Journal Pre-proof

Home-based exercise in autoimmune myasthenia gravis: a

randomized controlled trial

S Birnbaum , R Porcher , P Portero , B Clair , S Demeret ,

B Eymard , M Gargiulo , E Louët , S Berrih-Aknin , R Le Panse ,
P Aegerter , JY Hogrel , T Sharshar , the MGEX Study Group

PII: S0960-8966(21)00127-9
DOI: https://doi.org/10.1016/j.nmd.2021.05.002
Reference: NMD 3996

To appear in: Neuromuscular Disorders

Received date: 16 November 2020

Revised date: 7 May 2021
Accepted date: 11 May 2021

Please cite this article as: S Birnbaum , R Porcher , P Portero , B Clair , S Demeret , B Eymard ,
M Gargiulo , E Louët , S Berrih-Aknin , R Le Panse , P Aegerter , JY Hogrel , T Sharshar , the
MGEX Study Group, Home-based exercise in autoimmune myasthenia gravis: a randomized con-
trolled trial, Neuromuscular Disorders (2021), doi: https://doi.org/10.1016/j.nmd.2021.05.002

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2021 Published by Elsevier B.V.

Highlights
 Quality of life is reduced in myasthenia gravis
 The role of exercise therapy to improve quality of life has not been explored
 This is the first prospective randomized controlled trial of home-based, unsupervised
exercise therapy in MG
 Home-based exercise therapy was well-tolerated but did not improve quality of life
Title: Home-based exercise in autoimmune myasthenia gravis: a randomized controlled trial
Authors : Birnbaum S, Porcher R, Portero P, Clair B, Demeret S, Eymard B, Gargiulo M,

Louët E, Berrih-Aknin S, Le Panse R, Aegerter P, Hogrel JY, Sharshar T* and the MGEX

Study Group.

Abstract

The tolerance of exercise and its effects on quality of life in myasthenia gravis are not

currently backed up by strong evidence. The aim of this study was to determine whether

exercise as an adjunct therapy is well tolerated and can improve health-related quality of life

(HRQoL) in stabilized, generalized autoimmune myasthenia gravis (gMG).

We conducted a parallel-group, multi-center prospective RCT using computer-generated

block randomization. Adults with stabilized, gMG, and no contra-indication to exercise, were

eligible. Participants received usual care alone or usual care and exercise. The exercise

intervention consisted of 3-weekly 40-minute sessions of an unsupervised, moderate-intensity

home rowing program over 3 months. The primary endpoint was the change in HRQoL from

randomization to post-intervention. Assessor-blinded secondary endpoints were exercise

tolerance and effects on clinical, psychological and immunological status.

Of 138 patients screened between October 2014 and July 2017, 45 were randomly assigned to

exercise (n=23) or usual care (n=20). Although exercise was well tolerated, the intention-to-

treat analysis revealed no evidence of improved HRQoL compared to usual care (MGQOL-

15-F; mean adjusted between-groups difference of -0.8 points, 95%CI -5.4 to 3.7). Two

patients hospitalized for MG exacerbation were from the usual care group.

Trial registration: ClinicalTrials.gov identifier NCT02066519

Keywords: Myasthenia gravis, Exercise, randomized controlled trial, Autoimmune disease,

Quality of life

2
Introduction

Myasthenia gravis (MG), a rare autoimmune disorder of the neuromuscular junction is

characterized by fluctuating weakness involving variable combinations of ocular, bulbar,

respiratory and limb muscles. Treatment of MG primarily relies on immunosuppressive

therapies, notably the association of corticosteroids with non-steroidal immunosuppressive

agents [1]. These therapeutic strategies aim to reduce symptoms, improve functional status

and health-related quality of life (HRQoL); three dimensions which can be assessed with

specific scores [2]. Despite appropriate therapeutic management, individuals with MG still

have reduced HRQoL which is not improving despite therapeutic advances [3]. Therefore,

adjuvant therapy for improving HRQoL in MG is warranted. Whilst rest is indicated during

an exacerbation, exercise may be a relevant option in stabilized MG as it targets muscle-

related symptoms, i.e. weakness and fatigability.

Exercise therapy has been shown to improve strength and functional capacity in various

neuromuscular diseases such as facioscapulohumeral disease [4] and other muscular

dystrophies [5]. There are physiological and clinical arguments for its therapeutic use in MG.

It induces remodeling of the neuromuscular junction in experimental models [6] and

interventional human studies have shown improved lower limb strength [7, 8], walking

capacity and overall function [9-11] and, exercise self-efficacy [11]. However, these studies

were non-randomized [7-9, 11] and included small numbers [7-11].

In addition, exercise has positive psychological and immunological effects, which are of

potential benefit to individuals with MG. Symptoms of depression and anxiety have been

shown to decrease with exercise in chronic illness [12-14]. Favorable immunomodulatory

effects have been induced by the release of myokines during repetitive muscle contractions

[15, 16]. Because of its pluripotent effects, it is conceivable that exercise could improve

HRQoL in MG. The objective of this study was to evaluate whether an unsupervised home

3
exercise program (ExP) could improve MG-related HRQoL of individuals with stabilized and

gMG, with reduced HRQoL. Secondary objectives were to assess exercise tolerance and the

impact of exercise on physical, psychological and immune status.

Methods

Study design

We conducted a two-arm, parallel group, multi-center randomized controlled trial (RCT)

comparing a home-exercise program to usual care (UC). Full details of the study rationale,

design, methods and statistical analysis have been previously published [17]. The overall

study duration for an individual participant was 36 weeks. Total intervention time was 12

weeks beginning at randomization at month 3 (M3), ending at month 6 (M6). This was

followed by a 3-month follow-up period (with no intervention), ending at month 9 (M9).

Writing of this manuscript follows the CONSORT guidelines [18]. Ethics approval to carry

out the study on human subjects was granted by the French regulatory board (Comité de

Protection des Personnes Ile de France XI under the CPP number #13064 on 13/12/2013 and

authorized by the Agence National de Sécurité du Médicament et des produits de santé on

11/10/ 2013). The trial was conducted in accordance with the Helsinki Declaration and was

prospectively registered in ClinicalTrials.gov (NCT02066519).

Participants and setting

Three specialized neuromuscular centers in France screened patients with autoimmune MG

during regular outpatient consultations and study information was diffused to MG-patient

groups. Patients aged 18-70 years with a confirmed diagnosis of mild-moderately-severe,

gMG (class II-III according to the Myasthenia Gravis Foundation of America (MGFA)

classification) were eligible. Six months of MG disease stability, determined by the treating

neurologist, and reduced HRQoL (MGQOL-15-F score ≥ 15), was required for inclusion.

4
Exclusion criteria included any contra-indication to exercise which precluded participation in

the experimental arm, other neuromuscular pathology, disabling rheumatological or

orthopedic condition, chronic pain, recent hospitalization for a serious medical or surgical

condition, anemia (hematocrit < 30%), pregnancy or severe cognitive impairment

necessitating specific protection (Table S1). Written, informed consent was obtained from all

participants prior to enrolment.

Randomization, allocation concealment, blinding

Following the 3-month evaluation, if MG stability was confirmed, consecutive participants

were randomized to either the intervention (UC + exercise) or control (UC) group with a 1:1

allocation ratio as per a computer-generated randomization schema stratified by center using

permuted blocks of randomly varying sizes. To ensure concealment, block sizes were not

disclosed. The randomization list was constructed prior to the beginning of the study by a

statistician with no clinical involvement in the study. Group allocation was revealed

exclusively to the non-blinded physiotherapist at the moment of randomization (M3), via a

software (CleanWeb), and was concealed until the moment of group allocation. Investigators,

evaluators and study personnel were all masked to group assignment for the duration of the

study. Only the physiotherapist performing the neuromuscular evaluations and managing the

training program was aware of group allocation.

Intervention

All participants received UC and evaluations as per the study protocol. The control group

were instructed to maintain their regular physical activity without starting a new exercise

program or increasing baseline activity. Baseline activity was measured over one week, prior

to randomization, using a tri-axial accelerometer (Movemonitor, McRoberts, Netherlands).

The intervention group was invited to participate in a 12-week home-based aerobic exercise

program (ExP) using a rowing ergometer (Concept2™), Table S2. The ExP consisted of three

5
40-minute sessions per week, i.e. a total of 36 sessions over the intervention period. A chest

heart rate (HR) monitor (Garmin®) was worn by participants and training HR was visualized

on a screen for direct feedback. Each 40-minute moderate-intensity rowing session consisted

of a 10 minute warm-up to reach each participants’ individual target HR (i.e. 70% of their

maximal HR [HRmax], using the equation 220-age as their HRmax), followed by a 20-minute

plateau phase of constant aerobic activity, rowing at their individual target HR, followed by a

5-minute power interval phase (5 sets of ten consecutive pulls at maximum effort performed

at the beginning of each minute, followed by regular intensity strokes for the remainder of

each minute) and finally, a 5-minute active cool-down period where rowing was continued at

a slow pace. Data (rowing distance and session duration) were recorded on an individual

logcard for each training session. Participants were advised to organize their training sessions

on alternate days. The initial 1-3 sessions took place at a different site (Rothschild Hospital,

Paris) to study visits, to maintain investigator blinding. These individualized sessions were

carried out by a trained physiotherapist to ensure feasibility, to teach participants the ExP,

how to use the HR monitor, the rowing machine (Concept2™) and how to perform the rowing

movement. Once the physiotherapist and participant were satisfied that the ExP could be

performed unsupervised at home, the rowing machine was then delivered to the participant’s

home for the duration of the intervention period (3 months). The Concept 2™ rowing

machine was selected as it is a stationary ergometer where one is seated and simultaneously

exercises upper and lower limbs and the trunk thus targeting the multiple muscles which can

be affected in MG. Adherence was defined as having completed at least 20 (frequency) 30-

minute (duration) sessions.

Outcomes

The primary endpoint was the adjusted mean between-group difference in MG-specific

HRQoL, assessed with the French version of the Myasthenia Gravis Quality of Life

6
questionnaire (total score) [19] (MGQOL-15-F; range 0-60; higher scores represent lower

perceived HRQoL) between the ExP and UC group from M3 (baseline) to M6 (post-

intervention).

To assess tolerability of the ExP compared to UC, adverse events were recorded at each

monthly visit except serious adverse events such as MG crisis or MG exacerbation with

hospitalization which were notified immediately. MG crisis was defined as rapid clinical

deterioration necessitating either non-invasive or invasive mechanical ventilation. MG

exacerbation was defined as clinical deterioration with beyond usual impact on daily life, with

or without modification of medications. Tolerance was evaluated monthly by assessing

cardiorespiratory status (cardiac and thoracic auscultation, vital signs, electrocardiogram,

cardiothoracic pain, dyspnea, laboratory tests (serum hematology, creatine kinase),

muscular/articular pain (self-report), and medication dosage (corticosteroids and/or AChEi)

modification.

Prespecified secondary efficacy endpoints included the adjusted mean between-group

difference between the ExP compared to UC from M3 to M6 in: MG symptom severity

evaluated with the quantitative Myasthenia Muscle Score [20] (MMS; range 0-100; lower

scores represent greater symptom severity), impact of MG on daily function, evaluated with

the Myasthenia Gravis Activities of Daily Living scale [21] (MG-ADL; range 0-24, higher

scores represent greater impact of MG on function), physical capacity evaluated with the 6-

minute walk test (6MWT) for walking endurance [22, 23], maximal voluntary isometric

strength of elbow flexion and knee extension using a Biodex (Shirley, USA) and handgrip

strength using a MyoGrip and respiratory function (Forced Vital Capacity, maximal

inspiratory and expiratory pressures evaluated using a Vitalograph spirometer and a

MicroRPM). In addition, change in non-specific QoL (WHO-QOL Bref) [24, 25] and

psychological status; depression (Beck Depression Inventory) [26], anxiety (State Trait

7
Anxiety Inventory) [27] and self-esteem (Self-Esteem Inventory scale) [28] before and after

the ExP were assessed. Exploratory endpoints consisting of cytokine levels of immunological

markers, IL-6 and TNF-α were evaluated between M3 and M6.

Statistical analysis

Sample size calculations were described previously [17]. The primary analysis was conducted

in the intention-to-treat (ITT) population, where all randomized participants were analyzed in

the group they were allocated to irrespective of any protocol deviation. The primary outcome

was analyzed using analysis of covariance (ANCOVA), with changes in MGQOL-15-F as the

dependent variable, and randomization group, MGQOL-15-F score at baseline and center as

covariates. A similar model was used for secondary and exploratory outcomes defined as

changes from M3. For binary secondary outcomes, absolute risk difference and relative risks

were estimated. For the monthly MMS and MG-ADL scores, statistical tests were based on

linear mixed models on the ranked scores, with random subject intercept and slope, and a

separate fixed-effect for each time as well as interactions with the randomization arm from

M4 onwards, to force arms to be similar at randomization and to attribute any observed

difference to random variation. Magnitude of treatment effect was estimated by differences in

least-squares group means in mixed-effect models with bootstrap confidence interval.

Cytokines were analyzed after log-transformation and adjusted ratios of geometric means are

presented. For values under the lower limit of detection, ANCOVA was replaced by tobit

models, with the same formulation. For safety analyses, the numbers of participants with at

least one AE or SAE were compared by Fisher’s exact tests, and the number of AEs by

Poisson regression. Missing data were handled by multiple imputation by chained equations.

In addition to baseline subject characteristics, the imputation model considered height,

weight, MGQOL-15-F, MMS, MG-ADL, WHO-QOL, STAI, Beck, MINI, SEI, 6-minute

walking distance (6MWD), and adherence status in the intervention arm at M3, M6 and M9.

8
Imputations were performed separately in each randomization group as recommended [29].

Imputation of continuous variables was performed using predictive mean matching. Forty

imputed datasets were generated and analyzed. Results were then pooled. Physiological (e.g.

strength), biological (cytokines) and safety outcomes were not imputed. No allowance was

made for multiplicity for secondary endpoints. All tests were two-tailed with p values less

than 0.05 defined as significant. A compliers average causal effect (CACE) analysis was

carried out using an instrumental variable approach. Notification of AEs and visit tracking

was done electronically.

Results

Participants

Figure 1 presents the study flow diagram. Between October 3rd, 2014 and July 7th, 2017, 138

patients were screened, of which 45 were enrolled in the study with 23 being randomly

assigned to the exercise intervention and 20 to the UC control group. Forty-one participants

(91%) completed the study. Two participants withdrew before randomization. One participant

refused to undertake the ExP but remained in the study. Two participants (control group) were

lost to follow-up, one following randomization (M3) and the other in their last month of the

study (between M8 and M9). Baseline characteristics are presented in Tables 1-2.

Intervention adherence

Of the 23 participants randomized to the ExP, a mean of 24 (range 0-38) forty-minute

sessions were completed. Sixteen (70%) participants adhered to the ExP (Table S3). Reasons

for non-adherence were mainly work commitments. Reasons for missing occasional sessions

included: the flu, weekend away, menstrual pain/tiredness. Participant characteristics were

similar between adherers and non-adherers (Table S4).

Primary endpoint (Intention-to-treat)

9
There was an adjusted between-groups mean difference of -0.8 points (95%CI -5.4 to 3.7) on

the MGQOL-15-F score at the end of the intervention period (M6) and an adjusted between-

groups mean difference of 3.9 points (95%CI -1.7 to 9.5) at follow-up (M9) (Table 2),

indicating that there was no difference in MG-QOL-15-F score between the two groups at

either time point.

Exercise tolerance

Two participants from the control group necessitated hospitalization for an acute MG

exacerbation. No exacerbation was reported in the ExP group throughout the entire study

(Tables 3-4). There was no between-groups difference with regards to exercise tolerance

including cardiorespiratory indicators, muscular or articular pain (Table 3), nor concerning

non serious adverse events such as infection, allergic reaction, cephalalgia (Table 4). There

was no evidence of raised creatine phosphokinase levels with exercise (Table S5).

Secondary endpoints

There was a significant adjusted between-groups mean difference of -1.9 for the MG-ADL

(95%CI -3.0 to -0.9) and 27.7 m for the 6-minute walking distance (95%CI 7.2 to 48.1) in

favor of the ExP group at M6 but these differences were not maintained at M9 (Table 2).

Knee extension strength increased in the ExP group without reaching statistical significance

(8.8 Nm, %95 CI 0.1 to 17.5), Table 2. No between-groups difference was observed for the

MMS, elbow or grip strength, pulmonary function, general QoL (WHO-QOL BREF) or

psychological status (STAI, BDI-13, SEI) (Table 2, Table S6). Further, there was no between-

groups difference in prednisone dose -0.4 (95%CI -2.1 to 1.2) nor AChEi dose -0.1 (95%CI -

0.6 to 0.4) at M6, nor at M9 for prednisone -0.2 (95%CI -1.5 to 1.0) or, AChEi -0.0 (95%CI -

0.5 to 0.4).

Exploratory endpoints

10
There was no between-groups difference in adjusted ratios of geometric means from baseline

in cytokines IL-6; 1.09 (95%CI 0.97 to 1.23) nor TNF-α; 0.97 (95%CI 0.86 to 1.10), at M6,

nor at M9, IL-6; 1.07 (95%CI 0.95 to 1.21), TNF-α; 1.02 (95%CI 0.91 to 1.13).

Discussion

This randomized, clinical trial of individuals with mild-moderately severe, generalized and

stabilized MG, demonstrated that a 3-month unsupervised home exercise program did not

improve HRQoL (MGQOL-15-F) compared to UC. The exercise program was, however, well

tolerated, with improvement in the perceived impact of MG on daily life (MG-ADL score)

and walking capacity (6MWD), without a between-group difference in psychological or

immunological markers.

We believed that HRQoL would be an appropriate primary endpoint for measuring the

pluripotent effects of exercise on both physical and psychological domains. Quality of life

measures provide information with regards to how satisfied one is with their current level of

functioning compared to what they perceive to be possible or ideal [30]. However, self-

perceived HRQoL is a complex construct integrating multiple different dimensions and is

subject to a response shift phenomenon whereby one’s internal reference can change over

time or as a result of an intervention (herein exercise), or placebo [31]. We were not able to

assess to what extent this phenomenon occurred. It is possible that the absence of a significant

change in MGQOL-15-F was due to the intervention itself. The frequency, mode and duration

of the exercise program may not have been sufficient to shift this complex construct nor, to

improve psychological status. As reported elsewhere [17], we initially planned to include

patients with an MGQOL-15-F score above 30 in order to have a greater chance of detecting a

benefit from exercise. It appeared that for the majority of patients with an MGQOL-15-F

11
score > 30, MG was not stabilized, which we considered a contra-indication to the moderate-

intensity exercise intervention.

The improvement of walking endurance evaluated with the 6MWT and trend toward an

increase in knee extension strength indicates that our exercise program had an impact on

muscle performance, further confirmed by the CACE analyses (Table S8). Importantly, this

demonstrates a transfer from a rowing exercise to walking capacity, which we did not

specifically train. One may argue that an increase of 27.7m over six minutes is small however,

it is not only clinically relevant [32] but it should be noted that the distance walked in this

cohort was high prior to randomization (498.1±83.4m), making it more difficult to have such

a significant change [33]. Given that our exercise intervention was aerobic, these effects are

likely due to improvements in muscle metabolism and efficiency. Improvement in

cardiorespiratory fitness may also have occurred as we found significant cardiorespiratory

gains demonstrated by an increase in rowing distance for the same target heart rate.

Lower limb strength and functional gains (30 second chair stand test) have been demonstrated

in previous studies of exercise interventions in MG [7, 8]. We would like to emphasize that

any gain in strength for MG patients is important as weakness likely contributes to fatigue and

reduced daily function [34]. Consistent with our results, although the training programs in

these studies included upper limb training, no improvements were observed in the upper limb.

These studies were within-subject control (contralateral limb tested) [7] and repeated

measures design [8], each consisting of 11 participants. Thus it was important to confirm

findings in a methodologically rigorous randomized, controlled trial with a larger cohort.

Further, our exercise intervention was unsupervised and there was no specific strength

training as per the other two protocols [7, 8].

12
Tolerance and safety were thoroughly assessed throughout this study via clinical

examinations, vital parameters as well as biological indicators. One of the major findings in

this clinical trial is the tolerance of the exercise program. We did not observe any MG

deterioration nor increased muscular or articular pain or, cardiorespiratory symptoms

compared to UC. This is likely to be related to our choice of a moderate-intensity program

with specific training in using the rowing machine and the individualized nature of the

program. We are not able to determine to what extent the high tolerance option may have

compromised efficiency. Only a trial comparing an exercise program with various intensities

and durations could address this issue. However, good tolerance is a major argument for

recommending this type of exercise program in a larger population. The rowing machine is a

popular device for home exercise as it is simple to learn and perform.

When choosing an unsupervised, home-based intervention there is greater risk of poor

adherence than a supervised programme. Adherence should be taken into account when

generalization is considered. Seven (30%) participants were considered as non-adherers. Their

reported reasons were unrelated to the exercise program nor to their MG and baseline

characteristics between the two groups were similar (Table S4). The non-adherence in our

study is considered acceptable considering the intervention was home-based and unsupervised

and prior studies have shown similar compliance for exercise interventions [35]. Our

definition of non-adherence was arbitrary and the small number of participants limits

comparison between adherers and non-adherers, in determining the minimal exercise dosage

for efficacy. This RCT was designed with intention-to-treat (ITT) analyses however, CACE

analyses provides further insight into the effects of the exercise according to compliance. The

difference in the results of the CACE and ITT analysis suggests that it would be worth

optimizing adherence of the exercise intervention in future studies. This could possibly be

13
achieved with direct supervision or with remote monitoring using wearables, keeping in mind

that device adherence would also need to be taken into consideration.

This study presented some methodological limitations. Whilst our sample size could be

considered modest, there was an a priori calculated sample size, which is comparable to other

studies of exercise in rare disease [36, 37]. Despite the fact that MG is more prevalent in

women before the age of 50 (juvenile and early MG represents 86% of this cohort) [38], there

still remains disproportionately more women (40) than men (3) in this study, and all men were

randomly assigned to the control group. We note that this sex imbalance was also present in a

previous study suggesting possible selection bias which should be taken into account in future

studies of exercise interventions in MG [7]. We acknowledge that our results should be

confirmed in a larger population, notably to ensure the applicability and tolerance of our

exercise program particularly in men whom were largely lacking in this trial. Whilst it would

be of interest to assess whether the effects of exercise varies according to specific clinical or

immunological phenotypes of MG, our aim was to first evaluate the tolerance in all types of

MG. Due to the type of intervention, blinding of participants was not possible and some

improvement could be related to positive expectation bias particularly for those who were

motivated to perform exercise [39].

At this stage more research is required to answer questions concerning the type of exercise,

intensity, duration and frequency and the impact of each aspect on individuals and subgroups

of MG. It may be important to consider current lifestyle and physical activity levels for better

tailoring of an exercise intervention. In addition, we selected patients with stabilized MG.

Therefore, the feasibility, tolerance and effectiveness of exercise in non-stabilized MG will

need to be specifically assessed. Strengths of this study include the rigorous study design and

the comprehensive assessments of the functional, immunological and psychological impact of

14
exercise using validated measures. Participant retention throughout the study was good with

only two participants (control group) withdrawing.

Conclusion

This randomized clinical trial with intention-to-treat analyses is the first to demonstrate that a

3-month, home-based unsupervised exercise program is well tolerated and does not contribute

to exacerbations in adults with stabilized, gMG. This is a major result as benefit and risk of

exercise is a major concern for individuals with MG and for which no specific

recommendations are currently available. Further research is necessary to evaluate different

dosages of exercise and their effects, in this population.

15
Figure legend

Figure 1. Participant flow CONSORT diagram. ITT: intention-to-treat.

Table legends

Table 1. Demographic, disease characteristics, medical history and activity of

participants at baseline i.e. all untraineda.

a
Data are mean (SD) and range or n (%) or median [IQR].

SD, standard deviation; IQR, interquartile range; BMI, body mass index; MGFA, Myasthenia

Gravis Foundation of America classification; AChR, muscle nicotinic acetylcholine receptor;

MuSK, muscle-specific tyrosine kinase; AchE, acetylcholinesterase.

Table 2. Intention-to-treat analysis of the primary outcome. Analyses of secondary

outcomesa.
a
Analyses of imputed data of primary efficacy endpoint (MGQOL-15-F: French version of the

15-item MG-specific quality of life scale). Secondary efficacy endpoints; symptom severity

(Myasthenia Muscle Score), limb strength, walking capacity (six-minute walking distance),

pulmonary function (Forced Vital Capacity, Maximal inspiratory pressure, Maximal

expiratory pressure). %pred, % predicted; SD, standard deviation; MD, mean difference; 95%

CI: 95% Confidence Interval; M3, Randomization, at month 3; M6, end of intervention

period, at month 6; M9, end of study at month 9. *Adjusted mean difference estimated by a

latent class mixed-effect models, and 95% confidence interval by bootstrap; **p-value

obtained by rank ANCOVA.

Table 3. Analysis of tolerance to exercise based on neurologist assessment at each

monthly visit.

16
RD: absolute risk difference; RR: relative risk; 95% CI: 95% Confidence Interval; M6, month

6; M9, month 9.

Table 4. Patient-reported adverse events throughout the entire study as reported by the

participant at each monthly visit. Serious adverse events reported at any time

throughout the study and declared to the sponsor.

62 adverse events were declared for 30 patients, among which 2 were considered a serious

adverse event. P values based on Fisher’s exact tests except * where P value is based on

Poisson regression. n, number of participants.

Funding

This study was sponsored by the Assistance Publique – Hôpitaux de Paris « Direction de la

Recherche Clinique et de l’Innovation”. This study was funded through a grant from the

Programme Hospitalier de Recherche Clinique - PHRC régional 2012 (Ministère de la Santé

AOR12149), in collaboration with the Institute of Myology. The funders had no role in the

design of the study; nor data analysis, data interpretation, nor in the writing of this

manuscript.

Disclosure of conflicts of interest

S. Birnbaum, R. Porcher, B. Clair, S. Demeret, B. Eymard, M. Gargiulo, E. Louët, S. Berrih-

Aknin, R. Le Panse, P. Aegerter, J.Y. Hogrel, T. Sharshar report no disclosures relevant to the

manuscript.

Data availability

Individual de-identified participant data may be shared for research/academic purposes once

all secondary analyses have been completed. Requests for access to data must be made to the

17
corresponding author and sponsor and should be accompanied by a methodologically sound

proposal. A signed data sharing agreement is required before access can be provided.

Authors' contributions see Table S9 in supplementary information

Acknowledgements

We sincerely thank all participants. We thank the APHP for the grant through the French

Ministry of Health grant program for Clinical Research (AOR12149). We also thank the

Institute of Myology for their financial contribution. We acknowledge the assistance of

Sophie Hue, Mathieu Surenaud, IMRB luminex platform, Creteil. We would also like to

acknowledge the nurses and administration staff who played a crucial role in this study.

MG Study group collaborators to be individually indexed in Pubmed: Sylvie Azerad,

Guillaume Bassez, Anthony Behin, Sonia Berrih-Aknin, Francis Bolgert, Nawal Derridj Ait-

Younes, Yasmine Domingo, Mélinée Frenkian, Diane Friedman, Asmaa Jobic, Pascal

Laforêt, Isabelle Ledoux, Judith Mendelson, Sandra Misdrahi, Cécilia Orblin Bedos,

Benjamin Rohaut, Jacques Ropers, Elodie Soler, Philippe Thoumie, Frédérique Truffault,

Nicolas Weiss, Linda William.

Supplementary Information

Changes to trial design

Supplementary data Table S1. Study eligibility criteria

Supplementary data Table S2. Description of exercise intervention according to Slade et al.

2016

Supplementary data Table S3. Adherence to exercise intervention

Supplementary data Table S4: Baseline characteristics associated with adherencea.

Supplementary data Table S5. Maximum creatine phosphokinase values between month 3

(randomization) and month 6 (end of exercise intervention).

18
Supplementary data Table S6. Analyses of secondary outcomes quality of life and
psychological status.

Supplementary data Table S7. Analysis of primary endpoint according to adherence status.

Supplementary data Table S8. Analysis of strength and walking capacity according to

adherence status.

Supplementary data Table S9. Author contributions

CONSORT checklist

Statistical Analysis Plan

Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal
relationships that could have appeared to influence the work reported in this paper.

References

[1] Sanders DB, Wolfe GI, Benatar M, Evoli A, Gilhus NE, Illa I, et al. International consensus

guidance for management of myasthenia gravis: Executive summary. Neurology

2016;87:419-25.

[2] Barnett C, Herbelin L, Dimachkie MM, Barohn RJ. Measuring Clinical Treatment Response in

Myasthenia Gravis. Neurologic clinics 2018;36:339-353.

[3] Boldingh MI, Dekker L, Maniaol AH, Brunborg C, Lipka AF, Niks EH, et al. An up-date on

health-related quality of life in myasthenia gravis -results from population based cohorts.

Health and quality of life outcomes 2015;13:115.

[4] Bankole LC, Millet GY, Temesi J, Bachasson D, Ravelojaona M, Wuyam B, et al. Safety and

efficacy of a 6-month home-based exercise program in patients with facioscapulohumeral

muscular dystrophy: A randomized controlled trial. Medicine 2016;95:e4497.

[5] Siciliano G, Schirinzi E, Simoncini C, Ricci G. Exercise therapy in muscle diseases: open issues

and future perspectives. Acta myologica : myopathies and cardiomyopathies : official journal

of the Mediterranean Society of Myology 2019;38:233-238.

19
[6] Cheng A, Morsch M, Murata Y, Ghazanfari N, Reddel SW, Phillips WD. Sequence of age-

associated changes to the mouse neuromuscular junction and the protective effects of

voluntary exercise. PloS one 2013;8:e67970.

[7] Lohi EL, Lindberg C, Andersen O. Physical training effects in myasthenia gravis. Archives of

physical medicine and rehabilitation 1993;74:1178-80.

[8] Westerberg E, Molin CJ, Sporndly Nees S, Widenfalk J, Punga AR. The impact of physical

exercise on neuromuscular function in Myasthenia gravis patients: A single-subject design

study. Medicine 2018;97:e11510.

[9] Wong SH, Nitz JC, Williams K, Brauer SG. Effects of balance strategy training in myasthenia

gravis: a case study series. Muscle & nerve 2014;49:654-60.

[10] Rahbek MA, Mikkelsen EE, Overgaard K, Vinge L, Andersen H, Dalgas U. Exercise in

myasthenia gravis: A feasibility study of aerobic and resistance training. Muscle & nerve

2017;56:700-709.

[11] Westerberg E, Molin CJ, Lindblad I, Emtner M, Punga AR. Physical exercise in myasthenia

gravis is safe and improves neuromuscular parameters and physical performance-based

measures: A pilot study. Muscle & nerve 2017;56:207-214.

[12] Schuch FB, Vancampfort D, Rosenbaum S, Richards J, Ward PB, Stubbs B. Exercise improves

physical and psychological quality of life in people with depression: A meta-analysis including

the evaluation of control group response. Psychiatry research 2016;241:47-54.

[13] Stubbs B, Vancampfort D, Rosenbaum S, Firth J, Cosco T, Veronese N, et al. An examination

of the anxiolytic effects of exercise for people with anxiety and stress-related disorders: A

meta-analysis. Psychiatry research 2017;249:102-108.

[14] Herring MP, O'Connor PJ, Dishman RK. The effect of exercise training on anxiety symptoms

among patients: a systematic review. Archives of internal medicine 2010;170:321-31.

[15] Kruger K, Mooren FC, Pilat C. The Immunomodulatory Effects of Physical Activity. Current

pharmaceutical design 2016;22:3730-48.

20
[16] Gleeson M, Bishop NC, Stensel DJ, Lindley MR, Mastana SS, Nimmo MA. The anti-

inflammatory effects of exercise: mechanisms and implications for the prevention and

treatment of disease. Nat Rev Immunol 2011;11:607-15.

[17] Birnbaum S, Hogrel JY, Porcher R, Portero P, Clair B, Eymard B, et al. The benefits and

tolerance of exercise in myasthenia gravis (MGEX): study protocol for a randomised

controlled trial. Trials 2018;19:49.

[18] Boutron I, Moher D, Altman DG, Schulz KF, Ravaud P. Extending the CONSORT statement to

randomized trials of nonpharmacologic treatment: explanation and elaboration. Annals of

internal medicine 2008;148:295-309.

[19] Birnbaum S, Ghout I, Demeret S, Bolgert F, Eymard B, Sharshar T, et al. Translation, cross-

cultural adaptation, and validation of the french version of the 15-item Myasthenia Gravis

Quality Of life scale. Muscle & nerve 2017;55:639-645.

[20] Sharshar T, Chevret S, Mazighi M, Chillet P, Huberfeld G, Berreotta C, et al. Validity and

reliability of two muscle strength scores commonly used as endpoints in assessing treatment

of myasthenia gravis. Journal of neurology 2000;247:286-90.

[21] Wolfe GI, Herbelin L, Nations SP, Foster B, Bryan WW, Barohn RJ. Myasthenia gravis activities

of daily living profile. Neurology 1999;52:1487-9.

[22] Enright PL. The six-minute walk test. Respiratory care 2003;48:783-5.

[23] Salci Y, Karanfil E, Balkan AF, Kutukcu EC, Ceren AN, Ayvat F, et al. Functional exercise

capacity evaluated by timed walk tests in myasthenia gravis. Muscle & nerve 2019;59:208-

212.

[24] Skevington SM, Lotfy M, O'Connell KA. The World Health Organization's WHOQOL-BREF

quality of life assessment: psychometric properties and results of the international field trial.

A report from the WHOQOL group. Quality of life research : an international journal of

quality of life aspects of treatment, care and rehabilitation 2004;13:299-310.

21
[25] Leplege A, Reveillere C, Ecosse E, Caria A, Riviere H. Propriétés psychométriques d'un nouvel

instrument d'évaluation de la qualité de vie, le WHO-QOL-26, à partir d'une population de

malades neuro-musculaires. L'Encephale 2000;XXVI:13-22.

[26] Collet L, Cottraux J. [The shortened Beck depression inventory (13 items). Study of the

concurrent validity with the Hamilton scale and Widlocher's retardation scale]. Encephale

1986;12:77-9.

[27] Bruchon-Schweitzer M, Paulhan I, Le manuel du STAI-Y de CD Spielberger, adaptation

française. 1993, Paris: ECPA.

[28] Coopersmith S, Inventaire d’estime de soi. 1984, Paris: Edition du Centre de Psychologie

Appliquée.

[29] Sullivan TR, White IR, Salter AB, Ryan P, Lee KJ. Should multiple imputation be the method of

choice for handling missing data in randomized trials? Statistical methods in medical research

2016:962280216683570.

[30] Cella DF, Tulsky DS. Measuring quality of life today: methodological aspects. Oncology

(Williston Park, N.Y.) 1990;4:29-38; discussion 69.

[31] Bonnetain F, Fiteni F, Efficace F, Anota A. Statistical Challenges in the Analysis of Health-

Related Quality of Life in Cancer Clinical Trials. Journal of clinical oncology : official journal of

the American Society of Clinical Oncology 2016;34:1953-6.

[32] Baert I, Freeman J, Smedal T, Dalgas U, Romberg A, Kalron A, et al. Responsiveness and

clinically meaningful improvement, according to disability level, of five walking measures

after rehabilitation in multiple sclerosis: a European multicenter study. Neurorehabil Neural

Repair 2014;28:621-31.

[33] Salci Y, Karanfil E, Fil Balkan A, Calik Kutukcu E, Ceren AN, Ayvat F, et al. Functional Exercise

Capacity Evaluated by Timed Walk Tests in Myasthenia Gravis. Muscle Nerve 2018.

[34] Dobkin BH. Fatigue versus activity-dependent fatigability in patients with central or

peripheral motor impairments. Neurorehabilitation and neural repair 2008;22:105-10.

22
[35] Kelley GA, Kelley KS. Dropouts and compliance in exercise interventions targeting bone

mineral density in adults: a meta-analysis of randomized controlled trials. J Osteoporos

2013;2013:250423.

[36] Bello-Haas VD, Florence JM, Kloos AD, Scheirbecker J, Lopate G, Hayes SM, et al. A

randomized controlled trial of resistance exercise in individuals with ALS. Neurology

2007;68:2003-7.

[37] Gellen B, Messonnier LA, Galacteros F, Audureau E, Merlet AN, Rupp T, et al. Moderate-

intensity endurance-exercise training in patients with sickle-cell disease without severe

chronic complications (EXDRE): an open-label randomised controlled trial. The Lancet.

Haematology 2018;5:e554-e562.

[38] Berrih-Aknin S, Le Panse R. Myasthenia gravis: a comprehensive review of immune

dysregulation and etiological mechanisms. Journal of autoimmunity 2014;52:90-100.

[39] Sackett DL. Why Did the Randomized Clinical Trial Become the Primary Focus of My Career?

Value in Health 2015;18:550-552.

Tables

Table 1. Demographic, disease characteristics, medical history and activity of participants at baseline i.e.
all untraineda.

All (n=43) Exercise (n=23) Control (n=20)

Mean age, years 45.5 (10.0) 29 to 70 47.1 (9.0) 34 to 64 43.7 (10.9) 29 to 70
Female 40 (93%) 23 (100%) 17 (85%)
Mean body mass index (kg/m2) 28.4 (5.5) 20.1 to 45.4 27.1 (4.7) 20.1 to 38.8 29.9 (6.0) 20.6 to 45.4
BMI ≥ 30 13 (32%) 5 (24%) 8 (40%)
MG subtype (MGFA class)
Class IIa/Class IIb 17 (40%) / 6 (14%) 9 (39%) / 4 (17%) 8 (40%) / 2 (10%)
Class IIIa/Class IIIb 14 (33%) / 6 (14%) 5 (22%) / 5 (22%) 9 (45%) / 1 (5%)
Autoantibody profile
AChR 35 (81%) 20 (87%) 15 (75%)
MuSK 3 (7%) 1 (4%) 2 (10%)
Seronegative for AChR and MuSK 5 (12%) 2 (9%) 3 (15%)
History of the disease
MG onset
Juvenile, < 18 years 7 (16%) 6 (26%) 1(5%)
Early, 18-50 years 30 (70%) 13 (57%) 17 (85%)
Late, > 50 years 6 (14%) 4 (17%) 2 (10%)
Mean disease duration, years 14.3 (11.0) 1.3 to 44.5 17.6 (13.0) 1.3 to 44.5 10.4 (6.6) 2.1 to 25.9
Thymectomy 28 (65%) 18 (78%) 10 (50%)
Treatment at randomization
AchE inhibitor 37 (86%) 19 (83%) 18 (90%)
Glucocorticoid 18 (42%) 10 (43%) 8 (40%)

23
 Prednisone dose (mg) 3.2 (4.3) 3.2 (4.3) 3.1 (4.4)
Immunosuppressor 28 (65%) 13 (57%) 15 (75%)
Comorbidities
Thyroid dysfunction 7 (16%) 4 (17%) 3 (15%)
Osteoporosis 10 (23%) 6 (26%) 4 (20%)
Hypertension 8 (19%) 5 (22%) 3 (15%)
Type 2 diabetes mellitus 4 (9%) 2 (9%) 2 (10%)
Accelerometer data 37 22 15
Days worn 6 [5 to 7] 7 [5 to 7]
Average hours worn per day 14.2 (0.8) 14.5 (0.8)
Minimum hours worn per day 13.1 (12 to 15) 13.2 (12 to 16)
Maximum hours worn per day 15.3 (14 to 16) 15.5 (14 to 16)
Average activity (count per minute 446 (355 to 553) 566 (354 to 696)
per day)
a
Data are mean (SD) and range or n (%) or median [IQR].
SD, standard deviation; IQR, interquartile range; BMI, body mass index; MGFA, Myasthenia Gravis Foundation
of America classification; AChR, muscle nicotinic acetylcholine receptor; MuSK, muscle-specific tyrosine
kinase; AchE, acetylcholinesterase.
Table 2. Intention-to-treat analysis of the primary outcome. Analyses of secondary outcomesa.

Exercise Control
Outcome n Mean (SD) n Mean (SD) Adjusted MD (95% CI) p value
MGQOL-15-F score
M3 23 20.0 (8.8) 20 24.6 (8.7) … …
M6 23 18.8 (10.7) 20 23.4 (9.6) -0.8 (-5.4 to 3.7) 0.72
M9 23 19.9 (12.4) 20 20.2 (9.3) 3.9 (-1.7 to 9.5) 0.17
Myasthenia Muscle Score
M3 23 85.3 (10.6) 20 88.0 (11.5) … …
M6 23 85.4 (13.7) 18 87.1 (13.3) 1.1 (-4.9 to 6.8)* 0.58**
M9 21 85.7 (13.1) 18 88.5 (12.8) -1.1 (-6.3 to 4.3)* 0.80**
MG-ADL
M3 23 2.7 (2.6) 20 2.4 (2.2) … …
M6 23 1.6 (1.3) 19 3.4 (2.9) -1.9 (-3.0 to -0.9)* 0.005**
M9 22 2.5 (2.8) 18 2.8 (2.8) -0.3 (-2.0 to 1.4)* 0.35**
Handgrip strength (kg)
M3 23 23.8 (4.7) 20 26.4 (8.0) … …
M6 23 25.1 (5.8) 19 26.4 (8.2) 1.3 (-1.2 to 3.9) 0.31
M9 23 24.8 (4.9) 17 26.0 (9.5) 1.4 (-0.9 to 3.7) 0.25
Elbow flexion (Nm)
M3 23 36.8 (7.1) 20 39.2 (11.5) … …
M6 23 37.6 (8.3) 19 40.8 (11.7) -0.8 (-3.5 to 1.9) 0.57
M9 23 37.0 (8.2) 17 40.6 (14.9) -1.3 (-4.8 to 2.1) 0.45
Knee extension (Nm)
M3 23 107.5 (27.9) 20 115.5 (43.8) … …
M6 23 114.0 (33.3) 19 115.7 (45.9) 8.8 (0.1 to 17.5) 0.06
M9 23 109.3 (33.0) 17 118.1 (48.6) -4.0 (-18.6 to 10.6) 0.60
6-minute walking distance (m)
M3 22 502.1 (72.8) 20 493.8 (95.5) … …
M6 23 516.8 (77.8) 19 486.0 (88.2) 27.7 (7.2 to 48.1) 0.01
M9 23 508.0 (80.1) 16 499.0 (92.5) -7.5 (-29.8 to 14.9) 0.52
Forced Vital Capacity (%pred)
M3 23 82.1 (10.8) 20 87.4 (15.2) … …
M6 23 80.7 (10.6) 19 84.9 (15.0) 0.1 (-2.6 to 2.9) 0.93
M9 23 79.8 (13.7) 18 83.5 (15.3) 0.7 (-3.6 to 5.0) 0.75
Maximal Inspiratory Pressure (%pred)
M3 23 88.7 (33.1) 20 98.7 (28.9) … …
M6 22 90.7 (36.3) 19 98.5 (35.0) -0.7 (-11.2 to 9.8) 0.90
M9 23 92.7 (35.9) 17 93.4 (34.5) 5.3 (-6.1 to 16.6) 0.37
Maximal Expiratory Pressure (%pred)
M3 23 89.2 (30.1) 20 104.2 (31.4) … …

24
 M6 23 97.0 (31.3) 19 103.0 (31.5) 2.6 (-10.7 to 15.9) 0.71
M9 23 96.6 (33.8) 17 101.5 (32.8) 4.4 (-7.6 to 16.5) 0.47
a
Analyses of imputed data of primary efficacy endpoint (MGQOL-15-F: French version of the 15-item MG-
specific quality of life scale). Secondary efficacy endpoints; symptom severity (Myasthenia Muscle Score), limb
strength, walking capacity (six-minute walking distance), pulmonary function (Forced Vital Capacity, Maximal
inspiratory pressure, Maximal expiratory pressure). %pred, % predicted; SD, standard deviation; MD, mean
difference; 95% CI: 95% Confidence Interval; M3, Randomization, at month 3; M6, end of intervention period,
at month 6; M9, end of study at month 9. *Adjusted mean difference estimated by a latent class mixed-effect
models, and 95% confidence interval by bootstrap; **p-value obtained by rank ANCOVA.

Table 3. Analysis of tolerance to exercise based on neurologist assessment at each monthly visit.
Exercise Control
n No. (%) n No. (%) RD (95% CI) RR (95% CI) p value
MG Exacerbations
No. (%) until M6 23 0 (0) 19 3 (16) -16% (-38 to 2) 0 (0 to 0.98) 0.084
No. (%) until M9 23 0 (0) 18 5 (28) -28% (-51 to -7) 0 (0 to 0.55) 0.011
Cardiac, respiratory, articular and
muscular symptoms up to M6
Cardio-thoracic pain — no. (%) 19 6 (32) 14 2 (14) 17% (-13 to 42) 2.21 (0.62 to 8.84) 0.42
ECG abnormalities — no. (%) 21 3 (14) 17 0 (0) 14% (-6 to 35) ∞ (0.70 to ∞) 0.24
Dyspnea exertional — no. (%) 23 5 (22) 19 7 (37) -15% (-40 to 12) 0.59 (0.23 to 1.51) 0.32
Articular pain 20 9 (45) 17 11 (65) -20% (-46 to 12) 0.70 (0.37 to 1.27) 0.32
Muscular pain 20 11 (55) 17 11 (65) -10% (-37 to 21) 0.85 (0.49 to 1.48) 0.74
RD: absolute risk difference; RR: relative risk; 95% CI: 95% Confidence Interval; M6, month 6; M9, month 9.

Table 4. Patient-reported adverse events throughout the entire study as reported by the participant at
each monthly visit. Serious adverse events reported at any time throughout the study and declared to the
sponsor.

Variable Exercise Control p value

(n = 23) (n = 20)
No. of adverse events* 31 31 0.58
Allergic reaction 1 1
Cardiac disorder 1 0
Cephalalgia 3 1
Infection or infestation 5 6
Musculoskeletal symptoms 6 13
MG worsening 2 2
Respiratory symptoms 3 0
Other 10 8
≥ 1 event - n (%) 16 (70) 14 (70) > 0.99
Serious adverse events
Death or life-threatening event - n (%) 0 (0) 0 (0) > 0.99
Disability or incapacity - n (%) 0 (0) 0 (0) > 0.99
Hospitalization - n (%) 0 (0) 2 (10) 0.21
Hospitalization for exacerbation of myasthenia gravis - n (%) 0 (0) 2 (10) 0.21
Other significant event — n (%) 0 (0) 0 (0) > 0.99
62 adverse events were declared for 30 patients, among which 2 were considered a serious adverse event. P
values based on Fisher’s exact tests except * where P value is based on Poisson regression. n, number of
participants.

25
26