Contemporary Clinical Trials 104 (2021) 106331

Contents lists available at ScienceDirect

Contemporary Clinical Trials

journal homepage: www.elsevier.com/locate/conclintrial

Exercise training and cognition in multiple sclerosis: The GET Smart

trial protocol
Charles H. Bombardier a, *, Robert W. Motl b, Ralph H.B. Benedict c, Nancy Temkin d,
Peiqing Qian e, Katharine Alexander a, Annabeth Evans a, Andrea Thomas a, Kristin Toms a,
Cathea M. Carey a, George H. Kraft a, f
a
Department of Rehabilitation Medicine, School of Medicine, University of Washington, Box 359612, Harborview Medical Center, 325 9th Ave, Seattle, WA 98104, USA
b
Department of Physical Therapy, University of Alabama, SHPB 336, 1716 9th Avenue South, Birmingham, AL 35233, USA
c
Departments of Neurology and Psychiatry, University at Buffalo, UBMD Physicians Group, 1001 Main Street, 4th Floor, Buffalo, NY 14203, USA
d
Departments of Neurological Surgery, School of Medicine and Biostatistics, School of Public Health, University of Washington, Box 359925, Harborview Medical Center,
325 9th Ave, Seattle, WA 98104, USA
e
Multiple Sclerosis Center, Swedish Neuroscience Institute, 1600 E Jefferson St, Seattle, WA 98122, USA
f
Departments of Rehabilitation Medicine and Neurology, University of Washington, Box 356490, UW Medical Center-Montlake, 1959 NE Pacific St., Seattle, WA 98195,
USA

A R T I C L E I N F O A B S T R A C T

Keywords: Background and objectives: Multiple sclerosis (MS) causes cognitive impairment in approximately 50% of cases.
Multiple sclerosis Disease modifying medications and cognitive rehabilitation produce only small positive effects on cognition in
Exercise MS. Converging animal and human research suggests that aerobic exercise may improve cognition in people with
Cognition
MS, but definitive trials are lacking. We describe the design of the GET Smart study, a randomized controlled trial
Aerobic
Research design
comparing the effects of aerobic exercise versus stretching and toning on cognition in MS.
Methods: The study is a single-blind, parallel group randomized (1:1) controlled trial that compares aerobic
exercise training with an active control group consisting of stretching and toning exercises for improving
cognition. Participants are nondepressed, ambulatory, non-exercising adults with MS aged 18–54 years who have
below average cognitive processing speed. Both treatments were designed to generate equivalent outcome ex­
pectancies and entailed supervised, progressive exercise programs, 3 times per week for up to 40 min over a 6
month period.
Projected patient outcomes: The primary hypothesis is that the aerobic training group will demonstrate signifi­
cantly greater cognitive processing speed compared with the control group at the end of the treatment phase (6
months) as measured by a composite of the Paced Auditory Serial Additon Test and the oral Symbol-Digit Mo­
dalities Test using intent-to treat analyses. Secondary outcomes are neuropsychological functioning and
cardiorespiratory fitness as well as participant reported outcomes such as depression, sleep, and fatigue. Study
findings will inform future research, patient education, clinical care and policymaking.
Trial Registration: ClinicalTrials.gov Identifier NCT02106052

1. Introduction episodic memory, and to a lesser extent executive function [3]. Cogni­
tive impairment is associated with poorer vocational and social func­
Multiple sclerosis (MS) is a chronic inflammatory and neurodegen­ tioning and greater psychiatric comorbidity in persons with MS. [4]
erative disease [1] with a prevalence exceeding 2.5 million people When the present study was designed, most disease modifying
world-wide and 1 million in the United States [2]. MS is a leading cause therapies had no proven effect on cognition in MS [5–8] and cognitive
of weakness, disability, fatigue, depression and cognitive impairment. rehabilitation was in its infancy [9]. We therefore examined converging
The prevalence of cognitive impairment in MS is approximately 50% animal and human research supporting physical activity, especially
and is characterized by impaired cognitive processing speed, and aerobic exercise training, as influencing chemical, cellular and

* Corresponding author at: Box 359612, Harborview Medical Center, 325 9th Ave, Seattle, WA 98104, USA.
E-mail address: [email protected] (C.H. Bombardier).

https://doi.org/10.1016/j.cct.2021.106331
Received 11 December 2020; Received in revised form 7 February 2021; Accepted 23 February 2021
Available online 27 February 2021
1551-7144/© 2021 Elsevier Inc. All rights reserved.
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

structural changes in the brain culminating in improved cognitive 2.1.2. Aim 2

functioning [10]. Aerobic exercise training increased brain-derived To explore whether improvement in cognitive functioning is posi­
neurotrophic factor (BDNF) and enhanced neurogenesis, cell survival, tively associated with improvement in cardiorespiratory fitness (VO2-
synaptic proliferation and angiogenesis in the brain [10]. Clinical trials peak).
indicated that aerobic exercise training improved cognitive functioning
in people with normal cognition as well as older adults with mild 2.1.3. Aim 3
cognitive impairment or dementia [11–15]. Preliminary research To explore whether improvement in cognitive functioning is asso­
demonstrated a relationship between cardiorespiratory fitness and ciated with baseline fitness, baseline cognitive functioning, or cognitive
cognitive test performance as well as brain structure and function in reserve, as indicated by years of education.
people with MS. [16,17] When this project was proposed, there were no
definitive trials of aerobic exercise training and cognition in MS, and this 2.1.4. Aim 4
supported an urgent need for research on exercise as a means of To determine whether the aerobic exercise induced improvements in
improving cognition in people with MS. cognitive processing speed, other cognitive functioning, cardiorespira­
We have subsequently learned that some disease modifying treat­ tory fitness and subjective outcomes are maintained 3 months after the
ments (DMTs) have a small effect on cognition, primarily improving end of training compared to the minimal exercise control group.
cognitive processing speed in relapsing-remitting MS. [18] Other drug We chose a randomized controlled design because it is the primary
therapies still have no demonstrable effect on cognition [19]. Cognitive method for producing Level I evidence. While it was not feasible to
rehabilitation has demonstrated only small beneficial effects on working conduct a double-blind, placebo-controlled trial, we used blinded
memory and visual learning [20]. The explosion of research on exercise outcome assessors and included several design features aimed at mini­
and cognition in humans has produced mixed results. One systematic mizing between-groups differences in outcome expectancies. Rather
review of 11 meta-analyses including 97 unique randomized controlled than describing the study as treatment vs. control, we described it as a
trials (RCTs) concluded that there is too little high-quality evidence comparison of two types of exercise training on cognitive functioning.
supporting that exercise improves cognition in disease-free older adults, We emphasized the credibility of the stretching and toning condition by
and there is clearer evidence of benefit in persons with mild cognitive explaining that similar interventions have improved cognition [26] and
impairment [21]. Among people with MS, the most recent meta-analysis neural connectivity [27]. We used an accepted manualized stretching
of 13 studies reported that the pooled effects of exercise training on and toning treatment that includes progressive, albeit mild, Theraband
cognition was null and the authors enumerated the many limitations of strengthening exercises over time [28]. We measured treatment credi­
included studies [22]. bility and outcome expectancies in both groups after an introduction to
Herein, we describe the rationale and design of the Graded Exercise the treatment condition and the first training session.
Training (GET) Smart trial to test the efficacy of aerobic exercise for
improving cognitive processing speed and other cognitive functions in 2.2. Study setting
people with MS. We include detailed information on the conduct of the
study, including required procedural revisions and lessons learned in The GET Smart trial was conducted within the University of Wash­
order to place the subsequent outcomes paper in the proper context and ington Medical Centers (UWMC). The exercise training was originally
inform future research in this area. designed to occur in a UWMC-affiliated outpatient rehabilitation facility
for people with MS overseen by physical therapists and trainers. How­
2. Methods ever, this made study participation inconvenient and limited recruit­
ment. Therefore, we expanded the training settings to include 14 YMCAs
2.1. Study aims, hypotheses and design in our local area. Each site had appropriate exercise facilities and access
to personal trainers who could be trained to deliver the intervention
The study is a single site, two-arm, randomized controlled efficacy conditions to participants.
study wherein participants are allocated 1:1 into 6-months of aerobic The fitness testing was conducted at a laboratory within the uni­
exercise training or a minimal exercise training attention control con­ versity medical center. That laboratory closed due to revenue shortfalls
dition. The specific aims and hypotheses are: and fitness testing was moved to the university-affiliated cancer research
center. At both sites, fitness testing was conducted by a trained exercise
2.1.1. Aim1 physiologist. However, the lab at the cancer center lacked access to
To determine whether aerobic exercise training significantly im­ physicians for EKG safety monitoring. As a result, with the move we
proves cognitive functioning in adults with MS. switched from a maximal fitness testing protocol to a sub-maximal
protocol for estimating aerobic fitness.
2.1.1.1. Primary study hypothesis. We hypothesize greater improvement
in cognitive processing speed from pre- to post-treatment in the aerobic 2.3. Participant eligibility
exercise group compared to the minimal exercise (stretching and toning)
control group. Cognitive processing speed will be measured by a com­ The study was initially designed for high internal validity with nar­
posite score that combines the Paced Auditory Serial Addition Test, 3-s row inclusion and exclusion criteria. Difficulty with recruitment led to
version (PASAT-3) [23–25] and the oral Symbol Digit Modalities Test loosening of these criteria as described below.
(SDMT) [24,25].
2.4. Sources of participants
2.1.1.2. Secondary study hypotheses. We hypothesize greater improve­
ment in other cognitive domains and patient reported outcomes from We planned to recruit participants through three local MS clinics
pre- to post-testing in the aerobic exercise group versus minimal exercise (ours followed 1230 unique patients per year), the NMSS chapter
control group. Other cognitive domains will be measured by the Mini­ (12,000 members) and a registry of prior MS research study participants
mal Assessment of Cognitive Function in MS (MACFIMS) test battery (n = 1009). We posted flyers in clinics, asked clinicians to inform their
[25]. patients about the study, emailed and called prior research participants,
posted ads on the NMSS website, sent email blasts to NMSS members and
gave presentations. One highly committed physician paid to promote
the study (PQ) contributed substantially to participant recruitment.

2
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

Participant recruitment expectations were based on informal survey Table 2

data showing that 77%–86% of people with MS are interested in having Original and final inclusion and exclusion criteria.
help to exercise, 82% were using disease-modifying therapies and 76% Original inclusion criteria Final inclusion criteria
were within the age range targeted by this grant proposal as well as
Women and men in proportion to their
published research on the prevalence of key inclusion/exclusion criteria estimated prevalence in the population
such as major depression [43] and cognitive processing speed impair­ (3:1). We considered an exclusively
ment [31]. female sample because at that time
most research suggested larger effect of
exercise on cognition among women
2.5. Screening, informed consent and enrollment procedures versus men [30]. However, this would
come at the cost of poorer
Screening for the trial was a four-step process. In step one, research generalizability to people with MS
staff interviewed potential participants to determine whether they met overall. We considered equal numbers
of men and women. This would result in
basic demographic, medical history, psychiatric history, and exercise
equal power to detect significant effects
related inclusion/exclusion criteria. Second, we obtained written of aerobic exercise in both sexes but
permission to review their medical record and contact the person’s MS also reduce the predicted effect size,
physician to confirm the diagnosis of MS, date of onset, and type as well require a larger sample size and make it
as the absence of any disqualifying comorbid medical conditions. The more difficult to recruit the target
sample size.
physician determined whether the person is safe to initiate an aerobic Demonstrate impaired cognitive processing Ultimately, we included those who
exercise program. The third step was an in-person prescreening session speed or working memory based on either were below mean for age, sex and
during which research staff conducted a near vision eye test, the EDSS the PASAT-3 or the SDMT (i.e., z < − 1.5 education (z < 0.0) at baseline.
exam, and the SCID major depression module if the person screened controlling for age, education and sex)
[31]. We proposed procedures
positive for major depression on the PHQ-9 during step one. Then the
previously shown to identify a mildly
PASAT-3 and the SDMT were administered twice (see Table 2 for test impaired group and excluded those
forms used). The first administration was used to identify those who met with more severe impairment because
the eligibility criterion of having below average cognitive processing they may not be able to participate fully
speed (using regression-based norms) on either measure. The extra or benefit [5].
Aged 18–54. We restricted age to avoid We expanded to 18–59 to facilitate
administration of the PASAT and SDMT was used to generate practice potential age-related cognitive recruitment and because age-related
effects prior to the baseline assessment. Those who remained eligible impairment or Alzheimer’s dementia changes in cognition seem to begin
were invited to return for the baseline assessment within 1–2 weeks. At that is more likely to be present in older even before middle age and we screen
the beginning of the baseline assessment, which also served as the fourth adults. out person with dementia.
English is primary language as cognitive tests
screening step, a research staff member fully explained study partici­
were not normed on other language
pation and obtained written informed consent. Afterwards subjects un­ groups.
derwent the baseline neuropsychological assessment. Those who met At least 9th grade education.
criteria for dementia based on that assessment (see exclusion criteria) Physician confirmed, clinically definite MS
were not randomized. (See Table 1.) diagnosis using the revised McDonald
criteria[32].
MS diagnosis at least 5 years prior. We changed this to 6 months post-
2.6. Randomization, allocation concealment and procedures to minimize diagnosis when data became available
bias regarding the safety of exercise in MS
and specifically that exercise does not
increase the risk of relapse [33]. We
Central computer-generated randomization was performed by the
also reasoned that, if successful,
biostatistician and then transmitted to the exercise trainer via secure exercise and might be employed as an
email. We used permuted block randomization (with block sizes of 2 or 4 early intervention.
randomly chosen) to balance group sizes within cells. The ratio of Withdraw and replace subjects who relapse. Because exercise does not increase risk
allocation to aerobic exercise versus stretching/toning was 1:1 to of relapse and participants wanted to
resume exercise, we allowed them to
maximize the precision of any between-groups differences. Randomi­ resume with physician approval and
zation was stratified based on sex because there is some evidence that extended their exercise training period
women and men may differ on magnitude of exercise-induced cognitive to make up for lost time.
improvement [12]. Able to walk at least 100 m without
assistance (Expanded Disability Severity
The persons trained to perform baseline and outcome assessments
Scale (EDSS) score equivalent 0–5.5)
were kept unaware of the study participant’s condition. We sought to [34]. We excluded people with greater
maintain blinding of the assessor by having other personnel manage the disability because it is uncertain
participant’s progress through the trial and deal with study-related whether they could tolerate the aerobic
questions. We held separate staff meetings from which the assessor conditioning program. We chose a
subgroup most likely to respond and to
was excluded. We endeavored kept the assessor physically separate from reduce the risk of null or weak results.
staff who managed participant progress. We monitored instances of All MS subtypes will be included.
unblinding. Be on disease-modifying therapies (DMTs). We dropped this requirement because
there was no research to suggest
exercise effects would be dependent on
2.7. Study intervention
being on a DMT. We recorded DMT
status and type for post-hoc analyses
2.7.1. Treatment conditions outcome predictors.
Both the aerobic exercise training and the stretching/toning atten­ Currently exercising less than public health We required no periods of exercise
tion control conditions were delivered though the same outpatient clinic recommendations[35](less than 30 min training within the last 6 months.
of structured physical activity less than 3
and YMCAs. The program was manualized for reproducibility, and times per week during the past 6 months)
delivered and documented with the assistance of an exercise trainer.
(continued on next page)
Heart rate monitors and individualized treatment manuals were sent to
the participant’s chosen exercise location. The manual consisted of a

3
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

Table 2 (continued ) Table 2 (continued )

Original inclusion criteria Final inclusion criteria Original inclusion criteria Final inclusion criteria

in order to avoid a ceiling effect on potential effects on depression-related

aerobic training. cognitive impairment [38].
Physician clearance to engage in aerobic Have current alcohol or other drug abuse. In the case of marijuana use, which
exercise training. Included to ensure We used the AUDIT-C [39] and CAGE became more popular and then legal to
participant safety. questionnaire [40] to exclude for use recreationally, we initially excluded
Willing and able to participate in either We modified the location so that alcohol abuse and the WHO Alcohol, those who used unless by prescription.
exercise training program 3 days per week participants could exercise at their Smoking and Substance Involvement Ultimately we excluded chronic regular
for 6 months at the Outpatient preferred YMCA. Screening Test (WHO ASSIST; V3.0) to users and included occasional users
Rehabilitation Clinic. exclude for other drug use (excluded if regardless of prescription status due to
Willing not to undertake additional they ever used drug in the past three evidence of prolonged cognitive
structured exercise or leisure time physical months) [41] because of the potential impairment [42].
activity during the 6-month trial. influence on study participation and
cognitive testing.
Original exclusion criteria Final exclusion criteria
Received steroids in last 30 days or a relapse
in the last 90 days. People who were
otherwise eligible were deferred until Table 3
they met these criteria.
Screening and primary outcome measurement plan.
Contra-indications for exercise training.
Based on American College of Sports Prescreen Trial 1a PASAT-3” Rao Form A [24]; SDMT Smith Form [29]
Medicine screening criteria using Prescreen Trial 2b PASAT Gronwall 2.4” [44]; SDMT Rao Form 1 [24]
Physical Activity Readiness Baseline PASAT-3” Rao Form B [24]; SDMT Benedict Form 1 [45]
-Questionnaire (PAR-Q) [35]. Primary outcome PASAT-3” Rao Form A [24]; SDMT Smith Form [29]
Current Dementia. Criteria from a To include more people who might Follow-up PASAT-3” Rao Form B [24]; SDMT Benedict Form 1 [45]
definition validated in people with MS benefit given the efficacy of exercise for a
using the MACFIMS [25] (> 2 SD below persons with mild dementia we adopted
Used to exclude unimpaired persons.
b
the mean on at least one memory test more restrictive criteria (i.e., z < 1 Used as run-in, practice trial.
and > 2 SD below the mean on at least percentile of composite of CVLT Total,
one neuropsychological test in another CVLT Delay, BVMT Total, BVMT Delay
domain) [36]. and z < 1 percentile on another test in Table 4
another domain except SDMT & PASAT Aerobic training schedule.
controlling for age, education and sex).
Have had neuropsychological testing within We changed to the past 6 months Program stage Week Days per Intensity (% Duration (min/
the past year. week HRR) d)
Near visual acuity with correction 20/70 Initial stage 1 3 40–50 15–20
or better (sufficient to perform 2 3 40–50 20–25
cognitive testing). 3 3 50–60 20–25
Has neurological disease other than MS that 4 3 50–60 25–30
may impact cognitive status (Alzheimer’s Improvement 5–7 3 60–70 25–30
disease, Parkinson’s, stroke, TIA, stage 8–10 3 60–70 30–35
Vascular Dementia, Huntington’s, 11–13 3 65–75 30–35
traumatic brain injury, learning 14–16 3 65–75 30–35
disorder, seizure disorder, or chronic 17–20 3 65–75 35–40
CNS infection). 21–24 3 65–75 35–40
Using any medication known to have adverse To enhance the representativeness of
effects on motor or cognitive function (i.e., study findings, we switched from
monoamine oxidase inhibitors, excluding people taking these
personalized exercise prescription, 24 copies of the appropriate activity
sympathomimetics, antipsychotic medications to requiring them to be on
agents, modafinil, oxybutynin, tricyclic stable doses for at least 6 weeks prior to log, 12 self-challenge worksheets, 6 goals worksheets, a reaction to
antidepressants, benzodiazepines, baseline. We assessed change in treatment questionnaire, a list of the benefits of their respective exercise
cholinesterase inhibitors, and medications and documented all type, a copy of the combined exercise handouts, and the appropriate
anticonvulsants other than gabapentin medications being used at 6 and 9 participant manual. Trainers supervised the 72 planned exercise sessions
and pregabalin). months. For PRN drugs, we asked
participants not to use within 24 h of
on a tapered schedule: every session for sessions 1–7, every third session
testing. No alcohol or caffeine were to for sessions 8–22, every sixth session for sessions 23–34, and every 9–10
be used the day of exercise testing. sessions for sessions 35–72. Participants were asked not to undertake
Have prior history of serious mental illness or We removed learning disorder as an additional exercise during the study and this will be monitored by
developmental or learning disorder prior to exclusion.
comparing an exercise history [46,47] and the Godin Leisure-Time Ex­
MS diagnosis (e.g., obsessive-
compulsive disorder, schizophrenia, ercise Questionnaire administered at baseline, post-treatment and three-
other psychotic disorders, bipolar month follow-up [48–50].
disorder or major depressive disorder
by self-report). These conditions may 2.7.2. Aerobic exercise condition
affect participation and confound
The exercise condition is a graduated program of supervised aerobic
interpretation of study results.
Have current major depressive disorder exercise up to one hour per day, three days per week and lasting 6
(MDD). As indicated by a total score of months. The time, intensity and duration of the exercise program was
10 or greater on the Patient Health based on American College of Sports Medicine (ACSM) [51] guidelines
Questionnaire-9 followed by a
for improving cardiorespiratory fitness in sedentary persons (i.e.,
Structured Clinical Interview for DSM-5
(SCID) major depressive disorder VO2peak) and consistent with procedures used in successful controlled
module [37]. We excluded for MDD trials of aerobic exercise for cognitive improvement in older adults
because we did not want to confound [12,30] The program was tailored for people with MS who are sedentary
the potential effects of exercise on MS and may have difficulties with fatigue, heat sensitivity and mild cogni­
related cognitive impairments with the
tive impairment (See Table 4) [46,52].
The regimen was 3 days per week and the structured exercise portion

4
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

Table 1 Table 1 (continued )

Overview of methods: prior limitations, suggestions and current study Prior study General methods to Study-specific methods to
characteristics. limitations overcome limitations overcome limitations
Prior study General methods to Study-specific methods to fitness testing results.
limitations overcome limitations overcome limitations Parallel standardized,
Sample related progressive stretching and
Small sample size Perform a priori sample Power calculation performed toning program used for
size calculation with based on effect sizes from control group.
Bonferroni correction for prior research *Unsupervised or *Both treatment and *Used identical supervision
multiple tests *Power calculation was differentially control groups are plan for both groups
adjusted for heterogeneity in supervised groups supervised equally
effect sizes in men versus Lack of control for Using activity trackers or/ Assessed self-reported
women and the anticipated physical activity and questionnaires physical activity in both
ratio of women to men (3:1) beyond the treatment and control
in the sample intervention groups
*Attrition *Plan for attrition *Target sample size set at *Poor adherence to *Used social cognitive
125 to retain 100 (80%) treatment theory strategies to boost
Inclusion of Cognitive impairment as Included persons who scored adherence in both groups
cognitively intact an inclusion criterion below age, sex, and Study design related
subjects education regression-based Poor construction of Follow recommended Used design principles
means on either the PASAT-3 study designs guidelines (e.g. PRISMA consistent with PRISMA and
[23,24] or SDMT [29] [20], TESTEX [22]) TESTEX and will report
Very heterogenous Use more selective Included persons with EDSS results using those schemes
sample (e.g., MS inclusion criteria (e.g. 0–5.5 (able to ambulate No blinding of Consider blinding the Assessor was blinded
subtypes, EDSS) selecting MS subtypes, a unassisted for at least 100 m participants, assessor and when possible *In lieu of blinding
narrower EDSS range) therapists, and try to blind participants to participants, active control
*Excluded those who already assessors the study hypothesis group was described in ways
met public health physical to minimize between-groups
activity guideline differences in subject
* Excluded persons with expectations of benefit
major depression such that *Assessed subject
any observed improvement expectations of benefit in
in cognition is not due to the both groups
antidepressant effects of Very long testing Create more focused MACFIMS [25] is a brief yet
exercise periods could testing procedures comprehensive battery of
Assessment of cognition related impact study neuropsychological tests
Lack of validated and Choice of specific test Utilized NINDS Common outcomes tailored to people with MS
recommended test batteries suited to the Data Element (CDE) tests of Lack of control group Use a passive control group Progressive stretching-
batteries targeted population cognitive processing speed or only active controlling for social toning group used to control
for the primary outcome control group exposure or Hawthorne for social exposure and
(PASAT-3, SDMT) effect with the possible subject expectations of
Utilized MACFIMS [25] addition of a third group benefit
which consists of NINDS CDE being an active control
tests to measure secondary *No assessment of *Include potential *We measured depression,
outcomes potential mediator variables affect, sleep, and fatigue at
*All tests have age, sex, and mediators all time points
education-based norms for
Table adapted from: Gharakhanlou and colleagues 2020 [22].
people with MS *
These items represent additional methods we recommend and used in this
*Chose single primary
outcome that was a trial.
composite of two measures,
PASAT-3 and SDMT of the sessions was 15–20 min at 40%–50% heart rate reserve (HRR) and
*Improvement *Minimize practice effects *Used tests with alternate
progressively increase up to 35–40 min in duration at 65–75% HRR
confounded by forms
practice effects *Excluded persons exposed during the last month of the program. When study procedures changed
to the same tests within the from maximal to submaximal fitness testing, participants were allowed
past 6 months to work out outside of their prescribed heart rate zones and use RPE and
*Generate practice effects *Exposed participants to the Talk Test to regulate their workout intensity.
before baseline assessment primary outcome measures
twice prior to baseline
The training was divided into two stages: an initial conditioning and
adjustment stage that is preparatory for the improvement stage. The
Exercise intervention related
sessions began and ended with 5 min of warm-up, proceed with the
Inadequate Choice of standardized and Planned VO2-max fitness
assessment to well-controlled tests (e.g. testing at all time points structured portion of the exercise training and ended with 5 min of cool-
prescribe Cardiopulmonary Exercise (Switched to submaximal down. The training focused on large, dynamic movements of the lower
individual exercise Testing until exhaustion) to VO2-peak estimation when lab extremities (e.g., treadmill walking/running, leg cycling ergometry).
intensity define exercise intensity closed and physician Participants used exercise logs to record training parameters (e.g.,
for an aerobic exercise oversight of EKG was not
intervention available)
intensity, duration, and frequency) for each session over the 6-month
Lack of detailed Precise and comprehensive Used standardized, period. The exercise prescription is tailored to each person’s capacity
description of description of intensity (e. progressive, two stage (i.e., % HRR) and adaptability. After the first session the trainer
exercise g. percentage of VO2Peak), (initial and improvement) administered an adaptation of section A of the Reaction to Treatment
interventions frequency, mode, and aerobic conditioning
Questionnaire [53] a measure of treatment credibility and outcome
duration of exercise program designed for
interventions) sedentary adults and tailored expectancies.
to the individual’s baseline
heart rate reserve from 2.7.3. Stretching/toning attention control condition
We included a minimal exercise comparison arm to control for

5
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

expectations regarding the benefits of engaging in exercise training, and Adherence to the 6- and 9-month outcome assessment. During the
for nonspecific treatment factors, such as trainer attention and social last year of the trial when several participants were declining the
contact, participation in regular, minimal physical activity as well as outcome assessments, we boosted participant retention by increasing
natural improvement and regression to the mean on outcome measures. the financial incentive to complete the 6- and 9-month assessments from
This is the standard control condition for studies on exercise training to $30–40 to $200.
improve cognition in older adults [12,30]. It will further provide an The trial was conducted using the intent-to-treat principle. We did
estimate of the spontaneous deterioration or improvement on outcome not stop the exercise training program for poor compliance with treat­
measures over 6 months. A minimal exercise condition was ethical ment and we made every effort to obtain outcome data on all partici­
because, as yet no type or dose of physical activity had been proven to be pants regardless of exercise participation. We required participants to
effective for improving cognition in MS. stop exercise for relapse or injury on a case-by-case in consultation with
The minimal exercise/stretching and toning program was delivered the participant and health care provider. If exercise was stopped,
in the same settings, by the same exercise trainers, and using the same physician approval was required to resume exercise training. We
frequency and duration as in the aerobic training program. The extended the exercise program to make up for lost training time. Exer­
stretching exercises were based on a manual published by the National cise training was not resumed if the program was interrupted for 3
Multiple Sclerosis Society [28]. The program was standardized and months or more because by then the program would have to be restarted
manualized for reproducibility and progresses by increasing session time from the beginning.
from 15 to 40 min, including more stretches and sets with Therabands We also planned a completers analysis. Completers were defined as
for resistance as well as some basic Yoga poses (e.g., Savasana, Tada­ those who participated in 75% or more of all planned sessions and did
sana, Sukhasana and Vajrasana) over the course of the 6-month period. not drop out due to injury or relapse.
Five minutes of warm-up and cool-down on an ergometric machine was
allowed before and after the stretching sessions. Participants used ex­ 2.8. Data collection procedures and measures
ercise logs to record their exercises and training parameters (e.g.,
duration and frequency) for each session over the 6-month period. Study assessments were planned to occur at baseline, at the end of
Early on in the trial, participants criticized the stretching and toning the 6-month treatment period and at follow-up (3 months after the end
condition for requiring them to attend sessions that were only 15 min in of treatment). The assessment windows were set to be within one week
length and focused on stretching only the head, neck and shoulders, plus before or after the target date. When participants had relapses or injuries
warm-up and cool-down. Some refused to comply with the program and requiring an extension of their exercise program, the assessments were
threatened to withdrawal altogether. Therefore, similar to how we also delayed correspondingly. When the fitness testing lab closed there
adjusted the aerobic exercise prescriptions based on %HRR, we allowed were delays in outcome assessments. When the post-treatment assess­
participants to move ahead more quickly on the stretching and toning ment was delayed participants were instructed to continue their
protocol based on subjective tolerance for these exercises. assigned exercise program until they could undergo the post-treatment
assessment. In cases where fitness testing could not occur within win­
2.7.4. Safety and adherence dow, self-report and neuropsychological testing were administered
We monitored subjects for safety and compliance throughout the separately and as close to the target date.
trial. Safety monitoring included obtaining the date, type cause, study Assessments consisted of self-report measures, followed by neuro­
relatedness and outcome of any adverse or serious adverse events psychological testing, and then fitness testing. Neuropsychological
including relapses and injuries. Originally the trainer maintained an testing was performed by research staff trained and supervised by the
exercise log of every session for each subject. When the intervention was study neuropsychologist (RHBB). The primary outcome measures
moved to multiple YMCAs around the region, we relied on participants (PASAT, SDMT) were administered twice before the baseline assessment
to keep exercise logs and monitor adverse events. Participants used to minimize the role of practice effects and increase detection of training
exercise logs to track the date, time, resting heart rate (RHR), duration effects during the trial. Serial testing studies show that the largest
and Rating of Perceived Exertion (RPE) of each session as well as post- practice effects occur between the first and second test exposures [57].
exercise feeling (− 5 to +5), enjoyment (1–7), mental fatigue (0− 10), Whenever possible, participants were tested by the same research staff
and physical fatigue (0–10). Participants in the control group further member and at the same time of day at each assessment point. Neuro­
reported which stretches were performed and any balance or yoga psychological tests were administered in the standard fixed order. Self-
engagement. report measures were administered in an interview format by research
The trial included the use of specific motivational strategies to pro­ staff to ensure data completeness. Maximal, incremental exercise testing
mote safety, enjoyment and adherence based on social cognitive theory was performed by an ACSM-certified clinical exercise physiologist in a
[54]. Trainers were provided with a treatment manual instructed to use dedicated physiological testing lab within the Clinical Research Unit at
strategies described therein to enhance safety and comfort, to set UW Medical Center.
SMART goals, to form appropriate outcome expectations, bolster exer­ Originally, we utilized a maximal testing protocol with EKG moni­
cise self-efficacy, and increase enjoyment (manual available from cor­ toring for adverse events by a cardiologist, as required by our local
responding author). human subjects review board. However, the testing lab closed due to loss
Initially, we planned to have trainers employ motivational in­ of funding during the study. Exercise testing was moved to the only
terventions for those with attendance below 90% in either arm, or other research lab in our system at Fred Hutchinson Cancer Research
failure to improve for three successive weeks in terms of duration or Center. That lab did not have physician availability to oversee EKG
intensity in the aerobic training arm. This approach is consistent with safety monitoring. Therefore, the study adopted a sub-maximal exercise
what has been done in previous research on exercise training and testing protocol, and this involved monitoring HR during the submaxi­
cognition in older adults [55,56]. When the intervention moved to mal test performed until 85% of age-predicted HR maximum, and then
YMCA centers, direct contact between study staff and trainers was often linear estimation of peak oxygen consumption. This permitted estima­
not possible. We switched to indirect participant monitoring and tion of the same outcome as maximal exercise testing, but using a sub­
compliance management through the manager at each YMCA. Access to maximal test.
YMCA managers was challenging and there was high turnover in the
group. Therefore, an unblinded study staff member was hired to call 2.9. Outcome measures
participants monthly to monitor exercise participation and promote
adherence to the exercise program. All primary and secondary outcome measures consist of

6
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

neuropsychological tests from the Minimal Assessment of Cognitive disease course, MRI detected lesion burden and work disability
Function in MS MACFIMS [25]. The MACFIMS is a relatively brief yet [25,59,63].
comprehensive battery of neuropsychological tests constituted specif­
ically for the assessment of people with MS by a consensus of experts in 2.9.2. Secondary outcomes
MS neurology and neuropsychology. The MACFIMS includes the PASAT- Our decision to include four other MACFIMS tests was based on a
3′′ and SDMT as well as five other tests (see Secondary Outcomes Mea­ meta-analysis of exercise-related cognitive improvement which
sures, below). The MACFIMS was chosen because it is comprised of demonstrated positive effects in all domains, albeit larger effects in the
commonly used, standardized neuropsychological tests that have good domain of our primary outcome [12]. The MACFIMS consists of the
test-retest reliability, well established validity and equivalent alternate following tests and corresponding outcome variables. The Delis Kaplan
forms [25,31]. Alternate forms were used to minimize practice effects Executive Function System (DKEFS) [69] Sorting Test [70] will be used
associated with repeat testing [57]. The PASAT-3′′ has two equivalent to assess higher executive function. The dependent variables are the
forms (Rao forms A, B) [24] while the two most equivalent forms of the total number of correct sorts and the total verbal description score be­
SDMT are the Smith [29] and Benedict [45] versions [45]. The plan for tween the two decks. The California Verbal Learning Test, second edi­
using the alternate forms is described in Table 3. The other MACFIMS tion (CVLT2) [71] will be used to measure verbal learning, the total
tests also have original and alternate versions. We used the original number of words recalled during the initial learning trials and verbal
versions at baseline, the alternate versions at post-treatment and the recall, the number of words recalled after the delay interval. The Brief
original versions at follow-up. The study neuropsychologist (RHBB) Visuospatial Memory Test Revised (BVMTR) [72] measures visual
trained study staff to administer and score the neuropsychological tests learning via the total figure reproduction score following three learning
via instruction, observation, and coaching until the staff member ach­ trials as well as figural recall, the total number recalled after a 20–25
ieved competence. The neuropsychologist also made final decisions min delay. Verbal fluency will be measured with the Controlled Oral
regarding test scoring questions. Word Association Test (COWAT) following the method of Arthur Benton
[73]. The outcome is the total number of correct words over the three
2.9.1. Primary outcome trials. Spatial processing will be assessed with the Judgment of Line
The primary outcome will be a composite score based on the PASAT- Orientation Test (JLO) [73]. The outcome measure will be the total
3′′ [24,58] and the SDMT (oral versions) [29]. The PASAT-3′′ is an number of correct responses over 30 items.
auditory processing speed measure in which subjects are exposed to
single digit numbers voiced every three seconds. The main score is the 2.9.3. Tertiary outcome measure
number of correct responses. The SDMT will be used to measure visual Cardiorespiratory fitness will be measured at baseline and 6-month
processing speed. This test presents a stimulus key of numbers paired outcome as a manipulation check, that is, to demonstrate that the aer­
with abstract symbols at the top of a page. Participants scan the page obic training group improved cardiorespiratory fitness to a greater
below the key that has rows of symbols without the paired numbers. The extent than among those in the control condition. Cardiorespiratory
task is to generate the associated numbers orally as fast as possible. To fitness is defined as peak oxygen consumption (VO2peak) using an in­
create the composite primary outcome, based on prior work from our cremental exercise test on an electronically-braked, computer-driven
group [45,59] and elsewhere [60], we will transform results from each cycle ergometer. The exercise scientist will fit participants to the bicycle
measure into z-scores (with a mean of 0 and a standard deviation of 1) ergometer and explain the test procedures including how to provide
and compute the average of the two z-scores into a mean z-score. ratings of perceived exertion (RPE). After giving the participant a chance
The rationale for the primary outcome is as follows. The PASAT-3′′ to ask questions, the exercise scientist will occlude the participant’s nose
and SDMT are highly reliable, valid and sensitive measures of MS related and insert a mouthpiece for collecting expired gases. Participants will
cognitive impairment [25]. The cognitive domains measured by these start with a 5-min warm-up at a work rate of 0 W. Thereafter, the work
tests, processing speed and working memory, are the most commonly rate will continuously increase at a rate of 15 W⋅min− 1 until the
impaired cognitive processes affected by MS. [61] These two measures participant reaches the point of volitional termination.
are complementary in that both measure cognitive processing speed and Using an open-circuit spirometry system, oxygen consumption
working memory in the auditory (PASAT) and visual (SDMT) modalities. (VO2), carbon dioxide production (VCO2), ventilation (VE), and respi­
The PASAT-3′′ and SDMT are correlated with MS-related changes in ratory exchange ratio (RER) will be measured every 20 s. Heart rate will
brain structure and functioning as measured by MRI and fMRI [62–64]. be displayed using a Polar heart rate monitor, and HR and RPE will be
The PASAT-3′′ is correlated with work disability [25]. Worsening per­ recorded every minute during the test. The main outcome, of this test,
formance on the SDMT predicts loss of employment over a three-year VO2peak will be defined as the highest recorded VO2 value expressed in
period [65]. In addition, faster processing speed on measures such as ml/kg/min when 2 of the following 3 criteria are satisfied: RER ≥1.10;
the PASAT-3′′ are correlated with higher levels of cardiorespiratory peak HR within 10 beats/min of age-predicted maximum (i.e., ~1 SD);
fitness in persons with MS. [66] Aerobic conditioning improves perfor­ or peak RPE ≥17. The resting and peak HR will be used for prescribing
mance on the SDMT in older adults with mild cognitive impairment exercise training intensity based on the Karvonen method of HR reserve
[30]. (Target Heart Rate = ((HRmax − HRrest) × % intensity) + HRrest)
We chose to use a composite primary outcome variable that com­ described below and used in the aging literature on exercise and
bines these two measures for several reasons. First, these two measures cognition. This same protocol has been used in our previous MS research
have been combined into a composite score for previous cardio- [66,74,75], is well tolerated and safe. Nevertheless, the maximum
respiratory related research in people with MS. [59,66] Second, these testing protocol included EKG monitoring to conform to local safety
two tests measure complementary aspects (sensory and motor) of the requirements.
same domain so as to improve the reliability with which we can measure
cognitive speed of processing. Third, these tests have been combined to 2.9.4. Participant reported outcomes
measure information processing speed and efficiency in studies of We obtained self-report data at baseline, 6 months and 9 months on
cognitive deterioration in MS. [60,67] Finally, whereas the PASAT-3′′ selected-variables. Subjective impairment was measured with the Mul­
was single cognitive measure included in the MS Functional Composite, tiple Sclerosis Neuropsychological Screening Questionnaire [76] and the
the “gold standard” multidimensional measure of impairment and Perceived Deficits Questionnaire [77]. The Patient Health
disability in people with MS [68], the SDMT has some advantages Questionnaire-9 was used as a measure of depression symptomatology
including acceptability to patients, ease of administration and slightly [37]. The Pittsburgh Sleep Quality Index [78], the Fatigue Severity Scale
better reliability and validity, including predicting disease status, [79], and the Positive and Negative Affect Scale [80] were also

7
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

administered. significance level of p < .05 will be used for testing the primary hy­
pothesis. The primary analysis will be run in accordance with the intent
2.9.5. Process measures to treat principle, that is we will perform the outcome assessment on all
We used the Godin Leisure-Time Exercise Questionnaire [48] and the randomized people at 6 months regardless of their compliance, dropping
International Physical Activity Questionnaire-Short Form [81] to out of the treatment, or MS relapses and include all people in the group
monitor overall physical activity retrospectively in both the treatment to which they were assigned. We will perform a secondary “completers”
and control groups. We measured subjective outcome expectancies and analysis excluding those who did not receive an adequate dose of
treatment credibility were measured after the first assigned treatment treatment, completion of at least 75% of assigned treatment sessions. We
session using an adaptation of section A of the Reaction to Treatment would also exclude those who started an exercise program outside their
Questionnaire (RTQ) [53]. Responses on the RTQ will be used to assigned treatment or who had an MS relapse.
compare participant beliefs of the two treatments and to predict
outcomes. 2.10.3.3. Secondary study hypotheses. We hypothesize that improve­
ment in other cognitive domains captured by the MACFIMS (e.g.,
learning and memory, visual-spatial processing, executive functioning
2.10. Statistical analyses and expressive language), subjective cognitive impairment and depres­
sion from pre- to post-treatment will be significantly greater in the
2.10.1. Statistical power aerobic exercise group compared to the minimal exercise/attention
With a sample size of 50 subjects per group (100 total), the study has control group.
an 80% chance to detect a significant between-groups effect-size (p <
.05), based on a regression analysis testing the group effect on the post- 2.10.3.4. Secondary statistical analyses. Analyses will be similar to the
intervention composite PASAT-3”/SDMT scores after controlling for pre- approach outlined for the primary outcome. The significance level for
intervention scores, the stratification factor (women vs. men) and any these multiple tests will be subject to Bonferroni correction.
key demographic (age, years of education) or disease severity (EDSS,
years since diagnosis) variable that was imbalanced at baseline despite 2.10.4. Aim 2
randomization. This is based on an effect sizes for women (0.67) and To explore whether improvement in cognitive functioning is related
men (0.29) from the Baker study [30], adjusted for the ratio of women to to improvement in VO2-peak.
men that we plan to recruit into the sample (3:1). The resulting esti­
mated ES is 0.575. This ES is more conservative than the ES reported in a
2.10.4.1. Aim 2 statistical analyses. These analyses will be similar to
meta-analysis of studies like ours that treated mostly females (0.60),
that indicated for the primary aim but including change in VO2 peak in
used 31–45 min sessions (0.61) and exercised for at least 6 months
addition to assigned treatment. Cardiorespiratory fitness will be
(0.67) [12]. The estimated power is also conservative in that the com­
considered a mediator if it is significantly related in this model and
posite measure should have lower variability than the SDMT score
assigned group shows a diminished effect from the primary analysis.
alone.
2.10.5. Aim 3
2.10.2. Randomization effectiveness
To explore whether greater improvement in cognitive functioning is
The use of stratification should assure similar proportions of women
associated with baseline fitness and cognitive reserve.
and men in the aerobic training vs. minimal exercise/attention control
groups. However, data analyses will be conducted to determine
2.10.5.1. Aim 3 statistical analyses. These analyses will be similar to
comparability of the groups at pre-treatment. We will compare groups
that indicated for the primary aim but including a potential moderator
on sex, age, years of education, EDSS, MS type, pre-treatment minutes of
and the interaction of that moderator with treatment in addition to
moderate to vigorous physical activity per week, baseline VO2peak and
assigned treatment. The variable will be considered a moderator of the
baseline cognitive processing speed (PASAT-3′′ and SDMT composite
interaction with treatment is significant.
score). Appropriate analyses (chi-squares for categorical variables, t-
tests for continuous variables with reasonably normal distributions or
2.10.6. Aim 4
Mann-Whitney tests for ordered categorical and decidedly non-normal
To determine whether 6-month aerobic exercise training results in
variables) will be performed comparing the groups on these important
improved cognitive speed processing and improved cardiorespiratory
variables. Variables that are significantly different between the groups
fitness that persist three months after the end of training compared to
at baseline (defined as p ≤ .05), will be reported and potentially entered
the minimal exercise control group.
as a covariate in all planned outcome analyses to control for its potential
confounding effect.
2.10.6.1. Aim 4 statistical analyses. These analyses will be similar to the
primary study analyses except we will examine change in neuropsy­
2.10.3. Aim 1
chological functioning and fitness from baseline to three months after
To determine whether exercise training significantly improves
the end of exercise training.
cognitive functioning in adults with MS.

2.10.3.1. Primary study hypothesis. We hypothesize that improvement 2.11. Trial status
in cognitive processing speed (as measured by a standardized composite
z-score combining performance on the PASAT-3′′ and SDMT from The University of Washington Human Subjects Division approved
baseline to post-treatment) will be greater in the aerobic exercise group the study protocol on January 21, 2014. Recruitment began in May 2014
compared to the minimal exercise/attention control group. and ended in May 2019. The final 9-month assessment was delayed by
the COVID-19 pandemic and was completed in August 2020.
2.10.3.2. Primary statistical analysis. We will test this hypothesis using a
mixed-effects linear regression on the adjusted change in the PASAT-3’/ 3. Discussion and conclusions
SDMT composite score from baseline to post-treatment. The analyses
will be adjusted for baseline covariates such as those that were imbal­ This study protocol describes the rationale, design and necessary
anced between treatment groups despite the randomization. A modifications of a trial to determine whether aerobic exercise improves

8
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

cognitive processing speed and other cognitive functions in individuals An advantage of the dosing study is that it would identify the minimally
with MS. The GET Smart trial was designed to exceed contemporary efficacious dose, if one existed and does not assume we know the correct
methodological standards and can potentially inform future studies. dose for this population. The design with a no treatment arm is more
This trial also exposes common challenges investigators face conducting relevant to answering the “real world” question about whether adding
rigorous trials and describes efforts to complete the study while main­ an exercise program would improve cognition. The main limitation of
taining trial integrity. these alternative designs is that they require larger sample sizes to
The rationale for this trial remains compelling. Cognitive impair­ adequately power the studies and recruiting sufficiently large samples is
ment is a major under-treated aspect of MS and there is still no widely already quite challenging for two-arm studies.
accepted treatment to reverse impairment or prevent deterioration. Finally, this study faced a number of common and uncommon
Evidence for the efficacy of DMTs to improve cognition appears limited. operational challenges. Participant recruitment is a universal challenge
A recent meta-analysis restricted to people with RRMS found only 14 of in exercise trials as demonstrated by the small sample sizes in most
55 treatment samples were RCTs and 41 studies measured cognitive studies [22]. Despite a large population of people with MS in our area
processing speed with either the PASAT of SDMT. The authors reported a and local survey data indicating overwhelming demand for help to ex­
small effect (g = 0.27, 0.28) of platform (e.g., b-interferon) and esca­ ercise, we found it extremely challenging to recruit for this study. To
lation (e.g., natalizumab or alemtuzumab) DMTs, respectively on boost recruitment we eliminated or modified overly restrictive inclu­
cognitive processing speed [18]. The effects of other agents such as sion/exclusion criteria and advertised the trial via local MS providers
acetylcholinesterase inhibitors and neurostimulants on cognition in MS and healthcare institutions, regional MS associations, and the National
are not superior to placebo [19]. Cognitive rehabilitation, including Multiple Sclerosis Society. A major barrier to participation was
both computer-based cognitive exercises as well as strategy-based geographic distance from our exercise center. Thus, we shifted the
training, has demonstrated only small effects in selected domains in intervention to the regional network of YMCAs. While this strategy
people with RRMS [20]. improved recruitment, it also taxed our capacity to train, supervise and
With regard to aerobic exercise, the failure of most studies to incentivize the personal trainers at each site who delivered the in­
demonstrate a positive effect on cognition may be due to fundamental terventions and to closely follow participants. After we exhausted all of
flaws in the underlying theory. For example, Diamond and Ling [82] our recruitment sources, we partnered with a single MS neurologist (PQ)
argue that simple aerobic exercise has weak effects on executive func­ who specialized in recently diagnosed patients and was extremely
tioning because during the course of simple exercise executive functions enthusiastic about personally recruiting participants for an exercise
are not challenged. They believe that activities that engage executive trial. With her assistance and two no cost extensions we were able to
functions by combining mindfulness practices and movement have more approximate our target enrollment.
potential to improve these processes [82]. Complaints about the stretching and toning control condition
The absence of a more robust aerobic exercise effect on cognition emerged early in the trial. Threats of nonadherence or dropout led to our
may be due to methodological weaknesses of past research. The current being more permissive about how quickly participants progressed into
trial attempted to meet not only the PRISMA [83] and TESTEX [84] longer sessions of stretching and toning exercises involving more muscle
guidelines, but also addressed a number of more nuanced methodolog­ groups as well as simple Yoga poses. This change was essential to
ical considerations that may have undermined prior studies. For continue the trial, but may have the effect of increasing the exercise dose
example, our primary outcome combines the two most widely used tests within the control group and narrowing differences relative to the aer­
of cognitive processing speed into a single measure. This composite obic conditioning group.
measure incorporates central, auditory, visual and oral-motor process­ Study retention was also challenging. The 6-month training period
ing speed elements and is expected to be more reliable. was difficult for participants to sustain. Six months is longer than in most
Practice effects on neuropsychological tests can further confound studies and the most recent meta-analysis indicates that duration of the
estimates of cognitive improvement or decline over time. Therefore, we exercise training period does not moderate cognitive outcomes [22].
used tests that had alternative forms and excluded persons who had The requirement to perform follow-up fitness training was another
recently been exposed to the same tests. Moreover, we gave all partici­ barrier to retention. Our institution required EKG monitoring and
pants repeated practice on the two primary outcome measures before cardiologist supervision for maximum fitness testing. Cardiologist
the baseline assessment. By removing practice effects prior to baseline, availability was in short-supply and the University closed the only on-
we hoped to minimize the amount of observed improvement on these site testing laboratory forcing us to go off-site where cardiologist
measures between baseline, end of treatment and follow-up that is availability was non-existent. Alternatives to fitness testing are needed
attributable to practice effects. We recognize that repeated exposure to to objectively measure the impact of aerobic training on the theoretical
the outcomes could diminish the sensitivity of the measures for mechanism of action, cardiorespiratory fitness. Future investigators
capturing changes with the intervention. should consider body worn accelerometry as a proxy for cardiorespira­
While a double-blinded study was not possible, we designed all tory fitness [86]. Accelerometry could also serve as an objective
study-related communications to minimize between-group differences manipulation check, that is evidence of between group differences in
in outcome expectations. We highlighted the uncertainty in the scientific minutes of moderate to vigorous physical activity during the training
literature about the effects of exercise on cognition as well as evidence of period.
an association between light intensity physical activity and improved In conclusion, the efficacy of aerobic training to improve cognition
cognition [85] to describe both exercise conditions as potentially remains uncertain. The GET Smart trial was designed with high internal
beneficial for cognition. We also measured initial subjective outcome validity and to exceed prior scientific standards in order to answer this
expectations within each group after subjects began their assigned question. For feasibility reasons the trial evolved to have greater
treatment. These and other methodological considerations listed in external validity. The way this study was designed, its methods and how
Table X can be used to inform future research in this area. it ultimately was conducted can inform future research.
Other research designs could be considered in future studies, that
were not adopted for this trial. For example, we could have performed a Funding
dosing study in which exercise frequency, intensity or duration was
varied systematically to determine the relationship between dose and This investigation was supported by grant RG 4887 from the Na­
changes in cognition. Another approach would have been to add a no tional Multiple Sclerosis Society, U.S.A.. The funding source had no role
treatment or usual care arm to our trial to measure the potential effects in the decision to submit the article for publication.
of both treatment programs relative to what people would do normally.

9
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

Declaration of Competing Interest [19] J. Cotter, N. Muhlert, A. Talwar, K. Granger, Examining the effectiveness of
acetylcholinesterase inhibitors and stimulant-based medications for cognitive
dysfunction in multiple sclerosis: a systematic review and meta-analysis, Neurosci.
Dr. Bombardier-no disclosures. Dr. Motl—no disclosures. Dr. Bene­ Biobehav. Rev. 86 (Mar 2018) 99–107, https://doi.org/10.1016/j.
dict received honoraria, speaking, or consulting fees from Biogen, Cel­ neubiorev.2018.01.006.
gene, EMD Serono, Genentech, Medday, Novartis, and Roche; research [20] E.S. Gromisch, J.M. Fiszdon, M.M. Kurtz, The effects of cognitive-focused
interventions on cognition and psychological well-being in persons with multiple
support from Biogen, Genentech, and Novartis; and royalties from Psy­ sclerosis: a meta-analysis, Neuropsychol Rehabil 30 (4) (May 2020) 767–786,
chological Assessment Resources. Dr. Temkin—no disclosures. Dr. https://doi.org/10.1080/09602011.2018.1491408.
Qian—no disclosures. Dr. Kraft—no disclosures. [21] D.T. Turner, M.X. Hu, E. Generaal, D. Bos, M.K. Ikram, A. Heshmatollah, L. Fani, M.
A. Ikram, B. Penninx, P. Cuijpers, Physical exercise interventions targeting
cognitive functioning and the cognitive domains in nondementia samples: A
Appendix A. Supplementary data systematic review of meta-analyses, J. Geriatr. Psychiatry Neurol. (Apr 15 2020),
891988720915523, https://doi.org/10.1177/0891988720915523.
[22] R. Gharakhanlou, L. Wesselmann, A. Rademacher, A. Lampit, R. Negaresh,
Supplementary data to this article can be found online at https://doi. M. Kaviani, M. Oberste, R.W. Motl, B.M. Sandroff, J. Bansi, J.S. Baker, C. Heesen,
org/10.1016/j.cct.2021.106331. P. Zimmer, F. Javelle, Exercise training and cognitive performance in persons with
multiple sclerosis: A systematic review and multilevel meta-analysis of clinical
trials, Mult. Scler. (May 11 2020), 1352458520917935, https://doi.org/10.1177/
References 1352458520917935.
[23] D.M. Gronwall, Paced auditory serial-addition task: a measure of recovery from
[1] C.A. Dendrou, L. Fugger, M.A. Friese, Immunopathology of multiple sclerosis, Nat. concussion, Percept. Mot. Skills 44 (2) (Apr 1977) 367–373.
Rev. Immunol. 15 (9) (Sep 15 2015) 545–558, https://doi.org/10.1038/nri3871. [24] S.M. Rao, A manual for the brief, repeatable battery of neuropsychological tests in
[2] G.B.D.M.S. Collaborators, Global, regional, and national burden of multiple multiple sclerosis, National Multiple Sclerosis Society, 1991.
sclerosis 1990–2016: A systematic analysis for the global burden of disease study [25] R.H.B. Benedict, J.S. Fischer, C.J. Archibald, P.A. Arnett, W.W. Beatty, J. Bobholz,
2016, Lancet Neurol. 18 (3) (Mar 2019) 269–285, https://doi.org/10.1016/S1474- G. Chelune, J.D. Fisk, D.W. Langdon, L.S. Caruso, F.W. Foley, N.G. LaRocca,
4422(18)30443-5. L. Vowels, A. Weinstein, J. DeLuca, S.M. Rao, F.E. Munschauer, Minimal
[3] M.P. Amato, D. Langdon, X. Montalban, R.H. Benedict, J. Deluca, L.B. Krupp, A. neuropsychological assessment of ms patients: a consensus approach, Clin.
J. Thompson, G. Comi, Treatment of cognitive impairment in multiple sclerosis: Neuropsychol. 16 (2002) 381–397.
Position paper, J. Neurol. (Nov 23 2012), https://doi.org/10.1007/s00415-012- [26] L. Yaguez, K.N. Shaw, R. Morris, D. Matthews, The effects on cognitive functions of
6678-0. a movement-based intervention in patients with alzheimer’s type dementia: a pilot
[4] N.D. Chiaravalloti, J. DeLuca, Cognitive impairment in multiple sclerosis, Lancet study, Int J Geriatr Psychiatry 26 (2) (Feb 2011) 173–181, https://doi.org/
Neurol. 7 (12) (Dec 2008) 1139–1151, https://doi.org/10.1016/S1474-4422(08) 10.1002/gps.2510.
70259-X. [27] M.W. Voss, R.S. Prakash, K.I. Erickson, C. Basak, L. Chaddock, J.S. Kim, H. Alves,
[5] L.B. Krupp, C. Christodoulou, P. Melville, W.F. Scherl, L.Y. Pai, L.R. Muenz, D. He, S. Heo, A.N. Szabo, S.M. White, T.R. Wojcicki, E.L. Mailey, N. Gothe, E.A. Olson,
R.H. Benedict, A. Goodman, S. Rizvi, S.R. Schwid, B. Weinstock-Guttman, H. E. McAuley, A.F. Kramer, Plasticity of brain networks in a randomized intervention
J. Westervelt, H. Wishart, Multicenter randomized clinical trial of donepezil for trial of exercise training in older adults, Front. Aging Neurosci. 2 (2010), https://
memory impairment in multiple sclerosis, Neurology 76 (17) (Apr 26 2011) doi.org/10.3389/fnagi.2010.00032.
1500–1507, https://doi.org/10.1212/WNL.0b013e318218107a. [28] National Multiple Sclerosis Society, Stretching for people with ms: An illustrated
[6] S.A. Morrow, T. Kaushik, P. Zarevics, D. Erlanger, M.F. Bear, F.E. Munschauer, R. manual, NMSS, New York, 2004.
H. Benedict, The effects of l-amphetamine sulfate on cognition in ms patients: [29] A. Smith, Symbol digit modalities test: Manual, Western Psychol. Serv., Los
results of a randomized controlled trial, J Neurol 256 (7) (Jul 2009) 1095–1102, Angeles, 1982.
https://doi.org/10.1007/s00415-009-5074-x. [30] L.D. Baker, L.L. Frank, K. Foster-Schubert, P.S. Green, C.W. Wilkinson,
[7] C.T. Bever Jr., P.A. Anderson, J. Leslie, H.S. Panitch, S. Dhib-Jalbut, O.A. Khan, A. McTiernan, S.R. Plymate, M.A. Fishel, G.S. Watson, B.A. Cholerton, G.
R. Milo, J.R. Hebel, K.L. Conway, E. Katz, K.P. Johnson, Treatment with oral 3,4 E. Duncan, P.D. Mehta, S. Craft, Effects of aerobic exercise on mild cognitive
diaminopyridine improves leg strength in multiple sclerosis patients: results of a impairment: A controlled trial, Arch. Neurol. 67 (1) (Jan 2010) 71–79, 67/1/71
randomized, double-blind, placebo-controlled, crossover trial, Neurology 47 (6) [pii], https://doi.org/10.1001/archneurol.2009.307.
(Dec 1996) 1457–1462. [31] R.H. Benedict, D. Cookfair, R. Gavett, M. Gunther, F. Munschauer, N. Garg,
[8] P.M. Rossini, P. Pasqualetti, C. Pozzilli, M.G. Grasso, E. Millefiorini, A. Graceffa, G. B. Weinstock-Guttman, Validity of the minimal assessment of cognitive function in
A. Carlesimo, G. Zibellini, C. Caltagirone, Fatigue in progressive multiple sclerosis: multiple sclerosis (macfims), J. Int. Neuropsychol. Soc. 12 (4) (Jul 2006) 549–558.
results of a randomized, double-blind, placebo-controlled, crossover trial of oral 4- [32] C.H. Polman, S.C. Reingold, G. Edan, M. Filippi, H.P. Hartung, L. Kappos, F.
aminopyridine, Mult. Scler. 7 (6) (Dec 2001) 354–358. D. Lublin, L.M. Metz, H.F. McFarland, P.W. O’Connor, M. Sandberg-Wollheim, A.
[9] A.R. O’Brien, N. Chiaravalloti, Y. Goverover, J. Deluca, Evidenced-based cognitive J. Thompson, B.G. Weinshenker, J.S. Wolinsky, Diagnostic criteria for multiple
rehabilitation for persons with multiple sclerosis: A review of the literature, Arch. sclerosis: 2005 revisions to the “Mcdonald criteria”, Ann. Neurol. 58 (6) (Dec 2005)
Phys. Med. Rehabil. 89 (4) (Apr 2008) 761–769. S0003-9993(08)00003–8 [pii], htt 840–846, https://doi.org/10.1002/ana.20703.
ps://doi.org/10.1016/j.apmr.2007.10.019. [33] L.A. Pilutti, M.E. Platta, R.W. Motl, A.E. Latimer-Cheung, The safety of exercise
[10] C.H. Hillman, K.I. Erickson, A.F. Kramer, Be smart, exercise your heart: exercise training in multiple sclerosis: A systematic review, J. Neurol. Sci. 343 (1–2) (Aug
effects on brain and cognition, Nat. Rev. Neurosci. 9 (1) (Jan 2008) 58–65, https:// 15 2014) 3–7, https://doi.org/10.1016/j.jns.2014.05.016.
doi.org/10.1038/nrn2298. [34] J.F. Kurtzke, Rating neurologic impairment in multiple sclerosis: an expanded
[11] M. Angevaren, G. Aufdemkampe, H.J. Verhaar, A. Aleman, L. Vanhees, Physical disability status scale (edss), Neurology. 33 (11) (Nov 1983) 1444–1452.
activity and enhanced fitness to improve cognitive function in older people without [35] S. Thomas, American college of sports medicine guidelines for exercise testing and
known cognitive impairment, Cochrane Database Syst. Rev. 3 (2008), CD005381, prescription (Baltimore), 2006.
https://doi.org/10.1002/14651858.CD005381.pub3. [36] R.H. Benedict, J.H. Bobholz, Multiple sclerosis, Semin. Neurol. 27 (1) (Feb 2007)
[12] S. Colcombe, A.F. Kramer, Fitness effects on the cognitive function of older adults: 78–85, https://doi.org/10.1055/s-2006-956758.
a meta-analytic study, Psychol. Sci. 14 (2) (Mar 2003) 125–130. [37] K. Kroenke, R.L. Spitzer, J.B. Williams, The phq-9: validity of a brief depression
[13] A.F. Kramer, K.I. Erickson, Capitalizing on cortical plasticity: influence of physical severity measure, J. Gen. Intern. Med. 16 (9) (Sep 2001) 606–613.
activity on cognition and brain function, TRENDS Cog Sci 11 (8) (2007) 342–348. [38] I. Herrera-Guzman, E. Gudayol-Ferre, J.E. Herrera-Abarca, D. Herrera-Guzman,
[14] A.F. Kramer, K.I. Erickson, S.J. Colcombe, Exercise, cognition, and the aging brain, P. Montelongo-Pedraza, F. Padros Blazquez, M. Pero-Cebollero, J. Guardia-Olmos,
J. Appl. Physiol. 101 (4) (Oct 2006) 1237–1242, 00500.2006 [pii], https://doi. Major depressive disorder in recovery and neuropsychological functioning: effects
org/10.1152/japplphysiol.00500.2006. of selective serotonin reuptake inhibitor and dual inhibitor depression treatments
[15] P.J. Smith, J.A. Blumenthal, B.M. Hoffman, H. Cooper, T.A. Strauman, K. Welsh- on residual cognitive deficits in patients with major depressive disorder in
Bohmer, J.N. Browndyke, A. Sherwood, Aerobic exercise and neurocognitive recovery, J. Affect. Disord. 123 (1–3) (Jun 2010) 341–350, https://doi.org/
performance: a meta-analytic review of randomized controlled trials, Psychosom. 10.1016/j.jad.2009.10.009.
Med. 72 (3) (Apr 2010) 239–252, https://doi.org/10.1097/ [39] K.A. Bradley, A.F. DeBenedetti, R.J. Volk, E.C. Williams, D. Frank, D.R. Kivlahan,
PSY.0b013e3181d14633. Audit-c as a brief screen for alcohol misuse in primary care, Alcohol. Clin. Exp. Res.
[16] R.W. Motl, L.A. Pilutti, The benefits of exercise training in multiple sclerosis, Nat. 31 (7) (Jul 2007) 1208–1217, https://doi.org/10.1111/j.1530-0277.2007.00403.
Rev. Neurol. (Jul 24 2012), https://doi.org/10.1038/nrneurol.2012.136. x.
[17] R.W. Motl, B.M. Sandroff, R.H. Benedict, Cognitive dysfunction and multiple [40] J.A. Ewing, Detecting alcoholism. The cage questionnaire, JAMA 252 (14) (Oct 12
sclerosis: developing a rationale for considering the efficacy of exercise training, 1984) 1905–1907.
Mult. Scler. 17 (9) (Sep 2011) 1034–1040, https://doi.org/10.1177/ [41] WHO Assist Working Group, The alcohol, smoking and substance involvement
1352458511409612. screening test (assist): development, reliability and feasibility, Addiction. 97 (9)
[18] N.C. Landmeyer, P.C. Burkner, H. Wiendl, T. Ruck, H.P. Hartung, H. Holling, S. (Sep 2002) 1183–1194 (185 [pii]).
G. Meuth, A. Johnen, Disease-modifying treatments and cognition in relapsing- [42] A.M. Schreiner, M.E. Dunn, Residual effects of cannabis use on neurocognitive
remitting multiple sclerosis: A meta-analysis, Neurology 94 (22) (Jun 2 2020) performance after prolonged abstinence: a meta-analysis, Exp. Clin.
e2373–e2383, https://doi.org/10.1212/WNL.0000000000009522. Psychopharmacol. 20 (5) (Oct 2012) 420–429, https://doi.org/10.1037/
a0029117.

10
C.H. Bombardier et al. Contemporary Clinical Trials 104 (2021) 106331

[43] Consensus G. Goldman, The goldman consensus statement on depression in [65] S.A. Morrow, A. Drake, R. Zivadinov, F. Munschauer, B. Weinstock-Guttman, R.
multiple sclerosis, Mult. Scler. 11 (3) (Jun 2005) 328–337, https://doi.org/ H. Benedict, Predicting loss of employment over three years in multiple sclerosis:
10.1191/1352458505ms1162oa. clinically meaningful cognitive decline, Clin. Neuropsychol. 24 (7) (Oct 2010)
[44] D.M.A. Gronwall, Paced auditory serial addition task: a measure of recovery from 1131–1145, 926682619 [pii], https://doi.org/10.1080/13854046.2010.511272.
concussion, Percept. Mot. Skills 1977 (44) (1977) 367–373. [66] R.S. Prakash, E.M. Snook, K.I. Erickson, S.J. Colcombe, M.W. Voss, R.W. Motl, A.
[45] R.H. Benedict, A. Smerbeck, R. Parikh, J. Rodgers, D. Cadavid, D. Erlanger, F. Kramer, Cardiorespiratory fitness: A predictor of cortical plasticity in multiple
Reliability and equivalence of alternate forms for the symbol digit modalities test: sclerosis, Neuroimage 34 (3) (Feb 1 2007) 1238–1244. S1053-8119(06)01018–4
Implications for multiple sclerosis clinical trials, Mult. Scler. (Jan 25 2012), [pii], https://doi.org/10.1016/j.neuroimage.2006.10.003.
https://doi.org/10.1177/1352458511435717. [67] J.F. Sumowski, G.R. Wylie, J. Deluca, N. Chiaravalloti, Intellectual enrichment is
[46] E. McAuley, R.W. Motl, K.S. Morris, L. Hu, S.E. Doerksen, S. Elavsky, J. linked to cerebral efficiency in multiple sclerosis: functional magnetic resonance
F. Konopack, Enhancing physical activity adherence and well-being in multiple imaging evidence for cognitive reserve, Brain J. Neurol. 133 (Pt 2) (Feb 2010)
sclerosis: a randomised controlled trial, Mult. Scler. 13 (5) (Jun 2007) 652–659, 362–374, https://doi.org/10.1093/brain/awp307.
1352458506072188 [pii], https://doi.org/10.1177/1352458506072188. [68] J.S. Fischer, R.A. Rudick, G.R. Cutter, S.C. Reingold, The multiple sclerosis
[47] R.W. Motl, D. Dlugonski, Y. Suh, M. Weikert, B. Fernhall, M. Goldman, functional composite measure (msfc): An integrated approach to ms clinical
Accelerometry and its association with objective markers of walking limitations in outcome assessment. National ms society clinical outcomes assessment task force,
ambulatory adults with multiple sclerosis, Arch. Phys. Med. Rehabil. 91 (12) (Dec Mult. Scler. 1999 (5) (Aug 1999) 244–250.
2010) 1942–1947. S0003-9993(10)00715-X [pii], https://doi.org/10.1016/j. [69] D.C. Delis, E. Kaplan, J.H. Kramer, Delis-kaplan executive function system, The
apmr.2010.08.011. Psychological Corporation, San Antonio, TX, 2001.
[48] G. Godin, R.J. Shephard, A simple method to assess exercise behavior in the [70] D.C. Delis, J.H. Kramer, E. Kaplan, J. Holdnack, Reliability and validity of the delis-
community, Can. J. Appl. Sport Sci. 10 (1985) 141–146. Kaplan executive function system: an update, J. Int. Neuropsychol. Soc. 10 (2) (Mar
[49] J.L. Gosney, J.A. Scott, E.M. Snook, R.W. Motl, Physical activity and multiple 2004) 301–303, https://doi.org/10.1017/S1355617704102191
sclerosis: validity of self-report and objective measures, Fam. Community Health. (S1355617704102191 [pii).
30 (2) (Apr-Jun 2007) 144–150. [71] D.C. Delis, J.H. Kramer, E. Kaplan, B.A. Ober, California verbal learning test
[50] R.W. Motl, E. McAuley, E.M. Snook, J.A. Scott, Validity of physical activity manual: Second edition, adult version, Psychological Corporation, San Antonio,
measures in ambulatory individuals with multiple sclerosis, Disabil. Rehabil. 28 TX, 2000.
(18) (Sep 30 2006) 1151–1156. Q77233344868T520 [pii], https://doi.org/10. [72] R.H.B. Benedict, D. Schretlen, L. Groninger, M. Dobraski, B. Shpritz, Revision of the
1080/09638280600551476. brief visuospatial memory test: studies of normal performance, reliability, and
[51] American College of Sports Medicine, ACSM’s guidelines for exercise testing and validity, Psychol. Assess. 1996 (8) (1996) 145–153.
prescription, 6th ed., Williams and Wilkins, Baltimore, 2000. [73] A.L. Benton, A.B. Sivan, K. Hamsher, N.R. Varney, O. Spreen, Contributions to
[52] R.W. Motl, J.L. Gosney, Effect of exercise training on quality of life in multiple neuropsychological assessment, Second ed, Oxford University Press, New York,
sclerosis: a meta-analysis, Mult. Scler. 14 (1) (Jan 2008) 129–135, 1994.
1352458507080464 [pii], https://doi.org/10.1177/1352458507080464. [74] R.W. Motl, M. Goldman, Physical inactivity, neurological disability, and
[53] C.S. Holt, R.G. Heimberg, The reaction to treatment questionnaire: measuring cardiorespiratory fitness in multiple sclerosis, Acta Neurol. Scand. 123 (2) (Feb
treatment credibility and outcome expectancies, Behav. Ther. 13 (1990) 213–214 2011) 98–104, https://doi.org/10.1111/j.1600-0404.2010.01361.x.
(222). [75] R.S. Prakash, E.M. Snook, R.W. Motl, A.F. Kramer, Aerobic fitness is associated
[54] A. Bandura, Health promotion by social cognitive means, Health Educ. Behav. 31 with gray matter volume and white matter integrity in multiple sclerosis, Brain
(2) (Apr 2004) 143–164, https://doi.org/10.1177/1090198104263660. Res. 1341 (Jun 23 2010) 41–51. S0006-8993(09)01276–1 [pii], https://doi.org/10
[55] K.I. Erickson, M.W. Voss, R.S. Prakash, C. Basak, A. Szabo, L. Chaddock, J.S. Kim, .1016/j.brainres.2009.06.063.
S. Heo, H. Alves, S.M. White, T.R. Wojcicki, E. Mailey, V.J. Vieira, S.A. Martin, B. [76] R.H.B. Benedict, D. Cox, L.L. Thompson, F.W. Foley, B. Weinstock-Guttman,
D. Pence, J.A. Woods, E. McAuley, A.F. Kramer, Exercise training increases size of F. Munschauer, Reliable screening for neuropsychological impairment in ms, Mult.
hippocampus and improves memory, Proc. Natl. Acad. Sci. U. S. A. 108 (7) (Feb 15 Scler. 2004 (10) (2004) 675–678.
2011) 3017–3022, https://doi.org/10.1073/pnas.1015950108. [77] R.W. Motl, Y. Suh, M. Weikert, Symptom cluster and quality of life in multiple
[56] D.E. Barnes, W. Santos-Modesitt, G. Poelke, A.F. Kramer, C. Castro, L.E. Middleton, sclerosis, J. Pain Symptom Manag. 39 (6) (Jun 2010) 1025–1032, https://doi.org/
K. Yaffe, The mental activity and exercise (max) trial: a randomized controlled trial 10.1016/j.jpainsymman.2009.11.312.
to enhance cognitive function in older adults, JAMA Intern Med. (Apr 1 2013) 1–8, [78] D.J. Buysse, C.F. Reynolds 3rd, T.H. Monk, S.R. Berman, D.J. Kupfer, The
https://doi.org/10.1001/jamainternmed.2013.189. Pittsburgh sleep quality index: a new instrument for psychiatric practice and
[57] L.J. Beglinger, B. Gaydos, O. Tangphao-Daniels, K. Duff, D.A. Kareken, J. Crawford, research, Psychiatry Res. 28 (2) (May 1989) 193–213.
P.S. Fastenau, E.R. Siemers, Practice effects and the use of alternate forms in serial [79] L.B. Krupp, N.G. LaRocca, J. Muir-Nash, A.D. Steinberg, The fatigue severity scale.
neuropsychological testing, Arch. Clin. Neuropsychol. 20 (4) (Jun 2005) 517–529, Application to patients with multiple sclerosis and systemic lupus erythematosus,
https://doi.org/10.1016/j.acn.2004.12.003. Arch. Neurol. 46 (10) (Oct 1989) 1121–1123.
[58] S.M. Rao, A Manual for the Brief, Repeatable Battery of Neuropsychological Tests [80] D. Watson, L.A. Clark, A. Tellegen, Development and validation of brief measures
in Multiple Sclerosis, National Multiple Sclerosis Society, New York, NY, 1991. of positive and negative affect: the panas scales, J. Pers. Soc. Psychol. 54 (6) (Jun
[59] A.S. Drake, B. Weinstock-Guttman, S.A. Morrow, D. Hojnacki, F.E. Munschauer, R. 1988) 1063–1070.
H. Benedict, Psychometrics and normative data for the multiple sclerosis functional [81] C.L. Craig, A.L. Marshall, M. Sjostrom, A.E. Bauman, M.L. Booth, B.E. Ainsworth,
composite: replacing the pasat with the symbol digit modalities test, Mult. Scler. 16 M. Pratt, U. Ekelund, A. Yngve, J.F. Sallis, P. Oja, International physical activity
(2) (Feb 2010) 228–237, 1352458509354552 [pii], https://doi.org/10.1177/1 questionnaire: 12-country reliability and validity, Med. Sci. Sports Exerc. 35 (8)
352458509354552. (Aug 2003) 1381–1395, https://doi.org/10.1249/01.MSS.0000078924.61453.FB.
[60] J.F. Sumowski, N. Chiaravalloti, G. Wylie, J. Deluca, Cognitive reserve moderates [82] A. Diamond, D.S. Ling, Aerobic-exercise and resistance-training interventions have
the negative effect of brain atrophy on cognitive efficiency in multiple sclerosis, been among the least effective ways to improve executive functions of any method
J. Int. Neuropsychol. Soc. 15 (4) (Jul 2009) 606–612, https://doi.org/10.1017/ tried thus far, Dev. Cogn. Neurosci. 37 (Jun 2019) 100572, https://doi.org/
S1355617709090912. 10.1016/j.dcn.2018.05.001.
[61] M.P. Amato, V. Zipoli, E. Portaccio, Multiple sclerosis-related cognitive changes: A [83] A. Liberati, D.G. Altman, J. Tetzlaff, C. Mulrow, P.C. Gotzsche, J.P. Ioannidis,
review of cross-sectional and longitudinal studies, J. Neurol. Sci. 245 (1–2) (Jun 15 M. Clarke, P.J. Devereaux, J. Kleijnen, D. Moher, The prisma statement for
2006) 41–46. S0022-510X(06)00097–9 [pii], https://doi.org/10.1016/j.jns.200 reporting systematic reviews and meta-analyses of studies that evaluate health care
5.08.019. interventions: Explanation and elaboration, Ann. Intern. Med. 151 (4) (Aug 18
[62] R.H. Benedict, J.M. Bruce, M.G. Dwyer, N. Abdelrahman, S. Hussein, B. Weinstock- 2009) W65–W94, https://doi.org/10.7326/0003-4819-151-4-200908180–00136.
Guttman, N. Garg, F. Munschauer, R. Zivadinov, Neocortical atrophy, third [84] N.A. Smart, M. Waldron, H. Ismail, F. Giallauria, C. Vigorito, V. Cornelissen,
ventricular width, and cognitive dysfunction in multiple sclerosis, Arch. Neurol. 63 G. Dieberg, Validation of a new tool for the assessment of study quality and
(9) (Sep 2006) 1301–1306, 63/9/1301 [pii], https://doi.org/10.1001/archneur. reporting in exercise training studies: Testex, Int. J. Evid. Based Healthc. 13 (1)
63.9.1301. (Mar 2015) 9–18, https://doi.org/10.1097/XEB.0000000000000020.
[63] R.H. Benedict, B. Weinstock-Guttman, I. Fishman, J. Sharma, C.W. Tjoa, R. Bakshi, [85] S. Lee, A. Yuki, Y. Nishita, C. Tange, H. Kim, R. Kozakai, F. Ando, H. Shimokata,
Prediction of neuropsychological impairment in multiple sclerosis: comparison of Research relationship between light-intensity physical activity and cognitive
conventional magnetic resonance imaging measures of atrophy and lesion burden, function in a community-dwelling elderly population-an 8-year longitudinal study,
Arch. Neurol. 61 (2) (Feb 2004) 226–230, https://doi.org/10.1001/ J. Am. Geriatr. Soc. 61 (3) (Mar 2013) 452–453, https://doi.org/10.1111/
archneur.61.2.226 61/2/226 (pii). jgs.12119.
[64] C. Christodoulou, L.B. Krupp, Z. Liang, W. Huang, P. Melville, C. Roque, W. [86] R.W. Motl, B.M. Sandroff, L.A. Pilutti, R.E. Klaren, T. Baynard, B. Fernhall, Physical
F. Scherl, T. Morgan, MacAllister WS, L. Li, L.A. Tudorica, X. Li, P. Roche, activity, sedentary behavior, and aerobic capacity in persons with multiple
R. Peyster, Cognitive performance and mr markers of cerebral injury in cognitively sclerosis, J. Neurol. Sci. 372 (Jan 15 2017) 342–346, https://doi.org/10.1016/j.
impaired ms patients, Neurology 60 (11) (Jun 10 2003) 1793–1798. jns.2016.11.070.

11