Primed Physical Therapy Enhances Recovery of Upper Limb Function in Chronic Stroke Patients

Abstract

Background. Recovery of upper limb function is important for regaining independence after stroke. Objective. To test the effects of priming upper limb physical therapy with intermittent theta burst stimulation (iTBS), a form of noninvasive brain stimulation. Methods. Eighteen adults with first-ever chronic monohemispheric subcortical stroke participated in this randomized, controlled, triple-blinded trial. Intervention consisted of priming with real or sham iTBS to the ipsilesional primary motor cortex immediately before 45 minutes of upper limb physical therapy, daily for 10 days. Changes in upper limb function (Action Research Arm Test [ARAT]), upper limb impairment (Fugl-Meyer Scale), and corticomotor excitability, were assessed before, during, and immediately, 1 month and 3 months after the intervention. Functional magnetic resonance images were acquired before and at one month after the intervention. Results. Improvements in ARAT were observed after the intervention period when therapy was primed with real iTBS, but not sham, and were maintained at 1 month. These improvements were not apparent halfway through the intervention, indicating a dose effect. Improvements in ARAT at 1 month were related to balancing of corticomotor excitability and an increase in ipsilesional premotor cortex activation during paretic hand grip. Conclusions. Two weeks of iTBS-primed therapy improves upper limb function at the chronic stage of stroke, for at least 1 month postintervention, whereas therapy alone may not be sufficient to alter function. This indicates a potential role for iTBS as an adjuvant to therapy delivered at the chronic stage.

Keywords

theta burst stimulation transcranial magnetic stimulation stroke rehabilitation functional magnetic resonance imaging

Introduction

Upper limb (UL) impairment is common after stroke and recovery of function is important for regaining independence in activities of daily living.¹ Rehabilitation of the UL involves repetitive motor practice to promote use-dependent neuroplasticity and functional recovery, and primarily occurs in the first 6 months after stroke.^2-4 Whether further gains are possible beyond this time has been a matter of ongoing debate.⁵ Therapy may need to be primed in order to realize the potential for further gains in function at the chronic stage.⁶ Noninvasive brain stimulation techniques can be used to prime the motor cortex by promoting long-term potentiation–like plasticity⁷ and rendering the primary motor cortex (M1) more receptive to input from other cortical areas for a greater response to therapy.^8-11

In healthy individuals, the balance of excitability between the 2 cerebral hemispheres is symmetric. At the chronic stage after stroke, the ipsilesional M1 is typically underexcitable and interhemispheric inhibition between the hemispheres is asymmetric, reinforcing an imbalance in corticomotor excitability between hemispheres.^12,13 Better clinical outcomes for the affected hand and arm are seen when asymmetry of corticomotor excitability is reduced.¹⁴

Noninvasive brain stimulation techniques that increase the excitability of the ipsilesional motor cortex may promote reorganization within ipsilesional M1 and improve the symmetry of corticomotor excitability between hemispheres.^9,10 A protocol of repetitive transcranial magnetic stimulation (rTMS), called intermittent theta burst stimulation (iTBS), may act as a priming stimulus to facilitate excitability and promote use-dependent plasticity.^15,16 Ipsilesional M1 iTBS followed by a single dose of UL practice at the chronic stage after stroke is more beneficial than UL practice alone.^17,18 However, one study has investigated the effects of multiple sessions combining iTBS with UL therapy in chronic stroke patients, with a negative result.¹⁹ There was no difference between real and sham treatment groups for any hand function outcome measure.

The aim of this study was to examine the effects of priming UL physical therapy with iTBS of ipsilesional M1 in subcortical stroke patients at the chronic stage. We hypothesized that UL function would be improved immediately and one month after intervention in the PRIMED group (receiving real iTBS and physical therapy) and exceed any benefit made by the CONTROL group (receiving sham iTBS and physical therapy). We also hypothesized improved UL function may be associated with balancing of cortical activity toward symmetry between the hemispheres, assessed with neurophysiology and neuroimaging measures.

Methods

Participants

Eighteen adults with UL impairment (Fugl-Meyer [FM] score >20) at least 6 months after first-ever monohemispheric subcortical stroke (Figure 1) participated in this randomized, sham-controlled, single-centred, triple-blinded trial. Volunteers were excluded if they had a brainstem or cerebellar infarction, had been diagnosed with another neurological condition, were undertaking any formal rehabilitation, or had contraindications to transcranial magnetic stimulation (TMS) or magnetic resonance imaging (MRI). They were also excluded if they were on medications that interfered with the interpretation of the neurophysiological results, or had significant aphasia or impaired cognition precluding informed consent. This study was approved by the regional ethics committee, and all participants gave written informed consent in accordance with the Declaration of Helsinki.

Figure 1.

Lesion overlay map. Images have been normalized to MNI (Montreal Neurological Institute) space, flipped so all lesions appear in the right hemisphere and overlaid on the T1 of a representative participant. Note the absence of motor cortex involvement.

Design and Procedures

The trial design and procedures are outlined in Figure 2. Interventions and assessments are described below. Participants were randomly allocated to PRIMED (real iTBS + physical therapy) or CONTROL (sham iTBS + physical therapy) groups after baseline assessments were completed. Customized software (http://www.rando.la) was used to allocate participants and minimize between-group differences in age, baseline Action Research Arm Test (ARAT) score, the presence of a motor-evoked potential (MEP) in the paretic first dorsal interosseous (FDI) muscle in response to TMS of ipsilesional M1, and fractional anisotropy (FA) asymmetry in the posterior limbs of the internal capsules.²⁰ Participants, assessors and the physiotherapist were blinded to group allocation. Participants were naïve to iTBS.

Figure 2.

Trial profile and procedures. (A) Trial profile. (B) Trial procedures. Clinical and neurophysiological assessments were completed before (Base1, Base2) and during (MID) the intervention, and immediately (IMMED), 1 month (1M) and 3 months (3M) after the intervention. Functional magnetic resonance imaging was completed at baseline (MRIBASE) and 1 month after the intervention (MRIPOST). Intervention was either Real or Sham intermittent theta burst stimulation (iTBS) prior to 45-minute upper limb therapy for 10 consecutive weekdays.

Intervention

Real or sham iTBS was delivered immediately before a 45-minute session of individualized UL physical therapy for 10 consecutive weekdays. Intermittent TBS (600 stimuli¹⁵) was delivered to ipsilesional M1 with a biphasic Rapid Stimulator (Magstim, Dyfed, UK) by an investigator blinded to all other aspects of data collection. Sham iTBS was delivered with a sham coil (Magstim, Wales, UK). The site of stimulation for ipsilesional M1 was that which produced the largest MEP amplitude in paretic FDI using single-pulse TMS (ie, the “hotspot”). When no MEPs could be elicited in paretic FDI, iTBS was applied to ipsilesional M1 at the mirror location of the “hotspot” for contralesional M1. Intermittent TBS intensity was set to 90% active motor threshold (AMT) of the nonparetic FDI. AMT was obtained as the participant performed isometric contraction of the nonparetic FDI at around 20% maximum voluntary contraction. AMT was defined as the minimum stimulus intensity that produced a peak-to-peak MEP amplitude >100 μV in 4 of 8 consecutive trials, similar to recent guidelines.²¹ Delivery site and nonparetic AMT were determined at the beginning of every session. Participants remained at rest during and for 5 minutes after iTBS delivery.

Upper limb therapy was commenced 5 minutes after iTBS, to allow time for consolidation before movement.²² Therapy consisted of 45 minutes of UL exercises (strengthening, task-specific and functional tasks) delivered by an experienced neurological physiotherapist. Therapy content was individualized to each participant, based on the therapist’s own assessment.

Assessment

Clinical and neurophysiological assessments were completed before the intervention period on 2 occasions, separated by 2 weeks (Base1 and Base2). Assessments were repeated midway through the intervention (MID) and immediately (IMMED), 1 month (1M) and 3 months (3M) postintervention. MRI studies were conducted 2 weeks before (MRI_Base) and 1 month after (MRI_POST) the intervention.

The National Institutes of Health Stroke Scale (NIHSS) and the modified Rankin Scale (mRS) were used to evaluate stroke severity and disability at baseline. UL function and impairment were evaluated with the Action Research Arm Test (ARAT²³) and the UL component of the Fugl-Meyer Scale (FM²⁴), respectively. Assessors were blinded to group allocation and not involved in treating participants.

Transcranial magnetic stimulation was used to evaluate corticomotor excitability. MEPs were recorded from FDI bilaterally using standard surface electromyography (EMG) techniques. Signals were amplified using Grass P511 amplifiers (Grass Instrument Division, Warwick, RI), bandpass filtered at 20 to 1000 Hz, sampled at 2 kHz, and stored for subsequent analysis. Rest motor threshold (RMT) was determined for each FDI, defined as the minimum stimulus intensity that produced a peak-to-peak MEP amplitude ≥50 μV in 4 of 8 consecutive trials, similar to recent guidelines.²¹ Stimulus-response curves were constructed by recording blocks of 12 MEPs (4-5 seconds between stimuli) at intensities −5%, +5%, +15%, +25%, and +35% of maximum stimulator output (MSO) relative to RMT, with the order of stimulus intensities randomized. Trials were rejected online when root mean square EMG (rmsEMG) calculated over a 100 ms window prior to the stimulus exceeded 10 µV. The average MEP amplitude at each stimulus intensity was calculated, and the slope of the linear portion of the stimulus-response curve was estimated. Slope was set to zero when no MEPs could be elicited. Interhemispheric balance of corticomotor excitability was quantified by calculating an asymmetry index from the slope value (mV/10%MSO) for the contralesional and ipsilesional hemisphere: CE_AI = (CE_contra − CE_ipsi)/(CE_contra + CE_ipsi). The asymmetry index was set to 1.0 when no MEPs could be elicited from the ipsilesional M1. The change in corticomotor excitability asymmetry index (∆CE_AI) was calculated for participants with MEPs in the paretic FDI.

For neuroimaging, all T1-weighted and diffusion-weighted images were acquired with a Siemens 1.5 T Avanto scanner. Axial T1-weighted images were used to identify lesion location and had 1.0 × 1.0 × 1.0 mm voxels, a 256 mm field of view, repetition time (TR) = 11 ms, and echo time (TE) = 4.94 ms. Diffusion-weighted images had 1.8 × 1.8 × 3.0 mm voxels, a 230 mm field of view, b = 2000 s/mm², TR = 6700 ms, TE = 101 ms, 30 gradient directions, and 2 averages. All image processing was carried out with the Oxford FMRIB Software Library.²⁵ The mean FA was calculated within the posterior limb of each internal capsule (PLIC) by warping a template PLIC volume of interest to the participants’ images.²⁶ The structural integrity of the PLICs was quantified by calculating an asymmetry index from the mean FA values: FA_AI = (FA_contra − FA_ipsi)/(FA_contra + FA_ipsi).^20,26

The functional MRI (fMRI) experiment was an event-related handgrip task.²⁷ Blood oxygen level–dependent (BOLD) contrast images were acquired using a T2*-weighted single-shot gradient echo EPI sequence (TR = 3000 ms, TE = 50 ms, 3 mm isotropic voxels, axial slices, 90° flip angle, 64 × 64 matrix). MRI compatible force transducers (Biopac Ltd) were placed in each hand, with the transducer in the paretic hand used to generate the visual feedback signal,²⁷ and the other transducer used to monitor the presence of mirror movements. Each scanning run lasted 6 minutes and contained 45 events separated by interstimulus intervals of 7 ± 2 seconds. Two scanning runs were performed with each hand for a total of 90 events per hand.

Each event consisted of paretic hand grip to a specified level of force that was displayed visually using custom software and a projection system. Force targets were varied pseudo-randomly between 10%, 15%, 25%, 30%, and 35% of maximum voluntary contraction for the paretic hand. Each target force was maintained for 1.7 s until a cue to “relax” appeared. Rest intervals varied between 7 and 9 seconds and served as an implicit baseline.

Image processing for fMRI was performed using FMRIB’s software library and FMRI expert analysis Tool, (FEAT V5.98).^28,29 Images were motion and slice-time corrected, spatially smoothed (5 mm) and then coregistered to the T₁-weighted image and spatially normalized to the Montreal Neurological Institute (MNI) template.³⁰ Each grip was modelled with the height (force) scaled relative to the maximum voluntary contraction and convolved with a gamma-shaped hemodynamic response function, along with its time derivative. The 6 motion parameters computed by MCFLIRT were modeled as nuisance variables, as was the handgrip force signal from the good hand (not involved in the task). fMRI statistical analyses were determined from a contrast of handgrip force computed for each scan (first-level analysis). A second-level analysis combined both scans in a fixed effects model for each participant at each time point (MRI_Base and MRI_POST). Both first- and second-level analyses used a corrected cluster threshold of Z > 2.3 and significance threshold of P = .05. For participants with a left hemisphere lesion, images were left-right flipped such that lesions were always on the right side to assist with group-level analyses.

In FEAT, MRIpost − base difference images were computed from second-level analyses, and a 2-group unpaired t test was conducted at a third-level (PRIMED vs CONTROL). An additional third-level analysis was conducted independent of group to determine which brain areas (Post) were positively associated with ∆ARAT and ∆FM (fixed effect, single group plus covariate de-meaned). Results of third-level analyses were masked using the Hand Motor Area Template with a corrected cluster threshold of Z > 2.3 and significance threshold of P = .05. Anatomical locations of activation foci were identified probabilistically using the Juelich Histological Atlas.³¹

Supra-threshold voxels from Z-statistic images were identified within the Hand Motor Area Template,³² which consists of bilateral primary motor (M1) and sensory (S1) cortex, dorsal and ventral premotor (PMd, PMv) cortex, and pre- and proper supplementary motor area (preSMA, SMA). A laterality index (LI) was calculated as follows: LI = (NV_contra − NV_ipsi)/(NV_contra + NV_ipsi), where NV_contra and NV_ipsi are the number of suprathreshold voxels within contralesional areas and ipsilesional areas, respectively.^20,33 ∆LI was calculated (MRI_POST − MRI_BASE) such that negative values reflect a change toward more normal activation during the task.²⁷

Statistical Analysis

The primary endpoint was the improvement in UL function at 1 month, measured with the ARAT. The secondary endpoints were UL impairment (FM), corticomotor excitability asymmetry index (CE_AI), and FMRI lateralization (LI) at 1 month. ∆ARAT, ∆FM, and ∆CE_AI were calculated at each time-point, by subtracting the baseline from the scores at MID, IMMED, 1M, and 3M. For fMRI lateralization, ∆LI was calculated as LI_POST − LI_BASE.

Baseline, ∆ARAT and ∆FM data were analyzed using repeated-measures analysis of variance (RM ANOVA) with the between-subject factor group (PRIMED, CONTROL), and time and hand (when appropriate) as the within-subject factors. Kolmogorov-Smirnov tests were used to confirm normality. Post hoc 2-tailed independent t tests and 2-tailed 1-sample t tests were used to investigate significant effects. Two-tailed independent t tests were conducted to compare ∆LI between groups.

Regression analyses were used to determine associations between the change in clinical measures (∆ARAT and ∆FM) and the change in corticomotor excitability asymmetry (∆CE_AI) immediately and 1M after the intervention, independent of group. Regression analyses were also used to determine associations between the change in clinical measures and the change in lateralization of brain activation from FMRI (∆LI) at 1M, independent of group. For the fMRI data, a further analysis was conducted independent of group to determine which brain areas at MRI_POST were positively associated with ∆ARAT and ∆FM at 1M.

Baseline measures were analyzed with independent 2-sided t tests for linear continuous variables and 2-tailed Pearson chi-square tests for nominal and ordinal variables, except when expected cell counts were less than 5, in which case 2-tailed Fisher’s exact tests were used. Modified Bonferroni correction was used for multiple comparisons.³⁴ Significance level was P = .05.

Results

The PRIMED and CONTROL groups were well matched at baseline (Table 1). RMT was analyzed with an RM ANOVA, with factors hand (paretic, nonparetic), group, and time (Base1, Base2). RMT was higher in the paretic FDI (68% ± 6% MSO) than the nonparetic FDI (40% ± 2% MSO), as expected, F(1, 16) = 19.2, P < .001. For each hemisphere RMT was similar between groups and stable across baseline (both Ps > .05), and there were no interactions (all Ps > .4). Contralesional FDI AMT was similar for the PRIMED (51% ± 7% MSO) and the CONTROL groups (49% ± 8% MSO) (2-tailed independent t test, P > .5). Intermittent TBS stimulation intensity was similar for the PRIMED (46% ± 6% MSO) and CONTROL groups (44% ± 7% MSO) (2-tailed independent t test, P > .5). Intermittent TBS protocols were well tolerated by participants, with no adverse effects. There was no mortality, seizures, hospital visits, or recurrent strokes in either group.

Table 1.

Baseline Characteristics for Randomized Participants.

	PRIMED	CONTROL	P Value
Age in years, median (range)	61 (21-80)	71 (38-79)	.23
Sex, n	3 female, 6 male	3 female, 6 male	1.00
Lesioned hemisphere	3 left, 6 right	3 left, 6 right	1.00
Months poststroke, median (range)	20 (6-72)	18 (7-56)	.58
NIHSS, median (range)	3 (0-4)	2 (0-5)	.18
mRS, median (range)	2 (2-4)	2 (1-4)	.78
UL-FM, mean (range)	38 (23-63)	40 (21-56)	.88
ARAT, mean (range)	31 (13-51)	30 (5-55)	.93
Paretic FDI MEPs	6 yes, 3 no	7 yes, 2 no
CE asymmetry, mean (range)	0.887 (0.33-1.00)	0.541 (–0.16 to 1.00)	.08
FA asymmetry, mean (range)	0.215 (0.050-0.571)	0.258 (–0.033 to 0.485)	.61

Abbreviations: NIHSS, National Institutes of Health Stroke Scale; mRS, modified Rankin Scale; UL-FM, upper limb Fugl-Meyer Scale; ARAT, Action Research Arm Test; FDI, first dorsal interosseous; MEPs, motor-evoked potentials; CE, corticomotor excitability; FA, fractional anisotropy of the posterior limb of the internal capsule.

Independent 2-sided t tests were used for linear continuous variables. Two-sided Pearson chi-square tests were used for nominal and ordinal variables, except when expected cell counts were less than 5, in which case 2-sided Fisher’s exact tests was used.

Upper Limb Function

The second baseline ARAT scores (Base2) were used to calculate ∆ARAT, as there was a significant effect of time on the 2 baseline scores, F(1, 16) = 8.4, P = .010. A small increase in mean ARAT score was seen (Base1 = 28.9, 95% CI = 20.6-37.2; Base2 = 30.4, 95% CI = 22.2-38.7), probably due to increasing familiarity with the testing procedure. There was no effect or interaction with group (both P > .3).

Upper limb function improved in the PRIMED but not CONTROL groups immediately and at 1 month after intervention. There was a main effect of group, F(1, 16) = 14.2, P = .002, and no effect or interaction with time (both Ps > .2) (Figure 3A). The main effect of group arose because ∆ARAT was greater for the PRIMED than the CONTROL group (PRIMED = 1.9, 95% CI = 1.1–2.8; CONTROL = −0.19, 95% CI = −1.0 to 0.65). The increase in ARAT score was greater than zero immediately after the intervention (IMMED) for the PRIMED group—mean ∆ARAT = 3.2, t(1, 8) = 3.7, P = .006—but not the CONTROL group—mean ∆ARAT = −0.4, t(1, 8) = −1.1, P = .312. The increase in ARAT was also greater than zero at the primary endpoint (1M after treatment concluded) for the PRIMED group—mean ∆ARAT = 2.0, t(1, 8) = 4.0, P = .004—but not the CONTROL group—mean ∆ARAT = 0.2, t(1, 8) = 0.4, P = .719. In addition, the change in ARAT score was higher for the PRIMED than CONTROL group immediately after the intervention: IMMED, t(1, 16) = 3.8, P = .002; and at the primary endpoint: 1M, t(1, 16) = 2.3, P = .036). ∆ARAT did not differ from baseline; nor did it differ between groups, at the MID or 3M time points (all Ps > .1).

Figure 3.

Upper limb function and impairment. (A) Change in Action Research Arm Test score (∆ARAT) for the PRIMED (black bars) and CONTROL (light gray bars) groups across time. There was a main effect of group (P = .002). ARAT score improved (positive ∆ARAT) for the PRIMED compared with the CONTROL group. For the PRIMED group ∆ARAT was positive immediately and 1 month after intervention, and was significantly greater than for the CONTROL group (all P < .04). (B) Change in upper limb (UL) Fugl-Meyer score (∆FM) across time (patterned bars). There was a main effect of time (P = .025). FM score increased (positive ∆FM) compared with baseline immediately after intervention. For both figures: MID, midway through intervention; IMMED, immediately after intervention; 1M, 1 month after intervention. 3M, 3 months after intervention. Error bars = standard error; ^P < .05, 2-tailed t test comparison with baseline. *P < .05, **P < .01, 2-tailed t test comparison between groups. (C) Regression analysis between ∆ARAT score (positive values indicate improvement; negative values, deterioration) and ∆ corticomotor excitability asymmetry index (CE_AI) (negative values indicate less asymmetry) at 1 month after intervention (1M) for the PRIMED (black symbols) and CONTROL (light gray symbols). Data points are participants.

Upper Limb Impairment

The 2 groups were matched at baseline for UL impairment, with no effect of time or group, and no interaction between them (all Ps > .4). For consistency, FM scores at the second baseline were used to calculate ∆FM.

Upper limb impairment improved in both the PRIMED and CONTROL groups immediately after intervention. There were no differences between the PRIMED and CONTROL groups for the change in FM score, or any interaction with time (both Ps > .7). There was an effect of time, F(3,48 = 3.4, P = .025, as FM score was greater immediately after the intervention—IMMED, t(1, 17) = 3.4, P = .004, but at no other time points (all Ps > .07) (Figure 3B).

Regression Analyses

Greater improvement in ARAT score was associated with a shift toward more balanced corticomotor excitability (negative ∆CE_AI) 1 month after intervention. There was a weak negative correlation between ∆CE_AI and ∆ARAT at the 1M primary endpoint (R² = .269, P = .028) (Figure 3C). There was no association between ∆CE_AI and ∆ARAT, or between ∆CE_AI and ∆FM, at the IMMED time point (both R²s < .16, P > .8). MEPs were present in the paretic FDI in 13 participants. Slope was set to zero and CE_AI was set to 1.0 for the remaining 5 participants.

Functional Magnetic Resonance Imaging

At baseline, paretic hand task performance was associated with bilateral cortical activity, which was more lateralized to the ipsilesional hemisphere for less impaired participants (Figure 4).

Figure 4.

Functional magnetic resonance imaging (fMRI) results. Event-related fMRI was performed for visuomotor force handgrip task performed with the paretic hand (n = 16). (A) Group average activation at baseline and (B) Primary endpoint. Colors show range above threshold (2.3) to maximum = 17.3. The background anatomical image is the MNI152_1mm_brain, radiological convention (right hemisphere, image left). Transparent colours over frontal and parietal cortical regions outline the 12 areas of the Hand Motor Area Template (HMAT) used for prethreshold masking of Z-statistics at the group level. (C) Individual fMRI results from 2 participants (red = mild impairment; blue = severe impairment) at baseline (overlaid on MNI template and HMAT as above). Crosshair is located at MNI coordinate (mm 35, −22, 57) from group peak of activation at baseline (A above). Colors show range above threshold (2.3) to maximum = 10.3.

At baseline, the group average activation (maximum Z = 12.2) had peaks probabilistically located within the ipsilesional hemisphere across the corticospinal tract white matter (78%), primary motor cortex BA4a (60%) and BA4p (7%), and premotor cortex BA6 (14%) (Figure 4A). One month after the intervention the group average activation (maximum Z = 16.1) had peaks similarly located within the ipsilesional hemisphere in the primary motor cortex BA4p (80%) and BA4a (46%), primary somatosensory cortex BA3b (34%) and corticospinal tract white matter (24%) (Figure 4B). From inspection of Figure 4A and B, the activation spanned both hemispheres into all regions of the Hand Motor Area Template, with the peak of activation centered near the hand knob,³⁵ and M1/S1 border of the central sulcus of the ipsilesional hemisphere. Conversely, in the contralesional hemisphere, there was activation in the premotor, supplementary, and secondary sensory regions, but activation was largely absent along the M1/S1 boundary and central sulcus. This indicates the contralesional activation pattern was unlikely to represent nonparetic hand involvement.

Results from 2 participants at baseline are shown in Figure 4C and illustrate the extent to which activation was lateralized depending on impairment level. The red-yellow map is the task-related activation from a participant with mild impairment and good function at baseline (ARAT = 55; FM = 54). Activation was more lateralized and contained predominantly within the ipsilesional hemisphere. The blue–light blue map is from a participant with more severe impairment and poor function (ARAT = 14; FM = 26). This participant had more bilateral task-related activation including cortical areas beyond those defined by the Hand Motor Area Template.

The change in the laterality index after the intervention did not differ between the PRIMED and CONTROL groups (∆LI; P = .52). When the groups were combined, there was no significant activation in the average pre-post and post-pre contrasts, indicating there was no consistent pattern of cortical reorganization across participants.

Changes in UL function and impairment were related to 2 distinct patterns of activation postintervention (Figure 5). Improvements in ARAT were associated with activation peaks (2193 voxels, maximum Z = 6.54) in ipsilesional premotor cortex BA6 (52%) and primary motor cortex BA4a (4%). Improvements in FM were associated with activation peaks (1857 voxels, maximum Z = 4.9) in ipsilesional primary somatosensory cortex BA1 (50%), BA2 (25%), and BA3b (19%), primary motor cortex BA4a (13%) and BA4p (5%) and superior parietal lobule 5L (4%). Activation in these areas was associated with positive linear trends in the respective clinical scores.

Figure 5.

Functional magnetic resonance imaging (fMRI) covariate analyses. Group-level covariate analyses postintervention found that change in Action Research Arm Test score (∆ARAT; blue–light blue) was associated with increased activation in ipsilesional premotor cortex, and change in upper limb (UL) Fugl-Meyer score (∆FM; red-yellow) was associated with increased activation in ipsilesional sensorimotor cortex. Colors show range above threshold (2.3) to maximum = 10.3. The background anatomical image is the MNI152_1mm_brain, radiological convention (right hemisphere, image left). Transparent colors over frontal and parietal cortical regions outline the 12 areas of the Hand Motor Area Template (HMAT) used for prethreshold masking of Z-statistics at the group level.

Discussion

Two weeks of iTBS-primed physical therapy improved UL function in people who were on average 20 months poststroke. The improvement in ARAT score was modest; however, it exceeded the intrarater limits of agreement.³⁶ Physical therapy alone was not sufficient to produce functional gains, possibly because the dose was inadequate. This study provides some insight into adequate doses of iTBS-primed therapy, as 2 weeks, but not 1 week, of treatment were required to produce improvement in UL function. Ten days of iTBS-primed therapy produced improvements that persisted for at least 1 month, but not 3 months. Longer treatment periods may produce more lasting benefits, and regular retreatment may be needed to maintain these benefits long term.

For a lower therapy dose, priming may be required to maximize recovery potential in the chronic phase after stroke. Two weeks of high-dose UL therapy such as modified constraint-induced movement therapy can produce lasting functional gains but is only suitable for people with relatively mild impairment.³⁷ A more recent study found that 1 hour of UL therapy delivered daily for 2 weeks did not alter motor function in people only 8 months poststroke.³⁸ However, when iTBS was delivered on the same days as therapy, motor function and reaction time of the paretic hand improved. The timing of iTBS relative to therapy was not reported in this previous study making it difficult to identify the potential mechanism underlying these gains. The present results indicate that iTBS may specifically prime the ipsilesional hemisphere in a way that increases the effectiveness of a low dose of UL therapy.

Intermittent TBS may have created a permissive environment for cortical reorganization in response to therapy by facilitating cortical excitability, promoting long-term potentiation–like plasticity⁷ and modifying the receptiveness of M1 to input from other cortical areas.¹¹ These effects may have contributed to improvements in the planning, fractionation, and coordination of paretic UL movement, reflected by an increase in ARAT score that persisted for at least 1 month. This interpretation is supported by our finding that improvement in UL function was related to rebalancing of M1 excitability, as hypothesized. Improvement in UL function was also related to an increase in ipsilesional dorsal premotor cortex activation during paretic hand grip. Dorsal premotor cortex is implicated in organizing and selecting motor output, and through its dense connections with M1 is thought to make an important contribution to the recovery of motor function after stroke.³⁹

Upper limb impairment was reduced immediately after the intervention, regardless of whether participants were primed with real or sham iTBS, but this improvement did not persist 1 month later. Therapy is likely to have had both central and peripheral effects. Centrally, the increase in FM score was related to increased ipsilesional sensorimotor cortex activation during paretic hand grip but not rebalancing of M1 excitability. The former may reflect enhanced processing of sensory feedback during paretic hand use. Peripheral effects such as reconditioning of muscles, joints, and connective tissue may also contribute to reduced impairment. Although 2 weeks of therapy temporarily reduced impairment, this did not translate to improved function unless therapy was primed with iTBS.

The strengths of this study are the blinding of participants, therapists, and assessors, and the balancing of groups at baseline. The multimodal design provided some insight into the underlying mechanisms of the observed clinical benefits of the intervention, which was brief, noninvasive, and well-tolerated. A potential limitation is that the resources required to conduct the repeated neurophysiological and neuroimaging measures constrained the study’s sample size. However, future trials may focus on clinical outcomes, rather than underlying mechanisms, enabling larger sample sizes.

In conclusion, iTBS priming of the ipsilesional hemisphere is an effective adjuvant to upper limb physical therapy for patients at the chronic stage. Larger gains in motor function at the chronic stage may be realized with primed therapy, as opposed to therapy alone. It appears that more than 1 week of primed therapy is required to produce improvements in motor function, and more than 2 weeks of primed therapy may be required to produce lasting benefits. Primed therapy promoted rebalancing of corticomotor excitability, and this was associated with greater improvements in UL function, confirming previous work with patients at the chronic stage of recovery. Improvements in UL function and impairment were associated with different patterns of cortical reorganization, and this may be a useful area of future research.

Footnotes

Acknowledgements

The authors thank Raewyn Hopkins for assistance with data collection and analysis.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by a Neurological Foundation of New Zealand project grant (NF0931_PG).

References

Veerbeek

Kwakkel

van Wegen

Ket

Heymans

MW.

Early prediction of outcome of activities of daily living after stroke: a systematic review. Stroke. 2011;42:1482-1488.

Kreisel

Hennerici

Bäzner

Pathophysiology of stroke rehabilitation: the natural course of clinical recovery, use-dependent plasticity and rehabilitative outcome. Cerebrovasc Dis. 2007;23:243-255.

Nakayama

Jorgensen

Raaschou

Olsen

TS.

Recovery of upper extremity function in stroke patients: the Copenhagen Stroke Study. Arch Phys Med Rehabil. 1994;75:394-398.

Kwakkel

Kollen

BJ.

Predicting activities after stroke: what is clinically relevant?

Int J Stroke. 2013;8:25-32.

Teasell

Mehta

Pereira

. Time to rethink long-term rehabilitation management of stroke patients. Top Stroke Rehabil. 2012;19:457-462.

Stinear

Barber

Coxon

Fleming

Byblow

WD.

Priming the motor system enhances the effects of upper limb therapy in chronic stroke. Brain. 2008;131:1381-1390.

Huang

Chen

Rothwell

Wen

HY.

The after-effect of human theta burst stimulation is NMDA receptor dependent. Clin Neurophysiol. 2007;118:1028-1032.

Ackerley

Stinear

CM.

Stimulating stimulation: can we improve motor recovery following stroke using repetitive transcranial magnetic stimulation?

Phys Ther Rev. 2010;15:302-308.

Butler

Shuster

O’Hara

Hurley

Middlebrooks

Guilkey

A meta-analysis of the efficacy of anodal transcranial direct current stimulation for upper limb motor recovery in stroke survivors. J Hand Ther. 2013;26:162-170.

10.

Hsu

Cheng

Liao

Lee

Lin

YY.

Effects of repetitive transcranial magnetic stimulation on motor functions in patients with stroke: a meta-analysis. Stroke. 2012;43:1849-1857.

11.

Ackerley

Stinear

Barber

Byblow

WD.

Priming sensorimotor cortex to enhance task-specific training after subcortical stroke. Clin Neurophysiol. 2013;125:1451-1458.

12.

Murase

Duque

Mazzocchio

Cohen

LG.

Influence of interhemispheric interactions on motor function in chronic stroke. Ann Neurol. 2004;55:400-409.

13.

Swayne

Rothwell

Ward

Greenwood

RJ.

Stages of motor output reorganization after hemispheric stroke suggested by longitudinal studies of cortical physiology. Cereb Cortex. 2008;18:1909-1922.

14.

Takeuchi

Oouchida

Izumi

Motor control and neural plasticity through interhemispheric interactions. Neural Plast. 2012;2012:823285.

15.

Huang

Edwards

Rounis

Bhatia

Rothwell

JC.

Theta burst stimulation of the human motor cortex. Neuron. 2005;45:201-206.

16.

Bolognini

Pascual-Leone

Fregni

Using non-invasive brain stimulation to augment motor training-induced plasticity. J Neuroeng Rehabil. 2009;6:8.

17.

Talelli

Greenwood

Rothwell

JC.

Exploring theta burst stimulation as an intervention to improve motor recovery in chronic stroke. Clin Neurophysiol. 2007;118:333-342.

18.

Ackerley

Stinear

Barber

Byblow

WD.

Combining theta burst stimulation with training after subcortical stroke. Stroke. 2010;41:1568-1572.

19.

Talelli

Wallace

Dileone

. Theta burst stimulation in the rehabilitation of the upper limb: a semirandomized, placebo-controlled trial in chronic stroke patients. Neurorehabil Neural Repair. 2012;26:976-987.

20.

Stinear

Barber

Smale

Coxon

Fleming

Byblow

WD.

Functional potential in chronic stroke patients depends on corticospinal tract integrity. Brain. 2007;130:170-180.

21.

Huang

Rothwell

Edwards

Chen

RS.

Effect of physiological activity on an NMDA-dependent form of cortical plasticity in human. Cereb Cortex. 2008;18:563-570.

22.

Lyle

RC.

A performance test for assessment of upper limb function in physical rehabilitation treatment and research. Int J Rehabil Res. 1981;4:483-492.

23.

Fugl-Meyer

Jaasko

Leyman

Olsson

Steglind

The post-stroke hemiplegic patient I. A method for evaluation of physical performance. Scand J Rehabil Med. 1975;7:13-31.

24.

Rossini

Barker

Berardelli

. Non-invasive electrical and magnetic stimulation of the brain, spinal cord and roots: basic principles and procedures for routine clinical application. Report of an IFCN committee. Electroencephalogr Clin Neurophysiol. 1994;91:79-92.

25.

Smith

Jenkinson

Woolrich

. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage. 2004;23(suppl 1):S208-S219.

26.

Petoe

Byblow

de Vries

. A template-based procedure for determining white matter integrity in the internal capsule early after stroke. Neuroimage Clin. 2014;4:695-700.

27.

Ward

Newton

Swayne

. The relationship between brain activity and peak grip force is modulated by corticospinal system integrity after subcortical stroke. Eur J Neurosci. 2007;25:1865-1873.

28.

Beckmann

Jenkinson

Smith

SM.

General multilevel linear modeling for group analysis in FMRI. Neuroimage. 2003;20:1052-1063.

29.

Smith

Beckmann

Ramnani

. Variability in fMRI: a re-examination of inter-session differences. Hum Brain Mapp. 2005;24:248-257.

30.

Jenkinson

Bannister

Brady

Smith

Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage. 2002;17:825-841.

31.

Eickhoff

Paus

Caspers

. Assignment of functional activations to probabilistic cytoarchitectonic areas revisited. Neuroimage. 2007;36:511-521.

32.

Mayka

Corcos

Leurgans

Vaillancourt

DE.

Three-dimensional locations and boundaries of motor and premotor cortices as defined by functional brain imaging: a meta-analysis. Neuroimage. 2006;31:1453-1474.

33.

Cramer

Nelles

Benson

. A functional MRI study of subjects recovered from hemiparetic stroke. Stroke. 1997;28:2518-2527.

34.

Rom

DM.

A sequentially rejective test procedure based on a modified Bonferroni inequality. Biometrika. 1990;77:663-665.

35.

Yousry

Schmid

Alkadhi

. Localization of the motor hand area to a knob on the precentral gyrus. A new landmark. Brain. 1997;120(pt 1):141-157.

36.

Van der Lee

De Groot

Beckerman

Wagenaar

Lankhorst

Bouter

LM.

The intra- and interrater reliability of the Action Research Arm Test: a practical test of upper extremity function in patients with stroke. Arch Phys Med Rehabil. 2001;82:14-19.

37.

McIntyre

Viana

Janzen

Mehta

Pereira

Teasell

Systematic review and meta-analysis of constraint-induced movement therapy in the hemiparetic upper extremity more than six months post stroke. Top Stroke Rehabil. 2012;19:499-413.

38.

Sung

Wang

Chou

Chen

Chang

Tsai

PY.

Efficacy of coupling inhibitory and facilitatory repetitive transcranial magnetic stimulation to enhance motor recovery in hemiplegic stroke patients. Stroke. 2013;44:1375-1382.

39.

Fridman

Hanakawa

Chung

Hummel

Leiguarda

Cohen

LG.

Reorganization of the human ipsilesional premotor cortex after stroke. Brain. 2004;127:747-758.