Time-efficient, high-resistance inspiratory muscle strength training increases cerebrovascular reactivity in midlife and older adults

Abstract

Aging is associated with increased risk for cognitive decline and dementia due in part to increases in systolic blood pressure (SBP) and cerebrovascular dysfunction. High-resistance inspiratory muscle strength training (IMST) is a time-efficient, intensive respiratory training protocol (30 resisted inspirations/day) that lowers SBP and improves peripheral vascular function in midlife/older adults with above-normal SBP. However, whether, and by what mechanisms, IMST can improve cerebrovascular function is unknown. We hypothesized that IMST would increase cerebrovascular reactivity to hypercapnia (CVR to CO₂), which would coincide with changes to the plasma milieu that improve brain endothelial cell function and enhance cognitive performance (NIH Toolbox). We conducted a 6-wk double-blind, randomized, controlled clinical trial investigating high-resistance IMST [75% maximal inspiratory pressure (PI_max); 6×/wk; 4 females, 5 males] vs. low-resistance sham training (15% PI_max; 6×/wk; 2 females, 5 males) in midlife/older adults (age 50–79 yr) with initial above-normal SBP. Human brain endothelial cells (HBECs) were exposed to participant plasma and assessed for acetylcholine-stimulated nitric oxide (NO) production. CVR to CO₂ increased after high-resistance IMST (pre: 1.38 ± 0.66 cm/s/mmHg; post: 2.31 ± 1.02 cm/s/mmHg, P = 0.020). Acetylcholine-stimulated NO production increased in HBECs exposed to plasma from after vs. before the IMST intervention [pre: 1.49 ± 0.33; post: 1.73 ± 0.35 arbitrary units (AU); P < 0.001]. Episodic memory increased modestly after the IMST intervention (pre: 95 ± 13; post: 103 ± 17 AU; P = 0.045). Cerebrovascular and cognitive function were unchanged in the sham control group. High-resistance IMST may be a promising strategy to improve cerebrovascular and cognitive function in midlife/older adults with above-normal SBP, a population at risk for future cognitive decline and dementia.

NEW & NOTEWORTHY Midlife/older adults with above-normal blood pressure are at increased risk of developing cognitive decline and dementia. Our findings suggest that high-resistance inspiratory muscle strength training (IMST), a novel, time-efficient (5–10 min/day) form of physical training, may increase cerebrovascular reactivity to CO₂ and episodic memory in midlife/older adults with initial above-normal blood pressure.

INTRODUCTION

Aging is the primary risk factor for cerebrovascular diseases (e.g., stroke), cognitive decline, Alzheimer’s disease, and other related dementias (1, 2). The greater risk for these disorders with advancing age is driven, in part, by age-related increases in systolic blood pressure (SBP) and cerebrovascular dysfunction (3, 4). The rapid aging of our population intensifies the need for strategies that can combat these risk factors and dampen the projected rise in cognitive impairment and dementia burden (2, 5).

Conventional aerobic exercise training is a well-established strategy for improving cardiovascular, metabolic, and skeletal muscle health and function. Importantly, regular aerobic exercise also may promote cerebrovascular and cognitive health with aging (6). However, despite the expansive health benefits of aerobic exercise, adherence is poor at the population level because of the weekly time commitment required to meet current guidelines (7, 8). As such, novel therapies that are both efficacious and time efficient are needed. Our laboratory studies high-resistance inspiratory muscle strength training (IMST), an abbreviated (5–10 min/day) form of high-intensity physical training comprising inspiratory efforts against resistance (9, 10). In a pilot study (11), we found that 6 wk of high-resistance IMST (30 inspirations/session; 6 days/wk) was highly adherable, lowered SBP by an average of 9 mmHg, and improved in vivo endothelial function in midlife/older (ML/O) adults (age 50–79 yr) with initial above-normal SBP (≥120 mmHg), i.e., a group at increased risk for cognitive decline and cerebrovascular diseases (4). We also found that endothelial cell function, defined as acetylcholine-stimulated nitric oxide (NO) production in cultured human umbilical vein endothelial cells, was improved after exposure to serum from participants after IMST, suggesting that IMST may induce changes to the circulating milieu that improve endothelial cell function (11).

It has been reported that moderate-intensity inspiratory muscle training improves markers of dynamic cerebrovascular regulation in older women (12, 13). However, it is unknown whether time-efficient, high-resistance IMST improves cerebrovascular function with aging in humans. Importantly, the mechanisms by which high-resistance IMST may mediate increases in cerebrovascular function, including changes in the circulating milieu, have not been investigated.

Circulating concentrations of brain-derived neurotrophic factor (BDNF), an important growth factor involved in neurogenesis, synaptic function, and neurovascular health, decline with advancing age (14, 15). Some forms of physical training, including high-intensity interval training (HIIT), increase BDNF levels (16), but it is unknown whether the hypothesized improvements in cerebrovascular function with IMST are associated with corresponding increases in BDNF. Finally, because cerebrovascular dysfunction is thought to be a mechanism contributing to impaired cognitive function with aging (17), it is possible that IMST-related improvements in cerebrovascular function may correspond with increases in cognition. However, this has never been assessed.

Here we tested the hypothesis that 6 wk of high-resistance IMST would improve cerebrovascular function in vivo, as characterized by an increase in cerebrovascular reactivity to hypercapnia (CVR to CO₂), in otherwise healthy midlife/older adults with above-normal SBP. We also hypothesized that this effect would be associated with improvements in brain endothelial cell function induced by changes in the circulating plasma milieu, increases in plasma BDNF, and/or enhanced cognitive performance.

METHODS

This study was approved by the Institutional Review Board at the University of Colorado Boulder (IRB No. 17-0151). Participants from a randomized, double-blind, sham-controlled parallel group designed pilot trial to determine the efficacy of IMST for lowering SBP underwent exploratory investigation of cerebrovascular and cognitive function before and after 6 wk of high-resistance IMST or low-resistance sham training (11). Because of a hardware failure of the transcranial Doppler, it was not possible to perform pre- and postintervention cerebrovascular measurements in all participants. As such, here we are reporting on a subset of the overall cohort (11). All data related to cerebrovascular and cognitive function have not been reported previously, whereas participant characteristics data have been published for the overall study cohort.

Sixteen men and postmenopausal women (age 50–79 yr; ≥1 yr amenorrheic and not taking hormone replacement therapy) performed either high-resistance IMST [all IMST participants started at 55% maximal inspiratory pressure (PI_max) and increased to 65% PI_max during week 2 and then 75% PI_max during weeks 3–6 of training] or low-resistance sham training (15% PI_max) using the POWERbreathe K3 tapered-resistance device (POWERbreathe, United Kingdom), a device that enables storage of training session data for monitoring adherence. Both groups performed 30 resisted inspirations/day, 6 days/wk (1 supervised training session and 5 at-home training sessions), for 6 wk; expiration was unimpeded. Participants were generally healthy and were screened for cognitive impairment with the Mini Mental State Examination (average score = 29/30). Group randomization was performed with a block randomization scheme based on age, sex, and baseline SBP by a study team member who was not involved in analyzing study outcomes. PI_max was assessed weekly in all participants, and absolute resistance was adjusted by an unblinded research assistant. Participants taking medications (e.g., antihypertensives or statins) were enrolled if their treatment regimen was stable for >3 mo and remained stable throughout the study. Investigators involved in data collection and analysis for all vascular, cognitive, cell culture, and plasma outcomes were blinded to participant group allocation.

Measurements

All data were collected at the University of Colorado Boulder Clinical Translational Research Center or Integrative Physiology of Aging Laboratory after written informed consent was obtained. Measurements were performed before and after 6 wk of high-resistance IMST or low-resistance sham training. Vascular function was measured and blood samples were collected after participants abstained from food and caffeine (water allowed) for ≥12 h and alcohol, strenuous exercise, and over-the-counter medications and supplements for ≥24 h. Cognitive function was assessed ∼2 h after a small meal or snack and ≥24 h after abstaining from alcohol and strenuous activity. Experiment start times were standardized for each participant across the intervention. Postintervention testing was performed 24–48 h after the last IMST or sham training session. Physical activity was measured by accelerometry (GENEActiv; Activinsights, UK) before and after the intervention to ensure that levels were unchanged. Training power and lung volume were calculated automatically by the POWERbreathe K3 device. Training adherence was defined as the number of training sessions completed/number of training sessions prescribed (36 sessions).

Participants rested in a dark, quiet, temperature-controlled (21°C) room for at least 10 min before vascular measurements. Endothelium-dependent dilation was measured via flow-mediated dilation of the brachial artery in response to 5 min of forearm cuff occlusion by high-resolution ultrasonography (PowerVision 6000, Toshiba, Tokyo, Japan or iE33, Philips, Eindhoven, The Netherlands), as previously described (11). Middle cerebral artery blood velocity (MCAv) was measured with a 2-MHz transcranial-Doppler probe (Neurovision, Multigon, Elmsford, NY) placed on the left transtemporal window while participants were in a semirecumbent position (18). The coefficient of variation for within-subject MCAv in our laboratory is 3 ± 2%. Individuals breathed through a mask attached to a one-way nonrebreathing valve (Hans Rudolph, Shawnee, OK) for 5 min each of normocapnia (room air) and hypercapnia (5% CO₂, 21% O₂, balance nitrogen). End-tidal pressure of carbon dioxide ($\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$) was measured through a mouthpiece attached to a breath-by-breath CO₂ gas analyzer (17630; VacuMed, Ventura, CA). Mean arterial blood pressure (MAP) was measured with an automated oscillometric sphygmomanometer (Cardiocap 5; General Electric, Boston, MA) during normocapnia and hypercapnia, and MAP reactivity was calculated as ΔMAP/Δ$\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$. MCAv and $\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$ were averaged over the last minute of normocapnia and hypercapnia. CVR to CO₂ was determined as ΔMCAv/Δ$\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$ (absolute response) or %ΔMCAv/Δ$\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$ (relative response) from normocapnia to hypercapnia and interpreted as an index of cerebrovascular function and health (19). To account for potential changes in perfusion pressure occurring with the IMST intervention, CVR to CO₂ was normalized to the change in MAP from normocapnia to hypercapnia and expressed as cerebrovascular conductance index (CVCi) reactivity to CO₂. Pulsatility index, a measure of pulsatile blood velocity entering the cerebrovasculature that is strongly related to arterial stiffness, was calculated as (MCAv_systole − MCAv_diastole)/MCAv_mean (20).

To determine whether changes to the circulating milieu following the high-resistance IMST intervention may contribute to improvements in brain endothelial cell function, ex vivo experiments were performed whereby human brain endothelial cells (HBEC-5i, CRL-3245; ATCC, Manassas, VA) were treated with plasma obtained from participants before and after IMST or sham training. This cell line was isolated from the cerebral cortex of a brain devoid of any pathological abnormalities. These cells have been validated for their maintenance of endothelial cell characteristics and used in multiple in vitro models of the human blood-brain barrier (21, 22). HBECs were plated in 96-well culture plates and incubated under standard conditions (37°C, 5% CO₂) for 2 h in basal media supplemented with 10% participant plasma. Plasma samples for each participant were tested using three experimental replicates and averaged. After treatment, cells were incubated with a fluorescent probe (DAR-4M AM; Enzo Life Sciences, Farmingdale, NY) to detect NO production before and 6 min after addition of 100 µM acetylcholine (Sigma-Aldrich, Burlington, MA) to the cell culture media. Our laboratory has verified the specificity of this florescent NO marker, as the acetylcholine-stimulated increase in signal is abolished in the presence of the NO synthase inhibitor nitro-l-arginine methyl ester (l-NAME) (23). As previously published, acetylcholine-stimulated NO production was interpreted as an ex vivo model of endothelial cell function (11). Plasma BDNF was measured by ELISA (212166; Abcam, Cambridge, UK) to investigate its role in IMST-related improvements in cerebrovascular function.

Domains of cognitive function were assessed via the National Institutes of Health (NIH) Toolbox Cognition Battery on the NIH Toolbox iPad app (24). We used the core tests within this standardized battery, each of which has established excellent response reliability and validity (24, 25). The battery was administered by a trained study team member matched for pre- and postintervention visits. This battery takes our participants ∼60 min to complete. The flanker test measured inhibitory control and attention, the picture sequence memory test measured episodic memory, the list sorting test measured working memory, the picture vocabulary and oral reading recognition tests measured language, the dimensional change card sort test measured executive function, and the pattern comparison test measured processing speed. Uncorrected standard scores were used to assess changes after versus before the IMST or sham interventions [normative mean = 100, standard deviation (SD) = 15].

Statistical Analyses

Statistical analyses were performed with SPSS version 27 (IBM, Armonk, NY). The impact of IMST on outcome variables was evaluated by repeated-measures analysis using two-way ANOVA, with group (IMST vs. sham), time (pre- vs. postintervention), and group × time interactions entered as the independent variables. As this was an exploratory arm of a larger clinical trial, Fisher’s least significant differences post hoc analyses were used for within- and between-group comparisons regardless of significant interaction effects. Repeated-measures analysis of covariance (ANCOVA) was used to assess changes across the intervention while controlling for baseline MCAv. Between-group mean differences in training variables (e.g., training load) were assessed with unpaired t tests because these values were only assessed during the intervention. To gain insight into the magnitude of the effect of our intervention on the outcomes of interest, Cohen’s d effect sizes were calculated. Cohen’s d = 0.2–0.49 was considered a small effect, d = 0.5–0.79 was considered a medium effect, and d ≥ 0.8 was considered a large effect. Data are means ± SD. Statistical significance was set a priori at α = 0.05.

RESULTS

Participants

Participant characteristics are shown in Table 1. There was a main effect of time for SBP (P = 0.006), with the IMST group showing a decrease in SBP post- versus preintervention (P = 0.003) in post hoc comparisons. There was a main effect of time (P = 0.037) and an interaction effect (P = 0.028) for brachial artery flow-mediated dilation, such that the IMST group had increased brachial artery flow-mediated dilation post- versus preintervention (P = 0.003). There was an interaction effect for PI_max (P = 0.021). Post hoc analyses indicated that PI_max differed between groups at baseline (P = 0.034) and increased in the IMST group post- versus preintervention (P = 0.006). There was a main effect of time for physical activity (P = 0.049). No other group, time, or interaction effects, or post hoc comparisons, were significant for participant characteristics. Because of the COVID-19 shutdown, one participant (IMST) was not able to provide a posttesting plasma sample or perform cognitive testing; thus cerebrovascular function outcomes are shown for 16 participants but plasma experiments and cognitive function data are shown for 15 participants.

Table 1.

Participant characteristics

	Sham		IMST
	Baseline	Posttesting	Baseline	Posttesting
N	7	-	9	-
Age, yr	63 ± 6	-	65 ± 9	-
Sex, females/males	2/5	-	4/5	-
BMI, kg/m2	27.8 ± 4.1	28.2 ± 4.3	24.9 ± 2.4	25.0 ± 2.1
Heart rate, beats/min	60 ± 3	59 ± 6	70 ± 13	67 ± 12
SBP, mmHg	131 ± 6	128 ± 10	133 ± 8	123 ± 9*
DBP, mmHg	81 ± 5	81 ± 2	81 ± 7	80 ± 8
Total cholesterol, mg/dL	158 ± 26	161 ± 28	176 ± 15	172 ± 20
HDL cholesterol, mg/dL	44 ± 13	43 ± 12	53 ± 18	53 ± 20
LDL cholesterol, mg/dL	94 ± 18	96 ± 18	104 ± 14	99 ± 37
Triglycerides, mg/dL	98 ± 39	111 ± 37	93 ± 37	100 ± 56
Fasting blood glucose, mg/dL	94 ± 7	91 ± 6	89 ± 6	88 ± 8
Flow-mediated dilation, Δ%	6.5 ± 3.4	6.4 ± 3.2	5.7 ± 2.5	8.4 ± 3.4*
Activity, total minutes light, moderate, vigorous	586 ± 147	704 ± 358	402 ± 164	649 ± 214
PImax, cmH2O	83.3 ± 12.2#	80.9 ± 10.7	62.2 ± 21.2	73.7 ± 18.2*
Respiratory rate, breaths/min	8.4 ± 3.4	8.3 ± 4.0	9.9 ± 3.0	9.5 ± 3.6
Training load, cmH2O	-	17.4 ± 2.1#	-	66.8 ± 16.5
Training power, W	-	5.7 ± 2.8	-	7.5 ± 3.6
Training lung volume, L	-	2.3 ± 0.5#	-	1.5 ± 0.5
Training adherence, %	-	88 ± 12	-	96 ± 8
Antihypertensive treatment, n	2	-	5	-
Statin treatment, n	1	-	1	-

Values are means ± SD; n, number of participants. BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HDL, high-density lipoprotein; LDL, low-density lipoprotein; PI_max, maximal inspiratory pressure. Statistics are 2-way repeated-measures ANOVA with Fisher’s least significant difference post hoc analysis or unpaired t test for single time point variables. *P < 0.05 vs. within-group baseline; #P < 0.05 vs. inspiratory muscle strength training (IMST) at matched time point.

In Vivo Cerebrovascular Reactivity to Hypercapnia

Absolute CVR to CO₂ increased after 6 wk of high-resistance IMST (pre: 1.38 ± 0.66 cm/s/mmHg, post: 2.31 ± 1.02 cm/s/mmHg; within group, P = 0.020; d = 1.08) but was unchanged with low-resistance sham training (pre: 1.20 ± 0.21 cm/s/mmHg, post: 1.48 ± 0.62 cm/s/mmHg; within group, P = 0.504; d = 0.60) (Fig. 1A). There was a main effect of time for absolute CVR to CO₂ (P = 0.041) but no main effect of group (P = 0.055) or group × time interaction (P = 0.244). When controlling for baseline MCAv, the effect of IMST on absolute CVR to CO₂ remained (within group: P = 0.004). There was a main effect of time (P = 0.020) and a group × time interaction (P = 0.039), but no main effect of group (P = 0.240), for absolute CVR to CO₂ in the ANCOVA model.

Figure 1.

A and B: cerebrovascular reactivity to hypercapnia (CVR to CO₂; A) and cerebrovascular conductance index reactivity (CVCiR; B) to hypercapnia before and after 6 wk of low-resistance sham training (n = 7) or high-resistance inspiratory muscle strength training (IMST) (n = 9). C: fold change in acetylcholine (ACh)-stimulated nitric oxide (NO) production of cultured human brain endothelial cells exposed to plasma obtained from participants before and after 6 wk of low-resistance sham training (n = 7) or high-resistance IMST (n = 8). Each plasma sample was tested using 3 experimental replicates and averaged. Dark gray circles indicate individual data at baseline, and gold circles indicate individual data at posttesting. Statistics are 2-way repeated-measures ANOVA with Fisher’s least significant difference post hoc analysis.

Relative CVR to CO₂ was lower in the IMST group compared with the sham group at baseline (IMST, pre: 2.79 ± 0.94%, sham, pre: 5.21 ± 2.37%; between group, P = 0.014). After 6 wk, relative CVR to CO₂ increased significantly in the IMST group (post: 5.64 ± 3.29%; within group, P = 0.015; d = 1.18) but did not change in the sham group (post: 4.87 ± 2.21%; within group, P = 0.775; d = 0.15). There were no group (P = 0.375), time (P = 0.130), or group × time interaction (P = 0.060) effects. Importantly, with statistical adjustment for baseline MCAv, the relative CVR to CO₂ response was no longer different between groups at baseline (IMST, pre: 3.13 ± 2.00%; sham, pre: 4.78 ± 2.01%; between group, P = 0.178) and significantly increased in the IMST group (post: 6.90 ± 2.83%; within group, P = 0.006; d = 1.54), with no change in the sham training group (post: 3.26 ± 2.94%; within group, P = 0.278; d = 0.60), such that CVR to CO₂ was greater in the IMST group compared with the sham training group after the intervention (between group: P = 0.046). There was a group × time interaction (P = 0.020) but no main effects of group (P = 0.347) or time (P = 0.063).

CVCi reactivity increased after versus before the IMST intervention (pre: 0.009 ± 0.006 cm/s/mmHg², post: 0.018 ± 0.012 cm/s/mmHg²; within group, P = 0.025; d = 0.93) but was unchanged with sham training (pre: 0.009 ± 0.002 cm/s/mmHg², post: 0.010 ± 0.005 cm/s/mmHg²; within group, P = 0.683; d = 0.44) (Fig. 1B). There were no group (P = 0.160), time (P = 0.069), or group × time interaction (P = 0.201) effects. The effects of IMST on CVCi reactivity were similar when controlling for baseline MCAv (within group: P = 0.033). There were no group (P = 0.593), time (P = 0.261), or group × time interaction (P = 0.199) effects for CVCi reactivity in the ANCOVA model.

MAP reactivity was not different after versus before the intervention in either group (IMST, pre: 0.65 ± 0.94 mmHg/mmHg, post: 1.13 ± 0.48 mmHg/mmHg; within group, P = 0.090; d = 0.65; sham, pre: 0.78 ± 0.70 mmHg/mmHg, post: 1.09 ± 1.16 mmHg/mmHg; within group, P = 0.486; d = 0.32). There were no group (P = 0.656), time (P = 0.095), or group × time interaction (P = 0.442) effects.

MCAv, CVCi, MCA pulsatility index, MAP, and $\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$ during normocapnia and hypercapnia can be found in Table 2. There was a main effect of group for MCAv (normocapnia, P = 0.006; hypercapnia, P = 0.008) but no time (normocapnia, P = 0.783; hypercapnia, P = 0.338) or group × time interaction (normocapnia, P = 0.060; hypercapnia, P = 0.590) effects. Post hoc tests revealed differences in MCAv at baseline (normocapnia, between group, P = 0.004; hypercapnia, between group, P = 0.017) and after the intervention (normocapnia, between group, P = 0.024; hypercapnia, between group: P = 0.009). There was a main effect of group for CVCi (normocapnia, P = 0.027; hypercapnia, P = 0.026) but no time (normocapnia, P = 0.579; hypercapnia, P = 0.838) or group × time interaction (normocapnia, P = 0.123; hypercapnia, P = 0.465) effects. Post hoc tests showed differences in CVCi at baseline (normocapnia, between group, P = 0.021; hypercapnia, between group, P = 0.035) and select differences after testing (normocapnia, between group, P = 0.068; hypercapnia, between group, P = 0.044). There were no group (normocapnia, P = 0.275; hypercapnia, P = 0.162), time (normocapnia, P = 0.147; hypercapnia, P = 0.075), or group × time interaction (normocapnia, P = 0.815; hypercapnia, P = 0.471) effects for MCA pulsatility index. There were no group (P = 0.560), time (P = 0.849), or group × time interaction (P = 0.316) effects for MAP during normocapnia. There was a group × time interaction effect (P = 0.045) but no group (P = 0.699) or time (P = 0.152) effects for MAP during hypercapnia. Post hoc tests showed greater MAP during hypercapnia post versus pre in the IMST group (within group, P = 0.021). There were no group (normocapnia, P = 0.989; hypercapnia, P = 0.969), time (normocapnia, P = 0.332; hypercapnia, P = 0.906), or group × time interaction (normocapnia, P = 0.192; hypercapnia, P = 0.421) effects for $\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$.

Table 2.

Hemodynamic and gas exchange variables during the cerebrovascular reactivity to hypercapnia test

	Sham		IMST
	Baseline	Posttesting	Baseline	Posttesting
Normocapnia
MCAv, cm/s	25.5 ± 6.8#	31.1 ± 3.8#	49.5 ± 17.6	45.2 ± 14.3
CVCi, cm/s/mmHg	0.27 ± 0.07#	0.30 ± 0.04	0.49 ± 0.22	0.43 ± 0.17
MCA pulsatility index, AU	1.46 ± 0.82	1.27 ± 0.70	1.14 ± 0.39	1.00 ± 0.18
MAP, mmHg	108 ± 6	106 ± 6	108 ± 12	108 ± 14
$\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$, mmHg	36.0 ± 3.9	38.6 ± 2.9	37.5 ± 1.7	37.1 ± 5.1
Hypercapnia
MCAv, cm/s	37.4 ± 6.3#	42.5 ± 8.0#	61.8 ± 23.1	63.3 ± 16.6
CVCi, cm/s/mmHg	0.34 ± 0.08#	0.38 ± 0.06#	0.57 ± 0.25	0.55 ± 0.20
MCA pulsatility index, AU	1.29 ± 0.64	1.04 ± 0.34	0.98 ± 0.28	0.87 ± 0.15
MAP, mmHg	115 ± 6	114 ± 10	113 ± 11	117 ± 15*
$\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$, mmHg	45.4 ± 3.4	46.4 ± 2.8	46.3 ± 1.9	45.5 ± 3.4

Values are means ± SD. Data were collected with participants in the semirecumbent position. MCAv, middle cerebral artery blood velocity; CVCi, cerebrovascular conductance index; AU, arbitrary units; MAP, mean arterial pressure; $\text{P}\text{E}{\text{T}}_{{\text{CO}}_2}$, end-tidal pressure of carbon dioxide. Statistics are 2-way repeated-measures ANOVA with Fisher’s least significant difference post hoc analysis. *P < 0.05 vs. within-group baseline; #P < 0.05 vs. inspiratory muscle strength training (IMST) at matched time point.

Ex Vivo Brain Endothelial Cell Function

We next assessed the role of intervention-induced changes to the circulating plasma milieu in transducing beneficial effects on cultured brain endothelial cell function. NO production in response to acetylcholine increased in HBECs exposed to plasma from participants after versus before the IMST intervention [pre: 1.49 ± 0.33, post: 1.73 ± 0.35 arbitrary units (AU); within group, P < 0.001; d = 0.69] and decreased in HBECs exposed to plasma from participants after versus before the sham intervention (pre: 1.56 ± 0.27, post: 1.41 ± 0.25 AU; within group, P = 0.006; d = 0.59) (Fig. 1C). There was a group × time interaction (P < 0.001) but no main effects of group (P = 0.424) or time (P = 0.245).

Brain-Derived Neurotrophic Factor

We also investigated whether BDNF, an important molecule promoting neurovascular health, may play a role in the beneficial effects of IMST on cerebrovascular function. However, plasma BDNF was unchanged after 6 wk of IMST (pre: 4,280 ± 2,171 pg/mL, post: 4,465 ± 2,641 pg/mL; within group, P = 0.911; d = 0.08) and sham training (pre: 6,228 ± 4,052 pg/mL, post: 7,329 ± 4,170 pg/mL; within group, P = 0.537; d = 0.27). There were no group (P = 0.073), time (P = 0.598), or group × time interaction (P = 0.706) effects.

Cognitive Function

Picture sequence memory score, a measure of episodic memory, increased modestly after IMST (within group, P = 0.045; d = 0.53) but was unchanged after sham training (within group, P = 0.416; d = 0.22). There were no group (P = 0.838), time (P = 0.053), or group × time interaction (P = 0.382) effects. All other domains of cognitive function assessed by the NIH Toolbox were not different after IMST (P > 0.291) or sham training (P > 0.088) (Table 3).

Table 3.

National Institutes of Health Toolbox Cognition Battery scores

Cognitive Test	Cognitive Domain	Sham		IMST
		Baseline	Posttesting	Baseline	Posttesting
Flanker	Inhibitory control and attention	102 ± 4	105 ± 5	99 ± 5	99 ± 6
List sorting	Working memory	98 ± 12	102 ± 6	100 ± 13	104 ± 10
Picture sequence memory	Episodic memory	99 ± 11	102 ± 18	95 ± 13	103 ± 17*
Picture vocabulary	Language	123 ± 7	121 ± 10	122 ± 7	123 ± 7
Oral reading recognition	Language	114 ± 3	115 ± 5	117 ± 3	116 ± 4
Dimensional change card sort	Executive function	106 ± 9	105 ± 9	105 ± 7	104 ± 8
Pattern comparison	Processing speed	103 ± 13	106 ± 7	94 ± 16	97 ± 10

Values are means ± SD and uncorrected standard scores: normative mean = 100, and SD = 15. IMST, inspiratory muscle strength training. Statistics are 2-way repeated-measures ANOVA with Fisher’s least significant difference post hoc analysis. *P < 0.05 vs. within-group baseline.

DISCUSSION

The present findings show that 6 wk of high-resistance IMST increases CVR to CO₂, induces changes to the circulating plasma milieu that promote ex vivo brain endothelial cell function, and may modestly improve episodic memory in otherwise healthy ML/O adults with above-normal SBP. Collectively, these results suggest that high-resistance IMST may be a promising strategy for improving cerebrovascular health, including cerebrovascular endothelial cell function, and possibly episodic memory in a population at increased risk of future cognitive decline and dementia because of elevated SBP.

In Vivo Cerebrovascular Reactivity to Hypercapnia

This is the first study of its type to assess the effects of high-resistance IMST on cerebrovascular function. We observed an increase in CVR to CO₂ after only 6 wk of 5–10 min/day of high-resistance IMST in ML/O adults with above-normal SBP. CVR to CO₂ is an important marker of cerebrovascular function, with higher values associated with reduced risk of Alzheimer’s disease, dementia, and mortality (26, 27); thus, high-resistance IMST may be an efficacious, time-efficient strategy for improving cerebrovascular health.

To our knowledge, one previous study has assessed the effects of a HIIT on CVR to CO₂ in older adults (28). The authors reported only a slight increase in relative CVR to CO₂ (4%) compared with the 120% increase observed in the present study. Moreover, the time commitment was 60–90 min/wk for 3 mo compared to only ∼30 min/wk for 6 wk with the present intervention. A recent meta-analysis found that 3–7 mo of more time-intensive, moderate-intensity aerobic exercise training (∼90–200 min/wk) failed to evoke a statistically significant improvement in CVR to CO₂ compared with nonexercise control subjects (29). Importantly, because participants’ overall physical activity patterns, objectively assessed by accelerometry, were similar across both our IMST and sham interventions, any improvement in cerebrovascular function with IMST is not attributable to enhanced general physical activity. A larger, multiyear clinical trial is currently underway to evaluate the effects of high-resistance IMST on cerebrovascular function and establish its comparative effectiveness to conventional aerobic exercise (NCT04848675). Although high-resistance IMST should not be viewed as a replacement for conventional aerobic exercise, its time efficiency, portability, safety, and relative cost-effectiveness make it an attractive option for individuals limited by those barriers.

The mechanisms by which high-resistance IMST increases CVR to CO₂ require further investigation. There is evidence that IMST at moderate intensities (e.g., 50% PI_max) can modulate cardiac autonomic control and improve respiratory muscle pump responses (30), which may underlie improvements in components of dynamic cerebrovascular function (12, 13). Additionally, reductions in intrathoracic (31) and intracranial (32) pressure with resistance breathing may promote oscillations in cerebral blood flow (33–35) and increase shear stress within the cerebrovasculature (36), which could enhance arterial dilation and cerebrovascular function (37, 38). Previous studies using the same protocol as the present intervention have observed decreases in muscle sympathetic nerve activity and systemic vascular resistance after 6 wk of high-resistance IMST (39, 40), which may be involved in the blood pressure-lowering effects of IMST. We did not observe any changes in circulating catecholamines in this group of midlife/older adults (11), though it is possible that decreased sympathetic outflow and vascular resistance may be involved in IMST-associated increases in CVR to CO₂. Finally, although the 6-wk IMST intervention did not improve arterial stiffness, as assessed by either carotid-femoral pulse wave velocity or carotid artery compliance (11), whether IMST induced changes to MCA vessel structure is unknown. However, it is generally hypothesized that longer intervention durations (>3 mo) are needed to evoke structural changes to the vasculature. Future trials are needed to better understand the stimulus/stimuli elicited by high-resistance IMST that evoke(s) favorable adaptations in cerebrovascular function.

Ex Vivo Brain Endothelial Cell Function

Exposure to plasma obtained from participants after high-resistance IMST increased acetylcholine-stimulated NO bioavailability in brain endothelial cells compared with a modest decrease in the low-resistance sham group, suggesting that changes to the circulating milieu induced by high-resistance IMST can promote cerebral endothelial cell function. NO is a crucial vaso- and neuroprotective molecule involved in endothelial cell health and the prevention of cerebrovascular diseases, such as stroke. Although NO may play a permissive, rather than obligatory, role in CVR to CO₂ (41), NO production in response to chemical or mechanical stimuli is a key component of endothelium-dependent dilation and vascular health (42). Thus, the results of our ex vivo endothelial function experiments are viewed as an initial step in understanding the role of the circulating milieu in modulating brain endothelial cell function in humans, rather than as an attempt to establish NO as a specific mechanism of the IMST-related improvements in CVR to CO₂.

Future research should identify which circulating factors may be responsible for the beneficial effects of IMST on cerebral endothelial cell function. We investigated circulating BDNF, a neuroprotective molecule that is reported to increase in response to some HIIT interventions in healthy young adults (16) and patient populations [e.g., stroke (43)]. However, our results indicate that 6 wk of high-resistance IMST does not change concentrations of resting plasma BDNF in healthy ML/O adults. The divergent findings between IMST and whole body HIIT may be due to the sample type used for BDNF analyses (plasma vs. serum), the timing of sample collection after exercise, the population tested, and/or differences in the work or metabolic demand associated with the two types of training. Changes to oxidative stress and inflammation are two potential mechanisms that may be modulated in the circulation in response to IMST. Additionally, peripheral levels of factors involved in blood vessel growth and neuronal health (e.g., vascular endothelial growth factor or insulin-like growth factor) may play a role in exercise-induced structural and functional changes in the brain (44, 45). Whether IMST modulates these, or other, circulating factors requires further investigation.

Cognitive Function

Our findings suggest that episodic memory, a domain of cognitive function closely linked to age-related cognitive decline and Alzheimer’s disease (46), may be improved slightly by high-resistance IMST. Midlife hypertension is associated with increased risk of late-life dementia (4). Thus, our study participants, ML/O adults with above-normal SBP, are considered a population at risk of future dementia. It is estimated that 40% of dementia cases may be prevented or delayed by the management of risk factors such as hypertension and vascular dysfunction (47). Thus, identifying effective strategies to lower SBP and improve vascular health has great potential for lowering dementia burden. Our laboratory and others have shown consistent effects of high-resistance IMST for lowering SBP (11, 48). To our knowledge, this is the first study to gain initial insight into the effects of high-resistance IMST on cognitive function. It is unclear why IMST improved only one domain of cognitive function in our study. One possibility is that physical activity mainly affects brain structures sensitive to neurodegeneration such as the hippocampus (49), the brain region largely responsible for episodic memory, suggesting that IMST also may have select impacts on brain health. The improvement also may be mediated by changes in cerebrovascular function. Longitudinal evidence suggests that declines in CVR to CO₂ are related to declines in episodic memory (50). Thus, CVR to CO₂ may be a useful predictor of future cognitive decline, and interventions that modulate CVR to CO₂ may hold promise for improving cognitive performance as well. The sham group did not significantly improve episodic memory from pre- to postintervention. This may be the result of an insufficient stimulus compared with IMST, although a slightly (nonsignificant) higher baseline score in the sham group versus the IMST group may have influenced the response. As such, these data should be interpreted with caution, as our subset of participants was underpowered to detect differences between groups. Future studies with larger sample sizes are needed to rigorously assess the effects of IMST on cognitive function and ensure the absence of learning effects.

Experimental Considerations

We recognize that the inspiratory resistance elicited by our sham intervention may acutely affect cerebral blood flow (36). However, we observed no changes in cerebrovascular function or cognitive performance in our low-resistance sham group, indicating that this training intensity served as an appropriate control. Nonetheless, future trials may consider implementing a different control group when assessing the cerebrovascular effects of high-resistance IMST. Because of the small sample size and exploratory nature of these outcomes, the chance for group differences in baseline MCAv was inherently possible; however, statistical adjustment for these differences did not change the findings. The use of transcranial Doppler limits our ability to determine whether group differences in MCAv were due to differences in MCA diameter or whether IMST induced changes to vessel diameter or dilation responses. This method operates under the assumption that vessel diameter does not change, although vasoactive stimuli such as hypercapnia can evoke changes to MCA diameter (51, 52). As such, it is possible that the present measurements of CVR to CO₂ using blood velocity may have underestimated IMST-induced improvements in cerebrovascular function. Assessment of CVR to CO₂ using methods that measure vessel diameter (e.g., duplex ultrasound or MRI) are warranted to determine the potential effects of IMST on cerebral blood flow and vasodilation. Cerebrovascular measurements were made in the semirecumbent position, which may have resulted in lower MCAv values compared with supine posture (53); however, the observed values are within the physiologically plausible range for seated older adults (54). There was a main effect of time for accelerometry-measured physical activity, primarily driven by increased light-intensity physical activity. However, modest increases were observed in both the IMST and sham groups, suggesting that changes in physical activity could not explain the beneficial cerebrovascular effects observed with IMST. Finally, the present study was not properly powered to determine sex-specific effects of the IMST intervention on our outcomes of interest. Postmenopausal women exhibit greater lifetime burden of stroke and make up nearly two-thirds of patients with Alzheimer’s disease in the United States (1, 2, 55). As such, interventions that improve cerebrovascular and cognitive function in women with aging are a biomedical research priority.

Conclusions

Together, the findings from this pilot study suggest that high-resistance IMST may be a promising, time-efficient form of physical training for improving cerebrovascular function in ML/O adults with above-normal SBP, a population at increased risk of cerebrovascular diseases such as stroke as well as Alzheimer’s disease and other related dementias. Moreover, changes to the circulating plasma milieu after high-resistance IMST may be responsible for improvements in cerebrovascular endothelial cell function, as assessed ex vivo. Finally, short-term (6 wk) IMST may modestly improve episodic memory in this population. These preliminary findings are in line with our observations in the peripheral vasculature, suggesting that similar benefits can be evoked by high-resistance IMST in the cerebral circulation.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

This work is supported by National Institutes of Health Grants R21AG061677, R01AG071506, T32DK007135, UL1TR002535, F31HL154782, K01HL153326, and K01DK115524 and American Heart Association Grant 18POST33990034.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

K.A.F., D.H.C., M.J.R., E.F.B., and D.R.S. conceived and designed research; K.A.F., D.H.C., T.C.H., R.A.J., L.R.J., and K.R.L. performed experiments; K.A.F. and D.H.C. analyzed data; K.A.F., D.H.C., and D.R.S. interpreted results of experiments; K.A.F. prepared figures; K.A.F. and D.H.C. drafted manuscript; K.A.F., D.H.C., T.C.H., M.J.R., R.A.J., L.R.J., K.R.L., M.C., E.F.B., and D.R.S. edited and revised manuscript; K.A.F., D.H.C., T.C.H., M.J.R., R.A.J., L.R.J., K.R.L., M.C., E.F.B., and D.R.S. approved final version of manuscript.