Effect of intermittent theta-burst stimulation on akinesia and sensorimotor integration in patients with Parkinson’s disease


  • Adrian Degardin,

    1. Université Lille-Nord de France, EA 4559, CHRU de Lille, Hôpital Roger Salengro, F-59037 Lille Cedex, France
    2. Université Lille-Nord de France, CHRU de Lille, Neurologie, Hôpital Roger Salengro, Lille Cedex, France
    Search for more papers by this author
  • David Devos,

    1. Université Lille-Nord de France, EA 4559, CHRU de Lille, Hôpital Roger Salengro, F-59037 Lille Cedex, France
    2. Université Lille-Nord de France, CHRU de Lille, Neurologie, Hôpital Roger Salengro, Lille Cedex, France
    Search for more papers by this author
  • Luc Defebvre,

    1. Université Lille-Nord de France, EA 4559, CHRU de Lille, Hôpital Roger Salengro, F-59037 Lille Cedex, France
    2. Université Lille-Nord de France, CHRU de Lille, Neurologie, Hôpital Roger Salengro, Lille Cedex, France
    Search for more papers by this author
  • Alain Destée,

    1. Université Lille-Nord de France, CHRU de Lille, Neurologie, Hôpital Roger Salengro, Lille Cedex, France
    Search for more papers by this author
  • Lucie Plomhause,

    1. Université Lille-Nord de France, EA 4559, CHRU de Lille, Hôpital Roger Salengro, F-59037 Lille Cedex, France
    Search for more papers by this author
  • Philippe Derambure,

    1. Université Lille-Nord de France, EA 4559, CHRU de Lille, Hôpital Roger Salengro, F-59037 Lille Cedex, France
    Search for more papers by this author
  • Hervé Devanne

    1. Université Lille-Nord de France, EA 4559, CHRU de Lille, Hôpital Roger Salengro, F-59037 Lille Cedex, France
    2. Université Lille-Nord de France, ULCO, rue Ferdinand Buisson, Calais Cedex, France
    Search for more papers by this author

Dr A. Degardin, 1Université Lille-Nord de France, CHRU de Lille, Neurophysiologie Clinique, as above.
E-mail: adrian.degardin@chru-lille.fr


High-frequency (HF) repetitive transcranial magnetic stimulation (rTMS) over the primary motor cortex (M1) has been shown to reduce akinesia in Parkinson’s disease (PD). Given that the processing of sensory afferents is deficient in PD and might be involved in akinesia, we sought to determine whether or not the application of very HF rTMS [intermittent theta-burst stimulation (iTBS) protocol] over the M1 affected sensorimotor integration (SMI) and akinesia. The experiments were carried out in: (i) 11 patients taking their usual dopaminergic treatment (‘on-drug’); (ii) eight of the latter patients after withdrawal of dopaminergic treatment (‘off-drug’); and (iii) 10 de novo (drug-naive) patients. Sham stimulation was applied in 11 other patients in the ‘on-drug’ condition. SMI was investigated by conditioning a supra-threshold transcranial magnetic stimulation pulse in the motor region controlling the abductor pollicis brevis with a nerve shock over the median nerve at time intervals corresponding to short- and long-latency afferent inhibition (SAI and LAI) and afferent-induced facilitation (AIF). Akinesia was assessed with a pointing test. In on-drug, off-drug and de novo patients, akinesia in the contralateral arm was lower after iTBS. Sham stimulation had no effect. In on-drug patients (but not other groups), SMI was also influenced by iTBS, with an increase in AIF. No changes in SAI and LAI were observed. Our data suggest that iTBS might improve both akinesia and sensory processing in patients with PD taking levodopa.


The effects of repetitive transcranial magnetic stimulation (rTMS) in Parkinson’s disease (PD) have been thoroughly investigated. As Pascual-Leone et al. (1994)’s initial observation of a shortened reaction time and movement time during 5-Hz rTMS, many authors have reported significant reductions in akinesia after high-frequency (HF) rTMS (de Groot et al., 2001; Khedr et al., 2003; Lefaucheur et al., 2004). Some authors have even suggested that the method could be a promising treatment for the motor symptoms in PD (Elahi et al., 2009; Lefaucheur, 2009). The safety of HF rTMS (even with frequencies as high as 50 Hz) was recently confirmed in patients with PD (Benninger et al., 2009, 2011). HF rTMS is able to modify the excitability of local interneuron networks, as demonstrated by an increase in the duration of the silent period (SP; Siebner et al., 2000) and the potentiation of short-latency intracortical inhibition (SICI; Fierro et al., 2008) and intracortical facilitation (ICF; Lefaucheur et al., 2004). Furthermore, HF rTMS is known to induce changes in the excitability of spatially distant cortical areas that are functionally interconnected with the targeted zone (Hallett, 2000). For instance, 5-Hz rTMS over the premotor cortex was able to increase ipsilateral primary motor cortex (M1) excitability (Rizzo et al., 2004), and repetitive M1 stimulation led to excitability changes in sensory areas (Murakami et al., 2008). There is also convergent evidence to show that the application of HF rTMS over cortical layers has a remote influence on striatal function. In a study in the rat, application of 20-Hz rTMS over the frontal cortex induced an elevation in striatal dopamine levels (probably via glutamatergic frontostriatal activation; Keck et al., 2001). In humans, dopamine release in the ipsilateral putamen (as demonstrated by a decrease in [11C] raclopride binding) is thought to occur after application of 10-Hz rTMS (Strafella et al., 2003) and 5-Hz rTMS (Kim et al., 2008) over M1.

Whereas the influence of rTMS has been tested on many circuits known to be impaired in PD, the potential effect of stimulation on networks involved in sensorimotor integration (SMI) has not yet been addressed. Impaired sensory processing in PD has been extensively demonstrated by studies of two-point discrimination, roughness discrimination and kinesthesia (Schneider et al., 1986; Klockgether et al., 1995; Demirci et al., 1997; Rickards & Cody, 1997; Sathian et al., 1997; Zia et al., 2000; Maschke et al., 2003). These sensory impairments may be involved in the pathophysiology of akinesia and rigidity. Indeed, it has been suggested that the dysfunction of basal ganglia caused by nigrostriatal degeneration in PD accounts for the patient’s inability to match the correct motor response to the sensory input and thus execute movement effectively (Tatton et al., 1984). Using TMS, changes in the inhibitory influence of somesthetic inputs on M1 excitability have also been observed in patients with PD (Sailer et al., 2003). The latter study also demonstrated an alteration in short-latency and long-latency afferent inhibition (SAI and LAI, respectively) in patients with PD.

Given that rTMS over M1 is able to induce changes in local networks and distant interconnected areas (such as the basal ganglia, premotor cortex and sensory areas involved in the processing of somesthetic afferent inputs), we hypothesized that excitatory rTMS should modify SMI in parallel with its action on M1 excitability in PD. Of the various rTMS protocols known to exert a transient effect on cortical network excitability, intermittent theta-burst stimulation (iTBS) is especially useful because it requires a shorter application time than conventional rTMS (Huang et al., 2005) and appears to be at least as effective (Zafar et al., 2008; Di Lazzaro et al., 2011). Hence, we sought to determine whether iTBS over the M1 area influences: (i) akinesia; (ii) the excitability of intracortical inhibitory and excitatory circuits; and/or (iii) SMI in PD by conditioning a TMS-evoked motor response with a peripheral stimulus over the median nerve.

Materials and methods


Thirty-two patients with PD participated in the study. Eleven of the patients (referred to as the ‘on-drug’ group) were studied during their usual dopaminergic treatment comprising levodopa and (in some cases) a dopaminergic agonist or entacapone. To investigate the effects of levodopa, which modifies akinesia but whose effects on sensory processing are subject to debate (O’Suilleabhain et al., 2001; Shin et al., 2005; Jacobs & Horak, 2006; Degardin et al., 2009; Mongeon et al., 2009), eight patients from the ‘on-drug’ group also participated in an experimental session after withdrawal of their drug treatment (forming the ‘off-drug’ group). Three of the eight patients were firstly recorded ‘off-drug’ and then ‘on-drug’, whereas the reverse was true for the other five patients. For the ‘off-drug’ experiment, levodopa administration had been withdrawn for at least 12 h, and patients had not taken dopaminergic agonists or entacapone for a time period corresponding to at least five half-lives of the respective drugs. Ten de novo (i.e. drug-naive) patients with PD were also studied, in order to compare the effect of iTBS in untreated patients with different disease durations (i.e. the ‘off-drug’ group vs. the de novo group). Lastly, 11 other patients were recorded with sham stimulation during their usual dopaminergic treatment (the ‘sham on-drug’ group). Table 1 shows the groups’ clinical characteristics. The on-drug groups (receiving either real iTBS or sham stimulation) comprised patients suffering from akinetic-rigid predominant PD and moderate disability (i.e. a motor UPDRS scale of between 10 and 30). All the patients were treated with levodopa, which was combined in some cases with dopaminergic agonists or entacapone. The de novo group comprised patients suffering from recently diagnosed akinetic-rigid predominant PD and who had not received any antiparkinsonian medications. Patients with the tremor predominant form of PD were excluded. Patients were also excluded if they had ferromagnetic implants or deep brain stimulation, a history of seizures, major head trauma, dementia or severe depression, or if they were taking antidepressants. Each patient gave his/her written informed consent to participation. The study was carried out in accordance with the Declaration of Helsinki, and the local ethic committee had approved the protocol.

Table 1.   Patients’ clinical characteristics
Group N GenderAge (years)Disease duration (years)Daily levodopa treatment equivalence (mg)Dopaminergic treatmentAkineto rigid/tremor dominant/mixed parkinsonismHoehn and YahrUPDRS-III
  1. PD, Parkinson’s disease.

PD on drug114 males 7 females61.5 ± 8.56.8 ± 2.7785 ± 418levodopa (n = 11), entacapone (n = 6), pramipexole (n = 3), ropinirole (n = 2), piribedil (n = 1), bromocriptine (n = 1)akineto rigid (n = 7) mixed (n = 4)2 ± 0.6317.4 ± 8.1
PD off drug83 males 5 females61.3 ± 9.66.2 ± 2.5646 ± 282levodopa (n = 8), entacapone (n = 4), pramipexole (n = 2), ropinirole (n = 1), piribedil (n = 1), bromocriptine (n = 1)akineto rigid (n = 5) mixed (n = 3)2 ± 0.7529.3 ± 8.6
PD de novo108 males 2 females60.6 ± 11.81.8 ± 10 akineto rigid (n = 7) mixed (n = 3)1.3 ± 0.4813.6 ± 4.5
PD on drug with sham117 males 4 females61.5 ± 9.98.2 ± 5.2754 ± 485levodopa (n = 11), entacapone (n = 7), pramipexole (n = 4), ropinirole (n = 3), piribedil (n = 1)akineto rigid (n = 6) mixed (n = 5)2.2 ± 0.6319.5 ± 11.6

Electromyographic (EMG) recordings

The patient was seated comfortably in an armchair in a semi-reclined, resting position, with his/her forearm placed on an armrest. Ag–AgCl surface electrodes were used to record the EMG of the abductor pollicis brevis (APB) muscle in a belly–tendon montage. The electrodes were connected to an isolated preamplifier (Digitimer, Hertfordshire, UK). A large reference electrode (connected to the shared preamplifier input) was placed around the wrist. The signal was amplified (× 1000), filtered (10 Hz–1 kHz), digitized at 2 kHz (1401 Micro MKII; Cambridge Electronic Design, Cambridge, UK) and stored on a computer for off-line analysis with customized Signal software (Cambridge Electronic Design).

Single-pulse TMS

Magnetic stimuli were delivered using a figure-of-eight focal coil (external diameter – 9.5 cm) connected to two Magstim 200 stimulators via a bistim module (The Magstim Company, Whitland, UK). The subject wore a polyester swimming cap on which surface markings served as a visual reference; the coil was positioned by the experimenter over the M1 contralateral to the hand most affected by parkinsonism. In order to identify the optimal locations for producing motor-evoked potentials (MEPs) in the APB, fine adjustment of the coil position was performed at the beginning of the experimental session. Stimulation was applied over this scalp point with the coil held tangentially to the scalp, with the handle pointing backwards and sideways (at a 45° angle to the midline). We measured the APB’s resting motor threshold (RMT), defined as the lowest possible stimulus intensity capable of inducing MEPs greater than 50 μV in at least five out of 10 trials. We also measured the APB’s active motor threshold (AMT), defined as the minimum stimulus intensity required to produce a linear motor-evoked response (about 200 μV in 50% of trials) during isometric contraction at about 20% of the maximal voluntary force. SICI and ICF were measured according to the method described by Kujirai et al. (1993). Eight interstimulus intervals (ISIs) were evaluated (2, 3, 4, 7, 11, 13, 15 and 20 ms). Within the series, each ISI was used eight times in pseudo-random order. Unconditioned stimuli were also delivered pseudo-randomly throughout the block. The conditioning stimulus was set to 90% of the AMT, and the test stimulus was set to the minimum intensity required to obtain a 1-mV MEP in the APB. For each ISI, the average peak-to-peak MEP amplitude was measured and expressed as a percentage of the unconditioned MEP. The SP was recorded while subjects maintained a tonic contraction of the APB muscle during thumb–index pinching at about 20% of the maximal voluntary contraction. Eight stimuli were delivered at the intensity needed to obtain a 1-mV MEP in the APB at rest. The duration of the SP was defined as the period between the end of the MEP and the resumption of sustained EMG activity.

Median nerve stimulation (MNS)

The influence of MNS on the APB MEP amplitude was studied while the muscle was at rest. Constant-current, 0.2-ms square-wave single pulses of electrical stimulation were delivered through standard bar electrodes to the median nerve at the wrist, using a constant-current stimulator (the DS7A from Digitimer). The MNS stimulus intensity was set so as to obtain a slight contraction of the APB (i.e. thumb abduction). Nine time intervals between MNS and TMS were explored – 20, 45, 50, 55, 60, 65, 70, 100 and 200 ms. SAI has been shown to occur at ISIs as short as 20 ms (Delwaide & Olivier, 1990; Tokimura et al., 2000). Although the latency of maximal inhibition should vary from one subject to another (due to differences in the sensory fibres’ conduction velocity), SAI is generally present at an ISI of 20 ms (Tokimura et al., 2000; Sailer et al., 2002). At longer ISIs (e.g. 100–200 ms), LAI was consistent, as reported by Chen et al. (1999). In the upper limb, Devanne et al. (2009) recently described afferent-induced facilitation (AIF) at ISIs ranging from 45 to 70 ms. The TMS test stimulus was adjusted so as to obtain a 1-mV MEP in the APB muscle. Eight conditioned stimuli (intermingled with unconditioned stimuli) were delivered within the block in pseudo-random order. For each ISI, the average peak-to-peak MEP amplitude (MNS-conditioned MEP) was measured and expressed as a percentage of the unconditioned MEP.


Each patient underwent iTBS over the M1 contralateral to the hand most affected by parkinsonism at an intensity of 80% of the AMT. Magnetic stimulation was delivered using a 9.5-cm external diameter figure-of-eight focal coil connected to a Medtronic stimulator. The coil was held at an angle of 45 º to the midsagittal line and was positioned over the TMS hotspot for the APB (i.e. the position producing the largest MEPs). Given that: (i) the coil shape differs for single-pulse and rTMS; and (ii) the hotspot may differ for monophasic and biphasic pulses, we checked for the optimal APB stimulation point with the repetitive coil before the iTBS session. The AMT was again determined with this stimulator, as described above. The TBS session consisted of repeated bursts of three pulses at 5 Hz. The pulses within bursts were delivered at 50 Hz. We delivered a series of 20 2-s trains with an interval of 10 s (i.e. a total of 600 pulses; Huang et al., 2005). For the sham on-drug group, sham stimulation was applied over the M1 contralateral to the most affected side. The sham coil had the same shape as the active TMS coil and produced the same noise.


The subject was seated comfortably in an armchair. Sensorimotor performance and akinesia were evaluated in a pointing test with a mechanical tapping device. The test had been inspired by the recommendations of the Core Assessment Program for Surgical Interventional Therapies in PD (Defer et al., 1999). It consisted of alternately touching two buttons (set 30 cm apart) 10 times each and as rapidly as possible with the index finger of the hand most affected by PD. The time to complete the test was measured twice, and the shorter of the two completion times was considered in our analysis. We then evaluated part III of the United Parkinson’s Disease Rating Scale (the UPDRS-III score) in the patients with PD, and determined the AMT and RMT. Next, paired-pulse series, SP series and SMI series were performed in random order. The series were followed by iTBS over M1 for 192 s. After iTBS, the same series were again performed – measurement of the RMT and AMT, SMI series, paired-pulse series and the SP series, followed by two pointing tests, and the evaluation of akinesia and rigidity in both hands with the UPDRS-III score. The latter was always rated by the same neurologist, who was blinded to whether sham or real iTBS had been performed.

Statistical analysis

Statistical analysis was performed with Statistica software, version 10 (StatSoft, Tulsa, OK, USA). The different outcome measures were compared in an analysis of variance for repeated-measures (RM anova, with a linear general model). Data on the MNS-conditioned MEP amplitude were analysed using a two-way RM anova with ISI (three levels – SAI, AIF and LAI) and session (pre-iTBS and post-iTBS) as within-subject repeated-measures, and subject group as a between-subject factor. We also compared the influence of the L-dopa treatment and its interaction with iTBS and ISI by performing a three-way RM anova with ISI (SAI, AIF and LAI), session (pre-iTBS and post-iTBS) and medication (On drug and Off drug) as within-subject repeated-measures. We increased the power of the statistical two-way and three-way anova by reducing the number of ISIs by averaging data from intervals 50–65 ms for AIF and 100–200 ms for LAI. Likewise, RM anova was performed to assess the influence of iTBS, sham stimulation and L-dopa treatment on SICI and ICF. We also grouped the results from intervals 2–4 ms for SICI and ISIs 11–15 ms for ICF. Mauchly’s sphericity test was used to assess the homogeneity of covariances for different levels of the repeated measures factor. When sphericity was not achieved, a Greenhouse–Geisser correction was applied. The mean amplitude (± standard error) at each ISI was presented as the conditioned/unconditioned MEP ratio, in order to minimize interindividual variability. The influence of iTBS on motor thresholds, SP durations and clinical scores was tested with a one-way RM anova with session (pre-iTBS and post-iTBS) as a within-subject repeated-measure and subject group as a between-subject factor. When a group effect was present, we tested for contrasts between pre- and post-iTBS sessions. We looked for baseline differences between groups in terms of age, UPDRS score and motor thresholds with a one-way anova (or an anova on ranks, when required). Differences in disease duration between on-drug with iTBS datasets and on-drug with sham stimulation datasets were examined in a Mann–Whitney rank sum test. Baseline differences between on-drug and off-drug states were tested with a paired t-test or a Wilcoxon signed-rank test, depending on the normality and variance of the data distribution. Lastly, we used Spearman’s correlation test to look for correlations between SAI, LAI and/or AIF on one hand and the UPDRS motor score on the other. The threshold for statistical significance was set to P < 0.05.


The patients’ baseline measurements

Only eight of the 11 on-drug patients were recorded off-drug. The withdrawal of drug treatment was not well tolerated in two patients. One other patient presented too many artefacts, due to severe tremor. No adverse events were noted during or after iTBS. The groups did not differ in terms of age, RMT, AMT and mean levodopa dose. There was a baseline difference between groups in the pointing task before iTBS (F3 = 2.91, = 0.047); the post hoc test indicated a significant difference between the on-drug group receiving sham stimulation and the de novo group (Bonferroni, t21 = 2.82; = 0.045). No other inter-group differences were observed. In particular, on-drug patients receiving iTBS and those receiving sham stimulation did not differ in terms of disease duration (= 126, = 0.521) or the UPDRS score (t21 = 0.75, = 0.46).

The effect of iTBS on motor performance

The time to complete the pointing task was significantly influenced by the factors ‘group’ (F3,1 = 5.11; = 0.0045) and ‘stimulation session’ (F3,1 = 43.61; < 0.0001). There was a significant interaction (F3,1 = 4.21; = 0.011) between the two factors. Pointing test performance was improved by 17% after iTBS in the on-drug group (F1 = 98.71; P < 0.0001); it was improved by 11% in the off-drug group (F1 = 8.44; = 0.0061) and 15% in the de novo group (F1 = 15.44; = 0.0004), but did not change significantly in the sham group (F1 = 0.71; = 0.40). Figure 1 shows the performance levels in all groups before and after iTBS. In the off-drug group, dopaminergic treatment enhanced the mean pointing test score by 12.7%.

Figure 1.

 Box plot of the motor performance of patients with PD, as measured by the time taken to complete a pointing test. For each box, the mean and median are represented by the dashed white line and the solid black line, respectively. The lower and upper boundaries indicate the 25th and 75th percentiles, respectively. Whiskers above and below the box indicate the 90th and 10th percentiles, respectively. The asterisk means that there was a significant difference (P < 0.05) between pre- and post-intervention values. iTBS, intermittent theta-burst stimulation.

The effect of iTBS on arm akinesia and rigidity (by rating the UPDRS-III finger tapping, hand movement and arm rigidity items from 0 to 4) in the different groups is presented in Table 2. Contralateral finger tapping and hand movement were significantly improved in the on-drug group only. In the on-drug and de novo groups, rigidity was significantly decreased after iTBS (F1 = 6.99; = 0.012 and F1 = 13.68; = 0.0007, respectively), but did not change in the two other groups. No significant effect was noted in the ipsilateral arm.

Table 2.   Effect of iTBS on akinesia and rigidity
 Before iTBSAfter iTBSComparison before/after iTBS
  1. iTBS, intermittent theta-burst stimulation; PD, Parkinson’s disease.

Contralateral finger tapping (0–4)PD on drug1.730.31.090.21 = 0.03
PD off drug1.750.71.50.9Ns
PD de novo1.
PD on drug sham1.810.601.630.8Ns
Ipsilateral finger tapping (0–4)PD on drug10.230.630.15Ns
PD off drug1.620.741.570.53Ns
PD de novo10.2610.26Ns
PD on drug sham1.
Contralateral hand movement (0–4)PD on drug1.540.280.910.16 = 0.01
PD off drug20.531.750.46Ns
PD de novo1.30.1510.21Ns
PD on drug sham1.630.671.450.82Ns
Ipsilateral hand movement (0–4)PD on drug0.910.210.720.14Ns
PD off drug1.870.831.620.51Ns
PD de novo0.
PD on drug sham1.
Contralateral upper limb rigidity (0–4)PD on drug10.330.450.21 = 0.0072
PD off drug1.
PD de novo1. = 0.0003
PD on drug sham1.630.51.450.8Ns
Ipsilateral upper limb rigidity (0–4)PD on drug0.450.20.270.14Ns
PD off drug0.750.70.620.5Ns
PD de novo0.
PD on drug sham1.

A two-way RM anova with session (pre-iTBS and post-iTBS) and L-dopa medication (on drug and off drug) as repeated-measures showed that performance in the pointing task was improved by iTBS (F1,1 = 28.79; < 0.0001), but not by L-dopa. The three other motor scores (finger tapping, hand movement and rigidity) were pooled before the statistical analysis. The resulting overall motor score was improved by iTBS (F1,1 = 9.12; = 0.0078), but not by L-dopa treatment (F1,1 = 2.36; = 0.14). Whatever the parameter tested, there was no interaction between the factors.

Lastly, a comparison of off-drug patients with de novo patients revealed that there was a significant influence of iTBS (F1,1 = 24.17; = 0.00016) and group (F1,1 = 5.51; = 0.032) on the pointing test score, but no interaction between these factors. These data suggest that after iTBS (but not before), de novo patients performed the pointing test more rapidly than more akinetic, advanced PD patients (F1,1 = 7.05, = 0.017), and that iTBS is as efficient at the beginning of the disease as it is in advanced disease stages.

The effect of iTBS on MNS-conditioned MEPs

Figure 2 depicts the raw data obtained in a representative on-drug patient before and after iTBS. In the on-drug group, a two-way RM anova indicated that iTBS (but not ISI) had a significant effect (F1,2 = 10.40; = 0.0091). In this group, the conditioned MEP amplitude appeared to have shifted toward higher values, whatever the ISI considered (Fig. 3A). In fact, changes in SAI and LAI did not achieve statistical significance level (SAI before iTBS – 78.61 ± 15.11%; SAI after iTBS – 89.92 ± 21.31%, F1 = 2.93, = 0.118; LAI before iTBS – 84.45 ± 8.68%; LAI after iTBS – 106.36 ± 13.12%, F1 = 3.87, = 0.077), whereas AIF was significantly increased by iTBS (AIF before iTBS – 97.82 ± 15.21%; AIF after iTBS – 129.61 ± 23.82%, F1 = 7.26, = 0.022). After iTBS (but not before), motor performance as evaluated by the pointing test was significantly correlated with the amplitude of AIF (= 11, = 0.0014) in the on-drug group (Fig. 4).

Figure 2.

 Recordings of the APB MEPs evoked by suprathreshold TMS, preceded by MNS at increasing interstimulus intervals (ISIs) in a subject before intermittent theta-burst stimulation (iTBS; left) and after iTBS (right). The lower plots show the control (unconditioned) MEPs.

Figure 3.

 Modulation of abductor pollicis brevis (APB) motor-evoked potentials (MEP) amplitude following MNS in on-drug (A), off-drug (B) and de novo patients (C), either before (black bars) or after intermittent theta-burst stimulation (iTBS; grey bars). (D) The influence of sham stimulation in a group of on-drug patients. The two bars on the left of each graph were obtained at an interstimulus interval (ISI) of 20 ms. The bars in the middle were obtained by averaging the conditioned MEPs at intervals of 50–65 ms (AIF) in all the subjects of a given group. The bars on the right were obtained by averaging conditioned MEPs at ISIs of 100–200 ms (LAI). Bars are the mean (± standard error) MEP amplitude expressed as a percentage of the control (unconditioned) MEP. Asterisks indicate a significant difference (P < 0.05). PD, Parkinson’s disease.

Figure 4.

 Scatter plot showing the significant correlation between the motor score obtained in the pointing task (completion time in seconds) and the afferent-induced facilitation (AIF) after intermittent theta-burst stimulation (iTBS; black circles and solid line). There was no correlation between the two before iTBS (grey circles and dashed line).

In the off-drug group, ISI (but not iTBS) had a significant effect (Fig. 3B). Both before and after iTBS, conditioned MEPs at AIF intervals were significantly more intense (88.01 ± 15.47% and 120.94 ± 26.58%, respectively) than conditioned MEPs at an ISI of 20 ms (52.57 ± 6.27% and 72.12 ± 22.54%, respectively; F1,2 = 33.02, = 0.0007 and F1,2 = 8.68, = 0.022, respectively). In de novo patients, neither iTBS nor ISI had a significant influence on conditioned MEPs (Fig. 3C). Lastly, sham stimulation did not have any influence on SMI (Fig. 3D).

The three-way RM anova used to compare on-drug vs. off-drug patients with L-dopa treatment as an additional repeated factor indicated that the amplitude of MNS-conditioned MEPs was significantly influenced by iTBS (F1,1,2 = 6.99; = 0.017), but not by ISI (F1,1,2 = 2.45; P = 0.10) or L-dopa treatment (F1,1,2 = 1.08; = 0.31). There was no interaction between the different factors.

The effect of iTBS on motor thresholds, SP and paired-pulse TMS

Despite slight changes in the motor thresholds after iTBS, no significant effect of the repetitive train was noted in any of the four groups, neither for AMT (F3,1 = 0.42; = 0.738) nor RMT (F3,1 = 0.08; = 0.972; Table 3). Whatever the group considered, SICI and ICF were never modified by iTBS (F3,1,1 = 3.35, = 0.077; Fig. 5). Similarly, the duration of the SP was not changed after iTBS (F3,1 = 1.92, = 0.146).

Table 3.   Values of RMT, AMT and SP before and after iTBS
 RMT (%)AMT (%)SP (ms)
  1. All values are mean ± SE.

  2. AMT, active motor threshold; iTBS, intermittent theta-burst stimulation; RMT, resting motor threshold; SP, silent period.

On drug with iTBS40.9 ± 2.641.3 ± 2.536.2 ± 2.436.7 ± 2.3138.0 ± 6.1124.0 ± 4.9
Off drug41.4 ± 7.738.1 ± 8.938.8 ± 7.736.8 ± 6.9132.6 ± 15.5130.4 ± 13.2
De novo 39.2 ± 2.239.8 ± 2.734.7 ± 2.034.8 ± 2.2121.7 ± 5.4124.4 ± 6.7
On drug with sham39.4 ± 6.939.3 ± 6.736.4 ± 6.936.2 ± 6.6137.8 ± 17.1139.1 ± 18.4
Figure 5.

 Short-latency intracortical inhibition (SICI) and intracortical facilitation (ICF) before intermittent theta-burst stimulation (iTBS; black bars) and after iTBS (grey bars) in on-drug (A), off-drug (B) and de novo patients (C). (D) SICI and ICF before and after sham stimulation in on-drug patients. Each bar was obtained by averaging the conditioned motor-evoked potentials (MEPs) obtained in the different subjects for ISIs of 2, 3 and 4 ms for SICI, and ISIs of 11, 13 and 15 ms for ICF. Bars are mean (± standard error) MEP amplitude, expressed as a percentage of the control (unconditioned) MEP. APB, abductor pollicis brevis; PD, Parkinson’s disease.


The purpose of this study was to investigate the effects of iTBS on akinesia, motor cortex excitability and the networks involved in SMI. We found that a single session of iTBS applied over M1 in patients with PD led to a reduction of akinesia and rigidity in the contralateral arm. Despite the lack of effects of iTBS on motor thresholds, SICI, ICF and the duration of the SP, the technique markedly enhanced SMI in patients with PD taking dopaminergic treatment, but not in off-drug or de novo patients.

Although some studies have reported that HF rTMS did not significantly alleviate motor symptoms in PD, others have demonstrated changes in motor performance and motor cortex excitability after one or more rTMS sessions (for a review, see Elahi et al., 2009; Lefaucheur, 2009). HF rTMS is often presented as a promising tool for compensating for the changes in motor cortex excitability that accompany the progression of PD (Edwards et al., 2008; Lefaucheur, 2009; Benninger et al., 2011). Of the various protocols designed to increase cortical excitability, iTBS offers a number of real advantages – it has a shorter application time and lower stimulus intensity (80% of the AMT), and enhances cortex excitability more effectively and rapidly than conventional HF rTMS (Huang et al., 2005; Katayama & Rothwell, 2007, Di Lazzaro et al., 2011). To the best of our knowledge, only three studies have investigated the effect of iTBS applied over M1 in PD (Rothkegel et al., 2009; Benninger et al., 2011; Kishore et al., 2011). All three studies were disappointing, with rather weak clinical benefits. For example, Rothkegel et al. (2009) only noted an improvement in performance in the Purdue Pegboard Task, with no changes in other motor tasks, corticospinal excitability, gait or mood. Using a circular coil centred over M1, Benninger et al. (2011) observed a beneficial effect of iTBS on mood in patients with PD, but no improvement in motor function. However, the latter study was performed with repeated sessions of iTBS, and some recent work has suggested that this technique has no cumulative effects (Gamboa et al. (2011). Lastly, Kishore et al. (2011) have shown that neither intermittent nor continuous TBS led to long-term potentiation (LTP)- and long-term depression-like plasticity in M1 in de novo PD patients, whereas modulation of MEP amplitude was observed in healthy age-matched volunteers after both types of stimulation. The latter authors additionally stated that a single dose of L-dopa did not restore TBS-induced plasticity, even though the motor signs of PD were drug-dependent.

The present study demonstrated a reduction in akinesia (as evaluated by performance in a pointing test and the UPDRS-III score) after iTBS over M1, with greater effects in on-drug PD patients than off-drug patients. It is also noteworthy that sham stimulation did not produce this type of effect. In PD, akinesia is generally thought to be caused by changes in the basal ganglia/thalamocortical circuits and subsequent hypoactivation of the supplementary motor area. Repetitive motor cortex stimulation was found to have some benefit in a primate model of PD (Drouot et al., 2004). In parallel with a reduction in akinesia and bradykinesia, HF stimulation of M1 led to normalization of metabolic activity in the supplementary motor area, and of the electrophysiological activity in the internal globus pallidus and subthalamic nucleus. One possible explanation for the effect of iTBS on akinesia relates to a remote action of the stimulus train on striatal dopamine release. This mechanism has been shown to operate in the rat after application of 20-Hz rTMS over the frontal cortex (probably via glutamatergic frontostriatal activation; Keck et al., 2001). In humans, the decrease in [11C] raclopride binding observed after 10-Hz rTMS (Strafella et al., 2003) and 5-Hz rTMS (Kim et al., 2008) over M1 also suggests dopamine release in the ipsilateral putamen. Our present data showed that iTBS was more efficient in on-drug patients than off-drug patients. The greater efficacy of rTMS in patients with PD on dopaminergic treatments was also noted in a previous study using interventional paired associative stimulation (Ueki et al., 2006; Kuo et al., 2008). Given that PD is characterized by a progressive decrease in nigrostriatal presynaptic dopamine storage (Kempster et al., 1989), these findings prompt one to consider the dopamine release hypothesis. However, the latter appears to be improbable because: (i) dopamine release has also been observed after sham stimulation in PD (Strafella et al., 2006); and (ii) we did not observe a significant improvement in motor performance after sham stimulation.

Nonetheless, it is very likely that the iTBS effects observed in the on-drug patients (and not observed in the off-drug ones) are related to dopaminergic neurotransmission or neuromodulation. Dopamine seems to be able to potentiate and prolong the excitatory effect of paired-associative stimulation (PAS), which is believed to induce plasticity in specific motor cortical synapses receiving proprioceptive inputs (Kuo et al., 2008). It has been suggested that this effect mimics the result obtained in animal studies, in which dopamine facilitates associative LTP in vivo (Jay et al., 1996; Gurden et al., 2000). We can thus hypothesize that L-dopa exerts a permissive effect on iTBS-induced plasticity within the neural networks involved in processing proprioceptive inputs. This action could then lead to a reduction in akinesia.

One explanation for the effects of iTBS in on-drug patients with PD relates to a potential local action on M1 and functional reinstatement of the synaptic efficacy of projections from other areas (such as basal ganglia-thalamocortical or corticocortical areas).

In PD, stimulation of the subthalamic nucleus or the internal globus pallidus has been shown to facilitate MEP amplitude – most likely via facilitation of interneurons that induced later I-waves in response to TMS (Hanajima et al., 2004). Given that iTBS also influences intracortical neurons responsible for later I-waves (Di Lazzaro et al., 2008), one can suggest that stimulation helps to restore functional interactions between deep brain nuclei and/or cortical areas mediating afferent inputs and cortical networks within M1. However, in accordance with previous studies using a similar methodology (Rothkegel et al., 2009; Kishore et al., 2011) and that emphasized the lack of changes in M1 excitability following TBS, we found that neither M1 excitability (RMT and AMT) nor intracortical network excitability (SICI or ICF) were modified by iTBS in our study, suggesting that local effects were either weak or absent. One other possible explanation relates to the fact that our methodology prevented us from accurately probing changes in motor cortex excitability. Indeed, we measured changes in RMT and AMT (rather than changes in MEP amplitude) after iTBS, which are less sensitive measures of excitability than MEP size.

It is important to note the observed lack of influence of iTBS on intracortical circuits. This finding suggests that the reduction in akinesia observed in the present study was not due to the restoration of inhibitory and facilitatory intracortical mechanisms, as was suggested in previous work using conventional rTMS. Indeed, our results contrast with the restoration of intracortical inhibitory mechanisms (Siebner et al., 2000; Fierro et al., 2008) and facilitatory mechanisms (Lefaucheur et al., 2004) seen in conventional HF rTMS studies of patients with PD. Literature data on the effects of iTBS on paired-pulse responses are very scarce. In healthy subjects, iTBS was not followed by changes in ICF (Huang et al., 2005), and the stimulation’s influence on SICI was slight and transient (Huang et al., 2005; Murakami et al., 2008; Suppa et al., 2008). During our experimental sessions, paired-pulse and SP series were recorded after the SMI series (i.e. up to 10 min after the end of the TBS train). This time lag (reportedly about 12 min in healthy subjects) might also explain the lack of influence of iTBS on intracortical circuitry (Huang et al., 2005).

One of the most striking findings of the present study relates to the influence of iTBS on SMI in patients with PD receiving their usual dopaminergic treatment. In healthy subjects, MNS leads to complex, time-dependent modulation of motor cortex excitability, with two distinct inhibition phases – SAI at ISIs of about 20 ms (Delwaide & Olivier, 1990; Tokimura et al., 2000); LAI at ISIs up to 100 ms (Chen et al., 1999); and powerful facilitation at ISIs of between 45 and 70 ms (Roy & Gorassini, 2008; Devanne et al., 2009). We recently showed that LAI and AIF are less intense in healthy elderly subjects than in young subjects, whereas SAI was unaffected (Degardin et al., 2011). In fact, the AIF and LAI patterns in healthy elderly subjects were similar to those observed in patients with PD – suggesting that both mechanisms were defective at baseline in PD. This point must be explored further by carefully comparing patients with PD with healthy, age-matched controls. Again, the fact that iTBS influences AIF but neither SAI nor LAI (at least in the on-drug group) suggests a possible influence of the repetitive train on distant networks (e.g. premotor and secondary sensory areas). There are several arguments in favour of a remote effect of iTBS, although this matter is subject to debate. For example, some recent studies have demonstrated a remote decrease in contralateral M1 excitability following iTBS over M1 (Di Lazzaro et al., 2008; Suppa et al., 2008). Furthermore, iTBS over the lateral cerebellum was seen to enhance M1 excitability (Koch et al., 2008). The failure of iTBS to modify somatosensory-evoked potentials (SEPs) when applied over M1 suggests that primary sensory processing (as reflected by a normal N20 wave) was not modified by this type of stimulation in healthy subjects (Katayama & Rothwell, 2007). However, Murakami et al. (2008) showed that although SEPs were not modified by iTBS applied over M1, the HF oscillations that are superimposed on the ascending slope of N20 were enhanced. The latter authors also saw a change in SICI, and further suggested that iTBS may have increased the excitability of the somatosensory cortex via inhibitory γ-aminobutyric acid (GABA)ergic interneurons (Murakami et al., 2008). Afferent input processing in secondary sensory areas might thus be modified by iTBS over the M1.

In conclusion, our study demonstrated a reduction in akinesia and changes in SMI processing (but not in the excitability of local M1 networks) after a single session of iTBS over M1. In line with the previous observation of an increased effect of PAS in patients with PD on dopaminergic medications, we suggest that L-dopa has a permissive effect on iTBS-induced plasticity within the neural networks involved in processing proprioceptive inputs. Further studies could now address the question of whether or not: (i) repetitive iTBS sessions might have a greater influence on sensorimotor circuits; and (ii) iTBS has effects on other sensorimotor impairments in PD.


This work was funded by a grant from the Nord-Pas de Calais regional council. The authors wish to thank David Fraser for helpful comments on the manuscript’s English, and Alain Duhamel for his help with the statistical analysis.


afferent-induced facilitation


active motor threshold


abductor pollicis brevis






intracortical facilitation


interstimulus interval


intermittent theta-burst stimulation


long-latency afferent inhibition


long-term potentiation


primary motor cortex


motor-evoked potential


median nerve stimulation


paired-associative stimulation


Parkinson’s disease


resting motor threshold


repetitive transcranial magnetic stimulation


short-latency afferent inhibition


somatosensory-evoked potentials


short-latency intracortical inhibition


sensorimotor integration


silent period