The effects of an exercise intervention on executive function among overweight adults with obstructive sleep apnea

Abstract

Obstructive sleep apnea (OSA) is associated with poorer executive function. This study examined the effects of a comprehensive exercise intervention on executive function in overweight adults with mild and moderate-to-severe OSA. Participants aged between 30 and 65 years, with a body mass index (BMI) ranging from 27 to 42 kg/m², participated in a 6-week exercise program. Standardized polysomnographic recording methods provided total Apnea–Hypopnea Index (AHI) and level of hypoxemia. Executive function was assessed using the NIH Toolbox Flanker Inhibitory Control Test. A submaximal treadmill exercise test evaluated cardiorespiratory fitness. Participants with baseline total AHI between 5 and 14.9 events/h were classified as mild OSA and participants with baseline total AHI 15 ≥ events/h were classified as moderate-to-severe OSA. Fifteen participants completed 18 exercise sessions. Significant differences between OSA categories at baseline were observed for sleep characteristics, but not for fitness or executive function. Wilcoxon signed rank tests showed significant increases in median values for the Flanker Test in the moderate-to-severe category only, z = 2.429, p < 0.015, η² = 0.737. Six weeks of exercise improved executive function in overweight individuals with moderate-to-severe OSA, but not in those with mild OSA.

Introduction

Obstructive sleep apnea (OSA) is a common disorder, characterized by repeated episodes of upper airway obstruction, reductions in ventilation, arousals, and oxyhemoglobin desaturations during sleep. It has a prevalence of 37% and 50% in men and women, respectively [1]. In addition to its cardiometabolic sequela, OSA is associated with neurocognitive impairment that may contribute to the development of neurodegenerative disease [2].

Specifically, an abundance of evidence suggests that OSA may impair executive function [2–4], the domain of cognitive functioning that focuses on the conscious control of thought and action [5]. Further, cognitive impairment increases with greater OSA severity [6] and may be moderated by three key mechanisms (i.e., intermittent hypoxia, excessive daytime sleepiness, and sleep fragmentation) that jointly contribute to the cardiovascular and metabolic injury that ultimately cause cognitive dysfunction [2].

On the contrary, exercise training offers a plethora of benefits for those with OSA. That is, it increases slow wave sleep, increases sleep efficiency, decreases daytime sleepiness, increases upper airway dilator muscle tone, reduces fluid accumulation in the neck, reduces body weight/ body fat, and reduces systemic inflammatory responses [7, 8]. Because of its beneficial effects, exercise is often recommended as a first line treatment for OSA [9]. Moreover, exercise has been shown to improve executive function among healthy older adults and in older adults with non-OSA-related cognitive impairments [10–12]. Colcombe and Kramer [13] found that the effects of exercise training on executive function were larger than those observed for the other domains of cognitive functioning. Moreover, Ludyga et al. [12] found that even a single bout of moderate aerobic exercise produced small to moderate improvements in executive function and that these improvements were larger in older than in younger individuals. Importantly, the mechanisms responsible for the detrimental effects of OSA on executive function (i.e., sleep fragmentation and hypoxemia) are improved with exercise training [14].

While previous experimental studies [6, 15] have examined the effects of exercise training on executive function among those suffering from OSA, findings have been conflicting. For example, Ueno-Pardi et al. [7] found improved executive function in adults with moderate to severe OSA following a 6-month exercise program consisting of both aerobic and strength components. In contrast, despite implementing a similar exercise program in a group OSA patients, Kline et al. [15] failed to observe any change in executive function after the 12-week intervention. While acknowledging that the length of the intervention differed between studies, we emphasize that the mean AHI of patients enrolled in each study differed considerably (Ueno-Pardi et al. [7] 45 events/hour vs Kline et al. [16], 32 events/hour), suggesting that beneficial effects of exercise on executive function may vary in relation to the severity of OSA. Indeed, sleep disordered breathing has been shown to moderate the relationship between fitness/exercise and executive functioning [16, 17].

Accordingly, the primary aim of this exploratory study was to examine the effects of a 6-week American College of Sports Medicine (ACSM) exercise program on executive function in overweight adults diagnosed with either mild or moderate-to-severe OSA. Second, we sought to determine whether the degree of improvement in executive function was related to the severity of OSA (as defined by AHI and level of hypoxemia) at baseline. To address these aims, we chose a highly reliable and valid measure of executive function, the NIH Toolbox Flanker Test, likely to reflect improvements in executive function following exercise training [18, 19].

Materials and methods

Participants

Participants aged between 30 and 65 years, with a body mass index (BMI) between 27 and 42 kg/m² were recruited from the campus and surrounding community to participate in a 6-week exercise program, designed to examine whether OSA severity would predict post-training vascular function changes [ClinicalTrials.gov ID: NCT03219749]. The study design, exclusion criteria and results of the parent study can be found elsewhere [20]. Of the 55 who enrolled in the parent study, 15 (5 males, 10 females) volunteered to participate in the ancillary study, addressing executive function. Written informed consent was obtained from all participants. The study was approved by the Institutional Review Board at Towson University (Approval number: 16-A095).

Polysomnography assessment

To assess OSA severity, polysomnography was performed on all participants prior to beginning the intervention. Standardized recording methods were utilized, and included continuous monitoring (Alice 6, Philips-Respironics) of left and right electro-oculogram, submental EMG, F3/A2, C3-A2, O1-A2, F4/A1, C4/A1, O2A1 electroencephalogram, anterior tibialis electromyogram, oronasal airflow as assessed by pressure sensitive nasal cannula and a thermistor, pulse oximetry and thoracic and abdominal movements with piezo-electric gauges, and a modified V5 electrocardiogram lead for cardiac rhythm monitoring. Sleep studies were scored using criteria set forth by the American Academy of Sleep Medicine scoring manual Version 2.5. Hypopneas were scored when associated with a 3% desaturation or arousal based on the recommended definition. Sleep characteristics, including total AHI (calculated as the number of apneas and hypopneas per hour of sleep), mean oxyhemoglobin saturation (Mean SpO₂) during sleep, minimum oxyhemoglobin saturation (Minimum SpO₂), and percentage of total sleep time under 90% oxyhemoglobin saturation (Under 90%), were examined as markers of OSA severity.

Executive function

Executive function (i.e., attention and inhibitory control) was assessed using an iPad app running the Flanker Inhibitory Control Test (i.e., the Flanker Test). The Flanker Test is an assessment tool included in the NIH Toolbox’s Cognition Battery; Version 1.19; www.Healthmeasures.net/nih-toolbox). The Flanker Test has been used in previous studies demonstrating deficits in cognitive functioning in individuals with OSA. For example, Tulek et al. [21] measured executive function using the Simon, Flanker, and Stroop Tests in 14 healthy adults and 24 participants with OSA. The study found that those with OSA performed worse on the Flanker Test but not on the Stroop or Simon Test compared to controls. Similarly, Chou et al. [22] assessed executive function via Flanker task in 25 patients with OSA and 12 controls matched for age, sex, and education level. The study reported poorer performance on the Flanker Test in those with OSA than the controls.

As per the NIH Toolbox Scoring and Interpretation Guide, the Flanker Test “requires the participant to focus on a given stimulus while inhibiting attention to stimuli… flanking it. Sometimes the middle stimulus is pointing in the same direction as the ‘flankers’ (congruent) and sometimes in the opposite direction (incongruent)” [23] [pp. 6–7]. The Flanker Test consists of twenty trials and takes approximately three minutes to administer. The Flanker Test provides both computed and normative score output. As per the NIH Toolbox Scoring and Interpretation Guide, because they “provide a way of gauging raw improvement or decline from Time 1 to Time 2” [23] [p. 8] computed (i.e., absolute) scores for the Flanker Test were the focus of this study. The NIH Toolbox Cognition Battery has demonstrated high levels of reliability and validity [18, 19]. Specifically, Ott [18] found small to moderate correlations between Flanker Test scores and scores on several other commonly used tests of attention an executive function, including the Stroop Interference, the Trails B, and the Dimensional Change Card Sort tests.

Aerobic fitness

A submaximal treadmill exercise test evaluated cardiorespiratory fitness at baseline and following the intervention. Heart rate and oxygen consumption were measured continuously throughout the test and recorded at 20-s intervals. Participants were encouraged to continue exercising until they reached 85% of their heart rate reserve, at which point the exercise test was stopped and participants began a 5-min cool down. Maximal oxygen consumption was estimated by linear extrapolation of the line expressing the relationship between heart rate and oxygen consumption.

Intervention

A 6-week exercise intervention was carried out in accordance with the recommendations of the ACSM. Briefly, all participants attended three monitored exercise sessions per week (18 sessions in total). For the aerobic phase, participants spent 40 min per session (120 min × week ⁻¹) engaged in moderate to vigorous intensity exercise, defined as 60–90% of heart rate reserve. Each exercise session was monitored by an exercise physiologist to ensure adherence to the desired intensity. Resistance level was initially set at 60% 1 repetition maximum and progressively increased. Each exercise session ended with a set of abdominal crunches to fatigue and five minutes of cool down on an aerobic apparatus. The entire exercise session lasted between 60 and 65 min. Complete details of the exercise training stimulus can be found here [20].

Statistical analysis

Following clinical guidelines establishing 15 events/h as a threshold for initiating OSA treatment in adults [9], participants were classified into two OSA severity categories. Participants with baseline total AHI (events/hour) between 5 and 14.9 events/h were classified as having mild OSA. Participants with baseline total AHI 15 > events/h were classified as having moderate-to-severe OSA. Analyses were performed on all completers using SPSS version 24 and, given the small sample size, the nonparametric approach was utilized for all analyses conducted.

Mann–Whitney U tests were used to assess differences in baseline characteristics (measures of executive function, cardiorespiratory fitness and sleep) according to categories of OSA severity. Eta squared provided effect size estimates.

Wilcoxon signed rank tests were used to assess the effects of 6 weeks of exercise on measures of executive function and cardiorespiratory fitness according to categories of OSA severity. Two-sided p values were reported according to an alpha level of 0.025 with Bonferroni correction for two independent tests (i.e., baseline versus 6 weeks for the low severity category and baseline versus 6 weeks for the high severity category) per variable. Eta squared provided effect size estimates.

Kendall’s tau-b correlations were used to determine the relationships between sleep measures and changes (i.e., 6 weeks minus baseline/baseline) in executive function and cardiorespiratory fitness. Positive change scores indicated an increase, and negative change scores indicated a decrease in executive function and/ or cardiorespiratory fitness from baseline to 6 weeks.

Results

Fifteen participants (age: 51.6 ± 9.9 years; BMI: 35.4 ± 3.4 kg/m²; total AHI: 21.9 ± 12.9 events/h) completed the 18 exercise sessions. Table 1 presents baseline measures of sleep characteristics, executive function, and cardiorespiratory fitness, according to OSA severity. Per our design, median values for mean SpO₂ and minimum SpO₂ were lower in the high severity category (U = 45, z = − 3.188, p = 0.0005, η² = 0.726, and U = 8.5, z = − 2.186, p = 0.026, η² = 0.341, respectively). Also consistent with our design, median values for time spent below 90% oxygen saturation and total AHI were higher in the high severity category (U = 53, z = 3.067, p = 0.001, η² = 0.672, and U = 54, z = 3.185, p = 0.0005, η² = 0.725, respectively). No significant differences between OSA categories were observed for cardiorespiratory fitness or executive function at baseline.

Table 1.

Baseline executive function, cardiorespiratory fitness, and sleep expressed according to OSA severity

	OSA severity levels				p value#
	Total AHI: 5–14.9 events/h (n = 6, 2 M/4 F)		Total AHI: ≥ 15 events/h (n = 9, 3 M/6 F)
	Median	95% CI	Median	95% CI
Sleep
Mean SpO2 (%)	96	95–96	92	92–95	0.0005*
Minimum SpO2 (%)	89	86–90	82	79–87	0.026*
Under 90% SpO2 (% TST)	1	0–3	3	2–5	0.001*
Total AHI (events/ hour)	12	6–13	29	24–34	0.0005*
Executive function
Flanker inhibitory control test (i.e., computed scores, AU)	8.3	7.8–8.6	8.0	7.9–8.4	0.388
Cardiorespiratory fitness
BMI (kg/m2)	36	36–38	36	37–42	0.699
VO2 max (ml/kg/min)	28	27–30	29	25–33	0.852

AU arbitrary units, BMI body mass index, SpO₂ oxygen saturation, TST total sleep time, AHI Apnea–Hypopnea Index

^#P values for Mann–Whitney U Tests comparing total AHI categories

^*p < .05

Baseline and 6-week executive function and cardiorespiratory fitness according to OSA severity

Table 2 presents baseline and 6-week measures of executive function and cardiorespiratory fitness according to OSA severity. Wilcoxon signed rank tests for differences between baseline and 6 weeks showed significant increases in executive function in the moderate-to-severe OSA severity category. That is, there was a statistically significant increase in the median values for the Flanker Test in the moderate-to-severe category, z = 2.429, p < 0.015, η² = 0.737. However, there were no significant changes in either BMI, z = − 1.294, p < 0.196, or VO₂ max, z = 1.12, p < 0.263 in the moderate-to-severe category. Wilcoxon signed rank tests for differences between baseline and 6 weeks showed no significant changes in executive function or cardiorespiratory fitness in the mild severity category. That is, there were no significant changes in the Flanker test, z = 0.946, p < 0.344, BMI, z = 0.001, p < 1.00, or VO_2max, z = 1.782, p < 0.075, in the mild severity category.

Table 2.

Baseline and 6-week executive function and cardiorespiratory fitness expressed according to OSA severity

	Total AHI: 5–14.9 events/h					Total AHI: ≥ 15 events/h
	Baseline (n = 6)		6 weeks (n = 6)			Baseline (n = 9)		6 weeks (n = 9)
	Median	95% CI	Median	95% CI	p value#	Median	95% CI	Median	95% CI	p value#
Executive function
Flanker Test (AU)	8.3	7.8—8.6	8.3	7.9—8.5	.344	8.0	7.9—8.4	8.3	8.2—8.9	.015*
Cardiorespiratory fitness
BMI (kg/m2)	36	36 -38	36	35—39	1.00	37	37—42	35	33—38	.196
VO2max (ml/kg/min)	28	27—30	35	29—38	.075	29	25—33	30	27—34	.263

BMI body mass index, AU arbitrary units

^#P values for Wilcoxon signed rank tests for differences between baseline and 6 weeks

^*p < .025

Relationships between sleep and changes in executive function and cardiorespiratory fitness

Table 3 presents correlations between sleep characteristics and changes in cardiorespiratory fitness and executive function. As expected, several significant correlations were observed between various measures of sleep. Interestingly, a significant, positive correlation was observed between the amount of total sleep time spent under 90% SpO₂ and changes in Flanker test performance, τb = 440, p = 0.023.

Table 3.

Correlations between sleep characteristics, changes in cardiorespiratory fitness, and changes in executive function

	Minimum SpO2	Mean SpO2	Under 90% SpO2	Total AHI	∆Flanker Inhibitory Control test scores	∆BMI scores
Mean SpO2	0.380
Under 90% SpO2 (% TST)	− 0.563*	− 0.435*
Total AHI (events/h)	− 0.398*	− 0.348	0.606*
∆Flanker Inhibitory Control test scores	− 0.116	− 0.317	0.440*	0.230
∆BMI scores	0.108	0.161	0.043	− 0.053	0.298
∆VO2 max scores	0.322	0.167	− 0.309	− 0.221	− 0.055	− 0.226

P values for Kendall’s tau-b correlations

^*p < .05

Discussion

In this exploratory study, we found that 6 weeks of exercise improved executive function in overweight adults with moderate-to-severe OSA. Our study extends earlier work in several meaningful ways. First, we did not observe an improvement in executive function among those with mild OSA. Second, in contrast to Ueno-Pardi et al. [6], we found that improvement in executive function occurred independently of exercise-related changes in cardiorespiratory fitness (which occurred similarly across all participants). Third, we found that across the entire sample a higher level of hypoxemia (i.e., amount of time spent under 90% SpO₂ at baseline), but not total AHI, was correlated with a greater improvement in executive function. Last, the magnitude of the effect of exercise training on executive function was in line with the magnitude of the effect of traditional (i.e., continuous positive airway pressure [CPAP]) treatment on executive function [24].

Several experimental studies have examined the effects of exercise on executive function. Of note, Kleinloog et al. [25] reported decreased response latency following an 8-week exercise training intervention in healthy older adults. Also, Smiley-Oyen, Lowry, et al. [26] demonstrated similar effects on executive function after a 10 month exercise training intervention in healthy older adults. Both interventions utilized an intensity of exercise similar to the present intervention, e.g., (50 min duration/three times a week/70% peak VO₂) [27]. Baker et al. [28] examined the effects of a 6-month exercise intervention (45 min a day/4 days a week/75–85% of heart rate reserve) on several domains of cognitive functioning (including measures of executive function) in individuals with mild cognitive impairment. They found moderate to large effect sizes, indicative of exercise-related improvements in executive function.

Moreover, a number of other studies have found improvements in additional domains of cognitive functioning among individuals with cognitive impairments [29, 30]. On the contrary, a recent study employed a 6-week exercise intervention with older individuals with moderate dementia and found no positive effects on any domain of cognitive functioning [31]. These findings contrast with our findings and the previous work and it is possible that exercise stimulus, that is 30 min a day/5 days a week/self-selected walking speed, may not have been of sufficient intensity to impact cognitive functioning [31]. The observation that moderate-to-severe, but not mild, OSA patients improved executive function is a novel aspect of the current study. Ueno-Pardi et al. [6] focused on those with moderate-to-severe OSA. Although executive function and cardiorespiratory fitness were essentially the same across OSA severity groups at baseline, only those with more severe OSA showed improved executive function. This trend was further demonstrated by the observed correlation between time spent under 90% SpO₂ and changes in executive function. Together, these findings suggest that those with more severe OSA are more responsive to exercise training developed using ACSM endorsed exercise guidelines.

The observed improvement in executive function in those with moderate-to-severe, but not mild, OSA was not explained by differential increases in cardiorespiratory fitness. Our finding is consistent with those from previous studies showing that exercise training improved executive function independent of its impact on cardiorespiratory fitness [26, 32]. Cardiorespiratory fitness has previously been viewed as a physiological mediator that explains the cognitive performance benefits of physical activity [33]. Yet the scientific literature fails to support this “cardiorespiratory fitness hypothesis” [34]. Instead, it has been argued that cardiorespiratory fitness itself is not sensitive enough to predict changes in cognitive function with exercise, but rather leads to more specific mechanistic changes that result in enhanced delivery of oxygen to the brain [34]. To this end, a randomized, controlled crossover trial examining the effects of aerobic exercise training on cerebral blood flow found that exercise-related improvements in executive function were accompanied by concomitant increases in cerebral blood flow (increased oxygenation) in the frontal lobes [25]. Whether the observed correlations between OSA severity at baseline and degree of improvement in executive function with exercise could be explained by positive changes in cerebrovascular blood flow is a hypothesis that needs further investigation.

Interestingly, the effects of exercise training on executive function appear to be comparable to those of traditional treatment. CPAP treatment has been shown to improve executive function [24, 35]. Jiang et al. [24] reported moderate to large effect sizes for the effects of CPAP treatment on executive function, with larger effect sizes associated with long-term CPAP treatment. We observed a similar, large effect size for the effects of exercise training on executive function in those with moderate-to-severe OSA.

Our study had several strengths. First, all participants had a polysomnographic assessment of their sleep characteristics at baseline. This assessment provided an index of OSA severity, as well as sleep fragmentation and hypoxemia. Second, participants underwent a well-designed, supervised exercise intervention. It was individualized for their initial level of cardiorespiratory fitness and adjusted as they progressed. In addition, our exercise intervention included not only aerobic, but strength training activities. Third, we utilized a participant-friendly, reliable and valid assessment of executive function.

Our study also had several limitations. First, due to a small sample size, it was not feasible to explore more subtle (i.e., small effect size) research questions. That is, there was insufficient power to explore additional effects of the exercise intervention and/ or additional predictors of changes in executive function. Second, while there are multiple domains of cognitive functioning that are negatively impacted by OSA, we focused only on executive function. We made this choice primarily because of the detrimental effects of OSA on executive function, but also because of the beneficial effects of exercise training on executive function. Future research should examine the effects of exercise on other domains of cognitive functioning. Third, we did not repeat polysomnography testing following the intervention and therefore could not determine with certainty whether OSA improved with exercise. That is, we collected sleep data on our participants solely for the purpose of determining the severity of their OSA symptomatology and not for the purpose of determining the effects of exercise training on polysomnographic data.

In conclusion, we found that a 6-week exercise intervention improved executive function in overweight adults with moderate-to-severe, but not mild, OSA. The observed improvement in executive function occurred independent of changes in cardiorespiratory fitness and was associated with the severity of hypoxemia at baseline.

Funding

The study was sponsored by the National Heart, Lung, and Blood Institute under grant number 1R15HL133884-01 (Dobrosielski: PI). Support for the study was also provided by the College of Health Professions, Towson University. In addition, Author CP is a recipient of the Instituto de Salud Carlos III Miguel Servet fellowship (grant CP 19/00189).

Data availability

The dataset generated during and/or analysed during the current study is available from the corresponding author on reasonable request.

Declarations

Conflict of interest

KK declares that she has no conflict of interest. HP declares that she has no conflict of interest. SP declares that he has no conflict of interest. CP declares that he has no conflict of interest. DD declares that he has no conflict of interest.

Ethical approval

All procedures performed were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The study was approved by the Institutional Review Board at Towson University (Approval number: 16-A095).

Informed consent

Written informed consent was obtained from all participants included in the study.