Abstract
Background Despite mounting evidence that physical activity has positive benefits for brain and cognitive health, there has been little characterization of the relationship between cardiorespiratory (CR) fitness and cognition-associated brain activity as measured by functional magnetic resonance imaging (fMRI). The lack of evidence is particularly glaring for diseases such as Alzheimer disease (AD) that degrade cognitive and functional performance.
Objective The aim of this study was to describe the relationship between regional brain activity during cognitive tasks and CR fitness level in people with and without AD.
Design A case-control, single-observation study design was used.
Methods Thirty-four individuals (18 without dementia and 16 in the earliest stages of AD) completed maximal exercise testing and performed a Stroop task during fMRI.
Results Cardiorespiratory fitness was inversely associated with anterior cingulate activity in the participants without dementia (r=−.48, P=.05) and unassociated with activation in those with AD (P>.7). Weak associations of CR fitness and middle frontal cortex were noted.
Limitations The wide age range and the use of a single task in fMRI rather than multiple tasks challenging different cognitive capacities were limitations of the study.
Conclusions The results offer further support of the relationship between CR fitness and regional brain activity. However, this relationship may be attenuated by disease. Future work in this area may provide clinicians and researchers with interpretable and dependable regional fMRI biomarker signatures responsive to exercise intervention. It also may shed light on mechanisms by which exercise can support cognitive function.
Alzheimer disease (AD) is a pervasive disease typically associated with memory loss. However, executive dysfunction, including deficits in inhibition, attention allocation, and response planning, is common and may be one of the earliest manifestations of the disease.1–5 Recognition and remediation of executive impairment in people with AD is particularly important as it is these cognitive faculties that underpin functional independence.6–9 Evidence suggests that both functional and exercise training—the domain of physical therapists—can positively affect performance of activities of daily living in people with AD.10–15 As the percentage of the population over 65 years of age continues to grow, 1 in 8 Americans over age 65 now has AD.16 Physical therapists are uniquely qualified to prescribe individualized aerobic exercise programs for clinical populations and will be increasingly called upon to provide exercise prescription for these individuals,17 which may support brain health and functional independence.
The hypothesis that exercise and cardiorespiratory (CR) fitness are associated with healthy brain aging and reduced AD progression is biologically plausible and supported by observational and epidemiological studies. Increased physical activity has been postulated to have a trophic effect on the brain, particularly the hippocampus. For instance, exercise is associated with increased brain-derived neurotrophic factor18 and other important neurochemicals19 supporting brain growth and survival. Exercise appears to stimulate neurogenesis,20 as evidenced by increased counts of new neurons in adult animals on an exercise regimen. Exercise also is associated with enhanced neuronal survival21 and increased synaptic development and plasticity.22 Additionally, voluntary exercise may mediate the amyloid cascade by reducing the production of beta-amyloid and thus have a direct effect on AD pathophysiology.23
In older adults without dementia, there is accumulating evidence that physical activity, such as walking, benefits overall brain health and executive function in particular.24–27 Fitness has been associated with maintenance or increase in brain gray matter volume28–31 in normal aging. Furthermore, regular physical activity may reduce the risk of cognitive decline32,33 and delay dementia onset.34 In one study, researchers found greater activation in the right middle frontal cortex and better performance of individuals who were more highly fit when they engaged in a task assessing conflict resolution.25
In people who have already been diagnosed with dementia of various causes, meta-analysis of the small scale and methodologically diverse extant literature has shown that exercise may have small effects on physical function, cognition, and behavior.35 We have reported that in people with AD, higher CR fitness levels are associated with increased whole brain volume36 and hippocampal volume.37 We also found that lower CR fitness levels were associated with faster dementia progression over 2 years.31 However, a causal link between exercise and improved cognition, especially for the prevention or treatment of AD, is far from established.38 Thus, the literature warrants a broad investigation of the association between CR fitness and cognitive health and its potential as a remediating factor for those in the early stages of the disease.
Aerobic exercise may preferentially support executive function,39 and there is considerable interest in exercise as a remediating intervention for AD.11 Furthermore, functional magnetic resonance imaging (fMRI) may offer a sensitive means of capturing early exercise-related neuroplastic change in cognition.40,41 In prior cross-sectional analyses of individuals who were cognitively healthy, right frontal cortical activity was correlated with CR fitness during a task requiring resolution of conflicting stimuli.25 Higher CR fitness was associated with decreased activity in superior frontal, supramarginal, occipital, and superior temporal gyri42 and in the anterior cingulate cortex (ACC).25 Exercise intervention appears to enhance these relationships in a task-dependent manner.25,43 The definitive measure of the effects of exercise on cognition would be improved cognitive performance. However, frequently, there is a need in both research and clinical practice for additional biomarkers of intervention responsiveness. Neuroimaging may provide one measure of responsiveness to exercise intervention. Recently, Kramer and colleagues highlighted the need for neuroimaging studies44 of the role of fitness on the performance of everyday tasks.45
Few imaging studies of executive neural activity have been performed in people with AD. In the available literature, differences in activation patterns with executive processes have been noted in individuals with mild cognitive impairment (MCI) and AD compared with controls without dementia.46–48 For example, Rosano and colleagues46 described increased activity in the dorsolateral prefrontal cortex and the posterior parietal cortex of individuals with MCI during an executive task (prepotency of response). Similar regions were active in those with MCI when a Stroop task of attentional inhibition was performed.47 Another study48 showed increased activity in frontal regions, especially the medial frontal cortex, in participants with AD compared with those without dementia or MCI during a working memory task. Thus, the available evidence suggests that individuals with cognitive impairment activate additional frontoparietal regions during executive tasks in comparison with their peers without dementia. However, it remains unclear whether CR fitness is associated with brain activity during executive tasks in people with AD.
To our knowledge, there have been no reports of the relationship between CR fitness and executive brain activity in people with AD. Because the fMRI signal appears to be altered by AD, the association of CR fitness and fMRI signal must be quantified before an intervention effect can be reliably interpreted. Therefore, we had participants with and without AD complete maximal exercise testing and a Stroop task during fMRI. We hypothesized that CR fitness would be associated with brain activity during stimulus conflict resolution (ie, the Stroop interference task) in both those with AD and those without dementia specifically in the ACC and middle frontal (MidF) regions. We also assessed the superior parietal lobule as a secondary region of attentional control modulated by CR fitness. The importance of the knowledge gained in this study is thus twofold: (1) it provides additional evidence toward our understanding of the CR fitness and brain function relationship, and (2) it provides additional guidance toward the use of neuroimaging to evaluate therapeutic interventions.49
Method
Participants
Forty participants, both with and without AD (AD and ND groups, respectively), aged 60 to 85 years were recruited as a sample of convenience from a registry of individuals with an interest in participating in research at the University of Kansas Alzheimer's Disease Center. Participants for this case-control study were recruited from November 2009 through May 2011. Inclusion criteria were normal cognition or impaired cognition possibly or probably related to AD based on a clinical evaluation within the previous 6 months, independent ambulation without an assistive device, and an informant regularly in contact with the participant (>3 days/week) able to be present at the clinical evaluation. Exclusion criteria were neurological disorders other than AD that have the potential to impair cognition (eg, Parkinson disease, stroke), insulin-dependent diabetes mellitus, a recent history (<3 years) of cardiovascular or pulmonary disease, significant orthopedic issues that could limit performance on the maximal exercise test, clinically significant depressive symptoms (Geriatric Depression Scale score >4), abnormalities in B12 or thyroid function that may account for cognitive symptoms, use of psychoactive or investigational medications, any magnetic resonance imaging (MRI) exclusion, and significant visual or auditory impairment. Informed consent was obtained from all participants before enrollment in the study.
All participants first underwent a clinical evaluation as part of the Alzheimer's Disease Center Registry and additional measures such as the Mini-Mental State Examination (MMSE)50 and Geriatric Depression Scale.51 Participants were offered the opportunity to enroll at that time. All but 2 participants had undergone Clinical Dementia Rating (CDR) scale testing within 2 months of starting the exercise testing or MRI. These 2 participants were enrolled from inquiries they made to the center regarding ongoing studies and had undergone a clinical evaluation and CDR scale testing within the previous 6 months. Once enrolled, MRI and CR fitness assessment visits were scheduled within 2 weeks of each other without regard to order, based on laboratory and participant availability.
Clinical Evaluation
Dementia status and diagnosis of AD were based on clinical evaluation by trained clinicians.52 The assessment included a semistructured interview with the participant and the informant. Alzheimer disease diagnosis was determined by a single clinician based on established diagnostic criteria.53 Data on medications, past medical history, education, demographic information, and family history were collected. Dementia severity was determined using the CDR scale.54,55 A global CDR scale score is derived from individual ratings in each domain such that a CDR scale rating of 0 indicates no dementia, 0.5 indicates very mild, 1 indicates mild, 2 indicates moderate, and 3 indicates severe dementia. Individuals with AD met criteria for very mild or mild dementia (global CDR scale score 0.5 or 1.0). All participants were community dwelling.
CR Fitness Assessment
Prior to testing, the Physical Performance Test (PPT)56–58 was administered to index physical function and to allow study staff to subjectively assess potential ability during the maximal exercise test. Anderson et al59 and Billinger et al60 have previously shown that maximal exercise testing is reliable and feasible in people with AD. Although our laboratory has no hard rule for PPT scores that would raise concerns about exercise testing performance, the PPT items allow study staff to observe the participant's power, agility, gait, balance, and ability to follow commands. Study staff had no concerns about any participant's ability to satisfactorily complete testing.
Cardiorespiratory fitness was assessed as peak oxygen consumption (V̇o2peak; mL × kg−1 × min−1) during a symptom-limited, graded treadmill test using a modified Bruce protocol designed for older adults.61 Maximal exercise testing has been shown to be reliable in people with early-stage AD.59 Individuals were instructed to abstain from consuming food and caffeine 3 hours prior the scheduled test. Calibration procedures were performed on the metabolic cart before each test according to the manufacturer specifications. An exercise physiologist familiarized each participant with the exercise equipment and testing protocol and explained the Borg Rating of Perceived Exertion (RPE) scale. Participants began walking at a pace of 1.7 miles (1 mile=1.6 km) per hour at a 0% incline. At each 2-minute interval, the grade, speed, or both was increased. Participants were attached to a 12-lead electrocardiograph to continuously monitor heart rate and rhythm. A 2-way, nonrebreathing valve, headgear, mouthpiece, and noseclip were worn to collect expired air. Blood pressure and RPE scale score were acquired during the last 30 seconds of each stage. Expired gases were collected continuously, and oxygen uptake (V̇o2) and carbon dioxide production were averaged at 15-second intervals (ParvoMedics TrueOne 2400, ParvoMedics, Sandy, Utah). The exercise test was terminated if the participant reached volitional exhaustion or met absolute test termination criteria according to American College of Sports Medicine guidelines.62 Cardiorespiratory fitness was indexed as V̇o2peak, the highest V̇o2 value achieved during the test.
Stroop Task During MRI
A structural MRI scan was obtained using a Siemens 3.0 Tesla Allegra MRI scanner. High-resolution T1-weighted images were collected for anatomic localization and co-registration (magnetization prepared rapid acquisition gradient echo: 1.3- × 0.9- × 1.0-mm voxels, repetition time/echo time=2,300/3.05 milliseconds, 8° flip angle, field of view=240 mm, 208 slices). Functional imaging data were collected as axial echo-planar images using a single-shot, blipped gradient, echo-planar pulse sequence (3 × 3 × 3.5 mm, 0.5-mm gap, repetition time/echo time=2,000/30 milliseconds, 90° flip angle, field of vision=192 mm, 34 slices, 337 volumes). Foam pads were placed around the participant's head to reduce movement.
One run of the Stroop word/arrow task was performed following anatomical imaging. The Stroop word/arrow task63 assesses response inhibition and selective attention. Each stimulus consists of a word (“right” or “left”) and an arrow (← or →), stacked one above the other at the center of the screen. The stacking order was random. The 4 event conditions were generated by crossing stimulus attention (word versus arrow) and congruency (congruent versus incongruent) factors. The event conditions were equally represented in a rapid event-related paradigm, designed for maximal efficiency using event-related fMRI (Chris Rorden, http://www.mccauslandcenter.sc.edu/CRNL/tools/fmrisim). The task was synchronized to MRI acquisition. Stimuli were back-projected onto a translucent plate and viewed on an adjustable mirror mounted above the participant's head.
Participants were initially instructed to pay attention and respond to the direction indicated by the word using the left or right index finger on a button box (Current Designs Inc, Philadelphia, Pennsylvania). After each block of 10 trials (12 blocks total), an instruction was presented to attend to a new stimulus object (word or arrow). Congruency conditions and relevant stimulus location were randomized but balanced over trials. Trials were separated by a random, interstimulus interval of 5 to 7 seconds.
Functional images were preprocessed using Analysis of Functional NeuroImages software.64 The first 4 functional volumes of the time series were discarded to allow the magnetic field to stabilize. Data sets for each individual were first processed using the standard uber_subject.py pipeline settings with an 8-mm3 full-width half-maximum smoothing kernel. We modeled activation for each participant using 4 event condition onsets convolved with the gamma function, the 6 motion parameters, and third-order polynomials, censoring repetition times with >1 mm of motion. Resulting contrasts of incongruent events > congruent events (0.5 0.5 −0.5 −0.5) then were passed to the group analysis.
Structural images were aligned to a standard template in Talairach space as part of the uber_subject.py pipeline. Additionally, structural images were divided by tissue type and region using Freesurfer.65,66 The standard Freesurfer processing pipeline, recon_all, separates the anatomical image into individualized regions based on gyral and sulcal structure and tissue type.65,66
Data Analysis
Published estimates of sample size for fMRI experiments with a conservative random effect model and moderate effect sizes indicate that 80% power can be achieved using a threshold of 0.002 with approximately 20 participants per group.67 Thirty-four individuals were included in the final analysis. Four individuals with AD were excluded with accuracy on the Stroop task below 66%. One individual with AD experienced a fall in the home prior to exercise testing and withdrew. One individual without AD declined to participate in the MRI after consenting.
For our primary hypothesis, we found the median percent signal change in our a priori hypothesized regions of interest (ROIs) constructed using the Freesurfer segments.65,66 We extracted the median percent signal change values from the following regions of the Destrieux Atlas: left and right middle frontal gyri, superior parietal lobule, anterior cingulate gyri and sulci, and precentral gyri as a control region. We then calculated the difference in median percent signal change (dPSC) between conditions (incongruent − congruent).
We assessed group differences in demographics and task performance with appropriate tests (analysis of variance, Pearson chi-square, Mann-Whitney). To test our primary hypothesis that CR fitness would be associated with greater ROI activation, we correlated dPSC with CR fitness (V̇o2peak; mL × kg−1 × min−1), correcting for age using partial correlation. We also tested dPSC against the response time cost of incongruency. Tolerance of type I error was set at α=.05. Statistical analyses were performed using SPSS version 20 (IBM Corp, Armonk, New York).
We then assessed group differences in imaging space (ie, AFNI program 3dttest++), including age and V̇o2 peak as covariates. First, we assessed the main effect of incongruency (incongruent > congruent) pooling all participants. We then compared group activation patterns. Finally, we specifically assessed the effect of incongruency with regard to CR fitness within each group separately. Because AD-associated atrophy can result in spurious findings, we inclusively masked all imaging analyses using a binary mask of voxels representing gray matter in at least 50% of the AD group. A voxel-wise threshold of P<.001 uncorrected and cluster size (k) of 10 voxels or greater were set for these analyses.68
Data Management
Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Kansas.69 Imaging data were archived on the Extensible Neuroimaging Archive Toolkit (XNAT version 1.5).70
Role of the Funding Source
Portions of this work were supported by the following grants from the National Institutes of Health: R01AG033673, R01AG034614, KL2TR000119, and UL1TR00001. Dr Vidoni was supported by a New Investigator Fellowship Training Initiative Award from the Foundation for Physical Therapy during this study. Dr Vidoni and Dr Burns are supported by the University of Kansas Alzheimer's Disease Center (P30AG035982). Dr Honea is supported by National Institutes of Health grant K01AG035042.
Results
All participants were community dwelling and independently ambulating without an assistive device. The groups (ND and AD) did not differ in age, V̇o2 peak, sex, or performance on the Stroop task. The groups were different in MMSE, PPT, and global CDR scale score, as expected. We also saw an increased response time for incongruent stimuli across all participants. Table 1 includes demographic and performance data for the groups.
Demographic and Performance Measures for Participants Without Dementia (ND Group) and Those With Alzheimer Disease (AD Group)a
We saw no difference between left and right dPSC in ACC and MidF and, therefore, averaged dPSC sides for the analyses. Our primary hypothesis was that higher CR fitness would be associated with greater activity in the ACC and MidF across all individuals. However, when we correlated V̇o2peak with ACC dPSC corrected for age, we found group differences in the relationship of fitness level and activity (Fig. 1). In the AD group, V̇o2peak was not associated with dPSC in any ROI (P>.9). The ND group showed a significant inverse relationship in the ACC, with higher CR fitness level associated with a lower dPSC in the ACC (r=−.48, P=.05; Fig. 1). No other ROIs were significantly associated with V̇o2peak in the ND group.
Alzheimer disease modulates the relationship of cardiorespiratory (CR) fitness and bilateral anterior cingulate (ACC) difference in percent signal change (dPSC). Participants without dementia (ND group) (open circles, dashed line) with greater CR fitness (x-axis, toward the right) showed less difference in activity (r=−.48, P=.05). Individuals with AD (filled circles, solid line) showed no such relationship (P=.9). V̇o2peak=peak oxygen uptake.
We also compared dPSC with the response time cost of stimulus incongruency. In the AD group, response time cost again was not associated with dPSC in any ROI (P>.3). However, the ND group displayed a positive relationship in the MidF (r=.57, P=.02) and superior parietal lobule (r=.48, P=.05), with greater response time cost associated with increased dPSC. No other regions approached significance in the ND group (P>.13).
We next performed our whole brain imaging analysis. First, we assessed the main effect of incongruency > congruency after pooling all participants. Bilateral MidF and bilateral inferior frontal gyri showed increased activation, even after controlling for age and V̇o2peak, demonstrating that our paradigm elicited activation comparable to the findings of previous studies (Tab. 2A, Fig. 2A). No differences were detected between groups, although there were trends (P<.005 uncorrected, k≥10) for greater activity by the ND group in the left precentral gyrus (Brodmann area 4 and 6) and right precuneus (Brodmann area 7) (Tab. 2B). No region showed a significant CR fitness × group interaction.
Regions of Increased Activity in the Incongruent Versus Congruent Stimulus Conditions for Participants Without Dementia (ND Group) and Those With Alzheimer Disease (AD Group)a
(A) Across participants, stimulus incongruency increased activity in bilateral middle frontal cortex (P<.001, k>10). (B) We found a trend (P<.005, k=7) for increased activity to be associated with greater peak oxygen uptake in the left middle frontal gyrus only in the participants without dementia. The color bar represents difference in percent signal change. All displayed voxels are members of significant clusters overlaid on an average anatomical image of all participants with Alzheimer disease.
To explore CR fitness-related brain activity beyond our a priori–defined ROIs, we regressed CR fitness against whole brain contrast maps in each group separately. In the AD group, CR fitness was unrelated to regional activity, although there was a trend for decreased activity to be associated with higher CR fitness in the right inferior temporal gyrus (P<.005, k=5; Tab. 2C). Likewise, in the ND group, no regions were significantly associated with CR fitness. However, there was a trend in the ND group for those participants with higher CR fitness to demonstrate increased activity in left MidF gyrus (Brodmann area 9, P<.005 uncorrected, k=7; Tab. 2D; Fig. 2B).
Discussion
The results of this study reflect a complex relationship among CR fitness, cognitive performance, and AD. To our knowledge, only one other study has compared CR fitness and functional imaging markers of cognition in people with cognitive impairment,71 and none have specifically assessed CR fitness or executive function. In our study, we focused on ROIs that are commonly activated during Stroop tasks.72,73 We found that CR fitness was associated with increased activation during Stroop interference, echoing prior findings of a fitness and brain activity association in people without dementia.25 However, this relationship was not present in individuals with AD. This disease-related mitigation of the CR fitness/fMRI signal suggests that fMRI could be used to monitor exercise intervention response. That is, if exercise interventions in people with AD resulted in a “normalization” of the CR fitness and brain activity association similar to older adults without dementia, it would suggest a positive response to the intervention, especially if coupled to behavioral improvement.
CR Fitness and Brain Activity During the Stroop Task
Both increased and decreased regional activation have now been reported to be associated with CR fitness.25,42 The conflicting reports may reflect differences in functional task, age of the participants, or analysis methods. For example, both Colcombe et al25 and Voelcker-Rehage et al42 reported right MidF activation to be positively associated with CR fitness during a task requiring resolution of conflicting stimuli. However, Voelcker-Rehage et al also reported that individuals with increased CR fitness had decreased activation in superior frontal and MidF cortex.42 Colcombe et al25 also reported an inverse relationship in the ACC, whereas Voelcker-Rehage et al42 found no relationship. Our results, albeit some at a trend level, are consistent with the findings of Colcombe et al25: increased MidF and superior parietal lobule and decreased ACC activity are associated with CR fitness during an executive task.
Executive function encompasses a broad set of cognitive functions that few measures can fully quantify. We chose to use a well-known task paradigm (Stroop interference) that specifically tests the executive functions of selective attention and conflict resolution. These executive functions are highly applicable to functional independence. For example, a person must rely heavily on selective attention and conflict resolution when evaluating the many directional arrows on roadway signs when driving on the highway. Although our results cannot speak to a CR fitness relationship to other executive functions, such as working memory or scheduling, or to ecologically valid activities such as driving, they do support further investigation of the executive faculties and activities.
Does AD Modulate the CR Fitness and Brain Activity Relationship?
Cardiorespiratory fitness and brain activity associations evident in those participants without dementia were not evident in the AD group. We noted that bilateral MidF activation was increased with greater task difficulty across participants. However, we saw no group-based differences, and the increased activity was unrelated to CR fitness in the AD group. Notably absent was an association between CR fitness and decreased ACC activity. Based on these results, we suggest that AD alters or overrides the relationship of CR fitness with brain function. We readily acknowledge that absence of evidence is not evidence of absence of a relationship. However, we were particularly careful to account for various confounding factors that might influence this association discussed in our “Limitations” section. Thus, our results likely reflect a diminished fitness effect associated with AD.
These findings do not preclude the possibility that exercise training could entrain a more typical relationship and positively alter brain activity, especially in the ACC. Indeed, there is preliminary evidence that exercise may support functional plastic change in the earliest stages of AD.71 In that randomized controlled trial, resistance training in individuals with mild cognitive impairment improved performance on both the Stroop task and an associative memory task. Increased activity in lingual and temporal regions was positively correlated with improved memory task performance in those individuals who resistance trained.
Strengths and Limitations
The fMRI signal can change with age and disease. With this factor in mind, we were careful to test a priori hypotheses and relied on several strategies recommended to account for age-related signal change.74 We focused our imaging analyses on a comparison of 2 conditions, rather than task performance to a no-task baseline, and used percent signal change, which is relative to the individual. Thus, we analyzed an interaction of condition and diagnosis rather than a main effect of diagnosis—an important distinction that resolves some concerns regarding disease-related hemodynamic differences. Because age and CR fitness are correlated, our assessment using age, a covariate of no interest, is conservative. Prior studies have not controlled for age, which may explain some differences in our study. Tighter age groups would alleviate this problem but decrease generalizability to the wide age range of AD onset. A second limitation of our study is that we cannot rule out that our group without AD was not in the preclinical stages of AD through biomarker assessment. However, we are confident that our cohort without dementia represents a group without cognitive impairment based on a thorough clinical interview that includes corroboration with someone who knows the participant well. Finally, we stress that this is a cross-sectional assessment and thus cannot inform our understanding of any causal relationship between CR fitness and brain function. As we have noted, future randomized controlled studies can build upon these observational findings and investigate causal or intervention effects of exercise.
Conclusions
Taken together, the results of this study support previous findings that suggest that in individuals without dementia, greater CR fitness is associated with increased middle frontal and decreased anterior cingulate activity during tasks that tax attention and conflict resolution, 2 executive functions. We know of no prior studies that have examined the relationship between CR fitness and executive function using fMRI in early-stage AD. This gap in the literature remains despite evidence that CR fitness supports executive cognitive performance in nondemented aging.39 The present study provides further evidence that these regions are appropriate candidates for monitoring functional neuroplastic response to exercise in both people without dementia and those in the earliest stages of AD. If CR fitness is ultimately shown to support executive function in people with AD, it may have significant implications as an adjunct therapy for maintaining functional independence. Physical therapists could incorporate both functional and aerobic training in their plans of care for those in the early stages of AD to maximize performance of instrumental activities of daily living and independence. Future work in this area should focus on providing clinicians and researchers with interpretable and dependable regional fMRI biomarker signatures responsive to exercise intervention. Additionally, follow-up studies should emphasize exploration of mechanisms by which exercise can support cognitive function.
Footnotes
Dr Vidoni, Dr Savage, and Dr Burns provided concept/idea/research design and project management. Dr Vidoni, Mr. Gayed, Dr Savage, and Mr Hobbs provided writing. Dr Vidoni provided data collection. Dr Vidoni, Mr. Gayed, Dr Honea, Dr Savage, and Mr Hobbs provided data analysis. Dr Vidoni and Dr Burns provided fund procurement, study participants, facilities/equipment, and institutional liaisons. Dr Savage and Dr Burns provided consultation (including review of manuscript before submission).
The study was approved by the Human Subjects Protection Committee of the University of Kansas Medical Center.
Portions of this work were supported by the following grants from the National Institutes of Health: R01AG033673, R01AG034614, KL2TR000119, and UL1TR00001.
Dr Vidoni was supported by a New Investigator Fellowship Training Initiative Award from the Foundation for Physical Therapy during this study. Dr Vidoni and Dr Burns are supported by the University of Kansas Alzheimer's Disease Center (P30AG035982). Dr Honea is supported by National Institutes of Health grant K01AG035042.
- Received November 19, 2012.
- Accepted April 2, 2013.
- © 2013 American Physical Therapy Association
References
Issue highlights
