Content uploaded by Nabin Koirala
Author content
All content in this area was uploaded by Nabin Koirala on Oct 12, 2017
Content may be subject to copyright.
Abstract— Deep brain stimulation of subthalamic nucleus (STN-
DBS) became a standard therapeutic option in Parkinson's disease
(PD), even though the underlying modulated network of STN-DBS
is still poorly described. Probabilistic tractography and
connectivity analysis as derived from diffusion tensor imaging
(DTI) were performed together with modelling of implanted
electrode positions and linked postoperative clinical outcome.
Fifteen patients with idiopathic PD without dementia were
selected for DBS treatment. After pre-processing, probabilistic
tractography was run from cortical and subcortical seeds of the
hypothesized network to targets represented by the positions of the
active DBS contacts. The performed analysis showed that the
projections of the stimulation site to supplementary motor area
(SMA) and primary motor cortex (M1) are mainly involved in the
network effects of STN-DBS. An involvement of the
“hyperdirected pathway” and a clear delimitation of the cortico-
spinal tract were demonstrated. This study shows the effects of
STN-DBS in PD distinctly rely on the network connections of the
stimulated region to M1 and SMA, motor and premotor regions.
I. INTRODUCTION
High-frequency deep brain stimulation of the subthalamic
nucleus (STN-DBS) is an effective therapy option for
patients with Parkinson’s disease (PD) [1]. The effect on
major clinical motor symptoms as tremor, rigidity, and
bradykinesia is unequivocal. Furthermore DBS improve
non-motor symptoms of the disease, possibly through its
network effects [2]. Despite the clinical success of DBS, its
mechanisms at the local and systemic level have not been
fully elucidated. Furthermore there is no clear data on the
anatomical structures targeted. A complex modulation of
the basal ganglia loops or the cortico-subcortical networks
is hypothesised [3]. Early studies attested that high-
frequency stimulation might modulate the neuronal activity
within the STN [4]. However, DBS presumably does not
only change the neural activity in the nuclei but furthermore
targets the fibre tracts entering, exiting, or passing the
stimulation site [5]. Moreover an interaction with the
pathological oscillations in distinct brain networks might be
a critical feature of the DBS induced clinical response in PD
[7, 8].
*Koirala N, Fleischer V, Muthuraman M, Groppa S, are with the Johannes
Gutenberg university hospital, 55131, Mainz and Granert O,Deuschl G is with
University clinic Schleswig-Holstein, 24105,Kiel,Germany; corresponding e-
mail: nkoirala@ui-mainz.de,Vinzenz.Fleischer@unimedizin-mainz.de,
mmuthura@uni-mainz.de,segroppa@uni-mainz.de, o.granert@neurologie.uni-
kiel.de, g.deuschl@neurologie.uni-kiel.de .
Studies on primates and recent studies on humans attested
the existence of the so-called hyperdirect cortical STN
projections [6]. Furthermore this pathway might be of
special importance for the effects of STN-DBS [7].
Here we analyse the connections of the electrode regions
by the use of diffusion tensor imaging (DTI) and
probabilistic tractography. Therefore we focus on cortical
and subcortical projection to the electrode site that might be
involved in the clinical effect of STN-DBS. We hypothesise
that these connections play an important role for STN-DBS
and their characteristics are crucial for the clinical outcome
and stimulation parameters. Furthermore we consider the
connectivity data as possible predictors that relate to
postoperative clinical outcome scores such as UPDRS-III
(UPDRS – motor score) or stimulation parameters for an
optimal clinical response.
II. METHODS
A. Data Acquisition
Fifteen patients with idiopathic PD without dementia
selected for DBS treatment (11 males, age 63.3±8.2, Hoehn
and Yahr 3.5± 0.8) selected for DBS treatment were
randomly included in this study. All patients received, after
clinical and neuropsychological assessment, bilateral STN
electrodes. The UPDRS-III values in the medication OFF,
stimulation ON state have been used to calculate the
quotient to the preoperative UPDRS-III score in medication
ON state (mentioned further as qUPDRS) have been
selected for the parameter of clinical outcome and included
into further analysis. The surgical procedure has been
previously described in detail [8]. The study protocol used
was approved by the local ethics committee and all patients
have signed a written consent regarding the procedure.
All patients underwent a preoperative high resolution
MRI (3T) using an 8-channel SENSE head coil. We
acquired diffusion sensitive MRI of the whole brain at 2
mm isometric voxel resolution covering a field of view of
224 x 224 mm. DTI included three acquisitions of 32
gradient directions plus 5 b0 images for each acquisition (b
value 1000 s/mm2, TE = 59 ms, TR = 11855 ms, fat
saturation “on”, 60 contiguous slices). Moreover we
obtained a high-resolution T1-weighted structural image of
the whole brain using a standard MPRAGE sequence (TR =
7.7 ms, TE = 3.6 ms, flip angle = 8°). The T1-scan
consisted of 160 contiguous sagittal slices with 1 mm
isometric voxels and a field of view = 240 x 240 mm. On
the first postoperative day a further recording was
Network effects and pathways in Deep brain stimulation in
Parkinson’s disease*
Koirala N, Fleischer V, Granert O, Deuschl G, Muthuraman M, Groppa S
performed on a 1.5 T scanner with a protocol consisting of a
T1-weighted structural image of the whole brain using a
standard MPRAGE sequence (TR = 10.7 ms, TE = 1.96 ms,
flip angle = 8°). The structural brain scan consisted of 160
contiguous coronal slices with 2 mm isometric voxel size
and a field of view = 256 x 256 mm.
B. Preprocessing
Electrode positions and electrode trajectory were
determined by performing the following image analysis
procedure:
First step: Post-operative T1 images were used to determine
the position of the electrode lead. The lead was
mathematically modelled by a straight line and the position
was determined from a set of manually placed 3D space
points (markers) along the electrode trajectory. The
electrode trajectory was determined within the MRI T1
weighted images using all three orthogonal views (sagittal,
coronal and transversal). Markers were placed at the target
points, near the points of exit and uniformly along the
trajectory artefact. Finally, a three dimensional least square
optimization procedure was used to determine the exact
position of the trajectory. Based on the optimized lead
position the T1 intensity profile was extracted along the
trajectory. The exact electrode positions were then
determined by shifting the four contacts manually along the
lead such that the center of the intensity dip, apparent in the
extracted intensity profile, was in correspondence with the
lead contacts.
Second step: Geometrically determined electrode contact
positions were used to create spatially Gaussian weighting
masks. The masks were calculated by specifying the
following two standard deviations: I. along the lead to
model contact dimensions, known from manufacturer’s
annotations; II. in orthogonal directions to model
stimulation depth. The multivariate Gaussians were
centered at the contact positions determined as described in
the first step. We restricted our analysis to a mask with an
extension of two standard deviations along the lead and two
standard deviations in depth (isometric mask with 4.7 mm
full width at half maximum (FWHM), corresponding to a
radius of ~2.4 mm). These parameters were selected
considering existing literature that attests that neural
elements up to a distance of 2 mm from the active contact
might be excited by DBS [9]. The generated Gaussian
masks were then used for further analysis. To allow bias-
free definition of seed and target areas unaffected by
subjective judgments about anatomical correspondences, we
built masks for cortical seeds from anatomical coordinates
known from a meta-analysis for activation studies [10]. The
generated masks were spheres with a radius of 5 mm
centered at the following MNI coordinates: primary motor
cortex [M1 (-37 -21 58)], dorsal and ventral premotor
cortex [PMd (-30 -4 58), PMv (-50 5 22)] and SMA (-2 -7
55) [11]. The coordinates were transformed into MNI space
using GingerALE [12]. ROIs of Globus Pallidus internus
(GPi) and Globus Pallidus externus (GPe) were generated
from the MNI probability atlas by including the entire areas
[13].
C. Tractography analysis
The aim of our tractography analysis was to generate
voxel-based connectivity index maps in the regions of the
DBS electrodes. We used all voxels in the basal ganglia and
the midbrain structures as generated by the MNI atlas and
defined this area as target region [14]. A multi-fiber model
was fit to the diffusion data at each voxel, allowing for
tracing through regions of crossing fibers [15]. Here, we
drew 5,000 streamline samples from our seed voxels to form
an estimate of the probability distribution of connections
from each seed voxel using FSL (v 4.1). Tracts generated
are volumes wherein values at each voxel represent the
number of samples (or streamlines) that passed through that
voxel. For the elimination of spurious connections,
tractography in individual subjects was restricted to include
only voxels through which at least 10 percent of all
streamline samples had passed [16]. The probability of
connection to the target mask was obtained from the
proportion of samples that reached each of the voxels. The
individual maps were then normalized to calculate a tract
probability at each voxel of the target region for each tract
and subject. This connectivity values were then extracted
from the Gaussian masks and feed into further analysis.
Further we performed another tractography analysis to
generate a voxel-based connectivity index map to delimitate
the cortical connections to the electrode regions from the
cortico-spinal tract. The analyzed tract started as well from
the M1-mask but passed a conjunction mask of the
ipsilateral cerebral peduncle region and internal capsule
[17]. The connectivity values of the electrode regions have
been then extracted as described above.
D. Statistical analysis
The correlation analysis was performed using SPSS
software (Version 16.0, SPSS Inc, Chicago, IL, USA). To
improve the statistical power of the data analysis, we pooled
the data from both sides. Stimulation intensity amplitudes
and the quotient of the post- to preoperative UPDRS-III
(qUPDRS) were introduced into further analysis of
covariance (ANCOVA). T-test has been calculated for the
clinical outcome measurements. For the cortico-spinal
tractography analysis, we calculated two single linear
regression analyses with the connectivity data from the
cortico-spinal tract and “DBS intensity” and “qUPDRS”-
values.
III. RESULTS
A. Tractography analysis
The analysis of covariance (ANCOVA) including the
continuous factor DBS stimulation intensity revealed a
significant main effect for the factor Seed [F(5, 140)=2.35,
p<0.05]. The interaction between the factors Seed and DBS
intensity was also significant [F(5, 140)=2.30, p<0.05].
Model correlation with the continuous factor DBS intensity
was significant for the variables of connectivity indices from
M1 (r=0.45, F=7.30, p<0.05) and SMA (r=0.39, F=5.28,
p<0.05, Figure 1A & 1B).
Figure 1. Correlation analysis of connectivity ratios for M1 (A) and SMA
(B) to stimulation intensity at the active contact.
The ANCOVA including the continuous factor qUPDRS
revealed a significant main effect for the factor Seed [F(5,
140=8.81, p<0.05] but showed no other significant
interactions between terms. No correlations with
connectivity values from other cortical (PMd, PMv) or
subcortical seeds (GPi or GPe) achieved statistical
significance in the above mentioned analyses.
B. Delimitation of the cortico-spinal tract
The two linear regression analyses for the connectivity
values from the cortico-spinal tract and “DBS intensity”
(t=-1.2, p>0.1) and “qUPDRS” (t=-0.4, p>0.1) were not
significant. On the visual inspection of the tract probability
maps and VTA (volume tissue activation) positions, the
corticospinal tract was positioned more laterally, while for
electrodes localised in the neighbourhood of the cortico-
spinal tract lower stimulation intensities have been chosen,
possibly to reduce the effects (Figure 2).
Figure 2. Probabilistic tractography results with schematic presentation of
the binarized cortico-spinal tract (blue) and electrode region. Color bar:
connectivity values represent the number of subjects with positive voxels and
current intensity at the effective electrode.
IV. DISCUSSION
Using diffusion MRI, we show that the connectivity
pattern as derived from probabilistic tractography from M1
and SMA directly correlate with the applied voltage at the
active contact for an optimal clinical effect after STN-DBS.
The connectivity profile from these two cortical regions
might become important predictors for STN-DBS. The
main purpose of this study was to reconstruct the
anatomical network modulated by STN-DBS. So far little is
known about the systemic mechanisms of the STN-DBS.
Important data on possible network interactions was
obtained from animal studies, while direct translations to
human models are lacking [18].
The direct involvement and imperative role of the
primary motor cortex for the effects of the STN-DBS have
been demonstrated in two recent studies. They showed that
DBS induced antidromic spikes in Layer V pyramidal cells
triggered a dampened oscillation of local field potentials in
cortex with a resonant frequency around 120 Hz [19]. With
optogenetics and solid-state optics a direct activation of
cortical afferents from M1 projecting to the STN region was
observed and an explicitly associated with therapeutic
benefit was determined [20]. Seminal data on the role of the
motor cortex for the effects of the STN-DBS provided a
study on Parkinsonian rats, that showed that the corrective
action is upon the cortex, where stochastic antidromic
spikes originating from the STN directly modify the firing
probability of the corticofugal projection neurons, destroy
the dominance of beta rhythm [7]. In summary these studies
together with our probabilistic and structural data suggest
that STN-DBS specifically modulates the M1-STN and
SMA-STN connections via either the hyperdirect pathway
or possible loops, functionally related circuits in a way that
normalise the overall cortico-basal-ganglia-cortical network
and the circulating pathological activity.
In our view, the correlative evidence of the stimulation
parameters and connectivity values explains the fiber tract
integrity and the associated modifiability profile of the
connection to the motor and premotor areas through STN-
DBS. Our results are supported by existing effective
connectivity data too, showing an increased cortical output
to STN via hyperdirect tract area in the Parkinsonian
primates compared to the control group [21]. Stronger
structural connectivity in these circuits might be further the
basis for the oscillations that have to be counteracted by
STN-DBS.
An important point for the discussion of STN-DBS
effects in the light of these results is the possible direct
activation of the pyramidal tract fibers that might worsen
bradykinesia and akinesia and negatively influence the
clinical outcome [22]. Since both regions M1 and SMA
present wide corticospinal projections the importance of
these for the effects of STN-DBS could not be completely
ruled out. Nonetheless the adjacent studied premotor
cortical areas (PMd and PMv) are similarly to M1 and SMA
sources for dense corticospinal projections [23]. The
connectivity analysis of these regions to the electrode sites
did not depict any correlative relationships. Furthermore a
direct stimulation of the pyramidal tract would activate
cranial or spinal motoneurons and lead to muscle
contractions, which was not the case in all of our patients as
well. The performed analysis of the connectivity data of the
generated cortico-spinal tract and the lack of the correlation
to the stimulation parameters or clinical outcome make the
direct involvement of this pathway for the STN-DBS
effects.
V. CONCLUSION
In conclusion our data suggests that the effects of STN-
DBS in PD distinctly depend on the network connections of
the stimulated region to M1 and SMA, motor and premotor
regions. We observed no correlations with connectivity
values from other cortical (PMd, PMv) or subcortical seeds
(GPi or GPe). Furthermore we witnessed DTI and
probabilistic tractography as the important tools that can be
used to refine STN-DBS targeting and better elucidate the
achieved systemic effects.
REFERENCES
[1] G. Deuschl, C. Schade-Brittinger, P. Krack, J. Volkmann, H.
Schafer, K. Botzel, et al., "A randomized trial of deep-brain
stimulation for Parkinson's disease," New England Journal of
Medicine, vol. 355, pp. 896-908, Aug 31 2006.
[2] C. C. McIntyre and P. J. Hahn, "Network Perspectives on the
Mechanisms of Deep Brain Stimulation," Neurobiol Dis, vol. 38,
pp. 329-37, Jun 2010.
[3] C. C. McIntyre and P. J. Hahn, "Network perspectives on the
mechanisms of deep brain stimulation," Neurobiol Dis, vol. 38,
pp. 329-337, 2010.
[4] L. Garcia, J. Audin, G. D'Alessandro, B. Bioulac, and C.
Hammond, "Dual Effect of High-Frequency Stimulation on
Subthalamic Neuron Activity," The Journal of Neuroscience, vol.
23, pp. 8743-8751, September 24, 2003 2003.
[5] V. Gradinaru, M. Mogri, K. R. Thompson, J. M. Henderson, and
K. Deisseroth, "Optical deconstruction of parkinsonian neural
circuitry," Science, vol. 324, pp. 354-9, Apr 17 2009.
[6] A. Nambu, H. Tokuno, and M. Takada, "Functional significance
of the cortico-subthalamo-pallidal [] hyperdirect'pathway,"
Neuroscience Research, vol. 43, pp. 111-117, 2002.
[7] Q. Li, Y. Ke, Danny C. W. Chan, Z.-M. Qian, Ken K. L. Yung, H.
Ko, et al., "Therapeutic Deep Brain Stimulation in Parkinsonian
Rats Directly Influences Motor Cortex," Neuron, vol. 76, pp.
1030-1041, 12/6/ 2012.
[8] B. Schrader, W. Hamel, D. Weinert, and H. M. Mehdorn,
"Documentation of electrode localization," Mov Disord, vol. 17
Suppl 3, pp. S167-74, 2002.
[9] J. B. Ranck, "Which elements are excited in electrical stimulation
of mammalian central nervous system: a review," Brain Research,
vol. 98, pp. 417-440, 1975.
[10] M. A. Mayka, D. M. Corcos, S. E. Leurgans, and D. E.
Vaillancourt, "Three-dimensional locations and boundaries of
motor and premotor cortices as defined by functional brain
imaging: A meta-analysis," NeuroImage, vol. 31, pp. 1453-1474,
2006.
[11] M. Argyelan, M. Carbon, M. Niethammer, A. M. Ulug, H. U.
Voss, S. B. Bressman, et al., "Cerebellothalamocortical
connectivity regulates penetrance in dystonia," J Neurosci, vol. 29,
pp. 9740-9747, 2009.
[12] S. B. Eickhoff, A. R. Laird, C. Grefkes, L. E. Wang, K. Zilles, and
P. T. Fox, "Coordinate-based ALE meta-analysis of neuroimaging
data: A random-effects approach based on empirical estimates of
spatial uncertainty," Human Brain Mapping, vol. 30, p. 2907,
2009.
[13] J. Mazziotta, A. Toga, A. Evans, P. Fox, and J. Lancaster, "A
probabilistic atlas of the human brain: theory and rationale for its
development the international consortium for brain mapping
(ICBM)," NeuroImage, vol. 2, pp. 89-101, 1995.
[14] J. L. Lancaster, M. G. Woldorff, L. M. Parsons, M. Liotti, C. S.
Freitas, L. Rainey, et al., "Automated Talairach atlas labels for
functional brain mapping," Hum Brain Mapp, vol. 10, pp. 120-
131, 2000.
[15] T. Behrens, H. J. Berg, S. Jbabdi, M. Rushworth, and M.
Woolrich, "Probabilistic diffusion tractography with multiple fibre
orientations: What can we gain?," Neuroimage, vol. 34, pp. 144-
155, 2007.
[16] E. D. Boorman, J. O'Shea, C. Sebastian, M. F. S. Rushworth, and
H. Johansen-Berg, "Individual differences in white-matter
microstructure reflect variation in functional connectivity during
choice.," Curr Biol, vol. 17, pp. 1426-1431, 2007/08/21 2007.
[17] S. Jbabdi, M. W. Woolrich, J. L. R. Andersson, and T. E. J.
Behrens, "A Bayesian framework for global tractography,"
Neuroimage, vol. 37, pp. 116-129, 2007.
[18] T. Hershey, F. J. Revilla, A. R. Wernle, L. McGee-Minnich, J. V.
Antenor, T. O. Videen, et al., "Cortical and subcortical blood flow
effects of subthalamic nucleus stimulation in PD," Neurology, vol.
61, pp. 816-21, Sep 23 2003.
[19] Q. Li, Y. Ke, D. C. W. Chan, Z. M. Qian, K. K. L. Yung, H. Ko,
et al., "Therapeutic Deep Brain Stimulation in Parkinsonian Rats
Directly Influences Motor Cortex," Neuron, vol. 76, pp. 1030-
1041, 2012.
[20] V. Gradinaru, M. Mogri, K. R. Thompson, J. M. Henderson, and
K. Deisseroth, "Optical Deconstruction of Parkinsonian Neural
Circuitry," Science, vol. 324, pp. 354-359, April 17, 2009 2009.
[21] A. Moran, E. Stein, H. Tischler, and I. Bar-Gad, "Decoupling
neuronal oscillations during subthalamic nucleus stimulation in the
parkinsonian primate," Neurobiology of Disease, vol. 45, pp. 583-
90, Jan 2012.
[22] W. Xu, S. Miocinovic, J. Zhang, K. B. Baker, C. C. McIntyre, and
J. L. Vitek, "Dissociation of motor symptoms during deep brain
stimulation of the subthalamic nucleus in the region of the internal
capsule," Experimental Neurology, vol. 228, pp. 294-7, Apr
2011.
[23] H. Tokuno and A. Nambu, "Organization of nonprimary motor
cortical inputs on pyramidal and nonpyramidal tract neurons of
primary motor cortex: An electrophysiological study in the
macaque monkey.," Cereb Cortex, vol. 10, pp. 58-68, 2000/01/
2000.
