Can menopausal status moderate the effects of aerobic exercise on executive functions? evidence from a randomized controlled trial

Abstract

Background

Physical exercise is widely recognized for its cognitive benefits; however, the effect of menopausal status in modulating the cognitive effects of exercise is not definitively established.

Objective

To examine the cognitive benefits of two 6-month physical exercise programs in cognitively healthy older women across adulthood, and whether menopause status moderates these benefits.

Methods

In a post hoc analysis of a randomized controlled trial, 93 cognitively healthy women (aged 20–67; 43% at post-menopause) were assigned to either aerobic exercise (AE) or stretching/toning (ST) 4 days a week for six months. Neuropsychological assessment, cardiorespiratory exercise test, and blood draw were performed at baseline, 3-months, and 6-months. Linear mixed-effects regression models assessed whether menopausal status moderated the impact of exercise on executive functions and processing speed.

Results

Seventy-six participants (81.7%) completed the intervention. A time-by-group-by-menopause interaction emerged after 3 months (β = −0.89; p = 0.001) and 6 months (β = −0.67, p = 0.016). Post-menopausal women in the AE group showed greater improvement in executive functions compared to the ST group and pre-menopausal women. Models controlled for age, education, and baseline cognitive performance.

Conclusions

Our results provide novel evidence that AE improves cognition with pronounced executive functions benefits in post-menopausal women, a population at higher risk for dementia. Since women are at a higher risk of developing Alzheimer's disease compared to men, these findings support AE as a relevant strategy to promote women's brain health. Although this is a secondary analysis, it may inform future exercise trials targeting women.

Introduction

Approximately 45% of dementia risk is linked to 14 modifiable factors, ¹ among them physical inactivity in midlife and later life. ¹ In this context, physical exercise is a beneficial lifestyle intervention, as it is accessible to diverse populations. Furthermore, data suggest that physical exercise, particularly aerobic exercise (AE), promotes cardiovascular and brain health, including cognitive benefits.^2,3 These benefits have been more consistently reported in executive functions, although evidence exists in other domains, such as processing speed, attention, and episodic memory.^2–6 Critically, an extensive literature^3,6 indicates several physiological effects of AE that may support these cognitive benefits, including improved cerebral blood flow and VO₂peak, increased gray matter volume and thickness, reduction in adipose tissue, and upregulation of neurotrophic factors such as brain-derived neurotrophic factor (BDNF), critical for neuroplasticity and hippocampal neurogenesis.

Despite the established cognitive benefits of AE, its effects can vary^4,7 due to differences in prescription parameters (e.g., type, frequency, intensity, and duration) and individual factors, such as baseline cognitive abilities and biological sex. ³ Biological sex influences various aspects of lifestyle and biological aging, including resilience to proteinopathies, brain changes, cognitive decline, and ultimately dementia risk.^1,8 Furthermore, previous evidence indicates that women (defined in this report as biologically female) persistently lag behind men (defined here as biologically male) in physical activity engagement, suggesting a “sex/gender gap” that begins early in life and continues throughout adulthood. ⁹ Women's exercise habits during menopause are reported to further decline, including a decrease in physical activity and exercise engagement.^10,11 This decline has been attributed to the discomfort of some physical and emotional menopause symptoms, but also a lack of adequate opportunities and access to facilities, high cost, safety fears, and a busy schedule. Critically, other evidence indicates that women at and after menopause represent the segment of the population with the lowest levels of physical activity. ¹²

Of note, previous evidence indicates that women have double the risk of developing dementia compared to men. ⁸ A meta-analysis suggests sex-based differences in the effects of exercise on executive functions, visuospatial abilities, and word fluency. ¹³ Our previous report, which examined sex moderation of AE effects, found greater cognitive benefits in men than in women across adulthood, particularly in executive functions and processing speed. ¹⁴ Although in our previous report, older age was associated with greater AE benefits on cognition in both men and women, in women, this benefit was only observed for participants 50 years and above. ¹⁴ Even though our previous report is based on a small sample, it suggests that further research should investigate if there is a unique biological process in women that may interact with AE effects on cognition after the age of 50. However, it is also relevant to consider that older adults, regardless of sex, may have greater potential for cognitive gain after AE due to lower baseline aerobic capacity and cognitive performance.

Menopause, clinically defined as 12 consecutive months without menses, ¹⁵ is a transition in a woman's life, typically occurring around age 51 in the US (with a natural occurrence typically between 45 and 55 years old). ¹⁵ The Stages of Reproductive Aging Workshop (STRAW) classification ¹⁶ has been widely adopted in clinical and research settings, and defines seven stages of adult women's lives into three broad categories: reproductive (pre-menopause), menopausal transition (peri-menopause), and post-menopause.

During perimenopause, fluctuating estrogen levels have been correlated with hot flashes, sleep disturbances, and cognitive decline, including executive functions, attention, and memory. ¹⁷ Other common symptoms, such as mood swings, migraines, and dizziness, also may negatively impact quality of life. ¹⁵ Factors such as age, socioeconomic status, and pre-existing health conditions can exacerbate these symptoms, underscoring the need for personalized treatments. ¹⁵ Furthermore, women may experience widespread changes in brain structure and function during perimenopause, including cognitive decline. ³ While cognitive issues usually resolve in post-menopausal years, some women may experience persisting symptoms and are at increased risk of long-term cognitive impairment. ¹⁸

Although it is still an area of debate, there is evidence that a decline in estrogen levels at menopause increases the risk for several conditions, including cardiovascular diseases, cognitive decline, and neurodegenerative diseases, including Alzheimer's disease and related dementia (ADRD).^17,19 Post-menopausal women experience over twice as many cardiovascular events compared to pre-menopause, ²⁰ and a higher burden of white matter hyperintensities, also linked to ADRD risk. ²¹ Conversely, estrogen is consistently associated with neuroprotective properties, including brain plasticity, brain structure, and cognitive health.^22,23 Nevertheless, the independent effect of menopause status on brain health, cardiovascular health, and ADRD risk remains unclear, as chronological age is an inherent confounder of menopause effects and a strong risk factor for several health conditions, including ADRD.

Given the potential for physical activity, including AE, to counteract or at least minimize the detrimental effects of menopause on neurocognitive and cardiovascular health, ²⁴ a growing body of evidence is beginning to show that these benefits may be particularly relevant for women undergoing the menopausal transition. For instance, randomized clinical trials (RCTs) involving physical activity in postmenopausal women observed benefits on executive functions, speed processing, and episodic memory,^25,26 although this research area remains sparse. ²⁷ The potential cognitive benefits of physical exercise in postmenopausal women are consistent with reduced neuroplasticity in hippocampal regions that may be partially compensated with exercise-induced increases in brain-derived neurotrophic factor (BDNF) and synaptogenesis, ²⁸ improved cerebrovascular function, ²⁹ anti-inflammatory and neuroprotective effects on brain tissues.^30,31 Therefore, executive functions, processing speed, and episodic memory are relevant candidates for AE-related benefits in peri and post-menopausal women.

The detrimental effects of menopause on women's overall health highlight the critical need to investigate hormonal changes and/or menopause stages when investigating the cognitive effects of physical exercise. Nevertheless, menopause is frequently overlooked in exercise research, with a surprising lack of clinical trials focused on the cognitive benefits of exercise in pre-/peri- and postmenopausal women. This research gap is crucial for advancing ADRD prevention. The current study aims to address this gap by investigating the potential effects of menopausal status on AE-cognitive effects. Furthermore, as an exploratory aim, we also assessed AE effects on estradiol levels. This post-hoc analysis examined data from a previously completed RCT designed to investigate the cognitive benefits of AE across adulthood.^6,14

Methods

Study design

This study is a randomized, parallel-group, observer-masked controlled trial. The detailed methodology and design are described in the original report ⁶ and presented in Figure 1. The trial's primary aim was to compare the cognitive effects of six months of AE against a stretching/toning (ST) control condition in cognitively healthy adults with below-average fitness. In the original report, ⁶ 132 participants, aged 20 to 67 years, were equally randomized into the AE or ST exercise conditions. Randomization was stratified by age and biological sex. For the current analysis, only female participants’ data were analyzed (N = 93), and we compared the cognitive effects of AE in menopausal and non-menopausal women. This did not include non-cisgender individuals. Of note, although in our previous report the cognitive benefits were associated with changes in VO₂ peak, ⁶ the goal of the current report was only the cognitive outcomes.

Figure 1.

Study design.

BPASS: Baecke Physical Activity Sports Score; CPET: Cardiopulmonary Exercise Test to determine maximal oxygen consumption (VO₂peak); DRS: Mattis Dementia Rating Scale.

The study protocol and informed consent form were approved by the Institutional Review Board of the New York State Psychiatric Institute, and all participants provided written informed consent before enrollment. The trial was registered on Clinicaltrials.gov (NCT01179958) and was overseen by an independent data and safety monitoring board under the auspices of the National Institute on Aging. This study report adheres to the Consolidated Standards of Reporting Trials (CONSORT) guidelines ³² and the Consensus on Exercise Reporting Template (CERT). ³³

Recruitment

As previously reported, ⁶ participants were recruited by social media and posted flyers. Eligibility criteria included being healthy (e.g., absence of cardiovascular disease, brain disorder, and psychiatric condition, and several other conditions and medications listed in Supplemental Material and on ClinicalTrials.gov), cognitively intact, nonsmoking, with below-average cardiovascular fitness based on CPET assessment, following the American Heart Association and American College of Sports Medicine standards (VO₂peak < 35.2, 33.8, 32.3, 29.4, and 26.6 mL/kg/min for women age 20–29, 30–39, 40–49, 50–59, and 60–69 years, respectively). Post-menopausal women were required not to be on hormonal replacement therapy (HRT - estrogen). Initial telephone screening determined that participants met basic inclusion criteria. Participants were excluded if they reported regularly exercising (Baecke Physical Activity Sports Score >2 on this 5-point scale) or presented MRI contraindications, required for ancillary imaging studies.

Study assessments and randomization

Potential participants attended an in-person visit where they provided written informed consent and completed the Mattis Dementia Rating Scale. Participants with scores below 135 were excluded to ensure a cognitively healthy sample. ⁶ After screening, participants completed baseline assessments to determine fitness level (maximal oxygen consumption or VO₂peak) through the Cardiorespiratory Exercise Test (CPET), neuropsychological functioning, and health profile, including menopausal status and blood draw to measure ovarian hormone levels. Subsequently, they were randomized into one of two six-month conditioning programs: aerobic exercise (AE) or stretching/toning (ST), using random block assignment stratified by sex. At 3-month and 6-month assessments, the participants repeated the neuropsychological assessment, CPET, and blood draw. Furthermore, although MRI scans were performed at baseline and at 6 months; however, this data was not included in this report. Details ⁶ are added as Supplemental Material and Supplemental Figure 1.

Cardiorespiratory exercise test

Maximal oxygen consumption (VO₂peak) was measured by CPET on an Ergoline 800S electronic-braked cycle ergometer (SensorMedics Corporation, Anaheim, CA). An individualized ramping protocol (10, 15, or 20 watts each two minutes) was selected according to each participant's perceived exercise capacity (ascertained by an exercise physiologist based on the participant's age, prior exercise history, and familiarity with cycling) to yield a test duration of approximately 10 min. Each subject began the test with a 3-min warm-up against no resistance, and the work rate was then linearly increased at the individualized ramp rate. Participants were encouraged to maintain a pedal cadence >50 rpm, when they dropped below this threshold, the test was terminated (volitional fatigue).

Maximal exercise capacity was determined by having all subjects achieve at least 2 of the following: >90% peak predicted heart rate (220 – age), maximal exertion with limitation due to dyspnea, maximum exertion with muscle fatigue, RER >1.10, or a plateau in the slope of the VO₂-work-rate relationship. The highest 15-breath moving median in the last minute of exercise was considered VO₂peak. The baseline CPET was used to establish safe exercise parameters and heart rate (HR) targets for training and to ensure that participants had below-average fitness by American College of Sports Medicine standards, as detailed in our original publication. ⁶ In addition, a normal baseline electrocardiogram (ECG) was required. Any ischemic changes, abnormal blood pressure responses, or significant ectopy resulted in exclusion.

Neuropsychological assessment

A full description and references of the selected tasks are contained in the original paper. ⁶ The Wechsler Test of Adult Reading (WTAR) was administered at the initial assessment to estimate IQ. Tests were chosen from domains sensitive to AE manipulations, including executive function, processing speed, attention, working memory, and episodic memory. ⁶ The measures used for each cognitive domain were selected before initiating analyses. This selection was validated using factor analysis of the baseline cognitive data, which produced groupings comparable to those in the cognitive domains listed below. Of note, the factor analysis confirmed that the local switch cost loaded on the executive functions factor and not the processing speed factor, supporting their empirical separability despite the reaction-time component of switch costs. Standardization of each test was based on the mean and standard deviation of baseline values. Mean values of the standardized outcomes of the tests in each domain were used for analysis.

A detailed description of each cognitive test used is presented in the Supplemental Material. In brief, executive functions measures included a local switch cost from a Set-Switching task and total rule break errors across five trials of the Groton Maze Learning Test (CogState). Processing speed measures included the number of symbols copied from the Digit Symbol task (WAIS-III Digit Symbol subtest); the number of clicks/second from the Groton Maze Chase Test (CogState); and reaction time from the Identification Task (CogState). Language measures included total words produced across three letters on the Controlled Oral Word Association and animal names produced in 60 s on the Animal Naming. Attention was assessed with the total number of correct cancellations within 5 min on the Two and Seven Test. Working Memory was assessed with the total number of correct sequence trials on the WAIS-III Letter Number Sequencing and correct responses on the N-back task (CogState). Episodic Memory measures included delayed recall from the Modified Rey Auditory Verbal Learning Test and the number of errors from the Continuous Paired Associate Learning task (CogState). Cogstate tasks had multiple forms across visits; all other tasks remained the same.

Hormone level assays

Blood samples were analyzed using multiplex assays using the Luminex™ 200 system (Luminex, Austin, USA), and a Milliplex human cytokine kit (MilliporeSigma, Burlington, USA) by Eve Technologies Corporation (Calgary, Canada). For each serum sample, duplicate assays were conducted to ensure the reliability of the results. Assays were conducted according to manufacturers’ instructions using assay kits. Assay details are available on the website. Serum estradiol was measured using radioimmunoassays.

Menopausal status classification

Menopausal status was classified based on a self-report binary question on whether participants were at menopause or not (i.e., yes or no). Although our study follows the STRAW criteria, ¹⁶ the self-report data collected were not specific enough to differentiate the three broad phases: reproductive, menopausal transition, and post-menopause. Only one participant voluntarily declared to be in peri-menopause; the others reported yes or no to the menopausal status question. Therefore, we are considering two groups in the current analysis: pre- versus post-menopause. The participant that self-declare in perimenopause was classified as pre-menopause for the current analysis. Furthermore, two participants who did not provide data on menopausal status were classified as post-menopause due to their ages (64 and 67 years old). We acknowledge that age alone does not definitively confirm menopausal status; however, at ages 64 and 67, post-menopausal classification is highly probable.

Intervention procedures

Run-in period

Qualified participants then proceeded to the two-week run-in period, during which they attended their choice of a YMCA fitness center in Manhattan (NYC), four times/week for two weeks. Each participant was assigned a fitness center coach who provided detailed instruction on stretches, consisting of arm circles, neck rotations, toe reach, gluteal stretches, lateral leg swings, Achilles stretches, and ankle rolls. All participants wore a Polar heart rate monitor during the stretching sessions. Coaches monitored participants’ adherence to the protocol via gym attendance records and heart rate (HR) monitor data. Only those participants who attended at least 7 sessions were permitted to continue in the study. In our experience, including the run-in period resulted in improved adherence.

Exercise program structure and heart rate measurement

After the run-in period, participants in both groups started the exercise programs at the YMCA Fitness Centers for 4 sessions/week over 6 months (24 weeks), resulting in 96 sessions. The training sessions were individual, one hour in length, and adapted to participant's scheduling preferences. The fitness center trainers introduced the exercise programs to each participant. In both exercise conditions, the sessions consisted of 10–15 min of warm-up/cool-down and 45 min of workout. During each exercise session, participants’ HR was recorded with a Polar HR monitor (model RS400, Polar Electro, Kempele Finland). To account for missing data due to equipment malfunction, the date and time of the exercise session were cross-checked with the diary log. For additional details, see Supplemental Material.

Aerobic exercise group

To perform the AE, participants were asked to choose a training equipment of preference, e.g., elliptical trainer, treadmill, or stationary bicycle, to increase the level of compliance with the protocol. Participants were also asked to refrain from weightlifting or bodyweight workouts. For each participant, the exercise intensity (prescribed as a percentage of maximal heart rate - HR_max), was progressively increased each week: 55%–65% (1^st and 2^nd weeks), 65%–75% (3^rd and 4^th weeks), and 75% (5^th to 24^th weeks) of the measured HR_max achieved during the study entry CPET.

Stretching/toning group

Participants were asked to engage in a series of stretching and toning exercises designed to promote flexibility and improve core strength. Participants started their exercise sessions with 10 min of low-intensity warm-up on training equipment of their choice (e.g., elliptical trainer, treadmill, or stationary bicycle). They then proceeded to complete the stretching portion of the session. Each stretch position was specific to the subject's range of motion and was held for 20–30 s. Exercises targeted the entire body and focused on flexibility enhancement. As the participants’ level of flexibility increased, stretches with increasing levels of difficulty were incorporated into the program. Lastly, they completed core strengthening exercises, including abdominal, back, and pelvic muscles. The complete list of stretching/toning exercises is provided in Supplemental Figure 1.

Adherence

During both interventions, participants were contacted either through e-mail or phone calls by the research team. All participants were contacted at least once a week. Contact with the participants included the use of techniques designed to promote good adherence and encourage improvement if needed. Data from the exercise sessions were reviewed by the fitness coach in every session, and feedback was provided to participants to avoid exercise session issues (e.g., not reaching the target HR) from being repeated.

Statistical analyses

Using data from the previously completed trial, ⁶ we examined the effect of menopausal status on the observed exercise-related cognitive benefits. A modified intention-to-treat (available-case) approach was used, including all randomized participants with available data under a missing at random assumption. No data were imputed. Our analysis focused on investigating menopause moderation on the effects of exercise on executive functions and speed of processing, as these cognitive domains were the ones previously reported to be improved after our AE study.^6,14 Power analysis is not provided in this report, as post hoc power calculation is not recommended and may be misleading. ³⁴ Nevertheless, according to sample size rules of thumb for pilot RCTs ³⁵ that suggest at least 30 subjects are needed to estimate a parameter or 12 subjects per treatment arm, our analysis may be considered appropriately powered (details on sample sizes in Supplemental Table 1).

The baseline characteristics of the sample were described as a function of menopausal status using ANOVA and chi-square tests. To examine the main effects and interactions among menopausal status, time, intervention group, and cognition, we performed a linear mixed-effect regression including time (baseline versus 3-months versus 6-months), group (AE versus ST), menopause status (pre versus post) and their 2- and 3-way interactions as a fixed effect and random intercepts of subjects; controlling for baseline age, education and baseline cognition (model: Cognitive Change∼Menopausal Status*Time*Group + Baseline Age + Education + Baseline performance). Based on our original trial and previous reports,^6,14 the a-priori and primary analyses were focused on executive functions and processing speed domains; however, we also ran an exploratory analysis extending to the other cognitive domains: episodic memory, working memory, attention, and language. All cognitive scores are presented with Z-scores. Critically, we also ran a sensitivity analysis adding the potential age interaction to the previous model, which could help disentangle menopause from age moderations.

Furthermore, an exploratory analysis was run to test the interaction of exercise training and menopausal status on changes in estradiol levels, with group, time, menopausal status, and their 2- and 3-way interactions controlling for baseline age. The relevance of this analysis is that exercise may increase estrogen levels, ³⁶ so this could be a potential mechanism of cognitive change after AE. All statistical analyses were conducted using RStudio.

Results

Participants

Table 1 describes the baseline characteristics of the participants, with similar sample sizes in each exercise condition (N = 47 in AE; and N = 46 in ST); and no group differences regarding age, race/ethnicity, education, BMI, VO₂peak, and cognitive performance. Furthermore, the sample was described as a function of menopausal status (Table 2), with a similar percentage of post-menopause women in both exercise conditions: 38.2% (N = 18) in the AE group, and 35.5% (N = 16) in the ST group. As would be expected based on the classification criteria, the pre-menopause groups in both exercise conditions were younger and presented higher levels of estradiol than the post-menopause groups (p < 0.001). In addition, at baseline, the pre-menopause group showed better cognitive performance across both exercise conditions, particularly for executive functions, speed, and episodic memory (Table 2). Critically, there was no difference at baseline on menopause status regarding cardiovascular fitness, as all participants’ fitness level was classified at study enrollment, so VO₂peak was expected to be low in the entire sample (i.e., on average, lower than 25 ml/kg/min). In addition, as seen in Table 2, there is an age-range overlap between pre- and post-menopause groups, which mitigates the VO₂peak differences as a function of menopausal status.

Table 1.

Baseline women's characteristics by exercise conditions.

	Aerobic Exercise N = 47	Stretching/Toning Exercise N = 46	p
Age, years [Range]	42.98 (12.87) [20–64]	41.87 (15.03) [21–67]	0.70
Race/Ethnicity			0.09
Asian	5 (38.5%)	8 (61.5%)
Black	16 (64%)	9 (36%)
White	19 (54.3%)	16 (45.7%)
Latinx	4 (25%)	12 (75%)
Other	3 (75%)	1 (25%)
Education, years	16.17 (2.62)	16.59 (1.90)	0.38
Estimated IQ	111.79 (12.77)	110.96 (13.83)	0.76
BMI	26.63 (5.31)	25.84 (4.99)	0.47
VO2 Peak	23.62 (3.87)	24.98 (4.24)	0.36
Executive Functions	−0.24 (0.73)	−0.43 (0.53)	0.88
Speed	−0.02 (0.78)	0.10 (0.70)	0.39
Episodic Memory	0.13 (0.73)	0.09 (0.79)	0.79
Working Memory	−0.01 (0.78)	−0.00 (0.74)	0.94
Attention	0.13 (0.62)	0.02 (0.76)	0.44
Language	0.06 (0.94)	0.09 (0.93)	0.87

Missing data for BMI (body mass index): Aerobic Exercise = 2; Stretching/Toning group = 1.

Table 2.

Characteristics of the women participants at baseline as a function of menopausal status.

	Aerobic exercise N = 47			Stretching/Toning exercise N = 45
	Pre-menopause N = 29 (61.7%)	Post-menopause N = 18 (38.2%)	p	Pre-menopause N = 29 (64.4%)	Post-menopause N = 16 (35.5%)	p
Age, years [Range]	34.48 (7.54) [20–49]	56.67 (5.73) [44–64]	<0.001	32.17 (8.90) [21–51]	56.94 (5.79) [46–67]	<0.001
Education, years	16.55 (2.48)	15.56 (2.79)	0.20	16.21 (1.63)	17.38 (2.18)	0.04
Estimated IQ	110.72 (12.26)	113.50 (13.72)	0.47	110.00 (12.97)	112.88 (15.93)	0.51
BMI	26.63 (5.89)	26.63 (4.47)	0.99	25.98 (5.61)	25.54 (3.89)	0.78
VO2max	26.70 (7.17)	23.40 (3.91)	0.08	27.84 (6.63)	24.86 (4.48)	0.10
Estradiol	3.49 (1.01)	1.73 (1.07)	<0.001	3.36 (1.00)	2.01 (1.19)	<0.001
Speed Processing	0.25 (0.63)	−0.37 (0.84)	<0.001	0.33 (0.62)	−0.32 (0.67)	0.002
Exec. Functions	0.24 (0.48)	−0.42 (0.89)	0.002	0.07 (0.53)	−0.29 (0.47)	0.024
Episodic Memory	0.36 (0.65)	−0.22 (0.79)	0.006	0.21 (0.71)	−0.11 (0.93)	0.195
Working Memory	0.04 (0.82)	−0.10 (0.74)	0.549	0.03 (0.81)	−0.06 (0.67)	0.680
Attention	0.11 (0.68)	0.18 (0.53)	0.712	0.07 (0.53)	−0.01 (0.47)	0.705
Language	0.001 (0.15)	0.15 (0.88)	0.587	0.01 (0.89)	0.27 (1.02)	0.373

Missing data for the Aerobic Exercise group: BMI (body mass index) = 2; Estradiol = 1. Missing data for the Stretching/Toning group: BMI = 1; Estradiol = 1; Estradiol values are log-transformed. Significant p-values < 0.05 are highlighted in bold.

The CONSORT diagram describes the selection process and overall reasons for exclusion in the study^6,14 (Figure 2). For the women sample, 81.7% (76/93) completed 3 months of exercise (N = 38 in AE; and N = 38 in ST), and 70.9% (66/93) completed the 6-month intervention (N = 32 in AE; and N = 34 in ST). There was no significant difference in the % of menopausal status across time within each group (AE: χ² = 0.55, p = 0.76; ST: χ² = 0.16, p = 0.92); for additional details of sample size by group and menopausal status, see Supplemental Table 1. Of note, in the original trial, our previous analysis of dropouts^6,14 revealed no difference by group, baseline cognitive performance, or demographic variables such as age, sex, ethnicity, or years of education. No adverse events occurred during exercise.

Figure 2.

CONSORT flow diagram by sex.

The flowchart is adapted from Stern et al., 2019. ⁶

Menopause moderation on exercise response

The linear mixed-effects models indicated a significant group × time × menopause interaction at both 3-months (β = −0.86; p = 0.002) and 6-months (β = −0.57; p = 0.04) in predicting executive functions (Table 3). As presented in Figure 3, these findings indicate that executive functions were more improved in the AE group relative to the ST group for the post-menopause participants than for the pre-menopause participants, controlling for the covariates (age, education, and baseline executive functions). No interaction with menopausal status was observed for the speed composite score at 3-months (β = −0.01; p = 0.95) and 6-months (β = −0.11; p = 0.59) (Table 4 and Figure 4). Of note, the menopause factor was not associated with executive functions or speed when testing the model excluding the interaction terms (for model details, see Supplemental Tables 3 and 4).

Table 3.

Statistics for linear mixed model with self-reported menopausal status (SMS) predicting executive functions.

Factors	Executive functions
	Estimate	Confidence Interval	p
Age	−0.01	−0.02 – −0.00	0.005
Education, y	0.01	−0.02–0.04	0.508
Self-Report Menopausal Status (SMS)	0.11	−0.26–0.49	0.551
Executive Functions Baseline	0.66	0.54–0.78	<0.001
Group	−0.08	−0.33–0.17	0.526
Time (3 months)	0.12	−0.12–0.35	0.164
Time (6 months)	0.34	0.08–0.60	0.010
Group × SMS	0.14	−0.28–0.56	0.505
Time (3 months) × SMS	0.40	0.02–0.77	0.038
Time (6 months) × SMS	0.34	0.05–0.73	0.087
Group × Time (3 months)	0.23	−0.09–0.56	0.164
Group × Time (6 months)	0.09	−0.27–0.44	0.628
Group × Time (3-months) × SMS	−0.86	−1.39 – −0.32	0.002
Group × Time (6-months) × SMS	−0.057	−1.12 – −0.01	0.045

Figure 3.

Executive functions change by menopausal status at 3-months and 6-months.

*p = 0.04, **p = 0.002.

Table 4.

Statistics for linear mixed model with self-reported menopausal status (SMS) predicting processing speed.

Factors	Speed
	Estimate	Confidence Interval	p
Age	−0.01	−0.02 – −0.00	0.021
Education, y	0.00	−0.02–0.02	0.932
Menopausal Status (MS)	0.13	−0.13–0.40	0.322
Speed Baseline	0.92	0.83–1.0	<0.001
Group	−0.01	−0.20–0.17	0.876
Time (3 months)	0.07	−0.10–0.25	0.415
Time (6 months)	0.09	−0.11–0.28	0.393
Group × MS	0.05	−0.26–0.35	0.764
Time (3 months) × MS	−0.03	−0.31–0.26	0.839
Time (6 months) × MS	−0.03	−0.33–0.27	0.846
Group × Time (3 months)	0.06	−0.19–0.31	0.623
Group × Time (6 months)	0.15	−0.12–0.42	0.267
Group × Time (3-months) × MS	−0.01	−0.42–0.40	0.956
Group × Time (6-months) × MS	−0.11	−0.54 – −0.31	0.596

Figure 4.

Processing speed change was not moderated by menopausal status.

Furthermore, exploratory analysis revealed no significant group × time × menopause interactions for all other cognitive domains, such as episodic memory (3-months: β = −0.16; p = 0.49 and 6-months: β = 0.05; p = 0.83), working memory (3-months: β = −0.06; p = 0.86 and 6-months: β = −0.14; p = 0.72), attention (3-months: β = 0.64; p = 0.30 and 6-months: β = 0.76; p = 0.23), and language (3-months: β = 0.23; p = 0.37 and 6-months: β = 0.11; p = 0.76).

Sensitivity analysis: age moderation

As expected by the menopause classification (based on self-report), menopausal status was correlated with age (r = 0.84; p < 0.001). Therefore, a sensitivity analysis was run, adding the potential age interaction to the previous model (already adjusted by age, sex, and education). This model could help disentangle the effects of menopause from age moderation. The fact that we observed a significant 3-way interaction (β = −0.86; p = 0.002) while accounting for the 4-way interaction (time by group by menopause by age moderation: β = −0.057; p = 0.045) in the executive functions model strongly suggests that menopausal and age moderations are independent. See Table 3 for detailed statistics. The linear mixed model indicated that the group × time × menopause interaction remained significant at 3 months (p = 0.001) and 6 months (p = 0.003) in predicting executive functions.

Exploratory analysis on changes in estradiol levels

To further understand the effect of menopause on aerobic exercise and executive functions, we ran an exploratory analysis to test changes in estradiol after the exercise program at 6 months, as estradiol levels may increase with exercise. ³⁷ Furthermore, whether time, group, and menopausal status moderated the change in estradiol was also investigated. We did not observe significant estradiol changes after both exercise programs (p > 0.05) nor any moderation effects (p > 0.05), or correlation with cognitive change (p > 0.05). For additional details, see Supplemental Table 2 and Supplemental Figure 2).

Discussion

To our knowledge, this study represents one of the few studies examining whether menopausal status influences the effects of aerobic exercise on cognition using data from a randomized clinical trial. After adjusting for key confounders of age, education, and baseline cognitive performance, we found that post-menopausal women showed a greater benefit of AE on executive functions in comparison to pre-menopausal women. These effects were larger than those observed in the control condition (ST group). Furthermore, although our report is based on a small sample, a strength of our study is that the results remain significant when we tested age moderation (sensitivity analysis), indicating a unique menopause moderation on the cognitive effects after three and six months of aerobic exercise.

Of note, the cognitive benefits observed were specific to executive functions, but not processing speed, as also hypothesized. The benefits were also not extended to other domains in the exploratory analysis, such as attention, working memory, language, or episodic memory. This result is consistent with our studies that suggest executive functions are the primary cognitive domain to benefit from AE.^2,3 Furthermore, our earlier reports from this clinical trial have noted greater benefits for executive functions over speed or other cognitive domains, ⁶ and sex-based differences in AE's effects were observed only in executive functions. ¹⁴ These findings hold significant clinical relevance as executive functions reflect crucial higher-order cognitive processes (e.g., planning, task-switching, resistance to interference, inhibition, and mental flexibility) that are largely governed by the frontal lobes. These functions are essential for maintaining functional independence, ³⁸ yet they are known to decline with aging ³⁹ and are impacted by the menopausal process.^15,40

Future studies should consider the characteristics and dosage of the exercise protocol adopted. This study implemented the regimen of 4 sessions per week, 1-h per session, and incorporated several strategies to boost adherence. While this dosage aligns with previous literature, the exact dose-benefit relationship of AE on cognition remains unclear.^3,41 Notably, this trial design is consistent with public health recommendations from the US Department of Health and Human Services, American Heart Association, and American College of Sports Medicine, which suggest at least 150 min per week of moderate-intensity exercise or 75 min of vigorous exercise for physical health benefits.

These results are also relevant considering the menopause-associated cognitive difficulties. Reviews and longitudinal cohort studies indicated that not only verbal memory, but also executive functions and complex attention are frequent cognitive complaints during perimenopause^42–44 and are associated with estrogen-related fluctuations and decline.^17,45 However, this literature is complicated by confounders, as cognitive difficulties during menopause transition can be exacerbated by other menopause symptoms, such as insomnia, mood, and vasomotor symptoms. For instance, a previous study ⁴² found that subjective memory complaints during perimenopause are not associated with objective verbal memory difficulties, but rather with difficulties in executive functions and complex attention tasks, which can also impact learning. The study also demonstrated that memory complaints in perimenopause are associated with symptoms of depression, anxiety, somatic complaints, and sleep disturbance.

Although it remains unclear the pathways through which AE may have contributed to the greater improvement in executive functions in post-menopausal than pre-menopausal women, it is possible that there was more ‘room for improvement’ in the postmenopausal-AE group, as this was the subgroup with the worst baseline performance on executive functions. Notably, despite postmenopausal women lacking relevant protection for cardiovascular and brain health^46,47 may influence exercise response, our findings contradict previous observations that the positive effect of physical activity on the cardiovascular system declines within a few years after the menopausal transition. ⁴⁸ We would expect cognitive benefits to follow a similar trend observed for cardiovascular health, and to share potential mechanisms. This type of observation is supported by findings that the measures of cardiovascular health improved more after exercise when estrogen is supplemented. ⁴⁹ This has led to the proposal of the “exercise timing hypothesis”, suggesting that the ideal time for women to start being more physically active is before estrogen production ceases, when estrogen hormonal levels and estrogen receptor expression are still high, as estrogen receptors are also relevant in the cellular response to exercise. ⁵⁰ It is plausible that the effects of AE on physical capacity and increased cerebral flow are critical in our findings, as we previously reported greater improvement in VO₂peak (on CPET), and a reduction in BMI after AE, in comparison to ST.^6,14 These findings are in line with the ‘cardiorespiratory/CPET hypothesis’, which suggests that higher levels of fitness are related to increased cerebral blood flow. ⁵¹ This hypothesis is supported by multiple studies showing that individuals (young, middle-aged, and older adults) with higher fitness demonstrate increased cerebral oxygenation compared to lower fit groups. ⁵² This is particularly relevant as the increased risk for cardiovascular diseases coincides with menopause, ¹⁵ including hypertension, diabetes, obesity, and metabolic syndrome, all of which are associated with increased risk of ADRD. ¹ In women, midlife comprises a critical window for exercise intervention as this is a period in life during which the accumulation of cardiovascular risk factors is strongly associated with cognitive impairment and biomarkers of Alzheimer's disease start to accumulate ³ (e.g., amyloid-β deposition and elevated tau protein).

The relationship between estradiol and neurocognitive functioning has been widely described, indicating the benefits of estrogen on cognition, brain health,^22,23 and neuroprotective effects, including the inhibition of beta-amyloid formation and oxidative cell damage. ³ This evidence is consistent with findings that a decline in estradiol contributes to midlife cognitive complaints and difficulties and that women in menopause experience a constellation of cognitive changes often referred to as “brain fog”. ⁵³ As expected, our data noted significant differences in estradiol levels and baseline cognition in the sample at pre-menopause and post-menopause. However, changes in estradiol levels after the exercise programs in either pre- or post-menopausal women were not observed. These observations suggest that estradiol is unlikely to explain the greater executive functions after AE; however, we are aware of the limited power of our exploratory analysis.

The present study was not designed to investigate the effects of AE on sleep and mood; nevertheless, these factors should be carefully considered in future trials investigating the interaction of menopause and exercise effects on cognition. Among menopausal symptoms, sleep disturbances and “psychological distress” (e.g., irritability, anxiety, tension, and depressive symptoms) are common ¹⁵ and can benefit from AE intervention. For instance, a meta-analysis indicates that low- and medium-intensity AE intervention is beneficial for sleep disorders in menopausal women, particularly in AE training regimens of 3 times/week and 70–90 min/session duration. ⁵⁴ Our protocol was similar, as the participants underwent an AE program of 4 times/week, duration of 60 min/session, over 6 months. Regarding emotional health, previous literature indicates AE can reduce symptoms of depression, ⁵⁵ anxiety, ⁵⁶ and relieve stress by lowering chronic cortisol levels. ⁵⁶ Although mood symptoms are typically reported to be more prevalent and intense during the transition to menopause, they can persist and become more prevalent in comparison to the pre-menopause stage. ⁵³ It is possible that our intervention improved sleep quality and emotional health, which may have confounded the benefits observed, and should be carefully assessed in future trials. Furthermore, although it is a strength that our RCT involved an active control group (ST), and was able to control for the aerobic component only present in the AE group, we should consider the potential cognitive benefit that the ST may have had, as it involved low-intensity toning (without weight lifting), an aspect of resistance training. This is relevant as previous evidence indicates cognitive benefits of resistance training,^57,58 an aspect that should be considered for future exercise trials.

The current study has limitations. First, it is a secondary analysis of an RCT designed to assess cognitive changes after exercise programs, but not originally designed to test the interaction of menopausal status with the effects of exercise on cognition; therefore, the statistical power is limited, although consistent with sample size rules of thumb for pilot RCTs. ³⁵

Critically, the study compared pre- and post-menopausal women but did not assess different stages of menopausal transition (e.g., perimenopause), or menopause history (e.g., age and duration of menopause), relevant factors that should be investigated in future studies, which we acknowledge as a limitation in our study. Despite that, we believe this study will advance women's brain health research by highlighting the benefits of exercise to a population at higher risk for dementia: post-menopausal women with below-average cardiovascular fitness. Even though our models controlled for age and our sensitivity analysis suggests there is an independence of age, menopause interaction, and age interaction, it is impossible to completely disentangle the age and menopause effects in this study. Despite that, the current report contributes novel findings for generating hypotheses and designing future intervention studies. Furthermore, the wide age range in this report (20 to 67 years old) is a limitation, and exercise trials examining the menopause status on cognition should consider a smaller age range (e.g., 45 to 60 years old), more in line with the menopause research. Despite these limitations, this study provides critical evidence for generating hypotheses and pathways for future trials. Furthermore, there are limitations regarding the hormonal data available: 1) estradiol was not collected considering the menstrual cycle phase, which can interfere with estradiol levels in premenopausal women; 2) the study did not control for medications, including those that affect estradiol levels (e.g., selective estrogen reception modulators, aromatase inhibitors); 3) the study did not measure estrone, the predominant circulating estrogen after menopause, which limits the ability to detect differences/associations with estrogen; 4) the estradiol exploratory analysis is limited in power, due to the degree of missing data; 5) although the participants were required to be free of HRT at study entry, it is unclear whether any participant initiated HRT during the trial. This represents a limitation, given the potential effects of HRT on cognitive and vascular outcomes, which should be accounted for in future studies.

In conclusion, the findings from a randomized clinical trial suggest that women in pre- and post-menopause may benefit differently from regular exercise, in particular AE training. Menopause associations with cognition are frequently overlooked in exercise research, which limits brain health prevention strategies specific to women. The demonstration that AE affected primarily executive functions for post-menopausal women should be further investigated and may have a significant impact on women's brain health recommendations. This report, despite the limitations, reinforces the need for future trials to investigate the hormonal and physiological mechanisms of exercise interventions on cognition in females. This is critical as exercise is a low-cost and accessible intervention, and a well-established modifiable factor to promote cardiovascular health and ADRD prevention.