Abstract
Background and Purpose A pilot study to determine feasibility of detecting changes in structural connectivity (SC) and resting-state functional connectivity (RSFC) occur alongside motor improvements after participation in the Targeted Ballet Program (TBP) in adults with relapsing-remitting multiple sclerosis (RRMS).
Methods Five participants (four female) with RRMS between the ages of 38-64 with the following characteristics at baseline: Expanded Disability Status Scale 2.0-6.0, International Cooperative Ataxia Rating Scale (ICARS) > 7, Symbol-Digit Modality Test > 22, and no relapses or initiation of medications indicated to affect mobility within the past 30 days. Participants were asked to complete 12 weeks (one hour, twice per week) of the TBP. Magnetic resonance imaging data was collected pre- and post-intervention for SC and RSFC network analysis.
Results Increases in two RRMS-related graph theoretical measures (mean strength and mean clustering coefficient) for RSFC (p < 0.05) are detectable alongside significant reduction in ataxia (ICARS: p = 0.01012, Smoothness Index: p = 0.04995), and increase in balance (Mini-BESTest: p = 0.01474) following participation in the well-tolerated TBP.
Discussion and Conclusions Significant increases in mean strength and mean clustering coefficient of RSFC suggest functional neurological improvements after participation in the TBP. The relationship between these network changes and clinical improvements in balance and amelioration of ataxia after participation in the TBP requires a larger randomized-controlled clinical trial of the TBP in persons with RRMS.
Introduction
Multiple sclerosis (MS) is an autoimmune-mediated disease with brain demyelination and axonal loss causing impaired mobility and ataxia, which affect an estimated 75% and 80% of persons with MS respectively 1-3. Ataxia presents as varied combinations of unsteady gait, limb trajectory errors, trunk instability, motor sequencing and timing errors, speech impairment, decreased tone, abnormal stretch reflexes, and body or head tremor 4. An estimated 900,000 persons in the U.S. suffer from MS, which has no known cure 5 or effective pharmacological treatment for restoring mobility and decreasing ataxia 6,7. Pharmacological treatment of MS consists of disease-modifying drugs that slow down disease progression by targeting immune system function. However, these drugs do not induce remyelination of the CNS and, therefore, loss of motor and sensory functions persists.
Studies in mice and in healthy adult humans suggest that myelination is required for motor learning 8. Learning complex motor tasks – such as juggling, balancing on an unstable board, rhythmically cued finger movements, and dance – has been shown to increase white matter integrity and structural connectivity (SC) in fiber tracts in the cerebellum and areas involved in visuo-motor coordination and intra- and interhemispheric communication in both healthy adults and after stroke 8-17. Resting-state functional connectivity (RSFC) 18,19 analyses have shown changes in magnetic resonance imaging (MRI) in healthy adults due to bimanual visual tracking training and musical performance training. However, it is not yet known whether learning complex motor movements causes similarly significant changes in brain connectivity in persons with MS.
A classical ballet-based program for complex motor learning delivered in a group setting, targeted to improve balance and reduce ataxia in persons with MS has shown positive results beyond those of other physical rehabilitation interventions 20. Regarding dance research, evidence shows positive training effects of classical ballet on balance21, postural control22, joint position sense23, and a shift from vision to proprioception in the coordination of movements24,25 in healthy adults, beyond the existing evidence for the benefits of social dance forms (e.g. salsa and ballroom dancing) 23-32. The foundation of the targeted ballet program (TBP) lies in complex motor learning intrinsic to western classical ballet training 20. Motor learning in ballet training organizes movement and its instruction hierarchically while requiring active allocation of proprioceptive, auditory, visuospatial, emotional, and attentional resources. Classical ballet training hierarchy is based on a set of postures organized in anatomical Cartesian planes. Trained ballet postures are akin to letters in the alphabet. Learning complexity evolves by connecting postures to create movement words and phrases20. This approach to movement composition resembles theories of motor control based in the compositionality of dynamic primitives: movements that connect consecutive postures may be thought of as the dynamic ballet primitives 33-35. By modulating the dynamics of movement phrases, specific movement qualities such as smoothness and static and dynamic equilibrium are explicitly trained. This process is amenable to the training of movement patterns for transfer to everyday functional mobility, thereby unlocking ballet’s potential for movement rehabilitation. Movement narratives with meaning during training enhance motivation and cognitive engagement during complex motor learning.
In a pilot study, this TBP was well-tolerated and showed evidence that it leads to improvements in motor measures in persons with relapsing-remitting MS (RRMS) 20. Prior to this work, the largest improvements due to physical activity interventions in MS were 11.5% for balance and 15.2% for a timed 10-meter walk test 36,37. TBP increased Mini-Balance Evaluation Systems Test (Mini-BESTest) scores by 42% (p = 1e-4, Cohen’s d = 1.2) and decreased International Cooperative Ataxia Rating Scale (ICARS) scores by 58% (p = 7.11e-5, Cohen’s d = 2.6) 20. However, we did not collect data on concurrent brain connectivity changes.
Brain imaging of SC before and after participation in TBP may provide evidence of neuroplasticity underlying motor improvements in persons with MS. Along with SC, RSFC measures the effective active communication between regions 38. RSFC correlates with SC 39, but may provide differential sensitivity to indirectly connected networks in MS. Previous research supports the potential of this method in MS. RSFC has been shown to be reproducible in patients with stable RRMS according to a recent study of 20 RRMS patients compared with 14 healthy controls 40. Due to simultaneous use of multiple sensory modalities and cognitive networks, we expect increases in RSFC network measures after participating in the TBP.
Methods
This research was approved by the Institutional Review Board at the University of Illinois at Urbana-Champaign in accordance with the Declaration of Helsinki. All participants gave voluntary informed consent after approval for participation. This preliminary study is registered in ClinicalTrials.gov under the trial identifier NCT04073940 (https://clinicaltrials.gov/ct2/show/NCT04073940).
Participants and Setting
Inclusion criteria were: age 18-64 years, informed consent, confirmation of RRMS diagnosis and movement training approval from participant’s neurologist, Expanded Disability Status Scale scores of 2.0-6.5, relapse-free for 30 days, and absence of other central nervous system conditions. Exclusion criteria were: Symbol Digit Modalities Test < 23, pregnancy, education level < 8th grade, change in disease modifying therapy within 6 months, initiating medications that influence mobility within 30 days, orthopedic or musculoskeletal conditions, and contra-indications for MRI. Recruitment began upon IRB approval on June 6th of 2018 and ended on September 19th of 2019.
Pre-intervention assessments began on August 20th of 2019 and ended on September 19th of 2019. Post-intervention assessments were performed between December 3rd of 2019 and December 13th of 2019. All motor assessments and TBP classes were performed at the University of Illinois at Urbana-Champaign in the ballet studio of the Neuroscience of Dance in Health and Disability Laboratory. MRI data was collected at the Biomedical Imaging Center of the Beckman Institute for Advanced Science and Technology at the University of Illinois at Urbana-Champaign. The University of Illinois at Urbana-Champaign is located in a semi-rural community.
Intervention
Participants were asked to perform 12 weeks (one hour, two days/week) of group TBP classes led by a Bolshoi-certified ballet instructor (CLO). Weekly make-up classes were made available to account for pre-existing time conflicts expressed by multiple participants. TBP classes were taught from August 28th, 2019 to December 2nd, 2019. Every class consisted of a seated warm up and ballet technique (20 min), followed by exercises using ballet barres (15 min), exercises across the floor (20 min), and a cool-down (5 min) 20. The seated warm-up focuses on articulating feet and hands, with progression from distal to proximal movements throughout the body, emphasizing selective control of each joint. As participants gain selective joint control, the warm-up transitions to smoothly connected, whole-body movements that link proximal to distal joints and vice versa. During the seated ballet section of the TDP, participants execute movements starting with the with the lower limbs in parallel position and progressing to first position of the feet with external rotation at the hips, from which they perform exercises in an adapted manner. The seated exercises were developed to provide the motor coordination foundations for the ballet barre and across the floor exercises. Movement adaptations focus on maintaining key characteristics of each ballet movement 41. Exercises were selected to target movement deficits observed in the group. Whenever possible, participants performed all movements to the front, side, and back directions. Movements were performed in the allowable range of motion of the joints of each participant, to minimize the possibility of injury. All instruction conformed to the principles of classical ballet technique and focus on control of the trunk with awareness of all joints in postures and movements. Movement difficulty was comparable to the Royal Academy of Dance and the Cecchetti Council of America Ballet I-II Syllabi 42,43. The instructor and assistants ensured that participants performed the intervention with the equivalent of the moderate level of the Borg Rating of Perceive Exertion Scale44, with focus on motor learning over aerobic or muscular exertion. University-level graduate and undergraduate trained assistants at a ratio of 1:1 to 2:1 facilitated fidelity in the execution of the movement presented and adherence to the prescribed movement. Movements were performed in time with piano recordings from Dmitri Roudnev’s Ballet Class Music Series45
Main Outcome Measures
Feasibility
Recruitment and retention rates were calculated as basic measures of trial feasibility. A retention rate of 80% or better was considered feasible due to similar rates observed in prior MS studies in our setting. Adherence to protocol was determined by the number of hours of TBP instruction time completed by each participant in comparison to the 24 hours of prescribed instruction time. Other dance forms have been well-tolerated in MS 46, thus we expected the TBP to be well-tolerated as seen in our prior pilot study 20. Tolerability was measured by the ability of participants to complete each TBP class without related adverse events or issues of thermoregulation preventing participation. Testing burden was deemed feasible if participants could complete all testing measures within the prescribed testing window without requesting adjustment to protocol. Detection of significant motor improvements and changes in SC and RSFC metrics served as critical measures of effect-detection feasibility.
Motor function
We clinically assessed ataxia using ICARS, the leading comprehensive clinical measure of ataxia in MS with strong inter-evaluator reliability and validity 47-49. Higher ICARS scores indicate increased ataxia. We assessed balance ability using the Mini-BESTest, which has been validated in MS 50-52. Lower scores for the Mini-BESTest indicate better control of balance. Using a Qualisys motion capture system (Qualisys AB, Goteborg, Sweden), we assessed ataxia quantitatively using the spectral arc length metric yielding a bilateral smoothness index (s-index) 20,53,54 calculated using the SpectralArcLength.m function from https://github.com/siva82kb/smoothness 53.
MRI Connectivity
We calculated three graph theory metrics (GTMs): mean strength, global efficiency, and mean clustering coefficient that have been used previously in MS to characterize disruptions in SC and RSFC in relation to sensorimotor function 55-58. We expected to detect increases in these GTMs for both SC and RSFC based on large increases in associated motor function seen in our prior pilot of the TBP 20.
Acquisition
For diffusion-weighted imaging (DWI) on the Siemens 3 T Prisma (Siemens, Erlangen, Germany) using a 64-channel head coil, the CMRR multiband sequence (https://www.cmrr.umn.edu/multiband/; 59,60) was used for both the fMRI and DWI data achieving whole brain coverage with 2.5 mm isotropic resolution, multiband factor of 4. DWI data had 64 diffusion encoded directions in 2 shells (b-value = 1500 and 3000 s/mm2). Blood-oxygen-level dependent (BOLD) fMRI data (TR 2 s, TE 40 ms, 96 slices of 2.5 mm slice thickness, flip angle 52°) was collected for a 9-minute acquisition using a visual fixation point as performed for MS in Bollaert et al 61. Standard structural data including a 0.9 mm isotropic T1-weighted 3D MPRAGE (TR 2.3 s, TE 2.32 ms, TI 900 ms, flip angle 8°) and 1.0 mm isotropic T2-weighted flow attenuated inversion recovery (FLAIR) acquisition were acquired as in our prior RRMS research 62.
Preprocessing for Resting-State Functional Connectivity
Structural and fMRI data were preprocessed with fMRIPrep 20.0.6 63, based on Nipype 64 1.4.2. T1-weighted (T1w) volumes were corrected for intensity non-uniformity using N4BiasFieldCorrection 65 and skull-stripped using antsBrainExtraction.sh v2.1.0 using the NKI template. Brain surfaces were reconstructed using FreeSurfer recon-all v6.0.1 66. Spatial normalization to the ICBM 152 Nonlinear Asymmetrical template version 2009c 67 was performed through nonlinear registration with antsRegistration 68.
Functional data was slice time corrected using 3dTshift from AFNI v16.2.07 69 and motion corrected using mcflirt 70. Motion correcting transformations, field distortion correcting warp, BOLD-to-T1w transformation and T1w-to-template (MNI) warp were concatenated and applied using antsApplyTransforms using Lanczos interpolation.
Resting-State Functional Connectivity
RSFC processing was performed using the 36 parameter confound regression and motion de-spiking pipeline design (see github.com/PennBBL/xcpEngine/blob/master/designs/fc36p_despike.dsn) in xcpEngine 71.Pearson’s correlations were then quantified for regions of interest in the Automated Anatomical Labeling Atlas 72. Network-based statistics 55,56 were calculated for the whole-brain network comprised of these regions of interest using the Brain Connectivity Toolbox (BCT) 55,73.
Preprocessing for Structural Connectivity
Preprocessing was performed using QSIPrep 0.12.2, which is based on Nipype 1.5.1 64. Many internal operations of QSIPrep use nilearn 74 and DIPY 75. The T1-weighted (T1w) image was corrected for intensity non-uniformity (INU) using N4BiasFieldCorrection65, and used as T1w-reference throughout the workflow. The T1w-reference was then skull-stripped using antsBrainExtraction.sh (ANTs 2.3.1), using OASIS as target template. Spatial normalization to the ICBM 152 Nonlinear Asymmetrical template version 2009c 67 was performed through nonlinear registration with antsRegistration68 (ANTs 2.3.1), using brain-extracted versions of both T1w volume and template. Brain tissue segmentation of cerebrospinal fluid (CSF), white-matter (WM) and gray-matter (GM) was performed on the brain-extracted T1w using FAST76 (FSL 6.0.3:b862cdd5). The DWI time-series were resampled to ACPC, generating a preprocessed DWI run in ACPC space.
Structural Connectivity
Reconstruction was performed using QSIprep 0.12.2, which is based on Nipype 1.5.1 64 Gorgolewski et al. (2018). Multi-tissue fiber response functions were estimated using the Dhollander algorithm. Fiber orientation distributions (FODs) were estimated via constrained spherical deconvolution 77,78 (CSD) using an unsupervised multi-tissue method 79. Reconstruction was done using MRtrix380. FODs were intensity-normalized using mtnormalize 81. Three GTMs describing network properties that reflect communication disruption caused by MS were calculated using the BCT 55,73.
Statistical Analysis
Outcome differences were tested for normality using the Shapiro-Wilk Test (p < 0.05) and visual assessment of QQ-plots and boxplots 82-84. We then conducted single-tailed paired t-tests in R (R Core Team, Vienna, Austria) 85. Given that this was a single-center pilot study, feasibility measures were not statistically tested due to low generalizability of such statistics.
Results
Twenty-one potential participants expressed interest in the study, six of whom were available to regularly attend the TBP classes and complete eligibility screening. Safety precautions related to COVID-19 prevented offering TBP classes for four potential participants. One potential participant was ineligible due to age outside of the range approved for inclusion. Five participants were eligible, provided informed consent, and completed the study: four female, one left-handed, two retired and three employed, three Caucasian, one Hispanic/Latinx, and one African-American. Participants had the following characteristics at baseline and hours of participation (see Table 1):
One participant stopped at 50% of the intervention duration due to an unrelated knee injury. However, this participant performed all post-intervention testing and is included in statistical analysis. This retention rate of 80% in the intervention is in line with previous studies on MS in our group and setting. The TBP was well-tolerated, with no adverse events related to participation and with successful thermoregulation of the dance studio and availability of trained assistants for comfortable completion of each class. All motor testing was completed within a two-hour block of testing and MRI scans were completed in less than one hour per session.
Paired t-tests showed reduction in ataxia (ICARS: p = 0.01012; bilateral s-index: p = 0.04995) and increased balance (Mini-BESTest: p = 0.01474). Mean strength and mean clustering coefficient increased for RSFC data (p < 0.05) (see Table 2). No significant changes were detected in structural connectivity analysis. Raw data for these outcomes are available upon request.
Discussion
Originally, additional rounds of the TBP were planned to begin in the Spring semester with rolling-enrollment allowing testing and the intervention to minimize conflict with Winter holidays and maximize participant recruitment and adherence. However, recruitment for this pilot study was limited by the ongoing COVID-19 pandemic. The TBP was well-tolerated and participants were able to complete all testing. The intervention retention rate of 80% is in line with prior MS research in our setting and was due to outside circumstances unrelated to the intervention. As was originally planned for this pilot, future trials will be scheduled to avoid holiday conflicts and provide a longer period of TBP class availability to maximize adherence and consistent intervention delivery.
Motor improvements seen here are consistent with our pilot study of TBP in females with RRMS 20, including for the first time a male with RRMS. While still a very large effect, the ICARS Cohen’s d = 1.67 was lower than the corresponding Cohen’s d = 2.6 from our pilot study of TBP 20. These smaller effect sizes were expected due to lower baseline levels of ataxia and ∼25% fewer intervention hours spent than in the prior study 20. This effect size is also limited by less than full completion of the intervention by one participant. Improvements in the bilateral s-index (Cohen’s d = 0.95) and Mini-BESTest (Cohen’s d = 1.48) were similar to the respective large and very large effect sizes seen previously for TBP 20. Sample size bias-corrected Hedge’s g 86 yielded large effect sizes for all clinical outcomes and a medium effect size for the s-index (see Table 2).
As expected, alongside motor improvements, we observed increases in mean strength and mean clustering coefficient in the RSFC data. These metrics have been shown to be lower in persons with MS than healthy controls 57,58. Thus, their increases suggest improvement in persons with MS, bringing these measures closer to those seen in healthy sensorimotor function 57,58. The extent to which each GTM represents disease response to intervention is limited by largely cross-sectional current literature.
The lack of expected significant increases in SC measures may be due to the small sample size of this preliminary study. Based on these results and accounting for the 80% intervention completion rate, a future longitudinal randomized-controlled trial with a repeated measures ANOVA statistical design would require recruitment of n = 212 participants achieve a significant change at power = 0.80 and α = 0.005 (Bonferroni-corrected for 10 simultaneous comparisons) in the GTMs of interest for SC that showed non-negligible changes in this pilot. The same statistical design would require recruitment of n = 20 participants for a trial to detect significant changes in GTMs for RSFC. However, we acknowledge that this estimation is limited in generalizability to longitudinal and multi-site trials due to the single-center, pre- and post-intervention design of this pilot.
With only five participants, statistical power of a small sample limits conclusions about the general RRMS population. We acknowledge that significance of outcome changes is limited by the need for appropriate correction for multiple simultaneous comparisons. However, the detection of within-participant differences in motor measures, mean strength, and mean clustering coefficient in RSFC supports a larger longitudinal study of the TBP in persons with RRMS. This would be a longitudinal randomized-controlled trial of the TBP in persons with RRMS, comparing the effects of the TBP on the aforementioned connectivity measures in SC and RSFC and motor outcomes to those of an intensity-matched version of the National Multiple Sclerosis Society Stretching for People with MS group class augmented by walking and lower extremity functional exercises known to improve mobility in persons with RRMS 36,87.
Conclusions
The detection of significant increases in mean strength and mean clustering coefficient of resting-state functional connectivity coinciding with clinical improvements in balance and amelioration of ataxia after participation in the TBP supports a larger randomized-controlled clinical trial of the TBP in persons with RRMS. Effect sizes seen in this preliminary study may be skewed due to a small number of participants but remain large/medium after correction for small sample size. Although generalizability to multi-site randomized-controlled trials is limited, enrollment, retention rates, and well-tolerated intervention and testing suggest that a larger clinical trial is feasible.
Data Availability
Data related to this study is available upon request to the authors
Authors’ Roles
CLO conceived, organized, and executed the project with assistance from PBC and BPS. BPS designed the MRI acquisition and analysis strategy with assistance from PBC and CLO. PBC processed and analyzed data with design, review, and critique from CLO and BPS. PBC wrote the manuscript with critique and assistance from CLO and BPS.
Financial Disclosures
PBC was funded by the International Graduate Mentoring Program as a Graduate Mentor and by the Graduate College Distinguished Fellowship (University of Illinois at Urbana-Champaign). BPS received funding from the National Institutes of Health (R01DE027989-01A1, 1R21HD095314-01, R01AG053952, 1R03CA235109-01, 1R21EB026238-01A1, 1R01AG059878-01, 1RF1AF062666, R01MH116226); the Center for Nutrition, Learning, and Memory at the University of Illinois; and the Jump Applied Research for Community Health through Engineering and Simulation Research Grant. He also has a research agreement with Siemens Healthineers. CLO received honoraria from the International Graduate Mentoring Program at the University of Illinois at Urbana-Champaign, grants from the National Multiple Sclerosis Society (#PP3418), National Science Foundation (Co-PI Federal Award ID 2024837), the Arnold and Mabel Beckman Foundation through the Beckman Institute for Advance Science and Technology, OSF Healthcare: Jump Applied Research for Community Health through Engineering and Simulation Research Grant (CHI 64809549v7-239), and a Fellowship from the Center for Advanced Study at the University of Illinois at Urbana-Champaign.
Acknowledgements
We thank the Arnold and Mabel Beckman Foundation for financial support through the Beckman Institute for Advanced Science and Technology. We greatly appreciate the Biomedical Imaging Center and the Neuroscience of Dance in Health and Disability Laboratory staff, especially Andrea Rivera-Maza and Nina Crouchelli.
Footnotes
Financial Disclosure/Conflict of Interest: The authors have no conflicts of interest to declare relative to this research.
Funding Sources:Arnold and Mabel Beckman Foundation through the Beckman Institute for Advanced Science and Technology.
Presentations:This work has not been presented in other publications or scientific meetings to date.
References
- 1.↵
- 2.
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.
- 27.
- 28.
- 29.
- 30.
- 31.
- 32.↵
- 33.↵
- 34.
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.
- 49.↵
- 50.↵
- 51.
- 52.↵
- 53.↵
- 54.
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.
- 84.↵
- 85.↵
- 86.↵
- 87.↵