1 Hz profile, and all nine participants tested completed the 0.05 Hz profile.

End-tidal forcing.

We also successfully clamped both etCO₂ and etO₂ (Fig. 2) at resting baseline values for all three protocols.

Figure 2.

Respiratory gas responses to end-tidal forcing. Absolute responses are shown in A and C, and relative (% Δ from baseline) data are show in B and D. etCO₂, end-tidal CO₂; etO₂, end-tidal O₂. Absolute data were analyzed using a two-way linear mixed model, followed by a priori planned post hoc tests between profiles (0 Hz vs. 0.1 Hz vs. 0.05 Hz) at baseline and lower body negative pressure (LBNP), and within profile (0 Hz vs. 0.1 Hz vs. 0.05 Hz) between baseline and LBNP, with a Holm correction for multiple comparisons. Percent change from baseline data were analyzed using a one-way linear mixed model followed by Tukey post hoc tests. Data are presented as means ± SD. Exact P values are presented.

Systemic Cardiovascular Responses

Stroke volume decreased during exposure to all three LBNP profiles (Table 1); however, this reduction was attenuated in both the 0.1 Hz and 0.05 Hz profiles (Fig. 3A). Both pulse pressure and mean arterial pressure responses directly reflect the stroke volume response (Fig. 3, C and E). The subsequent heart rate response was increased above baseline for all profiles (Table 1), but was increased to a greater extent in the 0 Hz profile compared with the oscillatory profiles (Fig. 3B). The compensatory increases in heart rate in response to the decreases in stroke volume resulted in similar cardiac output responses for all three profiles (Table 1; Fig. 3D). Overall, systemic vascular resistance did not change across the three profiles (Table 1), but in relative terms, the increase in systemic vascular resistance was greater during the 0.1 Hz profile compared with the 0 Hz profile (Fig. 3F).

Table 1.

Cardiovascular and cerebrovascular responses to oscillatory LBNP

		Time
	Profile	Baseline	Late LBNP	Profile	Time	Profile × Time	%Δ from Baseline	P Value
Cardiovascular responses
Stroke volume, mL	0 Hz	76.4 ± 15.5	41.3 ± 9.1*	<0.001	<0.001	0.07	−45.5 ± 10.0	<0.001
	0.1 Hz	80.7 ± 14.4	55.2 ± 9.6*†				−31.4 ± 7.2†
	0.05 Hz	82.6 ± 18.9	59.0 ± 15.5*†				−28.6 ± 7.6†
Heart rate, beats/min	0 Hz	66.1 ± 9.7	105.8 ± 16.7*	<0.001	<0.001	<0.001	61.1 ± 21.0	<0.001
	0.1 Hz	66.7 ± 8.8	83.0 ± 10.7*†				24.7 ± 9.1†
	0.05 Hz	65.1 ± 7.8	82.1 ± 14.8*†				26.2 ± 17.2†
Pulse pressure, mmHg	0 Hz	56.9 ± 5.3	37.7 ± 7.4*	0.003	<0.001	0.03	−33.7 ± 11.2	0.003
	0.1 Hz	59.3 ± 7.2	45.8 ± 9.9*†				−23.3 ± 10.4†
	0.05 Hz	56.3 ± 9.5	45.7 ± 8.7*†				−19.0 ± 6.1†
Mean arterial pressure, mmHg	0 Hz	98.5 ± 7.1	85.0 ± 10.4*	<0.001	<0.001	0.10	−13.7 ± 8.7	0.01
	0.1 Hz	104.3 ± 6.4†	97.7 ± 9.9*†				−6.4 ± 6.6†
	0.05 Hz	98.7 ± 8.4	90.4 ± 10.7*†§				−8.5 ± 5.9†
Cardiac output, L/min	0 Hz	5.1 ± 1.4	4.3 ± 1.1*	0.12	<0.001	0.81	−12.9 ± 15.4	0.91
	0.1 Hz	5.3 ± 0.9	4.6 ± 0.8*				−13.3 ± 12.9
	0.05 Hz	5.4 ± 1.3	4.9 ± 1.2				−9.7 ± 6.3
Systemic vascular resistance, mmHg/L/min	0 Hz	20.8 ± 5.8	21.0 ± 6.9	0.78	0.20	0.66	1.5 ± 18.0	0.05
	0.1 Hz	20.0 ± 4.2	22.0 ± 5.6				9.5 ± 15.1†
	0.05 Hz	19.3 ± 5.7	20.3 ± 7.8				3.7 ± 11.1
Cerebrovascular responses
Peak ICA velocity, cm/s	0 Hz	43.1 ± 4.9	44.1 ± 5.4	0.18	0.66	0.89	2.6 ± 7.4	0.78
	0.1 Hz	41.5 ± 5.1	41.6 ± 5.2				0.6 ± 8.8
	0.05 Hz	43.2 ± 4.2	43.3 ± 4.2				0.5 ± 5.8
ICA diameter, cm	0 Hz	0.52 ± 0.05	0.52 ± 0.06	0.68	0.75	0.93	−0.12 ± 2.17	0.90
	0.1 Hz	0.52 ± 0.05	0.52 ± 0.05				−0.43 ± 1.96
	0.05 Hz	0.49 ± 0.03	0.49 ± 0.03				−0.09 ± 2.05
ICA resistance, mmHg/mL/min	0 Hz	0.38 ± 0.11	0.32 ± 0.08*	0.004	0.001	0.35	−15.28 ± 10.15	0.04
	0.1 Hz	0.41 ± 0.08	0.38 ± 0.08†				−5.54 ± 9.19†
	0.05 Hz	0.41 ± 0.09	0.37 ± 0.08				−8.08 ± 5.26
ICA shear rate, s−1	0 Hz	333.2 ± 53.0	347.7 ± 67.4	0.21	0.39	0.80	4.3 ± 10.1	0.74
	0.1 Hz	322.3 ± 51.3	325.9 ± 53.8				1.6 ± 10.7
	0.05 Hz	349.4 ± 18.9	352.5 ± 30.5				1.0 ± 7.7
MCA resistance index, mmHg/cm/s	0 Hz	1.7 ± 0.4	1.4 ± 0.3	<0.001	0.01	0.60	−12.7 ± 5.5	0.05
	0.1 Hz	1.9 ± 0.4	1.8 ± 0.4†				−5.8 ± 8.5
	0.05 Hz	1.7 ± 0.5	1.6 ± 0.4				−5.2 ± 3.8†
PCAv, cm/s	0 Hz	37.7 ± 8.4	35.7 ± 7.2	0.11	0.33	0.83	−4.0 ± 12.7	0.44
	0.1 Hz	43.4 ± 4.7	43.3 ± 4.2				−0.2 ± 5.5
	0.05 Hz	40.3 ± 8.2	38.7 ± 6.9				−3.5 ± 5.3
PCA resistance index, mmHg/cm/s	0 Hz	2.80 ± 0.84	2.40 ± 0.60	0.89	0.04	0.40	−12.97 ± 7.92	0.17
	0.1 Hz	2.45 ± 0.30	2.29 ± 0.27				−6.11 ± 11.73
	0.05 Hz	2.59 ± 0.57	2.49 ± 0.40				−2.72 ± 5.98
PCA CVC index, cm/s/mmHg	0 Hz	0.38 ± 0.10	0.44 ± 0.10	0.82	0.05	0.41	16.17 ± 12.70	0.22
	0.1 Hz	0.41 ± 0.05	0.44 ± 0.05				7.79 ± 14.20
	0.05 Hz	0.40 ± 0.08	0.41 ± 0.06				3.10 ± 6.51
Total cerebral Hb concentration, μM	0 Hz	44.1 ± 9.7	43.7 ± 9.7	0.31	0.94	0.95	−0.8 ± 3.3	0.35
	0.1 Hz	45.6 ± 10.5	45.9 ± 10.4				0.8 ± 3.4
	0.05 Hz	42.7 ± 11.7	43.1 ± 11.6				0.9 ± 2.5
Cerebral oxy Hb, μM	0 Hz	29.8 ± 8.3	29.0 ± 8.5	0.37	0.73	0.90	−3.3 ± 5.9	0.20
	0.1 Hz	30.9 ± 8.7	30.7 ± 8.8				−0.8 ± 4.0
	0.05 Hz	28.8 ± 8.5	28.9 ± 8.3				0.7 ± 5.1
Cerebral deoxy Hb, μM	0 Hz	14.3 ± 3.3	14.8 ± 3.2	0.06	0.07	0.81	4.0 ± 5.1	0.15
	0.1 Hz	14.7 ± 3.5	15.2 ± 3.4				3.9 ± 4.1
	0.05 Hz	14.0 ± 3.9	14.2 ± 4.1				1.1 ± 4.2
Respiratory responses
Respiratory rate, breaths/min	0 Hz	12.6 ± 3.6	16.5 ± 7.1*	0.69	0.01	0.13	30.7 ± 32.7	0.15
	0.1 Hz	13.7 ± 4.6	14.6 ± 5.7				10.4 ± 31.3
	0.05 Hz	13.3 ± 3.7	14.4 ± 6.4				6.3 ± 28.8
Plasma samples
Nitrite, μM	0 Hz	3.0 ± 0.9	3.5 ± 0.6	0.05	0.46	0.21	18.4 ± 21.1	0.04
	0.1 Hz	3.6 ± 1.6	3.3 ± 0.7				−2.7 ± 21.3†
	0.05 Hz	2.8 ± 0.5	3.0 ± 0.4				11.8 ± 12.8
Endothelin-1, pg/mL	0 Hz	1.39 ± 0.43	1.45 ± 0.30	0.56	0.03	0.30	11.31 ± 27.96	0.39
	0.1 Hz	1.26 ± 0.18	1.66 ± 0.80				34.07 ± 65.78
	0.05 Hz	1.28 ± 0.30	2.03 ± 1.45				65.45 ± 133.07

Absolute data were analyzed using a two-way linear mixed model, followed by a priori planned post hoc tests between profiles at baseline and lower body negative pressure (LBNP), and within profile between baseline and LBNP, with a Holm correction for multiple comparisons. Percent change from baseline data were analyzed using a one-way linear mixed model followed by Tukey post hoc tests. Data are presented as means ± SD. etO₂, end-tidal O₂; CVC index, cerebrovascular conductance index; Hb, hemoglobin; ICA, internal carotid artery; MCA, middle cerebral artery; PCA, posterior cerebral artery; PCAv, posterior cerebral artery velocity. *P ≤ 0.06 vs. baseline; †P ≤ 0.07 vs. 0 Hz; §P ≤ 0.07 vs. 0.1 Hz.

Figure 3.

Systemic cardiovascular responses to oscillatory lower body negative pressure (LBNP). A: stroke volume. B: heart rate. C: pulse pressure. D: cardiac output. E: mean arterial pressure (MAP). F: systemic vascular resistance (SVR). Data were analyzed using a one-way linear mixed model, followed by Tukey post hoc tests. Data are presented as means ± SD. Exact P values are reported.

Cerebrovascular Responses

As a result of the successful end-tidal forcing of etCO₂, ICA blood flow, MCAv, and PCAv did not change from baseline during LBNP and there were no differences between profiles (Fig. 4, Table 1). As a result of this maintenance of cerebral blood flow during the hypovolemic stress of LBNP, ScO₂ did not change from baseline either, and was not different between profiles (Fig. 4). Interestingly, the relative increases in ICA conductance and MCAv conductance index were greater during the 0 Hz profile (vs. 0.1 Hz for ICA conductance, and vs. 0.05 Hz for MCAv conductance index), but these responses did not translate to higher flow or velocity in either vessel (Table 1; Fig. 4).

Figure 4.

Cerebral responses to oscillatory lower body negative pressure (LBNP) with clamped end-tidal CO₂ (etCO₂). Absolute responses are shown in A, C, E, G, and I, and relative (% Δ from baseline) data are show in B, D, F, H, and J. ICA, internal carotid artery; MCA CVCi, middle cerebral artery cerebrovascular conductance index; MCAv, middle cerebral artery velocity; ScO₂, cerebral tissue oxygenation. Absolute data were analyzed using a two-way linear mixed model, followed by a priori planned post hoc tests between profiles (0 Hz vs. 0.1 Hz vs. 0.05 Hz) at baseline and LBNP and within profile (0 Hz vs. 0.1 Hz vs. 0.05 Hz) between baseline and LBNP, with a Holm correction for multiple comparisons. Percent change from baseline data were analyzed using a one-way linear mixed model followed by Tukey post hoc tests. Data are presented as means ± SD. Exact P values are presented.

Frequency Domain Responses

As intended by the experimental design, mean arterial pressure oscillations in the LF range and at 0.1 Hz were increased during the 0.1 Hz profile (Table 2). Oscillations in mean arterial pressure in the VLF range and at 0.05 Hz were also increased during the 0.05 Hz profile (Table 2). This same pattern of response occurred for MCAv during the 0.1 Hz profile, but was only observed in the 0.05 Hz profile at the point measure of 0.05 Hz and not across the VLF range (Table 2). PCAv responses differed slightly in that LF oscillations and increased from baseline for both the 0 Hz and 0.1 Hz profiles (Table 2), although oscillations at the point frequency of 0.1 Hz only increased with the 0.1 Hz profile. Oscillations for PCAv only showed an increase at the point measure of 0.05 Hz during the 0.05 Hz profile, but not across the VLF range (Table 2).

Table 2.

Frequency domain cardiovascular and cerebrovascular responses to oscillatory LBNP

		Time
	Profile	Baseline	LBNP	Profile	Time	Profile × Time	%Δ from Baseline	P Value
MAP LF, mmHg2	0 Hz	9.4 ± 7.1	25.3 ± 17.2	0.02	<0.001	0.04	187.6 ± 176.7	0.01
	0.1 Hz	13.0 ± 19.4	46.4 ± 27.6*†				582.2 ± 545.7†
	0.05 Hz	10.6 ± 9.9	20.6 ± 19.5§				135.5 ± 148.5§
MAP at 0.1 Hz, mmHg2	0 Hz	96.9 ± 92.0	392.9 ± 409.9	<0.001	<0.001	<0.001	394.2 ± 481.8	0.03
	0.1 Hz	274.9 ± 637.9	2,222.6 ± 1714.2*†				3,132.6 ± 4163.2†
	0.05 Hz	130.6 ± 209.1	320.0 ± 256.7§				313.0 ± 361.3§
MAP VLF, mmHg2	0 Hz	12.6 ± 4.9	14.7 ± 12.6	0.01	0.02	0.04	18.8 ± 93.4	0.09
	0.1 Hz	10.7 ± 4.4	14.0 ± 8.5				39.1 ± 80.2
	0.05 Hz	14.6 ± 4.9	46.2 ± 49.2*†§				231.0 ± 364.2
MAP at 0.05 Hz, mmHg2	0 Hz	137.7 ± 75.7	157.9 ± 140.2	0.003	0.01	0.005	43.9 ± 134.5	0.003
	0.1 Hz	110.7 ± 75.6	231.3 ± 175.1				167.8 ± 254.3
	0.05 Hz	169.4 ± 70.9	2,104.0 ± 2442.5*†§				1,329.1 ± 1472.3†§
MCAv LF, (cm/s)2	0 Hz	4.0 ± 2.5	13.0 ± 13.4	0.87	<0.001	0.39	196.0 ± 193.3	0.03
	0.1 Hz	3.5 ± 1.9	15.5 ± 10.2*				358.8 ± 194.0
	0.05 Hz	5.1 ± 3.2	9.8 ± 5.2				125.8 ± 112.9§
MCAv at 0.1 Hz, (cm/s)2	0 Hz	44.6 ± 26.6	241.5 ± 245.7	0.03	<0.001	0.02	415.3 ± 488.2	0.002
	0.1 Hz	44.1 ± 33.1	739.6 ± 661.9*†				2,070.7 ± 1461.3†
	0.05 Hz	44.5 ± 19.5	174.6 ± 143.9§				308.9 ± 307.7§
MCAv VLF, (cm/s)2	0 Hz	8.8 ± 5.1	8.6 ± 5.1	0.24	0.16	0.51	14.9 ± 62.3	0.44
	0.1 Hz	7.8 ± 5.0	11.0 ± 7.0				72.0 ± 99.7
	0.05 Hz	9.5 ± 5.4	14.5 ± 14.6				92.9 ± 200.9
MCAv at 0.05 Hz, (cm/s)2	0 Hz	66.5 ± 35.5	72.0 ± 46.2	0.01	0.03	0.04	38.9 ± 125.3	0.01
	0.1 Hz	85.7 ± 65.1	94.7 ± 105.0				32.6 ± 87.7†§
	0.05 Hz	95.8 ± 67.4	309.7 ± 296.9*†§				298.7 ± 309.2
PCAv LF, (cm/s)2	0 Hz	1.8 ± 1.5	4.5 ± 3.7*	0.29	<0.001	0.13	175.2 ± 231.7	0.70
	0.1 Hz	2.1 ± 0.2	7.1 ± 2.6*				232.8 ± 111.9
	0.05 Hz	2.1 ± 1.6	4.1 ± 1.9				141.4 ± 143.1
PCAv at 0.1 Hz, (cm/s)2	0 Hz	15.2 ± 7.7	78.0 ± 81.0	0.003	<0.001	<0.001	362.5 ± 444.2	0.004
	0.1 Hz	18.6 ± 9.9	312.0 ± 177.2*†				2,201.7 ± 1716.6†
	0.05 Hz	20.1 ± 11.2	83.5 ± 81.2§				241.6 ± 166.7§
PCAv VLF, (cm/s)2	0 Hz	3.2 ± 2.0	2.8 ± 1.9	0.29	0.95	0.53	9.0 ± 92.6	0.95
	0.1 Hz	4.1 ± 1.7	5.1 ± 3.2				21.6 ± 48.4
	0.05 Hz	4.2 ± 2.8	3.7 ± 1.5				13.5 ± 51.9
PCAv at 0.05 Hz, (cm/s)2	0 Hz	22.8 ± 20.9	22.9 ± 13.0	0.004	0.004	0.02	44.2 ± 98.4	0.13
	0.1 Hz	44.6 ± 24.1	74.9 ± 52.7				165.1 ± 289.7
	0.05 Hz	32.1 ± 25.5	115.0 ± 70.5*†				410.7 ± 488.3

Absolute data were analyzed using a two-way linear mixed model, followed by a priori planned post hoc tests between profiles at baseline and lower body negative pressure (LBNP) and within profile between baseline and LBNP, with a Holm correction for multiple comparisons. Percent change from baseline data were analyzed using a one-way linear mixed model followed by Tukey post hoc tests. Data are presented as means ± SD. LF, low frequency; MAP, mean arterial pressure; MCAv, middle cerebral artery velocity; PCAv, posterior cerebral artery velocity; VLF, very low frequency. *P ≤ 0.04 vs. baseline; †P ≤ 0.07 vs. 0 Hz; §P ≤ 0.05 vs. 0.1 Hz.

Circulating Vasoactive Mediators

Although there was no overall increase in plasma nitrite from baseline to LBNP for any of the profiles (Table 1), there was a difference in the relative responses, with an increase in the 0 Hz and 0.05 Hz profiles, but a small average decrease in the 0.1 Hz profile. Although plasma endothelin-1 did increase with LBNP (Table 1), the relative change in plasma endothelin-1 concentrations did not differ between profiles, likely due to very high interindividual variability.

DISCUSSION

The main findings of this study are: 1) neither cerebral blood flow nor cerebral tissue oxygen saturation decreased with oscillatory or nonoscillatory LBNP when etCO₂ was clamped at baseline values, and subsequently; 2) oscillating arterial pressure and cerebral blood flow did not increase cerebral blood flow measured as either ICA blood flow or MCAv; 3) mean arterial pressure and stroke volume were protected during LBNP with oscillatory arterial pressure and blood flow, and; 4) circulating concentrations of nitrite and endothelin-1 did not meaningfully differ between the oscillatory and nonoscillatory LBNP profiles.

In this study, we sought to test whether oscillations in arterial pressure and blood flow could be a mechanism for increasing cerebral blood flow during simulated hemorrhage, without the confounding effect of LBNP-induced hypocapnia. Arterial partial pressure of CO₂ is a powerful regulator of cerebral blood flow (19, 20), and hypocapnia could potentially account for over 50% of the reduction in cerebral blood flow observed during LBNP (13). In our previous studies using oscillatory arterial pressure and cerebral blood flow during LBNP, the hypocapnia induced by application of LBNP was not prevented, and etCO₂ decreased by the same magnitude for all conditions (10, 11). In our first study (10), we also observed an oscillatory pattern of the etCO₂ signal at the same frequencies as the oscillatory LBNP profiles, which could have been contributing to the oscillations in cerebral blood flow. To account for the known effects of hypocapnia on the decreases in cerebral blood flow during LBNP, and the oscillatory effect of etCO₂ with forced oscillations, we used dynamic end-tidal forcing to clamp etCO₂ [and therefore $\text{P}{\text{a}}_{\text{C}{\text{O}}_2}$ (15)] at resting eucapnic values. Interestingly, all indices of cerebral blood flow (ICA blood flow, MCAv, and PCAv) did not change from baseline during application of LBNP, despite reductions in mean arterial pressure, stroke volume, and cardiac output during each protocol. Subsequently, oscillating arterial pressure and cerebral blood flow did not affect indices of cerebral blood flow compared with the control condition (0 Hz LBNP), and there was no change in cerebral tissue oxygenation. The maintenance of cerebral blood flow despite decreases in cerebral perfusion pressure (i.e., mean arterial pressure) suggests that cerebral autoregulation was also intact under all conditions.

These findings highlight a potential use of hypercapnia as a therapeutic approach for preserving both cerebral blood flow and cerebral tissue oxygenation with reductions in central blood volume (e.g., hemorrhage, sepsis). The effect of either CO₂ supplementation or end-tidal CO₂ clamping on tolerance to LBNP has been explored by several investigators. Howden et al. (21) assessed the effectiveness of 5% CO₂ supplementation on tolerance to LBNP. CO₂ supplementation increased LBNP tolerance time, but the physiological mechanism for this response was not investigated. Lucas et al. (22) also tested the viability of supplemental CO₂ as a means for improving tolerance to LBNP with the added physiological stress of heat. Participants in this study were exposed to two LBNP conditions to presyncope—heated with hypercapnia, or heated without hypercapnia. Despite increases in etCO₂ and MCAv, LBNP tolerance was not altered, which was contrary to the Howden et al. finding. In a similar study, also with the combined stress of LBNP plus heat, Shibasaki et al. (23) also showed no improvement in LBNP tolerance, despite protection of volumetric flow through the ICA and vertebral artery at submaximal LBNP stages. Instead of supplemental CO₂, Lewis et al. (13) clamped etCO₂ at resting values and compared MCAv and LBNP tolerance to a free-breathing condition. Again, tolerance was no different between these conditions despite an attenuated reduction in MCAv. In the current study, where we also clamped etCO₂, 64% of participants (7 of 11) completed the entire 0 Hz profile. In comparison, in our previous study where etCO₂ was not clamped and participants became hypocapnic, 50% of participants (7 of 14) completed the 0 Hz profile. There is no difference in these outcomes between the eucapnic and hypocapnic protocols (P = 0.16; assessed via a χ² test), further supporting the conclusion that CO₂ supplementation does not improve tolerance to LBNP, but does protect cerebral blood flow.

A consistent finding across all three of our studies assessing the effects of oscillatory hemodynamics during simulated hemorrhage is the improvement in tolerance to this stress. In our first study in normoxic conditions, 7 of 14 (50%) participants completed the entire 0 Hz profile, 11 of 14 (79%) participants completed the 0.1 Hz protocol, and 13 of 14 (93%) participants completed the 0.05 Hz protocol (10). In our second study in hypoxic conditions, 6 of 8 (75%) participants completed the 0 Hz condition, and 7 of 8 (88%) participants completed the 0.1 Hz profile. In our current study, 7 of 11 (64%) participants completed the 0 Hz profile, and all participants completed the 0.1 Hz profile (11/11, 100%) and 0.05 Hz profile (9/9, 100%). Other studies have also demonstrated improvement in tolerance to central hypovolemia by inducing oscillatory hemodynamics. Tolerance time to combined head-up tilt and LBNP was increased in a study by Lucas et al. (24) who used a slow breathing protocol (6 breaths/min, or 0.1 Hz) to induce hemodynamic oscillations. Similarly, tolerance to combined head-up tilt and LBNP was increased in a study by Hockin and Claydon (25) who employed the use of intermittent calf compression with a 15-s cycle (0.07 Hz). Taken together, low-frequency (around 0.05–0.1 Hz) hemodynamic oscillations appear to show promise in improving tolerance to these hypovolemic stimuli.

With both 0.1 Hz and 0.05 Hz oscillations during LBNP, stroke volume was protected, similar to our prior study in normoxic conditions (10). This protection of stroke volume resulted in an attenuated compensatory increase in heart rate with these two profiles compared with the 0 Hz profile, and no difference in the subsequent cardiac output responses across profiles. Despite the similar cardiac output and systemic vascular resistance responses, however, mean arterial pressure was protected in both the oscillatory LBNP profiles compared with the 0 Hz profile. This finding differs from our two previous studies, where there was no difference in the reduction in mean arterial pressure between profiles (10, 11). This discrepancy is likely explained by 1) the separation of LBNP profiles onto different days in our current study versus our previous study conducted in normoxic conditions (10), and 2) testing in normoxic versus hypoxic conditions (11). In our normoxic study, all LBNP profiles (0 Hz, 0.05 Hz, and 0.1 Hz) were run within the same experimental session, with a randomized order. This approach resulted in an elevation in sympathetic activity during recovery, which did not return to baseline before starting the next profile (indexed by circulating norepinephrine concentrations) (10). This sympathoexcitation could have resulted in a “priming” of the baroreflex, and protection of arterial pressure in all subsequent profiles, regardless of the forced hemodynamic oscillations. Similarly, sympathetic activity was most likely elevated in our second study conducted at high altitude (11), due to the effects of hypoxia on the sympathetic nervous system (26). As an increase in sympathetic activity will increase the amplitude of endogenously generated ∼0.1 Hz arterial pressure oscillations in humans (27), forcing oscillations around this frequency may also augment sympathetic activity, resulting in a protection of mean arterial pressure compared with the nonoscillatory condition. Interestingly, due to the greater decrease in arterial pressure during the 0 Hz profile, ICA conductance increased (presumably due to vasodilation distal to the ICA) compared with the 0.1 Hz profile to maintain similar blood flow to the brain.

Plasma concentrations of nitrite did not differ meaningfully between profiles. Although the percent change in nitrite concentration differed between the 0 Hz and 0.1 Hz profiles, the magnitude of change was no more than what might be seen in daily fluctuations in basal nitrite plasma concentrations (28). Plasma endothelin-1 concentrations did not differ between profiles but did increase overall with LBNP (Table 1). This is in line with a previous observation from Kaufmann et al. (14) who reported an increase in plasma endothelin-1 during head-up tilt in healthy participants. Together, however, hemodynamic oscillations did not appear to exert any cardiovascular effects through release of these vasoactive mediators.

Methodological Considerations

Vertebral artery blood flow was not assessed, so the effect of forced arterial pressure and blood flow oscillations on volumetric blood flow into the posterior regions of the brain were not examined. However, PCAv was also maintained at baseline values, following the same pattern of response as both ICA blood flow and MCAv, so we assume vertebral artery blood flow would show a similar response.

The number of female participants was low in this study, so it makes generalizing the findings to both sexes more difficult. However, the two female participants had similar cardiovascular responses as their male counterparts. Future work should seek to clarify whether the cardiovascular responses to forcing hemodynamic oscillations during simulated hemorrhage are altered by sex. In addition, the influence of the menstrual cycle could be explored as hormonal changes across the menstrual cycle may change vascular responsiveness to forced hemodynamic oscillations.

Although direct validation of ModelFlow-derived measurements of stroke volume and cardiac output have not been made during oscillatory LBNP, Shibasaki et al. (29) demonstrated that under normothermic conditions, the Modelflow method of measuring the decrease in stroke volume and cardiac output during LBNP is the same as using thermodilution. Furthermore, Thomas et al. (30) demonstrated that the reduction in mean arterial pressure measured via Finometer was identical to that measured invasively with a radial arterial catheter during the combined stress of head-up tilt plus LBNP (−41% vs. −42% at presyncope). Accordingly, we are confident that our estimates of stroke volume, cardiac output, and systemic vascular resistance are accurate during LBNP.

Conclusions

In conclusion, neither cerebral blood flow nor cerebral oxygenation decreased during any LBNP profile, most likely due to the induced state of isocapnia. However, forcing hemodynamic oscillations did protect stroke volume and mean arterial pressure, and also improved tolerance to the simulated hemorrhage protocols. This improvement in tolerance was not associated with protection of cerebral blood flow, cerebral tissue oxygenation, or changes in plasma concentrations of nitrite or endothelin-1. Future studies should explore mechanisms that may play a role in protecting mean arterial pressure with oscillatory blood flow, such as changes in sympathetic nerve activity and/or neurovascular transduction.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

G.K.A. and K.A.D. were funded by a National Institutes of Health (NIH)-supported Neurobiology of Aging Training Grant T32 AG020494 (Principal Investigator: N. Sumien), and G.K.A. was also funded by an American Heart Association Predoctoral Fellowship 20PRE35210249. K.A.D was also funded by an American Heart Association Predoctoral Fellowship 23PRE1018469. A.J.R. was funded by a Ruth L. Kirchstein NRSA F32 Postdoctoral Fellowship 1F32 HL144082-01A1. C.A.R. was funded by an American Heart Association Grant-in-Aid 17GRNT33671110 and an American Heart Association Transformational Project Award 19TPA34910073.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

G.K.A., A.J.R., and C.A.R. conceived and designed research; G.K.A., K.A.D., N.B., R.R., A.J.R., and C.A.R. performed experiments; G.K.A. analyzed data; G.K.A., K.A.D., N.B., and C.A.R. interpreted results of experiments; G.K.A. prepared figures; G.K.A. drafted manuscript; G.K.A., K.A.D., N.B., R.R., A.J.R., and C.A.R. edited and revised manuscript; G.K.A., K.A.D., N.B., R.R., A.J.R., and C.A.R. approved final version of manuscript.