Participants
Patients with stroke were recruited from the rehabilitation ward of Chang Kung Memorial Hospital. Inclusion criteria were: (1) first ever cerebral stroke; (2) under stable condition; (3) unilateral hemiplegia or hemiparesis due to unilateral cerebral stroke; (4) Brunnström stage of the affected upper limb ≥ 3, and (5) 30 to 70 years of age. Exclusion criteria were: (1) brainstem or cerebellar stroke; (2) history of seizure, brain aneurysm or arteriovenous malformation; (3) active psychiatric disease; (4) progressive neurodegenerative disease impairing cognitive function; (5) communicated disorders such as aphasia; (6) severe or active medical problems such as cardiac disease or pneumonia; (7) heavy metal implant; (8) pregnancy, (9) severe visual impairment; and (10) inability to follow instructions. All participants had signed the informed consent. The study was approved by Chang Gung medical foundation institutional review board and was registered under ClinicalTrials.gov ID No. NCT03350087.
Design and experimental procedure
This study was a prospective, double-blinded and randomized controlled trial. Patients were randomly assigned to iTBS or sham stimulation in addition to VCT and were blind to the type of stimulation delivered. Randomized allocation was performed by generating a random sequence on the (https://www.randomizer.org/) website. Figure 1 and 2 illustrate schematic overviews of the randomized allocation and experimental procedure, respectively. Each patient received iTBS or sham stimulation before the 60-minute VCT program on the same day for 15 consecutive working days (3 weeks). To avoid the contamination of physical activities on the effects of TBS, patients were told to avoid any movement of the affected upper limb before, during, and after the stimulation. We try to avoid subjects’ physical activities and consolidate the effects of TBS in the period between TBS and VCT. Patients were then moved from the site of TBS stimulation to that of VCT by a wheelchair, and the distance between two sites was around 2 meters. The training of VCT program was started as early as the setting of VCT was completed and the vital signs of patients were checked. In general, it took around 10 minutes between the end of TBS and the beginning of training. Patients were evaluated within 3 days before and after completing the therapy. The outcome measures were administered by raters, occupational therapists, who contacted patients only during assessment and were blind to group assignment. The raters were trained before the experiment and evaluated by the written exam and reliability test. A 10-patient reliability test, measuring both intra-rater and inter-rater reliability, was conducted at 7-day intervals. The intra-rater/inter-rater reliability of the MAS-UE, FMA-UE, BBT and ARAT were analyzed by intra-class correlation as 0.841/0.841, 0.984/0.992, 1.000/0.998, and 0.986/0.998.
Virtual reality-based cycling training
The VCT program comprised a warm-up exercise for 5 minutes, a 10-min weight training for upper limb including muscle strengthening, a 40-min cycling program composed of warm up, strength, and endurance training, and a 5-min cool down exercise. Dr. Hsieh-Ching Chen integrated virtual reality program with arm cycle (BK0010, X-BIKE Fitness Technology Company Limited) to comprise the virtual reality-based cycling system. The setup of the VCT was demonstrated in figure 3. During the training of the VCT program, patients would see themselves controlling the handlebar of a bicycle while riding on the road in different types of sceneries. The visual speed of the virtual scene was altered according to the signal of the cycling speed transmitted to the computer, increasing participants’ interest and motivation. Participants underwent low to moderate resistance and high revolutions per minute cycling exercise during the VCT program. Participants were encouraged to raise rpm during the program, aiming for the target heart rate based on the Karvonen Formula [22]. Thus, to ensure that participants achieved the target heart rate, the resistance was adjusted according to each participant’s clinical condition. To ensure participants’ safety, blood pressure, heart rate and oxygen saturation were monitored throughout the whole training program.
Intermittent theta burst stimulation paradigm
iTBS was delivered over the hand motor area of the affected hemisphere by a handheld 70-mm standard, figure-of-eight coil connected to a MagPro X100 package (Magventure, USA). The optimal coil positioning over the scalp region was the motor hot spot, where the transcranial magnetic stimulation (TMS) evoked the largest MEP in the contralateral first dorsal interosseous (FDI) muscle with the patient at rest. Active motor threshold (AMT) was measured before each intervention, and was defined as the minimum TMS intensity required to evoke MEPs (≥200 μV) in at least 5 of 10 successive trials from the slightly contracted (approximately 10-20% of maximal strength) FDI muscle. True stimulation was applied over the identified motor hot spot at an intensity of 80% AMT, with the coil placed tangentially to the skull, at a 45° angle to the midsagittal axis, generating posterior-anterior current flow at targeted area in the brain. If MEPs could not be elicited, stimulation was applied to the mirror site of the motor hot spot over the unaffected hemisphere, as previous studies [19, 20], at an intensity of 70% RMT. Sham stimulation was administered at the same site with identical flip coil, resulting in a 78% output elicited by non-flip side, at a lower intensity (60% AMT) equivalent to 46.8% AMT [23]. The sham stimulation with intensity lower than 70% AMT has no effect on MEPs, as demonstrated by a previous study [24], but produces indistinguishable sensation and sound compared to real stimulation. Several previous studies administered a similar sham stimulation method, and found it to be useful [19, 23, 25]. All patients were seated comfortably with their hands as relaxed as possible throughout the experiment after the AMT was recorded. An iTBS session comprised 2-second train of bursts, containing three pulses at 50Hz, repeated at intervals of 200ms, every 10 seconds for 20 times (a total of 600 pulses). Learning from studies on spaced TBS [15, 16], two sessions of iTBS were applied with a 10-to-15-min break for a total of 1200 pulses to consolidate and enhance the effects. Patients were instructed to rest and not to move during the 10-min break in order to minimize cofounding factors and not to disrupt any ongoing plasticity [26]. Real or sham iTBS was delivered for 15 consecutive work days.
Outcome measures
Based on the conceptual framework of International Classification of Functioning, Disability and Health (ICF) [27], body function, activities and participation of upper limb motor function before and after the intervention were evaluated using seven measures, namely MAS-UE, FMA-UE, ARAT, BBT, and NHPT, MAL, and SIS. Among these aspects, primary outcome measure was the domain of body function, and secondary outcome measures were the domains of activities and participation.
Co-primary outcome measures were the improvement of body function, measuring by MAS-UE and FMA-UE. MAS-UE, which is scored from 0 to 4 (0, 1, 1+, 2, 3, 4), was used to assess UL spasticity and resistance during passive joint movement [28, 29]. The affected finger flexor muscles, wrist and elbow were evaluated. These MAS-UE scores were summed to represent the UL spasticity, with 1+ calculated as 1.5. FMA-UE is a performance-based scale, particularly for patients with stroke, to assess sensorimotor function including motor function, joint function, sensation and balance [30]. This study only evaluated the UL motor function of FMA.
The improvement of activity was evaluated using ARAT, BBT, NHPT and MAL. ARAT measures UL motor function, and comprises 19 items divided into 4 subsets: grasp, grip, pinch, and gross movement (GM) [31]. BBT is a functional test to measure unilateral gross manual dexterity [32], in which patients have to move as many blocks from one box to another box as possible in 60 seconds only by the affected hand, which task requires grasping, transporting and releasing. NHPT is a timed test performed to evaluate manual dexterous function [33], in which patients insert nine pegs into nine holes of the pegboard and then pick them up as quickly as possible. Because the patients with severe motor impairment might be unable or needed a long time to complete the task, patients were asked to perform NHPT within two minutes. To distinguish the ability of the patients who could not complete the task in two minutes, the number of pegs being placed and removed was calculated. The outcome variable is the number of pegs/minute, and more pegs/minute indicates better dexterity. MAL was assessed to determine patients’ real life functional performance involving the affected arm based on 14 daily activities [34], including amount of use and quality of movement.
SIS, a patient-reported questionnaire, was performed to evaluate participation in patients with stroke [35]. SIS is a measure specific in patients with stroke, and higher scores reflect greater participation.
Statistical analysis
All statistical analyses were conducted with SPSS version 21 (SPSS Inc., Chicago, Illinois). To determine the baseline between-group differences of demographic characteristics, Chi-square tests were applied for the categorical variables and independent two-sample t-tests were conducted for the continuous variables. Paired t-tests were run to test whether each groups showed significant improvement after the therapy. Analysis of covariance (ANCOVA) was conducted to control the variance in the covariate, and was applied to assess whether the iTBS group had greater therapeutic effect than the control group. The pre-test scores were the covariates, the post-test scores were the dependent variables, and the group was the independent variable. The effect size (η2) was calculated to assess the degree of between-group differences, which were classified as large (η2 ≥ 0.138), moderate (0.059 ≤ η2 < 0.138), and small (0.01 ≤ η2 < 0.059) [36]. Statistical significance level was set at p < 0.05 (one-tailed) [37] for all analyses under directional hypothesis [38].