- Email: [email protected]
Motor learning after stroke: Is skill acquisition a prerequisite for contralesional neuroplastic change?
Neuroscience Letters 482 (2010) 21–25
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Motor learning after stroke: Is skill acquisition a prerequisite for contralesional neuroplastic change? Lara A. Boyd a,b,∗ , Eric D. Vidoni c , Brenda D. Wessel a a b c
Department of Physical Therapy, University of British Columbia, Vancouver, British Columbia, Canada Brain Research Centre, University of British Columbia, Vancouver, British Columbia, Canada Department of Neurology, University of Kansas Medical Center, Kansas City, KS, United States
a r t i c l e
i n f o
Article history: Received 4 November 2009 Received in revised form 27 April 2010 Accepted 29 June 2010 Keywords: Stroke Learning Motor cortex Rehabilitation fMRI
a b s t r a c t Limited data directly characterize the dynamic evolution of brain activity associated with motor learning after stroke. The current study considered whether sequence-specific motor skill learning or increasing non-specific use of the hemiparetic upper extremity drive functional reorganization of the contralesional motor cortex after stroke. Eighteen individuals with chronic middle cerebral artery stroke practiced one of two novel motor tasks; a retention test occurred on a separate fifth day. Using the hemiparetic arm, participants performed a serial targeting task during two functional MRI scans (day one and retention). Participants were randomized into either a task-specific group, who completed three additional sessions of serial targeting practice, or a general arm use group, who underwent three training sessions of increased but non-task specific use of the hemiparetic arm. Both groups performed a repeated sequence of responses that may be learned, and random sequences of movement, which cannot be learned. Change in reaction and movement time for the repeated sequence indexed motor learning; shifts in the laterality index (LI) within primary motor cortex (M1) for repeated and random sequences illustrated training effects on brain activity. Task-specific practice of the repeated sequence facilitated motor learning and shifted the LI for M1 as shown by a reduced volume of contralesional cortical activity. Random sequence performance did not stimulate motor learning or alter the LI within the task-specific training group. Further, between-group comparisons showed that increasing general arm use did not induce motor learning or alter brain activity for either random or repeated sequences. Motor skill learning of a repeated sequence altered cortical activation by inducing a more normal, contralateral pattern of brain activation. Our data suggest that task-specific motor learning may be an important stimulant for neuroplastic change and can remediate maladaptive patterns of brain activity after stroke. © 2010 Elsevier Ireland Ltd. All rights reserved.
How the brain compensates for damage after stroke is unclear, in part because the relationship between behavior and brain function are not well understood. One mechanism among many, that may help explain recovery after stroke is experience-dependent cortical plasticity [22], yet no pattern of motor-related brain activation has emerged that explains improved function. Some neuroplastic changes after stroke are helpful for recovery; others, such as bilateral cortical activity during unimanual movements, have uncertain value [7,11]. More than 40% of individuals with stroke have chronically impaired upper extremity (UE) function [13]. In the healthy brain, movement in one UE is associated with activity in the contralateral
∗ Corresponding author at: University of British Columbia, 212-2277 Wesbrook Mall, Vancouver, British Columbia V6T 1Z3, Canada. Tel.: +1 604 822 7197; fax: +1 604 822 1860. E-mail address: [email protected] (L.A. Boyd). 0304-3940/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.neulet.2010.06.082
primary motor cortex (M1). However, it is common for bilateral M1 to be activated during UE movements after both cortical [7,11] and subcortical stroke [23]. Recent work suggests that bilateral activity in M1 after stroke may relate to altered interhemispheric excitability and a loss of transcallosal inhibition from the ipsi- to the contralesional cortex [4]. One metric that captures this maladaptive intercortical relationship is the laterality index (LI), which is often presented as a ratio of the volume or intensity of activity in the motor cortices [29]. Despite the abnormal patterns of neural activation that occur [7], it is clear that stroke does not abolish motor learning ability [1]. Robust demonstrations of motor learning and improved function have been consistently shown in individuals with chronic stroke [1], suggesting that the functional organization of the motor system can be modified by use. However, our understanding of brain function is limited by an absence of data that directly characterizes the dynamic evolution of brain activity associated with motor learning after stroke. Understanding the conditions that normalize patterns
22
L.A. Boyd et al. / Neuroscience Letters 482 (2010) 21–25
Table 1 Participant characteristics. Group
Gender
Age
Affected side
Hand dominance
MMSE
UE Fugl-Meyer
Orpington
SP
5M 4F
62.7
3R 6L
8R 1L
28.3 (1.8)
58 (8.5)
2.0 (.37)
GP
7M 2F
61.1
2R 7L
9 R
28.8 (1.2)
58 (6.5)
2.0 (.39)
Mimi-mental status exam (MMSE) range 0–30, score >24 indicates absence of dementia; UE motor portion of the Fugl-Meyer range 0–66, higher score indicates less impairment; Orpington prognostic scale, range 1.6–6.8, higher scores indicate severer stroke. Mean (standard deviation).
of brain activity may be important as contralesional cortical activity may stimulate increased transcallosal inhibition back onto the ipsilesional motor cortex [10] and represent a source of motor control deficits after stroke. In the current study we begin to resolve this issue by simultaneously mapping changes in motor-related cortical activation and task-related behavioral improvements that reflect motor learning. Specifically, we considered whether sequencespecific motor learning, non-specific motor learning, or increased use of the hemiparetic UE could drive functional brain reorganization. We hypothesized that contralesional cortical activity in the primary motor cortex (M1) is modifiable and would be reduced by sequence-specific motor skill learning as evidenced by change in the laterality index [8], derived from functional magnetic resonance imaging (fMRI). To test our hypothesis we compared changes in behavior and brain activity (1) within a group that practiced both repeated and random sequences of movement, and (2) between a group who received task-specific training for repeated and random sequences and another that underwent increased non-task specific use of the hemiparetic UE. Eighteen individuals with a first middle cerebral artery stroke who met the inclusion criteria for functional MRI participated. To avoid confounding changes in brain activation patterns associated with learning with physiologic recovery, we tested individuals who were at least 6 months post-stroke [14]. All were free from neglect, aphasia, hemianopsia, and dementia. We categorized arm function and stroke severity the Fugl-Meyer [12] and Orpington scales [15] (Table 1). Participants were randomly assigned into either a task-specific training group (SP) or an increased general arm use group (GP). Independent t-tests verified that there were no between-group differences in age, arm function or stroke severity. Five sessions were completed on separate days over a two-week period. Using the hemiparetic UE, all participants performed a serial targeting (ST) task during functional MRI (fMRI) scanning on day 1 and at a retention test on day 5; three intervention sessions were performed on days 2–4. Individuals in the SP group practiced the ST task over 3 days; those in the GP group performed the same number of hemiparetic arm movements but in a non-specific fashion (details below). Institutional review boards at University of Kansas Medical Center and University of British Columbia approved all procedures. To test individuals without fractionated finger movement, ST task responses were made using a non-ferrous joystick (Resonance Technology) in combination with a task that did not rely on finger or wrist movements. When cued, participants made flexion/extension movements of the stroke-affected elbow and shoulder to guide a cursor (Fig. 1A) to one of three 2 cm targets located equidistant (10 cm) from the start position when it was highlighted in yellow. To ensure that movements were made to, and not through the target, participants held their cursor inside the highlighted target for 500 ms (Fig. 1B) before being cued to return to home (Fig. 1C). Following a short delay (200 ms), the next cue to move (new target highlighted) was delivered. All participants were instructed to respond as fast and accurately as possible. Joystick position sampling and all stimuli were presented at 40 Hz
using custom software developed in LabView (National Instruments Co.). Participants alternated between responding to a repeated sequence of targets (8-elements long) and random sequences. Using the same joystick on the practice days (2–4), the SP group performed the repeating sequence 40 times (320 responses); 80 random responses were also completed (400 total movements per day). Comparing performance from the random and repeated sequences allowed the separation of the effects of nonspecific learning from those associated sequence-specific learning [2]. Participants in the GP group underwent 3 days of increased use of their hemiparetic UE not specific to the ST task. The intervention was delivered by a physical therapist. Non-specific training tasks were patterned on prior reports [28] and to match the experimental ST task, also relied on elbow and shoulder flexion and extension. To mimic the movements of the SP group, the GP group made equidistant movements to three targets to accomplish bean-bag pushing, cone stacking and erasing; movement excursion was restrained to maintain similarity with the ST task. In addition the number of movements during the GP intervention (400/day) was matched to the number made by the SP group. Participants in the GP group were instructed to be as fast and accurate as possible. To address the possibility that contralesional cortical activity was the result of bilateral, “mirror” movements participants were screened for their ability to move the hemiparetic UE without movement of the opposite UE. During practice, surface electromyographic recordings of the flexor carpi ulnaris, extensor carpi radialis, posterior and anterior deltoid from the non-stroke UE verified the absence of mirror movements in the non-hemiparetic arm. The same motor task was performed at both fMRI sessions (days 1 and 5); each trial took 6 s to complete. Participants lay supine, with their elbow flexed to 45◦ and forearm in a resting position on their stomach while gripping the non-ferrous joystick. The joystick sent behavioral data to a computer. Before the first session, all participants were familiarized with the motor task. Visual stimuli were presented using a Resonance Technologies VisuaStim goggle system. A 3.0 T head only scanner equipped with a three-axis local gradient radio frequency coil collected whole brain fMRI (29 axial, 4 mm slices, skip 0.5 mm). Functional imaging data were collected as axial echo-planar images, using a single-shot, blipped gradientecho echo-planar pulse sequence (TE = 30 ms, TR = 2.0 s, 90◦ flip angle, FOV = 256 mm, 64 × 64). Four, 6 min 42 s runs of functional data were completed (201 images). Each run contained three periods of rest (19 images), a block of tracking to randomly highlighted targets (72 images), and another block of tracking following the repeating sequence (72 images). Prior to functional imaging, highresolution 3D T1 images were collected for anatomic localization and co-registration (TE = 5 ms, TR = 24 ms, 40◦ flip angle, NEX = 1, thickness = 1.2 mm, FOV = 256 mm, 256 × 256). Mean reaction time (RT) and movement time (MT) were calculated for the random and repeating sequences. RT was the time from target highlight to the beginning of participant’s response (movement onset). MT was movement onset to target hit. A change score (day 5 to day 1) was calculated for RT and MT to index the effect
L.A. Boyd et al. / Neuroscience Letters 482 (2010) 21–25
23
Fig. 1. Experimental task: (A) start position; (B) participant moves to and holds on target; and (C) cued to return to home.
of practice condition on responding; separate change scores were computed for repeated and random sequences. All fMRI data processing was performed using Analysis of Functional NeuroImages software [6]. Functional images were generated by condition (Rest, Random, Repeated), spatially registered to correct for head motion, and smoothed using a 4 mm fullwidth-half-maximum Gaussian kernel. A boxcar function modeled reactions, producing estimates of the blood oxygen level dependent response relative to baseline. To reduce the impact of any head motion signal predictors of no interest were included to account for translational and rotational motion in the x, y, and z planes. A region of interest (ROI) was defined using structural landmarks for M1 in both hemispheres for individual participants [9]. The area of M1 was defined as extending from the anterior bank of the central sulcus to the anterior edge of the precentral gyrus [9]. Within M1, ROI brain activation was represented by percent signal change (PSC), calculated from fMRI signal relative to baseline (Fig. 2A). Using the product of PSC and volume (in l) of significantly activated
voxel clusters within our M1 ROI, the laterality index (LI) [(contralateral − ipsilateral)/(contralateral + ipsilateral)] was calculated for each hemisphere by condition (Repeated, Random) [8]. The LI produces a range from −1 to +1, with a negative number indicating primarily ipsilateral activation and a positive number indicating activation primarily contralateral to the UE performing the task. We first assessed the impact of our practice conditions on motor learning. Sequence-specific motor learning was indexed using paired t-tests to compare within group change in RT for repeated versus random sequences at the retention test. The same test was repeated with MT as the dependent measure; a p-value of .025 was employed to account for multiple comparisons. To verify differences between practice conditions a multivariate ANOVA indexed the amount of between-group change in RT and MT. After establishing that sequence-specific motor learning occurred for the repeated sequence in the SP group, we tested our hypothesis that contralesional cortical activity is modifiable by motor skill acquisition as evidenced by change in the LI. A Group (SP, GP) by Sequence
Fig. 2. (A) M1 ROIs pre and post 3 days of practice. Less bilateral M1 activation was noted after practice in the SP but not the GP participant (p > .05). I = ipsilesional corresponding to the arm used; C = contralesional. (B) RT Change. (C) MT Change. (D) LI change. Learned repeated sequences by the SP group showed shifts in LI towards ipsilesional M1 because of reduced contralesional activity (1.0 = ipsilesional M1 activation).
−.087 0.08 (0.01) 5468.43 (658.84)
Data are testing session and sequence condition (repeated versus random) for the M1 ROI. Mean (standard error).
4585.71 (695.18) 0.04 (0.01) 5890.80 (550.28)
6161.20 (741.31)
0.04 (0.01) −.022
0.09 (0.01)
−.123 0.06 (0.01) 5562.43 (434.30) 4341.86 (701.06) 0.05 (0.01) 6149.20 (781.11)
6452.60 (395.75)
0.04 (0.01) −.024
0.07 (0.02)
−.091 0.09 (0.02) 5475.00 (696.70) 0.11 (0.02) 4556.71 (689.54) .735 0.06 (0.01) 0.06 (0.01) 5850.20 (549.89)
890.00 (437.47)
0.09 (0.01) 5428.71 (431.13) 4518.29 (701.80) 0.07 (0.01)
Repeated sequence fMRI #1 Repeated sequence fMRI #2 retention Random sequence fMRI #1 Random sequence fMRI #2 retention
6289.40 (734.46)
6452.60 (395.75)
0.06 (0.01) −.012
0.07 (0.01)
GP LI PSC
GP contralesional hemisphere
Volume (l) PSC
GP ipsilesional hemisphere
Volume (l)
SP LI PSC
SP contralesional hemisphere
Volume (l) Volume (l)
PSC
SP ipsilesional hemisphere
(Repeated, Random) ANOVA with a repeated measures correction and change in LI as the dependent measure determined whether motor learning would differentially alter the pattern of M1 activity across the hemispheres. Within groups, sequence-specific motor learning was shown by the SP group via significantly faster RTs (p = .005) and MTs (p = .001; Fig. 2B and C) for repeated as compared to random sequences at the retention test; no difference in repeated versus random sequences was demonstrated by the GP group (RT p = .263; MT p = .335). Between-group comparisons confirmed the beneficial nature of task-specific practice for task performance. The SP group made more change in RT and MT at retention as compared to the GP group (RT: F(1,16) = 11.041, p = .004; MT: F(1,16) = 6.631, p = .020). Motor learning associated with task-specific practice of the repeated sequence induced positive change in the LI, shifting it towards 1.0, for the SP but not the GP group during repeated sequence performance (Group by Sequence interaction F(1,16) = 5.345, p = .034; Fig. 2D). Change in LI for the SP group during the performance of learned repeated sequences resulted from reduced volume of activity in the contralesional cortex at the retention test fMRI session (Table 2), rather than shifts in volume or PSC in the ipsilesional cortex. Neither task specific (SP group) or general practice (GP group) altered the LI for random sequences (Group by Sequence interaction p = .857; Table 2). Several factors limited previous studies examining brain activation patterns after stroke. Critically, the network sub-serving motor performance is differentially engaged by task and level of performance. Thus, networks activated during initial performance are not necessarily those that underpin the recovery of skilled movement or learning. Because previous work has either examined single points in time [18], or grossly related an intervention with altered brain activity [5], the direct relationship between skill acquisition and changes in brain activation patterns after stroke remain unclear. Our main aim was to determine whether motor skill learning using the hemiparetic UE alters contralesional cortical activation differentially for repeated versus random sequences after stroke. We confirmed that this is the case showing that motor learning of repeated sequences reduces contralesional M1 activity. Practice of random sequences by either the SP or GP group did not alter contralesional cortical activity. Increasing non-specific arm use did not shift the pattern of activity within M1 during the ST task. Together, these findings support our contention that skill acquisition can stimulate neuroplastic change in contralesional M1 after stroke. The phenomenon of contralesional cortical activation acutely after stroke is common regardless of lesion location [7,11]. In some individuals contralesional cortical activity diminishes across the first months after injury, while in others this abnormal pattern persists [11], or worsens [17]. Decreased contralesional cortical activation with recovery may relate to physiologic recovery. Feydy et al. (2002) reported “re-focusing” of brain activation to the ipsilesional hemisphere in 8 of 10 patients without M1 injury and persistence of contralesional cortical activation in 3 of 4 individuals with M1 damage [11]. However, these findings were made in the absence of any intervention. The present study extends this work by demonstrating that “re-focusing” can be stimulated by sequencespecific motor skill learning even in individuals with chronic stroke. Hypotheses explaining contralesional M1 activation after stroke center on the concept that activation in the intact contralesional cortex reflects physiologic changes across interactive brain systems. In this case, the non-injured M1 may aid hemiparetic UE control either via ipsilateral projections [3], or through lost inhibition from ipsilesional M1 acting through transcallosal connections [27]. In fact, motor deficits associated with stroke may be worsened by increased transcallosal inhibition from contralesional M1 back onto ipsilesional M1 [10]. An alternate, and not mutually
−.091
L.A. Boyd et al. / Neuroscience Letters 482 (2010) 21–25
Table 2 Volume, PSC and LI for the ipsilesional and contralesional cortices.
24
L.A. Boyd et al. / Neuroscience Letters 482 (2010) 21–25
exclusive, explanation centers on the possibility that contralesional M1 activation after stroke represents an emerging compensatory strategy in response to perceived task difficulty [26]. Our data cannot distinguish between physiologic explanations; however, we do demonstrate that contralesional M1 activity after stroke is modifiable by motor learning. Converging lines of research support the hypothesis that contralesional M1 activity might be in part stimulated at by task difficulty. Normally, motor activation shifts from the contralateral hemisphere to become somewhat bilateral during complex tasks [25]. More specifically, activation in bilateral SMA and PMC correlate with task difficulty in neurologically intact subjects [16]. Across practice of our experimental motor task, decreased contralesional M1 activation for repeated sequences (which may be learned) but not for random sequences (which cannot be learned) suggests that activity in the intact contralesional cortex is modifiable by skill acquisition. It has become apparent that changes in motor behavior can alter representational maps in the motor cortex corresponding to the side being moved [19,20]. It is less clear which specific aspects of motor behavior stimulate such change, and whether these effects are similar for both the contra- and ipsilesional cortices. In animal models repetitive motor activity alone does not stimulate reorganization of ipsilesional cortical maps [21]. This finding led to a “learning dependent hypothesis” of cortical plasticity [24]. Our data suggest that specificity of motor learning is a key factor in this process. In fact, we speculate that if we had tested the tasks practiced by our GP group in the scanner we would likely show the opposite pattern of results; change in LI for the GP group. Together, the data presented in the present work further extend our understanding of the relationship between neuroplastic change and motor skill acquisition, and suggest that specificity of task practice that leads to motor learning after stroke is important for altering contralesional patterns of brain activity. Acknowledgement and funding This work was funded by a Scientist Development Grant (0530022N) to LAB from the American Heart Association. References [1] L.A. Boyd, J.D. Edwards, C.S. Siengsukon, E.D. Vidoni, B.D. Wessel, M.A. Linsdell, Motor sequence chunking is impaired by basal ganglia stroke, Neurobiology of Learning and Memory 92 (2009) 35–44. [2] L.A. Boyd, C.J. Winstein, Impact of explicit information on implicit motorsequence learning following middle cerebral artery stroke, Physical Therapy 83 (2003) 976–989. [3] P. Brodal, The ponto-cerebellar projection in the rhesus monkey: an experimental study with retrograde axonal transport of horseradish peroxidase, Neuroscience 4 (1979) 193–208. [4] C.M. Butefisch, M. Wessling, J. Netz, R.J. Seitz, V. Homberg, Relationship between interhemispheric inhibition and motor cortex excitability in subacute stroke patients, Neurorehabilitation and Neural Repair 22 (2008) 4–21. [5] J.R. Carey, T.J. Kimberley, S.M. Lewis, E.J. Auerbach, L. Dorsey, P. Rundquist, K. Ugurbil, Analysis of fMRI and finger tracking training in subjects with chronic stroke, Brain 125 (2002) 773–788. [6] R.W. Cox, AFNI: software for analysis and visualization of functional magnetic resonance neuroimages, Computers and Biomedical Research 29 (1996) 162–173. [7] S.C. Cramer, K.R. Crafton, Changes in lateralization and somatotopic organization after cortical stroke, Stroke 35 (2004) 240.
25
[8] S.C. Cramer, G. Nelles, R.R. Benson, J.D. Kaplan, R.A. Parker, K.K. Kwong, D.N. Kennedy, S.P. Finklestein, B.R. Rosen, A functional MRI study of subjects recovered from hemiparetic stroke, Stroke 28 (1997) 2518–2527. [9] P. Dassonville, S.M. Lewis, X.H. Zhu, K. Ugurbil, S.G. Kim, J. Ashe, The effect of stimulus-response compatibility on cortical motor activation, NeuroImage 13 (2001) 1–14. [10] J. Duque, F. Hummel, P. Celnik, N. Murase, R. Mazzocchio, L.G. Cohen, Transcallosal inhibition in chronic subcortical stroke, NeuroImage 28 (2005) 940–946. [11] A. Feydy, R. Carlier, A. Roby-Brami, B. Bussel, F. Cazalis, L. Pierot, Y. Burnod, M.A. Maier, Longitudinal study of motor recovery after stroke – recruitment and focusing of brain activation, Stroke 33 (2002) 1610–1617. [12] A.R. Fugl-Meyer, L. Jaasko, I. Leyman, S. Olsson, S. Steglind, The post-stroke hemiplegic patient: a method for evaluation of physical performance, Scandanivan Journal of Rehabilitation 7 (1975) 13–31. [13] G.E. Gresham, P.W. Duncan, W.B. Stason, H.P. Adams, A.M. Adelman, D.N. Alexander, D.S. Bishop, L. Diller, N.E. Donaldson, C.V. Granger, A.L. Holland, M. Kelly-Hayes, F.H. McDowell, L. Myers, M.A. Phipps, E.J. Roth, H.C. Siebens, G.A. Tarven, C.A Trombly, Post-Stroke Rehabilitation, U.S. Department of Health and Human Services, Public Health Service, Agency for Health Care Policy and Research, Rockville, MD, 1995. [14] H.S. Jorgensen, H. Nakayama, H.O. Raaschou, J. Vivelarsen, M. Stoier, T.S. Olsen, Outcome and time-course of recovery in stroke: 2. Time-course of recovery – the Copenhagen stroke study, Archives of Physical Medicine and Rehabilitation 76 (1995) 406–412. [15] L. Kalra, P. Crome, The role of prognostic scores in targeting stroke rehabilitation in elderly patients, Journal of the American Geriatrics Society 41 (1993) 396–400. [16] C.D. Marsden, L. Deecke, H.J. Freund, M. Hallett, R. Passingham, H. Shibasaki, J. Tanji, M. Wiesendanger, The Functions of the Supplementary Motor Area. Supplementary Sensorimotor Area, Lippincott-Raven, Philadelphia, 1996, pp. 477–487. [17] G. Nelles, G. Spiekermann, M. Jueptner, G. Leonhardt, S. Muller, H. Gerhard, H.C. Diener, Evolution of functional reorganization in hemiplegic stroke: a serial positron emission tomographic activation study, Annals of neurology 46 (1999) 901–909. [18] G. Nelles, G. Spiekermann, M. Jueptner, G. Leonhardt, S. Muller, H. Gerhard, H.C. Diener, Reorganization of sensory and motor systems in hemiplegic stroke patients – a positron emission tomography study, Stroke 30 (1999) 1510–1516. [19] D.A. Nowak, C. Grefkes, M. Ameli, G.R. Fink, Interhemispheric competition after stroke: brain stimulation to enhance recovery of function of the affected hand, Neurorehabilitation and Neural Repair 23 (2009) 641–656. [20] R.J. Nudo, G.W. Milliken, W.M. Jenkins, M.M. Merzenich, Use-dependent alterations of movement representations in primary motor cortex of adult squirrel monkeys, Journal of Neuroscience 16 (1996) 785–807. [21] R.J. Nudo, E.J. Plautz, G.W. Milliken, Adaptive plasticity in primate motor cortex as a consequence of behavioral experience and neuronal injury, Seminars in Neuroscience 9 (1997) 13–23. [22] R.J. Nudo, B.M. Wise, F. SiFuentes, G.W. Milliken, Neural substrates for the effects of rehabilitative training on motor recovery after ischemic infarct, Science 272 (1996) 1791–1794. [23] R. Pineiro, S. Pendlebury, H. Johansen-Berg, P.M. Matthews, Functional MRI detects posterior shifts in primary sensorimotor cortex activation after stroke – evidence of local adaptive reorganization? Stroke 32 (2001) 1134–1139. [24] E.J. Plautz, G.W. Milliken, R.J. Nudo, Effects of repetitive motor training on movement representations in adult squirrel monkeys: role of use versus learning, Neurobiology of Learning and Memory 74 (2000) 27–55. [25] H. Shibasaki, N. Sadato, H. Lyshkow, Y. Yonekura, M. Honda, T. Nagamine, S. Suwazono, Y. Mageta, A. Ikeda, M. Miyazaki, H. Fukuyama, R. Asato, J. Konishi, Both primary motor cortex and supplementary motor area play an important role in complex finger movement, Brain 116 (1993) 1387–1398. [26] H. Shibasaki, N. Sadato, H. Lyskow, Y. Yonekura, M. Honda, Both primary motor cortex and supplementary motor areas play an important role in complex finger movement, Brain 116 (2002) 1387–1398. [27] T. Shimizu, A. Hosaki, T. Hino, M. Sato, T. Komori, S. Hirai, P.M. Rossini, Motor cortical disinhibition in the unaffected hemisphere after unilateral cortical stroke, Brain 125 (2002) 1896–1907. [28] G.T. Thielman, C.M. Dean, A.M. Gentile, Rehabilitation of reaching after stroke: task-related training versus progressive resistive exercise, Archives of physical medicine and rehabilitation 85 (2004) 1613–1618. [29] A.C. Zemke, P.J. Heagerty, C. Lee, S.C. Cramer, Motor cortex organization after stroke is related to side of stroke and level of recovery, Stroke 34 (2003) e23–e28.
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Motor learning after stroke: Is skill acquisition a prerequisite for contralesional neuroplastic change? Lara A. Boyd a,b,∗ , Eric D. Vidoni c , Brenda D. Wessel a a b c
Department of Physical Therapy, University of British Columbia, Vancouver, British Columbia, Canada Brain Research Centre, University of British Columbia, Vancouver, British Columbia, Canada Department of Neurology, University of Kansas Medical Center, Kansas City, KS, United States
a r t i c l e
i n f o
Article history: Received 4 November 2009 Received in revised form 27 April 2010 Accepted 29 June 2010 Keywords: Stroke Learning Motor cortex Rehabilitation fMRI
a b s t r a c t Limited data directly characterize the dynamic evolution of brain activity associated with motor learning after stroke. The current study considered whether sequence-specific motor skill learning or increasing non-specific use of the hemiparetic upper extremity drive functional reorganization of the contralesional motor cortex after stroke. Eighteen individuals with chronic middle cerebral artery stroke practiced one of two novel motor tasks; a retention test occurred on a separate fifth day. Using the hemiparetic arm, participants performed a serial targeting task during two functional MRI scans (day one and retention). Participants were randomized into either a task-specific group, who completed three additional sessions of serial targeting practice, or a general arm use group, who underwent three training sessions of increased but non-task specific use of the hemiparetic arm. Both groups performed a repeated sequence of responses that may be learned, and random sequences of movement, which cannot be learned. Change in reaction and movement time for the repeated sequence indexed motor learning; shifts in the laterality index (LI) within primary motor cortex (M1) for repeated and random sequences illustrated training effects on brain activity. Task-specific practice of the repeated sequence facilitated motor learning and shifted the LI for M1 as shown by a reduced volume of contralesional cortical activity. Random sequence performance did not stimulate motor learning or alter the LI within the task-specific training group. Further, between-group comparisons showed that increasing general arm use did not induce motor learning or alter brain activity for either random or repeated sequences. Motor skill learning of a repeated sequence altered cortical activation by inducing a more normal, contralateral pattern of brain activation. Our data suggest that task-specific motor learning may be an important stimulant for neuroplastic change and can remediate maladaptive patterns of brain activity after stroke. © 2010 Elsevier Ireland Ltd. All rights reserved.
How the brain compensates for damage after stroke is unclear, in part because the relationship between behavior and brain function are not well understood. One mechanism among many, that may help explain recovery after stroke is experience-dependent cortical plasticity [22], yet no pattern of motor-related brain activation has emerged that explains improved function. Some neuroplastic changes after stroke are helpful for recovery; others, such as bilateral cortical activity during unimanual movements, have uncertain value [7,11]. More than 40% of individuals with stroke have chronically impaired upper extremity (UE) function [13]. In the healthy brain, movement in one UE is associated with activity in the contralateral
∗ Corresponding author at: University of British Columbia, 212-2277 Wesbrook Mall, Vancouver, British Columbia V6T 1Z3, Canada. Tel.: +1 604 822 7197; fax: +1 604 822 1860. E-mail address: [email protected] (L.A. Boyd). 0304-3940/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.neulet.2010.06.082
primary motor cortex (M1). However, it is common for bilateral M1 to be activated during UE movements after both cortical [7,11] and subcortical stroke [23]. Recent work suggests that bilateral activity in M1 after stroke may relate to altered interhemispheric excitability and a loss of transcallosal inhibition from the ipsi- to the contralesional cortex [4]. One metric that captures this maladaptive intercortical relationship is the laterality index (LI), which is often presented as a ratio of the volume or intensity of activity in the motor cortices [29]. Despite the abnormal patterns of neural activation that occur [7], it is clear that stroke does not abolish motor learning ability [1]. Robust demonstrations of motor learning and improved function have been consistently shown in individuals with chronic stroke [1], suggesting that the functional organization of the motor system can be modified by use. However, our understanding of brain function is limited by an absence of data that directly characterizes the dynamic evolution of brain activity associated with motor learning after stroke. Understanding the conditions that normalize patterns
22
L.A. Boyd et al. / Neuroscience Letters 482 (2010) 21–25
Table 1 Participant characteristics. Group
Gender
Age
Affected side
Hand dominance
MMSE
UE Fugl-Meyer
Orpington
SP
5M 4F
62.7
3R 6L
8R 1L
28.3 (1.8)
58 (8.5)
2.0 (.37)
GP
7M 2F
61.1
2R 7L
9 R
28.8 (1.2)
58 (6.5)
2.0 (.39)
Mimi-mental status exam (MMSE) range 0–30, score >24 indicates absence of dementia; UE motor portion of the Fugl-Meyer range 0–66, higher score indicates less impairment; Orpington prognostic scale, range 1.6–6.8, higher scores indicate severer stroke. Mean (standard deviation).
of brain activity may be important as contralesional cortical activity may stimulate increased transcallosal inhibition back onto the ipsilesional motor cortex [10] and represent a source of motor control deficits after stroke. In the current study we begin to resolve this issue by simultaneously mapping changes in motor-related cortical activation and task-related behavioral improvements that reflect motor learning. Specifically, we considered whether sequencespecific motor learning, non-specific motor learning, or increased use of the hemiparetic UE could drive functional brain reorganization. We hypothesized that contralesional cortical activity in the primary motor cortex (M1) is modifiable and would be reduced by sequence-specific motor skill learning as evidenced by change in the laterality index [8], derived from functional magnetic resonance imaging (fMRI). To test our hypothesis we compared changes in behavior and brain activity (1) within a group that practiced both repeated and random sequences of movement, and (2) between a group who received task-specific training for repeated and random sequences and another that underwent increased non-task specific use of the hemiparetic UE. Eighteen individuals with a first middle cerebral artery stroke who met the inclusion criteria for functional MRI participated. To avoid confounding changes in brain activation patterns associated with learning with physiologic recovery, we tested individuals who were at least 6 months post-stroke [14]. All were free from neglect, aphasia, hemianopsia, and dementia. We categorized arm function and stroke severity the Fugl-Meyer [12] and Orpington scales [15] (Table 1). Participants were randomly assigned into either a task-specific training group (SP) or an increased general arm use group (GP). Independent t-tests verified that there were no between-group differences in age, arm function or stroke severity. Five sessions were completed on separate days over a two-week period. Using the hemiparetic UE, all participants performed a serial targeting (ST) task during functional MRI (fMRI) scanning on day 1 and at a retention test on day 5; three intervention sessions were performed on days 2–4. Individuals in the SP group practiced the ST task over 3 days; those in the GP group performed the same number of hemiparetic arm movements but in a non-specific fashion (details below). Institutional review boards at University of Kansas Medical Center and University of British Columbia approved all procedures. To test individuals without fractionated finger movement, ST task responses were made using a non-ferrous joystick (Resonance Technology) in combination with a task that did not rely on finger or wrist movements. When cued, participants made flexion/extension movements of the stroke-affected elbow and shoulder to guide a cursor (Fig. 1A) to one of three 2 cm targets located equidistant (10 cm) from the start position when it was highlighted in yellow. To ensure that movements were made to, and not through the target, participants held their cursor inside the highlighted target for 500 ms (Fig. 1B) before being cued to return to home (Fig. 1C). Following a short delay (200 ms), the next cue to move (new target highlighted) was delivered. All participants were instructed to respond as fast and accurately as possible. Joystick position sampling and all stimuli were presented at 40 Hz
using custom software developed in LabView (National Instruments Co.). Participants alternated between responding to a repeated sequence of targets (8-elements long) and random sequences. Using the same joystick on the practice days (2–4), the SP group performed the repeating sequence 40 times (320 responses); 80 random responses were also completed (400 total movements per day). Comparing performance from the random and repeated sequences allowed the separation of the effects of nonspecific learning from those associated sequence-specific learning [2]. Participants in the GP group underwent 3 days of increased use of their hemiparetic UE not specific to the ST task. The intervention was delivered by a physical therapist. Non-specific training tasks were patterned on prior reports [28] and to match the experimental ST task, also relied on elbow and shoulder flexion and extension. To mimic the movements of the SP group, the GP group made equidistant movements to three targets to accomplish bean-bag pushing, cone stacking and erasing; movement excursion was restrained to maintain similarity with the ST task. In addition the number of movements during the GP intervention (400/day) was matched to the number made by the SP group. Participants in the GP group were instructed to be as fast and accurate as possible. To address the possibility that contralesional cortical activity was the result of bilateral, “mirror” movements participants were screened for their ability to move the hemiparetic UE without movement of the opposite UE. During practice, surface electromyographic recordings of the flexor carpi ulnaris, extensor carpi radialis, posterior and anterior deltoid from the non-stroke UE verified the absence of mirror movements in the non-hemiparetic arm. The same motor task was performed at both fMRI sessions (days 1 and 5); each trial took 6 s to complete. Participants lay supine, with their elbow flexed to 45◦ and forearm in a resting position on their stomach while gripping the non-ferrous joystick. The joystick sent behavioral data to a computer. Before the first session, all participants were familiarized with the motor task. Visual stimuli were presented using a Resonance Technologies VisuaStim goggle system. A 3.0 T head only scanner equipped with a three-axis local gradient radio frequency coil collected whole brain fMRI (29 axial, 4 mm slices, skip 0.5 mm). Functional imaging data were collected as axial echo-planar images, using a single-shot, blipped gradientecho echo-planar pulse sequence (TE = 30 ms, TR = 2.0 s, 90◦ flip angle, FOV = 256 mm, 64 × 64). Four, 6 min 42 s runs of functional data were completed (201 images). Each run contained three periods of rest (19 images), a block of tracking to randomly highlighted targets (72 images), and another block of tracking following the repeating sequence (72 images). Prior to functional imaging, highresolution 3D T1 images were collected for anatomic localization and co-registration (TE = 5 ms, TR = 24 ms, 40◦ flip angle, NEX = 1, thickness = 1.2 mm, FOV = 256 mm, 256 × 256). Mean reaction time (RT) and movement time (MT) were calculated for the random and repeating sequences. RT was the time from target highlight to the beginning of participant’s response (movement onset). MT was movement onset to target hit. A change score (day 5 to day 1) was calculated for RT and MT to index the effect
L.A. Boyd et al. / Neuroscience Letters 482 (2010) 21–25
23
Fig. 1. Experimental task: (A) start position; (B) participant moves to and holds on target; and (C) cued to return to home.
of practice condition on responding; separate change scores were computed for repeated and random sequences. All fMRI data processing was performed using Analysis of Functional NeuroImages software [6]. Functional images were generated by condition (Rest, Random, Repeated), spatially registered to correct for head motion, and smoothed using a 4 mm fullwidth-half-maximum Gaussian kernel. A boxcar function modeled reactions, producing estimates of the blood oxygen level dependent response relative to baseline. To reduce the impact of any head motion signal predictors of no interest were included to account for translational and rotational motion in the x, y, and z planes. A region of interest (ROI) was defined using structural landmarks for M1 in both hemispheres for individual participants [9]. The area of M1 was defined as extending from the anterior bank of the central sulcus to the anterior edge of the precentral gyrus [9]. Within M1, ROI brain activation was represented by percent signal change (PSC), calculated from fMRI signal relative to baseline (Fig. 2A). Using the product of PSC and volume (in l) of significantly activated
voxel clusters within our M1 ROI, the laterality index (LI) [(contralateral − ipsilateral)/(contralateral + ipsilateral)] was calculated for each hemisphere by condition (Repeated, Random) [8]. The LI produces a range from −1 to +1, with a negative number indicating primarily ipsilateral activation and a positive number indicating activation primarily contralateral to the UE performing the task. We first assessed the impact of our practice conditions on motor learning. Sequence-specific motor learning was indexed using paired t-tests to compare within group change in RT for repeated versus random sequences at the retention test. The same test was repeated with MT as the dependent measure; a p-value of .025 was employed to account for multiple comparisons. To verify differences between practice conditions a multivariate ANOVA indexed the amount of between-group change in RT and MT. After establishing that sequence-specific motor learning occurred for the repeated sequence in the SP group, we tested our hypothesis that contralesional cortical activity is modifiable by motor skill acquisition as evidenced by change in the LI. A Group (SP, GP) by Sequence
Fig. 2. (A) M1 ROIs pre and post 3 days of practice. Less bilateral M1 activation was noted after practice in the SP but not the GP participant (p > .05). I = ipsilesional corresponding to the arm used; C = contralesional. (B) RT Change. (C) MT Change. (D) LI change. Learned repeated sequences by the SP group showed shifts in LI towards ipsilesional M1 because of reduced contralesional activity (1.0 = ipsilesional M1 activation).
−.087 0.08 (0.01) 5468.43 (658.84)
Data are testing session and sequence condition (repeated versus random) for the M1 ROI. Mean (standard error).
4585.71 (695.18) 0.04 (0.01) 5890.80 (550.28)
6161.20 (741.31)
0.04 (0.01) −.022
0.09 (0.01)
−.123 0.06 (0.01) 5562.43 (434.30) 4341.86 (701.06) 0.05 (0.01) 6149.20 (781.11)
6452.60 (395.75)
0.04 (0.01) −.024
0.07 (0.02)
−.091 0.09 (0.02) 5475.00 (696.70) 0.11 (0.02) 4556.71 (689.54) .735 0.06 (0.01) 0.06 (0.01) 5850.20 (549.89)
890.00 (437.47)
0.09 (0.01) 5428.71 (431.13) 4518.29 (701.80) 0.07 (0.01)
Repeated sequence fMRI #1 Repeated sequence fMRI #2 retention Random sequence fMRI #1 Random sequence fMRI #2 retention
6289.40 (734.46)
6452.60 (395.75)
0.06 (0.01) −.012
0.07 (0.01)
GP LI PSC
GP contralesional hemisphere
Volume (l) PSC
GP ipsilesional hemisphere
Volume (l)
SP LI PSC
SP contralesional hemisphere
Volume (l) Volume (l)
PSC
SP ipsilesional hemisphere
(Repeated, Random) ANOVA with a repeated measures correction and change in LI as the dependent measure determined whether motor learning would differentially alter the pattern of M1 activity across the hemispheres. Within groups, sequence-specific motor learning was shown by the SP group via significantly faster RTs (p = .005) and MTs (p = .001; Fig. 2B and C) for repeated as compared to random sequences at the retention test; no difference in repeated versus random sequences was demonstrated by the GP group (RT p = .263; MT p = .335). Between-group comparisons confirmed the beneficial nature of task-specific practice for task performance. The SP group made more change in RT and MT at retention as compared to the GP group (RT: F(1,16) = 11.041, p = .004; MT: F(1,16) = 6.631, p = .020). Motor learning associated with task-specific practice of the repeated sequence induced positive change in the LI, shifting it towards 1.0, for the SP but not the GP group during repeated sequence performance (Group by Sequence interaction F(1,16) = 5.345, p = .034; Fig. 2D). Change in LI for the SP group during the performance of learned repeated sequences resulted from reduced volume of activity in the contralesional cortex at the retention test fMRI session (Table 2), rather than shifts in volume or PSC in the ipsilesional cortex. Neither task specific (SP group) or general practice (GP group) altered the LI for random sequences (Group by Sequence interaction p = .857; Table 2). Several factors limited previous studies examining brain activation patterns after stroke. Critically, the network sub-serving motor performance is differentially engaged by task and level of performance. Thus, networks activated during initial performance are not necessarily those that underpin the recovery of skilled movement or learning. Because previous work has either examined single points in time [18], or grossly related an intervention with altered brain activity [5], the direct relationship between skill acquisition and changes in brain activation patterns after stroke remain unclear. Our main aim was to determine whether motor skill learning using the hemiparetic UE alters contralesional cortical activation differentially for repeated versus random sequences after stroke. We confirmed that this is the case showing that motor learning of repeated sequences reduces contralesional M1 activity. Practice of random sequences by either the SP or GP group did not alter contralesional cortical activity. Increasing non-specific arm use did not shift the pattern of activity within M1 during the ST task. Together, these findings support our contention that skill acquisition can stimulate neuroplastic change in contralesional M1 after stroke. The phenomenon of contralesional cortical activation acutely after stroke is common regardless of lesion location [7,11]. In some individuals contralesional cortical activity diminishes across the first months after injury, while in others this abnormal pattern persists [11], or worsens [17]. Decreased contralesional cortical activation with recovery may relate to physiologic recovery. Feydy et al. (2002) reported “re-focusing” of brain activation to the ipsilesional hemisphere in 8 of 10 patients without M1 injury and persistence of contralesional cortical activation in 3 of 4 individuals with M1 damage [11]. However, these findings were made in the absence of any intervention. The present study extends this work by demonstrating that “re-focusing” can be stimulated by sequencespecific motor skill learning even in individuals with chronic stroke. Hypotheses explaining contralesional M1 activation after stroke center on the concept that activation in the intact contralesional cortex reflects physiologic changes across interactive brain systems. In this case, the non-injured M1 may aid hemiparetic UE control either via ipsilateral projections [3], or through lost inhibition from ipsilesional M1 acting through transcallosal connections [27]. In fact, motor deficits associated with stroke may be worsened by increased transcallosal inhibition from contralesional M1 back onto ipsilesional M1 [10]. An alternate, and not mutually
−.091
L.A. Boyd et al. / Neuroscience Letters 482 (2010) 21–25
Table 2 Volume, PSC and LI for the ipsilesional and contralesional cortices.
24
L.A. Boyd et al. / Neuroscience Letters 482 (2010) 21–25
exclusive, explanation centers on the possibility that contralesional M1 activation after stroke represents an emerging compensatory strategy in response to perceived task difficulty [26]. Our data cannot distinguish between physiologic explanations; however, we do demonstrate that contralesional M1 activity after stroke is modifiable by motor learning. Converging lines of research support the hypothesis that contralesional M1 activity might be in part stimulated at by task difficulty. Normally, motor activation shifts from the contralateral hemisphere to become somewhat bilateral during complex tasks [25]. More specifically, activation in bilateral SMA and PMC correlate with task difficulty in neurologically intact subjects [16]. Across practice of our experimental motor task, decreased contralesional M1 activation for repeated sequences (which may be learned) but not for random sequences (which cannot be learned) suggests that activity in the intact contralesional cortex is modifiable by skill acquisition. It has become apparent that changes in motor behavior can alter representational maps in the motor cortex corresponding to the side being moved [19,20]. It is less clear which specific aspects of motor behavior stimulate such change, and whether these effects are similar for both the contra- and ipsilesional cortices. In animal models repetitive motor activity alone does not stimulate reorganization of ipsilesional cortical maps [21]. This finding led to a “learning dependent hypothesis” of cortical plasticity [24]. Our data suggest that specificity of motor learning is a key factor in this process. In fact, we speculate that if we had tested the tasks practiced by our GP group in the scanner we would likely show the opposite pattern of results; change in LI for the GP group. Together, the data presented in the present work further extend our understanding of the relationship between neuroplastic change and motor skill acquisition, and suggest that specificity of task practice that leads to motor learning after stroke is important for altering contralesional patterns of brain activity. Acknowledgement and funding This work was funded by a Scientist Development Grant (0530022N) to LAB from the American Heart Association. References [1] L.A. Boyd, J.D. Edwards, C.S. Siengsukon, E.D. Vidoni, B.D. Wessel, M.A. Linsdell, Motor sequence chunking is impaired by basal ganglia stroke, Neurobiology of Learning and Memory 92 (2009) 35–44. [2] L.A. Boyd, C.J. Winstein, Impact of explicit information on implicit motorsequence learning following middle cerebral artery stroke, Physical Therapy 83 (2003) 976–989. [3] P. Brodal, The ponto-cerebellar projection in the rhesus monkey: an experimental study with retrograde axonal transport of horseradish peroxidase, Neuroscience 4 (1979) 193–208. [4] C.M. Butefisch, M. Wessling, J. Netz, R.J. Seitz, V. Homberg, Relationship between interhemispheric inhibition and motor cortex excitability in subacute stroke patients, Neurorehabilitation and Neural Repair 22 (2008) 4–21. [5] J.R. Carey, T.J. Kimberley, S.M. Lewis, E.J. Auerbach, L. Dorsey, P. Rundquist, K. Ugurbil, Analysis of fMRI and finger tracking training in subjects with chronic stroke, Brain 125 (2002) 773–788. [6] R.W. Cox, AFNI: software for analysis and visualization of functional magnetic resonance neuroimages, Computers and Biomedical Research 29 (1996) 162–173. [7] S.C. Cramer, K.R. Crafton, Changes in lateralization and somatotopic organization after cortical stroke, Stroke 35 (2004) 240.
25
[8] S.C. Cramer, G. Nelles, R.R. Benson, J.D. Kaplan, R.A. Parker, K.K. Kwong, D.N. Kennedy, S.P. Finklestein, B.R. Rosen, A functional MRI study of subjects recovered from hemiparetic stroke, Stroke 28 (1997) 2518–2527. [9] P. Dassonville, S.M. Lewis, X.H. Zhu, K. Ugurbil, S.G. Kim, J. Ashe, The effect of stimulus-response compatibility on cortical motor activation, NeuroImage 13 (2001) 1–14. [10] J. Duque, F. Hummel, P. Celnik, N. Murase, R. Mazzocchio, L.G. Cohen, Transcallosal inhibition in chronic subcortical stroke, NeuroImage 28 (2005) 940–946. [11] A. Feydy, R. Carlier, A. Roby-Brami, B. Bussel, F. Cazalis, L. Pierot, Y. Burnod, M.A. Maier, Longitudinal study of motor recovery after stroke – recruitment and focusing of brain activation, Stroke 33 (2002) 1610–1617. [12] A.R. Fugl-Meyer, L. Jaasko, I. Leyman, S. Olsson, S. Steglind, The post-stroke hemiplegic patient: a method for evaluation of physical performance, Scandanivan Journal of Rehabilitation 7 (1975) 13–31. [13] G.E. Gresham, P.W. Duncan, W.B. Stason, H.P. Adams, A.M. Adelman, D.N. Alexander, D.S. Bishop, L. Diller, N.E. Donaldson, C.V. Granger, A.L. Holland, M. Kelly-Hayes, F.H. McDowell, L. Myers, M.A. Phipps, E.J. Roth, H.C. Siebens, G.A. Tarven, C.A Trombly, Post-Stroke Rehabilitation, U.S. Department of Health and Human Services, Public Health Service, Agency for Health Care Policy and Research, Rockville, MD, 1995. [14] H.S. Jorgensen, H. Nakayama, H.O. Raaschou, J. Vivelarsen, M. Stoier, T.S. Olsen, Outcome and time-course of recovery in stroke: 2. Time-course of recovery – the Copenhagen stroke study, Archives of Physical Medicine and Rehabilitation 76 (1995) 406–412. [15] L. Kalra, P. Crome, The role of prognostic scores in targeting stroke rehabilitation in elderly patients, Journal of the American Geriatrics Society 41 (1993) 396–400. [16] C.D. Marsden, L. Deecke, H.J. Freund, M. Hallett, R. Passingham, H. Shibasaki, J. Tanji, M. Wiesendanger, The Functions of the Supplementary Motor Area. Supplementary Sensorimotor Area, Lippincott-Raven, Philadelphia, 1996, pp. 477–487. [17] G. Nelles, G. Spiekermann, M. Jueptner, G. Leonhardt, S. Muller, H. Gerhard, H.C. Diener, Evolution of functional reorganization in hemiplegic stroke: a serial positron emission tomographic activation study, Annals of neurology 46 (1999) 901–909. [18] G. Nelles, G. Spiekermann, M. Jueptner, G. Leonhardt, S. Muller, H. Gerhard, H.C. Diener, Reorganization of sensory and motor systems in hemiplegic stroke patients – a positron emission tomography study, Stroke 30 (1999) 1510–1516. [19] D.A. Nowak, C. Grefkes, M. Ameli, G.R. Fink, Interhemispheric competition after stroke: brain stimulation to enhance recovery of function of the affected hand, Neurorehabilitation and Neural Repair 23 (2009) 641–656. [20] R.J. Nudo, G.W. Milliken, W.M. Jenkins, M.M. Merzenich, Use-dependent alterations of movement representations in primary motor cortex of adult squirrel monkeys, Journal of Neuroscience 16 (1996) 785–807. [21] R.J. Nudo, E.J. Plautz, G.W. Milliken, Adaptive plasticity in primate motor cortex as a consequence of behavioral experience and neuronal injury, Seminars in Neuroscience 9 (1997) 13–23. [22] R.J. Nudo, B.M. Wise, F. SiFuentes, G.W. Milliken, Neural substrates for the effects of rehabilitative training on motor recovery after ischemic infarct, Science 272 (1996) 1791–1794. [23] R. Pineiro, S. Pendlebury, H. Johansen-Berg, P.M. Matthews, Functional MRI detects posterior shifts in primary sensorimotor cortex activation after stroke – evidence of local adaptive reorganization? Stroke 32 (2001) 1134–1139. [24] E.J. Plautz, G.W. Milliken, R.J. Nudo, Effects of repetitive motor training on movement representations in adult squirrel monkeys: role of use versus learning, Neurobiology of Learning and Memory 74 (2000) 27–55. [25] H. Shibasaki, N. Sadato, H. Lyshkow, Y. Yonekura, M. Honda, T. Nagamine, S. Suwazono, Y. Mageta, A. Ikeda, M. Miyazaki, H. Fukuyama, R. Asato, J. Konishi, Both primary motor cortex and supplementary motor area play an important role in complex finger movement, Brain 116 (1993) 1387–1398. [26] H. Shibasaki, N. Sadato, H. Lyskow, Y. Yonekura, M. Honda, Both primary motor cortex and supplementary motor areas play an important role in complex finger movement, Brain 116 (2002) 1387–1398. [27] T. Shimizu, A. Hosaki, T. Hino, M. Sato, T. Komori, S. Hirai, P.M. Rossini, Motor cortical disinhibition in the unaffected hemisphere after unilateral cortical stroke, Brain 125 (2002) 1896–1907. [28] G.T. Thielman, C.M. Dean, A.M. Gentile, Rehabilitation of reaching after stroke: task-related training versus progressive resistive exercise, Archives of physical medicine and rehabilitation 85 (2004) 1613–1618. [29] A.C. Zemke, P.J. Heagerty, C. Lee, S.C. Cramer, Motor cortex organization after stroke is related to side of stroke and level of recovery, Stroke 34 (2003) e23–e28.