The purpose of this systematic review is to collate evidence regarding the

Transcription

1 Authors: Manuela Corti, PT Carolynn Patten, PhD, PT William Triggs, MD Affiliations: From the Neural Control of Movement Lab (MC, CP, WT), Brain Rehabilitation Research Center, Malcom Randall VAMC, Gainesville, Florida; and Departments of Physical Therapy (MC, CP) and Neurology (CP, WT), University of Florida, Gainesville, Florida. Correspondence: All correspondence and requests for reprints should be addressed to: Carolynn Patten, PhD, PT, Brain Rehabilitation Research Center of Excellence, Malcom Randall VA Medical Center; and Department of Physical Therapy, University of Florida, Box , UFHSC, Gainesville, FL Disclosures: Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article. M. Corti is supported by a University of Florida Fellowship. C. Patten receives grant support from the Department of Veterans Affairs, Rehabilitation Research & Development Service, including a Research Career Scientist Award (F7823S) /12/ /0 American Journal of Physical Medicine & Rehabilitation Copyright * 2012 by Lippincott Williams & Wilkins DOI: /PHM.0b013e318228bf0c LITERATURE REVIEW Stroke Repetitive Transcranial Magnetic Stimulation of Motor Cortex after Stroke A Focused Review ABSTRACT Corti M, Patten C, Triggs W: Repetitive transcranial magnetic stimulation of motor cortex after stroke: A focused review. Am J Phys Med Rehabil 2012;91:254Y270. Repetitive Transcranial Magnetic Stimulation (rtms) is known to modulate cortical excitability and has thus been suggested to be a therapeutic approach for improving the efficacy of rehabilitation for motor recovery after stroke. In addition to producing effects on cortical excitability, stroke may affect the balance of transcallosal inhibitory pathways between motor primary areas in both hemispheres: the affected hemisphere (AH) may be disrupted not only by the infarct itself but also by the resulting asymmetric inhibition from the unaffected hemisphere, further reducing the excitability of the AH. Conceptually, therefore, rtms could be used therapeutically to restore the balance of interhemispheric inhibition after stroke. rtms has been used in two ways: low-frequency stimulation (e1 Hz) to the motor cortex of the unaffected hemisphere to reduce the excitability of the contralesional hemisphere or high-frequency stimulation (91 Hz) to the motor cortex of the AH to increase excitability of the ipsilesional hemisphere. The purpose of this systematic review is to collate evidence regarding the safety and efficacy of high-frequency rtms to the motor cortex of the AH. The studies included investigated the concurrent effects of rtms on the excitability of corticospinal pathways and upper-limb motor function in adults after stroke. This review suggests that rtms applied to the AH is a safe technique and could be considered an effective approach for modulating brain function and contributing to motor recovery after stroke. Although the studies included in this review provide important information, double-blinded, sham-controlled Phase II and Phase III clinical trials with larger sample sizes are needed to validate this novel therapeutic approach. Key Words: Repetitive Transcranial Magnetic Stimulation, Rehabilitation, Stroke, Recovery, Interhemispheric Competition The purpose of this systematic review is to collate evidence regarding the concurrent safety and efficacy of high-frequency (i.e., 91 Hz) repetitive transcranial magnetic stimulation (rtms) on the primary motor cortices (M1) of the affected hemisphere (AH) in adults after stroke. Specifically, this review focuses on how rtms to the AH influences both the excitability of corticospinal 254 Am. J. Phys. Med. Rehabil. & Vol. 91, No. 3, March 2012

2 pathways and upper-limb motor function in adults after stroke. Stroke is the leading cause of adult disability worldwide. According to the World Health Organization, 15 million people have a stroke each year. Stroke impairs a number of neurologic domains, the most common of which is the motor system. 1 The spontaneous return of motor function may span from 1 to 3 mos after stroke; however, less than 40% of stroke survivors recover completely despite intensive rehabilitation training. 2 Brain areas may undergo maladaptive plasticity after stroke, affecting, in some cases, both neural activation and, ultimately, motor behavior. Impaired cortical excitability can be detected through various neurophysiologic parameters. The most commonly accepted marker of corticospinal pathway disruption is the amplitude of motor-evoked potentials (MEP). In addition to MEP amplitude, some authors report other markers including recruitment curve slope (also known as input-output curve), the percentage of stimulator output at resting (RMT) or active motor threshold (AMT), and the markers of inhibition and facilitation of activity in intracortical circuits (i.e., intracortical inhibition; silent period, and intracortical facilitation). Importantly, each of these parameters characterizes cortical activity in only one hemisphere. Other neurophysiologic techniques can be used to investigate the interaction between the two hemispheres (i.e., interhemispheric inhibition). Interhemispheric inhibition can be studied using either single-pulse (i.e., ipsilateral silent period) or paired-pulse techniques. After stroke, activity in the AH is disrupted not only by the infarct itself but also by inhibition from the unaffected hemisphere (UH), which further reduces the excitability of the AH. As first described by Ward and Cohen 3 and more recently stated by Nowak s 4 hypothesis of interhemispheric competition after stroke, the M1 of the UH becomes disinhibited and exerts exaggerated inhibition onto the M1 of the AH. Among several innovative, noninvasive techniques for improving motor recovery after stroke, rtms shows considerable promise. 5 rtms involves focused magnetic stimulation applied to the skull to target a particular brain area. 6 In healthy adults, rtms at frequencies less than 1 Hz can suppress the excitability of the motor cortex, causing an inhibitory effect, whereas at higher frequencies (e.g., 91 Hz) rtms can increase cortical excitability, causing facilitation. 7 The capacity for rtms to influence cortical excitability contributes to the rationale for its use as a therapeutic adjuvant that may enhance the efficacy of rehabilitation for persons after stroke. 4,8 Asymmetric cortical excitability resulting from stroke may promote maladaptive neuromotor strategies. Repeated use of maladaptive, compensatory motor strategies will disrupt normal physiologic activity in transcallosal pathways, producing an imbalance in the reciprocal inhibitory projections between hemispheres. 3,4 Modulation of cortical excitability through rtms may induce synaptic plasticity and promote physiologic activity in transcallosal pathways, which, taken together, will potentially limit the development of maladaptive neural strategies. 3,4 In this context, rtms has been also been proposed as a theoretical approach to restore the balance of interhemispheric inhibition after stroke (e.g., reduce interhemispheric competition). 3,4 The current literature reveals the positive effects of rtms after stroke, including modulation of cortical excitability (e.g., MEP amplitude, recruitment curves, and motor threshold) toward interhemispheric balance. However, it is important to note that no studies to date have directly investigated the effects of rtms on interhemispheric inhibition. Therefore, support for the theoretical explanation that rtms rebalances interhemispheric inhibition remains to be demonstrated. The current working hypothesis holds that inhibitory rtms over the UH reduces transcallosal inhibition from the unaffected to the affected/ipsilesional hemisphere and facilitatory rtms over the AH increases excitability of the AH and increases transcallosal inhibition from the affected to the unaffected/contralesional hemisphere. Consistent with effects noted in healthy individuals, constant high-frequency rtms (trains of stimuli separated by intertrain intervals) has been used in two ways in persons after stroke: low-frequency (e.g., e1 Hz) stimulation of the UH to reduce hyperexcitability of the contralesional hemisphere or highfrequency (e.g., 91 Hz) stimulation of the AH to increase excitability of the ipsilesional hemisphere. 9 A more recent form of stimulation is theta burst stimulation (TBS), which uses repeating bursts of very lowyintensity combined-frequency rtms. 10 Each burst consists of three stimuli (delivered at 50 Hz) repeating at 5 Hz. TBS is also used in two ways: a continuous train of 100 bursts (300 stimuli) is used to suppress corticospinal excitability; whereas an intermittent pattern (20 trains of ten bursts, varied interstimulus interval, total 600 pulses) is used to enhance corticospinal excitability. A significant concern when using rtms is the potential to induce seizures even in individuals without any predisposing or underlying risks for seizure. Although this risk is low, it may increase rtms of Motor Cortex after Stroke 255

3 after stroke because cellular biochemical dysfunction may lower seizure threshold within the brain. 11 A recent review of safety and application guidelines for rtms reported a slightly higher risk of seizures with high-frequency rtms and with TBS compared with low-frequency rtms. 11 However, as reported in Bae et al., 12 the probability of seizure induction remains very low: approximately crude perperson risk estimate in individuals with known epilepsy and less than 0.01 in healthy adults. 11 Crude per-person risk estimate reflects the probability of an individual to have a seizure during an rtms intervention. In this case, crude per-person risk estimate was calculated for each of the 30 studies included, and the crude risk average was weighted by sample size and stimulus number (total stimuli per subject) and was reported with 95% confidence intervals. 12 Some investigators favor low-frequency rtms over high-frequency rtms because of its wider toleration and fewer risks. However, recent studies have demonstrated that high-frequency rtms applied to the AH with stimulation parameters (intensity, frequency, train length, or intertrain duration) within the known, published safety guidelines 11 is both safe and able to increase AH excitability. Although a correlation has not been demonstrated, the studies suggest that there may be an association between these neurophysiologic effects (i.e., reduced motor threshold, altered MEP amplitude) and improved behavioral function of the affected hand. 4,10,13Y22 The remainder of this article is organized into three sections. Section 2 describes the methods used to locate, screen, and classify the articles that are considered in this review. Section 3 summarizes and evaluates each article, and Section 4 discusses the findings and suggests paths for future research. METHODS Search Strategy To gather relevant articles, the following searches were performed on the PubMed database in May 2010: 1. BStroke/rehabilitation[ [Mesh] OR BStroke/ therapy[ [Mesh] 2. BTranscranial Magnetic Stimulation[ [Mesh] 3. Repetitive transcranial magnetic stimulation 4. Motor cortex 5. Affected hemisphere OR stroke hemisphere OR lesioned hemisphere 6. High frequency 7. #2 OR #3 8. #5 OR #6 9. #1 AND #7 AND #8 Additional potential references were identified in the BRelated Articles[ section of the PubMed database, in the references of included articles and in our own collection of papers. Criteria for Considering Studies We included a study in this review if it met all of the following criteria: 1. The participants were adults who have had stroke with either cortical or subcortical infarct. 2. The study addressed the effects of rtms on the M1 of the AH to facilitate motor control/ functional behavior and/or transmission in corticospinal pathways. Our rationale for including only studies of rtms to the AH is motivated by the need to understand the safety and efficacy of stimulation to the AH after stroke. Most rtms studies to date have focused on UH stimulation because AH stimulation has been considered a strong precaution, if not a frank contraindication, after stroke. Concerns related to AH stimulation involve both the safety (i.e., seizures) and efficacy of stimulating the damaged M1. This review aims to meet this need. 3. The study outcomes investigated at least one of the following categories: - Motor control, such as reaction time, muscle strength, dexterity, and others; - Functional behavior, such as Wolf Motor Function Test score, Motor Activity Log results, and others; and - Transmission in corticospinal pathways, such as MEP amplitude, resting and active motor threshold, and others. We excluded studies that used alternative techniques such as epidural electrical stimulation or transcranial direct current stimulation. We screened articles by assessing their abstracts and obtained full papers for any references selected. We evaluated all papers that met the selection criteria for their scientific evidence and quality. Criteria for Classifying Study Quality and Strength of Evidence To consider a more comprehensive list of factors affecting the quality of the studies, we used four grading methods to assess each paper: 1. The Five-Phase Model for Clinical Outcomes Research (Robey 23 ). This scale is an adaptation 256 Corti et al. Am. J. Phys. Med. Rehabil. & Vol. 91, No. 3, March 2012

4 of the clinical outcomes research (Greenwald and Cullen, 1985) for audiologists and speechlanguage pathologists. It classifies articles into five categories: - Phase I Y The study determines whether a therapeutic effect is present and whether further investigation is warranted. - Phase II Y The study explores the dimensions of the therapeutic effect and makes preparation for a clinical trial. - Phase III Y The study conducts a clinical trial to test efficacy. - Phase IV Y The study conducts field research to test effectiveness. - Phase V Y The study assesses relative cost and value. 2. The four types of research designs defined by Meltzoff 24 - I. Nonexperimental design - II. Pre-experimental design - III. Quasiexperimental design - IV. Experimental design. 3. The Oxford Centre for Evidence-Based Medicine levels (Phillips 25 ). In this grading scale, evidence is rated for quality from level 1 (best quality) to level 5 (lowest quality) as described in BAppendix.[ 4. Physiotherapy Evidence Database (PEDro) scale. 26 This scale assesses the overall quality of the randomized controlled trial and is commonly used in physical therapy-based systematic reviews. It includes 11 questions and is based on a scale of 0 to 10. The first question is used to determine external validity and was not graded in the scale. The PEDro scale is described in BAppendix.[ RESULTS The original search generated a total 90 articles; however, only 12 articles met the inclusion criteria for this review. Table 1 summarizes the 12 included studies. We identified five of the articles through electronic search as described previously. We found three articles in the BRelated Articles[ section in PubMed and three additional articles in the reference list of articles identified by the electronic search. One article was taken from our own collection. Together, the studies analyzed 317 stroke participants, of which 163 were classified as having a cortical stroke, and 125 were classified as having a subcortical stroke. The remaining 29 participants included both cortical and subcortical strokes, but the exact partition was not reported. The articles included subjects in both acute and chronic phases after stroke; time poststroke ranged from 1Y10 days to 10 yrs. The average age of the stroke participants was yrs, with a standard deviation of 6.34 yrs. The target muscles for rtms were the first dorsal interosseous in ten studies, 10,13Y15,18Y20,27,28 the abductor digiti minimi in one study 16 and the flexor pollicis brevis in one study. 21 In nine of the studies, the researchers applied constant high-frequency rtms 13Y19,21,28 (trains of stimuli separated by intertrain intervals), whereas in the remaining three studies, combined variablefrequency rtms, such as the TBS, was applied. 10,20,22 Two studies applied rtms at 1 Hz: one used an intensity of 100% of RMT, 15 whereas the other used 30% of 2.3 T. 17 Three studies applied rtms at 3 Hz; two, at 130% RMT 14,15 ; and one, at 120% RMT. 16 Three studies applied rtms at 10 Hz; two at 80% of RMT 18,28 and one at 100% of RMT. 14 Three studies applied rtms at 20 Hz; two studies used 90% RMT, 13,19 whereas one used 120% and 130% RMT, 21 in random order, for each participant. One study applied rtms at 25 Hz with intensities of 120% and 130% of RMT 21 in a random order for each participant. In the three studies 10,20,22 using TBS, the researchers applied intermittent TBS (itbs) over the AH at 80% AMT using 20 trains of ten bursts (each burst = 3 pulses at 50 Hz, total of 600 pulses) at 5 Hz with intertrain intervals of 8 or 10 secs. Four studies included a sham group 14Y16,19 ; another four studies used a sham condition 10,18,21,28 (participants of the same group underwent both real rtms and sham TMS); one study used a healthy control group, 22 and three studies lacked any such controls. 13,17,20 Study Quality and Strength of Evidence Table 2 summarizes the scores for each article. Using the five-phase model of clinical outcomes research, we classified four articles 14Y16,19 as Phase II and the other eight articles 10,13,17,18,20Y22,28 as Phase I. Using Meltzoff s scheme for research design classification, we obtained a similar assessment of these articles: four 14Y16,19 were defined as experimental designs, and eight 10,13,17,18,20Y22,28 were defined as quasiexperimental designs. In contrast, the Oxford Centre for Evidence-Based Medicine levels scale assigned all articles an evidence score of 2b. Four articles have clearly better quality than others, although none of them reported confidence intervals; therefore, none of them could be assigned a higher level of evidence by this scale. Using the Oxford Centre for Evidence-Based Medicine grades of recommendation, we assigned all articles a grade of B. rtms of Motor Cortex after Stroke 257

5 The PEDro scale graded the quality of articles with more interarticle variability. This scale considers a randomized controlled trial to be high quality if its total score is 6 of 10 or better. One article was excluded from this evaluation because we did not consider it a randomized controlled trial. Table 3 reports the results of the individual items and the total PEDro scores. The total scores range from 2 of 10 to 8 of 10. Six of 12 articles 10,14Y16,19,22 are considered high quality. The articles that obtained the highest score in the PEDro scale were the same articles that obtained the highest scores in the other classification scales. The first question of the PEDro scale is used to evaluate the external TABLE 1 Description of the included studies Reference Study Design No. of Participants (Mean T SD Age, yrs) Control Lesion Location Time Since Stroke Area of Stimulation 1 Yozbatiran N, 2009 Nonblinded, noncontrolled longitudinal study 12 adults poststroke (67 T 12 yrs) No NA 94.7 T 4.5 yrs Posterior precentral gyrus of the hand knob 2 Khedr EM, 2009 Longitudinal single-blinded, randomized, sham-controlled study (2:1, 3 Hz and 10 Hz rtms:sham) 48 adults poststroke (59.52 T 13.10): 16 at 3 Hz, 16 at 10 Hz and 16 with sham rtms Yes 13 cortical; 35 subcortical 6.5 T 3.6 days Hand area of M1 3 Ameli M, 2009 Single-blinded, sham-controlled crossover (?) 29 adults poststroke (56 T 13) Participants served as their own control 16 with subcortical stroke and 13 with combined cortical-subcortical infarction. 22 T 226 wks (1) hand area of M1 (2) vertex (control stimulation) 4 Khedr EM, 2009 Longitudinal single-blinded, randomized, sham-controlled study (2:1, 1 Hz and 3 Hz rtms: sham) 36 adults poststroke (57.9 T 11 ): 12 at 1 Hz, 12 at 3 Hz and 12 with sham rtms Yes 19 cortical; 17 subcortical 17.1 T 3.6 days Hand area of M1 258 Corti et al. Am. J. Phys. Med. Rehabil. & Vol. 91, No. 3, March 2012

6 validity of the randomized controlled trial, and it was not included in the final score. Five articles 14Y16,19,22 of 11 satisfied the first item. This item is satisfied if the trial report describes the source of the participants and lists eligibility criteria. Most of the trials reported the eligibility criteria but omitted the source of participants. Safety and Effects of Intervention Single-Session Studies This review included four single-session studies 13,18,21,28 that aim to assess the safety and the effects of a single application of constant high frequency rtms. Kim et al. 18 applied 10 Hz stimulation with an intensity of 80% RMT (eight trains for 2 Muscle/s Recorded Stimulation Parameters No. of Sessions Hemisphere Frequency Intensity Pattern Outcome Measure and Results FDI 1 AH 20 Hz 90% of RMT or 60% of device output if RMT was not elicitable 40 trains of 40 pulses; intertrain interval of 28 secs FDI 1 every day for 5 days AH Group 1: 3Hz Group 2: 10 Hz Group 3: same as Group 1, coil angled away from head Group 1: 130% of RMT Group 2: 100% of RMT Group 3: same as Group 1, coil angled away from head Group 1: 50 trains; duration of 5 secs; total of 750 pulses Group 2: 37 trains; duration of 2 secs; total of 750 pulses Group 3: same as Group 1, coil angled away from head FDI 1 AH 10 Hz 80% of RMT 1000 pulses; 5-sec train duration; 25-sec intertrain interval FDI 1 every day for 5 days AH Group 1: 1Hz Group 1: 100% of RMT Group 2: 3Hz Group 2: 130% of RMT Group 3: Group 3: same same as as group 2, group 2, coil angled coil angled away from away from head head Group 1: 900 continuous pulses; duration of 15 mins & Group 2: 30 trains; 10-sec train duration; 2-sec intertrain interval; total of 900 pulses Group 3: same as group 2, coil angled away from head Systolic blood pressure increased 7 mm Hg after 1 min of rtms; no change for diastolic blood pressure and pulse; Fugl-Meyer score improved 3% 1 wk after rtms; grip strength improved 20% immediately after rtms; Nine-Hole Peg Test score improved 80% immediately and 120% 1 wk after rtms. In Group 1 (3 Hz), the improvements were greater those in sham for hand grip, shoulder abduction strength (at 1 mo, 3 mos, and 1 yr), hip-flexion strength (at 1 mo), NIHSS scale (at 1 mo and 1 yr), toe dorsiflex strength (at 1 mo) and Modified Rankin Scale (at 1 yr). In Group 2 (10 Hz), improvements were greater than those in sham for shoulder abduction and hip-flexion strength (at 1 mo, 3 mos and 1 yr), toe dorsiflexion strength, and Modified Rankin Scale (at 1 yr). There were no different effects between the 3-Hz and 10-Hz groups. In the real rtms group, the AH decreased RMT and AMT and increased MEP amplitude. In the sham group, the same was true but for the UH. rtms over M1 in subcortical stroke group improved kinematics of index finger and hand tapping in 14 of 16 participants and reduced activity of the contralesional M1 in 11 participants; in cortical-subcortical stroke group, kinematics deteriorated and caused a widespread bilateral recruitment of primary and secondary areas in 7 of 13 participants. M1 activity at baseline correlated with finger tapping frequency improvements. Real rtms resulted in greater improvements compared with sham in keyboard tapping and pegboard task; no differences in hand grip. Improvements in keyboard tapping and pegboard task were greater in 1-Hz than in 3-Hz groups at third month. All groups improved NIHSS and BI scales at 3 mos; improvements were greater for the real rtms than for sham and greater in 1 Hz than in 3 Hz. 1-Hz rtms-induced increased MEP amplitude and decreased AMT in the AH and the opposite in the UH; 3 Hz induced increased MEP amplitude and decrease in AMT only in the AH; no cortical excitability changes in sham groups. (Continued on next page) rtms of Motor Cortex after Stroke 259

7 TABLE 1 (Continued) Reference Study Design No. of Participants (Mean T SD Age, yrs) Control Lesion Location Time Since Stroke Area of Stimulation 5 Mally J, 2008 Single-blinded, nonrandomized, nonysham-controlled trial 6 Talelli P, 2007 Single-blinded, sham-controlled crossover (?) 64 adults poststroke (57.6 T 10.8). 17 in group A: paretic arm movement could be evoked by stimulation of both sides of the brain; 25 in group B: no movement induced in the paretic arm by stimulation of either side of the brain; 16 in group C: stimulation of the unaffected hemisphere could induce movement in the paretic arm; 6 in group D: stimulation of the affected hemisphere could induce movement in the paretic side. 6 adults poststroke (61 T 14) No Cortical 10 T 6.4 yrs Chosen according to visible movement in paretic arm evoked by TMS Participants served as their own control 3 cortical and 3 subcortical 12Y108 mos Hand area of M1 of both hemispheres 7 Kim Y-H, 2006 Single-blinded, sham-controlled crossover 15 adults poststroke (54 T 5) Participants serve as their own control 5 cortical and 10 subcortical 4Y41 mos Hand area of M1 in affected hemisphere 8 Khedr EM, 2005 Longitudinal single-blinded, randomized, sham-controlled study (1:1, sham-real rtms) 9 Malcolm MP, 2007 Longitudinal double-blinded, randomized, sham-controlled study (1:1, sham-real rtms) 52 adults poststroke: 26 in real rtms group (54 T 10) and 26 in sham rtms group (52 T 8) 19 adults poststroke (67 T 7): 9 rtms + CIT and 10 sham rtms + CIT Yes Yes 26 cortical and 26 subcortical 11 cortical and 8 subcortical 5Y10 days Hand area of M1 in affected hemisphere 4 T 3 yrs Hand area of M1 in affected hemisphere 10 Di Lazzaro V, 2009 Longitudinal nonblinded, noncontrolled study 17 adults poststroke (68 T 11.7) No Cortical and subcortical G10 days Hand area of M1 in affected hemisphere 260 Corti et al. Am. J. Phys. Med. Rehabil. & Vol. 91, No. 3, March 2012

8 Muscle/s Recorded Stimulation Parameters No. of Sessions Hemisphere Frequency Intensity Pattern Outcome Measure and Results None 2 every day for a week Group A: AH and UH Group B: UH Group C: UH Group D: AH 1 Hz 30% of 2.3 T 100 stimuli in each session Group A: Spasticity decreased but no changes in movement; Group B: improvements in spasticity, movement induction, and behavior of paresis; Groups C and D: Spasticity decreased during the first week, but the movement of the paretic arm improved only in group C. FDI 1 Experiment 1: UH Experiment 2: AH 80% of RMT Experiment 1: ctbs; continuous train of 100 bursts (1 burst = 3 pulses at 50 Hz); total of 300 pulses Experiment 3: sham stimulation; randomized between the two previous conditions Experiment 2: itbs; 20 trains of 10 bursts (1 burst = 3 pulses at 50 Hz) at 5 Hz every 8-sec intervals; total of 600 pulses Experiment 3: sham stimulation; randomized between the two previous conditions FDI 1 AH 10 Hz 80% of RMT 8 trains of 20 pulses; train duration of 2 secs; intertrain interval of 58 secs; total of 160 pulses ADM FDI 1 every day for 10 days 1 every day for 10 days AH 3 Hz 120% of RMT 10 trains; train duration of 10 secs; 50-sec intertrain interval AH 20 Hz 90% of RMT 50 trains; 2-sec train duration; 28-sec intertrain interval; total of 1200 pulses FDI 1 AH 80% of RMT itbs; 10 bursts of 3 pulses at 50 Hz given at 5 Hz every 10 secs; total of 600 pulses After itbs on the AH: improvements in speed (90% immediately after rtms lasting for up to 40 mins); no effects in peak grip force; increased rest MEP and reduced AMT; increased area under the Input-Output curve in the affected hand. After ctbs on the UH: no effect in speed and peak grip force; increased rest MEP and reduced AMT in healthy hand but no effects in the paretic hand. Real rtms resulted in larger improvements in MEP amplitude than sham rtms and larger increase in accuracy and movement time. Real rtms resulted in larger improvements than did sham rtms in Scandinavian stroke scale, NIH Stroke scale, and Barthel index 10 days after rtms Wolf Motor Function Test and Motor Activity Log improved in both groups without significant differences. The real rtms group showed a reduction of RMT. itbs produced an increase in MEP amplitude for the AH (that was correlated with recovery at modified Rankin score after 6 mos). (Continued on next page) rtms of Motor Cortex after Stroke 261

9 TABLE 1 (Continued) Reference Study Design No. of Participants (Mean T SD Age, yrs) Control Lesion Location Time Since Stroke Area of Stimulation 11 Lomarev MP, 2007 Single-blinded, sham-controlled crossover study (?) 7 adults poststroke (age range, 35Y65 yrs) Participants serve as their own control 5 cortical and 2 subcortical 1Y5 yrs Hand area of M1 in affected hemisphere 12 Di Lazzaro V, 2008 Single-blinded, nonysham-controlled, healthy-controlled crossover study 12 adults poststroke (69 T 9.5); 12 healthy age-matched control subjects (63.2 T 5.3) No sham-control group, but there is a healthy control group 4 cortical and 8 subcortical 1Y10 days Hand area of M1 in affected hemisphere FDI indicates first dorsal interosseous; ADM, abductor digiti minimi; M1, primary motor cortex; AH, affected hemisphere; UH, unaffected hemisphere; RMT, resting motor threshold; AMT, active motor threshold; ctbs, continuous theta burst stimulation; itbs, intermittent theta burst stimulation; rtms, repetitive transcranial magnetic stimulation; NIH, National Institutes of Health; NA, not available; BEB, brief electromyographic bursts; SE, spread excitation; FPB, flexor pollicis brevis; ECR, extensor carpis radialis; BB, biceps brachii, NIHSS, NIH Stroke Scale; MEP, motor-evoked potentials; BI, Barthel Index. secs) and evaluated the changes in the behavior in finger motor tasks and corticomotor excitability before and after the single-session intervention. They found that real rtms to the AH resulted in larger improvements in MEP amplitude, movement accuracy, and speed compared with sham rtms. These results suggest that motor function was markedly improved after high-frequency rtms in the AH. Similarly, Ameli et al. 28 applied 10-Hz rtms with an intensity of 80% RMT (5-sec stimulation, 25-sec intertrain intervals, total of 1000 pulses). Interestingly, they found different effects of rtms between participants with subcortical stroke only and those with combined subcortical and cortical stroke. In participants with subcortical stroke, high-frequency rtms improved the kinematics of index finger and hand tapping; these improvements were associated with reduced activity of the contralesional M1 as noted by functional magnetic resonance imaging. However, in individuals with combined cortical and subcortical stroke, the kinematics of the affected hand deteriorated after high-frequency rtms; these effects were associated with a widespread bilateral recruitment of primary and secondary motor areas. These changes were revealed after the stimulation of M1 but not after stimulation over the vertex (sham condition). This study suggested that the effectiveness of highfrequency, facilitatory rtms applied over ipsilesional M1 depends on the functional integrity of the stimulation site and/or the extent of the brain area affected by the stroke. Yozbatiran et al. 13 studied safety and behavioral effects of higher-frequency rtms. They applied 20-Hz stimulation (40 trains of 40 pulses for 20 mins) with an intensity of 90% RMT (or 60% of device output if RMT could not be 262 Corti et al. Am. J. Phys. Med. Rehabil. & Vol. 91, No. 3, March 2012

10 Muscle/s Recorded Stimulation Parameters No. of Sessions Hemisphere Frequency Intensity Pattern Outcome Measure and Results FPB, FDI, ECR, and BB muscles. 1 AH Condition A: 20 Hz Condition A: 120% of RMT Condition B: Condition B: 20 Hz 130% of RMT Condition C: Condition C: 25 Hz 110% of RMT Condition E Condition D: (sham): 120% of RMT same as Condition E condition A (sham): same Condition D: as condition A 25 Hz Condition A: 8 trains of 20 pulses; duration of 1 sec; intertrain interval of 5 mins Condition B: 8 trains of 10 pulses; duration of 0.5 sec; intertrain interval of 5 mins Condition C: 8 trains of 20 pulses; duration of 0.8 sec; intertrain interval of 5 mins Condition D: 8 trains of 10 pulses; duration of 0.4 sec; intertrain interval of 5 mins Condition E (sham): same as condition A FDI 1 Experiment 1: AH Experiment 2: UH 80% of AMT Experiment 1: itbs; 10 bursts of 3 pulses at 50 Hz given at 5 Hz every 10 secs; total of 600 pulses Experiment 2: ctbs; 3 pulses at 50 Hz, repeated every 200 millisecs; total of 600 pulses After all real rtms conditions, patients showed BEB and SE. BEB were observed after 17 of 88 trains and SE after 16 of 88 trains. There was a single clonic contraction only in one subject. No seizures were observed in any subject. No SE and only one BEB were recorded after sham rtms. There was a 71% MEP increase after the first rtms train. At the end of the experiment, MEP was reduced. Change of MEP amplitude changed from 24% increase to a 32% decrease. There was no improvement in pinch force after rtms, ranging from 9% increase to 20% decrease. Both the itbs and the ctbs produced a comparable rebalancing of cortical excitability: decrease of RMT in the AH and increase in the UH; increase in MEP amplitude of the AH and decrease in the UH. In addition, the effects observed in patients were comparable to the one observed in controls. elicited). They demonstrated that high-frequency rtms was well tolerated and did not cause any adverse symptoms. Systolic blood pressure increased 7 mm Hg after 1 min of stimulation; no changes were revealed for diastolic blood pressure and pulse, and none of the behavioral measures showed a decrement. TABLE 2 Scores of each article according to the Robey, Meltzoff, and Phillips scales Reference Robey RR, 2004 Meltzoff, 1998 Phillips B, Yozbatiran N, 2009 Phase I Quasiexperimental design 2b 2 Khedr EM, 2009 Phase II Experimental design 1b 3 Ameli M, 2009 Phase I Quasiexperimental design 2b 4 Khedr EM, 2009 Phase II Experimental design 1b 5 Mally J, 2008 Phase I Quasiexperimental design 2b 6 Talelli P, 2007 Phase I Quasiexperimental design 2b 7 Kim Y-H, 2006 Phase I Quasiexperimental design 2b 8 Khedr EM, 2005 Phase II Experimental design 1b 9 Malcolm MP, 2007 Phase II Experimental design 1b 10 Di Lazzaro V, 2009 Phase I Quasiexperimental design 2b 11 Lomarev MP, 2007 Phase I Quasiexperimental design 2b 12 Di Lazzaro V, 2008 Phase I Quasiexperimental design 2b rtms of Motor Cortex after Stroke 263

11 TABLE 3 The scores for each article according to the PEDro scale Items Yozbatiran N, 2009 Khedr EM, 2009 Ameli M, 2009 Khedr EM, 2009 Mally J, 2008 Talelli P, 2007 Kim Y-H, 2006 Khedr EM, 2005 Malcolm MP, 2007 Di Lazzaro V, 2009 Lomarev MP, 2007 Di Lazzaro V, Eligibility criteria were specified. NA Y N Y N N N Y y N N y 2. Subjects were randomly allocated NA y N y N Y N Y y N N N to groups. 3. Allocation was concealed. NA y N y N N N N N N N N 4. The groups were similar at baseline NA y N y y Y N Y N N N Y regarding the most important prognostic indicators. 5. There was blinding of the subjects. NA y y y N Y Y Y Y N Y Y 6. There was blinding of all the NA N N N N N N N N N N N therapists who administered the therapy. 7. There was blinding of all the assessors who measured at least one key outcome. 8. Measures of at least one key outcome were obtained from more than 85% of the subjects initially allocated to groups. 9. All subjects for whom outcome measures were available received the treatment or control condition as allocated, or, where this was not the case, data for at least one key outcome were analyzed using Bintention to treat[ analysis. 10. The results of between-group statistical comparisons are reported for at least one key outcome. 11. The study provides both point measures and measures of variability for at least one key outcome. NA N N N N N N Y Y N N Y NA y y y Y Y Y Y y Y N Y NA y y y Y Y Y Y Y Y N Y NA y N y Y N N Y Y N N y NA y y Y N Y Y Y Y Y Y y Total NA 8 of 10 4 of 10 8 of 10 4 of 10 6 of 10 4 of 10 8 of 10 7 of 10 3 of 10 2 of 10 7 of 10 The total score is determined by counting the number of criteria that are satisfied; however, item 1 is not used to generate the total score, so the total scores are of 10. Y indicates the criterion was clearly satisfied; N, that criterion was not satisfied. PEDro indicates Physiotherapy Evidence Database; NA, not available. 264 Corti et al. Am. J. Phys. Med. Rehabil. & Vol. 91, No. 3, March 2012

12 In terms of behavioral effects, modest improvements were observed in grip strength, range of motion, Nine-Hole Peg test score, and Fugl-Meyer score up to 1 wk after a single-session of high-frequency rtms. Similarly, Lomarev et al. 21 evaluated the safety of higher frequencies and intensities of highfrequency rtms applied over the AH. They tested five paradigms: (1) 20 Hz at 120% RMT (1-sec train, 20 pulses); (2) 20 Hz at 130% RMT (0.5-sec train, ten pulses); (3) 25 Hz at 110% RMT (0.80-sec train, 20 pulses); (4) 25 Hz at 120% RMT (0.4-sec train, ten pulses) and (5) 20 Hz at 130% RMT (1-sec train, 20 pulses) with sham stimulation (plane of the coil tilted). They found that, after all real rtms conditions, the participants demonstrated brief electromyographic bursts (BEB) at rest after 17 of 88 trains (range of BEB observed for each subject, 0Y7), possibly representing peripheral manifestation of after discharges and spread of excitation to new muscles not activated by singlepulse TMS after 16 of 88 trains. All subjects revealed at least one BEB or spread-of-excitation phenomenon, with the exception of one subject for BEB and another subject for spread of excitation. Although these phenomena were considered to be associated with a risk of seizure, no seizures were observed in any participants. After sham high-frequency rtms, no spread of excitation and only one BEB was recorded. rtms did not result in either an increase in motor cortex excitability or in an improvement in pinch-force dynamometry. The study of Lomarev et al. 21 concluded that these rtms parameters were not safe for individuals with chronic stroke because phenomena associated with increased seizure risk were observed. Three additional single-session studies 10,13,18, 20Y22,28 assessed the safety and the effects of rtms delivered as TBS. Talelli et al. 10 tested a group of persons after stroke under three conditions: excitatory itbs over AH (itbs AH ), inhibitory continuous TBS (ctbs UH ) over the intact hemisphere and sham stimulation. itbs AH consisted of 20 trains of ten bursts (1 burst = 3 pulses at 50 Hz) at 5 Hz with 8-sec intertrain intervals at an intensity of 80% of AMT, whereas ctbs UH consisted of continuous trains of 100 bursts at an intensity of 80% of AMT. After itbs AH, there were improvements in simple reaction time in the paretic hand (90% immediately after rtms, lasting up to 40 mins) compared with the sham stimulation. No effect in peak grip force was revealed. The amplitude of the MEPs at rest and during active muscle activation, and the area under the input-output curves, also increased on the lesioned side. ctbs UH did not affect reaction time and peak grip force, but it suppressed MEP amplitude in the nonparetic hand but not in the paretic hand. This study suggested that TBS is safe and that itbs AH transiently improved motor behavior and cortical spinal output in the paretic hands. Similarly, Di Lazzaro et al. 22 compared the application of itbs AH and ctbs UH. Deviating from the previous study, they found that both the facilitatory TBS on the affected motor cortex and the inhibitory TBS on the unaffected motor cortex produced a significant increase in the amplitude of MEPs evoked by the stimulation of the AH. RMT decreased in the AH and increased in the UH, whereas MEP amplitude increased in the AH and decreased in the UH. The authors concluded that TBS could enhance the excitability of the lesioned motor cortex. Because either excitatory itbs to the AH or inhibitory ctbs to the UH increased excitability of the AH, they concluded that the imbalance of the hemispheric excitability after stroke could be explained, at least in part, by an abnormally high interhemispheric inhibitory drive from the UH to the AH as suggested by Murase et al. 29 In a more recent study, Di Lazzaro at al. 20 correlated changes produced by itbs AH (using the same parameters as in their previous study) to outcomes at a 6-mo follow-up. They found that itbs AH produced increased MEP amplitude in the AH, which correlated with recovery measured using the Modified Rankin score (which is a scale for measuring the degree of disability or dependence) at follow-up. This study showed for the first time in humans that changes in long-term potentiation in the AH correlated with the long-term (6 mos) recovery of functional motor behavior. This finding supported the correlation between the ability to induce changes in cortical excitability using rtms and the process of motor recovery. Multiple-Session Studies This review included five multiple-session studies 14Y17,19 with the aim of assessing whether the effects from a single session accumulate, inducing more lasting functional improvements. In a longitudinal single-blinded, randomized, sham-controlled study, Khedr et al. 16 applied rtms daily for 10 consecutive days at 3 Hz and 120% RMT (10 trains of 10 secs, 50 interstimulus intervals) over the AH. They found that real rtms resulted in larger improvements on the Scandinavian Stroke Scale, the National Institutes of Health Stroke Scale, and the Barthel Index compared with sham rtms and that the effects persisted up to 10 days after stimulation. More recently, Khedr et al. 14,15 performed rtms of Motor Cortex after Stroke 265

13 two longitudinal single-blinded, randomized, shamcontrolled studies. The first study 15 compared 1-Hz rtms applied with continuous stimulation for 15 mins at 100% RMT over UH and 3-Hz rtms (30 trains, 10 secs each, 2-sec intertrain interval) at 130% RMT over the AH. Both treatment groups underwent one session daily for five days. The real rtms group experienced greater improvements in keyboard tapping, pegboard, National Institutes of Health Stroke Scale score and Barthel Index score than the sham group. In addition, at 3 mos after stimulation, these improvements were greater in the 1-Hz UH than in the 3-Hz AH group. In terms of cortical excitability, 1-Hz rtms induced an increase in MEP amplitude and a decrease of AMT in the AH concurrently with a decrease in MEP amplitude and an increase in AMT in the UH. In contrast, the 3-Hz rtms induced only an increase in MEP amplitude and a decrease of AMT in the AH. The second study conducted by Khedr et al. 14 compared 3-Hz rtms (5 secs for 50 trains) at 130% RMT with 10-Hz rtms (37 trains. 2 secs each) at 100% RMT, both applied to the AH daily for 5 consecutive days. Real rtms produced greater improvements in muscle strength and greater alleviation of disability measured with stroke severity and functional activity scales (National Institutes of Health Stroke Scale and modified Rankin scale) than did the sham stimulation, and these improvements were evident even at 1-yr follow-up. In addition, in the real rtms groups, both RMT and AMT decreased, and MEP amplitude increased in the AH. The authors did not find significant differences between 3- and 10-Hz stimulation, although the 3-Hz stimulation seemed to produce greater changes in strength and clinical rating scales. Mally et al. 17 examined whether active movement could be induced by rtms even several years after stroke and which hemisphere would be the best location for stimulation to attenuate spasticity and develop movement in the paretic arm. Their results suggested that spasticity could be modified by stimulation of either the AH or UH, but recovery of movement could be achieved only through stimulation of the intact hemisphere. Malcolm et al. 19 tested the potential adjuvant effect of rtms in people undergoing constraint-induced therapy (CIT) for upper-limb hemiparesis. Participants who have had stroke underwent one session per day for 10 consecutive days of 20 Hz rtms to the AH at 90% RMT (50 trains, 2-sec duration, with 28-sec intertrain intervals) followed immediately by CIT. Although significant differential effects were not revealed between participants receiving rtms and those receiving sham rtms, 6 mos after stimulation, the real rtms group showed greater improvement on clinical measures, including the Wolf Motor Function test, the Motor Activity Log, and Box and Block test. Mechanisms for Rebalancing Interhemispheric Competition after Stroke Decreased excitability of the ipsilesional cortex has been observed after stroke through electrophysiologic recordings, 30 cortical stimulation, 31 and functional neural imaging studies. 32 This decreased cortical excitability has been attributed to damage from glutamate receptor expression from neurons in the infarct zone. As a consequence, it is argued that there is reduced interhemispheric inhibition via transcallosal pathways from the AH to the UH. 33,34 Consequently, the UH becomes disinhibited, creating additional inhibition on the affected hemisphere. Indeed, the magnitude of transcallosal inhibition exerted from the UH is positively correlated with the severity of functional impairment of the affected hand. 4 Accordingly, the interhemispheric competition hypothesis suggests that balancing excitability between the AH and UH may improve functional behavior in people after stroke. 2,4Y7,9,11 Most rtms studies considered in this article focus on rebalancing cortical excitability through stimulating the AH and suggest that increasing the cortical excitability of the AH M1 may improve affected hand function after stroke. 2,4Y7,9,11 Consistent with the interhemispheric competition hypothesis, the increased excitability of the AH observed in some of these studies implied that transcallosal inhibition from the UH to the AH was reduced. Therefore, reduced interhemispheric competition was posed as a potential mechanism underlining the functional improvements after rtms intervention. 2,4Y7,9,11 Overactivity of the UH may be a consequence of decreased cortical excitability in the AH. Based on this premise, recent rtms studies have attempted to rebalance cortical excitability by increasing AH excitability rather than by inhibiting the UH. 10,13Y22,27,28 After stimulation with facilitatory rtms, these studies showed enhanced cortical excitability, reduced inhibition of the AH accompanied by improvement in functional motor performance. Mechanisms currently proposed to mediate the effects of facilitatory rtms include enhanced glutamergic neurotransmission and altered GABAergic effects on intracortical circuits that interact to enhance the cortical excitability of AH. 2, Corti et al. Am. J. Phys. Med. Rehabil. & Vol. 91, No. 3, March 2012

14 DISCUSSION The results of this systematic review are quite consistent. Both single and multiple sessions of highfrequency rtms over the AH demonstrate immediate and long-term improvements after stroke. Improvements are reported in dexterity, force and spasticity, kinematics of index finger and hand tapping, and cortical excitability of both the AH and UH. Generally, rtms to the AH resulted in either one or a combination of the following: increased MEP amplitude and decreased AMT in the AH and/or decreased MEP amplitude and increased AMT in the UH. Most of the studies suggested comparable improvements following various high-frequency paradigms including 1 Hz, 3 Hz, 10 Hz, and 20 Hz. However, one study 15 suggested greater improvements after 1-Hz compared with 3-Hz stimulations. Some authors 21 suggest that higher-frequency stimulation was associated with increased electromyographic bursting and spread of excitation, which may be associated with a higher risk of seizure. However, it is critical to note that no seizures were observed in any of the studies included in this review. In this context, it is important to mention that Lomarev et al. s 21 study associated high frequencies, ranging from 20 to 25 Hz, with high intensities, ranging from 110% to 130% RMT. Most of the other studies 14Y16,18,20,22,27,28,35 applied lower intensity levels (80%Y90% RMT) with stimulation frequencies at, or higher than, 20 Hz. Therefore, it is possible that electromyographic bursting and spread of excitation are not associated solely with high-stimulation frequencies but rather with the combination of high frequencies and high intensities. Improvements have also been shown after application of TBS, which is a novel form of rtms that uses very low intensity to increase or decrease motor cortical excitability. This review included three studies applying a single session of TBS, all of which demonstrated that TBS is safe and may induce improvement of cortical excitability and positive changes in the functional motor behavior of the arm. One study investigating TBS demonstrated that itbs to M1 of the AH resulted in greater improvement when compared with ctbs 10 to M1 of the UH. Despite converging evidence for the effectiveness of rtms on the AH, it is not yet clear which stimulation parameters are most effective and how a multiple-session intervention should be organized to maximize the behavioral response. We assessed only four studies as having adequate methodologic quality. Therefore, more high-quality phase II and phase III studies are needed before recommending widespread application of this technique in a clinical environment. Some questions remain unanswered; therefore, future trials could focus on the following potential issues: (1) optimization of parameters, (2) sham controls, (3) the role of lesion site, (4) the time since stroke, and (5) the potential adjuvant effect of rtms associated with other rehabilitation therapies. Is Facilitatory rtms Equally Effective for Cortical and Subcortical Stroke? One study 28 included in this review suggested a differential effect of rtms on cortical and subcortical strokes. Participants with subcortical stroke responded to high-frequency rtms applied over the AH with improved control of the contralesional hand movement. In contrast, none of the participants with cortical involvement showed improvement after AH rtms, and some even showed deterioration in contralesional hand movement. Importantly, at baseline, individuals with cortical and subcortical stroke did not differ regarding either the severity of their impairment or time since stroke. This study suggested that the effectiveness of facilitatory rtms applied over the ipsilesional motor cortex depends on the amount of neural activity of the motor cortex and the functional integrity of the stimulation site. Future research combining behavioral measures with neuroimaging may reveal differential effects of rtms in persons after stroke with diverse locations and sizes of the neural lesion. Corticospinal tract integrity may also be considered as a prognostic indicator for the benefits of facilitatory rtms over the affected hemisphere. There is also a need to evaluate whether inhibitory rtms over the UH would be more effective in improving hand motor function in individuals with cortical stroke. Is Facilitatory rtms Equally Effective in Acute and Chronic Stroke? The articles included in this review suggested improvement of the behavior of the affected hand and of the cortical excitability after rtms in both acute and chronic phases of stroke. However, none of the articles investigated the optimal time since stroke for facilitatory rtms application. Lefaucheur 5 reported that, in chronic stroke, the application of rtms over the lesioned cortical areas is meant to recruit or activate compensatory pathways and to promote adaptive plasticity. 5 In contrast, rtms applied in the acute phase of stroke is meant to limit neuronal loss. 5 Stimulating neurons in the perilesional zone could increase neuronal survival rate and facilitate clinical recovery. 5 Further studies should address how time since stroke rtms of Motor Cortex after Stroke 267