Article Text

Download PDFPDF

Interrogating cortical function with transcranial magnetic stimulation: insights from neurodegenerative disease and stroke
  1. Smriti Agarwal1,
  2. Giacomo Koch2,3,
  3. Argye E Hillis4,5,6,
  4. William Huynh1,
  5. Nick S Ward7,8,9,
  6. Steve Vucic10,
  7. Matthew C Kiernan1
  1. 1 Brain and Mind Centre, University of Sydney, and Institute of Clinical Neurosciences, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia
  2. 2 Non-Invasive Brain Stimulation Unit, Neurologia Clinica e Comportamentale, Fondazione Santa Lucia IRCCS, Rome, Italy
  3. 3 Stroke Unit, Department of Neuroscience, Policlinico Tor Vergata, Rome, Italy
  4. 4 Department of Neurology, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
  5. 5 Physical Medicine and Rehabilitation, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
  6. 6 Cognitive Science, Johns Hopkins University, Baltimore, Maryland, USA
  7. 7 Sobell Department of Motor Neuroscience, UCL Institute of Neurology, University College London, London, UK
  8. 8 UCL Partners Centre for Neurorehabilitation, UCL Institute of Neurology, University College London, London, UK
  9. 9 The National Hospital for Neurology and Neurosurgery, London, UK
  10. 10 Westmead Clinical School, University of Sydney, Sydney, Australia
  1. Correspondence to Dr Smriti Agarwal, Brain and Mind Centre, University of Sydney, Sydney, New South Wales, Australia; smriti.agarwal{at}


  • Contributors MCK and SA conceived the idea for the article. SA drafted the manuscript. All authors revised the manuscript critically for important intellectual content and gave final approval of the version to be published.

  • Funding This work was supported by funding to Forefront, a collaborative research group dedicated to the study of motor neuron disease, from the National Health and Medical Research Council of Australia program grant (#1037746), the Motor Neuron Research Institute of Australia Ice Bucket Challenge Grant and grant aid from Magnetic Health Science Foundation. SA was funded by the Ellison-Cliffe travelling fellowship from the Royal Society of Medicine, UK. AH was funded by NIH P50 DC014664 and NIH ROI DC05375.

  • Competing interests None declared.

  • Patient consent Not required.

  • Provenance and peer review Commissioned; externally peer reviewed.

View Full Text

Statistics from


The ability to modify human brain function is a long-held scientific aspiration. Centuries ago, cognitive neuroscientists used torpedo fish and eels to electrically stimulate the brain, while more conventional electricity was first used for brain stimulation in the 18th century. It was only three decades ago that Pat Merton and colleagues1 achieved electrical stimulation of the motor cortex through the intact scalp to generate a relatively synchronous muscle response. One of the issues with this methodology of transcranial electrical stimulation, however, was the stimulation of pain fibres on the scalp. Subsequently, Barker and his team2 became the first to use magnetic stimulation (transcranial magnetic stimulation (TMS)) in the human brain to achieve simultaneous muscle activity. Over 18 000 scientific publications relating to TMS have appeared (, topic = ‘transcranial magnetic stimulation’ search) since Barker’s first description, with over a third of these in the last five years alone, indicative of the pace at which the field is moving forward.

The aim of the present review is to provide the clinician with an overview of physiological considerations involved with TMS, including cortical output measures that provide important information regarding pathophysiological alterations in neurodegenerative disorders and poststroke reorganisation of neural structure and function. This review aims to provide an overview of TMS applications and their utility in providing a functional understanding of disease mechanisms and the potential for development of novel diagnostic and prognostic tools in neurological disease.

Measures of cortical function

TMS induces current flows in the brain by application of a pulsed magnetic field leading to depolarisation of the underlying cortical neurons (figure 1). The resultant electrical activity in the brain can be modified by the shape and orientation of the coil used, combined with underlying neuronal anatomy and orientation relative to the coil, magnetic pulse wave form, intensity, frequency and pattern of stimulation.3–6

Figure 1

Transcranial magnetic stimulation (TMS) using a circular coil showing the lines of flux of the magnetic field and directions of stimulating and induced currents.

The precise nature of the neuronal circuitry activated by TMS remains incompletely understood. Applying TMS over the motor cortex (figure 2) generates a corticomotor neuronal volleys which may be a result of direct excitation of cortical neurons (direct or D-waves) or trans-synaptic excitation (indirect or I-waves). The I-waves are thought to originate through a complex interaction between cortical output cells (Betz cells, layer V) and interneuronal cells.3 7–9

Figure 2

The paired-pulse threshold tracking transcranial magnetic stimulation (TT-TMS) paradigm to measure cortical excitability. (A) Short interval intracortical inhibition (SICI) occurs at an interstimulus interval (ISI) of 0–7 ms while intracortical facilitation (ICF) occurs at an ISI of 7–10 ms. (B) TMS coil placed over the vertex stimulates the motor cortex and the response is recorded from the opposite abductor pollicis brevis muscle. (C) Change in stimulus intensity required to achieve a target motor-evoked potential (MEP) of 0.2 mV (±20%) is used to quantify the SICI and ICF. RMT, resting motor threshold.

Following a brief overview of TMS output measures, their application as potential diagnostic and prognostic markers will be further considered.

A widely used experimental paradigm involves application of TMS to the motor cortex with recording electrodes placed over an intrinsic hand muscle in the contralateral limb (figure 2). The resultant motor-evoked potential (MEP) on electromyography (EMG) is typically recorded from the abductor pollicis brevis (APB), abductor digiti minimi or the first dorsal interosseous muscle. This paradigm can be applied to quantity excitability characteristics of the underlying motor cortex.

Motor threshold (MT) indicates the ease with which motor cortex output cells and corticomotor neurons can be excited. MT is thought to reflect the density of corticomotor neuronal projections onto the anterior horn cells. It thus follows that MTs tend to be lower in the dominant hand10 and correlate with the performance of fine motor tasks.11 MTs have the potential of providing a biomarker of cortical neuronal membrane excitability. Voltage-gated sodium channels are critical to cortical axon excitability12 while excitatory synaptic neurotransmission in the neocortex is mediated by the glutaminergic alpha-amino-3-hydroxy-5-methyl-4-isoxazoleproprionic acid (receptors.13 Thus voltage-gated sodium channel blocking drugs increase MT14 15 while glutaminergic agonists decrease it.16 Interestingly, neuromodulatory agents affecting GABA, dopaminergic, noradrenergic and cholinergic systems do not affect the MT.17

MT was initially defined as the minimum stimulation intensity (% maximum stimulator output) required to achieve an MEP response of (amplitude >50 μV) in the target muscle in 50% of stimulus trials.18 Evolving studies in threshold tracking TMS have led to redefinition of the MT as stimulus required to achieve and maintain a target MEP response of 0.2 mV (±20%).19 20 MT tends to be lower in a voluntarily contracting muscle (active MT) compared with that in a muscle at rest (resting motor threshold (RMT)).21

Single-pulse TMS measures

MEP amplitude represents summation of descending corticospinal volleys onto motor neurons comprising direct (D) and indirect (I) waves onto the spinal motor neurons.22 23 Increasing MEP amplitude with increase in stimulus intensity generates a sigmoid stimulus response curve.21 MEP may be represented as a percentage of peripheral stimulation-derived compound muscle action potential (CMAP), to account for the lower motor neuron contribution.

Although the MEP reflects the density of corticomotor neuronal projections onto motor neurons similar to the MT,24 the neurotransmitter pathways involved in the generation of the MEP are different. GABAergic agents acting via the GABAA receptor suppress the MEP while glutaminergic and noradrenergic agents increase the MEP amplitude.25 26

The main limitation in using the MEP response as a biomarker of cortical motor neuronal function is the significant intersubject and intertrial variability in MEP latency and amplitude.27

Central motor conduction time (CMCT) is a measure of the time taken by a neural impulse to travel from the motor cortex to stimulate the spinal or bulbar motor neuron, and thus, is also indicative of the integrity of corticospinal tracts.28 CMCT is an overall reflection of time to activation of the pyramidal cells and conduction time of neural impulses in the corticospinal tract.

In TMS studies, CMCT is usually calculated using the F-wave method or cervical nerve root stimulation method.29 30 Both these methods measure the delay between the MEP latency and time to generate a response using peripheral stimulation. The key distinction between these two methods is the inclusion of the spinal motor neuron while measuring the peripheral stimulation time. In the F-wave method, a peripheral nerve is supramaximally stimulated leading to antidromic stimulation which travels up the nerve root to the spinal motor neuron. This in turn stimulates the efferent root orthodromically, generating an F-wave. In the cervical nerve root stimulation, the peripheral conduction time is estimated as the time taken to generate a CMAP by directly stimulating the spinal nerve root. The CMCT can be variable with a range of physiological and subject-dependent factors such as age, gender, hand dominance and neck position

Cortical silent period (CSP) refers to a transient cessation of voluntary activity on EMG in a target muscle measured after magnetic stimulation of the contralateral motor cortex. CSP is a reflection of GABAB receptor-mediated cortical inhibition31 32 and also appears to be influenced by the density of corticomotor neuronal projections onto the spinal motor neuron.27 It is, thus, the longest in the upper limb muscles.

CSP is calculated as the time interval between the onset of the MEP response and resumption of voluntary EMG activity following TMS,31 and increases with stimulus intensity.

Paired pulse TMS paradigms

Paired pulse techniques provide insights into functioning of intracortical excitatory and inhibitory circuits27 by measuring the modulation of the cortical response to a test stimulus preceded by a conditioning stimulus. The two commonly applied paired pulse paradigms comprise are referred to as the constant stimulus33 and threshold tracking19 techniques. Either one can be used to measure the short interval intracortical inhibition (SICI), long interval intracortical inhibition (LICI) and intracortical facilitation (ICF), each of which is an index of cortical motor function.

Paired pulse TMS paradigms (figure 2) used to determine the SICI and ICF consist of a subthreshold conditioning stimulus followed, at prespecified intervals (interstimulus interval (ISI)), by a suprathreshold test stimulus. The constant stimulus paired pulse paradigms33 measure the variation in MEP responses, while keeping the test and conditioning stimuli constant. Inhibition is observed at an ISI of 0–5 ms facilitation at longer intervals between the stimuli. To overcome the issue of inherent MEP variability, which was used as an output measure in the constant stimulus protocols, threshold tracking protocols19 34 were developed. These rely on using a fixed target amplitude MEP response and track the test stimulus intensity required to achieve this response. Higher stimulus intensity required to maintain this target response indicates inhibition while a lower intensity suggests facilitation. The target MEP response is chosen from the steepest part of the stimulus response curve (figure 2C), thus reducing the variation in the outcome variable.

Studies using cervical epidural electrode recordings suggest that SICI is associated with a reduction in the amplitude of I-waves in a temporal pattern consistent with inhibitory postsynaptic potentials mediated via GABAA receptors.35 36 Drugs potentiating GABAA receptor-mediated neurotransmission, thus, increase the SICI. Other neurotransmitter systems may have an indirect role via modulation of GABAA receptors, as indicated by SICI alterations using glutaminergic agents, dopamine agonists and noradrenergic blockers.37 38 The cortical signature of SICI is likely to be a combination of synaptic processes, inhibitory interneuronal interactions and axonal refractoriness.20 39–41

The physiological processes driving ICF remain even less well understood. Interestingly, ICF is decreased by antiglutaminergic agents37 and is not associated with changes in I-waves27 which coincide with SICI.15

LICI occurs when a suprathreshold conditioning stimulus is followed by a test stimulus at an ISI of 50–300 ms.3 LICI seems to be mediated via GABAB receptors.42 43

Short latency afferent inhibition (SAI) is the suppression of TMS-induced MEP response after peripheral nerve stimulation.44 45 Thus, when a median sensory stimulation is administered approximately 20 ms prior to the TMS pulse over the contralateral motor cortex, the MEP response from the APB muscle is suppressed. It reflects inhibitory modulation of large sensory fibres on the motor cortex and is likely to involve central cholinergic transmission.46 47

Repetitive TMS paradigms

Repetitive TMS (rTMS) with applications of trains of TMS pulses over several minutes duration48 produces cortical changes that last beyond the duration of stimulation in a frequency-dependent manner.14 49 Simple rTMS protocols involve application of single stimuli at fixed ISI and their effects depend on the frequency of stimuli used. A low-frequency stimulation (≤1 Hz) depresses cortical excitability, while high-frequency (5–20 Hz) stimulation increases excitability (figure 1). Patterned rTMS protocols use a combination of different ISIs, a common example of this being theta burst TMS (TBS), that incorporates triplet TMS pulses (bursts of three pulses at 50 Hz repeated at 200 ms intervals) to induce longer lasting effects than conventional rTMS protocols for a relatively shorter duration of application.50 Continuous theta burst stimulation, usually involving trains of uninterrupted stimulation for 20–40 s, has an inhibitory effect on corticospinal excitability whereas intermittent theta burst stimulation has the opposite effect.

At a larger scale, TMS may enhance the understanding of system-level changes in brain circuitry. The application of rTMS over a specified cortical region has effects on remote brain areas51 that may modulate network activity in the brain leading to behavioural alterations not directly related to the area being stimulated by the TMS directly.52 In terms of specificity, the same output can be elicited using a variety of stimulation sites. For instance, motor activity changes are associated with stimulation of the primary motor cortex M1,50 supplementary motor area,53 dorsal pre-motor cortex,54 as well as non-motor areas such as the cerebellum55 and dorsolateral prefrontal cortex.56The potential for rTMS effects to last beyond the duration of stimulation has been observed in a number of therapeutic applications in neurological disorders.57 58 However, therapeutic applications of rTMS are outside the scope of this article.

Safety considerations

With the rapid increase in TMS applications in research and rehabilitation trials, safety in the clinical setting remains an important consideration. Although rare, seizure risk is mainly pertinent to rTMS protocols with an estimated risk in the region of 0.1%.59 60 Most reported cases of seizures with TMS occurred before 1998 when higher frequency trains were routinely administered and typically occurred in patients who had a history of seizures. Resting EEG abnormalities have been noted during TMS, though mostly in patients with epilepsy, and they do not predict occurrence of seizures.61 62 Isolated rare cases in patients have been reported since with concomitant seizure threshold lowering drugs (eg, Selective Serotonin Reuptake Inhibitors (SSRI)) or after sleep deprivation.59 Risk of minor adverse events such as mild headache, tinnitus, cutaneous discomfort, neck muscle contraction, nausea, light headedness or syncope, unilateral eye pain and lacrimation remains <5%. To put this into perspective, the risk of seizures with penicillins and carbapenem drugs is up to 5%63 and increases further with predisposing factors. To date, meta-analyses of published treatment trials of TMS64–66 have been reassuring and support safe use of TMS in patients and healthy volunteers.

TMS is considered safe in individuals with other stimulator devices such as Vagal Nerve Stimulator (VNS) systems, cardiac pacemakers and spinal cord stimulators provided that the TMS coil is not activated near the implanted wires.59 Due to risk of induced currents, TMS should be avoided in patients with Deep Brain Stimulator (DBS), cochlear implants and with epidural electrodes. Additional safety studies are required to establish safe levels of currents that could be used with these implanted devices. Ex vivo studies have, reassuringly, demonstrated minimal, well below prescribed safety limits, heating of metal stents and aneurysm clips with rTMS protocols that have current approval for clinical uses.67 68 However, caution is still warranted before more definitive evidence of safety becomes available from in vivo animal models and, subsequently, human studies.

Cortical dysfunction in neurodegenerative disease

Assessment of cortical function in neurodegenerative disease has provided valuable pathophysiological insights and has the potential for diagnostic applications (table 1).

Table 1

Cortical function alterations across neurodegenerative disorders

Emerging biomarkers in amyotrophic lateral sclerosis

Determining the relationship between upper and lower motor neuron dysfunction remains key to understanding the pathogenesis of amyotrophic lateral sclerosis (ALS).69 70 Initial studies using single-pulse TMS approaches demonstrated a reduction in MT and the CSP as features of early disease, providing preliminary evidence for cortical hyperexcitability in ALS.71 72 Paired pulse techniques have, subsequently, provided more detailed evidence cortical excitability in terms of reduction or absence of SICI and increase in ICF.19 SICI reductions precede electrophysiological evidence of peripheral neurodegeneration73 as well as clinical evidence of lower motor neuron dysfunction in ALS.74 SICI and ICF reduction are also seen in atypical variants of ALS with phenotypic predominance of lower motor neuron dysfunction,75 while these changes are not seen in ALS mimic disorders76 77 such as spinobulbar muscular atrophy, despite a comparable disease burden. These findings strongly support the notion of cortical primacy in ALS.78 Other contributory evidence for this theory is the demonstration of reduced transcallosal inhibition in ALS.79 Partial normalisation of SICI following the administration of riluzole,80 an antiglutaminergic drug used in ALS, points to a pathogenic role for cortical hyperexcitability in ALS. This also highlights the potential application of TMS parameters in future clinical trials of ALS.

SICI has been shown to be the greatest sensitivity and specificity for as a diagnostic marker in ALS.81 Combining TMS measures with peripheral neurophysiological measures can, thus, potentially greatly increase the diagnostic accuracy in ALS.82

Motor cortical alterations in Alzheimer’s disease

The appearance of motor signs in Alzheimer’s disease (AD) is a late event in the natural history of the illness83 and is likely due to the spread of pathology into the motor cortices and striatal structures with disease progression.84 TMS studies have demonstrated a bimodal pattern for changes in the MT in AD. RMT appears to be reduced in early AD and shows progressive decline despite anticholinergic treatment.85 86 The early changes may be related to modulation of glutaminergic pathways by changes in activity of muscarinic cholinergic receptors,87 suggesting a degree of functional reorganisation.88 89 In later stages of AD, the observed increase in MT is a likely due to cortical neuronal degeneration, indicative of more widespread cortical dysfunction.86 Evidence regarding SICI changes in AD is more variable.47 90 A more recent study has found alterations in LICI which correlate with cognitive scores.91

Loss of SAI appears to be a more consistent feature in AD47 92 93 and seems to be normalised by administration of cholinesterase inhibitors.47 SAI appears to be mediated by cholinergic neurons92 and indirectly by GABAergic interneuronal inputs to cholinergic pyramidal neurons.94 95 Muscarinic ACh receptor blockade with scopolamine specifically inhibits SAI, while not affecting the SICI, cortical silent period and ICF, which are believed to be mediated by GABAergic interneurons.39 Interestingly, SAI does not seem to be affected in frontotemporal dementia (FTD), a disorder which does not directly involve the cholinergic system96 unlike AD.97

SAI changes have also been demonstrated in patients with Down’s syndrome who are at risk of developing early-onset AD.98 These findings have the potential for translation to the clinic for differentiating FTD from AD and are likely to be more cost effective than imaging modalities such as PET.

TMS has also been used to demonstrate the disruption of long-term potentiation (LTP)-related cortical changes early on in the disease trajectory99 in keeping with animal models of AD.100 As such, LTP-like cortical alterations could provide a viable biomarker useful to assess synaptic impairment and predict subsequent cognitive decline progression in patients with AD.101

Quantifying motor cortex dysfunction in Parkinson’s disease and other movement disorders

While the degeneration of dopaminergic neurons in the substantia nigra and involvement of nigrostriatal pathways are the primary pathogenic changes in Parkinson’s disease (PD), functional changes in the motor cortices have been well recognised.102–104 SICI reductions have been reported in PD105 106 particularly at higher stimulus intensities,107 suggesting a dysfunction in intracortical facilitatory pathways. Longitudinal evaluation of cortical dysfunction in PD revealed alterations in CSP between the less and more affected brain hemispheres which correlate with motor progression.108 SAI reductions have also been documented in PD,109 particularly in the context of cognitive symptoms,110 111 suggesting a possible role for cholinergic pathways in the pathogenesis of cognitive dysfunction. TMS studies have also found alterations in interhemispheric inhibition, supporting the view that mirror movements in patients with PD originate from crossed corticospinal projections rather than unmasking of ipsilateral projections PD.112 113 In genetic forms of PD, distinct patterns have been found using TMS. Reductions in SICI recruitment have been found in asymptomatic Parkin mutation carriers, without significant changes in overall SICI, indicative of altered cortical function in asymptomatic carriers.114 SICI reduction has not been noted in Parkin patients. Given that SICI appears normal in Parkin patients and CMCT is prolonged, the reduced SICI recruitment may be indicative of a compensatory change in the motor cortex to subclinical dopaminergic dysfunction in mutation carriers.

On the other hand, patients with leucine-rich repeat kinase2 appear to have a markedly hyperexcitable motor cortex compared with those with idiopathic PD, which is a likely contributor to functional changes in patients.115

Motor cortical changes appear in the early stages of Huntington’s disease (HD) as shown by imaging studies 116 117 and pathological confirmation of neuronal loss in the primary motor and anterior cingulate cortices.118 Moreover, motor symptomatology correlates with primary motor cortex involvement119 120 while116 118 cognitive and behavioural features seem to correspond with changes of other regions including prefrontal and anterior cingulate cortical areas.116–118 TMS studies have captured early motor cortical dysfunction in HD including a higher MT and a reduced SAI, the latter being related to motor symptoms.119 In addition, cortical hyperexcitability in terms of decreased SICI and increased ICF120 121 has also been shown in HD, especially in the context of motor symptoms, indicating a potential role for both GABA122 and glutaminergic pathways in HD pathogenesis.

Atypical parkinsonian syndromes include progressive supranuclear palsy (PSP), corticobasal degeneration (CBD) and multiple system atrophy (MSA) and are clinically and pathologically heterogeneous disorders. Motor cortical and corticospinal involvement is seen in these disorders to varying degrees.123–125 Reduced SICI and abnormalities in interhemispheric inhibition have been demonstrated in PSP,126 127 the latter being more evident in the Richardson syndrome compared with parkinsonism predominant PSP.128 RMT is elevated in CBD126 129 and along with reduced SICI and may correlate with primary motor cortex atrophy,129 indicating more severe neuronal loss in the motor cortex in CBD. Increased MTs, reduced SICI and interhemispheric inhibition changes have also been demonstrated in MSA.126 130 131 However, the correlation between these changes and clinical features remains less clear,132 133 and findings regarding interhemispheric inhibition are inconsistent.134 Motor cortex functional alterations have also been reported in PSP127 and MSA.131 Overall, findings from TMS studies suggest that primary motor cortex disinhibition may be an early process in PSP. In contrast, in CBD, global changes in inhibitory process may be secondary to neurodegeneration in the motor cortex.

Novel insights into FTD

FTD encompasses three heterogeneous disorders including behavioural variant FTD, semantic dementia and progressive non-fluent aphasia. Characteristic phenotypic features in FTD include deficits in social cognition, executive function, language and behaviour. There is emerging evidence to suggest that ALS and FTD lie on a disease continuum with motor features prominent at one end and cognitive features at the other.135 136 Concurrence of these two conditions in patients with C9orf72 mutation,137 138 occurrence of TAR DNA binding protein-43 pathology in both conditions,139 clinical and electrophysiological evidence of upper motor neuron dysfunction in FTD,140 alongside evidence of behavioural and cognitive function in ALS, are all supportive of this notion.141 142

Motor cortex involvement in FTD occurs with the spread of pathology from frontal regions posteriorly,135 and anterior cingulate and M1 involvement on imaging overlaps with the imaging patterns seen in ALS.143 TMS studies have shown central motor circuit abnormalities in FTD (reduced or absent MEP, increased MEP latency, increased CMCT) even in the absence of clinical evidence of pyramidal tract involvement, while MT and SAI have been found to be normal.96 140 Earlier studies had found no significant changes in SICI and ICF, but more recent studies indicate SICI reductions in FTD.140 144 SICI reductions in FTD seem to occur to a lesser degree than those seen in ALS. The preservation of cholinergic pathways evidenced by relatively normal SAI in conjunction with abnormalities in SICI and ICF has been used to distinguish FTD from AD.144

Understanding and predicting recovery after stroke

Recovery from stroke is modulated by the intrinsic capacity of the brain to reorganise surviving brain networks. This process takes place through a variety of complex cellular processes including inflammation, growth factors, changes in excitatory and inhibitory neurotransmitters, transcriptional changes, axonal sprouting, neurogenesis, gliogenesis and synaptogenesis.145 While there is variation related to stroke subtype and individual patient factors,146 severity of the initial deficit after stroke is the predominant predictor of recovery, referred to as proportional recovery.147 148 The ability to elicit and MEP response after stroke is a predictor of proportional recovery, regardless of the severity of initial impairment.149 150

Studies in the motor domain indicate that patients with mild to moderate upper limb deficit are able to recover 70% of lost function in the first three months after stroke. However, in patients with severe stroke, recovery is proportional to initial severity in about half of the patients with the other half making no recovery at all. Stroke lesion induced structural and functional changes in the brain occur in the early phase after stroke coinciding with a period of heightened reorganisation, which can support some restoration of function referred to as spontaneous biological recovery.147 While the precise biological mechanisms underlying spontaneous biological recovery are incompletely understood, evidence from animal models151 suggests that behavioural training administered in a critical time window152 153 can facilitate this process. The overarching goal of neuromodulatory approaches is to augment the process of spontaneous recovery and to change the trajectory of poor recovery to proportional recovery.

Early after stroke, glutaminergic excitotoxicity leads to cell death and counteracts GABAergic inhibition.145 154 155 The balance between glutaminergic excitotoxicity and GABAergic inhibition can influence regenerative processes and may reverse in later phases of recovery. TMS-based approaches can be used to better understand these excitability changes and to guide therapeutic neuromodulation in an appropriate time window.

Increased transcallosal inhibition from the contralesional hemisphere156 157 may suppress excitability of the lesioned hemisphere. More recent work has determined that transcallosal inhibition from ipsilesional to contralesional hemisphere may increase in patients with chronic stroke.158 Both these patterns seem to interfere with functional recovery.159 160 A meta-analysis of TMS studies of poststroke cortical changes found no asymmetry in interhemispheric inhibition in patients with stroke in the small number of available studies. In terms of experimental rehabilitation programmes, facilitating affected M1 excitability directly may be more beneficial than suppressing unaffected M1 excitability to promote poststroke recovery.161 Contralesional activity may play some role in improving function.162 163 An important determinant of recovery that interacts with excitability changes is the extent of structural damage to key pathways.164 165 Current understanding of recovery is well described under the bimodal balance recovery model.166 This model suggests that changes in interhemispheric activity interact with the extent of surviving neural pathways, referred to as the ‘structural reserve’. Thus, in strokes with a smaller deficit and a large structural reserve, interhemispheric imbalance predicts poorer outcomes. In these patients, restoration of activity towards the physiological equilibrium should be a primary therapeutic goal. On the other hand, in strokes with more severe deficits and lower structural reserve, the interhemispheric imbalance may allow some compensatory changes leading to varying amounts of functional recovery.

TMS has been used to interrogate cortical reorganisation in patients with stroke and can be useful for prognosis. The ability to elicit an MEP response after stimulation of the lesioned motor cortex might help predict motor function recovery.167 168 Conversely, inability to elicit an MEP after ipsilesional TMS and increased MEP after contralesional stimulation seems to predict poorer recovery of motor function.169 170 Likewise, appearance of MEP responses after ipsilesional stimulation, when MEP responses were not elicited previously, is associated with better functional recovery.171 Alterations in cortical excitability in the lesioned hemisphere have been demonstrated using TMS in patients with stroke172 (figure 3). Prolongation of CSP in the lesioned hemisphere, indicating increased intracortical inhibition, has been demonstrated after subcortical stroke.173 On the other hand, SICI and LICI are suppressed in the affected hemisphere,174–176 while ICF seems to be unaltered after stroke.175 177–179 Contralesional changes in excitability are less marked. MEP responses and MTs appear to be largely intact167 178 180–183 in the paretic limb, while some studies suggest alteration in SICI.174 175 178 184 Indeed, recent work evaluating longitudinal changes in cortical excitability after stroke using TMS from as early as the first week after stroke up to a year afterwards shows that contralesional hyperexcitability evolves differently in patients with different stroke types and may have an adaptive role when ipsilesional pathways are significantly disrupted.176 184 SICI is decreased in both the affected and unaffected hemisphere after stroke, but tends to remain suppressed only in patients with larger strokes and more severe clinical deficits.184

Figure 3

Transcranial magnetic stimulation (TMS) may be used to stimulate the perilesional cortex after stroke and/or suppress excitability of the opposite hemisphere.

Clearer understanding of neuroplastic changes underlying recovery is essential for the development of personalised rehabilitation strategies for patients and application in clinical trials165 accounting for the topography of damaged and surviving neural pathways after a stroke. The predicting recovery potential algorithm illustrates how a sequential consideration of clinical, TMS and imaging factors can provide prognostic information for motor function recovery in stroke.185 186 The key factors incorporated into this algorithm are the extent of clinical weakness, ability to elicit an MEP response in the paretic hand and the degree of corticospinal tract involvement on diffusion tensor imaging. Such a sequential approach has been shown to increase therapy efficiency while achieving good clinical outcomes in poststroke rehabilitation.150

In summary, TMS has evolved as a readily accessible, non-invasive neurostimulation tool with potentially wide-ranging diagnostic and prognostic applications. Separately, TMS provides a unique research tool to investigate pathophysiological changes in the cortex in stroke and neurodegenerative disorders. Applications of TMS-based biomarkers in clinical trials are likely to emerge. In an evolving era of precision medicine, TMS-based approaches have the potential to make personalised rehabilitative and restorative interventions in the future a reality, with better understanding of mechanisms of loss of function in neurodegeneration and the trajectory of recovery in stroke.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
  148. 148.
  149. 149.
  150. 150.
  151. 151.
  152. 152.
  153. 153.
  154. 154.
  155. 155.
  156. 156.
  157. 157.
  158. 158.
  159. 159.
  160. 160.
  161. 161.
  162. 162.
  163. 163.
  164. 164.
  165. 165.
  166. 166.
  167. 167.
  168. 168.
  169. 169.
  170. 170.
  171. 171.
  172. 172.
  173. 173.
  174. 174.
  175. 175.
  176. 176.
  177. 177.
  178. 178.
  179. 179.
  180. 180.
  181. 181.
  182. 182.
  183. 183.
  184. 184.
  185. 185.
  186. 186.
View Abstract


  • Contributors MCK and SA conceived the idea for the article. SA drafted the manuscript. All authors revised the manuscript critically for important intellectual content and gave final approval of the version to be published.

  • Funding This work was supported by funding to Forefront, a collaborative research group dedicated to the study of motor neuron disease, from the National Health and Medical Research Council of Australia program grant (#1037746), the Motor Neuron Research Institute of Australia Ice Bucket Challenge Grant and grant aid from Magnetic Health Science Foundation. SA was funded by the Ellison-Cliffe travelling fellowship from the Royal Society of Medicine, UK. AH was funded by NIH P50 DC014664 and NIH ROI DC05375.

  • Competing interests None declared.

  • Patient consent Not required.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.