BOLD MRI responses to repetitive TMS over human dorsal premotor cortex Sven Bestmann, a,b, * Ju ¨ rgen Baudewig, a Hartwig R. Siebner, c John C. Rothwell, b and Jens Frahm a a Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut fu ¨ r Biophysikalische Chemie, 37070 Go ¨ ttingen, Germany b Sobell Department of Motor Neuroscience and Movement Disorders, Institute of Neurology, University College of London, UK c Institute of Neurology, University of Kiel, Germany Received 20 June 2004; revised 20 April 2005; accepted 20 May 2005 Available online 5 July 2005 Functional magnetic resonance imaging (fMRI) studies in humans have hitherto failed to demonstrate activity changes in the direct vicinity of transcranial magnetic stimulation (TMS) that cannot be attributed to re- afferent somatosensory feedback or a spread of excitation. In order to investigate the underlying activity changes at the site of stimulation as well as in remote connected regions, we applied short trains of high- intensity (110% of resting motor threshold) and low-intensity (90% of active motor threshold) repetitive TMS (rTMS; 3 Hz, 10 s duration) over the presumed location of the left dorsal premotor cortex (PMd) during fMRI. Signal increases in the direct vicinity of the stimulated PMd were observed during rTMS at 110% RMT. However, positive BOLD MRI responses were observed with rTMS at both 90% and 110% RMT in connected brain regions such as right PMd, bilateral PMv, supplemen- tary motor area, somatosensory cortex, cingulate motor area, left posterior temporal lobe, cerebellum, and caudate nucleus. Responses were generally smaller during low-intensity rTMS. The results indicate that short trains of TMS can modify local hemodynamics in the absence of overt motor responses. In addition, premotor rTMS cannot only effectively stimulate cortico-cortical but also cortico-subcortical con- nections even at low stimulation intensities. D 2005 Elsevier Inc. All rights reserved. Keywords: fMRI; EMG; SMA; BOLD; Echo-planar imaging; Motor cortex Introduction A number of recent combined transcranial magnetic stimulation (TMS) and functional magnetic resonance imaging (fMRI) studies have confirmed that TMS leads to changes in neural activity in structures beyond the site of stimulation (Bohning et al., 1998, 1999, 2000b; Bestmann et al., 2004). Although similar effects have been observed with positron emission tomography (PET), the ability to perform similar experiments in fMRI allows us safely to make repeated studies on the same individuals at higher temporal and spatial resolution than can be obtained in PET (Bohning et al., 1998, 1999, 2000b; Shastri et al., 1999; Baudewig et al., 2001; Nahas et al., 2001; Bestmann et al., 2004). However, despite their advantages, combined TMS-fMRI studies have produced results that have led to one puzzling question: what happens at the site of stimulation itself? A problem has been that most studies have used TMS over motor cortex, and stimulation here can lead to contralateral movements and sensory feedback that additionally contribute to BOLD MRI signal changes. In fact, stimulation of motor cortex at sub-movement threshold intensities fails to change BOLD activity under the TMS coil, even though changes can be seen at other distant connected sites (Bohning et al., 1998, 1999, 2000b; Baudewig et al., 2001; Kemna and Gembris, 2003; Bestmann et al., 2004). Interestingly, no significant changes in activity have been reported at sites of stimulation in the parietal and premotor cortex in a recent repetitive TMS (rTMS) study even when using intensities well above the threshold needed to activate the primary motor cortex (Kemna and Gembris, 2003). One possible explanation is that the threshold for motor cortex stimulation is not the same as the threshold in non-motor regions (Stewart et al., 2001). Thus, the absence of significant local activity changes in these experiments may simply reflect the fact that the intensity or duration was not sufficient to cause significant BOLD MRI response changes. Here, we performed TMS-fMRI over the presumed location of the dorsal premotor cortex in order to investigate the occurrence of activity changes in the direct vicinity of the stimulation site as well as remote brain regions. This stimulation site has an advantage over motor cortex stimulation in that it does not provoke movements of contralateral muscles and therefore avoids the confounding BOLD signals that result from reafferent feedback. Although the precise site of stimulation with the TMS coil is not well defined, the effects of 1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2005.05.027 * Corresponding author. Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut fu ¨r Biophysikalische Chemie, 37070 Go ¨ttingen, Germany. Fax: +49 551 201 1307. E-mail address: [email protected] (S. Bestmann). Available online on ScienceDirect (www.sciencedirect.com). www.elsevier.com/locate/ynimg NeuroImage 28 (2005) 22 – 29
8
Embed
BOLD MRI responses to repetitive TMS over human dorsal premotor cortex
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.elsevier.com/locate/ynimg
NeuroImage 28 (2005) 22 – 29
BOLD MRI responses to repetitive TMS over human
dorsal premotor cortex
Sven Bestmann,a,b,* Jurgen Baudewig,a Hartwig R. Siebner,c
John C. Rothwell,b and Jens Frahma
aBiomedizinische NMR Forschungs GmbH am Max-Planck-Institut fur Biophysikalische Chemie, 37070 Gottingen, GermanybSobell Department of Motor Neuroscience and Movement Disorders, Institute of Neurology, University College of London, UKcInstitute of Neurology, University of Kiel, Germany
Received 20 June 2004; revised 20 April 2005; accepted 20 May 2005
Available online 5 July 2005
Functional magnetic resonance imaging (fMRI) studies in humans have
hitherto failed to demonstrate activity changes in the direct vicinity of
transcranialmagnetic stimulation (TMS) that cannot be attributed to re-
afferent somatosensory feedback or a spread of excitation. In order to
investigate the underlying activity changes at the site of stimulation as
well as in remote connected regions, we applied short trains of high-
intensity (110% of resting motor threshold) and low-intensity (90% of
activemotor threshold) repetitive TMS (rTMS; 3 Hz, 10 s duration) over
the presumed location of the left dorsal premotor cortex (PMd) during
fMRI. Signal increases in the direct vicinity of the stimulated PMd were
observed during rTMS at 110% RMT. However, positive BOLD MRI
responses were observed with rTMS at both 90% and 110% RMT in
connected brain regions such as right PMd, bilateral PMv, supplemen-
tary motor area, somatosensory cortex, cingulate motor area, left
posterior temporal lobe, cerebellum, and caudate nucleus. Responses
were generally smaller during low-intensity rTMS. The results indicate
that short trains of TMS can modify local hemodynamics in the absence
of overt motor responses. In addition, premotor rTMS cannot only
effectively stimulate cortico-cortical but also cortico-subcortical con-
nections even at low stimulation intensities.
D 2005 Elsevier Inc. All rights reserved.
Keywords: fMRI; EMG; SMA; BOLD; Echo-planar imaging; Motor cortex
Introduction
A number of recent combined transcranial magnetic stimulation
(TMS) and functional magnetic resonance imaging (fMRI) studies
have confirmed that TMS leads to changes in neural activity in
structures beyond the site of stimulation (Bohning et al., 1998,
1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2005.05.027
* Corresponding author. Biomedizinische NMR Forschungs GmbH am
Activity evoked by suprathreshold rTMS over the left PMd
Anatomical/functional location xa ya za t valueb Volumec
L premotor (dorsal) �40 �11 54 8.24 438
L premotor (ventral) �49 2 17 10.29 1339
L auditory cortexd �45 �25 15 16.42 7836
L middle temporal gyrus �50 �52 3 9.26 1299
L thalamus �14 �17 8 6.04 150
L caudate nucleus �11 �4 18 6.72 68
L putamen �27 �11 3 5.82 276
L cerebellar hemisphere �6 �59 �20 6.99 355
L inferior colliculus �3 �20 �2 7.00 56
BL medial superior frontal
gyrus (SMA)
3 �5 55 8.73 898
BL cingulate gyrus 3 4 44 9.26 908
R premotor (dorsal) 43 �3 49 8.92 367
R premotor (dorsal) 42 7 26 7.64 238
R premotor (ventral) 54 0 13 9.14 1127
R auditory cortexd 50 �22 15 15.72 7311
R thalamus 12 �14 8 6.97 172
R caudate nucleus 12 0 14 7.42 630
R inferior colliculus 6 �25 �6 7.80 93
R cerebellar hemisphere 5 55 �15 8.38 701
a Coordinates correspond to center of gravity of respective activation
clusters.b Peak activation within cluster with P < 0.01, corrected.c Rescaled to voxel size 1 � 1 � 1 mm.d Including superior temporal gyrus, ventral part of parietal operculum
and planum temporale. L: left, R: right, BL: bilateral.Fig. 2. BOLD MRI responses to suprathreshold rTMS of left premotor
cortex (group analysis, n = 9, P < 0.01, corrected). (a) Sagittal (x = �40),
coronal ( y = �11), and transverse (z = 55) view of activity in the left PMd.
(b) Six transverse sections showing activity changes in the cingulate gyrus,
PMv, auditory cortex, caudate nucleus, left posterior temporal lobe, medial
geniculate nucleus, and cerebellum with coordinates indicated. Activation
maps are projected onto a template brain (Montreal Neurological Institute,
MNI). L: left, R: right.
S. Bestmann et al. / NeuroImage 28 (2005) 22–29 25
activity within activation clusters were markedly reduced during
low-intensity rTMS (Tables 1, 3). A direct comparison revealed
that stimulation at 110% RMT was associated with a stronger
BOLD MRI signal at the site of stimulation and in distant brain
regions than low-intensity rTMS, including lateral and mesial
premotor regions, left posterior temporal gyrus, and the thalamus
bilaterally (Fig. 4, Table 4). Neither high- nor low-intensity rTMS
evoked significant BOLD MRI signal changes in the caudally
adjacent M1 hand area (Fig. 5b).
Fig. 3. Brain regions with BOLD MRI responses to both suprathreshold
rTMS of left premotor cortex and voluntary finger movement of the right
hand (group analysis, n = 9; P < 0.01, corrected). The data are projected
onto a left-hemispheric 3D surface reconstruction (MNI) and reveal
enhanced activity in the stimulated left PMd (red circle). Additional
activity in the left hemisphere was evoked in PMv, SMA, cingulate gyrus,
auditory cortices, and left posterior middle temporal gyrus.
Discussion
The present study provides evidence that suprathreshold rTMS
over the left premotor cortex is able to elicit BOLD MRI signal
increases in the direct vicinity of the targeted region, as well as in a
range of cortical and subcortical distant brain regions. In contrast,
stimulation at 90% AMT increased activity in connected cortical
and subcortical areas but failed to induce significant BOLD MRI
changes in the directly targeted premotor region. Further ROI
analysis confirmed that no consistent change in average BOLD
MRI signal occurred under the probe in the low-intensity
condition. While the spatial pattern of distant activity changes
was comparable for both intensities, these were stronger during
rTMS at 110% RMT as compared to subthreshold stimulation.
Local effects of premotor rTMS
Previous studies reported a local increase in BOLD MRI signal
at the site of stimulation only when rTMS was applied to M1 at
intensities suprathreshold for evoking a contralateral muscle twitch
(Bohning et al., 1998, 1999, 2000a,b; Baudewig et al., 2001;
Kemna and Gembris, 2003; Bestmann et al., 2003b, 2004).
However, these effects most likely had confounding contributions
from somatosensory feedback caused by the contralateral muscle
Fig. 4. Comparison of BOLD MRI responses to supra- and subthreshold
rTMS over left premotor cortex (group analysis, n = 9; P < 0.01, corrected)
in a sagittal (a), coronal (b), and two transverse (c, d) sections of a reference
brain (MNI) with Talairach coordinates indicated. Suprathreshold rTMS
evoked significantly stronger activity changes than subthreshold rTMS at
the stimulated left PMd (red circle), the right PMd, SMA, cingulate gyrus,
right cerebellum, bilateral superior temporal plane (putative auditory cortex
and SII), bilateral caudate nucleus, and left posterior middle temporal gyrus.
Table 4
Differences in activity between supra- and subthreshold rTMS over left
Anatomical/functional location xa ya za t valueb Volumec
L premotor (ventral) �50 0 12 9.21 1047
L auditory cortexd �45 �27 16 13.93 6370
L middle temporal gyrus �50 �53 3 9.26 1014
L thalamus �12 �17 5 5.89 122
L putamen �26 �11 3 6.12 180
L inferior colliculus �5 �19 �3 6.47 119
BL medial superior frontal
gyrus (SMA)
2 �5 53 8.72 727
BL cingulate gyrus 3 5 44 8.39 679
R premotor (ventral) 53 1 14 9.14 620
R auditory cortexd 53 �22 15 14.11 5308
R caudate nucleus 13 �2 15 6.86 200
R cerebellar hemisphere 5 �55 �15 8.38 588
a Coordinates correspond to center of gravity of respective activation
clusters.b Peak activation within cluster with P < 0.01, corrected.c Rescaled to voxel size 1 � 1 � 1 mm.d Including superior temporal gyrus, ventral part of parietal operculum
and planum temporale. L: left, R: right, BL: bilateral.
S. Bestmann et al. / NeuroImage 28 (2005) 22–2926
movements. In the present study, short trains of rTMS at 110%
RMT induced consistent BOLD MRI signal increases in the
stimulated left PMd. This shows for the first time that rTMS
evokes a local hemodynamic response that can now be visualized
by concurrent fMRI. Our simultaneous EMG recordings indicate
that movement-evoked sensory feedback was not the source of
these effects.
Using suprathreshold rTMS, previous prefrontal rTMS-fMRI
studies have failed to provoke significant local BOLD MRI signal
changes with pulse trains of 1 s duration at 10 Hz (Baudewig et al.,
2001) or 4 Hz (Kemna and Gembris, 2003), respectively. This
suggests that not only the high intensity but also the total number
of pulses (i.e., the length of the rTMS train) is critical for inducing
significant activity changes (as indexed by the BOLD MRI signal).
A similar conclusion was reached in several previous PET studies
(Paus et al., 1998; Speer et al., 2003a,b). Dose-dependent effects of
TMS have also been found in the global mean field amplitude,
taken as an index of the overall brain response due to TMS
(Komssi et al., 2004).
Table 3
Activity evoked by subthreshold rTMS over the left PMd
Anatomical/functional location xa ya za t valueb Volumec
L premotor (ventral) �52 �3 15 7.30 811
L auditory cortexd �47 �23 14 12.86 5619
L middle temporal gyrus �48 �53 4 7.63 251
L thalamus �19 �15 4 6.03 60
BL medial superior frontal
gyrus (SMA)
�1 �6 52 6.02 75
BL cingulate gyrus 3 6 43 6.23 231
R premotor (ventral) 53 �1 11 7.79 575
R auditory cortexd 49 �26 16 12.29 5709
R thalamus 12 �16 9 6.63 103
a Coordinates correspond to center of gravity of respective activation
clusters.b Peak activation within cluster with P < 0.01, corrected.c Rescaled to voxel size 1 � 1 � 1 mm.d Including superior temporal gyrus, ventral part of parietal operculum
and planum temporale. L: left, R: right, BL: bilateral.
Evidence from electrophysiological studies of stimulation over
M1 suggests that there is a progressive increase in the
excitability of local circuits during rTMS especially during
stimulation at frequencies above 1 Hz (Pascual-Leone et al.,
1994; Wu et al., 2000). This may cause a progressive increase of
neural activity and subsequent positive BOLD MRI changes.
PMd
Anatomical/functional location xa ya za t valueb Volumec
L premotor (dorsal) �37 �10 53 5.08 157
L premotor (ventral) �51 5 12 5.52 530
L posterior temporal lobe �50 �51 3 4.96 485
L caudate nucleus �10 �5 19 4.14 55
L auditory cortexd �48 �24 16 5.30 532
BL medial superior frontal
gyrus (SMA)
3 �5 55 6.53 297
BL cingulate gyrus 2 5 47 5.40 319
R premotor (dorsal) 43 �8 52 6.79 130
R premotor (ventral) 41 7 26 5.80 130
R caudate nucleus 13 0 15 4.63 414
R auditory cortexd 51 �20 14 5.72 411
R cerebellar hemisphere 5 �55 �14 5.61 336
a Coordinates correspond to center of gravity of respective activation
clusters.b Peak activation within cluster with P < 0.01, corrected.c Rescaled to voxel size 1 � 1 � 1 mm.d Including superior temporal gyrus, ventral part of parietal operculum
and planum temporale. L: left, R: right, BL: bilateral.
Fig. 5. BOLD MRI signal intensity time courses (mean T SEM, n = 9) in (a) the left PMd and (b) the left M1 as determined from a priori anatomically defined
regions-of-interest (Bestmann et al., 2004). Voluntary finger movement (solid line) and suprathreshold rTMS (dashed line) evoked responses in the stimulated
left premotor cortex, while no significant signal changes were found during subthreshold rTMS (dotted line). In the left M1 caudal to the site of rTMS, only
finger movements resulted in significant activity changes.
S. Bestmann et al. / NeuroImage 28 (2005) 22–29 27
Another possibility is that with increasing numbers of TMS
pulses or higher intensities of individual pulses, there is a gradual
accumulation of oxygen debt that eventually triggers subsequent
blood flow changes. If this were the case, a certain amount of
TMS-evoked activity within a certain time window would be
required to trigger BOLD MRI signal changes. Both views
would be consistent with the lack of significant local BOLD
MRI response changes during short subthreshold TMS applica-
tions (Bohning et al., 2000b; Baudewig et al., 2001). The
absence of significant BOLD MRI changes during subthreshold
rTMS, however, does not exclude the possibility that stimulation
led to changes in local intracortical activity (Sanger et al., 2001).
However, this would mean that combined TMS-fMRI may be
blind to subtle and short-lasting activity changes within intra-
cortical circuits that can be readily detected using surface EMG
or cervical epidural recordings.
Distant effects of premotor rTMS
Remote cortical and subcortical activity changes occurred in
secondary motor areas including the contralateral PMd, the ventral
premotor cortex, cingulate, and SMA bilaterally, and subcortically,
the thalamus, caudate nucleus, and putamen. The nature of the
remote effect of TMS is not well understood. The presumed net
facilitatory effect on overall (inhibitory and excitatory) neural
activity in remote regions may be produced by transsynaptic or
direct (orthodromic or antidromic) activation of cortico-cortical, or
cortico-subcortical neurons. While a combination of these is likely
to account for remote activity changes at sites with relatively direct
connections, antidromic activation is a less likely candidate for a
polysynaptic propagation of activity to regions such as the
contralateral thalamus.
Activity increases in the homologous right PMd during
suprathreshold rTMS could be caused by direct activation of
callosal connections (Marconi et al., 2003) by each TMS pulse.
However, direct activation of transcallosal fibers during subthres-
hold stimulation is less likely given that these fibers have a smaller
diameter than the corticospinal fibers from M1 that are activated at
suprathreshold stimulation intensities. It seems more likely that
subthreshold rTMS changed the ongoing pattern of activity in
connections from the PMd to remote target regions by altering the
activity in local circuits intrinsic to PMd itself.
Notably, no significant activity changes were found in the
adjacent left M1/S1. This contrasts with recent studies that have
shown physiological effects on motor cortex excitability after
single pulse TMS (Civardi et al., 2001) and metabolic as well as
physiological effects on M1/S1 after long periods of premotor
rTMS (Chouinard et al., 2003; Siebner et al., 2003). The different
results in the present experiment may be due to a combination of
the fact that we used a different frequency of rTMS than had been
used in previous studies as well as a smaller number of pulses.
Since the pattern of excitatory and inhibitory activity produced by
rTMS is critically dependent on these parameters, it may have been
that we evoked activation of M1/S1 that resulted in no net change
in metabolic activity.
As TMS generates auditory and vibrotactile stimulation on the
scalp, this raises the question whether activity changes are
mediated by these factors rather than direct effective cortical
stimulation and its propagation into remote regions. In the present
experiments, the comparison of the two effective TMS intensities
revealed a number of secondary motor regions, including
contralateral PMd, PMv, SMA, cingulate motor region, and the
caudate nucleus, that were activated more strongly during supra-
threshold rTMS. In contrast, auditory activation was only margin-
ally larger during suprathreshold rTMS. The finding that there was
extensive overlap in auditory cortices even during finger tapping
(at which TMS was applied at 15% of stimulator output) suggests
that these activations are not due to the cortical stimulation itself,
but rather due to the auditory stimulation.
Notwithstanding the secondary activity changes related to
somatosensory and auditory stimulation, TMS applications to the
motor system target distinct cortical and subcortical circuits that
reflect the known anatomical relationship between such areas.
This means that it is possible to test the effective connectivity of
a variety of brain regions by using TMS to provide a controlled
input into a cortical circuit(s). Although the importance of
stimulus intensity on local and distal BOLD signal changes
needs to be explored further across a wide range of stimulus
intensities, the present results are proof of principle that combined
TMS-fMRI over non-primary motor regions can be used to
monitor functional connectivity in cortical circuits. In the future,
we hope to be able to apply the method to explore how these
systems react to acute and chronic perturbation of function in