Journal of the American Heart Association ORIGINAL RESEARCH Associations Between Cardiorespiratory Fitness, Cardiovascular Risk, and Cognition Are Mediated by Structural Brain Health in Midlife Goretti España-­Irla , MSc*; Joyce Gomes-­Osman , PhD*; Gabriele Cattaneo , PhD; Sergiu Albu , PhD; María Cabello-­Toscano , MSc; Javier Solana-­Sanchéz , PhD; María Redondo-­Camós , MSc; Selma Delgado-­Gallén , MSc; Vanessa Alviarez-­Schulze , MSc; Catherine Pachón-­García , MSc; Josep M. Tormos , PhD; David Bartrés-­Faz , PhD†; Timothy P. Morris , PhD† Álvaro Pascual-­Leone , PhD† BACKGROUND: Evidence in older adults suggests that higher cardiorespiratory fitness and lower cardiovascular risk are as- sociated with greater cognition. However, given that changes in the brain that lead to cognitive decline begin decades before the onset of symptoms, understanding the mechanisms by which modifiable cardiovascular factors are associated with brain health in midlife is critical and can lead to the development of strategies to promote and maintain brain health as we age. METHODS AND RESULTS: In 501 middle-­aged (aged 40–­65 years) adult participants of the BBHI (Barcelona Brain Health Initiative), we found differential associations among cardiorespiratory fitness, cardiovascular risk, and cognition and cortical thickness. Higher cardiorespiratory fitness was significantly associated with better visuospatial abilities and frontal loading abstract problem solving (β=3.16, P=0.049) in the older middle-­aged group (aged 55–­65 years). In contrast, cardiovascular risk was negatively associated with better visuospatial reasoning and problem-­solving abilities (β=−0.046, P=0.002), flexibility (β=−0.054, P<0.001), processing speed (β=−0.115, P<0.001), and memory (β=−0.120, P<0.001). Cortical thickness in frontal regions mediated the relationship between cardiorespiratory fitness and cognition, whereas cortical thickness in a disperse network spanning multiple cortical regions across both hemispheres mediated the relationship between cardiovascular risk and cognition. CONCLUSIONS: The relationships between modifiable cardiovascular factors, cardiorespiratory fitness, and cardiovascular risk, and cognition are present in healthy middle-­aged adults. These relationships are also mediated by brain structure highlighting a potential mechanistic pathway through which higher cardiorespiratory fitness and lower cardiovascular risk can positively impact cognitive function in midlife. Key Words: cardiorespiratory fitness ■ cardiovascular health ■ cognition ■ exercise ■ mediation ■ midlife ■ structural brain health U nderstanding factors associated with mainte- accompanied by an increase in health span,1 and brain-­ nance of cognitive brain health in aging is of related disorders are projected to account for half of great clinical and public health interest. An in- the worldwide economic impact of disability by 2030.2 crease in lifespan over the past century has not been Notwithstanding, the development of pathological loss Correspondence to: Timothy P. Morris, PhD, Department of Psychology, Northeastern University, 606 ISEC, 360 Huntington Avenue, Boston, MA 02115. E-­mail: t.morris@northeastern.eduand Álvaro Pascual-­Leone, PhD, Barcelona Brain Health Initiative, Institut Guttmann, Camí de Can Ruti, s/n, 08916 Badalona, Barcelona, Spain. E-­mail: apleone@guttmann.com * G. España-­Irla and J. Gomes-­Osman contributed equally as a first author. † D. Bartrés-­Faz, T.P. Morris and A. Pascual-­Leone contributed equally as senior author. Supplementary Material for this article is available at https://www.ahajo​urnals.org/doi/suppl/​10.1161/JAHA.120.020688 For Sources of Funding and Disclosures, see page 13. © 2021 The Authors. Published on behalf of the American Heart Association, Inc., by Wiley. This is an open access article under the terms of the Creative Commons Attribution-­NonCommercial-­NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-­commercial and no modifications or adaptations are made. JAHA is available at: www.ahajournals.org/journal/jaha J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206881 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure sociobehavioral proxies have been found to contribute CLINICAL PERSPECTIVE to the development of cognitive reserve.4 For instance, maintaining an active lifestyle by engaging in physical What Is New? exercise,5,6 promoting cardiovascular health,7 con- • We extend prior work by demonstrating that suming nutritious foods,8,9 assuring sufficient good-­ some of the well-­established relationships be- quality sleep,10–­12 and maintaining motor skills13 are tween determinants of cardiovascular health independently associated with better cognitive brain and brain health that exist in older age are al- health across one’s lifespan. The exact mechanisms ready present in middle age. by which modifiable sociobehavioral proxies influence • Cardiorespiratory fitness was associated with the development of cognitive reserve are not fully elu- frontal cognitive abilities, such as visuospatial cidated but can be attributed to the interplay between problem solving, but only in individuals aged 55 brain reserve and brain maintenance. That is, brain years and older. reserve is defined as the neurobiological capital, or • Cardiovascular risk was associated with a wide range of cognitive abilities within the whole structural integrity, of the many components of the ner- sample; these results suggest distinct, but syn- vous system at any given point in time, whereas brain ergistic effects of cardiovascular risk and cardi- maintenance is defined as the reduced development orespiratory fitness with cognitive brain health in of age-­related changes over time.4 Brain maintenance healthy middle-­aged adults. reflects the notion that the brain can be modified by experience, and many of the same lifestyle proxies that What Are the Clinical Implications? contribute to cognitive reserve also contribute to brain • Importantly, we advance existing knowledge by maintenance.4 Finally, it is also known that measurable revealing that such relationships driven by distinct changes in brain structure precede clinically measur- patterns of cortical thickness, specifically corti- able cognitive deficits by many years,14,15 and therefore cal thickness in frontal regions mediated the re- examining the relationships between modifiable fac- lationship between cardiorespiratory fitness and tors that may contribute to cognitive and brain reserve visuospatial problem solving, whereas cortical beginning in midlife may provide evidence to develop thickness in a disperse network spanning mul- and refine lifestyle strategies capable of promoting or tiple cortical regions across both hemispheres maintaining brain health in older age. mediated the relationship between cardiovascu- lar risk and multiple domains of cognition. There is strong evidence that cardiovascular health • The implications of our study lie within the poten- in midlife is a strong predictor of cognitive health in later tial importance of engaging in modifiable lifestyle life.7,16 One important domain of cardiovascular health behaviors that can promote heart health, early in is cardiorespiratory fitness (CRF). The gold-­standard midlife, long before the onset of measurable cog- measure of CRF is the maximum rate of oxygen con- nitive decline. sumption during incremental exercise, or VO2max, which measures the body’s efficiency to intake, circu- late, and use oxygen during exercise. CRF has been identified as a critical mechanism implicated in exer- cise’s effect on cognitive function.5,6,16–­18 Furthermore, Nonstandard Abbreviations and Acronyms numerous studies have associated CRF itself with cog- CPET cardiopulmonary exercise testing nitive functions, whereby rather than having a global effect on cognition, high levels of CRF later in life seem CRF cardiorespiratory fitness to be related to selective enhancement of cognitive CVR cardiovascular risk abilities more reliant on frontal brain areas such as ex- ecutive and reasoning abilities.5,6 Another important cardiovascular health predictor of brain health does not appear to be an obligatory is the risk of developing a future cardiovascular event, consequence of aging.3 Several lifestyle behaviors which can be calculated by measuring several factors have been found to be protective of age-­related and such as hypertension, cigarette smoking, diabetes, hy- pathological brain changes, which are referred to as perlipidemia, family history, and obesity.19 Interestingly, the concept of cognitive reserve.4 Cognitive reserve cardiovascular risk (CVR) factors overlap with cogni- helps to explain why certain individuals can withstand tive impairment risk factors,20–­24 further strengthening age-­related and pathological brain changes while the link between cardiovascular and cognitive health. maintaining their cognitive and physical functioning Evidence suggests that CVR later in life is associated and ultimately their independence with age.4 with more diffuse patterns of gray matter atrophy and Although cognitive reserve is a theoretical construct white matter lesions, thus potentially affecting cognitive and is rarely measured directly, several modifiable abilities in a more global manner.25–­27 As such, low CVR J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206882 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure burden in middle age might be associated with a more subjects. All procedures were approved by the eth- global effect on cognitive health and brain structure. ics and education committee of the Institut Guttmann In this study, our primary objective was to assess (Badalona, Spain). The Strengthening the Reporting the respective relationships between CRF and CVR of Observational Studies in Epidemiology (STROBE) and cognitive function in midlife in a sample of 501 checklist has been used for the reporting of the pre- adults aged 40 to 65 years. We further aimed to ex- sent study results.33 amine the mechanistic correlates of these relation- ships in midlife through measures of brain structure Neuropsychological Exam using magnetic resonance imaging (MRI), by testing Neurocognitive assessments were performed by 2 li- whether cortical thickness mediated the relationships censed neuropsychologists. Education in years was between each predictor (CVR and CRF) and cognitive assessed via an online questionnaire, and this informa- function. Although genetic predisposition influences tion was validated and corrected by a neuropsycholo- both CRF28,29 and CVR,30 these 2 factors are modifi- gist or physician during the in-­person assessments. able through lifestyle changes. Therefore, further elu- Paper and pencil evaluations consisted of a battery of cidating the relationships and potential mechanisms of well-­established neuropsychological tests. These in- these modifiable cognitive reserve protecting factors cluded Matrix Reasoning,34 Cancelation Test,34 Block in midlife can contribute to the development of more Design,34 Trail Making Test B,35 Trail Making Test effective and precise lifestyle interventions to maintain A,36 Digit Forward, Digit Backward, Letter–­ Number or improve cognitive brain health with age. Sequencing,37 Rey Auditory Verbal Learning Test,38 Digit Symbol Substitution,34 and Corsi Block-­Tapping Task.39 Tests were grouped in cognitive domains using METHODS a data-­driven approach with principal component anal- The data that support the findings of this study are ysis. Scores on individual tests were Z-­score normal- available from the corresponding author upon reason- ized before their inclusion in the principal component able request. analysis with Oblimin rotation, considering the prob- able correlation between latent factors.40 Based on the sample size, the acceptable level of factor loading was Study Design and Participants set at 0.30.41 Cognitive domains were then created as This was a cross-­sectional study that included data the composite sum of the Z scores for each test per collected on a subset of participants enrolled in the on- the results from the principal component analysis. The going BBHI (Barcelona Brain Health Initiative) (https:// principal component analysis indicated the presence bbhi.cat/en/), who were selected to participate in of 5 principal components for the cognitive scores. phase 2 of the initiative, which involved a comprehen- The first factor included the Digit Symbol Test (0.65), sive in-­person assessment.31,32 For a detailed descrip- the Cancellation Test (0.76) and the Trail Making Test tion of the cohort and study protocol see Cattaneo et A (0.80), likely reflecting visual searching, processing al.31,32 Inclusion criteria (assessed by a medical doc- speed, and attentional components. The second com- tor) for this study included: (1) age between 40 and ponent comprised all 3 measures of the Rey Auditory 65 years and (2) absence of any neurological or psy- Verbal Learning Test (immediate recall=−0.85, de- chiatric disorders. Exclusion criteria included any per- layed recall=−0.89, recognition=−0.81) creating a ver- son presenting with any contraindications for brain bal memory domain. The third component contained MRI and cardiopulmonary exercise testing (CPET) (see the Digit Forward (0.81), Digit Backward (0.66), and details below). We further excluded those participants Letter–­Number Sequencing (0.68), reflecting a working who did not meet the criteria for a completed CPET memory domain. Cognitive flexibility and set-­shifting evaluation (see CPET section). A cohort consort dia- abilities were reflected in the fourth component, which gram from the wider BBHI study and selection criteria included the Trail Making Test B (0.91) and the Trail for this analysis is shown in Figure 1. A total of 501 Making Test B-­A (1.02). Finally, a visuospatial reason- participants were eligible for this analysis based on ing and problem-­solving domain was found in the fifth having completed a full CPET evaluation. There were component comprising Wechsler Adult Intelligence incomplete data on a total of 114 subjects (74 subjects Scale Fourth Edition matrix reasoning (0.78), Block did not have full neuropsychological data and 40 sub- Design (0.74), and Corsi cubes (0.46). jects did not have sufficient information for the calcu- lation of the Framingham score) and were therefore excluded from the cognitive analyses. All participants Cardiopulmonary Exercise Testing gave written informed consent before participation in Before CPET evaluation, participants were assessed any study procedures, all of which conformed to the for potential absolute and relative contraindications for Declaration of Helsinki for research involving human maximal exhaustive exercise following the Guidelines of J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206883 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure Figure 1. Cohort consort diagram from the wider BBHI (Barcelona Brain Health Initiative) study and selection criteria for the current analysis. CPET indicates cardiopulmonary exercise testing; CRF, cardiorespiratory fitness; CVR, cardiovascular risk; HR, heart rate; MRI, magnetic resonance imaging; NP, neuropsychology assessment; and RER, respiratory exchange ratio. J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206884 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure the Spanish Society of Cardiology for Clinical Practice only 20.4% of participants reached a detectable VO2 pla- in Exercise Testing.42 The Physical Activity Readiness teau. To ensure that scaling VO2peak by total body mass Questionnaire43 was administered to assess for safety did not affect the associations with our outcomes, we to participate in the CPET. Additionally, participants replicated our results using allometric scaling48,49 (Data performed baseline spirometry (Ergoflow flowsensor; S1, Tables S1 through S3). Geratherm Respiratory, Bad Kissingen, Germany) and a baseline 12-­lead ECG recording before the test (WAM Medical Exam and Cardiovascular Risk Wireless Acquisition Module; Mortara, Milwaukee, WI). Assessment Individuals who had forced expiratory volume in 1 sec- A medical evaluation included a structured interview, ond of <80%, forced expiratory volume in 1 second/ which gathered past and present medical history (in- forced vital capacity ratio of >80%, or peak expiratory cluding diagnosis of diabetes), medication intake (in- flow of >75% did not complete the CPET evaluation. cluding antihypertensive drugs), alcohol and tobacco The CPET was performed using a modified consumption, absolute and relative risk factors for the Wasserman protocol44 on a cyclometer (Ergoselect 4 CPET, anthropometric measures (weight, height, body model; Ergoline, Bitz, Germany) with a respiratory gas mass index, and waist circumference), and blood pres- analysis system (Ergostik; Geratherm Respiratory). sure. Questionnaires about education history (including The modified Wasserman protocol44 consisted of number of years of formal higher education) and self-­ a 7-­minute warm-­ up phase (no load), a progressive reported physical activity (including the International workload phase, and a 5-­minute recovery phase (min- Physical Activity Questionnaire)50 were filled out by imal load). The slope of the progressive increase in workload was calculated individually by dividing the each participant. A fasting blood draw was performed expected maximum workload (calculated automatically to measure total cholesterol (millimoles per liter) and by the Bluecherry software [Geratherm Respiratory] high-­density lipoprotein (millimoles per liter). The modi- from height, weight, age, and sex) by 9, to derive a fied Framingham cardiovascular disease risk calculator progressive increase in workload that would result in a was then used to calculate the 5-­year risk of the de- maximal exercise test lasting ≈13 minutes. velopment of any cardiovascular disease,51 including Gas analysis was conducted using a tight-­fitting face the following variables: age (years), biological sex, total mask (Hans Rudolph, Shawnee, KS), and the following cholesterol, high-­ density lipoprotein, systolic blood measures were recorded continuously: oxygen con- pressure, treatment for hypertension, smoker status, sumption, oxygen uptake (efficiency slope), and respi- and diabetes status. In addition, we also we calculated ratory exchange ratio (VO₂/VCO₂), 12-­lead ECG, heart the modified Framingham cardiovascular disease risk rate (beats per minute, from a 12-­lead ECG), and pulse calculator using a formula that was adapted specifi- oximetry. Blood pressure, measured manually from the cally for the Catalan population (the Registre Gironí del left arm using a blood pressure cuff (Boso Medicus Cor52,53). The latter is presented in Data S2, Tables S4 X; Boso, Jungingen, Germany) and a handheld sphyg- through S6, and Figure S1. momanometer (MDF Instruments, Agoura Hills, CA) and perceived effort, measured via the Spanish trans- Structural MRI lation of the Borg scale,45 were recorded every 2.5 Participants underwent a high-­resolution minutes. Ventilatory thresholds (lactate threshold and (0.8×0.8×0.8 mm3) 3-­ dimensional magnetization-­ respiratory compensation point) were calculated using prepared rapid gradient-­echo T1-­weighted structural the V-­slope method.44 brain MRI session using a 3T Siemens Magnetom A test was considered complete under the following Prisma machine. A total of 208 contiguous axial slices criteria: verbal manifestation of exhaustion, Borg score of were obtained in ascending fashion (sequence param- ≥9, heart rate of ±10 bpm of heart rate max, or inability to eters of repetition time=2400 ms, echo time=2.22 ms, maintain pedal cadence (≈70 rpm). The highest full min- inversion time=1000 ms, flip angle=8º, slice thick- ute VO2 uptake (maximal oxygen consumption) value ob- ness=0.8 mm, and field of view=256 mm). Additionally, served during the final minute of the test was accepted as a high-­ resolution (0.8×0.8×0.8 mm3) 3-­dimensional the functional aerobic capacity (VO2 plateau). Whenever SPACE T2-­weighted structural brain MRI was under- a VO2max plateau could not be detected, we applied the taken, using the same device (sequence parameters following 2 metrics to determine the validity of the CPET of repetition time=3200 ms, echo time=563 ms, flip results: (1) the maximal respiratory exchange ratio (respi- angle=120º, slice thickness=0.8 mm, and field of ratory exchange ratio of ≥1.0, considered to be indicative view=256 mm). Image quality control measures were of true maximal oxygen uptake),46,47 and (2) the reached implemented manually by a trained MRI technician. target heart rate ≥80% of the maximum theoretical ex- Cortical reconstruction and volumetric segmenta- pected heart rate (220−age). We use the term VO2peak tion were performed with the Freesurfer image analy- (oxygen uptake during peak exercise) herein because sis suite, which is documented and freely available for J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206885 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure download online (http://surfer.nmr.mgh.harva​rd.edu/). Austria). The associations between predictor variables A 3-­dimensional cortical surface model was created (VO2peak, Framingham score) and outcome measures by running the recon-­all processing stream with de- (domain-­ specific cognitive performance and cortical fault parameters,54 except for the addition of the T2 flag thickness measures) were analyzed using multiple lin- for the improvement of pial surfaces reconstruction. ear regression, controlling for age, education, socioec- Therefore, inputs for this command were T1-­w volumes onomic status, body mass index, waist perimeter, and and T2-­w volumes. Briefly, automated Talairach trans- biological sex for the VO2peak models and education formation55 and intensity normalization56 were followed and socioeconomic status for the CVR models (age by non–­brain tissue removal,57 tessellation of the gray and biological sex are factors used to calculate the and white matter boundary, and automated topology Framingham score). Model assumptions were checked correction.58 Finally, surface deformation enabled the using Q-­ Q plots and fitted versus residual plots in detection of tissue boundaries; gray–­white and gray–­ R, and the normality of the residuals was formally cerebrospinal fluid (CSF) borders.54 The cortical sur- checked using Shapiro-­Wilk tests of normality. Outlier faces were then inflated and registered to a spherical observations that had influence on the models were re- atlas that used individual cortical folding patterns to moved using Cook’s distance (observed using Cook’s match cortical geometry across subjects.57,59,60 distance of >0.5) and R’s outlier package (upper limit of n=10 in any given model). To conform to the model Cortical Thickness Analyses assumptions, VO2peak and Framingham scores were log10 transformed before analyses. Model fitness is Individual cortical thickness maps were calculated as presented as adjusted R2 values, and significance was the closest distance from the gray–­white matter bound- considered at the P<0.05 level. We present standard- ary to the gray–­ cerebrospinal fluid boundary at each ized β coefficients as the strength of the relationship vertex on the tessellated surface.54 Then, a Gaussian between our predictor and outcome variables. That is, kernel of 10-­mm full width at half maximum was applied for every 1-­unit increase in the predictor, there is an X to these maps. Vertex-­wise general linear models were standard deviation increase in the outcome. Multiple run in FreeSurfer version 6.0, with cortical thickness as comparisons were corrected for using Bejamini and the dependent variable and either CVR or CRF as the Hochberg’s false discovery rate, at a q value of 0.05, independent variables, with education, age, body mass after pooling the P values from the regression analyses index, socioeconomic status, waist perimeter, and bio- for each predictor model. For the VO2peak models, the logical sex as controlling predictors of no interest. A total cohort was dichotomized into younger middle-­aged of 5 models were fitted: Models 1 and 2 included CVR as (aged 40–­54 years) and older middle-­aged (aged 55–­ the predictor of interest, using the Registre Gironí del Cor 65 years) groups to gain greater sensitivity to further and Framingham scores, respectively. Models 3, 4, and explore age-­related associations. 5 addressed CRF (ie, VO2peak) as the predictor of inter- Mediation analysis using the R mediation package62 est. Whereas Models 1, 2, and 3 were fitted for the whole was performed to assess whether cortical thickness set of observations; the fitting of Models 4 and 5 were mediated the associations between VO2peak and restricted to a dichotomization of the sample according Framingham and cognitive performance, taking into ac- to their age: younger middle-­aged (aged 40–­54 years) count all covariates (age, biological sex, socioeconomic and older middle-­aged (aged 55–­65 years), respectively. status, education, waist perimeter, and body mass For each model, regions where the predictor of interest index). The total effects (effect of X [predictor variable] significantly predicted cortical thickness were identified on Y [outcome variable]), direct effects (effect of X on using a method provided by FreeSurfer (ie, mri_glmfit-­ Y taking into account M [mediator] [average direct ef- sim). Here, multiple comparisons correction of whole-­ fect]) and indirect effects (or mediation effect, the total brain vertices was performed by computing P values for effect minus the direct effect [average causal mediation contiguous clusters of vertices based on Monte-­Carlo effect]) are reported. The presence of statistical media- Null-­Z simulations61 and permutation62 (with 10 000 itera- tion was determined through nonparametric bootstrap tions per simulation). This method assigns a P value to confidence intervals via 1000 bootstrap resamples of each resulting cluster. Consequently, we used a cluster-­ the estimated indirect effect. The estimated indirect (av- forming threshold of P<0.005 in cardiovascular risk mod- erage causal mediation effect) effect corresponds to the els (ie, Models 1 and 2), and P<0.05 in cardiorespiratory reduction in the independent variable effect on the de- fitness models (ie, Models 3, 4, and 5) and a cluster sig- pendent variable when adjusted for the mediator. nificance threshold of P<0.05 in all models. Statistical Analysis RESULTS All statistical analyses were performed in R version A total of 501 (248 women) participants with a mean±SD 3.6.3 (R Foundation for Statistical Computing, Vienna, age of 53.58±6.96 years (range, 40–­ 65 years) J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206886 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure completed the study. Our sample is generally charac- Table 1. Participant Characteristics terized by White, highly educated, and cognitive and Age cardiovascularly healthy individuals. Full demographic Age, y, mean±SD 53.58±6.96 information is found in Table 1. Aged 40–­54 y, n (%) 288 (54) Aged ≥55 y, n (%) 243 (46) Associations Between VO2peak, Sex, n (%) Framingham, and Cognitive Functions Men 283 (53) At the whole group level, no significant associations Women 248 (47) between VO2peak and cognitive functions were found Education, n (%) (Table 2). When we dichotomized our sample into Primary 16 (3) younger middle-­aged (aged 40–­54 years) and older Secondary 125 (24) middle-­aged (aged 55–­65 years) we found no signifi- Higher 390 (73) cant correlations between any cognitive domain and Cognitive profile, mean±SD (percentile) VO2peak in the younger group (Table 2). However, in Block design 12.12±3.06 (75) the older middle-­aged adults, we did find a signifi- Matrix reasoning 13.20±2.64 (84) cant and positive association between VO2peak and visuospatial reasoning and problem solving (β=3.16, Direct digits 10.72±3.05 (63) P=0.049), which remained significant after false dis- Indirect digits 11.16±2.62 (63) covery rate corrections (false discovery rate P=0.0499) Digit symbol 13.67±2.69 (91) (Figure 2). Letter–­number sequencing 14.38±2.56 (91) For CVR, we found a significant negative association Cancellation test 11.41±2.76 (63) between Framingham score and the following cognitive TMT-­A 11.26±2.77 (63) abilities: visuospatial ability (β=−0.046, P=0.002), TMT-­B 8.66±2.16 (37) processing speed (β=−0.115, P<0.001), flexibility Corsi cubes 13.99±2.52 (91) (β=−0.054, P<0.001), and verbal memory (β=−0.120, Fitness evaluation P<0.001), but not working memory (β=−0.010, VO2peak, mL/kg per min, mean±SD 24.9±7.18 P=0.502). Full model results are seen in Table 3 and 40–­54 y, VO2peak, mL/kg per min, 26.44±7.22 depicted in Figure 3. mean±SD 55 y and above VO2peak, mL/kg per 23.28±6.57 Cortical Thickness min, mean±SD At the whole group level, higher VO2peak was sig- BMI, kg/m2, mean±SD 25.40±4.01 nificantly associated with greater cortical thickness in IPAQ, METs, min/wk 2558.13±2486.57 the left prefrontal cortex (rostral middle frontal gyrus) Cardiovascular status (cluster-­ wise corrected with a vertex-­ wise thresh- Smoker, n (%) 58 (11) old P<0.05, cluster-­wise P<0.05) (Data S3, Table S7, Diabetes, n (%) 7 (1) Figure S2A). In the young middle-­aged group (aged Hypertension, n (%) 40 (8) 40–­54 years) VO2peak was not associated with any Systolic blood pressure, mean±SD 124±16.13 specific gyrus (Data S3, Table S8), whereas in the old Cholesterol, mg/dL, mean±SD 177.83±75.14 middle-­aged group (aged ≥55 years), associations with HDL, mm/dL, mean±SD 54.5±24.77 left prefrontal regions (left rostral middle frontal) and Framingham 5-­y risk, %, mean±SD 8.45±6.76 left temporal regions (superior temporal gyrus) were seen (Data S3, Table S9, Figure S2B). Moreover, in the Percentiles extracted from the Wechsler Adult Intelligence Scale Fourth Edition toolbox (Wechsler63). BMI indicates body mass index; HDL, high-­ older middle-­aged group, cortical thickness in the left density lipoprotein; IPAQ, International Physical Activity Questionnaire; prefrontal gyrus mediated the relationship between METs, metabolic equivalent of task; TMT-­A, Trail Making Test A; TMT-­B, Trail VO2peak and visuospatial reasoning abilities (Figure 4, Making Test B; and VO2peak, oxygen uptake during peak exercise. Data S3, Table S10). Higher Framingham risk score was significantly asso- flexibility, and memory after controlling for education ciated with lower cortical thickness across different and monthly incomes. In visuospatial problem solving, cortical regions (18 clusters) of both hemispheres in- the following regions were significantly mediating its cluding frontal, parietal, temporal, and medial (insula, relationship with Framingham: left postcentral gyrus, cuneus) cortices (cluster-­wise corrected with a vertex-­ left pars triangularis, left insula, left cuneus, left cau- wise threshold P<0.005, cluster-­wise P<0.05) (Data dal anterior cingulate gyrus, left transverse temporal S4, Table S11 and Figure S3). Cortical thickness sig- gyrus, and right supramarginal region. The relationship nificantly mediated the relation between Framingham between processing speed and Framingham was sig- and visuospatial problem solving, processing speed, nificantly mediated by right cuneus, whereas flexibility J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206887 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure Table 2. Associations Between VO2peak and Cognitive Domains β SE P value R2 Whole population Memory −2.070 1.158 0.074 0.181 Working memory −1.885 1.117 0.092 0.009 Flexibility −0.632 0.780 0.418 0.152 Processing speed <0.001 <0.001 0.715 0.193 Visuospatial problem solving 1.167 1.031 0.258 0.208 40–­54 y Memory −1.475 1.336 0.270 0.186 Working memory −2.229 1.542 0.149 0.007 Flexibility 0.211 0.891 0.813 0.115 Processing speed −1.035 1.374 0.452 0.164 Visuospatial problem solving −0.709 1.347 0.598 0.129 55–­65 y Memory −3.231 1.971 0.102 0.102 Working memory −1.284 1.697 0.450 −0.025 Flexibility −0.667 1.370 0.626 0.094 Processing speed 2.053 1.844 0.267 0.132 Visuospatial problem solving 3.165 1.604 0.049* 0.160 All models are controlling for age, biological sex, body mass index, waist perimeter, socioeconomic status, and education as covariates. R2 values are adjusted for all predictors. VO2peak indicates oxygen uptake during peak exercise. *Survives false discover rate corrections. had different gyri that mediated its relationship with thickness in a disperse network spanning multiple Framingham, in particular, left postcentral gyrus, left cortical regions across both hemispheres mediated insula, left caudal anterior cingulate gyrus, left trans- the relationship between CVR and multiple domains of verse temporal gyrus, right inferior parietal gyrus, right cognition. cuneus, right supramarginal region, and right supe- We found associations between CRF and frontal-­ rior frontal gyrus. Lastly, left triangularis and left and loading cognitive abilities (visuospatial reasoning) only right cuneus significantly mediated the relationship be- in those aged ≥55 years. These results are supported tween Framingham and memory (Figure 4, Data S4, by earlier work in older adults6,64,65 and more recent Table S12A through S12D). work in middle-­aged adults.66 We extend those pre- vious results in 2 important ways. First, although re- gional specificity of high CRF to the frontal lobe in older DISCUSSION adults has been reported,67–­71 the mediating effect of In the present study, we demonstrate that some of the cortical thickness in frontal regions on the relationship well-­established relationships between determinants between CRF and cognition in midlife is novel, extend- of cardiovascular health and brain health that exist in ing previous reports of a similar mediating effect in older age are already present in late middle age. In our older adults.71 Frontal regions are particularly suscep- sample of healthy middle-­aged adults, CRF and CVR, tible to age-­related cortical thinning,72 and critical for 2 independent clinical predictors of cardiovascular visuospatial73 problem solving and executive abilities.74 health, had distinct associations with neuropsychologi- High CRF has been shown to decrease small-­vessel cal metrics of cognitive brain health. CRF had domain-­ ischemic disease, which often preferentially affects specific associations with cognitive abilities highly the frontal/subcortical region of the brain,75 providing reliant on the frontal lobe, but only in individuals aged a possible explanation for the reported regional spec- ≥55 years. In contrast, CVR had domain-­general asso- ificity. Furthermore, white matter tracts have been ciations with various cognitive abilities within the whole implicated as an indirect path between CRF and bet- sample. Importantly, mediation analyses strengthened ter performance of frontal cognitive abilities.76 Future our findings by revealing that the relationships between planned studies will also examine the integrity of white each predictor and cognition were driven by distinct matter tracts in this population. patterns of cortical thickness. Cortical thickness in Second, the age-­ specific associations between frontal regions mediated the relationship between CRF CRF and cognitive abilities can be explained in several and visuospatial problem solving, whereas cortical ways. It is possible that our neuropsychological test J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206888 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure that variations in brain structure and function precede the onset of behavioral symptoms of cognitive decline by years,77–­79 further strengthening the importance of engaging in modifiable lifestyle behaviors relevant for the promotion and maintenance of brain health in early midlife. Given our analysis is cross-­ sectional, we can only speculate about the directionality of these re- sults. Based on our analyses alone, in addition to our interpretations herein, it is just as plausible that higher cognitive resources lead to higher levels of fit- ness. Although numerous interventional studies have demonstrated that aerobic fitness training can improve cognition,80,81 other modes of exercise have also been found to positively influence cognition.13 Furthermore, longitudinal studies have suggested that cognitive re- sources themselves82 are predictive of engagement in moderate-­intensity physical exercise beyond the age of 50 years (a key modifier of CRF).83 In addition, in a large longitudinal study, engagement in moderate physical exercise began to decline starting some 8 to 12 years Figure 2. Significant positive relationship between VO2peak before dementia diagnosis, and in those who did not (oxygen uptake during peak exercise) and visuospatial have an eventual dementia diagnosis, total physical reasoning and problem-­solving abilities in the older middle-­ activity continued to increase through older age.84 As aged group (aged 55–­6 5 years) after controlling for age and previously mentioned, physical activity is one of many biological sex. Survives false discovery rate multiple comparison correction factors found to improve CRF.85 Taken together, the re- (Table 2). lationship between CRF and cognition may ultimately be bidirectional, and because we cannot delineate this directionality, the result that these relationships exist battery may have been insufficiently sensitive for the in midlife in healthy adults is itself important to know younger subgroup (aged 40–­54 years), and a ceiling for targeting through longitudinal studies beginning in effect may have masked potential associations be- midlife or earlier. tween CRF and cognition. Conversely, and perhaps In contrast to the domain-­specific associations with more likely, the relationship between CRF and neuro- CRF, we found that CVR was associated with perfor- cognitive function may be stronger in late middle age, mance in many cognitive abilities, including visuospa- when measurable age-­related change in neurocogni- tial reasoning, but also cognitive flexibility, processing tive performance is more likely to be seen. Our sample speed, and memory. Similar findings have been widely of healthy adults scored in the higher percentiles for reported both in older adults86–­92 and in middle-­aged performance on these cognitive tasks (Table 1). One adults.7,93–­97 We build on these findings by demonstrat- implication of our findings is the existence of a period ing that cortical thickness in disperse cortical regions from early to late middle age when it becomes particu- across bihemispheric frontal, cuneus, parietal, temporal, larly critical to maintain CRF to optimize cognitive brain and cingulate areas mediated the relationship between health as we age. Longitudinal studies are needed to low CVR and better cognitive performance. The overlap explore this possibility further. One potential interpreta- between the clusters identified herein and cortical areas tion for this finding could reflect the growing evidence considered to be particularly sensitive to the effects of Table 3. Associations Between Framingham and Cognitive Domains β SE P value R2 Memory −0.120 0.016 <0.001* 0.149 Working memory −0.010 0.015 0.502 −0.0003 Flexibility −0.054 0.011 <0.001* 0.094 Processing speed −0.115 0.016 <0.001* 0.120 Visuospatial problem solving −0.046 0.015 0.002* 0.072 All models are controlling for socioeconomic status and education as covariates. R2 values are adjusted for all predictors. *Survives false discovery rate corrections. J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.0206889 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure Figure 3. Significant negative relationships between cardiovascular risk (Framingham 5-­year risk score) and multiple cognitive domains including flexibility, visuospatial problem-­ solving abilities, processing speed, and memory, after controlling for education (total number of years) and monthly incomes. All models survive false discovery rate multiple comparison corrections (Table 3). early cognitive impairment and Alzheimer’s dementia could be explained by the fact that CVR is mostly associ- pathology (ie, the inferior and anterior temporal lobe, ated with small lesions in cerebral white matter that exhibit inferior and superior temporal lobe, and posterior cin- a more disperse representation over striatal, cortico–­ gulate cortex),98 supports existing evidence that cardio- cortical, and cortical–­subcortical pathways.90 As men- vascular risk factors are also cognitive risk factors.21–­24 tioned, future studies will additionally assess the integrity The region-­general pattern of cortical thickness im- of white matter tracts. Importantly, it is noteworthy that plicated in the relationship between CVR and cognition management of CVR involves not one but many healthy J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.02068810 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure Figure 4. Cortical thickness in various regions mediated relationships between our predictors (VO2peak and Framingham) and cognitive domains. The relationship between each predictor and significant cortical thickness clusters (X [predictor variable] on M [mediator]) are found in Tables S7 through S9 along with full mediation model results (Table S10). Orange arrows depict the exact cluster, which mediates the relationship between X (predictor) and Y (cognitive domain outcome). The mediated effect is calculated as the difference between the estimates from the total and direct effects (see Tables S10 and S12A through S12D) which correspond to the reduction in the independent variable (X) effect on the dependent variable (Y) when adjusted for the mediator (M). The total effect (X on Y) is seen under the horizontal arrow representing the β coefficient followed by the 95% CIs in parentheses. The average causal mediation effect (X [predictor variable] on Y [outcome variable] including M [mediator]) is seen between square brackets following the direct effect. In the case of VO2peak (oxygen uptake during peak exercise) on visuospatial problem solving (top), of the estimated total effect (0.07, note this is the unstandardized β coefficient), an estimated 0.01 is because of the mediator (cortical thickness in the left prefrontal gyrus). behaviors (avoiding smoking, weight management, and One important point that distinguishes our results healthy eating habits, to name a few), and we found this from previous studies is that our sample was partic- collective effort to be manifested by diffuse patterns of ularly healthy from a heart-­health perspective. For brain structure that likely support the wide range of cog- instance, our sample had a relatively high group av- nitive abilities that were associated with CVR. erage in CRF (24.9±7.18 mL/kg per minute) (see age J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.02068811 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure and gender norms for the American population99) and cognitive brain health, our study has unique strengths. a low group average for CVR (estimated to be ≈8% risk We add to previous research by examining not just 1 of a future cardiovascular event in 5 years). The fact but 2 independent predictors (CRF, CVR), and by using that in this overall healthy sample, individual variations a detailed and comprehensive neuropsychological as- in CRF and CVR were still associated with cognitive sessment in over 500 healthy middle-­aged adults free behavior, and brain structure demonstrates that these from clinically detectable cognitive deficits. Finally, established biomarkers of heart health in older adults given that the relationship between cardiovascular may also be sufficiently sensitive for better under- health and cognition is likely to be underpinned by standing cognitive trajectories in early and late middle brain structure, we also advance previous studies by age. It is pertinent to highlight though that these same using neuroimaging and analytical methods to demon- characteristics of this sample may affect generalizabil- strate the mediating effect of brain structure on the re- ity to other populations. The BBHI sample is by de- lationships between CRF/CVR and cognition. sign particularly young and healthy, because our data There are also limitations to our study. Because are cross-­sectional samples of this longitudinal cohort of the cross-­ sectional nature of our results, it was study that aims to better understand and characterize not possible to make any kind of inference about neurobiological determinants of cognitive brain health casual relationships. In addition, the normalization from middle to late life. As such, this sample is exposed of VO2peak to total body mass (referred to as sim- to known environmental factors reported to strongly ple ratio standard) can produce confounding results contribute to cognitive brain health, such as adherence because of individual differences in adiposity levels.48 to a Mediterranean diet, engagement in physical ac- We aimed to minimize this source of bias by includ- tivity, and leisure activities. It is also important to note ing waist circumference as a covariate in all analyses. that this is a mostly White sample, which is relevant Furthermore, we replicated our results using allo- because cardiovascular risk has differential associa- metric scaling of VO2peak to ensure that scaling to tions with other racial and ethnic groups, particularly in total body mass did not confound the associations Black and Latino individuals and other minority groups. with cognition.48,49 Future studies are encouraged to As such, comprehensive and inclusive brain health measure adiposity levels and normalize VO2peak to strategies must also address this knowledge gap by fat-­free mass. Considerations about biological sex examining such associations between determinants interactions are critical in this work given reported of cardiovascular health and cognitive brain health in differences in CRF,101–­103 CVR,104 and trajectories of other racial and ethnic groups in midlife. cognitive performance105 between men and women. Importantly, although cognitive brain health is a top We will address biological sex interactions in a future health-­related priority for people when they reach older planned study. Finally, we did not assess other poten- age,100 our findings highlight the relevance of creating a tial factors that influence the relationships seen such cognitive brain health plan in middle age. Given growing as diet, physical activity levels, and motor skills. evidence demonstrating changes in the brain related to Taken together, our findings show that even in the onset of neurodegenerative disorders begin some younger and healthy middle-­aged adults with relatively 10 to 20 years before the onset of symptoms,77–­79 it is high CRF and low CVR, relationships between these critical that strategies to mitigate age-­related cognitive modifiable factors that may contribute to cognitive/brain decline and promote cognitive brain health need to be reserve and cognition exist. Furthermore, we shed light introduced decades earlier in midlife. CRF and CVR on a potential mechanistic pathway (cortical thickness) are both modifiable factors, and thus our results could that may contribute to this relationship. The implications potentially suggest that by adopting lifestyle changes of our study lie within the potential importance of engag- that promote heart health in middle age, it may be pos- ing in modifiable lifestyle behaviors that can promote sible to actively steer the course of one’s cognitive tra- heart health, early in midlife, long before the onset of jectory in later life. Our results (Data S5) also reproduce measurable cognitive decline, which can be assessed the ubiquitous association between greater CRF and in future longitudinal and interventional study designs. greater time practicing physical activity (Figure S4). Thus, engagement in a physically active lifestyle is a potential strategy (among many, including diet, sleep, ARTICLE INFORMATION and other cognitively stimulating activities), that are Received December 29, 2020; accepted July 13, 2021. likely to have a positive effect on cognitive brain health Affiliations even in midlife. Albeit these conclusions need to be Guttmann Brain Health Institute, Institut Guttmann, Institut Universitari de supported by longitudinal and interventional studies. Neurorehabilitació Adscrit a la UAB, Badalona, Spain (G.E., G.C., S.A., M.C., J.S., M.R., S.D., V.A., C.P., J.M.T., D.B., Á.P.); Department of Neurology, Although our results are complimentary to several University of Miami Miller School of Medicine, Miami, FL (J.G.); Department of previous and large population studies investigating Medicine, Universitat Autònoma de Barcelona, Bellaterra, Spain (G.E., G.C., associations between cardiovascular outcomes and S.A., J.S., M.R., S.D., V.A., C.P., J.M.T.); Department of Medicine, Facultat J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.02068812 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure de Medicina i Ciències de la Salut i Institut de Neurociències, Universitat de cognitive aging in the Northern Manhattan Study. J Am Heart Assoc. Barcelona, Spain (M.C., D.B.); Department of Neurology, Harvard Medical 2016;5:e002731. DOI: 10.1161/JAHA.115.002731. School, Boston, MA (Á.P.); Hinda and Arthur Marcus Institute for Aging 8. Féart C, Samieri C, Allès B, Barberger-­Gateau P. Potential benefits of Research and Deanna and Sidney Wolk Center for Memory Health, Hebrew adherence to the Mediterranean diet on cognitive health. Proc Nutr SeniorLife, Boston, MA (Á.P.); and Department of Psychology, Center for Soc. 2013;72:140–­152. DOI: 10.1017/S0029​66511​2002959. Cognitive and Brain Health, Northeastern University, Boston, MA (T.P.M.). 9. Hardman RJ, Kennedy G, Macpherson H, Scholey AB, Pipingas A. Adherence to a Mediterranean-­ style diet and effects on cognition Acknowledgments in adults: a qualitative evaluation and systematic review of longitu- A special thanks is extended to all participants and other partners (Ad-­ dinal and prospective trials. Front Nutr. 2016;3:22. DOI: 10.3389/ Salutem Institute, Sodexo, I.C.A. Informàtica y Comunicaciones Avanzadas, fnut.2016.00022. Neuroelectrics, Corporació Catalana de Mitjans Audiovisuals, Club 10. Pace-­Schott EF, Spencer RMC. Age-­related changes in the cogni- Metropolitan, Casa Ametller, and Agència de Qualitat i Avaluació Sanitàries tive function of sleep. Prog Brain Res. 2011;191:75–­89. DOI: 10.1016/ de Catalunya-­AQuAS) for their invaluable collaboration. B978-­0 -­444-­53752​-­2.00012​-­6. 11. Scullin MK, Bliwise DL. Sleep, cognition, and normal aging: integrat- Sources of Funding ing a half-­century of multidisciplinary research. Perspect Psychol Sci. The research leading to the reported results is funded by Q12 grants from 2015;10:97–­137. DOI: 10.1177/17456​91614​556680. the La Caixa Foundation (grant agreement LCF/PR/PR16/11110004), Institut 12. Spira AP, Chen-­Edinboro LP, Wu MN, Yaffe K. Impact of sleep on Guttmann, and Fundació Abertis. Dr Pascual-­Leone was partly supported the risk of cognitive decline and dementia. Curr Opin Psychiatry. by the National Institutes of Health (R24AG06142 and P01 AG031720) and 2014;27:478–­483. DOI: 10.1097/YCO.00000​0 0000​0 00106. the National Science Foundation. Dr Bartrés-­Faz was funded by the Spanish 13. Ludyga S, Gerber M, Pühse U, Looser VN, Kamijo K. Systematic Ministry of Science, Innovation, Universities (RTI2018-­ 095181-­B-­C21) review and meta-­analysis investigating moderators of long-­term ef- and ICREA Academia 2019 award research grants. Dr Tormos was partly fects of exercise on cognition in healthy individuals. Nat Hum Behav. supported by Fundació Joan Ribas Araquistain_Fjra, AGAUR, Agència de 2020;4:603– ­612. DOI: 10.1038/s4156​2-­020-­0851-­8. Gestió d’Ajuts Universitaris i de Recerca (2018 PROD 00172), FUNDACIÓ 14. Launer LJ. The epidemiologic study of dementia: a life-­long quest? LA MARATÓ DE TV3 (201735.10), and the European Commission (Call Neurobiol Aging. 2005;26:335–­ 340. DOI: 10.1016/j.neuro​ biola​ H2020-­SC1-­2016-­2017_RIA_777107). Dr Gomes-­Osman was supported ging.2004.03.016. by the National Center for Advancing Translational Sciences of the National 15. Sperling RA, Aisen PS, Beckett LA, Bennett DA, Craft S, Fagan AM, Institutes of Health under award KL2TR002737. The content is solely the Iwatsubo T, Jack CR, Kaye J, Montine TJ, et al. Toward defining the responsibility of the authors and does not necessarily represent the official preclinical stages of Alzheimer’s disease: recommendations from the views of the National Institutes of Health. National Institute on Aging-­Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. Disclosures 2011;7:280–­292. DOI: 10.1016/j.jalz.2011.03.003. Dr Pascual-­Leone is a cofounder of Linus Health and TI Solutions AG; serves 16. Newson RS, Kemps EB. Cardiorespiratory fitness as a predictor of on the scientific advisory boards for Starlab Neuroscience, Neuroelectrics, successful cognitive ageing. J Clin Exp Neuropsychol. 2006;28:949–­ Magstim Inc., and MedRhythms; and is listed as an inventor on several is- 967. DOI: 10.1080/13803​39059​1004356. sued and pending patents on the real-­time integration of noninvasive brain 17. Wendell CR, Gunstad J, Waldstein SR, Wright JG, Ferrucci L, stimulation with electroencephalography and magnetic resonance imaging. Zonderman AB. Cardiorespiratory fitness and accelerated cognitive The remaining authors have no disclosures to report. decline with aging. J Gerontol A Biol Sci Med Sci. 2014;69:455–­462. DOI: 10.1093/geron​a /glt144. Supplementary Material 18. Zhu N, Jacobs DR, Schreiner PJ, Yaffe K, Bryan N, Launer LJ, Whitmer Data S1–­S5 RA, Sidney S, Demerath E, Thomas W, et al. Cardiorespiratory fitness Tables S1–­S12 and cognitive function in middle age: the CARDIA study. Neurology. Figures S1–­S4 2014;82:1339–­1346. DOI: 10.1212/WNL.00000​0 0000​0 00310. References 108–­110 19. Gorelick PB, Furie KL, Iadecola C, Smith EE, Waddy SP, Lloyd-­ Jones DM, Bae H-­J, Bauman MA, Dichgans M, Duncan PW, et al.; American Heart Association/American Stroke Association. Defining optimal brain health in adults: a presidential advisory from the American Heart Association/American Stroke Association. Stroke. REFERENCES 2017;48:e284–­e303. DOI: 10.1161/STR.00000​0 0000​0 00148. 1. Crimmins EM. Lifespan and healthspan: past, present, and promise. 20. Hörder H, Johansson L, Guo X, Grimby G, Kern S, Östling S, Skoog Gerontologist. 2015;55:901–­911. DOI: 10.1093/geron​t /gnv130. I. Midlife cardiovascular fitness and dementia: a 44-­year longitudinal 2. Mathers CD, Loncar D. Projections of global mortality and burden population study in women. Neurology. 2018;90:e1298–­e1305. DOI: of disease from 2002 to 2030. PLoS Medicine. 2006;3:e442. DOI: 10.1212/WNL.00000​0 0000​0 05290. 10.1371/journ​al.pmed.0030442. 21. Launer LJ, Ross GW, Petrovitch H, Masaki K, Foley D, White LR, 3. Pascual-­Leone A, Freitas C, Oberman L, Horvath JC, Halko M, Eldaief Havlik RJ. Midlife blood pressure and dementia: the Honolulu-­Asia M, Bashir S, Vernet M, Shafi M, Westover B, et al. Characterizing aging study. Neurobiol Aging. 2000;21:49–­ 55. DOI: 10.1016/S0197​ brain cortical plasticity and network dynamics across the age-­span -­4580(00)00096​-­8. in health and disease with TMS-­EEG and TMS-­fMRI. Brain Topogr. 22. Mielke MM, Rosenberg PB, Tschanz J, Cook L, Corcoran C, Hayden 2011;24:302–­315. DOI: 10.1007/s1054​8-­011-­0196-­8. KM, Norton M, Rabins PV, Green RC, Welsh-­Bohmer KA, et al. Vascular 4. Stern Y, Arenaza-­Urquijo EM, Bartrés-­Faz D, Belleville S, Cantilon M, factors predict rate of progression in Alzheimer disease. Neurology. Chetelat G, Ewers M, Franzmeier N, Kempermann G, Kremen WS, 2007;69:1850–­1858. DOI: 10.1212/01.wnl.00002​79520.59792.fe. et al. Whitepaper: defining and investigating cognitive reserve, brain 23. Ott A, Breteler MM, de Bruyne MC, van Harskamp F, Grobbee DE, reserve, and brain maintenance. Alzheimers Dement. 2020;16:1305–­ Hofman A. Atrial fibrillation and dementia in a population-­based study. 1311. DOI: 10.1016/j.jalz.2018.07.219. The Rotterdam Study. Stroke. 1997;28:316–­ 321. DOI: 10.1161/01. 5. Colcombe S, Kramer AF. Fitness effects on the cognitive function of STR.28.2.316. older adults: a meta-­analytic study. Psychol Sci. 2003;14:125–­130. 24. Skoog I, Lernfelt B, Landahl S, Palmertz B, Andreasson LA, Nilsson L, DOI: 10.1111/1467-­9280.t01-­1-­01430. Persson G, Odén A, Svanborg A. 15-­year longitudinal study of blood 6. Pentikäinen H, Savonen K, Ngandu T, Solomon A, Komulainen P, pressure and dementia. Lancet. 1996;347:1141–­ 1145. DOI: 10.1016/ Paajanen T, Antikainen R, Kivipelto M, Soininen H, Rauramaa R. S0140​-­6736(96)90608​-­X. Cardiorespiratory fitness and cognition: longitudinal associations in 25. Flicker L. Cardiovascular risk factors, cerebrovascular disease bur- the FINGER study. J Alzheimers Dis. 2019;68:961–­968. DOI: 10.3233/ den, and healthy brain aging. Clin Geriatr Med. 2010;26:17–­27. DOI: JAD-­180897. 10.1016/j.cger.2009.12.005. 7. Gardener H, Wright CB, Dong C, Cheung K, DeRosa J, Nannery 26. Leritz EC, McGlinchey RE, Kellison I, Rudolph JL, Milberg M, Stern Y, Elkind MSV, Sacco RL. Ideal cardiovascular health and WP. Cardiovascular disease risk factors and cognition in the J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.02068813 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure elderly. Curr Cardiovasc Risk Rep. 2011;5:407–­412. DOI: 10.1007/s1217​ 50. Roman-­ Viñas B, Serra-­ Majem L, Hagströmer M, Ribas-­ Barba L, 0-­011-​­0189-­x. Sjöström M, Segura-­ Cardona R. International Physical Activity 27. Levin BE, Llabre MM, Dong C, Elkind MSV, Stern Y, Rundek T, Sacco Questionnaire: reliability and validity in a Spanish population. Eur J RL, Wright CB. Modeling metabolic syndrome and its association with Sport Sci. 2010;10:297–­304. DOI: 10.1080/17461​39090​3426667. cognition: the Northern Manhattan study. J Int Neuropsychol Soc. 51. Anderson KM, Odell PM, Wilson PWF, Kannel WB. Cardiovascular dis- 2014;20:951–­960. DOI: 10.1017/S1355​61771​4000861. ease risk profiles. Am Heart J. 1991;121:293–­298. DOI: 10.1016/0002-­ 28. Hoppeler H. Deciphering V̇O2, max: limits of the genetic approach. J 8703(91)90861​-­B. Exp Biol. 2018;221:jeb164327. DOI: 10.1242/jeb.164327. 52. Marrugat J, Solanas P, D’Agostino R, Sullivan L, Ordovas J, Cordón 29. Williams CJ, Williams MG, Eynon N, Ashton KJ, Little JP, Wisloff U, F, Ramos R, Sala J, Masià R, Rohlfs I, et al. Coronary risk estimation Coombes JS. Genes to predict VO2max trainability: a systematic re- in Spain using a calibrated Framingham function. Rev Esp Cardiol. view. BMC Genom. 2017;18:831. DOI: 10.1186/s1286​4-­017-­4192-­6. 2003;56:253–­261. DOI: 10.1016/s0300​-­8932(03)76861​-­4. 30. Kathiresan S, Srivastava D. Genetics of human cardiovascular dis- 53. Marrugat J, Vila J, Baena-­Díez JM, Grau M, Sala J, Ramos R, Subirana ease. Cell. 2012;148:1242–­1257. DOI: 10.1016/j.cell.2012.03.001. I, Fitó M, Elosua R. Relative validity of the 10-­year cardiovascular risk 31. Cattaneo G, Bartrés-­Faz D, Morris TP, Sánchez JS, Macià D, Tarrero estimate in a population cohort of the REGICOR study. Rev Esp C, Tormos JM, Pascual-­ Leone A. The Barcelona Brain Health Cardiol. 2011;64:385–­394. DOI: 10.1016/j.recesp.2010.12.011. Initiative: a cohort study to define and promote determinants of brain 54. Fischl B, Dale AM. Measuring the thickness of the human cerebral health. Front Aging Neurosci. 2018;10:321. DOI: 10.3389/fnagi.2018. cortex from magnetic resonance images. Proc Natl Acad Sci USA. 00321. 2000;97:11050–­11055. DOI: 10.1073/pnas.20003​3797. 32. Cattaneo G, Bartrés-­Faz D, Morris TP, Solana Sánchez J, Macià D, 55. Talairach J, Tournoux P. Co-­Planar Stereotaxic Atlas of the Human Tormos JM, Pascual-­Leone A. The Barcelona Brain Health Initiative: Brain. Thieme Medical Publishers; 1988. Available at: https://www. cohort description and first follow-­up. PLoS One. 2020;15:e0228754. thieme.com/books​-­m ain/neuro​surge​r y/produ​c t/414-­c o-­p lana​r-­stere​ DOI: 10.1371/journ​al.pone.0228754. otaxi​c-­atlas​-­of-­the-­human​-­brain. Accessed October 13, 2020. 33. Cuschieri S. The STROBE guidelines. Saudi J Anaesth. 2019;13:S31–­ 56. Sled JG, Zijdenbos AP, Evans AC. A nonparametric method for au- S34. DOI: 10.4103/sja.SJA_543_18. tomatic correction of intensity nonuniformity in MRI data. IEEE Trans 34. Wechsler D. Wechsler Adult Intelligence Scale-­ Fourth Edition Med Imaging. 1998;17:87–­97. DOI: 10.1109/42.668698. (WAIS-­IV) [database record]. APA PsycTests. 2008. 57. Ségonne F, Dale AM, Busa E, Glessner M, Salat D, Hahn HK, Fischl B. 35. Reitan RM, Wolfson D. The Halstead-­Reitan Neuropsycholgical Test A hybrid approach to the skull stripping problem in MRI. NeuroImage. Battery: Therapy and Clinical Interpretation. Neuropsychological 2004;22:1060–­1075. DOI: 10.1016/j.neuro​image.2004.03.032. Press; 1985. 58. Fischl B, Liu A, Dale A. Automated manifold surgery: construct- 36. Shimoyama I, Ninchoji T, Uemura K. The finger-­tapping test. A quan- ing geometrically accurate and topologically correct models of the titative analysis. Arch Neurol. 1990;47:681–­684. DOI: 10.1001/archn​ human cerebral cortex. IEEE Trans Med Imaging. 2001;20:70–­80. DOI: eur.1990.00530​0 6009​5025. 10.1109/42.906426. 37. Pena-­ Casanova J, Quinones-­ Ubeda S, Quintana-­ Aparicio M, 59. Desikan RS, Ségonne F, Fischl B, Quinn BT, Dickerson BC, Blacker D, Aguilar M, Badenes D, Molinuevo JL, Torner L, Robles A, Barquero Buckner RL, Dale AM, Maguire RP, Hyman BT, et al. An automated la- MS, Villanueva C, et al. Spanish Multicenter Normative Studies beling system for subdividing the human cerebral cortex on MRI scans (NEURONORMA Project): norms for verbal span, visuospatial span, into gyral based regions of interest. NeuroImage. 2006;31:968–­980. letter and number sequencing, trail making test, and symbol digit mo- DOI: 10.1016/j.neuro​image.2006.01.021. dalities test. Arch Clin Neuropsychol. 2009;24:321–­341. DOI: 10.1093/ 60. Fischl B, Sereno MI, Dale AM. Cortical surface-­based analysis: II: infla- arcli​n/acp038. tion, flattening, and a surface-­based coordinate system. NeuroImage. 38. Rey A. L’examen Clinique en psychologie. Presses Universitaires de 1999;9:195–­207. DOI: 10.1006/nimg.1998.0396. France; 1958. 61. Hagler DJ, Saygin AP, Sereno MI. Smoothing and cluster thresholding 39. Tamayo F, Casals-­Coll M, Sánchez-­Benavides G, Quintana M, Manero for cortical surface-­based group analysis of fMRI data. NeuroImage. RM, Rognoni T, Calvo L, Palomo R, Aranciva F, Peña-­Casanova J,. et 2006;33:1093–­1103. DOI: 10.1016/j.neuro​image.2006.07.036. al.Neurologia. 2012;27:319–­329. DOI: 10.1016/j.nrl.2011.12.020. 62. Tingley D, Yamamoto T, Hirose K, Keele L, Imai K. Mediation: R 40. Lee S, Jennrich RI. A study of algorithms for covariance structure anal- package for causal mediation analysis. UCLA Statistics/American ysis with specific comparisons using factor analysis. Psychometrika. Statistical Association. 2014. Available at: https://dspace.mit.edu/ 1979;44:99–­113. DOI: 10.1007/BF022​93789. handl​e/1721.1/91154. Accessed March 19, 2021. 41. Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate Data 63. Wechsler D. WAIS-­ IV. Escala de Inteligencia de Wechsler Para Analysis. 5th ed. Prentice Hall; 1998. Adultos-­IV. Manual de Aplicación y Corrección. NCS Pearson, Inc.; 42. Arós F, Boraita A, Alegría E, Alonso AM, Bardají A, Lamiel R, Luengo 2012. E, Rabadán M, Alijarde M, Aznar J, et al. [Guidelines of the Spanish 64. Barnes DE, Yaffe K, Satariano WA, Tager IB. A longitudi- Society of Cardiology for clinical practice in exercise testing]. Rev Esp nal study of cardiorespiratory fitness and cognitive function in Cardiol. 2000;53:1063–­1094. healthy older adults. J Am Geriatr Soc. 2003;51:459–­ 465. DOI: 43. Adams R. Revised Physical Activity Readiness Questionnaire. Can 10.1046/j.1532-­5415.2003.51153.x. Fam Physician. 1999;45:992, 995, 1004–­1005. 65. Colcombe S, Kramer AF. Fitness effects on the cognitive function 44. Wasserman K, Hansen J, Sue D, Whipp B, Casaburi R. Principles of of older adults. Psychol Sci. 2003;14:125. DOI: 10.1111/1467-­ 9280. Exercise Testing and Interpretation. 4th ed. Williams and Wilkins; 2004. t01-­1-­01430. 45. Borg GAV, Noble BJ. Perceived exertion. Exerc Sport Sci Rev. 66. Boots EA, Schultz SA, Oh JM, Larson J, Edwards D, Cook D, Koscik 1974;2:131–­153. DOI: 10.1249/00003​677-­19740​0 020-­0 0006. RL, Dowling MN, Gallagher CL, Carlsson CM, et al. Cardiorespiratory 46. Aspenes ST, Nilsen TIL, Skaug E-­A, Bertheussen GF, Ellingsen Ø, fitness is associated with brain structure, cognition, and mood in a Vatten L, Wisløff U. Peak oxygen uptake and cardiovascular risk middle-­ aged cohort at risk for Alzheimer’s disease. Brain Imaging factors in 4631 healthy women and men. Med Sci Sports Exerc. Behav. 2015;9:639– ­649. DOI: 10.1007/s1168​2-­014-­9325-­9. 2011;43:1465–­1473. DOI: 10.1249/MSS.0b013​e3182​0ca81c. 67. Colcombe SJ, Erickson KI, Scalf PE, Kim JS, Prakash R, McAuley E, 47. Paterson DH, Cunningham DA, Koval JJ, St Croix CM. Aerobic fitness Elavsky S, Marquez DX, Hu L, Kramer AF. Aerobic exercise training in a population of independently living men and women aged 55–­86 increases brain volume in aging humans. J Gerontol A Biol Sci Med years. Med Sci Sports Exerc. 1999;31:1813–­1820. DOI: 10.1097/00005​ Sci. 2006;61:1166–­1170. DOI: 10.1093/geron​a /61.11.1166. 768-­19991​2000-­0 0018. 68. Erickson KI, Prakash RS, Voss MW, Chaddock L, Hu L, Morris KS, 48. Lolli L, Batterham AM, Weston KL, Atkinson G. Size exponents for White SM, Wójcicki TR, McAuley E, Kramer AF. Aerobic fitness is as- scaling maximal oxygen uptake in over 6500 humans: a systematic sociated with hippocampal volume in elderly humans. Hippocampus. review and meta-­ analysis. Sports Med. 2017;47:1405–­ 1419. DOI: 2009;19:1030–­1039. DOI: 10.1002/hipo.20547. 10.1007/s4027​9 -­016-­0 655-­1. 69. Erickson KI, Voss MW, Prakash RS, Basak C, Szabo A, Chaddock L, 49. Vanderburgh PM, Mahar MT, Chou CH. Allometric scaling of grip Kim JS, Heo S, Alves H, White SM, et al. Exercise training increases strength by body mass in college-­age men and women. Res Q Exerc size of hippocampus and improves memory. Proc Natl Acad Sci USA. Sport. 1995;66:80–­84. DOI: 10.1080/02701​367.1995.10607658. 2011;108:3017–­3022. DOI: 10.1073/pnas.10159​50108. J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.02068814 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure 70. Erickson KI, Leckie RL, Weinstein AM. Physical activity, fitness, and 88. Exalto LG, Biessels GJ, Karter AJ, Huang ES, Katon WJ, Minkoff JR gray matter volume. Neurobiol Aging. 2014;35(suppl 2):S20–­S28. DOI: & Whitmer RA. Risk score for prediction of 10 year dementia risk in 10.1016/j.neuro​biola​ging.2014.03.034. individuals with type 2 diabetes: a cohort study. Lancet Diabetes 71. Weinstein AM, Voss MW, Prakash RS, Chaddock L, Szabo A, White Endocrinol. 2015;1:183–­190. DOI: 10.1016/S2213​-­8587(13)70048​-­2. SM, Wojcicki TR, Mailey E, McAuley E, Kramer AF, et al. The associ- 89. Anstey KJ, Cherbuin N, Herath PM, Qiu C, Kuller LH, Lopez OL, Wilson ation between aerobic fitness and executive function is mediated by RS, Fratiglioni L. A self-­report risk index to predict occurrence of de- prefrontal cortex volume. Brain Behav Immun. 2012;26:811–­819. DOI: mentia in three independent cohorts of older adults: the ANU-­ADRI. 10.1016/j.bbi.2011.11.008. PLoS One. 2014;9:e86141. DOI: 10.1371/journ​al.pone.0086141. 72. Salat DH, Buckner RL, Snyder AZ, Greve DN, Desikan RSR, Busa E, 90. Gorelick PB, Scuteri A, Black SE, DeCarli C, Greenberg SM, Iadecola Morris JC, Dale AM, Fischl B. Thinning of the cerebral cortex in aging. C, Launer LJ, Laurent S, Lopez OL, Nyenhuis D, et al. Vascular contri- Cereb Cortex. 2004;14:721–­730. DOI: 10.1093/cerco​r/bhh032. butions to cognitive impairment and dementia: a statement for health- 73. Watson CE, Chatterjee A. A bilateral frontoparietal network underlies care professionals from the American Heart Association/American visuospatial analogical reasoning. NeuroImage. 2012;59:2831–­2838. Stroke Association. Stroke. 2011;42:2672–­ 2713. DOI: 10.1161/ DOI: 10.1016/j.neuro​image.2011.09.030. STR.0b013​e3182​299496. 74. Nissim NR, O’Shea AM, Bryant V, Porges EC, Cohen R, Woods AJ. 91. Baumgart M, Snyder HM, Carrillo MC, Fazio S, Kim H, Johns H. Frontal structural neural correlates of working memory performance Summary of the evidence on modifiable risk factors for cognitive in older adults. Front Aging Neurosci. 2017;08:328. DOI: 10.3389/ decline and dementia: a population-­based perspective. Alzheimers fnagi.2016.00328. Dement. 2015;11:718–­726. DOI: 10.1016/j.jalz.2015.05.016. 75. Erkinjuntti T, Inzitari D, Pantoni L, Wallin A, Scheltens P, Rockwood 92. Qiu C, Fratiglioni L. A major role for cardiovascular burden in age-­ K, Desmond DW. Limitations of clincal criteria for the diagnosis of related cognitive decline. Nat Rev Cardiol. 2015;12:267–­ 277. DOI: vascular dementia in clinical trials. Is a focus on subcortical vascu- 10.1038/nrcar​dio.2014.223. lar dementia a solution? Ann N Y Acad Sci. 2000;903:262–­272. DOI: 93. Tarraf W, Kaplan R, Daviglus M, Gallo LC, Schneiderman N, Penedo 10.1111/j.1749-­6632.2000.tb063​76.x. FJ, Perreira KM, Lamar M, Chai A, Vásquez PM, et al. Cardiovascular 76. Oberlin LE, Verstynen TD, Burzynska AZ, Voss MW, Prakash RS, risk and cognitive function in middle-­aged and older hispanics/latinos: Chaddock-­ Heyman L, Wong C, Fanning J, Awick E, Gothe N, results from the Hispanic Community Health Study/Study of Latinos et al. White matter microstructure mediates the relationship be- (HCHS/SOL). J Alzheimers Dis. 2020;73:103–­ 116. DOI: 10.3233/ tween cardiorespiratory fitness and spatial working memory in JAD-­190830. older adults. NeuroImage. 2016;131:91–­ 101. DOI: 10.1016/j.neuro​ 94. Joosten H, Van Eersel MEA, Gansevoort RT, Bilo HJG, Slaets JPJ, image.2015.09.053. Izaks GJ. Cardiovascular risk profile and cognitive function in young, 77. Bateman RJ, Xiong C, Benzinger TLS, Fagan AM, Goate A, Fox NC, middle-­aged, and elderly subjects. Stroke. 2013;44:1543–­1549. DOI: Marcus DS, Cairns NJ, Xie X, Blazey TM, et al. Clinical and biomarker 10.1161/STROK​E AHA.111.000496. changes in dominantly inherited Alzheimer’s disease. N Engl J Med. 95. Reis JP, Loria CM, Launer LJ, Sidney S, Liu K, Jacobs DR, Zhu N, 2012;367:795–­804. DOI: 10.1056/NEJMo​a1202753. Lloyd-­Jones DM, He K, Yaffe K. Cardiovascular health through 78. Beason-­Held LL, Goh JO, An Y, Kraut MA, O’Brien RJ, Ferrucci L, young adulthood and cognitive functioning in midlife. Ann Neurol. Resnick SM. Changes in brain function occur years before the onset 2013;73:170–­179. DOI: 10.1002/ana.23836. of cognitive impairment. J Neurosci. 2013;33:18008–­ 18014. DOI: 96. Kivipelto M, Ngandu T, Laatikainen T, Winblad B, Soininen H, 10.1523/JNEUR​OSCI.1402-­13.2013. Tuomilehto J. Risk score for the prediction of dementia risk in 20 years 79. Younes L, Albert M, Moghekar A, Soldan A, Pettigrew C, Miller MI. among middle aged people: a longitudinal, population-­based study. Identifying change points in biomarkers during the preclinical phase Lancet Neurol. 2006;5:735–­741. DOI: 10.1016/S1474​-­4422(06)70537​ of Alzheimer’s disease. Front Aging Neurosci. 2019;11:74. 10.3389/ -­3. fnagi.2019.00074. 97. Exalto LG, Quesenberry CP, Barnes D, Kivipelto M, Biessels GJ, 80. Gomes-­Osman J, Cabral DF, Morris TP, McInerney K, Cahalin LP, Whitmer RA. Midlife risk score for the prediction of dementia four Rundek T, Oliveira A, Pascual-­Leone A. Exercise for cognitive brain decades later. Alzheimers Dement. 2014;10:562–­570. DOI: 10.1016/j. health in aging. Neurol Clin Pract. 2018;8:257–­ 265. DOI: 10.1212/ jalz.2013.05.1772. CPJ.00000​0 0000​0 00460. 98. Pettigrew C, Soldan A, Zhu Y, Wang M-­C, Moghekar A, Brown T, Miller 81. Kramer AF, Colcombe S. Fitness effects on the cognitive function of M, Albert M. Cortical thickness in relation to clinical symptom onset older adults: a meta-­analytic study-­revisited. Perspect Psychol Sci. in preclinical AD. Neuroimage Clin. 2016;12:116–­122. DOI: 10.1016/j. 2018;13:213–­217. DOI: 10.1177/17456​91617​707316. nicl.2016.06.010. 82. Cheval B, Orsholits D, Sieber S, Courvoisier D, Cullati S, Boisgontier 99. Ferguson B. ACSM’s guidelines for exercise testing and prescription MP. Relationship between decline in cognitive resources and phys- 9th Ed. 2014. J Can Chiropr Assoc. 2014;58:328. ical activity. Health Psychol. 2020;39:519–­528. DOI: 10.1037/hea00 100. National Association of Area Agencies on Aging NC on A and UH The ​0 0857. 2015 United States of Aging Survey. 83. Lin X, Zhang X, Guo J, Roberts CK, McKenzie S, Wu W-­C, Liu S, Song 101. Al-­Mallah MH, Juraschek SP, Whelton S, Dardari ZA, Ehrman JK, Y. Effects of exercise training on cardiorespiratory fitness and biomark- Michos ED, Blumenthal RS, Nasir K, Qureshi WT, Brawner CA, et ers of cardiometabolic health: a systematic review and meta-­analysis al. Sex differences in cardiorespiratory fitness and all-­cause mortal- of randomized controlled trials. J Am Heart Assoc. 2015;4:e002014. ity: the Henry Ford ExercIse Testing (FIT) Project. Mayo Clin Proc. DOI: 10.1161/JAHA.115.002014. 2016;91:755–­762. DOI: 10.1016/j.mayocp.2016.04.002. 84. Sabia S, Dugravot A, Dartigues J-­F, Abell J, Elbaz A, Kivimäki M, Singh-­ 102. Wang C-­Y, Haskell WL, Farrell SW, Lamonte MJ, Blair SN, Curtin LR, Manoux A. Physical activity, cognitive decline, and risk of dementia: 28 Hughes JP, Burt VL. Cardiorespiratory fitness levels among US adults year follow-­up of Whitehall II cohort study. BMJ. 2017;357:j2709. DOI: 20–­49 years of age: findings from the 1999–­2004 National Health and 10.1136/bmj.j2709. Nutrition Examination Survey. Am J Epidemiol. 2010;171:426–­435. 85. Perissiou M, Borkoles E, Kobayashi K, Polman R. The effect of an DOI: 10.1093/aje/kwp412. 8 week prescribed exercise and low-­ carbohydrate diet on cardio- 103. Weltman A, Weltman JY, Hartman ML, Abbott RD, Rogol AD, Evans respiratory fitness, body composition and cardiometabolic risk fac- WS, Veldhuis JD. Relationship between age, percentage body fat, fit- tors in obese individuals: a randomised controlled trial. Nutrients. ness, and 24-­hour growth hormone release in healthy young adults: 2020;12:482. DOI: 10.3390/nu120​20482. effects of gender. J Clin Endocrinol Metab. 1994;78:543–­548. DOI: 86. Barnes DE, Covinsky KE, Whitmer RA, Kuller LH, Lopez OL, Yaffe K. 10.1210/jcem.78.3.8126124. Predicting risk of dementia in older adults: the late-­life dementia risk 104. Mosca L, Barrett-­ Connor E, Wenger NK. Sex/gender differences index. Neurology. 2009;73:173–­179. DOI: 10.1212/WNL.0b013​e3181​ in cardiovascular disease prevention what a difference a decade a81636. makes. Circulation. 2011;124:2145–­2154. DOI: 10.1161/CIRCU​L ATIO​ 87. Reitz C, Tang MX, Schupf N, Manly JJ, Mayeux R, Luchsinger JA. NAHA.110.968792. A summary risk score for the prediction of Alzheimer disease in el- 105. McCarrey AC, An Y, Kitner-­Triolo MH, Ferrucci L, Resnick SM. Sex derly persons. Arch Neurol. 2010;67:835–­ 841. DOI: 10.1001/archn​ differences in cognitive trajectories in clinically normal older adults. eurol.2010.136. Psychol Aging. 2016;31:166–­175. DOI: 10.1037/pag00​0 0070. J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.02068815 España-­Irla et al Proposal: Fitness, Cognition, and Brain Structure 106. Vanderburgh PM, Katch FI, Schoenleber J, Balabinis CP, Elliott R. in Southern Chinese 8 to 16 years old. PLoS One. 2019;14:e0213674. Multivariate allometric scaling of men’s world indoor rowing champi- DOI: 10.1371/journ​al.pone.0213674. onship performance. Med Sci Sports Exerc. 1996;28:626–­630. DOI: 108. Román Viñas B, Ribas Barba L, Ngo J, Serra ML. Validación en 10.1097/00005​768-­19960​5000-­0 0015. población catalana del cuestionario internacional de actividad 107. Yu CCW, McManus AM, Au CT, So HK, Chan A, Sung RYT, Li AM. física. Gac Sanit. 2013;27:254–­ 257. DOI: 10.1016/j.gaceta.2012. Appropriate scaling approach for evaluating peak VO2 development 05.013. J Am Heart Assoc. 2021;10:e020688. DOI: 10.1161/JAHA.120.02068816 Supplemental Material Data S1. CRF allometric scaling models. VO₂ peak was allometrically scaled using the procedure described by Vanderburgh et al49,108and seen in multiple CRF papers.48,109Firstly, VO₂ peak and body weight were log- transformed. A log-linear regression model was constructed using log (VO₂ peak) as the dependent and log (body weight) as independent variables. The interaction effect of biological sex was tested and found to significantly modulate the association between body mass and VO₂ peak, justifying the need for biological sex specific exponent. For that reason, regressions were performed separately for men and women to ensure the models were appropriate. Homoscedasticity was assessed by plotting the standardized residuals against the standardized predicted value. The resulting beta coefficients were used as the allometric exponents. Thus, VO₂ peak can then be allometrically scaled using the following equation: 109 unscaled peak VO₂ allometrically scaled peak VO₂ = body mass exponent In addition, Pearson correlation analysis was used to examine the association of the scaled VO₂ peak with non-scaled VO₂ peak to verify the effectiveness of the allometric scaling approach for controlling for body size within the sample. There was a very strong correlation between VO₂ peak and allometric VO₂ (r = .92, p < 0.001), suggesting that total body mass did not strongly affect our VO₂ peak measure and therefore, the results have remained practically stable. CRF models has been replicated using the new VO₂ scaled value and the results are seen on Table S1, S2 and S3. Data S2. Cardiovascular risk as measured by the Catalan-adjusted Framingham risk score (REGICOR). To ensure our results were valid when adjusted for the Catalan population, we repeated our analyses with the REGICOR risk score53. The REGICOR (Registre Gironí del Cor) function is an adaptation of the Framingham function to the incidence of ischemic heart disease and prevalence of local risk factors taking into account the different epidemiological characteristics of Spanish population. The Framingham-based REGICOR CV risk function provides a good prediction of the incidence of the coronary events of the general population of a region in the northwest of Spain and having a high long-term follow-up rate53. We found similar results both for the cognitive analyses and the cortical thickness analysis. Data S3. Individual results for the cortical thickness analyses with the VO₂ peak groups and mediation analyses. We run these models to illustrate the relationship between each significant cluster and VO₂ peak. Significant correlations were seen between VO2 peak and left rostral middle frontal gyrus (r mean=0.118). The older middle age group (55 and above) showed that left rostral middle frontal gyrus (r mean= 0.172) and left superior temporal gyrus (r mean= 0.169) were positively associated to VO₂ peak. The results also shown that cortical thickness significantly mediated the relationship between CRF 55 and above years old group and visuo-spatial problem solving, after controlling for age, biological sex, monthly incomes, education, waist perimeter and body mass index. Data S4. Individual plots and table for the cortical thickness analyses with the cardiovascular risk (Framingham) score and mediation analyses. We run these models to illustrate the relationship between each significant cluster and cardiovascular risk (Framingham score). Distributed clusters across multiple cortical regions were associated with cardiovascular risk (Framingham 5-year risk score). Those specific clusters were left post central (r mean= -0.170), left pars triangularis (r mean= - 0.170), left insula (r mean= -0.170), left cuneus gyrus (r mean= -0.172), left lingual (r mean= -0.164), left caudal anterior cingulate gyrus (r mean= -0.184), left superior parietal gyrus (r mean= -0.158), left inferior parietal gyrus (r mean= -0.160), left transverse temporal gyrus (r mean= -0.169). left rostral middle frontal (r mean=-0.161) and left precentral gyrus (r mean= -0.170). On the right hemisphere, the correlations were in right inferior parietal gyrus (r mean= -0.176), para hippocampal region (r mean= -0.184), right cuneus (r mean= -0.192), right supramarginal gyrus (r mean= -0.175), right precentral gyrus (r mean= -0.165), right lateral occipital gyrus (r mean= -0.160), and right superior frontal gyrus (r mean= -0.163). The results also shown that cortical thickness significantly mediated the relation between CVH and visuo-spatial problem solving, processing speed, flexibility, and memory, after controlling for education and monthly incomes. Data S5. Self-reported physical activity and its association with cardiorespiratory fitness. Self-reported physical activity was assessed using the International Physical Activity Questionnaire (IPAQ), validated for the Spanish/Catalan population 50,110. Data collected from the self-administered IPAQ surveys were summed within each physical activity domain (walking, moderate-intensity and vigorous-intensity activities) to estimate the total metabolic equivalent of task (MET) in minutes/week spent performing physical activity related to occupational, transportation, household, and leisure activities. The questionnaire was scored and analysed using established methods, available on the IPAQ website (www.ipaq.ki.se). Here, data collected with the IPAQ have been reported as a continuous measure. Total scores have been calculated for walking, moderate- intensity activities, and vigorous-intensity activities, for each domain (work, transport, domestic and garden, and leisure) and for overall total physical activity MET- minutes/week score, calculated as: Total physical activity MET-minutes/week = sum of Total (Walking + Moderate + Vigorous) MET-minutes/week scores. Engagement in physical activity as measured by the total number of METs-min/week including ‘walking’, ‘moderate activity’ and ‘vigorous activity’ explained 46% of the variance in VO₂ peak in our cohort (β = 3.61, SE = 0.71, p = <.001, R2=0.46). Table S1. Associations between CRF whole group allometric scaling values and cognitive domains. β SE P R2 Memory -0.007 0.003 0.023 0.115 Working memory -0.002 0.002 0.387 0.006 Flexibility -0.001 0.002 0.487 0.142 Processing speed -0.0006 0.003 0.839 0.163 Visuo-spatial problem solving 0.005 0.002 0.046 0.198 All CRF allometric scaled models are controlling for age, education and socioeconomic status as a covariate. R2 are adjusted for all predictors. Table S2. Associations between CRF_40_55 group allometric scaling values and cognitive domains. β SE P R2 Memory -0.007 0.003 0.025 0.115 Working memory -0.001 0.003 0.062 -0.002 Flexibility 0.0008 0.002 0.704 0.106 Processing speed -0.006 0.003 0.232 0.103 Visuo-spatial problem solving 0.001 0.003 0.731 0.083 All CRF allometric scaled models are controlling for age, education and socioeconomic status as a covariate. R2 are adjusted for all predictors. Table S3. Associations between CRF_55 and above group allometric scaling values and cognitive domains β SE P R2 Memory -0.007 0.003 0.022 0.115 Working memory -0.002 0.004 0.539 0.006 Flexibility -0.001 0.003 0.706 0.142 Processing speed 0.005 0.005 0.330 0.117 Visuo-spatial problem solving 0.011 0.004 0.007** 0.169 All CRF allometric scaled models are controlling for age, education and socioeconomic status as a covariate. R2 are adjusted for all predictors. Table S4. Associations between Regicor and cognitive domains. β SE P R2 Memory -2.587 0.392 <0.001 0.128 Working memory -0.558 0.368 0.130 0.004 Flexibility -1.105 0.269 <0.001 0.081 Processing speed -2.331 0.394 <0.001 0.091 Visuo-spatial problem solving -1.253 0.365 0.0006 0.079 All models are controlling for monthly incomes and education as covariates. R2 are adjusted for all predictors. *survives false discovery rate (FDR) corrections Table S5. Regicor Standardized beta coefficients. β P Memory -0.303 <0.001 Working memory -0.074 0.130 Flexibility -0.193 <0.001 Processing speed -0.274 <0.001 Visuo-spatial problem solving -0.160 0.0006 All models are controlling for age and education as covariates. R2 are adjusted for all predictors. *survives FDR corrections Table S6. Associations between Regicor and anatomical regions of cortical thickness. Cluster Hemisphere Anatomical ROI Size 1 Left Postcentral 2525.44 2 Left Insula 2343.16 3 Left Pars triangularis 1995.18 4 Left Superior frontal 1103.54 5 Left Cuneus 1051.47 6 Left Inferior parietal 617.95 7 Left Insula 458.83 8 Left Superior parietal 436.36 9 Left Middle temporal 358.97 1 Right Inferior parietal 9013.89 2 Right Precuneus 1799.52 3 Right Superior temporal 1708.11 4 Right Para hippocampal 1537.87 5 Right Lateral occipital 935.62 6 Right Superior frontal 491.33 7 Right Precentral 401.51 Table S7. Associations between VO₂ peak and anatomical regions of cortical thickness in the whole sample. Cluster Hemisphere Anatomical ROI Size 1 Left Rostral middle frontal 1465.59 Table S8. Associations between VO₂ peak and anatomical regions of cortical thickness in the 40-54 years old group. - No significant results. Table S9. Associations between VO₂ peak and anatomical regions of cortical thickness in the 55 and above years old group. Cluster Hemisphere Anatomical ROI Size 1 Left Rostral middle frontal 1634.39 2 Left Superior temporal 1168.92 Outcomes Total effect ADE ACME Visuo-spatial Beta (95%CI) Beta (95%CI) Beta (95%CI) problem solving Left rostral middle 0.07(0.02, 0.12)* 0.05(0.006, 0.11)* 0.01(0.0003, 0.03)* frontal gyrus Left superior 0.07(0.02, 0.12)* 0.06(0.02, 0.11)* 0.007(-0.005, 0.02) temporal gyrus Table S10. Each model was adjusted for age, biological sex, monthly incomes, education, waist perimeter and body mass index. ADE = average direct effect; ACME = average causal mediation effect. Statistical significance at p < 0.05 and 95% CI not including 0. Table S11. Associations between Framingham and anatomical regions of cortical thickness. Cluster Hemisphere Anatomical ROI Size 1 Left Post central 3518.9 2 Left Pars triangularis 1928.37 3 Left Insula 1428.72 4 Left Cuneus 1181.96 5 Left Lingual 845.33 6 Left Caudal anterior cingulate 833.6 7 Left Superior parietal 589.08 8 Left Inferior parietal 587.35 9 Left Transverse temporal 509.91 10 Left Rostral middle frontal 376.1 11 Left Precentral 357.23 1 Right Inferior parietal 9432.87 2 Right Parahippocampal 1962.3 3 Right Cuneus 1806.08 4 Right Supramarginal 1367.08 5 Right Precentral 568.03 6 Right Lateral occipital 462.1 7 Right Superior frontal 448.49 Outcomes Total effect ADE ACME Beta (95%CI) Beta (95%CI) Beta (95%CI) Visuospatial problem solving Left postcentral gyrus -0.05(-0.08, -0.02)* -0.03(-0.07, 0.00)* -0.01(-0.02, 0.00)* Left parstriangularis -0.05(-0.08, -0.02)* -0.03(-0.07, 0.00)* -0.01(-0.02, 0.00)* Left insula -0.05(-0.08, -0.02)* -0.03(-0.07, 0.00) -0.01(-0.02, -0.01)* Left cuneus -0.05(-0.08, -0.02)* -0.04(-0.07, -0.01)* -0.01(-0.02, 0.00)* Left lingual -0.05(-0.08, -0.02)* -0.04(-0.08, -0.02)* -0.002(-0.01, 0.01) Left caudal anterior cingulate gyrus -0.05(-0.08, -0.02)* -0.03(-0.07, 0.00)* -0.01(-0.02, -0.01)* Left superior parietal -0.05(-0.08, -0.01)* -0.04(-0.08, -0.01)* -0.006(-0.01, 0.00) Left inferior parietal -0.05( -0.08, -0.02)* -0.04(-0.08, -0.01)* -0.005(-0.01, 0.01) Left transverse temporal gyrus -0.05( -0.08, -0.02)* -0.04(-0.07, -0.01)* -0.009(-0.01, 0.00)* Left rostral middle frontal gyrus -0.05( -0.08, -0.02)* -0.04(-0.08, -0.01)* -0.004(-0.01, 0.00) Left precentral gyrus -0.05( -0.08, -0.02)* -0.04(-0.08, -0.01)* -0.002(-0.01, 0.01) Right inferior parietal gyrus -0.05( -0.08, -0.02)* -0.04(-0.08, -0.01)* -0.009(-0.02, 0.00)* Right parahippocampal region -0.05( -0.08, -0.02)* -0.05(-0.08, -0.02)* -0.0001(-0.011, 0.01) Right cuneus -0.05( -0.08, -0.02)* -0.04(-0.07, -0.01)* -0.01(-0.02, 0.00) Right supramarginal gyrus -0.05( -0.08, -0.02)* -0.03(-0.07, 0.00)* -0.01(-0.02, -0.01)* Right precentral gyrus -0.05( -0.08, -0.02)* -0.05(-0.08, -0.01)* -0.0002(-0.008, 0.01) Right lateral occipital gyrus -0.05( -0.08, -0.02)* -0.05( -0.08, -0.02)* 0.0008(-0.006, 0.01) Right superior frontal gyrus -0.05( -0.08, -0.02)* -0.04( -0.08, -0.02)* -0.001(-0.10, 0.01) Table S12A. Each model was adjusted for monthly incomes and education. ADE = average direct effect; ACME = average causal mediation effect. Statistical significance at p < 0.05 and 95% CI not including 0. Outcomes Total effect ADE ACME Beta (95%CI) Beta (95%CI) Beta (95%CI) Processing speed Left postcentral gyrus -0.11(-0.14, -0.08)* -0.10(-0.14, -0.08)* -0.005(-0.18, 0.01) Left parstriangularis -0.11(-0.14, -0.08)* -0.10(-0.14, -0.07)* -0.008(-0.02, 0.00) Left insula -0.11(-0.14, -0.08)* -0.10(-0.14, -0.08)* -0.004(-0.01, 0.01) Left cuneus -0.11(-0.14, -0.08)* -0.10(-0.14, -0.08)* -0.003(-0.01, 0.00) Left lingual -0.11(-0.14, -0.08)* -0.11(-0.14, -0.09)* 0.003(-0.005, 0.01) Left caudal anterior cingulate gyrus -0.11(-0.14, -0.08)* -0.10(-0.14, -0.08)* -0.005(-0.01, 0.00) Left superior parietal gyrus -0.11(-0.14, -0.08)* -0.10(-0.14, -0.08)* -0.004(-0.01, 0.00) Left inferior parietal gyrus -0.11(-0.14, -0.08)* -0.11(-0.14, -0.08)* -0.001(-0.01, 0.01) Left transverse temporal gyrus -0.11(-0.14, -0.08)* -0.10(-0.14, -0.07)* -0.006(-0.01, 0.00) Left rostral middle frontal gyrus -0.11(-0.14, -0.08)* -0.11(-0.14, -0.08)* -0.001(-0.009, 0.01) Left precentral gyrus -0.11(-0.14, -0.08)* -0.11(-0.15, -0.08)* 0.0009(-0.006, 0.01) Right inferior parietal gyrus -0.11(-0.14, -0.08)* -0.10(-0.13, -0.07)* -0.01(-0.02, 0.00) Right parahippocampal region -0.11(-0.14, -0.08)* -0.10(-0.14, -0.08)* -0.005(-0.01, 0.01) Right cuneus -0.11(-0.14, -0.08)* -0.10(-0.13, -0.07)* -0.01(-0.02, 0.00)* Right supramarginal gyrus -0.11(-0.14, -0.08)* -0.10(-0.14, -0.08)* -0.006(-0.01, 0.00) Right precentral gyrus -0.11(-0.14, -0.08)* -0.11(-0.14, -0.08)* -0.001(-0.009, 0.00) Right lateral occipital gyrus -0.11(-0.14, -0.08)* -0.11(-0.14, -0.08)* 0.002(-0.007, 0.01) Right superior frontal gyrus -0.11(-0.14, -0.08)* -0.11(-0.14, -0.08)* -0.001(-0.01, 0.01) Table S12B. Each model was adjusted for monthly incomes and education. ADE = average direct effect; ACME = average causal mediation effect. Statistical significance at p < 0.05 and 95% CI not including 0. Outcomes Total effect ADE ACME Beta (95%CI) Beta (95%CI) Beta (95%CI) Flexibility Left postcentral gyrus -0.05(-0.07, -0.03)* -0.03(-0.06, -0.01)* -0.01(-0.02, 0.00)* Left pars triangularis -0.05(-0.07, -0.03)* -0.04(-0.07, -0.02)* -0.007(-0.01, 0.00) Left insula -0.05(-0.07, -0.03)* -0.04(-0.06, -0.02)* -0.007(-0.01, 0.00)* Left cuneus -0.05(-0.07, -0.03)* -0.04(-0.07, -0.02)* -0.005(-0.01, 0.00) Left lingual -0.05(-0.07, -0.03)* -0.05(-0.07, -0.03)* -0.001(-0.008, 0.01) Left caudal anterior cingulate gyrus -0.05(-0.07, -0.03)* -0.04(-0.06, -0.02)* -0.009(-0.01, 0.00)* Left superior parietal gyrus -0.05(-0.07, -0.03)* -0.05(-0.07, -0.03)* -0.001(-0.009, 0.00) Left inferior aprietal gyrus -0.05(-0.07, -0.03)* -0.05(-0.07, -0.02)* -0.001(-0.009, 0.01) Left transverse temporal gyrus -0.05(-0.07, -0.03)* -0.04(-0.06, -0.02)* -0.008(-0.01, 0.00)* Left rostral middle frontal gyrus -0.05(-0.07, -0.03)* -0.04(-0.07, -0.02)* -0.003(-0.009, 0.00) Left precentral gyrus -0.05(-0.07, -0.03)* -0.04(-0.07, -0.02)* -0.002(-0.008, 0.00) Right inferior parietal gyrus -0.05(-0.07, -0.03)* -0.03(-0.06, -0.01)* -0.01(-0.02,0.00)* Right parahippocampal region -0.05(-0.07, -0.03)* -0.04(-0.06, -0.02)* -0.007(-0.01, 0.00) Right cuneus -0.05(-0.07, -0.03)* -0.04(-0.06, -0.02)* -0.009(-0.01, 0.00)* Right supramarginal gyrus -0.05(-0.07, -0.03)* -0.04(-0.06, -0.01)* -0.01(-0.02,0.00)* Right precentral -<0.001(-<0.001, - -<0.001(-<0.001, - -<0.001(-<0.001, - gyrus 0.03)* 0.03)* 0.00) Right lateral occipital gyrus -0.05(-0.07, -0.03)* -0.05(-0.07, -0.03)* -0.0009(-0.00, 0.00) Right superior frontal gyrus -0.05(-0.07, -0.03)* -0.03(-0.06, -0.01)* -0.01(-0.02, 0.00) Table S12C. Each model was adjusted for monthly incomes and education. ADE = average direct effect; ACME = average causal mediation effect. Statistical significance at p < 0.05 and 95% CI not including 0. Outcomes Total effect ADE ACME Beta (95%CI) Beta (95%CI) Beta (95%CI) Memory Left postcentral gyrus -0.11(-0.15, -0.08)* -0.10(-0.14, -0.06)* -0.01(-0.02, 0.00) Left pars triangularis -0.11(-0.15, -0.08)* -0.10(-0.13, -0.06)* -0.01(-0.02, -0.01)* Left insula -0.11(-0.15, -0.08)* -0.10(-0.14, -0.07)* -0.008(-0.01, 0.00) Left cuneus -0.11(-0.15, -0.08)* -0.10(-0.14, -0.07)* -0.009(-0.02, 0.00)* Left lingual -0.11(-0.15, -0.08)* -0.11(-0.14, -0.08)* -0.002(-0.01, 0.01) Left caudal anterior cingulate gyrus -0.11(-0.15, -0.08)* -0.11(-0.14, -0.07)* -0.006(-0.01, 0.00) Left superior parietal gyrus -0.11(-0.15, -0.08)* -0.12(-0.15, -0.08)* -0.004(-0.00, 0.01) Left inferior parietal gyrus -0.11(-0.15, -0.08)* -0.11(-0.15, -0.07)* -0.001(-0.01, 0.01) Left transverse temporal gyrus -0.11(-0.15, -0.08)* -0.11(-0.14, -0.07)* -0.004(-0.01, 0.00) Left rostral middle frontal gyrus -0.11(-0.15, -0.08)* -0.11(-0.14, -0.07)* -0.004(-0.01, 0.00) Left precentral gyrus -0.11(-0.15, -0.08)* -0.11(-0.15, -0.08)* -0.0007(-0.00, 0.01) Right inferior parietal gyrus -0.11(-0.15, -0.08)* -0.10(-0.14, -0.06)* -0.01(-0.02, 0.00) Right parahippocampal region -0.11(-0.15, -0.08)* -0.11(-0.14, -0.07)* -0.006(-0.01, 0.00) Right cuneus -0.11(-0.15, -0.08)* -0.10(-0.13, -0.06)* -0.01(-0.02, 0.00)* Right supramarginal gyrus -0.11(-0.15, -0.08)* -0.11(-0.14, -0.07)* -0.006(-0.01, 0.00) Right precentral gyrus -0.11(-0.15, -0.08)* -0.11(-0.15, -0.08)* -0.001(-0.01, 0.00) Right lateral occipital gyrus -0.11(-0.15, -0.08)* -0.11(-0.15, -0.08)* -0.0007(-0.00, 0.01) Right superior frontal gyrus -0.11(-0.15, -0.08)* -0.11(-0.15, -0.08)* -0.0007(-0.00, 0.01) Table S12D. Each model was adjusted for monthly incomes and education. ADE = average direct effect; ACME = average causal mediation effect. Statistical significance at p < 0.05 and 95% CI not including 0. Figure S1. REGICOR and cortical thickness. When using the Catalan-population adjusted Framingham risk score, we see similar patterns of associations with cortical thickness compared to those when using Framingham. Figure S2. CRF and cortical thickness. (A). Significant correlations were seen between VO2 peak and left rostral middle frontal gyrus (r mean=0.118). (B) The older middle age group (55 and above) showed that left rostral middle frontal gyrus (r mean= 0.172) and left superior temporal gyrus (r mean= 0.169) were positively associated to VO₂ peak. Figure S3. Framingham and cortical thickness. All plots illustrating the relationship between each significant cluster and cardiovascular risk (Framingham score). Distributed clusters across multiple cortical regions were associated with cardiovascular risk (Framingham 5-year risk score). Those specific clusters were left post central (r mean= -0.170), left pars triangularis (r mean= -0.170), left insula (r mean= -0.170), left cuneus gyrus (r mean= -0.172), left lingual (r mean= -0.164), left caudal anterior cingulate gyrus (r mean= -0.184), left superior parietal gyrus (r mean= -0.158), left inferior parietal gyrus (r mean= -0.160), left transverse temporal gyrus (r mean= -0.169). left rostral middle frontal (r mean=-0.161) and left precentral gyrus (r mean= -0.170). On the right hemisphere, the correlations were in right inferior parietal gyrus (r mean= -0.176), para hippocampal region (r mean= - 0.184), right cuneus (r mean= -0.192), right supramarginal gyrus (r mean= -0.175), right precentral gyrus (r mean= -0.165), right lateral occipital gyrus (r mean= -0.160), and right superior frontal gyrus (r mean= - 0.163). Figure S4. Physical activity and cardiorespiratory fitness. A significant positive association between physical activity levels (total weekly MET [metabolic equivalent of task]) and V0₂ peak, controlling for age, biological sex, education, monthly incomes, BMI (body mass index), and waist was found.