Background
With an advancing aging population and associated rise in dementia prevalence in developed countries, the associated costs and disease burden have exerted significant pressure on economic and social systems [
1]. To help ensure an aging society live enjoyable and productive lives, research into treating or preventing conditions such as Alzheimer’s disease (AD) and other forms of age-related neurodegenerative diseases is an urgent public health priority. Today, longevity-related prevalence of neurodegenerative diseases and especially dementia, along with the current absence of a cure are among the top prominent societal health-related challenges acknowledged by the first G8 Summit on dementia held in London on December 11th, 2013 [
2].
Published diagnostic criteria for AD in Lancet Neurology diagnose Mild Cognitive Impairment (MCI) as the preclinical or prodromal stage of AD [
3]. In MCI, several of the clinical and neuropsychologic pathologic features are present prior to the onset of overt AD [
4]. Patients with MCI in the earliest stage of neurodegeneration can be clinically diagnosed, and represent a patient cohort consistently able to participate in a structured exercise programme. Consequently, there has been an increased research focus on both pharmacological and non-pharmacological strategies to optimise cognitive function and enhance ‘brain health’ in older age [
5], particularly for individuals at risk of developing AD [
6].
Previous cross-sectional studies have established that moderate activity during midlife is associated with a lower risk of having MCI in later life, with late-life participation in moderate exercise also associated with lower risk for MCI [
7]. A meta-analysis of randomised controlled trials (RCTs) by Heyn et al. [
8] reported beneficial effects of physical activity on physical fitness and cognitive function in adults with cognitive impairment. In recent years, a broad range of exercise intervention studies have demonstrated cognitive benefits can be achieved with varying exercise modalities in populations with MCI [
9‐
11]. Aerobic exercise has demonstrated significant improvements in global cognitive scores with a weak but significant effect on memory [
12]. A meta-analysis by Gates et al. [
13] examining the effects of chronic exercise training on cognitive function in older adults with MCI found research quality was modest, with many studies under-powered and only 8% of cognitive outcomes demonstrating statistically significant change. A limitation across a number of these studies is the small sample size and variation in MCI diagnostic criteria applied. The differing exercise approaches used across exercise intervention studies, coupled with the wide variation in cognitive tasks utilised make it difficult to summarise and synthesise research findings [
14].
There is converging evidence from animal and human studies that regular aerobic exercise acts as a promoter of ’brain health‘ mediating neural homeostasis and, via neuroprotective and neurorestorative mechanisms, thereby counteracting brain ageing. At the behavioural level, exercise has been found to upregulate affective states [
15,
16] and to improve cognition throughout different age phases [
17‐
19] and different dimensions, including spatial/associative learning [
20,
21], attentional processing [
22], and executive control [
23]. While animal research has allowed the unravelling of the underlying neurobiological mechanisms of exercise at the behavioural (e.g. water maze-type tests), cellular (e.g. neurogenesis, synaptogenesis, neuroangiogenesis), and humoral (e.g. neurotrophic factors, inflammatory cytokines) levels [
24,
25], so far the neurobiological and epigenetic effects of exercise remains poorly understood in humans. While preliminary results indicate that aerobic exercise inhibits the progression of AD-like neuropathology in an animal model [
26,
27], currently there is little information about the effects of regular physical exercise on the progression of both functional and structural markers at the pre-dementia and early dementia stages in humans.
The aim of this study is to compare a 12-month structured exercise programme (aerobic and stretching and toning group) to a control group for progression of cognitive decline in MCI. The stretching and toning group will act as a non-aerobic exercise group, controlling for the social effect of a structured group exercise programme. This type of low-intensity exercise intervention has previously shown some positive effects on cognitive outcomes, as it provides participants with equivalent opportunity for social interaction [
28]. Previous exercise intervention studies in MCI that have utilised non-aerobic exercise (e.g. stretching and toning) as a control intervention have noted some improvement in cognitive performance [
8,
29] while others have not demonstrated significant change in cognitive performance [
9,
10]. The comparison between aerobic exercise, non-aerobic exercise and a control group coupled with the 12-month intervention period, concomitant brain scanning, genetic and epigenetic analyses is innovative and will be a strong addition to the growing body of literature around exercise in MCI.
Primary outcome
Cognitive function
The primary outcome is cognitive performance. Cognitive performance will be assessed by a neuropsychological test battery measuring six cognitive domains. The test battery will consist of a computer based CogState Battery including the International Shopping List Task (ISLT) – immediate and delayed recall, Detection Task, Identification Task, One Back Task and One Card Learning Task (
https://cogstate.com/) [
37,
38], Verbal fluency [
39,
40] and Trail Making Test (TMT) [
41].
The allocation of to the tests to the six cognitive domains is based on the CogState Guidelines and conventional classification of neuropsychological tests [
42].
Verbal memory will be assessed by ISLT.
Psychomotor function will be measured with the Detection Task.
Executive function will include TMT-B, Letter Fluency and Category Fluency.
Attention will be assessed by the Identification Task and TMT-A.
Wor
king memory will be measured by One Back Task, and
Visual memory by the One Card Learning Task. A description of the tasks is described below.
The CogState battery will take approximately 30 min to complete. The ISLT is a 12 word, four trial (three learning trials and one delayed recall trial). Total number of correct responses made in remembering the list on three consecutive trials at a single session and after a delay will be recorded. The ISLT has been shown to have good sensitivity to verbal memory impairment [
43]. The Detection Task measures psychomotor functioning and speed of processing. Participants must respond as quickly as possible when the card shown face down in the centre of the screen flips over by pressing a button on the keyboard. Reaction time is measured with lower scores indicating better performance. The Identification Task measures visual attention. Participants must decide whether a playing card presented on screen is red, by pressing the ‘Yes’ or ‘No’ button. Reaction time is measured and lower scores indicate better performance. The One Back Task assesses working memory. Participants are presented with a sequence of playing cards in the centre of a screen and must decide if the card presented is the same as the one shown immediately before. The One Card Learning Task measures visual learning and memory. Participants are presented with a succession of playing card on screen, and must decide if the card currently displayed has been displayed previously. Accuracy of performance is measured, with higher scores indicating better performance. A number of studies have found that the CogState battery of tests are sensitive to detecting cognitive impairment in mild to moderate AD and amnestic MCI populations relative to healthy matched controls [
44]. CogState and has been validated across a broad range of cognitively impaired populations [
38].
Verbal fluency will be assessed with Letter Fluency [
39] and Category Fluency [
40]. For the Letter Fluency test participants are allowed one minute to generate as many words as possible that begin with a specific letter. This task will be repeated three times with three different letters (e.g. F, A, S). For the category fluency test, participants must give as many examples of animals as possible within one minute.
TMT will be completed as a paper and pen based task. The TMT consists of two sub trials. TMT-A require individuals to sequentially connect 25 encircled numbers on a sheet of paper, while TMT-B require participants to draw a line, alternating between numbers and letters in ascending order.
Secondary outcomes
Secondary outcomes will include global cognitive function, cardiovascular fitness, physical activity, quality of life, depression, measures of frailty and epigenetic changes.
MoCA screening tool will be used as a broad measure of global cognitive function. The MoCA is a one-page 30-point test administered in 10 min which consists of 13 tasks covering the following eight cognitive domains: visuospatial/executive functions, naming, verbal memory registration and learning, attention, abstraction, delayed verbal memory, and orientation. It has demonstrated high sensitivity and specificity as a cognitive screening instrument and has been validated to detect MCI [
31].
Cardiovascular fitness will be assessed using an incremental exercise test on a standard cycle ergometer. Participants at the German Sports University and Trinity College Dublin will complete a maximal test in accordance with the World Health Organisation Protocol [
45]. The test will commence with 3 min cycling unloaded, followed by the incremental phase of exercise during which the load will increase by 25 W every two minutes until the test is terminated. Blood lactate levels and participants reported BORG RPE scores will be measured at each stage of the test (2 min intervals). At the University of Nijmegen, aerobic fitness will be estimated from a submaximal exercise test completed according to the Astrand-Rhyming submaximal protocol [
46]. During the first two minutes, resistance of the ergometer will be increased until a steady state HR of 70% of the estimated maximal HR is reached. Participants continue pedalling for 6 min. HR and RPE will be recorded every minute. VO
2max will be estimated using the average HR of minute 5 and 6 and the work load in the Astrand Nomogram.
Physical activity will be assessed objectively using an activity accelerometer to be worn for seven consecutive days by study participants as each assessment time point and subjectively with the LASA Physical Activity Questionnaire (LAPAQ). The LAPAQ questionnaire is a valid and reliable self-reported questionnaire that captures physical activity over the preceding 14 days [
47].
Health related quality of life will be evaluated using the Health Related Quality of Life for People with Dementia (DemQOL). DemQOL is a 28 item interview administered questionnaire relating to different aspect of QOL. The DemQOL has been validated in a large sample of people with dementia and demonstrates good acceptability and internal consistency [
48]. It has also been used in older adults and in patients with MCI [
49]. In addition, the Center for Epidemiologic Studies Depression (CES-D) questionnaire will be administered to determine depressive symptoms. The CES-D Scale is a short self-report questionnaire that measures symptomatic depression [
50] that has been validated as a depression screening tool in older adults [
51]. Depressive symptoms are associated with increased risk of MCI [
52]. The association of depression with prevalent MCI and with progression from MCI to dementia, but not with incident MCI, suggests that while depression is prevalent in MCI, it does not precede it [
53].
Measures of frailty will include The Timed Up and Go (TUG) test, hand grip strength and 30 s chair stand. The TUG will assess the participant’s mobility and balance. The TUG is a reliable and valid test for quantifying functional mobility and is useful in following clinical change in frailty over time [
54]. Hand grip strength will be measured using a Jamer Digital Dynamometer as a measure of upper limb strength. A standardised approach will be taken to obtaining the measurement [
55]. Hand‐Grip strength has been shown to predict future outcomes in aging adults including mortality and future levels of disability [
56]. The 30 s chair stand will determine lower limb strength and endurance. Lower body strength is considered critical in evaluating the functional performance of older adults [
57].
Venous blood samples will be collected for genetic and epigenetics analysis performed on a Sequenom Massarray Analyzer 4 at the Department of Psychology University of Bonn, Germany. Genotyping of the Apolipoprotein E (APOE) and the Brain Derived Neurotrophic Factor (BDNF) genes and their epigenetic methylation patterns will be the primary focus. The genetic analyses will serve as predisposed risk and resilience factors for cognitive functioning and decline in MCI participants. The epigenetic findings will shed new light on the link between exercise and gene activation of relevant genes in the biochemical pathways underlying cognitive decline. Given the moderating effect of the common genetic variation of APOE via personality on AD onset [
58], the NEO-Five Factor Inventory (NEO-FFI) will be included to assess personality traits such as neuroticism or extraversion, predicting an earlier onset of Alzheimer and cognitive decline in elderly humans [
59]. NEO-FFI is a psychological personality inventory consisting of 60 items to measure five personality traits. The questionnaire will be completed following each blood draw to form part of the epigenetics analysis. The NEO-FFI is also discussed as a measure of emotional intelligence [
60].
Structure and function of related brain regions will be measured using functional MRI (fMRI) brain which is a non-invasive method for examining brain activity and structure. Structural imaging will include isotropic T1-weighted-, T2-weighted- and FLAIR- sequences with an isotropic spatial resolution of 1x1x1mm. The combination of the different image weightings allows for an automatic detection/volumetry of white matter lesions. Additionally, the combination of these sequences improves image segmentation. The structural protocol will be rounded up by a 3D-DTI sequence (60 diffusion directions, 1.7x1.7x1.7 mm) and a resting state fMRI Table
1.
Table 1
SPIRIT diagram outlining schedule of enrolment, interventions, and assessments for study participants
| | <13 days following screening | 6 months after study visit 1 | 12 months after study visit 1 |
Enrolment | Explain study | X | | | |
Screen eligibility criteria | X | | | |
Study Outcomes | Neuropsychological testing | | X | X | X |
MoCA | | X | X | X |
DemQOL | | X | X | X |
CES-D | | X | X | X |
Incremental Exercise Test | | X | X | X |
Physical Activity Tracker | | X | X | X |
LAPAQ | | X | X | X |
Blood sampling | | X | | X |
NEO-FFI | | X | | X |
MRI Braina
| | X | | X |
TUG | | X | X | X |
Hand Grip Strength | | X | X | X |
30 second chair stand | | X | X | X |
Randomisation (after completion of all baseline assessments) | | X | | |
| Exercise interventionb
| | X | X | X |
Safety
All serious adverse events (SAE)/adverse events (AE) will be recorded on study specific adverse event forms. All AE’s will be registered with the local principle investigator (PI). These will be discussed at regular team meetings and collected and registered at the end of the study. All SAE’s will be registered centrally. In the case of an SAE, all site PI’s will be informed both at time of occurrence (with 24 h) and for a final conclusion on causality.
Sample size
The sample size estimation was performed in “G*Power”, a statistical software program. The effect size was estimated, based on the effect size found in several studies examining the effect of exercise on cognition in elderly with MCI or increased risk for AD [
8‐
10,
61,
62]. Sample size was calculated based on a two-tail statistical t-test set 1-β = 0.80, α = 0.05, an effect size of 0.4, an allocation ratio N2/N1 of 2. A total sample size of 224 was calculated. Considering a correlation of 0.5 between the outcome measures at T0 and T2, the design design-factor D = 1-0.52 = 0.75. The expected dropout rate is 25%. 224*0.75*1.25 = 210. Considering the fact that the primary analysis will be a combination of the results of three different centres, the final sample size
n = 210 is rounded up to
n = 225.
Statistical analysis
The primary analysis of this study will be comparison of cognitive functioning (primary outcome measure) between all three intervention arms before (T0) and after (T2) the 1 year intervention. A composite score will be calculated by averaging all six domain scores into one overall cognition score. The obtained scores per test will be converted into z-scores based on the standard deviation and mean of the total sample at baseline. In case of multiple tests within one domain, the average z-score for the domain will be calculated. Secondary outcome measures are the six separate cognitive domain scores and the other parameters.
For the primary analysis we will have an intervention group of n = 150 (both exercise forms together) and a control group of n = 75 as input of an ANCOVA with dependent variable the change in cognitive composite score between T2 and T0, and as covariates baseline cognitive functioning, sex, age. In a secondary analysis the comparison between the two exercise forms will be carried out in a similar ANCOVA analysis.
Analyses for all secondary outcome parameters will be carried out with similar ANCOVA analyses. Secondary analyses will also elucidate the contrasts between T0 and T1 and T2 for the primary and secondary outcomes, and thus elucidate the course over time, without correction for multiple comparisons, as these analyses are exploratory.
Furthermore, to assess change in physical fitness, quality of life, cerebral structure and epigenetics a similar statistical approach will be used as for the primary study parameter. A p-value of < 0.05 will be used to assess statistical significance.
Data management
All data will be managed using unique study codes to protect participant confidentiality, which will be used to code and file all electronic information. The key linking this code to participant identity will be stored in a secured file, access to this key is available only to designated members of the research team at each site. Raw data will be stored in a file cabinet with a lock where only designated research team members will have access to the key.
Acknowledgments
Members of the NeuroExercise Study Group (in alphabetical order by name). Dr. Justine A. Aaronson, Department of Medical Psychology, Radboud University Medical Center, Nijmegen, The Netherlands; Dr. Vera Abeln, Institute of Movement and Neurosciences, German Sport University Cologne, Germany; Dr. Jurgen A.H.R Claassen, Department of Geriatric Medicine, Radboud University Medical Center, Donders Institute for Brain, Cognition and Behaviour, Nijmegen, The Netherlands; Dr. Robert F. Coen, Mercer’s Institute for Research on Ageing, St. James’s Hospital, Dublin, Ireland; Dr. Emer M. Guinan, School of Medicine, Trinity College Dublin, Ireland; Dr. Damien Ferguson, Department of Neurology, St. James’s Hospital, Dublin, Ireland; Prof. Roy P.C. Kessels, Department of Medical Psychology, Radboud University Medical Center, Donders Institute for Brain, Cognition and Behaviour, Nijmegen, The Netherlands; Prof. Romain Meeusen, Human Physiology Research Group, Vrije Universiteit Brussels, Belgium; Prof. Christian Montag, Institute of Psychology and Education, Ulm University, Germany and Key Laboratory for NeuroInformation/Center for Information in Medicine, and School of Life Science and Technology, University of Electronic Science and Technology of China, Chengdu, China; Dr Ross T. Murphy, St. James’s Hospital, Dublin, Ireland; Prof. M. Cristina Polidori, Ageing Clinical Research, University Hospital of Cologne, Germany; Prof. Martin Reuter, University of Bonn, Department of Psychology and Center for Economics & Neuroscience, Laboratory of Neurogenetics, Bonn, Germany; Prof. Heiko K. Strüder, Institute of Movement and Neurosciences, German Sport University Cologne, Germany; Tim Stuckenschneider, Institute of Movement and Neurosciences, German Sport University Cologne, Germany; Prof H.J. Thijssen, Department of Physiology, Radboud University Medical Center, Nijmegen, The Netherlands; Prof. Tobias Vogt, Institute for Professional Sport Education and Sport Qualifications, German Sport University Cologne, Germany; Prof. Cathal Walsh, University of Limerick, Ireland; Prof. Bernd Weber, Department of Epileptology, University Hospital Bonn, Germany.
We would like to acknowledge the following network of hospital sites and investigators who have assisted in recruiting participants to this study at the Dublin site; Dr Jennifer Hoblyn, Bloomfield Healthcare; Dr Andrew Eustace, Highfield Healthcare; Dr Cora McGreevy, Mater Misericordiae University Hospital; Dr Aisling Denihan, Old Age Psychiatry Navan; Dr Justin Kinsella, St. Vincent’s University Hospital; Dr Declan Lyons, St. Patrick’s University Hospital and Dr Sean Kennelly, Tallaght Hospital.