Aerobic exercise training reduces cerebrovascular impedance in older adults: a 1-year randomized controlled trial

Abstract

Older adults have higher cerebrovascular impedance than young individuals which may contribute to chronic brain hypoperfusion. Besides, middle-aged athletes exhibit lower cerebrovascular impedance than their sedentary peers. We examined whether aerobic exercise training (AET) reduces cerebrovascular impedance in sedentary older adults. We conducted a proof-of-concept trial that randomized 73 older adults to 1 yr of AET (n = 36) or stretching and toning (SAT, n = 37) interventions. Cerebrovascular impedance was estimated from simultaneous recordings of carotid artery pressure (CAP) via applanation tonometry and cerebral blood flow velocity (CBFV) in the middle cerebral artery via transcranial Doppler using transfer function analysis. Fifty-six participants completed 1-yr interventions, and 41 of those completed cerebrovascular impedance measurements. AET group showed a significant increase in V̇o_2peak after the intervention [estimated marginal mean (95% confidence interval); from 22.8 (21.6 to 24.1) to 24.9 (23.6 to 26.2) mL·kg⁻¹·cm⁻¹, P < 0.001], but not SAT [from 21.7 (20.5 to 22.9) to 22.3 (21.1 to 23.7) mL·kg⁻¹·cm⁻¹, P = 0.114]. Coherence between changes in CBFV and CAP was >0.90 in the frequency range of 0.78–3.12 Hz. The averaged cerebrovascular impedance modulus (Z) in this frequency range decreased after 1-yr AET [from 1.05 (0.96 to 1.14) to 0.95 (0.92 to 1.06) mmHg·s·cm⁻¹, P = 0.023], but not SAT [from 0.96 (0.87 to 1.04) to 1.01 (0.92 to 1.10) mmHg·s·cm⁻¹, P = 0.138]. Reductions in Z were correlated positively with reductions in carotid pulse pressure (r = 0.628, P = 0.004) and inversely with mean CBFV (r = −0.563, P = 0.012) in the AET group. One-year AET reduces cerebrovascular impedance in older adults, which may benefit brain perfusion.

NEW & NOTEWORTHY Estimation of cerebrovascular impedance is essential for understanding dynamic cerebral blood flow regulation. This randomized controlled trial demonstrated that aerobic exercise training reduced cerebrovascular impedance in older adults, which may benefit brain perfusion.

INTRODUCTION

The brain is continuously perfused with high volume of blood flow because of its high metabolic rate and relatively low vascular resistance (1, 2). Since cerebrovascular resistance is relatively low compared with the other vascular beds, small cerebral blood vessels are exposed to high-pressure fluctuations (3). Thus, the cushioning function of the large elastic central arteries (e.g., the proximal aorta and carotid arteries), the so-called Windkessel effects, likely plays an essential role in maintaining brain health (3). However, with advancing age, the Windkessel function of the central arteries may deteriorate due to arterial stiffening (4–6), which has been related to brain structural and functional abnormalities and increased risk of Alzheimer disease and related dementias (3, 7–9). Likewise, the Windkessel function of the intracranial large cerebral arteries is also likely to be diminished with aging due to arterial stiffening (10–14). The precise underlying mechanisms by which central and large cerebral arterial stiffening lead to brain structural and functional abnormalities remain not fully understood. Recent studies suggest that the elevated central arterial pressure and blood flow pulsatility associated with arterial stiffening are important hemodynamic changes, which may contribute to cerebral endothelial dysfunction and development of cerebral small vessel disease (15, 16).

Aerobic exercise training (AET) improves central arterial compliance even in late life (6, 17, 18). However, such evidence is lacking regarding the large cerebral arteries. Cross-sectional studies demonstrated that higher aerobic capacity was associated positively with cerebral arterial compliance among young, healthy adults (10, 19). Our studies also showed that middle-aged Masters athletes had lower cerebrovascular impedance than their sedentary age-matched controls, which suggested higher cerebral arterial compliance in the athlete group. Moreover, the lower cerebrovascular impedance in athletes was associated with higher cerebral blood flow velocity (CBFV) with transcranial Doppler (TCD) and cerebral cortical perfusion measured with magnetic resonance imaging (MRI) arterial spin labeling (ASL) (10). Given the limitations of these cross-sectional studies, it is important to determine whether AET can reduce cerebrovascular impedance in older adults using interventional design.

Several studies have evaluated the Windkessel function of the cerebrovasculature with pulsatile changes in cerebral blood flow (CBF) using MRI ASL (19) or CBFV using TCD (20) at a given change in blood pressure (BP) measured in the brachial artery. Although these studies have advanced our understanding of cerebral arterial compliance in human subjects, a major limitation was the measurement of peripheral rather than cerebral or central arterial pressure in these studies. Central arterial pressure does not correspond to peripheral arterial pressure due to the pulse pressure (PP) amplification as pressure waves propagate from the central to the periphery arteries (21). Thus, the accuracy of cerebral arterial compliance reported in these previous studies is a major concern. In addition, because of the presence of blood flow inertia, vascular resistance, and arterial distensibility, measurement of arterial compliance with instantaneous blood flow through a cardiac cycle divided by the corresponding change in BP may not be appropriate for describing the dynamic pressure-flow relationship (22). In this regard, estimation of arterial input impedance, a frequency-domain approach used to quantify the dynamic pressure-flow relationship of a vascular bed, reflects not only local arterial properties where BP and flow are measured but also the properties of the downstream vascular bed (i.e., vascular resistance and compliance) (21).

The purpose of this study was to determine the effects of 1-yr AET on cerebrovascular impedance in older individuals and its relationship with changes in mean CBFV and CBFV pulsatility. We hypothesized that 1) 1-yr AET would reduce cerebrovascular impedance when compared with the active control group of stretching and toning (SAT) and 2) decreases in cerebrovascular impedance are correlated with increases in mean CBFV measured in the middle cerebral artery (MCA).

METHODS

Study Design and Subjects

This was a 12-mo open-label randomized controlled trial (RCT) to investigate the effect of a 1-yr progressive, moderate-to-vigorous AET program on neurocognitive function in cognitively normal older adults who previously had a sedentary lifestyle (22). Neurocognitive function assessments included the domains of inductive reasoning, long-term episodic memory, working memory, processing speed, and verbal ability which were reported in a previous study (22). The present study reported changes in cerebrovascular impedance with AET and SAT, which was a secondary outcome of the parent study. This study was approved by the Institutional Review Board of the University of Texas Southwestern Medical Center and Texas Health Presbyterian Hospital Dallas in accordance with the guidelines of the Declaration of Helsinki and Belmont Report (STU 102010-069). All participants gave written informed consent before participation. This trial was not registered because at the time of trial initiation, registration of nonpharmacological interventional studies in healthy adults in a public database was neither required nor typical.

A flowchart for this RCT characterizing participant recruitment and inclusion is presented in Supplemental Fig. S1(all Supplemental material is available at https://doi.org/10.6084/m9.figshare.19666251). This study targeted cognitively normal sedentary but otherwise healthy men and women aged 60–80 yr. Recruitment was conducted in the Dallas-Fort Worth metropolitan area using community-based advertisements. At an initial telephone screening, participants were screened if they had subjective cognitive complaints or a history of major clinical conditions, regularly engaged in a structured exercise program, and could participate in a 1-yr AET or SAT program including the required visits for data collection. Subsequently, participants were asked to visit our clinical office and screened for the following exclusion criteria: 1) clinical diagnosis of major psychiatric or neurological disorders or medications causing major effects on cognition, 2) a history of active alcoholism or drug abuse, 3) a history of recurrent epilepsy, stroke, or head injury/trauma with a loss of consciousness ≥ 30 min, 4) Mini-Mental Status Examination (MMSE) score < 26 to exclude dementia, 5) uncontrolled hypertension (averaged three measurements of sitting systolic BP ≥ 140 or diastolic pressure ≥ 90 mmHg confirmed by 24-h ambulatory BP monitoring), 6) a diagnosis of diabetes mellitus (fasting glucose > 126 mg/dL or taking antidiabetic medications), 7) severe obesity with body mass index (BMI) ≥ 35 kg/m², 8) smoking within the past 5 yr of the study, 9) other major or unstable medical conditions such as a history of coronary bypass surgery or heart attack within the past year, ongoing chemotherapy, or severe lung, kidney and liver disease, 10) individuals who spent >90 min of moderate-to-vigorous physical activity [>4.0 metabolic equivalents (METs)] per week were excluded, as determined by a 1-wk physical activity monitoring using an accelerometer (Actical, Philips Respironics), and 11) individuals with physical disability, metal implants in the body, or claustrophobia precluding MRI scans were excluded. Among the 991 candidates screened by phone, 152 older adults were invited to the clinic for further in-person screening. Seventy-nine of these were excluded because they either did not meet the inclusion and exclusion criteria or did not complete baseline measurements.

Randomization and Blinding

Randomization was performed in SAS V9.2 using two stratification groups, sex (men and women), and education (10–14 and 15–20 yr), using a blocking factor of four. The randomization assignments were generated by the study statistician and placed in a sealed envelope so that the study personnel were blinded until opening the envelope for treatment assignment of an individual subject. Investigators conducting the primary and secondary outcome measurements were blinded to treatment assignment throughout the study. Participants were instructed to maintain normal daily activities aside from the assigned interventions and were instructed not to disclose group assignments or interventions during outcome measurements or to meet with other participants. Finally, 73 individuals were randomly assigned to SAT (n = 37; 28 women, 68 ± 5 yr of mean age) or AET (n = 36; 27 women, 69 ± 6 yr of mean age). Their demographic characteristics at baseline are summarized in Table 1.

Table 1.

Participants’ demographics by randomization groups

Variables	SAT	AET	P Value
n, men/women	37 (9/28)	36 (10/27)	0.947
Age, yr	68 ± 5	69 ± 6	0.419
Race, white/black	34/3	36/0	0.947
Education, yr	16 ± 2	17 ± 2	0.160
Height, cm	164.8 ± 8.6	165.9 ± 8.3	0.559
Weight, kg	74.0 ± 10.7	71.5 ± 15.6	0.423
Body mass index, kg/m2	27.3 ± 3.6	25.8 ± 4.4	0.120
Medications
Antihypertensives, n Cholesterol-lowering, n	5 12	8 8	0.331 0.328
Physical activity, min/day
Light (<4.0 METs)	240 ± 91	221 ± 81	0.372
Moderate (4.0–5.0 METs)	4 ± 4	4 ± 5	0.938
Vigorous (>5.0 METs)	2 ± 4	1 ± 3	0.474

Values represent means ± standard deviation. n is number of subjects. AET, aerobic exercise training; METs, metabolic equivalents; SAT, stretching and toning.

Intervention

AET and SAT programs were conducted with the same protocol as previously reported (23). The dose and intensity of the AET program were based on each individual’s fitness level assessed with peak oxygen uptake (V̇o_2peak) testing, and progressively increased as participants adapted to previous workloads. Specifically, the program started with a frequency of 3 exercise sessions per week for 25–30 min per session at the intensity of 75%–85% of maximal heart rate that was measured during V̇o_2peak testing at baseline. At week 11, participants started alternating between 3 and 4 exercise sessions per week for 30–35 min per session, and at the weeks in which they performed 3 exercise sessions per week, a high-intensity exercise session was introduced, which consists of 30 min of walking at the intensity of 85%–90% of maximal heart rate (e.g., brisk uphill walking). After week 26, participants performed 4–5 exercise sessions per week for 30–40 min, including two high-intensity sessions. Any modes of aerobic exercise were allowed as long as they maintained the prescribed training dose and intensity, as monitored by changes in heart rate during each of the exercise sessions (Polar RS400, Polar Electro). This AET program meets the national physical activity guidelines for older adults (24). It has been used in our previous studies that showed significant improvement of cardiorespiratory fitness in sedentary individuals older than 65 yr of age (25).

SAT was used as an active control group to keep participants engaged with the same level of attention from the investigators as those for the AET group. The frequency and duration of the SAT program were the same as the AET program. An SAT routine that focuses on the upper and lower body was used. In this group, participants were asked to keep their heart rate below 50% of the maximal heart rate during each session. At week 19, we introduced a second set of full body stretches that are more advanced than the previous set. At week 26, we introduced a set of low resistance TheraBand exercise that focuses on strengthening the upper and lower body.

In both AET and SAT programs, each participant was supervised by an exercise physiologist for the first 4–6 wk until they could comfortably and safely perform the assigned program by themselves either at a fitness center or home. During the study period, they were asked to perform an assigned intervention on top of their regular physical activities. To ensure adherence to each program, participants were required to make a training log in addition to heart rate monitoring. Each month, participants visited the clinic to download heart rate data and review their training log together with an exercise physiologist to ensure the implementation of the prescribed training programs. When adherence to exercise programs was not met with the prescribed intensity, duration, and frequencies, in-person and/or telephone meetings were held to solve the issues and encourage participants to continue the program. Training compliance was calculated by the ratio of prescribed exercise sessions over the completed exercise sessions in which participants achieved the prescribed target heart rate (23).

Data Collection

In each participant, cardio- and cerebrovascular testing and cardiorespiratory fitness measurement were performed before and 1 yr after the start of intervention. This RCT involved MRI, cognitive function, and cerebral autoregulation measurements in addition to data reported in this article. Each participant continued SAT or AET intervention till the completion of all postmeasurements. All data were collected in an ambient temperature-controlled laboratory (∼22°C). Subjects abstained from caffeinated beverages, alcohol, and vigorous exercise at least 24 h before testing.

Cardio- and Cerebrovascular Testing

All vascular measurements were performed under the supine position after quiet resting for ≥10 min with normal breathing. ECG (via the three-lead system, Hewlett-Packard), end-tidal CO₂ ($\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$, via a nasal cannula using capnography, Capnogard; Novametrix), and CBFV in the MCA via TCD were recorded simultaneously. CBFV was measured over the temporal window using a 2-MHz TCD probe (Multi-Dop X2; DWL, Singen, Germany). Briefly, the probe was securely attached to the temporal bone acoustic window by using either an individually created mold to fit the facial bone structure or a probe holder (Spencer Technologies, Seattle, WA) to keep the position and angle of the probe unchanged during the assessment. To ensure the same location of TCD recording in each subject before and after the intervention, we recorded the probe location, depth of insonation, gain, and bony landmarks during premeasurement and used precisely the same setup for the postmeasurement, as previously recommended (26). In addition, brachial cuff pressure (via electrosphygmomanography; Suntech) was acquired three times and calculated the average systolic and diastolic BP and PP. Using an applanation tonometry system (SphygmoCor 8.0; AtCor Medical), the right brachial and the carotid (ipsilateral to the CBFV measurement) arterial pressure waveforms were recorded over 10 s at least three times. A pressure sensor was directly placed on the skin and pressed on the arteries at a location where the strongest pulse was felt. During carotid arterial pressure (CAP) measurement, subjects were asked to raise their chin slightly. The probe was held directly over the pulse, staying as close to vertical to the vessel axis as possible, and using light pressure to applanate the underlying vessel wall (27). Recordings were taken only when a clean signal was obtained with high-amplitude excursion. Based on the quality-control process incorporated into the SphygmoCor system, the most reproducible waveforms over 10 s with higher pulse amplitude, smaller pulse amplitude variation, and smaller diastolic variation were extracted at the brachial and the carotid arteries (28). Mean arterial pressure (MAP) was defined as the time-averaged area under the brachial arterial pressure waveforms over 10 s, which was corrected by the cuff systolic and diastolic BP as previously reported (17, 29). CAP waveform recorded as the change in voltage was calibrated with the assumption that the mean and diastolic CAPs are equal to those of brachial arterial pressure (17, 29). CAP and CBFV waveforms were recorded continually for 10 s for spectral analysis. All signals were stored on a computer using a commercial software package for data acquisition (AcqKnowledge 4.2; Biopac Systems Inc.) with a sampling frequency of 1k Hz.

Data Analysis

Frequency-domain analysis.

Cerebrovascular impedance was evaluated by the transfer function method from pulsatile changes in CBFV (as an input signal) and CAP (as an output signal) for 10 s as we previously reported (10–12). For this calculation, auto-spectra and cross-spectra of CBFV and CAP were estimated using the Welch algorithm (30) as follows. Time series of CBFV and CAP waveforms were resampled at 100 Hz and were subdivided into 256-point segments (2.56 s) with 50% overlap for spectral estimation. To reduce the potential effects of including fractional cardiac cycles in these data segments on spectral estimation, each data segment was multiplied by a Hamming window before the periodogram estimation and average (31). This process resulted in a spectral resolution of 0.39 Hz. Because each data set is short (i.e., 10 s) and does not include slow fluctuations, results at the frequency <0.78 Hz were omitted. In cerebrovascular impedance analysis, the coherence function provides the strength of the linear relationship between CAP to CBFV. The transfer function method is valid only if the system to be identified is linear. Previous studies (10–12) demonstrated that in middle-aged and older subjects, coherence is higher than 0.9 at the frequency <3 Hz where most of energy was included and decreases gradually above 3 Hz. These results demonstrated a strong linear relationship between pulsatile changes in CBFV and CAP at this frequency range, suggesting the validity of using the transfer function method. Prior to the aforementioned analysis, we assessed the reproducibility of cerebrovascular impedance measurement with repeated CAP and CBFV measurements in nine healthy adults in the same study session that lasted from 5 to 10 min and confirmed that the impedance moduli obtained between 0.78 to 3.12 Hz had excellent reproducibility [R² = 0.894, mean difference and standard deviation (SD) of two measurements = 0.03 ± 0.05 mmHg²/(cm/s)²; Supplemental Figs. S2 and S3]. Accordingly, in the present study, we focused on the frequency range from 0.78 to 3.12 Hz. In addition, to highlight the cerebrovascular buffering function, we obtained impedance modulus at the frequency range from 0.78 to 1.56 Hz, where the fundamental first harmonics of CBFV and CABP corresponding to the resting heart rate were contained, as Z₁ (10). Even though CAP is likely to be a good estimate of MCA pressure given the fact that MCA is a direct branch of the internal carotid artery (ICA) and the distance between the ICA and MCA is relatively short (≈10 cm), a time delay of pressure wave propagation from the carotid artery to the MCA would be a confounding factor for the impedance phase estimation. Therefore, we decided not to report the phase results.

Time-domain analysis.

The averaged values of $\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$, heart rate, brachial BP and PP, carotid BP and PP, and CBFV were obtained from breath-by-breath or beat-by-beat values during the corresponding period of cerebrovascular impedance measurement. Cerebrovascular resistance index (CVRi) was calculated as a ratio of mean CAP to mean CBFV.

Cardiorespiratory Fitness

V̇o_2peak, the gold standard measure of cardiorespiratory fitness, was collected using a modified Astrand–Saltin protocol on a treadmill (32, 33). The treadmill grade was increased by 2% every 2 min until exhaustion while participants walked or jogged at a fixed speed, which was determined by the individual fitness level (34). Specifically, the treadmill speed was determined during a submaximal warm-up exercise test before V̇o_2peak testing to achieve a HR response of ∼65%–75% of an individual’s estimated maximal HR (25, 34). V̇o₂ was measured during the 2nd minute of each stage using the Douglas bag method. Also, the breath-by-breath V̇o₂, V̇co₂, respiratory exchange ratio (RER), and ventilation were continuously monitored using an online computer system. Gas fractions were analyzed by mass spectrometry (Marquette MGA 1100), and ventilatory volume was measured by a Tissot spirometer. BP, 12-lead ECG, and heart rate were monitored continuously during exercise testing to assess cardiovascular responses. V̇o_2peak was defined as the highest V̇o₂ measured from a > 30-s Douglas bag during the last stage of testing. The criteria to confirm that V̇o_2peak was achieved included an increase in V̇o₂ <150 mL despite increasing work rate of 2% grade, an RER > 1.1, and heart rate < 5 beats/min of age-predicted maximal values (e.g., 220 − age). In all cases, at least two of these criteria were achieved, confirming the identification of V̇o_2peak based on the American College of Sports Medicine guidelines (24). Our previous study showed that by using these methods, V̇o_2peak can be measured reliably in sedentary older adults (25, 34).

Statistics Analysis

To compare participants’ demographics between the AET and SAT groups, independent sample t test was used for continuous variables and χ² test was used for categorical variables. Data are presented as means and SD for continuous variables or frequency for categorical variables. The data analysis for the RCT was based on the intent-to-treat principle (35) using all available data from the randomized subjects (SAT: n = 37, AET: n = 36). For the analysis of cerebrovascular impedance [e.g., coherence and impedance modulus (Z)], a linear mixed model was used to analyze the main and interaction effects of group (AET vs. SAT), time (Pre vs. Post), and frequency (0.39 Hz-bin from 0.78 to 3.12 Hz). The linear mixed model was also used to analyze the main and interaction effect of group (AET vs. SAT) and time (Pre vs. Post) on Z₁ as well as other hemodynamic measures. Post hoc multiple pairwise comparisons were corrected by the Bonferroni method in case of a significant interactions. To confirm the result from the intent-to-treat analysis, repeated-measures analysis of variance (ANOVA) was performed on the complete case data (SAT: n = 22, AET: n = 19). Pearson’s product-moment correlation was used to determine the relationship between variables of interest. All statistical analyses were performed using SPSS 26.0 (Chicago, IL). Results from the linear mixed model analysis were reported as estimated marginal means and 95% confidence interval (CI). The statistical significance level was set to P < 0.05.

RESULTS

Twenty-eight (76%) and 28 (78%) participants completed the SAT and AET intervention programs, respectively. The averaged training compliance to aerobic exercise program was ∼ 81.3%, as calculated by the ratio of prescribed exercise sessions over the exercise sessions completed in which participants achieved their target heart rate. The average compliance to stretching program was ∼ 70.2%. There was no group difference in the training compliances. The participants’ demographic characteristics, who completed hemodynamics assessments, did not differ from those who participated in the parent study.

Table 2 presents results of systemic hemodynamics and cardiorespiratory fitness from intent-to-treat analysis. AET group had higher V̇o_2peak after the intervention when compared with their baseline level (P < 0.001) and with SAT (P = 0.006), whereas V̇o_2peak did not change with SAT program (P = 0.114). Change in V̇o_2peak in AET was larger than that in SAT in the complete-case comparison (P = 0.021, Cohen’s d = 0.645). A similar tendency was seen in subjects who were completed cerebrovascular impedance measurement (interaction: P = 0.081; Supplemental Table S1). Change in V̇o_2peak in AET tended to larger than that in SAT (P = 0.071, Cohen’s d = 0.596). Baseline systemic hemodynamics were similar between the randomization groups. After the 1-yr intervention, heart rate and brachial BP did not change significantly. Brachial PP showed a significant group-by-time interaction effect (P = 0.021), but no significant differences were observed from the post hoc pairwise comparisons after the Bonferroni correction. There was no significant group-by-time interaction effect in the complete-case analysis (Supplemental Table S1).

Table 2.

Systemic hemodynamics and cardiorespiratory fitness by randomization groups

	SAT			AET			Time	Group	Interaction
	n	EMM	95% CI	n	EMM	95% CI	F Value P value
Weight, kg
Pre	37	73.5	(69.2–77.7)	36	71.8	(67.5–76.2)	0.942	0.608	0.893
Post	28	73.5	(69.1–77.8)	28	70.4	(66.0–74.9)	0.336	0.438	0.349
Body mass index, kg/m2
Pre	37	27.3	(25.8–28.4)	36	25.8	(24.6–27.3)	1.120	2.236	0.390
Post	28	27.0	(25.6–28.4)	28	25.4	(24.1–26.8)	0.294	0.139	0.535
Heart rate, bpm
Pre	37	62	(59–65)	36	64	(61–66)	0.712	0.054	2.981
Post	28	63	(60–65)	28	62	(59–65)	0.403	0.816	0.090
Systolic blood pressure, mmHg
Pre	37	113	(109–117)	36	116	(112–120)	0.202	0.048	2.451
Post	28	115	(111–119)	28	113	(109–117)	0.655	0.828	0.123
Mean arterial pressure, mmHg
Pre	37	88	(85–90)	36	88	(85–90)	2.276	0.274	0.786
Post	28	87	(84–90)	28	85	(82–88)	0.137	0.602	0.379
Diastolic blood pressure, mmHg
Pre	37	70	(67–72)	36	67	(65–70)	1.370	1.174	0.745
Post	28	68	(66–71)	28	67	(65–70)	0.247	0.282	0.392
Pulse pressure, mmHg
Pre	37	43	(40–47)	36	49	(45–52)	0.012	0.879	5.605
Post	28	47	(43–50)	28	46	(42–49)	0.914	0.352	0.021
V̇o2peak, mL/kg/min
Pre	37	21.7	(20.5–22.9)	36	22.8	(21.6–24.1)	22.35	4.681	6.050
Post	28	22.3	(21.1–23.7)	28	24.9	(23.6–26.2)*†	0.000	0.034	0.017

Values represent estimated marginal means (EMM) [95% confidence interval (CI)]. F values (upper row) and P values (lower row) were calculated from the linear mixed model (LMM). n is the number of observations. The Bonferroni correction was applied for multiple pairwise comparisons. *Significant difference (P < 0.05) vs. baseline within the same group. †Significant difference (P < 0.05) vs. SAT. AET, aerobic exercise training; SAT, stretching-and-toning; V̇o_2peak, peak oxygen uptake.

Table 3 shows the results of carotid arterial and cerebrovascular testing. At the baseline, six participants in the SAT and six in the AET group did not complete cerebrovascular testing because of technical issues related to the measurements of CAT or CBFV. At the postintervention, two participants in the AET group did not have the cerebrovascular measurement and five participants in the SAT and seven in the AET group did not have reliable TCD measurements. Thus, 22 participants in the SAT and 19 in the AET group completed the pre- and postdata collections for cerebrovascular impedance analysis (Supplemental Fig. S1).

Table 3.

Carotid arterial and cerebrovascular measurements by randomization groups

	SAT			AET			P Value (LMM)
	n	EMM	95% CI	n	EMM	95% CI	Time	Group	Time × Group
Carotid systolic blood pressure, mmHg
Pre	31	113	(108–117)	30	115	(111–120)	0.495	0.762	0.087
Post	23	115	(110–120)	19	110	(105–116)
Carotid pulse pressure, mmHg
Pre	31	43	(39–47)	30	48	(43–52)	0.986	0.756	0.022
Post	23	47	(42–52)	19	44	(39–49)
Systolic CBFV, cm/s
Pre	31	80	(73–87)	30	80	(73–87)	0.617	0.896	0.770
Post	23	81	(74–89)	19	80	(72–88)
Mean CBFV, cm/s
Pre	31	52	(47–56)	30	51	(46–55)	0.613	0.737	0.907
Post	23	51	(46–55)	19	50	(45–55)
Diastolic CBFV, cm/s
Pre	31	31	(28–33)	30	30	(27–33)	0.096	0.685	0.941
Post	23	29	(26–32)	19	28	(25–31)
Pulsatile CBFV, cm/s
Pre	31	49	(44–54)	30	50	(45–55)	0.103	0.988	0.698
Post	23	52	(47–57)	19	52	(46–57)
CVRi, mmHg·s·cm-1
Pre	31	1.8	(1.6–1.9)	30	1.8	(1.7–2.0)	0.868	0.810	0.638
Post	23	1.8	(1.6–2.0)	19	1.8	(1.6–2.0)
Z1, mmHg·s·cm-1
Pre	31	1.02	(0.93–1.11)	30	1.12	(1.03–1.21)	0.697	0.669	0.020
Post	23	1.08	(0.98–1.18)	19	1.03	(0.93–1.14)
$\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$, %
Pre	31	37.0	(35.7–38.4)	30	36.7	(35.3–38.1)	0.093	0.294	0.410
Post	17	36.4	(34.7–38.2)	18	35.0	(33.2–36.7)

Values represent estimated marginal means (EMM) [95% confidence interval (CI)] and P values calculated from the linear mixed model (LMM). n is the number of observations. The Bonferroni correction was applied for multiple pairwise comparisons. AET, aerobic exercise training; CBFV, cerebral blood flow velocity; CVRi, cerebrovascular resistance index; $\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$, end-tidal carbon dioxide; SAT, stretching-and-toning; Z₁, the impedance modulus corresponded to the first harmonics of CBFV and carotid artery pressure (CAP; 0.78–1.56 Hz).

There were no significant differences in the baseline carotid arterial and cerebrovascular measurements between the randomization groups. After 1-yr intervention, carotid SBP, systolic, mean, diastolic, and pulsatile CBFV, CVRi, and $\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ did not show significant group-by-time interactions. In contrast, carotid PP exhibited a significant group-by-time interaction effect (P = 0.022). However, there were no significant differences from post hoc pairwise comparisons after the Bonferroni correction.

Figure 1 depicts group-averaged data of cerebrovascular impedance modulus and coherence. Coherence was higher than 0.9 irrespective of time and groups. Z exhibited a significant group-by-time interaction effect (P = 0.017). Post hoc pairwise comparisons showed a significant reduction of Z in the AET group (P = 0.023) but not in the SAT group (P = 0.138). Z₁ also showed a significant group-by-time interaction effect (P = 0.020, Table 3). Pairwise comparisons revealed a tendency to reduce Z₁ after 1-yr AET (P = 0.075), whereas no significant change was observed after the 1-yr SAT (P = 0.191). The complete-case analysis also showed a significant group-by-time interaction on Z₁ (P = 0.024; Supplemental Table S1). Average Z in the frequency range from 0.78 to 3.12 Hz (P = 0.009, Cohen’s d = 0.753) and Z₁ (P = 0.021, Cohen’s d = 0.856) exhibited significantly larger reductions with 1-yr AET when compared with those of SAT (Fig. 2).

Figure 1.

Group-averaged frequency plots of cerebrovascular impedance modulus (Z) and coherence before and after 1 yr of stretching and toning (SAT) and aerobic exercise training (AET) programs. Solid and broken lines represent estimated marginal means and 95% confidence intervals, respectively. Black lines and red lines are Pre and Post.

Figure 2.

Changes in cerebrovascular impedance modulus (Z) over 1 yr of stretching and toning (SAT, n = 22) and aerobic exercise training (AET, n = 19) programs. Data are means and standard deviation. P value was calculated by independent t test.

Changes in Z and Z₁ were positively correlated with changes in carotid PP in the AET (r = 0.628, P = 0.004 and r = 0.524, P = 0.022, respectively) and SAT (r = 0.483, P = 0.023 and r = 0.506, P = 0.016) groups, respectively (Fig. 3).

Figure 3.

Correlations between changes in carotid pulse pressure (PP) and cerebrovascular impedance modulus with stretching and toning programs (SAT, n = 22) and aerobic exercise training (AET, n = 19). Z is the averaged impedance modulus in the frequency range of 0.78–3.12 Hz. Z₁ is the impedance modulus that corresponds to the first harmonics of cerebral blood flow velocity and carotid arterial pressure.

Changes in Z and Z₁ were inversely correlated with changes in mean CBFV in the AET (r = −0.563, P = 0.012 and r = −0.657, P = 0.002, respectively; Fig. 4). Similar trends were observed in the SAT (r = −0.401, P = 0.065 and r = −0.370, P = 0.090, respectively). In addition, changes in Z and Z₁ were inversely correlated with changes in pulsatile CBFV in the SAT (r = −0.595, P = 0.003 and r = −0.614, P = 0.002, respectively; Fig. 4). In the AET group, changes in Z₁ (r = −0.479, P = 0.038), but not Z (r = −0.365, P = 0.124), inversely correlated with changes in pulsatile CBFV. The correlation between changes in Z and pulsatile CBFV became significant when one outlying data (pointing with a blue arrow) was removed (r = −0.607, P = 0.008).

Figure 4.

Correlations of changes in cerebrovascular impedance modulus with those in mean and pulsatile cerebral blood flow velocity (CBFV) with stretching and toning programs (SAT, n = 22) and aerobic exercise training (AET, n = 19). Z is the averaged impedance modulus in the frequency range of 0.78–3.12 Hz. Z₁ is the impedance modulus that corresponds to the first harmonics of cerebral blood flow velocity and carotid arterial pressure. A blue arrow points an outlying data.

DISCUSSION

The main findings of this study were twofold. First, 1-yr AET, but not SAT, reduced cerebrovascular impedance in older adults. Second, reductions in cerebrovascular impedance after AET were correlated positively with reductions of carotid PP, and negatively with elevations in mean CBFV.

We have shown that cerebrovascular impedance was higher in middle-aged and older sedentary than young sedentary adults (10–12). Furthermore, cerebrovascular impedance in middle-aged endurance trained athletes was equivalent to that of young sedentary adults (10). These findings together suggest that cerebrovascular impedance increases with aging and that prolonged engagement in AET may prevent or reduce increases in cerebrovascular impedance with aging. The present study extended these previous studies (10–12) by showing that cerebrovascular impedance was reduced by 1-yr AET in previously sedentary older individuals. Previous studies suggested that 3–4 mo of moderate-to-vigorous intensity AET reduced central arterial stiffness in middle-aged and older adults, which was related to improved endothelial function and decreased sympathetic neural activity (17, 36, 37). Therefore, AET-induced improvements in cerebral endothelial function and reductions in sympathetic neural activity may contribute to the observed decreases in cerebrovascular impedance.

Interestingly, CVRi remained unchanged after the 1-yr AET. Vascular resistance or vascular resistance index, as calculated by dividing MAP by mean blood flow or blood flow velocity, was often used to measure vascular tone or vasoconstriction (38). “Resistance” has an analogous meaning as “impedance,” but it is only applicable under nonoscillatory or steady-flow conditions. In this regard, resistance may be considered as the vascular impedance at zero frequency (21). Considering that vascular impedance modulus is determined by both the vascular resistance and stiffness, our observation of decreased cerebrovascular impedance without changes in CVRi suggests that AET reduces cerebral arterial stiffness, which may improve the buffering capability of the cerebral vasculature against dynamic pulsatile changes in arterial pressure and CBF. In support of this hypothesis, a recent study from our group showed that 1-yr AET reduced carotid arterial stiffness, increased global CBF, and reduced CBFV pulsatility in patients with mild cognitive impairment (MCI) (39).

We observed that changes in cerebrovascular impedance (e.g., Z and Z₁) were correlated positively with changes in carotid PP (i.e., central arterial PP) in both the AET and SAT groups. We speculate that in response to age-related elevations in central arterial stiffness and carotid PP (40, 41), the cerebrovascular bed may undergo compensatory remodeling (e.g., arterial hypertrophy, vasoconstriction, and increased cerebrovascular impedance) to protect the downstream microcirculation from the increased mechanical burden of age-related increases in central arterial PP (16) at the cost of increased brain hypoperfusion risk. Consistently with this hypothesis, even no main group effects of AET or SAT on carotid PP were observed in this study, at the individual level, the positive associations between reductions in cerebrovascular impedance and carotid PP in both the AET and SAT group suggest that changes in cerebrovascular impedance are coupled to changes in central arterial PP.

Of note, the trend of reduced carotid PP observed in the AET group (Table 3) was consistent with the observations from a large population-based longitudinal study, which showed that higher physical activity level in late-life, as well as habitual physical activity during mid- to late-life, was associated with lower central arterial PP (42). The relatively small sample size and shorter duration of AET in this and previous training studies may have led to inconsistent findings of the effects of AET on central arterial PP (6, 17, 18).

There is a growing body of evidence suggesting salutary effects of regular AET on brain perfusion and cerebrovascular reactivity to changes in arterial Co₂ (43), although the current findings are inconclusive (44). A cross-sectional study indicated that habitual physical activity might attenuate the age-related reduction in CBFV (45, 46). However, Murrell et al. (47) showed no change in CBFV measured in the MCA at rest after 3 mo of moderate-intensity AET in previously sedentary young and older adults. Conversely, Guadagni et al. (48) observed an increase in CBFV in the MCA in healthy middle-aged and older adults after a 6-mo moderate-intensity AET. In our previous studies, we found global CBF and cerebrovascular reactivity to changes in arterial CO₂ were improved in patients with MCI after 1-yr moderate-vigorous intensity AET although mean CBFV measured with TCD at the MCA remained unchanged (49). In addition, we observed that cerebral cortical perfusion measured with MRI ASL in middle-aged endurance athletes at rest was similar to those of age-matched sedentary adults (10) and that lower cerebrovascular impedance was associated with higher mean CBFV and cerebral perfusion among these subjects (10). These findings are consistent with the inverse relationship between changes in mean CBFV and cerebrovascular impedance after 1-yr AET in older adults, suggesting that AET reduces cerebrovascular impedance, which may benefit brain perfusion (11).

Although we found no interventional effects on pulsatile CBFV at the group level (Table 3), at the individual level, we observed that changes in Z₁ were associated inversely with changes in pulsatile CBFV in the AET and SAT groups. This is consistent with our previous findings that middle-aged Masters athletes had significantly lower cerebrovascular impedance than sedentary age-matched controls associated with higher pulsatile CBFV (10). Of note, excessive pulsatile flow has been proposed as a risk factor of cerebral microvascular and brain damage in older adults (3, 41, 50, 51). A recent study also reported that older adults with MCI had a higher cerebral pulsatility index when compared with the age-matched normal participants partly due to the increased systemic vascular stiffness and endothelial dysfunction (52). In this regard, our previous studies have reported that 1-yr moderate-to-vigorous AET reduced CBFV pulsatility in older adults with MCI (39). Thus, under these conditions, whether AET may reduce high level of pulsatile CBF associated with reduction in cerebrovascular impedance (i.e., due to reductions in cerebral arterial stiffness and increases in vascular buffering effects discussed above) needs to be determined in future studies.

We acknowledge several limitations in this study. First, changes in CBFV measured by TCD represent changes in CBF only when the insonated arterial diameter remains relatively constant. This study did not measure MCA diameter and could not determine whether volumetric CBF was altered after the intervention.(41) Second, we measured CBFV only at the MCA and could not determine how flow velocity or vascular impedance might change in different cerebral arteries (e.g., the anterior and posterior cerebral arteries after AET or SAT). Third, we could not examine the sex-related differences in exercise training effects because of the relatively small sample size, although we previously reported that cerebrovascular impedance differed between men and women (11). Furthermore, participants were mainly well-educated Caucasians. Thus, the findings of this study need to be confirmed in future studies with a large sample of diverse racial and ethnic backgrounds. Finally, both subject dropout (∼23%) and technical issues with TCD have led to a discrepancy in the sample size from the pre- to postintervention, which may affect the group intervention balance achieved with the randomization. We conducted both the intent-to-treat and the complete-case analysis to assess the sensitivity of the outcome measures to the differences in the sample size before and after 1-yr interventions.

In conclusion, this RCT found that 1-yr AET reduced cerebrovascular impedance in sedentary older adults. Given the importance of cerebrovascular function for brain health, regular aerobic exercise represents an important lifestyle strategy for preventing or slowing brain aging in our society.

SUPPLEMENTAL DATA

Supplemental Figs. S1–S3 and Supplemental Table S1: https://doi.org/10.6084/m9.figshare.19666251.

GRANTS

This study was supported in part by the National Institute of Health (NIH, R01HL102457, R.Z.) and the Japan Society for the Promotion of Science (JSPS, 16KK0011, 17H02186, J.S.).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

T.T. and R.Z. conceived and designed research; T.T., C.X., J.L., and R.Z. performed experiments; J.S., T.T., C.X., J.L., T.T., E.P.P., and R.Z. analyzed data; J.S., T.T., C.X., J.L., T.T., E.P.P., and R.Z. interpreted results of experiments; J.S. prepared figures; J.S. and T.T. drafted manuscript; J.S., T.T., C.X., J.L., T.T., E.P.P., and R.Z. edited and revised manuscript; J.S., T.T., C.X., J.L., T.T., E.P.P., and R.Z. approved final version of manuscript.