OUP user menu

Effects of non-invasive cortical stimulation on skilled motor function in chronic stroke

Friedhelm Hummel, Pablo Celnik, Pascal Giraux, Agnes Floel, Wan-Hsun Wu, Christian Gerloff, Leonardo G. Cohen
DOI: http://dx.doi.org/10.1093/brain/awh369 490-499 First published online: 5 January 2005


Stroke is a leading cause of adult motor disability. Despite recent progress, recovery of motor function after stroke is usually incomplete. This double blind, Sham-controlled, crossover study was designed to test the hypothesis that non-invasive stimulation of the motor cortex could improve motor function in the paretic hand of patients with chronic stroke. Hand function was measured using the Jebsen–Taylor Hand Function Test (JTT), a widely used, well validated test for functional motor assessment that reflects activities of daily living. JTT measured in the paretic hand improved significantly with non-invasive transcranial direct current stimulation (tDCS), but not with Sham, an effect that outlasted the stimulation period, was present in every single patient tested and that correlated with an increment in motor cortical excitability within the affected hemisphere, expressed as increased recruitment curves (RC) and reduced short-interval intracortical inhibition. These results document a beneficial effect of non-invasive cortical stimulation on a set of hand functions that mimic activities of daily living in the paretic hand of patients with chronic stroke, and suggest that this interventional strategy in combination with customary rehabilitative treatments may play an adjuvant role in neurorehabilitation.

  • cortical stimulation
  • motor control
  • rehabilitation
  • stroke
  • CS = conditioning stimulus
  • ICF = intracortical facilitation
  • JTT = Jebsen–Taylor Hand Function Test
  • MT = motor thresholds
  • RC = recruitment curves to TMS
  • SICI = short-interval intracortical inhibition
  • tDCS = transcranial direct current stimulation
  • TMS = transcranial magnetic stimulation
  • TS = test stimulus


Stroke is the leading cause of long-term disability among adults in industrialized countries, and is responsible for 2–4% of total health-care expenses (Turney et al., 1984; Whisnant, 1984; Jongbloed, 1986; Broderick et al., 1989; Dobkin, 1995; Taylor et al., 1996). More than 60% of stroke survivors suffer from persistent neurological deficits (Gresham et al., 1975) that impair activities of daily living (i.e. dressing, eating, self-care and personal hygiene) (Gresham et al., 1975; Carod-Artal et al., 2000; Clarke et al., 2002), underlining the need for development of new neurorehabilitative treatments (Nudo et al., 1996; Nudo, 2003).

Recent studies have demonstrated that non-invasive brain stimulation enhances the beneficial effects of motor training on cortical plasticity (Butefisch et al., 2004), visuo-motor coordination (Antal et al., 2004a, b), implicit motor learning (Nitsche et al., 2003c), probabilistic classification learning (Kincses et al., 2004) and analogic reasoning (Boroojerdi et al., 2001b) in healthy volunteers. In animal models, preliminary reports suggested that cortical stimulation could facilitate motor function in animals with focal brain lesions involving the primary motor cortex (Adkins-Muir and Jones, 2003; Kleim et al., 2003; Plautz et al., 2003; Teskey et al., 2003). Thus, it is possible that cortical stimulation could facilitate performance of skilled motor tasks in human stroke patients (Brown et al., 2003; Hummel and Cohen, 2005).

Transcranial direct current stimulation (tDCS) (Nitsche et al., 2003a; Paulus, 2003) is a non-invasive, painless cortical stimulation technique (Nitsche and Paulus, 2000, 2001) that is well tolerated, does not elicit auditory or somatosensory perceptions beyond the initial minute of application (thereby facilitating the design of Sham interventions) (Priori et al., 1998; Nitsche and Paulus, 2000; Nitsche et al., 2003b, c; Hummel et al., 2004) and exerts facilitatory effects on learning processes in healthy volunteers (Nitsche et al., 2003c; Kincses et al., 2004). In this study, we investigated the hypothesis that non-invasive, painless cortical stimulation (tDCS) delivered to the motor cortex of the affected hemisphere could improve performance of motor tasks that mimic activities of daily living in patients with chronic stroke.

Material and methods


Six patients with a history of a single ischaemic cerebral infarct (Table 1) aged 38–84 years (mean ± SE, 62.2 ± 7.56 years; two of them females, all but one right-handed) participated in the study. All gave written informed consent to each experiment according to the Declaration of Helsinki [http://www.wma.net/e/policy/b3.htm (1997)] and the NINDS Institutional Review Board approved the study protocol. Patients were tested at least 1 year after the stroke (3.7 ± 1.1 years, range 1.9–8.9; Table 1). All patients had single ischaemic subcortical strokes leading to initial severe upper arm motor paresis (MRC grade <2) that over time recovered to the point of being able to perform the required motor tasks. Modified Ashworth Scale for Grading Spasticity (Bohannon and Smith, 1987) ranged from 0–2 and upper arm Fugl-Meyer scale ranged from 91% to 99% (Fugl-Meyer et al., 1975). Mini-Mental State Examination (MMSE) ranged between 28 and 30 (Folstein et al., 1975) in both groups. Patients with severe depression, history of severe alcohol or drug abuse, severe language disturbances, particularly of a receptive nature, or serious cognitive deficits (MMSE <23/30 points) were not enrolled in the protocol.

View this table:
Table 1

Patient data

Patient 1Patient 2Patient 3Patient 4Patient 5Patient 6Mean ± SE
Age (years)41723872668462.2 ± 7.56
Time after stroke (months)483431232310744.3 ± 13.1
Lesion siteR frontal operculum, putamen, corona radiata and insulaR basal gangliaL subcortical frontal lobeL subcortical centrum semiovale, basal gangliaR thalamusL basal ganglia
Handedness (EDS)Left (12/50)Right (46/50)Right (49/50)Right (46/50)Right (49/50)Right (47/50)
MMSE30/3029/3029/3030/3028/3029/3029.2 ± 0.31
Motor function:
MRC4. ± 0.03
FMS (%)95969991959695 ± 1.0
Abil-Hand0.860.8810.840.940.970.92 ± 0.03
  • F = female; M = male; R = right; L = left; ASS = Ashworth Spasticity Score; EDS = Edinburgh Handedness Scale; FMS = Fugl-Meyer Scale; MMSE = Mini-Mental State Examination; MRC = scale to determine strength by the Medical Research Council (mean MRC value of the tested muscles).

Experimental design

tDCS and behavioural testing

Initially, all patients participated in a familiarization session (Session 1) in which they practised the Jebsen–Taylor Hand Function Test (JTT) (the endpoint measure of the study) 10 times, sufficient to reach stable motor performance in all individuals. Subsequently, they moved on to the double-blind crossover portion of the study consisting of two (Sessions 2 and 3) counterbalanced sessions (tDCS and Sham) separated by 10.3 ± 2.06 days (mean ± SE). Half of the patients did tDCS first and half did Sham first. All of the patients participated in Sessions 1–3, during which the behavioural measurement (JTT) was determined. Finally, five out of six patients participated in an additional session that tested the effects of tDCS, as administered in the crossover section of the study, on motor cortical excitability to transcranial magnetic stimulation (TMS) (Session 4). Each session started with a questionnaire (see below) followed by the three measurements of baseline JTT (JTT1–3), intervention (tDCS or Sham) and follow-up JTT measurements, with one JTT measurement during (JTT4) and two more after (JTT5–6) the intervention (Fig. 1). On average, JTT6 was tested 26.5 ± 3.4 min (mean ± SE) after the end of each intervention. Additionally, JTT was tested ∼10 days after each intervention (JTT7).

Fig. 1

(A) Experimental design. Patients participated in three sessions. In the first session, they familiarized themselves with the JTT and reached a stable level of performance. The second and third sessions started with questionnaires followed by baseline determinations of JTT (JTT1–3), cortical stimulation (tDCS) or Sham in a counterbalanced double-blind design and later by post-intervention JTT (JTT4–6), with JTT4 determined during stimulation and JTT5–6 after stimulation. Questionnaires (Q) in which patients characterized (self-report on visual analogue scales) level of attention and fatigue during the experiment were given at four different times in each session. A fourth session was included later to test the effects of tDCS on motor cortical function tested with TMS. (B) Subtests of the JTT (Jebsen et al., 1969) included: turning over cards, picking up small objects and placing them in a can, pick up beans with a tea spoon, placing them in a can (mimicking a feeding function), stacking chequers, moving large light cans, and moving heavy cans.

All patients described their level of attention toward the task (range 1–7; 1 = no attention, 7 = highest level of attention) and their perception of fatigue (range 1–7; 1 = highest level of fatigue, 7 = no fatigue; see Q1–4 in Fig. 1) four times in each session, and their sense of discomfort/pain after each session ended (range 1–10; 1 = no discomfort/pain, 10 = maximal discomfort/pain) using visual analogue scales that have good internal consistency, reliability and objectivity (Folstein and Luria, 1973; Gracely, 1999; Chibnall and Tait, 2001; Reisine et al., 2003; Floel et al., 2004). Additionally, after the completion of the study, patients were asked to identify in which session they received ‘real’ cortical stimulation (tDCS). Instructions to the patients were identical for Sessions 2 and 3 (tDCS and Sham).

Jebsen-Taylor Hand Function Test

The JTT is a widely used assessment of functional hand motor skills (Jebsen et al., 1969). It has good validity and reliability, and normative data are available for different ages and both genders (Jebsen et al., 1969; Hackel et al., 1992). We included in this study six of the seven JTT subtests: turning over cards, picking up small objects and placing them in a can, picking up small objects with a teaspoon and placing them in a can (mimicking a feeding function), stacking chequers, moving large light cans, and moving heavy cans (Fig. 1B). Since some patients were unable to perform writing tasks (the seventh JTT subtest) due to dominant hemisphere strokes, we excluded this particular subtest from the study. Patients were instructed to perform the tasks as rapidly and accurately as possible according to written standardized instructions in the testing set (Jebsen et al., 1969; Stern, 1992). Total JTT time and partial subtest JTT times (except for the writing task, which was not included) were recorded for analysis. Feedback on task performance was not provided. Dropping of an object (cards, ‘small objects’, cans) was counted as an accuracy error and analysed off-line.

Non-invasive cortical stimulation

tDCS was delivered through two gel-sponge electrodes (TransQE; IOMED®, Salt Lake City, UT, USA; surface area 25 cm2) embedded in a saline-soaked solution. The anode was positioned on the projection of the hand knob area (Yousry et al., 1997) of the primary motor cortex of the affected hemisphere on the patient's scalp, and the cathode on the skin overlying the contralateral supraorbital region. The hand knob area of the motor cortex was first identified on each patient's MRI and then co-registered to the scalp using a frameless neuronavigation system (Brainsight®; Rogue Research Inc., Montreal, Canada). Stimulating electrodes were centred on the projection of this anatomical site on each patient's scalp. Anodal tDCS was delivered for 20 min in the tDCS session and for up to 30 s in the Sham session using a Phoresor® II Auto (Model No. PM850; IOMED®). At the onset of both interventions (tDCS and Sham), current was increased in a ramp-like fashion (Nitsche et al., 2003a) eliciting a transient tingling sensation on the scalp that faded over seconds, consistent with previous reports (Nitsche et al., 2003c). Current (1 mA) remained on for 20 min in the tDCS session and for up to 30 s in the Sham session. In both sessions, currents were turned off slowly over a few seconds, a procedure that does not elicit perceptions (Nitsche et al., 2003c) and that was implemented out of the field of view of the patients. The investigator testing motor function (JTT) and the patients were blind to the intervention (tDCS or Sham), which was administered by a separate investigator who did not participate in motor testing or data analysis.

tDCS and corticomotor excitability

In session 4, we evaluated the effects of application of tDCS on measures of corticomotor excitability including motor thresholds (MT), recruitment curves to TMS (RC), short-interval intracortical inhibition (SICI) and intracortical facilitation (ICF) (Kujirai et al., 1993; Cohen et al., 1998; Chen, 2000). Measures of corticomotor excitability were obtained immediately before (baseline), immediately after (TMS1) and 25 min following the end (TMS2) of tDCS (post). TMS was delivered from a Magstim 200 (Magstim Co., Whitland, UK) through a figure-of-eight shaped 70-mm coil. Motor evoked potentials (MEP's) were recorded from the first digital interosseus muscle (FDI) of the paretic hand. The optimal coil position for stimulation was determined by moving the coil in 1-cm steps on the scalp to identify the optimal spot for activation of the FDI in the paretic hand. The magnetic coil was held tangentially to the scalp at an angle of 45° to the midline with the handle backwards. Resting MT was defined as the lowest stimulus intensity evoking a MEP of 50 μV in five of 10 trials in the relaxed FDI (Rossini et al., 1994). SICI and ICF were measured using the paired-pulse technique (Kujirai et al., 1993). In brief, a suprathreshold test stimulus adjusted to a MEP amplitude of ∼1 mV (TS; mean 129% ± 3.3 MT) was preceded by a subthreshold conditioning stimulus (CS; 80% MT) at interstimulus intervals of 3 and 10 ms, sampling inhibitory (3 ms, SICI) and excitatory (10 ms, ICF) windows, respectively. Ten stimuli were applied at each interval in a randomized order. MTs were determined separately before and after tDCS; as they did not change, it was not necessary to adjust stimulus intensities. For RC, the stimulation intensity was changed systematically in steps of 10% of the individual's motor threshold, between 100% and 150% MT. For analysis of the RC, MEP amplitudes obtained at different stimulus intensities (100–150% MT) were expressed relative to the MEP amplitude at 100% MT (e.g. RC at 150% MT = MEP amplitude150%MT/MEP amplitude100%MT × 100).

Data analysis

Data were normally distributed as evaluated by Kolmogorov–Smirnov test. Repeated measures ANOVA(RM) was used to evaluate the effects of TIMEBaseline,post and INTERVENTIONtDCS,Sham on total JTT time and the effects of TIMEbaseline,post, INTERVENTIONtDCS,Sham and SUBTESTcards,objects,feeding,chequers,lightcans,heavycans on subtest JTT time. Additionally, we evaluated the effects of TIMEbaseline,post and INTERVENTIONtDCS,Sham on JTT tasks grouped according to predominant reliance on fine distal or more proximal arm function MOTOR CONTROLfine distal,proximal and the effects of INTERVENTIONtDCS,Sham on attention (TIME-QUESTattention(Q1,Q2,Q3,Q4)) and fatigue (TIME-QUESTfatigue(Q1,Q2,Q3,Q4)). Paired t-tests were used to evaluate the effect of INTERVENTIONtDCS,Sham on discomfort/pain. To evaluate the effects of tDCS on motor cortical excitability, ANOVARM with factors TIMEbaseline,post and TMS INTENSITY110%,120%,130%,140%,150% MT was used to compare RC. Paired t-tests were used to evaluate the effect of tDCS on SICI and ICF. Conditioned on significant F-values (P < 0.05), post hoc testing was performed and corrected for multiple comparisons (Bonferroni). JTT changes in percentage and net changes in percentage were calculated according to the following equations (see Table 3): Math All data are expressed as mean ± SE.


Psychophysical data

ANOVARM did not show significant differences of factors TIME-QUEST, INTERVENTIONtDCS,Sham or TIME-QUEST × INTERVENTIONtDCS,Sham interaction on either attention [F(3,16) = 1.11; not significant (ns)] or fatigue [F(3,16) = 1.25; ns] (Table 2A). Discomfort/pain was negligible ranging between 1 and 2 out of 10, and it was comparable in the tDCS and Sham sessions (paired t-test, ns) (Table 2B). All patients were unable to distinguish the tDCS from the Sham session.

View this table:
Table 2A

Fatigue and attention

tDCSShamStatistics ANOVARM
Fatigue (1–7) (mean ± SE)5.2 ± 0.34.9 ± 0.55.2 ± 0.85.0 ± 0.35.1 ± 0.44.7 ± 0.45.1 ± 0.44.8 ± 0.5ns
Attention (1–7) (mean ± SE)5.0 ± 0.34.6 ± 0.34.7 ± 0.35.0 ± 0.44.9 ± 0.34.8 ± 0.44.4 ± 0.54.7 ± 0.5ns
  • See timing of questionnaires (Q) in Fig. 1. Fatigue scale (1–7; 1 = highest level of fatigue; 7 = no fatigue). Attention scale (1–7; 1 = no attention; 7 = highest level of attention to the task). ns = not significant.

View this table:
Table 2B


Patient 111
Patient 211
Patient 311
Patient 411
Patient 521.5
Patient 611
Mean ± SE1.17 ± 0.11.08 ± 0.5
  • Pain/discomfort scale (1–10; 1 = no pain; 10 = strongest imaginable pain).

Effects of non-invasive cortical stimulation on JTT time

Total JTT time improved initially, reaching stable levels in all patients and subtests during the initial familiarization session (Fig. 2). After the familiarization session, baseline total JTT time was comparable immediately preceding both the tDCS and the Sham sessions (Fig. 2A and B). Error rates were comparable (paired t-test, ns) during Sham (2.78 ± 1.90%) and tDCS (4.63 ± 1.71%). ANOVARM revealed a significant interaction TIMEbaseline,Post × INTERVENTIONtDCS,Sham on total JTT time [F(1,10) = 10.87; P < 0.01]. Post hoc testing showed that tDCS significantly reduced total JTT time (from 43.57 ± 2.36 s at baseline to 39.72 ± 2.15 s post-stimulation; P < 0.05; Fig. 2A, asterisk) in the absence of changes with Sham (from 41.87 ± 2.5 s baseline to 43.27 ± 2.19 s post-stimulation; ns; Fig. 2B). Additionally, tDCS led to more prominent reductions in JTT time relative to baseline than Sham (−6.19 ± 7.81 s, −4.78 ± 5.05 s and −4.03 ± 6.11 s for JTT4, JTT5 and JTT6, respectively; mean ± SD). Therefore, performance improvements were already evident at the measurement of JTT4 (during tDCS) and outlasted the stimulation period by at least 25 ± 2.9 min (Fig. 2A, inset). Retesting 11.3 ± 4.1 days later showed values comparable to those identified at the end of the familiarization session and at the beginning of the tDCS session (Fig. 2A). Improvements in total JTT time with tDCS were identified in every single patient (Fig. 3).

Fig. 2

Effects of tDCS/Sham on motor performance. JTT total time in the familiarization session (displayed in both A and B for comparison), at baseline (JTT1–3), during and following (JTT4–6) and >9 days after tDCS (A) and Sham (B). Note that patients reached stable JTT performance during the initial familiarization session that was comparable to JTT1–3 baseline levels in Sessions 2 and 3. Additionally, half of the patients did tDCS first and half did Sham first. tDCS (A, asterisk) but not Sham (B) resulted in shorter total times (JTT4–6) relative to baseline (JTT1–3). Performance improvements that appeared during tDCS, persisted beyond the stimulation period for at least 25 min (A, inset) and returned to baseline levels days later (JTT7).

Fig. 3

Effects of (A) tDCS and (B) Sham on total JTT time in individual patients. Note shortening of total JTT time with tDCS in all patients.

The interaction TIMEbaseline,post × INTERVENTIONtDCS,Sham × SUBTESTcards,objects,feeding,checkers,light cans,heavy cans was not significant [F(5,60) = 1.28, ns), indicating that there was no detectable differential effect of tDCS on the different individual subtests (Table 3). However, tests that require finer motor control (turn over cards, pick up small objects by hand and spoon: fine distal tasks) tended to improve more than those requiring more proximal arm motions (stacking chequers, moving light and heavy cans: proximal tasks). ANOVARM showed a significant interaction TIMEbaseline,post×INTERVENTIONtDCS,Sham [F(1,68) = 9.79; P < 0.01] and a trend for a interaction TIMEbaseline,post × INTERVENTIONtDCS,Sham × MOTOR CONTROLfine distal,proximal [F(1,68) = 2.91; P = 0.09; Table 3]. Post hoc testing showed a significant difference between tDCS-induced improvements in JTT for fine distal tasks versus proximal tasks (P < 0.05). tDCS-induced performance improvement in fine distal tasks/proximal tasks ratio correlated well with MRC scores (r2 = 0.70; P = 0.039) and showed a correlation trend with Fugl-Meyer scores (r2 = 0.61; P = 0.068).

View this table:
Table 3

Effects of tDCS and Sham on JTT subtests

tDCSShamNet changes in %
Baseline (JTT1–3)JTT4–6Change in %Baseline (JTT1–3)JTT4–6Change in %
Cards5.65 ± 0.474.78 ± 0.35−15.45.4 ± 0.415.23 ± 0.35−3.15−12.25
Objects10.63 ± 0.879.86 ± 0.77−7.2410.31 ± 0.8310.32 ± 0.640.1−7.34
Feed12.08 ± 0.6911.43 ± 0.68−5.3811.86 ± 0.713.31 ± 1.1212.23−17.61
Chequers5.69 ± 0.385.12 ± 0.32−10.025.35 ± 0.355.18 ± 0.29−3.18−6.82
Light cans4.77 ± 0.184.38 ± 0.2−8.184.61 ± 0.274.61 ± 0.20−8.18
Heavy cans4.75 ± 0.194.41 ± 0.2−6.174.65 ± 0.264.62 ± 0.2−1.08−5.09
  • All values are mean ± SE. Change in % = (JTT4–6/JTT1–3 × 100–100). Net changes in % = change in %tDCS – change in %Sham stimulation. Negative values indicate a reduction in total JTT time and consequently performance improvement.

Effects of non-invasive cortical stimulation on motor cortical excitability

In Session 4 MT did not change with tDCS (from 46 ± 2.6% to 45 ± 1.9%, ns). On the other hand, ANOVARM showed a significant effect of TIMEbaseline,post [F(1) = 13.69; P < 0.001] and a trend for a significant effect of TMS INTENSITY110%,120%,130%,140%,150% MT [F(4) = 2.39; P = 0.08], in the absence of a significant interaction TIMEbaseline,post × TMS INTENSITY110%,120%,130%,140%,150% MT [F(1,4) = 0.22; P = 0.92], indicating an overall increase of RC with tDCS (Fig. 4A). The tDCS-induced enhancement in RC slope correlated well with tDCS-induced improvements in JTT (r2 = 0.78; P < 0.05; Fig. 4B). tDCS led to a change of SICI (from 57.17 ± 7.97% to 68.13 ± 6.36% of the test unconditioned MEP) reflecting reduced inhibition, which was more prominent immediately after the end of tDCS (P < 0.05; Fig. 4C, TMS1), and to a non-significant increase in ICF (from 145.11 ± 4.12% to 175.12 ± 15.99% of the test unconditioned MEP; P = 0.19). TS MEP amplitude was 1.1 ± 0.11 mV at baseline and 1.33 ± 0.12 mV at post.

Fig. 4

Effects of tDCS on corticomotor and intracortical excitability. (A) Recruitment curve (RC). Data for different stimulus intensities (100% to 150% MT) were calculated and displayed in percentage of the MEP amplitude elicited by the test stimulus at 100% MT (e.g. RC at 150% MT = MEP amplitude 150% MT/MEP amplitude 100% MT × 100; y-axis) before (baseline, grey lines) and after (post, black lines) tDCS. Insets display raw data in one individual. (B) The abcissa displays the tDCS-induced increase in RC slope calculated as the ratio of the RC slope before (RC-slopebaseline) and after tDCS (RC-slopepost), corresponding to the equation RC-slopepost/RC-slopebaseline (values >1 indicate larger RC with tDCS, while values <1 reflect decreases in RC with tDCS). The ordinate displays tDCS-induced improvement in JTT measured as the average percentage improvement for all subtests in each individual. Note the significant correlation between tDCS-induced increase in RC slope and tDCS-induced improvement in JTT. (C) Intracortical inhibition (SICI). Magnitude of SICI expressed as percentage of test MEP amplitude before (black bar) and after (white bar) tDCS. TMS1 and TMS2 display measurements obtained immediately and ∼25 min after the end of tDCS, respectively. Note the decrease of SICI that was more prominent immediately after the end of tDCS (*P < 0.05).


The main finding of this double-blind, crossover study was that non-invasive cortical stimulation in the form of tDCS applied to the motor cortex of the affected hemisphere resulted in functional improvement in the paretic hand of chronic stroke patients that outlasted the stimulation period and was present in every patient tested.

The JTT assesses functional hand motor skills (Jebsen et al., 1969), has good validity and reliability (Jebsen et al., 1969; Hackel et al., 1992), and has been extensively studied in rehabilitative settings (Spaulding et al., 1988; Kraft et al., 1992; Neistadt, 1994; Alon et al., 2003). Subcomponents of the JTT mimic activities of daily living that require skilled hand and arm motor function. Performance of these motor tasks is conducted through fast cortico spinal projections (Muller and Homberg, 1992) originated in the primary motor cortex (Jancke et al., 2004). While some of these subtests rely predominantly on skillful control of distal hand function, as, for example, picking up small objects, others rely predominantly on more proximal arm control like moving light or heavy cans (Jebsen et al., 1969). In previous studies, improvement in JTT correlated well with functional gains during rehabilitative training after stroke (Kraft et al., 1992; Alon et al., 2003; Wu et al., 2004). Therefore, JTT is a valid measure of hand function in the recovery process following a brain lesion such as stroke.

Our results document that non-invasive stimulation of motor regions of the affected hemisphere in patients with chronic stroke results in functional gains in motor function of the paretic hand. This finding is consistent with previous studies showing that direct stimulation of a motor cortical representation in rodents elicits cortical reorganization (Nudo et al., 1990), while non-invasive cortical stimulation in humans influences motor cortical excitability (Chen et al., 1997; Pascual-Leone et al., 1998; Nitsche and Paulus, 2000, 2001). Additionally, non-invasive cortical stimulation can facilitate cortical plasticity elicited by motor training (Butefisch et al., 2004) and ischaemic nerve block (Ziemann et al., 1998b), and induce behavioural gains in the form of enhancement of implicit motor learning (Nitsche et al., 2003c), visuo-motor processing (Antal et al., 2004a, b), probabilistic learning (Kincses et al., 2004) and analogic reasoning (Boroojerdi et al., 2001b) in healthy human volunteers. These findings raised the hypothesis that non-invasive cortical stimulation could contribute to recovery of motor function in stroke patients, a proposal that gained support from preliminary reports of improvements in motor function in brain-lesioned rodents and primates (Adkins-Muir and Jones, 2003; Kleim et al., 2003; Plautz et al., 2003; Teskey et al., 2003) and in one patient with stroke (Brown et al., 2003) with stimulation delivered through epidural electrodes.

In our experimental design geared to test this hypothesis, patients initially familiarized themselves with the task and reached a stable performance level in the first session. Baseline levels (JTT1–3) measured days apart in the following two counterbalanced sessions (tDCS and Sham) were comparable to those determined at the end of the familiarization session, demonstrating test reproducibility over time (Fig. 2). JTT improvements in the tDCS condition (see inset in Fig. 2A) persisted for more than 25 min after the stimulation ended and returned to baseline levels when retested ∼10 days later, but we do not know the precise duration of the effect. The fact that discomfort/pain (at minimum levels of 1–2 out of 10), attention and fatigue were comparable in tDCS and Sham sessions, together with the finding that patients were unable to distinguish the tDCS from the Sham session, as well as the lack of side-effects and the stratified design by which intervention and testing were performed by different investigators are consistent with success of an experimental design geared to blind both patients and investigators. The finding that error rates were comparable in both sessions supports the view that improvements in JTT did not originate in a change of speed accuracy trade-off. The magnitude of tDCS-induced improvement in JTT, while modest (∼12%), was robust since it was documented in every single subject tested (Fig. 3), supporting the proposal that cortical stimulation combined with motor training could enhance functional gains in stroke patients beyond levels reported in this investigation with tDCS alone. Consistent with this proposal are recent reports showing that cortical stimulation applied in synchrony with motor training enhanced training-dependent plasticity in healthy human volunteers (Butefisch et al., 2004) and increased functional recovery in animals with focal motor cortical lesions (Adkins-Muir and Jones, 2003; Teskey et al., 2003). The finding of slightly longer JTT in three patients after Sham (Fig. 3) could reflect mild fatigue in these individuals, insufficient to reach overt perception in analogue scales. This result underlines the beneficial effect detected in the tDCS session, leading to clear improvements in every single subject.

While our results are suggestive of differential effects of tDCS on different JTT subtests, it remains to be determined whether functional improvements elicited by this form of stimulation are more prominent for tasks that involve fine distal hand movements than for those involving more proximal functions. In our patients with chronic subcortical stroke, severe weakness right after the ictal event, and relatively good motor function at the time of testing, the magnitude of tDCS-induced improvement in JTT (11.75 + 3.61%) was similar to that elicited by the same intervention in age-matched healthy volunteers (10.96+2.75%) (Hummel et al., 2004), suggesting a comparable ability for neuroplastic changes in the motor system, possibly contributing to successful recovery. It remains to be determined whether the beneficial effects of tDCS of the affected hemisphere on JTT performance are mediated through stimulation of primary motor cortex alone (Nudo et al., 1990; Werhahn et al., 2003; Murase et al., 2004) or in combination with ipsilesional dorsal premotor cortex (Liu and Rouiller, 1999; Fridman et al., 2004), both regions closely located and actively involved in recovery of motor function after stroke. The dimension of tDCS electrodes does not allow at this time more focal stimulation, an issue that could be addressed specifically using focal TMS (Siebner et al., 2001; Johansen-Berg et al., 2002; Fridman et al., 2004).

Interventional tDCS is easy to apply, painless, presents advantages for the design of Sham controls and can influence motor cortical function for up to 90 min (Nitsche and Paulus, 2001). tDCS effects on motor cortical function appear to rely to some extent on increased efficacy of NMDA receptor activity (Liebetanz et al., 2002; Nitsche et al., 2004a, b), a mechanism that also influences recruitment curves and intracortical inhibition (Ziemann et al., 1998a; Schwenkreis et al., 1999; Stefan et al., 2002). Therefore, our finding of enhanced recruitment curves that correlate with tDCS-induced performance improvements and of reduced short-interval intracortical inhibition suggest the involvement of NMDA (Liebetanz et al., 2002; Nitsche et al., 2004a, b) and possibly GABA (Boroojerdi et al., 2001a; Chen, 2004) receptor-dependent mechanisms on JTT improvements identified in this study. tDCS did not affect motor thresholds in our patients (which were comparable to MT reported in healthy volunteers; Peinemann et al., 2001; Jancke et al., 2004), but elicited a non-significant trend for increase in intracortical facilitation, both consistent with previous studies in healthy volunteers (Nitsche et al., 2004b, c). Overall, association between increases in motor cortical excitability and performance improvements in motor function or motor learning have been also described in healthy subjects (Muellbacher et al., 2001; Garry et al., 2004) and patients with brain lesions (Traversa et al., 1997; Liepert et al., 1998), but cause-effect links between both remain to be demonstrated. Finally, it is conceivable that factors such as lesion site, size, location, time after the ictal event or impairment levels (Fridman et al., 2004; Wu et al., 2004) influence functional results of cortical stimulation.

In summary, this study demonstrates that non-invasive cortical stimulation of motor regions of the affected hemisphere can beneficially influence skilled motor functions of the paretic hand in patients suffering from chronic stroke. This finding supports the hypothesis that non-invasive cortical stimulation combined with motor training could represent a useful adjuvant to traditional interventions in neurorehabilitation.


The authors thank Shashi Ravindran, RN, for patient recruitment, N. Dang for providing technical help, M. Lomarev for technical advice and Devee Schoenberg for skillful editing. This research was supported by a grant from the Alexander von Humboldt Foundation (Feodor-Lynen) to F.H.


View Abstract