Insulin Homeostasis Mediates the Relationship Between Cardiorespiratory Fitness and Cognitive Speed in Aging Adults

Abstract

Background:

Cardiorespiratory fitness (CRF) supports cognition, though it is unclear what mechanisms underly this relationship. Insulin resistance adversely affects cognition but can be reduced with habitual exercise.

Objective:

We investigated whether insulin resistance statistically mediates the relationship between CRF and cognition.

Methods:

In our observational study, we included n = 1,131 cognitively unimpaired, nondiabetic older adults from a cohort characterized by elevated Alzheimer’s disease (AD) risk. We estimated CRF (eCRF) using a validated equation that takes age, sex, body mass index, resting heart rate, and habitual physical activity as inputs. The Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) quantified insulin resistance. Standardized cognitive factor scores for cognitive speed/flexibility, working memory, verbal learning/memory, and immediate memory were calculated from a battery of neuropsychological tests. Linear regression models and bootstrapped estimates of indirect effects were used to determine whether HOMA-IR mediated significant relationships between eCRF and cognition.

Results:

eCRF was positively associated with cognitive speed/flexibility (p = 0.034). When controlling for HOMA-IR, eCRF was no longer associated with cognitive speed/flexibility (p = 0.383). HOMA-IR had a significant indirect effect on the eCRF-cognition relationship (B = 0.025, CI = [0.003,0.051]). eCRF was not associated with working memory (p = 0.236), immediate memory (p = 0.345), or verbal learning/memory (p = 0.650).

Conclusion:

Among older adults at risk for AD, peripheral insulin resistance mediates the relationship between CRF and cognitive speed.

Keywords

Alzheimer’s disease cognition cognitive decline insulin physical fitness

INTRODUCTION

There is emerging consensus that the optimal avenue for reducing the burden of Alzheimer’s disease (AD) on patients and healthcare systems may be primary prevention in the preclinical phase, prior to significant and irreversible neuronal damage [1]. Cardiorespiratory fitness (CRF) and peripheral insulin resistance are two promising targets for mitigating AD risk and delaying the onset of cognitive impairment. These factors are closely related and can be modified by lifestyle factors such as habitual exercise. Understanding how these factors intersect with cognitive decline may be crucial for designing effective lifestyle interventions to promote AD resilience.

CRF is associated with favorable outcomes in cognitive performance in aging adults. CRF refers to the ability of the cardiorespiratory system to deliver oxygen to skeletal muscle during aerobic exercise and can be improved through aerobic exercise training [2]. CRF has been shown to be positively associated with measures of both executive function and memory [3 –9] and may also diminish the deleterious effects of amyloid-β on cognition [10], effectively increasing resilience to cognitive decline despite AD-like pathology. The large and growing body of evidence demonstrating the association between aerobic health and measures of cognitive function warrants further investigation of CRF as a modifiable target for AD prevention.

In contrast to CRF, peripheral insulin resistance may increase risk of AD and cognitive decline. Insulin resistance refers to an impaired response to insulin signaling. Peripheral insulin resistance has been associated with lower cognitive performance in adults aged beyond midlife [11, 12]. Proposed mechanisms for this connection include the role of insulin resistance in amyloid-β deposition, inflammation, and oxidative stress [13].

It is well established that CRF is negatively associated with peripheral insulin resistance [14 –16], but it remains unclear whether peripheral insulin resistance may explain the link between CRF and cognitive function. At least one previous study has demonstrated that plasma insulin levels may mediate the relationship between CRF and memory in older adults [17]. However, this study was limited in sample size (n = 58). In the present study, we interrogate the mediating effects of peripheral insulin resistance on the relationships between estimated CRF and four measures of executive function and memory in a large cohort (n = 1,131) of late-middle-aged adults enriched for AD risk. We hypothesized that CRF would be positively associated with the cognitive abilities we tested, and that insulin resistance would be a partial statistical mediator of these associations.

METHODS

Participants

Participants (n = 1,131) were members of the Wisconsin Registry for Alzheimer’s Prevention (WRAP), a longitudinal registry of over 1,500 older adults [18]. Recruitment sources for the WRAP study include memory clinics in which a parent was diagnosed or treated, limited radio and newspaper advertisements, and word of mouth. Participants generally meet the following inclusion criteria at study entry: age 40–65 years; fluent English speaker; visual and auditory acuity adequate for neuropsychological testing; good health with no diseases expected to interfere with study participation over time [18]. Participants are excluded from enrollment if they have a prior diagnosis of dementia or evidence of dementia at baseline testing. As with the rest of the WRAP cohort, our sample was enriched with participants who have family history of AD and/or are carriers of the Apolipoprotein E4 (APOE4) allele [18]. All participants were cognitively unimpaired at the time of their study visit and had never received a formal diagnosis of diabetes. Participants were determined cognitively unimpaired by a diagnostic consensus conference, and based on intact performance on a comprehensive battery of neuropsychological tests, absence of functional impairment, and absence of neurologic/psychiatric conditions that might impair cognition [18].

WRAP participants were included in analysis for this study if they had all necessary data for key variables in our mediation model, including CRF estimation, insulin resistance quantification, cognitive score calculation, and data for control variables. All data included in our mediation analyses (including variables for CRF estimation, bloodwork for insulin resistance quantification, cognitive performance data, and demographic data for covariates in our statistical models) were collected at the same WRAP study visit. Therefore, all of the data in our mediation analyses are from one cross-section of participants’ ongoing participation in the WRAP study. For participants with multiple WRAP visits that had data fulfilling inclusion criteria, the most recent research visit during which all necessary data was collected was used for analyses. This meant that for newer enrollees of the WRAP study, data was pulled from visit two or three, whereas participants who have been WRAP study members for longer had data pulled from visit six or seven. This allowed us to maximize our sample size while also maximizing the age-at-visit of our participants. We chose to maximize age to improve the likelihood of seeing early signs of cognitive decline in our aging sample.

Non-exercise cardiorespiratory fitness measure

An estimation of CRF (eCRF) was calculated using an adaptation of the proposed equation by Jurca et al. [19], i.e., eCRF = 18.07 + Sex(2.77) –Age(0.10) –Body Mass Index(0.17) –Resting Heart Rate(0.03) + Self-Reported Physical Activity. The output of the equation is an estimate of peak metabolic equivalent of task (METs). Sex was coded as “1” for males and “0” for females. Body Mass Index (BMI) was calculated using the equation $BMI = \frac{Weight}{(Height)^{2}}$ with weight measured in kilograms and height in meters. Resting heart rate was measured using a GE Dinamap Pro 400 V2 Vital Signs Monitor with a GE Critikon Blood Pressure Cuff. Self-Reported Physical Activity was coded as “1” if participants met US Department of Health and Human Services (DHHS) Physical Activity Guideline recommendations and “0” if participants did not meet recommended physical activity [4], according to their self-assessment of habitual exercise. DHHS Physical Activity Guidelines for older adults recommend at least 150 minutes of moderate (or equivalent amounts of vigorous) physical activity per week [20, 21]. To estimate minutes of moderate and vigorous activity, participants were given a questionnaire that asked about the frequency and duration of their physical activity at different intensity levels throughout a typical week. A study in another sample of older adults has previously demonstrated that estimating CRF using a non-exercise equation is at least as accurate a measure of CRF as graded exercise testing (GXT) [22] while also being more cost-effective and easily reproducible. Use of the CRF estimation equation was previously validated in our cohort by comparing calculated eCRF values with experimental GXT data collected from a subset of our participants [4]. Because the average age of the present study’s mediation analyses sample is approximately 6 years older than the Boots et al. (2015) sample, we re-validated use of our eCRF equation in a subset of WRAP participants (n = 162) who underwent GXT within two years of a WRAP visit (mean time between GXT and WRAP visit was 0.87±0.50 years) from which sufficient data was collected to calculate eCRF. GXT methods were the same as those used in the study from Boots et al. (2015), but with data collected from more recent GXT visits.

Insulin resistance estimation

Insulin resistance was quantified using the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) [23]. HOMA-IR is calculated according to the formula, HOMA-IR $= \frac{[Glucose] [Insulin]}{405}$ , where “Glucose” represents plasma glucose in mg/dL following a 12-hour fast and “Insulin” represents plasma insulin in μIU/mL following a 12-hour fast. Interpretation of HOMA-IR is predicated on the notion that insulin and glucose should have an inverse relationship, such that higher insulin levels should correspond with lower plasma glucose. Higher HOMA-IR values correspond to higher levels of insulin resistance.

Cognitive measurements

Cognitive composite scores were based on performance on neuropsychological exams administered by trained technicians [24]. Composite scores were calculated for four cognitive factor scores measuring cognitive speed/flexibility, working memory, verbal learning and memory, and immediate memory [25]. The neuropsychological tests used to calculate each of these cognitive factor scores are shown in Table 1. Measures of cognitive performance were standardized across the WRAP database for each cognitive factor score. In some cases, a participant was missing data for a cognitive factor score. Of our 1,131 participants, 1,098 (97.1%) had cognitive speed/flexibility scores, 1,123 (99.3%) had working memory scores, 1,126 (99.6%) had verbal learning and memory scores, and 1,128 (99.7%) had immediate memory scores.

Table 1

Cognitive tests included in each cognitive factor score

Factor Scores
Cognitive Speed/Flexibility
–Stroop Color-Word Test Interference Trial (Trenerry et al., 1989)
–Trail-Making Test A and B (Reitan and Wolfson, 1993)
Working Memory
–Digit Span and Letter-Number Sequencing subtests from the Wechsler Adult Intelligence Scale, 3^rd edition (Wechsler, 1997)
Verbal Learning and Memory
–RAVLT Learning Trials 3-5 and Delayed Recall (M. Schmidt, 1996)
Immediate Memory
–RAVLT Learning Trials 1 and 2 (M. Schmidt, 1996)

Statistical analyses

To validate eCRF as a predictor of aerobic fitness, we used Pearson correlation to measure the relationship between eCRF and experimental VO₂peak data. Additionally, we regressed VO₂peak from the exercise visit on sex, age, BMI, resting heart rate, and self-reported physical activity from the nearest WRAP visit to create an equation with our own parameter estimates. We then used Pearson correlation to determine our equation’s agreement with both the original Jurca-based eCRF and experimental VO₂peak data.

To determine whether eCRF had a total effect on any of the cognitive scores, linear regression models were fit to measure the association of eCRF on each cognitive measure while controlling for sex, age, years of education, family history of AD, and APOE4 carriage. In cases where a participant did not have data available for a cognitive outcome, that participant was excluded from regression models. If a significant association between eCRF and a cognitive measure was found, we followed up with mediation analysis to test whether HOMA-IR statistically mediated the relationship. In the mediation model, a second linear regression model was fit to test the association between eCRF and HOMA-IR while controlling for the same covariates as the regression model for the total effect. A third regression model measured the effect of HOMA-IR on the cognitive outcome. The third model also remeasured the effect of eCRF on the cognitive score while controlling for HOMA-IR to determine the direct effect of eCRF. Bootstrapped estimates were used to evaluate the indirect effect of HOMA-IR on the eCRF-cognitive score relationship. All analyses were completed with IBM SPSS version 27.0. All regression models within the mediation analyses used the PROCESS Macro v3.4 (Hayes, 2017). Findings with p < 0.05 (2-tailed) were considered significant.

RESULTS

Participant characteristics

Participant characteristics of our mediation analyses sample are detailed in Table 2. Our sample included mostly females (69.9%) with a mean age of 64.7±7.1 years. The sample population had high rates of APOE4 carriage (38.5%) and family history of AD (71.8%), suggesting heightened risk for AD in our sample. Our sample was overweight on average with a mean BMI of 29.0±6.2.

Table 2

Participant characteristics –mediation analyses (n = 1,331)

Variable	Value
DEMOGRAPHIC DATA
Age (y)	64.7±7.1
Female (%)	69.9
Education (y)	16.3±2.8
APOE4 Carrier (%)	38.5
Family History of AD (%)	71.8
CRF ESTIMATE DATA
eCRF –full sample (METs)	5.9±1.9
eCRF –male (METs)	7.9±1.3
eCRF –female (METs)	5.1±1.5
BMI (kg/m²)	29.0±6.2
Resting Heartrate (bpm)	65.5±9.5
Moderate Physical Activity Equivalent (minutes/week)	165.9±191.2
GLUCOMETABOLIC DATA
Fasting plasma glucose (mg/dL)	96.6±12.4
Fasting plasma insulin (μIU/mL)	10.2±7.8
HOMA-IR	2.5±2.3

APOE4, Apolipoprotein E4 allele; AD, Alzheimer’s disease; BMI, body mass index; CRF, cardiorespiratory fitness; HOMA-IR, Homeostatic Model Assessment for Insulin Resistance. All values are Mean±Standard Deviation unless otherwise specified.

CRF verification

Participant characteristics of our eCRF validation sample are detailed in Table 3. In the subsample who completed exercise testing (n = 162), experimentally measured VO₂peak was 7.4±1.8 METs. Mean eCRF in this subsample was 6.4±1.9 METs. eCRF was significantly correlated with VO₂peak (r = 0.771, p < 0.001, Fig. 1). Our own regression coefficients estimated VO₂peak within our sample as eCRF = 17.437 + Sex(1.769) –Age(0.086) –BMI(0.148) –RHR(0.020)+Self-Reported Physical Activity(0.537). The equation made from our sample was significantly correlated with both the Jurca-based eCRF (r = 0.988, p < 0.001) and the experimental VO₂peak data (r = 0.780, p < 0.001).

Table 3

Participant characteristics –eCRF validation (n = 162)

Variable	Value
EXERCISE VISIT DATA
VO₂peak –full sample (METs)	7.4±1.8
VO₂peak –male (METs)	8.6±1.8
VO₂peak –female (METs)	6.8±1.5
Female (%)	66.0
Age (y)	63.8±6.1
BMI (kg/m²)	27.9±5.4
WRAP VISIT DATA
eCRF –full sample (METs)	6.4±1.9
eCRF –male (METs)	8.3±1.2
eCRF –female (METs)	5.5±1.5
Female (%)	66.0
Age (y)	64.1±6.1
BMI (kg/m²)	27.8±5.2
Resting Heartrate (bpm)	64.4±10.3
Moderate Physical Activity Equivalent (minutes/week)	255±246
Time between Exercise and WRAP visits (years)	0.87±0.50

Fig. 1

Experimental CRF is significantly correlated with eCRF. MET, Metabolic Equivalent of Task; eCRF, Estimated cardiorespiratory fitness according to adapted Jurca et al. (2005) equation.

CRF, HOMA-IR, and cognitive relationships

eCRF was positively associated with cognitive speed/flexibility (total effect: B = 0.047, SE = 0.022, p = 0.034). eCRF was not associated with working memory (B = 0.026, SE = 0.022, p = 0.236), immediate memory (B = 0.022, SE = 0.024, p = 0.345), or verbal learning/memory (B=-0.010, SE = 0.022, p = 0.650).

HOMA-IR was assessed as a statistical mediator between eCRF and cognitive speed/flexibility (Fig. 2). eCRF was negatively associated with HOMA-IR (B=-0.839, SE = 0.047, p < 0.001) and HOMA-IR was negatively associated with cognitive speed/flexibility (B=-0.030, SE = 0.014, p = 0.039). Adding HOMA-IR to the total effect regression model as a mediator nullified the association between eCRF and cognitive speed/flexibility (direct effect: B = 0.022, SE = 0.025, p = 0.383). The estimated indirect effect of HOMA-IR on the association between eCRF and cognitive speed/flexibility was significant (B = 0.025, CI = [0.003, 0.051]) and the parameter estimate of this indirect effect was equal to 53.2% of the total effect of eCRF on cognition.

Fig. 2

HOMA-IR statistically mediates the relationship between eCRF and cognitive speed/flexibility. Linear regression models were used to calculate parameter estimates. All data represent β±standard error. The total effect estimates the effect of estimated cardiorespiratory fitness (eCRF) on cognitive speed/flexibility without controlling for HOMA-IR. The direct effect estimates the effect of eCRF on cognitive speed/flexibility independent of HOMA-IR. The “a” pathway estimates the effect of eCRF on HOMA-IR, and the “b” pathway estimates the effect of HOMA-IR on cognitive speed/flexibility independent of eCRF. Regression models for the total effect and “a” pathway controlled for age, sex, years of education, familial history of Alzheimer’s disease, and apolipoprotein E4 carriage. The regression model used to calculate the direct effect and “b” pathway included the same covariates and HOMA-IR in the regression model. HOMA-IR acted as a significant mediator in the relationship between eCRF and cognitive speed/flexibility.

DISCUSSION

The major finding of this study was that in an aging cohort at risk for AD, peripheral insulin resistance statistically mediated the relationship between eCRF and cognitive speed/flexibility. eCRF did not have a significant total effect on measures of working memory, immediate memory, or verbal learning/memory.

Our finding that eCRF is positively associated with cognitive speed/flexibility agrees with several prior studies showing a relationship between CRF (and/or exercise) and executive function-based measures of cognition [3 , 27]. Our finding that the relationship between eCRF and cognitive speed/flexibility was mediated by HOMA-IR is in line with what we expected based on a previous study demonstrating that higher HOMA-IR corresponds to lower performance on tests of executive function [11], a broad cognitive domain that encompasses cognitive speed/flexibility.

Our findings did not support the existence of a relationship between CRF and performance on memory tests, a result that is shared by at least one other study [8]. Other studies have reported an association between CRF and memory [4 , 28]. Boots et al. (2015) used many of the same participants from the WRAP cohort in their study that we used in our study, though data from earlier visits were used and thus the study from Boots et al. had an average age of 59 years compared to our average age of 65 years. The Boots et al. (2015) study also only controlled for education and sex in their regression models to test the effect of eCRF on verbal learning and memory whereas we also controlled for age, APOE4 carriage, and family history of AD. Using the present study’s dataset, a regression model controlling only for education and sex showed a significant positive association between eCRF and verbal learning/memory (β=0.066, SE = 0.020, p = 0.001). Therefore, our decision to control for additional factors relevant to cognitive health seems to explain the difference in findings between our study and the earlier study by Boots et al. (2015).

The study from Dougherty et al. used participants from the WRAP cohort and demonstrated an association between VO₂peak and RAVLT scores in males. This study group included BMI as a covariate in their regression models between VO₂peak and cognitive performance. When we attempted this in our sample, we introduced significant issues with multicollinearity to our models and did not find any significant associations between eCRF or BMI and memory performance. It is not surprising that multicollinearity became an issue upon addition of BMI as a covariate because, in addition to age and sex, BMI is one of the variables that contributes to eCRF calculation. When BMI was excluded as a covariate, but age and sex were used as controls, the variance inflation factor of eCRF never rose above 2.785. We decided to keep age and sex as covariates because they are both well known to be closely associated with cognitive outcomes, and we wanted to see if our constructed eCRF was associated with cognitive outcomes even with additional control of these crucial predictors of cognitive performance.

A study similar in design to our own showed that plasma insulin levels may mediate the relationship between CRF and memory [17]. Notable differences between the sample from Tarumi et al. and our sample include the Tarumi et al. sample having average CRF equivalent to 7.3 METs in their sedentary group and 12.2 METs in their endurance-trained group whereas our sample’s average eCRF was only 6.0 METs. Additionally, the Tarumi et al. sample had higher mean plasma insulin levels (sedentary = 21.3 uIU/mL, endurance-trained=12.6 uIU/mL) than in our sample (10.2 uIU/mL). It seems that while the Tarumi sample had higher CRF, their sample was also more prone to insulin resistance. The difference in relation between CRF and insulin in the Tarumi et al. study compared to our study may account for some differences in our findings related to cognition.

While we did not find an association between eCRF and memory ability, it is important to consider that several other studies have shown a relationship between these variables in humans. The lack of an association between eCRF and memory in our sample may be due in part to the chronology of cognitive deficits from AD. At least one study has shown that executive function begins to decline prior to memory in preclinical AD [29]. In our study, this might mean that a follow up with the same participants after several years (when participants are older) could result in more clear associations between eCRF and memory.

There are some limitations to the findings of this study. While the regression models we used controlled for many known risk factors for AD, we did not control for cerebrovascular health-related risk factors. Cerebrovascular health likely contributes to our finding that eCRF is associated with cognitive speed/flexibility and should be investigated as another mediator of the relationship in future work. Our use of an estimation equation for CRF allowed for a much larger sample size, but experimentally measured VO₂peak is generally considered the gold-standard measure of CRF in healthy adults. Nevertheless, we believe our use of eCRF provided a very accurate estimate of CRF in our sample given previous validation of this method in the WRAP cohort from our lab group and this study’s revalidation in our more age-advanced participants. Our use of HOMA-IR as a measure of insulin resistance also helped simplify data collection and allowed for a large sample size, but it is important to note that HOMA-IR is considered a measure of peripheral insulin resistance and does not necessarily correspond to insulin resistance or insulin resistance-related inflammatory processes in the brain. This is important to consider for contemplation of mechanisms that could explain the association we have demonstrated between peripheral insulin resistance and cognitive speed/flexibility. Our sample was also composed of mostly white individuals with high levels of education and our findings therefore may not be generalizable to people of lower socioeconomic status or who belong to underrepresented groups.

Findings from this study highlight the need for more investigation in several areas. There is still a need to explore how insulin function in brain tissue compares to insulin function in the rest of the body. In particular, much work is still needed to determine how peripheral versus central insulin resistance might modify brain function. Potential next steps for exploring this topic might include use of neuroimaging tools to compare the associations of peripheral versus central insulin resistance on outcomes such as functional brain connectivity and brain glucose uptake. Additionally, further investigation into how insulin resistance interacts with dyslipidemia, inflammation, and oxidative stress to alter cognition in adults at risk for AD is warranted. It also remains unclear how CRF and insulin resistance might affect resilience to cognitive decline at different stages of AD pathology. Future studies should determine the significance of CRF and insulin resistance in participants with varying levels of amyloid-β deposition and tau tangle presence.

In conclusion, we have shown that peripheral insulin resistance mediates the relationship between eCRF and cognitive speed. eCRF is not associated with verbal learning/memory, immediate memory, or working memory in our sample, though CRF and insulin resistance may be relevant to these abilities in other samples, particularly those with more advanced age and/or amyloid-β deposition.

Footnotes

ACKNOWLEDGMENTS

We thank the staff and study participants of the Wisconsin Registry for Alzheimer’s Prevention and the Wisconsin Alzheimer’s Disease Research Center, without whom this work would not be possible.

FUNDING

This work was supported by National Institute on Aging grants R01 AG062167 (O.C.O.), R01 AG027161 (S.C.J.), and P30 AG062715 (S.A.); and a Clinical and Translational Science Award (UL1RR025011) to the University of Wisconsin, Madison. Portions of this research were supported by the Wisconsin Alumni Research Foundation; and the Veterans Administration, including facilities and resources at the Geriatric Research Education and Clinical Center of the William S. Middleton Memorial Veterans Hospital, Madison, WI.

CONFLICT OF INTEREST

Ozioma Okonkwo and Barbara Bendlin are Associate Editors of this journal but were not involved in the peer review of this manuscript nor had access to any information regarding the peer review. No author has any conflict of interest to report regarding this manuscript.

DATA AVAILABILITY

The data that support the findings of this study are available upon reasonable request from corresponding author, Ozioma Okonkwo, PhD.

References

Sperling

, Jack

Jr , Aisen

(2011) Testing the right target and right drug at the right stage, Sci Transl Med 3, 111cm33.

Kokkinos

(2014) Cardiorespiratory fitness, exercise, and blood pressure, Hypertension 64, 1160–1164.

Okonkwo

, Schultz

, Oh

, Larson

, Edwards

, Cook

, Koscik

, Gallagher

, Dowling

, Carlsson

, Bendlin

, LaRue

, Rowley

, Christian

, Asthana

, Hermann

, Johnson

, Sager

(2014) Physical activity attenuates age-related biomarkeralterations in preclinical AD, Neurology 83, 1753–1760.

Boots

, Schultz

, Oh

, Larson

, Edwards

, Cook

, Koscik

, Dowling

, Gallagher

, Carlsson

, Rowley

, Bendlin

, LaRue

, Asthana

, Hermann

, Sager

, Johnson

, Okonkwo

(2015) Cardiorespiratory fitness is associated with brain structure, cognition, and mood in a middle-aged cohort at risk for Alzheimer’s disease, Brain Imaging Behav 9, 639–649.

Dougherty

, Jonaitis

, Gaitán

, Lose

, Mergen

, Johnson

, Okonkwo

, Cook

(2021) Cardiorespiratory fitnessmitigates brain atrophy and cognitive decline in adults at risk forAlzheimer’s disease, Alzheimers Dement (Amst) 13, e12212.

Dougherty

, Schultz

, Boots

, Ellingson

, Meyer

, Van Riper

, Stegner

, Edwards

, Oh

, Einerson

, Korcarz

, Koscik

, Dowling

, Gallagher

, Carlsson

, Rowley

, Bendlin

, Asthana

, Hermann

, Sager

, Stein

, Johnson

, Okonkwo

, Cook

(2017) Relationships between cardiorespiratory fitness, hippocampal volume, and episodic memory in a population at risk for Alzheimer’s disease, Brain Behav 7, e00625.

Gaitán

, Boots

, Dougherty

, Oh

, Ma

, Edwards

, Christian

, Cook

, Okonkwo

(2019) Brain glucose metabolism,cognition, and cardiorespiratory fitness following exercise trainingin adults at risk for Alzheimer’s disease, Brain Plast 5, 83–95.

Pentikainen

, Savonen

, Ngandu

, Solomon

, Komulainen

, Paajanen

, Antikainen

, Kivipelto

, Soininen

, Rauramaa

(2019) Cardiorespiratory fitness and cognition: Longitudinal associations in the FINGER Study, J Alzheimers Dis 68, 961–968.

Gnosa

, Marczak

, Binder

, Adler

(2019) Impact of physical fitness on cognitive performance in patients at a memory clinic, Dement Geriatr Cogn Dis Extra 9, 129–135.

10.

Schultz

, Boots

, Almeida

, Oh

, Einerson

, Korcarz

, Edwards

, Koscik

, Dowling

, Gallagher

, Bendlin

, Christian

, Zetterberg

, Blennow

, Carlsson

, Asthana

, Hermann

, Sager

, Johnson

, Stein

, Okonkwo

(2015) Cardiorespiratory fitness attenuates the influence of amyloid on cognition, J Int Neuropsychol Soc 21, 841–850.

11.

Kim

, Lee

, Kim

, Mok

, Kim

, Park

, Lee

, Yoo

, Jeong

, Lee

, Park

(2015) The relationship between cognitive performance and insulin resistance in non-diabetic patients with mild cognitive impairment, Int J Geriatr Psychiatry 30, 551–557.

12.

Ekblad

, Rinne

, Puukka

, Laine

, Ahtiluoto

, Sulkava

, Viitanen

, Jula

(2017) Insulin resistance predicts cognitive decline: An 11-year follow-up of a nationally representative adult population sample, Diabetes Care 40, 751–758.

13.

Bosco

, Fava

, Plastino

, Montalcini

, Pujia

(2011) Possible implications of insulin resistance and glucose metabolism in Alzheimer’s disease pathogenesis, J Cell Mol Med 15, 1807–1821.

14.

Chen

C-N

, Chuang

L-M

, Wu

Y-T

(2008) Clinical measures of physical fitness predict insulin resistance in people at risk for diabetes, Phys Ther 88, 1355–1364.

15.

Leite

, Monk

, Upham

, Chacra

, Bergenstal

(2009) Low cardiorespiratory fitness in people at risk for type 2 diabetes: Early marker for insulin resistance, Diabetol Metab Syndr 1, 8.

16.

Castro

, Venutolo

, Yau

, Convit

(2016) Fitness, insulin sensitivity, and frontal lobe integrity in adults with overweight and obesity, Obesity 24, 1283–1289.

17.

Tarumi

, Gonzales

, Fallow

, Nualnim

, Lee

, Tanaka

, Haley

(2013) Aerobic fitness and cognitive function in midlife: An association mediated by plasma insulin, Metab Brain Dis 28, 727–730.

18.

Johnson

, Koscik

, Jonaitis

, Clark

, Mueller

, Berman

, Bendlin

, Engelman

, Okonkwo

, Hogan

, Asthana

, Carlsson

, Hermann

, Sager

(2018) The Wisconsin Registry for Alzheimer’s Prevention: A review of findings and current directions, Alzheimers Dement (Amst) 10, 130–142.

19.

Jurca

, Jackson

, LaMonte

, Morrow

, Blair

, Wareham

, Haskell

, Van Mechelen

, Church

, Jakicic

, Laukkanen

(2005) Assessing cardiorespiratory fitness without performing exercise testing, Am J Prev Med 29, 185–193.

20.

Bushman

(2019) Physical Activity Guidelines for Americans: The relationship between physical activity and health, ACSMs Health Fit J 23, 5–9.

21.

Nelson

, Rejeski

, Blair

, Duncan

, Judge

, King

, Macera

, Castaneda-Sceppa

(2007) Physical activity and public health in older adults: Recommendation from the American College of Sports Medicine and the American Heart Association, Med Sci Sports Exerc 39, 1435–1445.

22.

Mailey

, White

, Wójcicki

, Szabo

, Kramer

, McAuley

(2010) Construct validation of a non-exercise measure of cardiorespiratory fitness in older adults, BMC Public Health 10, 59.

23.

Matthews

, Hosker

, Rudenski

, Naylor

, Treacher

, Turner

(1985) Homeostasis model assessment: Insulin resistance and β-cell function from fasting plasma glucose and insulin concentrations in man, Diabetologia 28, 412–419.

24.

Sager

, Hermann

, La Rue

(2005) Middle-aged children of persons with Alzheimer’s disease: APOE genotypes and cognitive function in the Wisconsin Registry for Alzheimer’s Prevention, J Geriatr Psychiatry Neurol 18, 245–249.

25.

Koscik

, Rue La

, Jonaitis

, Okonkwo

, Johnson

, Bendlin

, Hermann

, Sager

(2014) Emergence of mild cognitive impairment in late-middle-aged adults in the Wisconsin Registry for Alzheimer’s Prevention, Dement Geriatr Cogn Disord 38, 16–30.

26.

Mekari

, Dupuy

, Martins

, Evans

, Kimmerly

, Fraser

, Neyedli

(2019) The effects of cardiorespiratory fitness on executive function and prefrontal oxygenation in older adults, GeroScience 41, 681–690.

27.

Dupuy

, Gauthier

, Fraser

, Desjardins-Crèpeau

, Desjardins

, Mekary

, Lesage

, Hoge

, Pouliot

, Bherer

(2015) Higher levels of cardiovascular fitness are associated with better executive function and prefrontal oxygenation in younger and older women, Front Hum Neurosci 9, 66.

28.

Rigdon

, Loprinzi

(2019) The association of cardiorespiratory fitness on memory function: Systematic review, Medicina (Kaunas) 55, 127.

29.

Harrington

, Chiang

, Pogoda

, Gomez

, Thomas

, Marion

SDB

, Miller

, Siddarth

, Yi

, Zhou

, Lee

, Arakaki

, Cowan

, Tran

, Charleswell

, Ross

, Fonteh

(2013) Executive function changes before memory in preclinical Alzheimer’s pathology: A prospective, cross-sectional, case control study, PLoS One 8, e79378.