Resting-state Functional Connectivity of the Motor and Cognitive Areas is Preserved in Masters Athletes

—Aging is characterized by a decline in physical and cognitive functions, often resulting in decreased quality of life. Physical activity has been suggested to potentially slow down various aspects of the aging process, a theory that has been supported by studies of Masters Athletes (MA). For example, MA usually have better cognitive and physical functions than age-matched sedentary and healthy older adults (OA), making them a valuable model to gain insights into mechanisms that promote physical and cognitive function with aging. The purpose of this study was to identify diﬀerences in resting-state functional connectivity (rs-FC) of motor and cognitive regions between MA and OA and determine if these diﬀerences in the resting brain are associated with diﬀerences in cognitive and physical performance between groups. Fifteen MA (9 males) and 12 age-matched OA (six males) were included. rs-FC images were compared to identify signiﬁcant between-groups diﬀerences in brain connectivity. There was higher connectivity between the cognitive and motor networks for the OA group, whereas the MA group had stronger connectivity between diﬀerent regions within the same network, both for the cognitive and the motor networks. These results are in line with the literature suggesting that aging reduces the segregation between functional networks and causes regions within the same network to be less strongly connected. High-level physical activity practiced by the MA most likely contributes to attenuating aging-related changes in brain functional connectivity, preserving clearer boundaries between diﬀerent functional networks, which may ultimately favor maintenance of eﬃcient cognitive and sensorimotor processing. (cid:1) 2024 The Authors. Published by Elsevier Inc. on behalf of IBRO. This is an open access article under the CC BY license (http://creativecommons. org/licenses/by/4.0/).


INTRODUCTION
Aging is a normal process associated with a decline in cognitive and physical functions, resulting in a decreased quality of life (Asakawa et al., 2000).Working memory, memory retrieval, visuospatial attention and reasoning are just a few examples of the cognitive abilities that have been shown to deteriorate with normal aging (Harada et al., 2013).Impairments in gait and posture, coordination, and slowing of movement all characterize the decline in physical function commonly experienced with normal aging (Laughton et al., 2003;Seidler et al., 2010).Cognitive decline has been repeatedly associated with hypometabolism of the anterior cingulate cortex (ACC) (Ishibashi et al., 2018;Pardo et al., 2020), a key cortical brain region for cognition (Braver et al., 2001;Wager and Smith, 2003).Similarly, atrophy of the primary motor cortex (M1), the supplementary motor area (SMA) and the premotor cortex (PMC) (Salat et al., 2004;Seidler et al., 2010) have also been observed with aging.Furthermore, aging has been shown to cause a loss of segregation of the different functional networks, meaning that the networks are less clustered and less distinguishable from the others (Damoiseaux, 2017).Indeed, in normal aging the connectivity observed within functional networks, including the cognitive and sensorimotor networks, is decreased (Wu et al., 2007a;Onoda et al., 2012;Madden et al., 2017); however, connectivity between regions of different functional networks is increased (Damoiseaux, 2017).All these differences may explain the motor and cognitive deficits experienced with aging.
We know that aging interacts with physical activity by slowing the onset of detrimental changes caused by aging (Galloza et al., 2017).For example, physical activity reduces cognitive decline (Landi et al., 2010;Klimova et al., 2017) and reduces the risk for Alzheimer's disease (Buchman et al., 2012;Klimova et al., 2017).Similarly, physical activity improves balance (Z ˇivkovic´et al., 2018), boosts motor learning (Hu¨bner and Voelcker-Rehage, 2017), prevents frailty, maintains mobility and reduces falls (Landi et al., 2010;Duray and Genc¸, 2017) with normal aging.These benefits on cognitive and physical function occur through the protective effect of physical activity on brain function (Colcombe et al., 2006;Rolland et al., 2010) due to elevated concentrations of growth factors (Cassilhas et al., 2007) and neurotrophins (Marosi and Mattson, 2015), and improved mitochondrial function (Kou et al., 2017) and cerebral hemodynamics (Bailey et al., 2013).Multiple studies have demonstrated that physical activity leads to increased gray and white matter volumes with aging in several brain regions including the ACC, M1 and SMA (Colcombe et al., 2006;Rehfeld et al., 2018).Furthermore, the ACC is less activated in physically active older individuals when performing cognitive tasks, suggesting a more efficient use of this cognitive brain region (Voelcker-Rehage et al., 2010).Physical activity also results in increased connectivity between regions within the cognitive network, including the ACC (Voss et al., 2010;Raichlen et al., 2016).Regular physical activity, therefore, offers major benefits to aging by helping to preserve cognitive and physical function through the years.
The understanding of the beneficial effects of physical activity on aging has been strengthened by the study of masters athletes (MA).MA are by definition a population that maintains a high level of physical activity with aging and who continue to compete in sport (Geard et al., 2017).With better self-reported physical, psychological, cognitive, and social functioning, MA have drawn attention as an example of ''successful aging" (Geard et al., 2017).Compared to sedentary or less active older individuals, MA have better working memory (Schott and Krull, 2019), executive memory (Zhao et al., 2016), inhibitory control (Schott and Krull, 2019), dual-tasking capacity (Dupuy et al., 2018), reaction times (Zhao et al., 2016;Schott and Krull, 2019), maximal walking speed (Glenn et al., 2015), and postural control (Brauer et al., 2008).Even though the physiology of MA has been extensively studied (for review: Tanaka and Seals, 2008;Valenzuela et al., 2020;Burtscher et al., 2022), the possible protective effects of physical activity on brain function in MA has never been studied.Two studies focusing on brain structure determined that although loss of brain tissue is not completely spared in MA (Tseng et al., 2013b), they maintain higher gray matter volume (Tseng et al., 2013b), better white matter microstructure in several brain regions (Tseng et al., 2013a), and less white matter hyperintensities (Tseng et al., 2013a) than sedentary individuals.Differences in brain function between MA and sedentary or less active older individuals have yet to be studied.The purpose of this study was therefore to determine whether there are differences in functional brain networks in MA compared to healthy older adults and to investigate whether such differences might be associated with superior cognitive and motor abilities usually observed in MA.
One novel way to study differences in brain networks between groups is through resting-state functional connectivity (rs-FC).rs-FC is a functional magnetic resonance imaging (fMRI) method that quantifies the strength of connectivity between brain regions at rest (i.e., in the absence of an experimental task).Unlike task-based fMRI, rs-FC can be used to study multiple neural systems simultaneously, making it particularly suitable for studying differences in cognitive and motor brain regions connectivity in MAs compared to healthy aged-matched older adults.Keeping in mind that physical activity generally minimizes the effects of aging on the brain (Colcombe et al., 2006;Rolland et al., 2010), and because it is well established that aging itself causes a decrease in rs-FC within functional networks and an increase in rs-FC between regions of different functional networks (i.e., loss of segregation) (Damoiseaux, 2017), we hypothesized that compared to age-matched controls, MAs would have: [1] less rs-FC between motor and cognitive regions, [2] increased rs-FC within the ACC and other cognitive regions, and [3] increased rs-FC within motor regions.Furthermore, it was hypothesized that [4] the differences in rs-FC would be positively correlated with cognitive and cardiovascular outcomes.

EXPERIMENTAL PROCEDURES Participants
Fifteen MAs and 14 age-matched sedentary, healthy older adults (OA) were recruited as part of a larger project investigating the physiological differences between MA and healthy OA participants (Taran, 2015;Power et al., 2016bPower et al., , 2016a;;Ubaida-Mohien et al., 2022).Two male participants in the OA group were excluded due to abnormal findings on the MRI (i.e., abnormal ventricle size (Birn, 2023;Reynolds et al., 2023)).Therefore, 15 MAs (eight females, mean age 80 [SD 5] years) and 12 OAs (six females, mean 81 [SD 4] years) were included in the analysis.All participants were screened for any existing neurological, musculoskeletal, cardiovascular, and pulmonary conditions based on a medical history form.The MA group consisted of track and field athletes who were ranked in the top five worldwide for their age and event (eight of whom were current world record holders for their event at the time of study), and thus, represent the very highest functioning individuals for their age.As described in Supplementary Table 1, MA were competing in different track and field events including jumping, throwing, and running for an average of 23 (SD 9) years, and most of them were active through-out their life.Their average current training volume was of 16 (SD 5) hours per week and included mainly running, strength training, hiking, and stretching.The participants in the OA group were living independently, with physical activity levels ranging from sedentary to recreationally active according to the Community Healthy Activities Model Program for Seniors (CHAMPS) Physical Activity Questionnaire (Stewart et al., 2001) and to their daily step count (mean: 2731 [SD 984]) acquired using an Actigraph GT3M accelerometer during a typical week (Tudor-Locke et al., 2011).
All participants gave their written informed consent before participating in the study, in accordance with the Declaration of Helsinki and the McGill Faculty of Medicine Institutional Review Board regulations for human subjects' studies.

Cardiovascular and cognitive assessments
Peak aerobic capacity (VO 2 max in mL/kg/min) was assessed using a continuous incremental cycling test to volitional exhaustion on an electronically braked cycle ergometer (Ergoline 800s), as reported previously for this same cohort of individuals (Ubaida-Mohien et al., 2022).Testing followed standard procedures (American Thoracic Society and American College of Chest Physicians, 2003) consisting of a 3-minute rest period and a 3-minute unloaded cycling warm-up, followed by a 20-Watt increase every two minutes until exhaustion.Peak work rate was also obtained from the incremental cycling test as the last and highest power production reached by the participant before exhaustion.
General cognitive functions were assessed using the Mini Mental State Exam (MMSE) (Folstein et al., 1975).A truncated version that did not include the first two questions of the test was used, resulting in a maximal score of 20 instead of the typical 30-point scale.The two questions were removed because they tested the ability to identify current location and time, of which the participants of the study were all assumed to be aware as they were able to meet the research team at the research site at a given date and time.This assumption was also supported by the results of a study conducted in 3,062 participants of all ages showing that response to these two orientation questions is not significantly impacted by age, whereas the other items (i.e., memory, language and visual construction) are (Escobar et al., 1986).The mean scores obtained on the MMSE for MA and OA in this study (19/20 [SD 2], 15/20 [SD 3], respectively) would therefore be 29/30 for the MA group and 25/30 for the OA group.

Image acquisition
The MRI scans were obtained on a Siemens 3T Trio Scanner (Siemens, Knoxville, TN) at the Montreal Neurological Institute (MNI) in Montreal, Canada.The protocol for each participant included T1-weighted anatomical images (acquisition time = 9:14, 192 slices, voxel size = 1 mm 3 isotropic, echo spacing = 7.1 ms, flip angle = 9 deg) and a blood-oxygen-level-dependent (BOLD) MOSAIC resting scan (acquisition time = 10:01, 328 volumes, voxel size = 4 mm 3 isotropic, echo spacing = 0.49 ms, 34 slices, flip angle = 90 deg).During the resting-state scan, participants were asked to lie still, stay awake, not think about anything, and keep their eyes open fixating on a cross positioned in front of them.

Image preprocessing
Data were analyzed with a resting-state pipeline developed by the Center for Research on Brain,

Seeds
A seed-to-voxel FC analysis was used to identity clusters of voxels temporally correlated with the mean time series of each region of interest (ROI), namely the anterior cingulate cortex and motor regions.The seeds were created for each ROI, for a total of eleven right hemisphere and eleven left hemisphere seeds.The anterior cingulate consisted of six bilateral seeds: Brodmann area (BA) 9, BA 10, BA 24 caudodorsal, BA 24 rostroventral, BA 32 pregenual, and BA 32 subgenual.The remaining five ROIs were the pre-SMA, SMA, PMC, M1 and the primary somatosensory cortex (S1).All seeds were in MNI standard space and linearly transformed to native space using FSL's Linear Image Registration Tool (FLIRT).Details about the seeds are presented in Table 1.

Functional connectivity analysis
To produce individual FC maps, a resting-state FC regression analysis was performed in native space.For each seed, the mean time series was calculated by averaging the BOLD signal from all voxels within the seed region.The following nuisance predictors were included in the regression analysis: cerebrospinal fluid, white matter, global signal, motion outlier volume masks, and six motion parameters (x, y and z translations and rotations).The resulting rs-FC individual maps were then linearly transformed to MNI standard space using FLIRT to prepare for group analysis.The FC analysis is further detailed in (Potvin-Desrochers et al., 2019).

Statistical analysis
Between-group analyses were performed using a mixedeffect model and Bayesian modeling scheme in FLAME, FSL (Woolrich et al., 2004).Correction for multiple comparisons was carried out using a Gaussian random field theory, with a cluster threshold of Z > 2.3, and a cluster significance of p < 0.05.Resulting clusters were identified as specific brain regions using the Anatomy toolbox in SPM12.
Relationships between FC and cardiovascular or cognitive outcomes were investigated with correlations using SPSS v25 (IBM, NY, USA).The Shapiro-Wilk normality test was used to assess the normality assumption of our dataset.Relative VO 2 max, height and body mass were normally distributed and group means were compared using an independent t-test.Peak work rate, MMSE and age were not normally distributed thus group means were compared using a Mann-Whitney U-test.Mean resting-state FC from each resulting cluster was correlated with relative VO 2 max using Pearson's r coefficient and with peak work rate and MMSE scores using Spearman coefficient.The level of significance was set at p < 0.05.

RESULTS
The participant demographics can be found in Table 2. Briefly, mean VO 2 max, peak work rate, and MMSE were all significantly higher in MA than in OA (p 0.002).
As shown in Table 3 and Fig. 1(A-B), MA had less rs-FC between cognitive and motor areas compared to OA.Specifically, connectivity was lower between the left BA24 and M1-PMC region and in several subregions of the ACC and with motor regions (i.e.SMA, PMC and M1) in MA when compared to OA.In addition, the right SMA was also significantly less connected to frontal cognitive areas in MA compared with OA.The connectivity between all these regions were negatively correlated to the fitness level (i.e.VO 2 max) (Fig. 1(C)) and some of them with cognitive function and peak work rate (Table 3).
MA had higher connectivity compared to OA within subregions of the ACC and other cognitive regions (Table 3, Fig. 2).Connectivity between the right caudal BA24 and a cluster comprising the orbitofrontal cortex and the ventral striatum was significantly higher in MA than in OA and positively correlated with levels of fitness (Fig. 2(C)) and cognition.Furthermore, MA demonstrated a marked negative connectivity (i.e.anticorrelation and anti-coupling) between the left BA32 and a cluster located in the parietal lobe that can be attributed to the posterior parietal cortex (PPC, Fig. 2 (B)).This connectivity was not observed in OA and was correlated with fitness level, as quantified by VO 2 max.
Finally, the left SMA was significantly more connected with a cluster comprising the PPC and occipital regions in MA, whereas they were anti-correlated in OA (Fig. 3(B)).Fitness and cognition levels were positively correlated with this rs-FC.

DISCUSSION
This was the first study to assess brain rs-FC in MA.In line with our hypotheses, at rest, MA had lower functional connectivity between cognitive and motor brain regions, whereas functional connectivity within cognitive regions and within motor regions was increased when compared to OA.Whether this is linked to the high level of performance of MA, to their lifelong 3 physical activity practice, or simply by their overall better cardiovascular fitness as quantified by VO 2 max remains to be determined.These results support the hypothesis of high-level physical activity attenuating aging-related rs-FC patterns.rs-FC between two subregions of the ACC and several motor regions (i.e., M1, SMA, PMC) was lower in MA compared to OA, as well as between the SMA and the dorsolateral prefrontal cortex (Fig. 1).These results corroborate our hypothesis, suggesting that the increased rs-FC observed between different functional networks in aging does not occur in MA.Indeed, rs-FC between cognitive and motor regions was positive in our control group, but negative and weak in MA (Fig. 1(B)).Negative connectivity occurs when the spontaneous activity of one region increases, and the spontaneous activity of the other ROI synchronously decreases, which is also known as anti-coupling or anti-correlation (Fox et al., 2005).Cognitive and motor regions are therefore anti-coupled at rest in MA, possibly meaning that MA are spared from the loss of segregation between cognitive and motor networks that normally occurs with aging (Damoiseaux, 2017).In healthy middle aged adults, the ACC is positively-coupled with the precentral gyrus and SMA (Yu et al., 2011).However, this previous study did not assess levels of physical activity.Thus, the negative rs-FC we observe in the current study may be due to the high-levels of physical activity in the MA group.This idea is supported by the multiple negative correlations we obtained between fitness levels and mean rs-FC (Fig. 1(C), Table 3).These differences in rs-FC between cognitive and motor regions may also explain ageassociated alterations in the control of movement.It is known that actions become less automatic and more cognitively controlled with normal aging (Heuninckx et al., 2005).Thus, we propose that lower rs-FC between cognitive and motor brain regions may be associated with a reduced need for cognitive control of motor actions in MA, which could in part, explain their better dual-tasking capacity (Dupuy et al., 2018), reaction times (Zhao et al., 2016;Schott and Krull, 2019), maximal walking speed (Glenn et al., 2015), and postural control (Brauer et al., 2008) compared to sedentary OA.
It is worth mentioning that one of the ACC subregions (i.e. the right BA24) was less strongly connected to a cluster comprising the PMC, but also the dorsolateral prefrontal cortex in MA compared to OA.This contradicts our second hypothesis that rs-FC is higher within cognitive regions in MA compared to OA.However, the other significant differences in connectivity within cognitive regions that we observed in the current study does support our hypothesis.With aging, it is known that the ACC loses functional connections (Wu et al., 2007a(Wu et al., , 2007b) ) which has been correlated with a loss in cognitive capacity (Onoda et al., 2012).However, adherence to a 1-year walking program reverses this loss in connectivity, and increases the functional connectivity between several cognitive brain regions in normal aging (Voss et al., 2010).In line with this study, our results demonstrate that higher rs-FC between cognitive regions is associated with higher fitness level and cognitive capacity.Specifically, in MA, compared to OA, rs-FC was higher between the right BA24 and a cluster comprising the orbitofrontal cortex and the ventral striatum (Fig. 2  (A)).These regions are key contributors to the reward system, with the orbitofrontal cortex being involved in decision making (Rudebeck and Rich, 2018) and the ventral striatum in the evaluation of the reward magnitude (Haber and Knutson, 2010).Although it has been found in healthy adults that the ACC is positively connected to these two regions (Yu et al., 2011), our results demonstrate that these regions were negatively connected in OA, but positively connected in MA (Fig. 2(B)).Thus, we propose that this difference between OA and MA may be due to the high level of physical activity performed by MA, delaying the effects of aging.
Similarly, MA showed a negative correlation in connectivity between the left BA32 and the PPC, whereas OA lacked connectivity between these regions (Fig. 2(B)).This finding further suggests that physical activity preserved the brain of MA, as these regions are normally anti-coupled in healthy young adults (Margulies et al., 2007).Moreover, the posterior cingulate cortex and the superior parietal area are even more negatively connected in highly trained individuals (Raichlen et al., 2016).Thus, the anti-coupling observed in MA should not be interpreted as a reduced rs-FC, but instead as a better coordination between the spontaneous activity of the ACC and the PPC.The latter is a brain area involved in working memory and attentional control (Wager and Smith, 2003), and both are key regions of the frontoparietal executive control network, which is essential for cognitive, attentional, and perceptual processing (Fox et al., 2006).Thus, we propose that these results support the second hypothesis of increased connectivity within cognitive regions and that this enhanced connectivity could be associated with the better cognitive abilities of MA (Zhao et al., 2016;Dupuy et al., 2018;Schott and Krull, 2019).High fitness levels could also possibly attenuate the decline in cognitive processing.
We also observed an increased rs-FC between the SMA and the PPC (Fig. 3), which positively correlated to fitness level and cognitive capacity (Table 3), and thus, supported our third hypothesis that motor regions would have higher rs-FC in MA compared to OA.Although not a component of the sensorimotor network, the PPC is functionally and structurally connected to multiple sensorimotor regions (Narayana et al., 2012;Vergani et al., 2014) with an unquestionable involvement in sensorimotor processing.In addition to its participation in executive functions, the PPC contributes to the planning of movement through its central role in visuospatial processing (Culham et al., 2006), sensory integration (Andersen, 1997), and generation of body schema (Parkinson et al., 2010).With aging, functional connectivity within the sensorimotor network decreases (Onoda et al., 2012;Madden et al., 2017) and the SMA exhibits less functional connections (Wu et al., 2007a(Wu et al., , 2007b)).In this study, we showed that in MA, the SMA is positively connected to the PPC, whereas they were anti-coupled in OA (Fig. 3(B)).This anti-correlation is in line with the results of previous studies demonstrating that in frail older individuals the SMA and the PPC are anti-coupled (Lammers et al., 2020), that this anti-correlation is reduced in non-frail older individuals (Lammers et al., 2020), and that young healthy adults exhibit a reverse pattern of positive rs-FC between superior parietal lobules and the SMA (Li et al., 2014).Thus, a high-level of fitness in later age may help prevent the shift from positive to negative rs-FC between the SMA and the PPC and contribute to a better integration of sensory input to generate motor programs and perform actions.This may have a substantial implication in motor deficits experienced in older age, as the integrity of sensorimotor function plays a central role in gait, balance impairments, and fall risk (Taylor et al., 2012).The increased rs-FC between the SMA and the PPC in MA may be an indication of the important contribution of physical activity in the maintenance of the sensorimotor network efficacy, ultimately explaining and supporting the role of physical activity in the prevention of gait and balance troubles (Z ˇivkovicé t al., 2018) and risk of falls (Duray and Genc¸, 2017).Despite the novelty of its results, this study has some limitations.To further determine the effect of high-level physical activity on brain aging, future studies should consider involving a larger sample size and including and comparing MA according to their specialization in different sports, as the type of physical activity practiced (e.g.strength vs. cardiovascular training, individual vs. team sports, etc.) could impact the aging process differently.Providing a detailed history of physical activity for each participant would also be important to determine if alterations in brain networks could be attributed to sports training and its physical or cognitive components, or simply to the physical training required for competition.Performing an extensive cognitive assessment of the participants could also help determine which components of cognition are more strongly altered by physical activity.Interventional studies could potentially investigate the impact of initiating intense physical activity in older individuals on alterations in rs-FC.This would provide information on the possibility of restoring brain connectivity using a training program.Finally, we cannot rule out contributions resulting from unique biology of MA that may not be consequent to their physical activity habits.For example, we have previously identified more than 300 differentially represented proteins in skeletal muscle of these same MA participants that were distinct from those identified as responsive to exercise (Ubaida-Mohien et al., 2022).These findings implicate nonexercise factors (e.g., gene x environment) as contributing to the high function of our MA participants and presumably a similar situation could also apply to differences in brain connectivity.The aforementioned physical activity intervention study could help address this issue.This is the first study examining the rs-FC of MA.Our results suggest a better preservation of functional connectivity of motor and cognitive regions in MA in comparison with OA.Specifically, reduced connectivity between motor and cognitive regions, and increased connectivity within motor and within cognitive regions were observed, all of which correlated with fitness level.Altogether, these results are consistent with the notion that maintaining a high-level of physical activity in older age better maintains brain functional connectivity to favor more efficient cognitive and sensorimotor processing.

Fig. 1 .
Fig. 1. rs-FC results between cognitive and motor brain regions, in line with hypothesis 1. A. Statistical maps (Z-score) for each seed showing significant between-group differences for hypothesis 1. rs-FC of the seeds is greater for OA versus MAs (OA > MA).Mask of the seeds are presented in white and axial slices show largest mask area.p < 0.05 corrected.B. Median rs-FC strength (Z-score) between the seeds showing significant between-group differences and their cluster for each group in line with hypothesis 1 (p < 0.05).Positive values represent a positive connectivity, while negative values represent a negative connectivity (i.e.anti-coupling).Boxes are the 25th to 75th percentiles and whiskers 5th to 95th percentiles.Black circles are individual datapoints.C. Significant correlations between the mean rs-FC of resulting clusters and clinical measures for hypothesis 4 (p < 0.05).R = right, L = left, OA = older adults, MA = master athletes, rs-FC = resting-state functional connectivity, SMA = supplementary motor area, VO2max = maximal oxygen consumption.

Fig. 2 .
Fig. 2. rs-FC results within cognitive brain regions, in line with hypothesis 2. A. Statistical maps (Z-score) for each seed showing significant between-group differences for hypothesis 2. OA > MA indicates greater rs-FC of the seeds for OA versus MAs, and MA > OA indicates greater rs-FC for MAs versus OA.Mask of the seeds are presented in white and axial slices show largest mask area.p < 0.05 corrected.B. Median rs-FC strength (Z-score) between the seeds showing significant between-group differences and their cluster for each group in line with hypothesis 2 (p < 0.05).Positive values represent a positive connectivity, while negative values represent a negative connectivity (i.e.anti-coupling).Boxes are the 25th to 75th percentiles and whiskers 5th to 95th percentiles.Black circles are individual datapoints.C. Significant correlations between the mean rs-FC of resulting clusters and clinical measures for hypothesis 4 (p < 0.05).R = right, L = left, OA = older adults, MA = master athletes, rs-FC = resting-state functional connectivity, vST = ventral striatum, PPC = posterior parietal cortex, VO 2max = maximal oxygen consumption.

Fig. 3 .
Fig. 3. rs-FC results within sensorimotor brain regions, in line with hypothesis 3. A. Statistical maps (Z-score) fort the seed showing a significant between-group differences for hypothesis 3. rs-FC of the seeds is greater for MAs versus OA (MA > OA).Mask of the seed is presented in white and axial slices show largest mask area.p < 0.05 corrected.B. Median rs-FC strength (Z-score) between the seed showing significant between-group difference and its cluster for each group in line with hypothesis 3 (p < 0.05).Positive values represent a positive connectivity, while negative values represent a negative connectivity (i.e.anti-coupling).Boxes are the 25th to 75th percentiles and whiskers 5th to 95th percentiles.Black circles are individual datapoints.C. Significant correlations between the mean rs-FC of the resulting cluster and clinical measure for hypothesis 4 (p < 0.05).R = right, L = left, OA = older adults, MA = master athletes, rs-FC = resting-state functional connectivity, SMA = supplementary motor area, PPC = posterior parietal cortex, VO 2 max = maximal oxygen consumption.

Table 1 .
Description of the seeds used in the rs-FC analysis

Table 2 .
Description of participant characteristics MA = Masters Athletes; OA = Older Adults; MMSE = Mini-Mental State Examination Data presented as absolute value (sample size and sex); missing data due to technical issue with the equipment on testing day, sample size for these outcomes: 1 n = 13 and 2 n = 11; *Significant group differences (p < 0.05).Language, and Music (www.crblm.ca)runningonFSL 5.0.8 (FMRIB Software Library, Oxford, UK), and MATLAB 2018b software (MathWorks Inc., Natick, MA, USA).The image preprocessing procedure is detailed in our previous study(Potvin-Desrochers et al., 2019).Briefly, preprocessing included signal stabilization, calculation of the linear registration transformations, slice-timing correction, brain extraction, motion correction, global intensity normalization, spatial smoothing, bandpass filtering, and removal of motion outlier volumes.All participants had minimal head motion during MRI and there was no significant difference between the absolute mean displacement of the two groups (MA mean 0.20 [SD 0.10] mm, OA mean 0.23 [SD 0.08] mm, p = 0.51).

Table 3 .
Significant differences in rs-FC between OA and MA and their correlation with cardiovascular and cognitive outcomes.