NSC 16813 No.

of Pages 11
8 January 2016

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
1
Neuroscience xxx (2016) xxx–xxx

2 DUAL-TASK-RELATED NEURAL CONNECTIVITY CHANGES

3 IN PATIENTS WITH PARKINSON’ DISEASE
4 G. VERVOORT, a* E. HEREMANS, a A. BENGEVOORD, a worse DT performance. Conclusion: PD showed FC alter-
5 C. STROUWEN, a E. NACKAERTS, a ations in DT-related networks, which were not correlated
6 W. VANDENBERGHE b AND A. NIEUWBOER a to DT performance. However, FOG-specific FC alterations
7 a
KU Leuven, Department of Rehabilitation Sciences, Leuven, in DT-related regions involving the precuneus and striatum
8 Belgium were correlated to worse DT performance, suggesting that
b the balance between cognitive and motor networks is
9 University Hospitals Leuven, Department of Neurology,
10 Leuven, Belgium altered. Ó 2016 Published by Elsevier Ltd. on behalf of IBRO.

11 Abstract—Background and objectives: Dual-task (DT) gait

impairment in people with Parkinson’s disease (PD) and Key words: Parkinson’s disease, freezing of gait, resting-
specifically in those with freezing of gait (FOG), reflects state fMRI, dual-tasking.
12
attentional dependency of movement. This study aimed to
elucidate resting-state brain connectivity alterations related
to DT gait abnormalities in PD with and without FOG.
INTRODUCTION 13
Methods: PD patients (n = 73) and healthy age-matched
controls (n = 20) underwent DT gait analysis and resting- Dopaminergic denervation in the basal ganglia is 14
state functional Magnetic Resonance Imaging (rs-MRI) while responsible for reduced movement automaticity in 15
‘off’ medication. Patients were classified as freezer (n = 13) patients with Parkinson’s disease (PD) and increased 16
or non-freezer (n = 60). Functional connectivity (FC) reliance on attention to execute movements (Wu and 17
alterations between PD and controls and between patient
Hallett, 2005). Movement automaticity is often assessed 18
subgroups were assessed in regions of interest (ROIs)
within the fronto-parietal and motor network. Results: PD
using dual-task (DT) paradigms. Dual-tasking can be 19

had longer stance times, shorter swing times and more step defined as the simultaneous execution of two tasks which 20

length asymmetry during DT gait and needed more time and have distinct goals and often involve motor and/or cogni- 21
steps during DT turning compared to controls. Additionally, tive task sets (McIsaac et al., 2015). Although several 22
freezers showed similar impairments and longer double studies have shown DT) interference on gait in PD 23
support times compared to non-freezers during DT gait. (Hausdorff et al., 2003; Rochester et al., 2004, 2014; 24
PD demonstrated hyper-connectivity between the inferior Yogev et al., 2005; Lord et al., 2011), the effect sizes 25
parietal lobule and premotor cortex (PMC) and between the are variable and strongly rely on the nature of the task, 26
cerebellum and the PMC and M1. FOG-specific hypo- cohort selection and medication state. Several studies 27
connectivity within the striatum and between the caudate
have established that dual-tasking aggravates freezing 28
and superior temporal lobe and hyper-connectivity between
the dorsal putamen and precuneus was correlated with
of gait (FOG) Camicioli et al., 1998; Hackney and 29
Earhart, 2010; Spildooren et al., 2010. FOG is defined 30
as a brief, episodic absence or marked reduction of for- 31
*Corresponding author. Address: Department of Rehabilitation
ward progression of the feet despite the intention to walk 32
Sciences, Tervuursevest 101/1501, 3001 Leuven, Belgium.
Tel: +32-16/37-65-75. (Nutt et al., 2011). It is associated with reduced functional 33
E-mail addresses: griet.vervoort@faber.kuleuven.be (G. Vervoort), mobility, increased fall risk and reduced quality of life 34
elke.heremans@faber.kuleuven.be (E. Heremans), aniek.bengevoord (Bloem et al., 2004; Rahman et al., 2008a; Kerr et al., 35
@faber.kuleuven.be (A. Bengevoord), carolien.strouwen@faber.
kuleuven.be (C. Strouwen), evelien.nackaerts@faber.kuleuven.be
2010). Nutt et al. (2011), proposed that an exaggerated 36

(E. Nackaerts), wim.vandenberghe@uzleuven.be (W. Vandenberghe), loss of automaticity and/or frontal executive dysfunction 37
alice.nieuwboer@faber.kuleuven.be (A. Nieuwboer). may partially explain FOG (Nutt et al., 2011), which also 38
Abbreviations: CV, coefficient of variability; D, dominant; DT, dual task; provides a possible rationale for its strong link with dual- 39
FAB, Frontal Assessment Battery; FC, functional connectivity; FD,
framewise displacement; FDR, false discovery rate; FOG, freezing of tasking. 40
gait; GP, globus pallidus; MDS-UPDRS, Movement Disorder Society- Several theoretical frameworks have been put forward 41
Unified Parkinson’s disease rating scale; MFG, middle frontal gyrus; to explain DT deterioration (Meyer and Kieras, 1997; 42
MMSE, Mini-Mental State Examination; MNI, Montreal Neurological
Institute; MoCA, Montreal Cognitive Assessment; ND, non-dominant;
Ruthruff et al., 2003; Tombu and Jolicoeur, 2003; 43

NFOG-Q, new freezing of gait questionnaire; PD, Parkinson’s disease; Wickens, 2008), but studies investigating its neural mech- 44
PMC, premotor cortex; PPN, pedunculopontine nucleus; ROI, region of anisms in PD are scarce. Wu et al. (2008) showed greater 45
interest; rs-fMRI, resting-state functional Magnetic Resonance activation in the prefrontal cortex, middle frontal gyrus 46
Imaging; SMA, supplementary motor area; SNc, substantia nigra
pars compacta; ST, single task; TMT, Trail-Making Test; TR, repetition (MFG), parietal cortex, temporal lobe and cerebellum in 47
time. PD compared to controls while performing a tapping task 48

http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
0306-4522/Ó 2016 Published by Elsevier Ltd. on behalf of IBRO.

1
 NSC 16813 No. of Pages 11
8 January 2016

2 G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx

49 together with letter counting (Wu and Hallett, 2008). In Bank criteria and were able to walk independently for at 108
50 addition, increased brain activity registered in the cerebel- least 10 min. Exclusion criteria were: a Mini-Mental 109
51 lum, the dorsolateral prefrontal cortex and precuneus was State Examination (MMSE) score <24; presence of a 110
52 interpreted to have a supervisory role in DT integration Deep Brain Stimulator (DBS); neurological or motor 111
53 both in PD and healthy subjects (Wu et al., 2008, 2013). comorbidities. The new freezing of gait questionnaire 112
54 As for FOG, a recent functional Magnetic Resonance (NFOG-Q) was used to classify patients as freezers 113
55 Imaging (fMRI) study showed decreased activation in pre- (NFOG-Q >1) (FOG: n = 13) or non-freezers (NFOG: 114
56 frontal areas, the pre-supplementary motor area (pre- n = 60) and was administered at the beginning of the 115
57 SMA) and posterior parietal cortex in freezers compared behavioral test session. In addition, presence of FOG 116
58 to non-freezers during a DT foot pedaling task performed was determined by the physician in all but one patient, 117
59 in supine position in the scanner (Shine et al., 2013a). In who reported to have FOG after watching the video of 118
60 addition, a Diffusion Tensor Imaging (DTI) study demon- the NFOG-Q. Disease severity was assessed by means 119
61 strated that reduced structural connectivity of the right of the Movement Disorder Society-Unified Parkinson’s 120
62 pedunculopontine nucleus (PPN) in freezers was strongly disease rating scale (MDS-UPDRS) part III ‘off’ 121
63 related to DT gait interference outcomes (Peterson et al., medication and disease duration was calculated from 122
64 2015). the moment of first motor symptom onset. All tests were 123
65 While these studies provided essential information performed ‘off’ medication, i.e. at least 12 h after the last 124
66 to unravel the neural correlates underlying DT intake of medication (short and long working action) 125
67 performance, translation to gait is not straightforward intake. All subjects gave written informed consent and 126
68 due to movement restrictions inherent to a scanner the study was approved by the ethics committee of the 127
69 environment(Hanakawa, 2006). While proxy-measures University Hospitals Leuven according to the declaration 128
70 such as mental imagery of gait have proven useful to of Helsinki. 129
71 investigate single-task (ST) walking Snijders et al.,
72 2011; Crémers et al., 2012; Peterson et al., 2014, they Gait analysis 130
73 are less suitable for DT conditions. To overcome the prob-
74 lem of performance dependence (Tessitore et al., 2014), Gait was assessed using the VICON 3D motion analysis 131

75 resting-state fMRI (rs-fMRI) measures brain networks system (ÓVicon Motion Systems Ltd., UK). Subjects 132

76 which remain temporally correlated during rest, providing were instructed to continuously walk back and forth over 133

77 a global window into functional connectivity (FC) changes. a 6-m trajectory in the gait laboratory. They had to walk 134

78 Therefore, this study used rs-fMRI to investigate FC in a straight line in the part of the laboratory where the 135

79 changes underlying DT gait in PD and more specifically camera view was optimal and turn at the outer parts of 136

80 in an exploratory fashion in those with FOG. So far, only the field of view. This was repeated until at least 30 137

81 two rs-fMRI studies highlighted decreased connectivity steps were recorded (i.e. approximately 10 trials). Only 138

82 in the visual and right fronto-parietal network and between straight line walking in the middle of the walkway was 139

83 the mesencephalic locomotor region and SMA in freezers included in the analyses. In addition, patients were 140

84 compared to non-freezers, without modeling direct corre- asked to turn in the center of the walkway for six trials. 141

85 lations with gait outcomes (Tessitore et al., 2012; Fling Sixteen reflective markers were placed on anatomical 142

86 et al., 2014). For this purpose, we analyzed connectivity landmarks of the lower limb and pelvis according to the 143

87 between regions of interest (ROIs) within the motor and Plugin Gait model and were recorded at a sample 144

88 fronto-parietal network, known to be implicated in DT frequency of 100 Hz. Marker trajectories were processed 145

89 and FOG (Wu and Hallett, 2008; Shine et al., 2013b,c) using Nexus software (version 1.8.5) to calculate the 146

90 and performed a correlation analysis with DT gait position and timing of the heel strikes and toe-offs of 147

91 outcomes obtained from 3D gait analysis in PD and each step. This information was used to calculate 148

92 age-matched controls. Based on the work of Wu and spatiotemporal gait kinematics(Pijnappels et al., 2001). 149

93 colleagues (Wu and Hallett, 2008; Wu et al., 2014, Outcome measures for gait at self-preferred speed were: 150

94 2015), we expected that reduced DT gait performance step time (s), % stance time, % swing time, % double sup- 151

95 in PD patients compared to controls would correlate with port time, step length (mm), step width (mm) and gait 152

96 hyper-connectivity in cognitive control areas. In addition, speed (m/s). The average and variability were calculated 153

97 we hypothesized that DT gait impairment would be more for the disease dominant (D) and non-dominant (ND) leg 154

98 pronounced in freezers compared to non-freezers and for all parameters. Variability was expressed as the coeffi- 155

99 that this impairment would correlate to more widespread cient of variability (CV) and asymmetry was calculated 156

100 FC alterations in cognitive and motor control areas between the D and ND leg for patients and between right 157

101 (Tessitore et al., 2012), including the prefrontal cortex, and left leg for controls for all parameters except step 158

102 precuneus and cerebellum. width and gait speed. Stance time, swing time and double 159
support time were expressed as a percentage of the total 160
gait cycle. To maximize the DT impact, gait tests were per- 161
103 EXPERIMENTAL PROCEDURES formed while ‘off’ medication at self-preferred gait speed 162
and during 360° turning (Spildooren et al., 2010; 163
104 Participants
Snijders, 2012). All subjects performed 360° turns to the 164
105 Seventy-three PD patients and 20 healthy age-matched left and right side (Spildooren et al., 2010). Outcome mea- 165
106 controls were recruited. Patients were included if they sures were step number, turn duration (s) and cadence 166
107 were diagnosed with PD according to the UK Brain (steps/min). 167

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx 3

168 ST and DT conditions were applied in a semi- 5-mm full width at half maximum Gaussian kernel. The 224
169 randomized order. An auditory version of the Stroop first four volumes of the rs-fMRI scan were omitted to 225
170 task was used as DT. During this task, subjects are account for saturation effects. First- and second-level 226
171 presented with the Dutch words for ‘high’ and ‘low’ analyses were performed using the CONN toolbox 227
172 through a wireless headset. During the congruent task, (v.15) (http://www.nitrc.org/projects/conn). All rs-fMRI 228
173 the word ‘high’ is pronounced at a high pitch and the images were motion-corrected using six translational 229
174 word low is pronounced at a low pitch. During the and rotational realignment parameters which were 230
175 incongruent task, the word ‘high’ is pronounced at a low visually inspected for excessive movement (>3 mm). 231
176 pitch and the word ‘low’ is pronounced at a ‘high’ pitch. Instantaneous head motion was accounted for by 232
177 Subjects were instructed to respond with the pitch that calculating frame-wise displacement (FD) for each 233
178 was presented though the microphone of the headset volume and performing all analyses on ‘scrubbed’ data 234
179 which was synchronized with the VICON system. The (Power et al., 2014), i.e. censoring frames displaying FD 235
180 task was practiced and recorded while sitting and was >0.5 mm or frame-wise changes in brain image intensity 236
181 subsequently performed during gait. The average and >0.5 %D BOLD. If the proportion of corrupted volumes 237
182 variability (CV) of the reaction time were chosen as exceeded 50%, subjects were excluded from the analysis. 238
183 outcome measures for DT performance. FD correction led to exclusion of nine patients (FOG: 239
n = 4; NFOG: n = 5). 240

184 Cognitive tests

ROI-to-ROI analysis 241
185 All subjects underwent a cognitive test battery to assess
186 general cognition (MMSE, Montreal Cognitive Core regions from the motor control and fronto-parietal 242
187 Assessment (MoCA) and SCOPA-COG), visuospatial network (Wu et al., 2012), that were likely to be involved 243
188 function (Complex Figure of Rey copying) general in DT gait, were selected as ROIs) in the rs-fMRI analysis 244
189 executive function (Frontal Assessment Battery (FAB)) (see Fig. 1). At the subcortical level the dorsal putamen, 245
190 and task switching (Trail-Making Test (TMT), parts A and ventral putamen, caudate, substantia nigra pars com- 246
191 B), Alternating Names Test (ANT) and Alternating Intake pacta (SNc), subthalamic nucleus (STN), globus pallidus 247
192 Test (AIT)). (GP), thalamus (parts projecting to premotor cortex 248
(PMC) and M1) and PPN) were included. On the cortical 249
level the M1, PMC, SMA), pre-SMA, dorsolateral pre- 250
193 Image acquisition
frontal cortex (DLPFC), MFG), inferior parietal lobule 251
194 On a separate day than the behavioral tests, all subjects (IPL), superior temporal lobe, precuneus, posterior cingu- 252
195 underwent a rs-fMRI scan while ‘off’ medication (Philips late cortex (PCC), anterior cingulate cortex (ACC) and 253
196 3T ACHIEVA MRI scanner (Best, The Netherlands)) insula were included. In addition, we also included the 254
197 using a T2*-weighted Gradient Echo Planar Imaging cerebellum (lobe VI) and cerebellar vermis. Subcortical 255
198 (FEEPI) sequence (duration: 435 s, slice number: 31; ROI coordinates were based on previous work from our 256
199 slice thickness: 4 mm; slice gap: 0 mm; repetition time lab (Vercruysse et al., 2013) and cortical ROI coordinates 257
200 (TR): 1700 ms; echo time (TE): 33 ms; flip angle: 90°; on Postuma and Dagher (2006) and Wu et al. (2012). 258
201 matrix: 64  64; FOV: 230  124  230 mm) combined Coordinates originally reported in Talairach space, were 259
202 with a high-resolution anatomical T1-weighted sequence transformed into MNI space using an application of the 260
203 (T1 Turbo Field Echo (TFE) sequence, duration: Yale BioImage Suite Package (http://noodle.med.yale. 261
204 383 ms; slice number: 182; slice thickness: 1.2 mm; edu/~papad/mni2tal/). ROIs were constructed as spheres 262
205 slice gap: 0 mm; TR: 9.624 ms; TE: 4.6 ms; flip angle: with 8-mm radius (except for the SNc, which had a 3-mm 263
206 8°; matrix: 256  256; FOV: 218.4  250  250 mm). radius(Wu et al., 2012) in standard MNI space using the 264
207 Six patients (FOG: n = 1; NFOG: n = 5) and one MarsBaR toolbox(Brett et al., 2002) and were subse- 265
208 control could not undergo the MRI scan due to quently co-registered to each subject’s structural and 266
209 claustrophobia or presence of MRI incompatible functional images. Each subject’s time series for the 267
210 implants. The MRI images of two patients (NFOG) and defined ROIs were Pearson correlated to all other ROIs. 268
211 one control subject were corrupted by large image
212 artifacts. Statistics 269

213 Image processing

Behavioral outcomes. Between-group comparisons 270
214 Rs-fMRI images were pre-processed using SPM8 for continuous subject characteristics were performed 271
215 (Wellcome Department of Imaging Neuroscience, using unpaired T-tests and for categorical subject 272
216 University College London, UK) implemented in characteristics chi-squared tests were used. None of the 273
217 MatlabÒ R2011b. All functional images were realigned outcome measures showed skewed distributions. 274
218 to the reference (mean) image and slice-time corrected. Therefore, spatiotemporal kinematics during gait and 275
219 The realigned images were co-registered to each turning and cognitive scores were compared between 276
220 subject’s T1 image, which were segmented to generate PD patients and controls using a MANOVA to account 277
221 a white matter, gray matter and cerebrospinal fluid for covariance between gait parameters. All variables 278
222 mask. All images were normalized to standard Montreal were entered in the initial model and this was reduced 279
223 Neurological Institute (MNI) space and smoothed with a in step-wise manner until the largest amount of 280

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

4 G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx

Fig. 1. FC matrices for the motor control and fronto-parietal network. Matrices represent the Fisher-transformed Pearson correlation coefficients of
the comparison time series of ROIs within the motor control and fronto-parietal network. (a) shows the FC matrix for PD patients compared to
controls. Orange and red colors represent positive beta-values for ROIs that were hyper-connected in PD compared to controls, while green and
blue colors represent negative beta-values for ROIs that were hypo-connected in PD compared to controls; (b) shows the FC matrix for freezers
compared to non-freezers. Orange and red colors represent positive beta-values for ROIs that had increased FC in freezers compared to non-
freezers, while green and blue colors represent negative beta-values for ROIs that had decreased FC in freezers compared to non-freezers. ACC,
anterior cingulate cortex; GP, globus pallidus; DLPFC, dorsolateral prefrontal cortex; IPL, inferior parietal lobule; M1, primary motor cortex; MFG,
middle frontal gyrus; PMC, premotor cortex; PPC, posterior cingulate cortex; PPN, pedunculopontine nucleus; SMA, supplementary motor area;
SNc, substantia nigra pars compacta; STN, subthalamic nucleus; thal-M1, thalamus projecting to M1; thal-PMC, thalamus projecting to PMC. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

281 variability was explained by the least amount of variables. check if correlations were specific to DT performance, 308
282 A similar approach was used to compare freezers to non- the same gait parameters that were included in the DT 309
283 freezers with the addition of age, disease duration and correlation analysis were also correlated to cognitive 310
284 MDS-UPDRS III scores as covariates. The level of outcomes and FC measures in the ST condition. 311
285 significance was set at a = 0.05. Cohen’s d effect sizes Correlations were considered if R P0.35 or R 6 0.35 312
286 were calculated for all behavioral outcomes. Unlike in and were corrected for multiple comparison by 313
287 previous work (Peterson et al., 2015), we did not calcu- accounting for the number of correlations performed 314
288 late% interference, because of the high within-subject using a Bonferroni correction (a = 0.005). 315
289 variability and low test–retest reliability of this parameter
290 (Strouwen et al., 2015a,b). RESULTS 316

291 Neuroimaging. Differences between patients and Subject characteristics 317

292 controls were tested using ANCOVA’s with age as Subject characteristics are shown in Table 1. PD patients 318
293 covariate and false discovery rate (FDR) corrected and controls were matched for age and sex, but PD 319
294 p-values with a = 0.05. Differences between freezers patients had lower MMSE-scores compared to controls. 320
295 and non-freezers were tested using ANCOVA’s with When comparing freezers and non-freezers, there were 321
296 age, disease duration and MDS-UDPRS III as no differences in disease duration, H&Y stage and 322
297 covariates and FDR-corrected values with a = 0.05 or Leveodopa Equivalent Dose (LED). However, freezers 323
298 uncorrected p-values with a = 0.001. In contrast to the were older and had higher MDS-UPDRS III scores. 324
299 MANOVA analysis, in which all subjects were included, Hence, we controlled for these factors in the analysis. 325
300 rs-fMRI analyses were performed on 8 freezers and 47
301 non-freezers, as 4 FOG and 7 NFOG were excluded
Behavioral analysis 326
302 due to image artifacts and 1 FOG and 4 NFOG due to
303 MRI contra-indications. The final models demonstrated that a much greater 327
number of outcomes discriminated between PD and 328
304 Correlations. Discriminatory PD-CTRL and FOG- controls and between FOG and NFOG during DT vs. ST 329
305 NFOG FC measures were Pearson correlated to conditions. The final DT MANOVA model (F = 2.81; 330
306 significant gait and cognitive outcomes of the final p = 0.004) indicated that lower MMSE, MoCA and FAB- 331
307 MANOVA DT model in the entire patient group. To scores and higher TMTa and TMTb completion times 332

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx 5

Table 1. Subject characteristics

PD (n = 73) CTRL (n = 20) FOG (n = 13) NFOG (n = 60) Between-group p Between- subgroup p

Age (years) 59.5 (±10.2) 57.7 (±8.8) 65.8 (±9.9) 58.2 (±9.9) 0.42 0.02
Sex (M/F) 43/30 14/6 1/12 31/29 0.81 0.01
Disease duration (years) 6.2 (±3.8) NA 7.9 (±5.4) 5.8 (±3.3) NA 0.21
Disease dominant side (L/R) 31/42 NA 7/6 24/36 NA 0.54
MDS-UPDRS III (0–132) 28.1 (±13.2) NA 38.3 (±13.9) 25.9 (±12.0) NA <0.01
H&Y (1–5) 2.0 (±0.7) NA 2.2 (±0.8) 2.0 (±0.6) NA 0.36
NFOG-Q (0–29) NA NA 15.8 (±7.5) NA NA NA
LED (mg/day) 444.9 (±284.1) NA 604.8 (±380.0) 409.7 (±248.9) NA 0.1
MMSE (0–30) 28.4 (±1.6) 29.3 (±0.8) 28.0 (±1.6) 28.5 (±1.6) <0.01 0.32
Means and standard deviations (SD) are shown. Between-group p-values represent comparison between PD patients and controls. Between-subgroup p-values represent
comparison between FOG and NFOG. CTRL: controls; FOG: Freezing of Gait; H&Y: Hoehn and Yahr; LED: Levodopa Equivalent Dose; MDS-UPDRS: Movement Disorder
Society Unified Parkinson’s Disease Rating Scale; MMSE: Mini-Mental State Examination; NA: not applicable; NFOG: No Freezing of Gait; NFOG-Q: new freezing of gait
questionnaire; PD: Parkinson’s Disease.

333 discriminated PD from controls. In addition, PD needed longer turn durations (ND) (Table 3). When we explored 359
334 more steps (D/ND) during turning compared to controls the direct comparison between FOG and CTRL (data 360
335 (Table 2). Although not significant, the final model also not shown) it appeared that the PD-CTRL contrast was 361
336 showed longer stance times (p = 0.06), shorter swing largely driven by the inclusion of freezers. 362
337 times (p = 0.08), increased step length asymmetry
338 (p = 0.08) and longer turn durations (p = 0.13) in PD
339 compared to controls. Reaction times on the auditory
340 Stroop were not retained in the model. The final ST Rs-fMRI analysis 363

341 MANOVA model (F = 2.10; p = 0.04) comparing PD Fig. 1a shows the FC matrix between DT-related ROIs for 364
342 patients and controls showed lower MMSE, MoCA and PD compared to controls. PD showed significant hypo- 365
343 FAB-scores and higher TMTa and TMTb completion connectivity between the left M1 and the right M1 and 366
344 times. In addition, PD patients had longer turn durations right PMC compared to controls. In addition, the left IPL 367
345 (D/ND) and an increased step number during turning (D/ was hypo-connected to the bilateral caudate, as well as 368
346 ND) compared to controls. to the left SNc and the right PPN. Hyper-connections 369
347 When comparing freezers to non-freezers, the final were found between the left cerebellum and the bilateral 370
348 DT MANOVA model (F = 2.41; p = 0.02), longer PMC and M1 in PD compared to controls. In addition, 371
349 stance times (D/ND), shorter swing times (D) and longer the right IPL was hyper-connected to the bilateral PMC 372
350 double support times (D/ND). Freezers additionally had and the left M1. The latter was also hyper-connected to 373
351 longer turn durations (D/ND) and needed more steps (D/ the left SNc (Table 4). 374
352 ND) during turning compared to non-freezers (Table 2). Fig. 1b shows the FC matrix between DT-related ROIs 375
353 These effects were generally stronger in the FOG- for freezers compared to non-freezers. Freezers had a 376
354 NFOG comparison (Cohen’s d: 0.45–1.19) compared to striking pattern of decreased FC between the left caudate 377
355 the PD-CTRL comparison (Cohen’s d: 0.27–0.93). and the right superior temporal lobe and left cerebellum 378
356 When comparing freezers to non-freezers, the final ST compared to non-freezers. The right caudate showed 379
357 MANOVA model (F = 3.64; p = 0.02) indicated larger decreased FC with the bilateral dorsal putamen, the left 380
358 swing time variability (D), shorter swing times (ND) and GP) and the bilateral superior temporal lobe. Freezers 381

Table 2. MANOVA models for DT gait with cognition included

DT gait

Model parameters PD (n = 73) CTRL (n = 20) p-value Cohen’s d Model FOG (n = 13) NFOG (n = 60) p-value Cohen’s d
parameters

MMSE (0–30) 28.4 (±1.5) 29.3 (±0.8) 0.01 0.75 Stance time D (%) 61.3 (±1.7) 59.7 (±1.2) 0.03 1.09
MoCA (0–30) 26.0 (±2.9) 27.5 (±1.5) 0.03 0.65 Stance time ND (%) 61.4 (±2.7) 59.9 (±1.2) 0.02 0.72
TMTa (s) 37.3 (±12.0) 27.8 (±8.0) <0.01 0.93 Swing time D (%) 38.9 (±2.6) 40.3 (±1.0) 0.02 0.71
TMTb (s) 82.2 (±42.1) 54.1 (±12.8) <0.01 0.90 Swing time ND (%) 39.3 (±2.6) 40.2 (±1.1) 0.04 0.45
FAB (0–18) 16.0 (±1.8) 17.2 (±1.0) <0.01 0.82 Double support time D (%) 11.4 (±2.9) 9.8 (±1.2) 0.03 0.72
Stance time D (%) 60.0 (±1.4) 59.4 (±0.6) 0.06 0.56 Double support time ND (%) 11.2 (±2.3) 9.9 (±0.9) 0.03 0.74
Swing time D (%) 39.9 (±1.4) 40.2 (±0.7) 0.08 0.27 Turn duration D (s) 8.7 (±3.4) 5.5 (±2.2) <0.01 1.12
Step length asymmetry (mm) 15.9 (±22.9) 6.7 (±5.6) 0.08 0.55 Turn duration ND (s) 9.0 (±3.9) 5.4 (±2.1) 0.04 1.15
Turn duration D (s) 5.9 (±2.7) 5.0 (±1.5) 0.13 0.41 Turn steps D 16.5 (±5.6) 11.1 (±3.2) 0.01 1.18
Turn steps D 11.8 (±4.1) 9.6 (±1.7) 0.05 0.70 Turn steps ND 16.9 (±6.7) 10.7 (±3.0) <0.01 1.19
Turn steps ND 11.5 (±4.3) 9.6 (±1.8) 0.02 0.58

Means and standard deviations (SD) are shown. P-values represent the individual parameters within the significant general MANOVA model. CTRL, controls; D, disease
dominant side; FAB, Frontal Assessment Battery; FOG, Freezing of Gait; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; ND, disease non-
dominant side; NFOG, No Freezing of Gait; PD, Parkinson’s disease; TMT, Trail Making Test.

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

6 G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx

382 showed increased FC compared to non-freezers between Longer DT stance and shorter swing times in PD 423
383 the right precuneus and the left dorsal putamen (Table 4). compared to controls have been reported in earlier 424
studies in addition to reductions in DT gait speed and 425
384 Correlation analysis stride length (Hackney and Earhart, 2010), increased gait 426
asymmetry and variability (Plotnik et al., 2011; Kelly et al., 427
385 Fig. 2a–f show the FC alterations that proved significantly 2012). Similarly, other work demonstrated that dual- 428
386 correlated to DT outcomes after Bonferroni correction. tasking induced more gait abnormalities in freezers com- 429
387 This shows that worse step and swing times during DT pared to non-freezers (Camicioli et al., 1998; Giladi and 430
388 gait were correlated with FOG-specific decreased FC Hausdorff, 2006; Rahman et al., 2008b) and that DT turn- 431
389 between the left caudate and superior temporal lobe. In ing performance was significantly worse in freezers 432
390 addition, FOG-specific hypo-connectivity between the (Spildooren et al., 2010). However, DT interference was 433
391 right caudate and left GP correlated to a higher step also dependent on the nature of the DT and cohort selec- 434
392 number during DT turning, In contrast, FOG-specific tion. This was apparent from recent papers in early 435
393 increased FC between the left dorsal putamen and right cohorts (Smulders et al., 2012; Rochester et al., 2014), 436
394 precuneus was correlated to worse DT step (D and comparable to our own population of early to mid-stage 437
395 ND), swing (D and ND) and double support times (D PD patients, which showed different interference effects 438
396 and ND). Fig. 3a–c show that the increased FC depending on the level of the cognitive DT. The fact that 439
397 between the left dorsal putamen and right precuneus our PF group included a group of freezers may also partly 440
398 was correlated to worse ST stance (ND) and swing (ND) account for the current results. The finding that, unlike in 441
399 times while hypo-connectivity between the right caudate straight line gait, group differences were present during 442
400 and left dorsal putamen correlated to lower FAB-scores. both ST and DT turning could be explained by the 443
401 There were additional correlations that were increased attentional demand during turning due to the 444
402 significant but did not survive Bonferroni correction. Of higher burden on the postural control system, even in 445
403 note, hyper-connectivity between right PMC and right ST Bengevoord et al. (2015). 446
404 IPL (discriminatory for PD-CTRL) correlated with a The rs-fMRI analysis showed a combined pattern of 447
405 decreased step number during DT and ST turning hypo- and hyper-connectivity in PD compared to 448
406 (R = 0.35; p = 0.01). Decreased FC (discriminatory controls in DT-related neural networks. In line with 449
407 for FOG-NFOG) between the right caudate and dorsal previous ‘off’ medication studies (Yu et al., 2007; 450
408 putamen was correlated to worse performance during Jahanshahi et al., 2010), PD patients showed hyper- 451
409 DT turning (R = 0.36; p < 0.01). In addition, connectivity between the cerebellum and cortical motor 452
410 correlations were found between worse DT stance areas compared to controls which was interpreted as 453
411 (R = 0.35; p < 0.01), swing (R = 0.36; p < 0.01) and compensatory for the reduced motor output stemming 454
412 double support times (R = 0.35; p < 0.01) and hypo- from the striato-thalamo-cortical system. However, our 455
413 connectivity between the right caudate and left GP. correlation analysis did not confirm a direct cerebellar 456
involvement in DT performance in PD (Wu et al., 2013). 457
414 DISCUSSION In contrast, better performance during both ST and DT 458

415 The present study showed that gait at self-preferred turning was correlated to hyper-connectivity between the 459

416 speed was more impaired in DT conditions in PD right IPL and PMC, although these correlations were not 460

417 patients compared to controls, whereas turning was corrected for multiple comparisons. The IPL is known 461

418 affected in both ST and DT conditions. As shown by for its role in conscious movement control as it integrates 462
spatial and sensory information to achieve optimal motor 463
419 Cohen’s d values, a similar but more pronounced
420 pattern was found when comparing freezers to execution (Deiber et al., 1991). In addition, conscious 464

421 non-freezers. Temporal gait kinematics were mostly control of movement in contrast to automatic performance 465
has been associated with greater activation in the PMC 466
422 affected by dual-tasking in freezers.

Table 3. MANOVA models for ST gait with cognition included

ST gait

Model parameters PD CTRL p-value Cohen’s d Model parameters FOG NFOG p-value Cohen’s d

MMSE (0–30) 28.4 (±1.5) 29.3 (±0.8) 0.04 0.75 Swing time ND (%) (40.0 ± 2.2) (41.2 ± 1.0) 0.05 0.70
MoCA (0–30) 26.0 (±2.9) 27.5 (±1.5) 0.05 0.65 Swing time CV D (%) (10.2 ± 6.6) (6.8 ± 2.9) 0.03 0.67
TMTa (s) 37.3 (±12.0) 27.8 (±8.0) <0.01 0.93 Turn duration ND (s) (9.9 ± 6.1) (4.9 ± 1.7) 0.06 1.12
TMTb (s) 82.2 (±42.1) 54.1 (±12.8) <0.01 0.90
FAB (0–18) 16.0 (±1.8) 17.2 (±1.0) <0.01 0.82
Turn duration D (s) (5.2 ± 1.7) (4.1 ± 1.2) 0.01 0.75
Turn duration ND (s) (5.3 ± 1.9) (4.2 ± 1.3) 0.03 0.68
Turn steps D (10.8 ± 2.8) (8.5 ± 1.2) <0.01 1.07
Turn steps ND (10.9 ± 1.9) (8.6 ± 1.6) <0.01 1.31
Means and standard deviations (SD) are shown. P-values represent the individual parameters within the significant general MANOVA model. MMSE, Mini-Mental State
Examination; MoCA, Montreal Cognitive Assessment; TMT, Trail Making Test; FAB, Frontal Assessment Battery; D, disease dominant side; ND, disease non-dominant side;
PD, Parkinson’s disease; CTRL, controls; FOG, freezing of gait; NFOG, no freezing of gait.

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx 7

467 and parietal cortex in PD compared to controls (Sabatini FAB-scores. Moreover, this hypo-connectivity pattern is 495
468 et al., 2000; Haslinger et al., 2001; Wu and Hallett, also in accordance with previous functional imaging stud- 496
469 2005). Although this hyper-activation was suggested to ies that showed hypo-metabolism in the basal ganglia in 497
470 compensate for basal ganglia dysfunction, it may also freezers compared to non-freezers (Bartels et al., 2006; 498
471 indicate less efficiently organized neural interactions dur- Imamura et al., 2012) and fits with the gradual striatal 499
472 ing movement execution (Wu et al., 2015). depletion model (Kish et al., 1988). Our results suggest 500
473 The FOG-specific results of this study were the most that more severe dopamine depletion is underlying DT 501
474 striking, especially given the fact that only a small group impairment, but also that this mechanism plays a greater 502
475 of freezers could be included due to head motion role in those with FOG. Furthermore, the correlation 503
476 artifacts. While the cerebellum was hyper-connected to between intra-striatal hypo-connectivity and FAB-scores 504
477 cortical motor areas in PD compared to controls, it corroborates previous functional imaging studies which 505
478 showed decreased FC with the left caudate in freezers found a correlation between reduced striatal dopamine 506
479 compared to non-freezers. This is in line with a study transporter availability and executive impairment (Polito 507
480 that found weakened striatal-cerebellar connections in et al., 2012; Pellecchia et al., 2015). In agreement, 508
481 PD patients with akinesia/rigidity (Wu et al., 2011) and fits Tessitore et al. (2012) demonstrated lower FAB-scores 509
482 within the predominant pattern of decreased FC that was in freezers compared to non-freezers as well as 510
483 present in freezers compared to non-freezers. While this decreased resting-state activity in the right fronto- 511
484 altered FC involving the left caudate might explain more parietal network (Tessitore et al., 2012). All this confirms 512
485 impaired motor control in freezers compared to cognitive involvement in FOG, in which fronto-striatal 513
486 non-freezers, the right caudate was specifically hypo- pathological changes in freezers might be mediated by 514
487 connected to the striatum and pallidum. In line with earlier striatal dopaminergic deficiency (Kehagia et al., 2010) 515
488 work on right-lateralization of neural impairment in FOG indicating increased dopaminergic denervation in freezers 516
489 (Fling et al., 2013; Shine et al., 2013b) and dual-tasking compared to non-freezers (Bartels et al., 2006). 517
490 (Peterson et al., 2015), these FC alterations might Dysconnections between the caudate and prefrontal 518
491 represent impaired cognitive function in freezers areas have been related to FOG pathophysiology as 519
492 compared to non-freezers. This finding is supported by dopamine release in both areas is necessary for gait 520
493 the correlation between the decreased FC between the execution (Ouchi et al., 2001; Shine et al., 2013c). How- 521
494 right caudate and left dorsal putamen and the lower ever, our results did not show altered FC between the 522

Table 4. Resting-state functional connectivity differences

Seed region Target region T-value FDR-corrected p Uncorrected p

PD (n = 57) < CTRL (n = 18)

Left M1 Right PMC 3.43 0.020 <0.001
Left M1 Right M1 3.01 0.037 0.002
Left IPL Right caudate 3.61 0.011 <0.001
Left IPL Left caudate 2.71 0.043 0.004
Left IPL Left SNc 3.02 0.036 0.002
Left IPL Right PPN 2.67 0.043 0.004

PD > CTRL
Left M1 Right IPL 2.90 0.041 0.002
Left M1 Left SNc 2.97 0.041 0.002
Left M1 Left cerebellum, lobule VI 2.84 0.041 0.003
Right M1 Left cerebellum, lobule VI 2.70 0.045 0.004
Left PMC Right IPL 3.84 <0.01 <0.001
Left PMC Left cerebellum, lobule VI 2.91 0.049 0.002
Right PMC Right IPL 2.79 0.046 0.003
Right PMC Left cerebellum, lobule VI 2.82 0.042 0.003

FOG (n = 8) < NFOG (n = 49)

Left caudate Right superior temporal lobe 3.48 0.017 <0.001
Left caudate Left cerebellum, lobule VI 2.99 0.039 0.002
Right caudate Right dorsal putamen 2.62 0.048 0.005
Right caudate Left dorsal putamen 2.61 0.048 0.005
Right caudate Left GP 2.63 0.048 0.005
Right caudate Right superior temporal lobe 2.69 0.037 0.004
Right caudate Left superior temporal lobe 2.78 0.048 0.003

FOG > NFOG

Left dorsal putamen Right precuneus 3.13 0.051 0.001
Functional connectivity (FC) differences between seed and target regions within the motor control and fronto-parietal network are shown for PD patients compared to
controls. Negative T-values represent hypo-connectivity in patients compared to controls while positive T-values represent hyper-connectivity in patients compared to
controls. CTRL, controls; FDR, false discovery rate; FOG, freezing of gait; GP, globus pallidus; IPL, inferior parietal lobule; M1, primary motor cortex; MFG, middle frontal
gyrus; NFOG, no freezing of gait; PCC, posterior cingulate cortex; PD, Parkinson’s disease; PMC, premotor cortex; PPN, pedunculopontine nucleus; SNc, substantia nigra
pars compacta.

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

8 G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx

Fig. 2. Bonferroni-corrected correlations between FC and DT behavioral measures. Figures represent correlations of discriminatory FC measures
between freezers and non-freezers with discriminatory behavioral measures. White dots and black squares represent non-freezers and freezers
respectively. Lines indicate correlations within the entire patient group and represent Pearson correlations between FC outcomes and DT behavioral
measures that discriminated between freezers and non-freezers. DT, dual-task; FC, Functional Connectivity; FOG, freezing of gait; GP, globus
pallidus; L, left; NFOG, No freezing of gait; R, right.

Fig. 3. Bonferroni-corrected correlations between FC and ST behavioral measures. Figures represent correlations of discriminatory FC measures
between freezers and non-freezers with discriminatory behavioral measures. White dots and black squares represent non-freezers and freezers
respectively. Lines indicate correlations within the entire patient group. (a) Represents a Spearman correlation between FAB-scores and FC
between the right caudate and left dorsal putamen. (b, c) Represent Pearson correlations between FC outcomes and ST measures. FAB, Frontal
Assessment Battery; FC, Functional Connectivity; FOG, Freezing of Gait; L, Left; ND, Non-dominant; NFOG, No freezing of gait; R, right; ST, single
task.

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx 9

523 caudate and prefrontal regions, but rather indicated alterations of the right caudate nucleus, cerebellum and 582
524 hyper-connectivity between the dorsal putamen and pre- superior temporal lobe when comparing patients with 583
525 cuneus. The precuneus has been suggested to play a and without FOG, supporting their role in executive 584
526 central executive role in DT performance and was more impairment and loss of movement automaticity in FOG. 585
527 activated in PD patients during a DT fMRI study (Wu However, the current study did not provide evidence for 586
528 and Hallett, 2008). Although we did not find hyper- direct involvement of frontal executive function in DT 587
529 connectivity involving the precuneus in the PD-CTRL performance in PD. We did demonstrate that the 588
530 comparison, the increased FC in freezers compared to precuneus, which is suggested to have an integrative 589
531 non-freezers might suggest an ineffective compensatory role in DT performance, showed hyper-connectivity 590
532 strategy as it proved to be inversely correlated to DT gait implying that the balance between the motor and 591
533 performance. It may indicate recruitment of additional cognitive control network is altered and may underlie DT 592
534 central executive capacity and attention to perform DTs impairment in PD, and especially in those who have FOG. 593
535 (Wenderoth et al., 2005). Lack of altered FC involving pre-
536 frontal areas might be explained by the relatively small
Acknowledgments—We would like to thank all subjects for their 594
537 ROI-sizes that were used for the relatively large prefrontal motivated and enthusiastic participation in this study. Support 595
538 areas. Future studies might benefit from using ROI-masks for this study was provided through a grant from the Special 596
539 that cover larger regions of the prefrontal cortex, although Research Fund of KU Leuven, Belgium (contract OT/11/091), 597
540 this might come at the cost of reduced spatial specificity. Malou-Malou funds of the King Baudouin Foundation, Research 598
541 Furthermore, we found decreased FC between the Foundation Flanders project G.0906.11 and funding through the 599
542 bilateral caudate and superior temporal lobe that Gabe Foundation. All authors report no conflict of interest. E. 600
543 correlated to worse DT gait quality. Since projections Heremans is a Postdoctoral Researcher, E. Nackaerts a 601

544 from superior temporal gyrus to the caudate are part of Research Assistant and W. Vandenberghe is a Senior Clinical 602

545 a complex associative network through which cognitive Investigator of the Research Foundation Flanders (FWO). 603

546 control areas may influence motor output (Selemon and

547 Goldman-Rakic, 1985), it is not surprising that this proved
548 to be a factor in DT impairment. This is in sharp contrast REFERENCES 604
549 with the increased activity within the temporal lobe, as
550 well as the MFG, cerebellum, prefrontal and parietal cor- Bartels AL, de Jong BM, Giladi N, Schaafsma JD, Maguire RP, 605
551 tex that was associated with combining finger tapping and Veenma L, Pruim J, Balash Y, Youdim MB, Leenders KL (2006) 606
552 letter counting in PD in earlier work using a task-based Striatal dopa and glucose metabolism in PD patients with freezing 607
of gait. Mov Disord 21:1326–1332. 608
553 analysis (Wu and Hallett, 2008).
Bengevoord A, Vervoort G, Spildooren J, Heremans E, 609
554 In contrast to previous studies (Shine et al., 2013d), Vandenberghe W, Bloem BR, Nieuwboer A (2015) Center of 610
555 we found no FC alterations involving the PPN, except mass trajectories during turning in patients with Parkinson’s 611
556 for hypo-connectivity between the PPN and IPL in PD disease with and without freezing of gait, Gait Posture, in press. 612
557 compared to controls. This brainstem area was also pre- Bloem BR, Hausdorff JM, Visser JE, Giladi N (2004) Falls and 613
558 viously shown to be hyper-connected to the SMA in freez- freezing of gait in Parkinson’s disease: a review of two 614
interconnected, episodic phenomena. Mov Disord 19:871–884. 615
559 ers compared to non-freezers in rest (Fling et al., 2014)
Brett M, Anton J-L, Valabregue R, Poline JB (2002) Region of interest 616
560 and showed asymmetric structural connectivity that corre- analysis using an SPM toolbox, 8th International Conference on 617
561 lated with DT interference in freezers (Peterson et al., Functional Mapping of the Human Brain, Sendai, Japan. 618
562 2015). Our lack of PPN-related results might be attributed Camicioli R, Oken BS, Sexton G, Kaye JA, Nutt JG (1998) Verbal 619
563 to the less severely affected freezers in the present study fluency task affects gait in Parkinson’s disease with motor 620
564 or to different coordinates used to define the PPN ROI freezing. J Geriatr Psychiatry Neurol 11:181–185. 621

565 compared to previous work(Fling et al., 2013). Crémers J, D’Ostilio K, Stamatakis J, Delvaux V, Garraux G (2012) 622
Brain activation pattern related to gait disturbances in Parkinson’s 623
566 Finally, the main limitation of our study was the small
disease. Mov Disord 27:1498–1505. 624
567 number of freezers. Next to the power issues in the FC Deiber MP, Passingham RE, Colebatch JG, Friston KJ, Nixon PD, 625
568 analyses, this may also have also led to a compromised Frackowiak RS (1991) Cortical areas and the selection of 626
569 detection of behavioral differences between freezers movement: a study with positron emission tomography. Exp 627
570 and non-freezers. Future studies should combine larger Brain Res 84:393–402. 628
571 sample sizes with imaging modalities that enable direct Fling BW, Cohen RG, Mancini M, Nutt JG, Fair DA, Horak FB (2013) 629
Asymmetric pedunculopontine network connectivity in 630
572 investigation of neural alterations during DT gait, using
parkinsonian patients with freezing of gait. Brain 136:2405–2418. 631
573 electroencephalography (EEG) Shine et al., 2014 or func- Fling BW, Cohen RG, Mancini M, Carpenter SD, Fair DA, Nutt JG, 632
574 tional Near Infrared Spectroscopy (fNIRS) Maidan et al., Horak FB (2014) Functional reorganization of the locomotor 633
575 2015; Mirelman et al., 2014 together with modalities that network in Parkinson patients with freezing of gait. PLoS One 9: 634
576 offer good anatomical resolutions (e.g. MRI). e100291. 635
Giladi N, Hausdorff JM (2006) The role of mental function in the 636
pathogenesis of freezing of gait in Parkinson’s disease. J Neurol 637
Sci 248:173–176. 638
577 CONCLUSION Hackney ME, Earhart GM (2010) The effects of a secondary task on 639
forward and backward walking in Parkinson’s disease. 640
578 our results support FC alterations in PD compared to Neurorehabil Neural Repair 24:97–106. 641
579 controls in areas that have previously been associated Hanakawa T (2006) Neuroimaging of standing and walking: special 642
580 with DT performance, such as the IPL, PMC and emphasis on Parkinsonian gait. Parkinsonism Relat Disord 12: 643
581 cerebellum. In addition, we found pronounced FC S70–S75. 644

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

10 G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx

645 Haslinger B, Erhard P, Kämpfe N, Boecker H, Rummeny E, patients with Parkinson’s disease suffering from motor response 716
646 Schwaiger M, Conrad B, Ceballos-Baumann AO (2001) Event- fluctuations. Exp Brain Res 208:169–179. 717
647 related functional magnetic resonance imaging in Parkinson’s Polito C, Berti V, Ramat S, Vanzi E, De Cristofaro MT, Pellicanò G, 718
648 disease before and after levodopa. Brain 124:558–570. Mungai F, Marini P, Formiconi AR, Sorbi S, Pupi A (2012) 719
649 Hausdorff JM, Balash J, Giladi N (2003) Effects of cognitive challenge Interaction of caudate dopamine depletion and brain metabolic 720
650 on gait variability in patients with Parkinson’s disease. J Geriatr changes with cognitive dysfunction in early Parkinson’s disease. 721
651 Psychiatry Neurol 16:53–58. Neurobiol Aging 33. 206.e229-239. 722
652 Imamura K, Okayasu N, Nagatsu T (2012) Cerebral blood flow and Postuma RB, Dagher A (2006) Basal ganglia functional connectivity 723
653 freezing of gait in Parkinson’s disease. Acta Neurol Scand based on a meta-analysis of 126 positron emission tomography 724
654 126:210–218. and functional magnetic resonance imaging publications. Cereb 725
655 Jahanshahi M, Jones CRG, Zijlmans J, Katzenschlager R, Lee L, Cortex 16:1508–1521. 726
656 Quinn N, Frith CD, Lees AJ (2010) Dopaminergic modulation of Power JD, Mitra A, Laumann TO, Snyder AZ, Schlaggar BL, Petersen 727
657 striato-frontal connectivity during motor timing in Parkinson’s SE (2014) Methods to detect, characterize, and remove motion 728
658 disease. Brain 133:727–745. artifact in resting state fMRI. NeuroImage 84:320–341. 729
659 Kehagia AA, Barker RA, Robbins TW (2010) Neuropsychological and Rahman S, Griffin HJ, Quinn NP, Jahanshahi M (2008a) Quality of life 730
660 clinical heterogeneity of cognitive impairment and dementia in in Parkinson’s disease: the relative importance of the symptoms. 731
661 patients with Parkinson’s disease. Lancet Neurol 9:1200–1213. Mov Disord 23:1428–1434. 732
662 Kelly VE, Eusterbrock AJ, Shumway-Cook A (2012) A review of dual- Rahman S, Griffin HJ, Quinn NP, Jahanshahi M (2008b) The factors 733
663 task walking deficits in people with Parkinson’s disease: motor that induce or overcome freezing of gait in Parkinson’s disease. 734
664 and cognitive contributions, mechanisms, and clinical Behav Neurol 19:127–136. 735
665 implications. Parkinsons Dis 2012:918719. Rochester L, Hetherington V, Jones D, Nieuwboer A, Willems AM, 736
666 Kerr GK, Worringham CJ, Cole MH, Lacherez PF, Wood JM, Silburn Kwakkel G, Van Wegen E (2004) Attending to the task: 737
667 PA (2010) Predictors of future falls in Parkinson disease. interference effects of functional tasks on walking in Parkinson’s 738
668 Neurology 75:116–124. disease and the roles of cognition, depression, fatigue, and 739
669 Kish SJ, Shannak K, Hornykiewicz O (1988) Uneven pattern of balance. Arch Phys Med Rehabil 85:1578–1585. 740
670 dopamine loss in the striatum of patients with idiopathic Rochester L, Galna B, Lord S, Burn D (2014) The nature of dual-task 741
671 Parkinson’s disease, Pathophysiologic and clinical implications. interference during gait in incident Parkinson’s disease. 742
672 N Engl J Med 318:876–880. Neuroscience 265:83–94. 743
673 Lord S, Baker K, Nieuwboer A, Burn D, Rochester L (2011) Gait Ruthruff E, Pashler HE, Hazeltine E (2003) Dual-task interference 744
674 variability in Parkinson’s disease: an indicator of non- with equal task emphasis: graded capacity sharing or central 745
675 dopaminergic contributors to gait dysfunction? J Neurol postponement? Percept Psychophys 65:801–816. 746
676 258:566–572. Sabatini U, Boulanouar K, Fabre N, Martin F, Carel C, Colonnese C, 747
677 Maidan I, Bernad-Elazari H, Gazit E, Giladi N, Hausdorff JM, Bozzao L, Berry I, Montastruc JL, Chollet F, Rascol O (2000) 748
678 Mirelman A (2015) Changes in oxygenated hemoglobin link Cortical motor reorganization in akinetic patients with Parkinson’s 749
679 freezing of gait to frontal activation in patients with Parkinson disease: a functional MRI study. Brain 123(Pt 2):394–403. 750
680 disease: an fNIRS study of transient motor-cognitive failures. J Selemon LD, Goldman-Rakic PS (1985) Longitudinal topography and 751
681 Neurol 262:899–908. interdigitation of corticostriatal projections in the rhesus monkey. J 752
682 McIsaac TL, Lamberg EM, Muratori LM (2015) Building a framework Neurosci 5:776–794. 753
683 for a dual task taxonomy. Biomed Res Int 2015:591475. Shine JM, Matar E, Ward PB, Bolitho SJ, Pearson M, Naismith SL, 754
684 Meyer DE, Kieras DE (1997) A computational theory of executive Lewis SJG (2013a) Differential neural activation patterns in 755
685 cognitive processes and multiple-task performance: Part 1. Basic patients with Parkinson’s disease and freezing of gait in 756
686 mechanisms. Psychol Rev 104:3–65. response to concurrent cognitive and motor load. PLoS One 8. 757
687 Mirelman A, Maidan I, Bernad-Elazari H, Nieuwhof F, Reelick M, e52602–e52602. 758
688 Giladi N, Hausdorff JM (2014) Increased frontal brain activation Shine JM, Matar E, Ward PB, Frank MJ, Moustafa AA, Pearson M, 759
689 during walking while dual tasking: an fNIRS study in healthy Naismith SL, Lewis SJ (2013b) Freezing of gait in Parkinson’s 760
690 young adults. J Neuroeng Rehabil 11:85. disease is associated with functional decoupling between the 761
691 Nutt JG, Bloem BR, Giladi N, Hallett M, Horak FB, Nieuwboer A cognitive control network and the basal ganglia. Brain 762
692 (2011) Freezing of gait: moving forward on a mysterious clinical 136:3671–3681. 763
693 phenomenon. Lancet Neurol 10:734–744. Shine JM, Moustafa AA, Matar E, Frank MJ, Lewis SJ (2013c) The 764
694 Ouchi Y, Kanno T, Okada H, Yoshikawa E, Futatsubashi M, role of frontostriatal impairment in freezing of gait in Parkinson’s 765
695 Nobezawa S, Torizuka T, Tanaka K (2001) Changes in disease. Front Syst Neurosci 7:61. 766
696 dopamine availability in the nigrostriatal and mesocortical Shine JM, Matar E, Ward PB, Bolitho SJ, Gilat M, Pearson M, 767
697 dopaminergic systems by gait in Parkinson’s disease. Brain Naismith SL, Lewis SJ (2013d) Exploring the cortical and 768
698 124:784–792. subcortical functional magnetic resonance imaging changes 769
699 Pellecchia MT, Picillo M, Santangelo G, Longo K, Moccia M, Erro R, associated with freezing in Parkinson’s disease. Brain 770
700 Amboni M, Vitale C, Vicidomini C, Salvatore M, Barone P, 136:1204–1215. 771
701 Pappatà S (2015) Cognitive performances and DAT imaging in Shine JM, Handojoseno AM, Nguyen TN, Tran Y, Naismith SL, 772
702 early Parkinson’s disease with mild cognitive impairment: a Nguyen H, Lewis SJ (2014) Abnormal patterns of theta frequency 773
703 preliminary study. Acta Neurol Scand 131:275–281. oscillations during the temporal evolution of freezing of gait in 774
704 Peterson DS, Pickett KA, Duncan RP, Perlmutter JS, Earhart GM Parkinson’s disease. Clin Neurophysiol 125:569–576. 775
705 (2014) Brain activity during complex imagined gait tasks in Smulders K, Esselink RA, Weiss A, Kessels RP, Geurts AC, Bloem 776
706 Parkinson disease. Clin Neurophysiol 125:995–1005. BR (2012) Assessment of dual tasking has no clinical value for fall 777
707 Peterson DS, Fling BW, Mancini M, Cohen RG, Nutt JG, Horak FB prediction in Parkinson’s disease. J Neurol 259:1840–1847. 778
708 (2015) Dual-task interference and brain structural connectivity in Snijders AH, Haaxma CA, Hagen YJ, Munneke M, Bloem BR (2012) 779
709 people with Parkinson’s disease who freeze. J Neurol Neurosurg Freezer or non-freezer: clinical assessment of freezing of gait. 780
710 Psychiatry 86:786–792. Parkinsonism Relat Disord 18:149–154. 781
711 Pijnappels M, Bobbert MF, van Dieën JH (2001) Changes in walking Snijders AH, Leunissen I, Bakker M, Overeem S, Helmich RC, Bloem 782
712 pattern caused by the possibility of a tripping reaction. Gait BR, Toni I (2011) Gait-related cerebral alterations in patients with 783
713 Posture 14:11–18. Parkinson’s disease with freezing of gait. Brain 134:59–72. 784
714 Plotnik M, Dagan Y, Gurevich T, Giladi N, Hausdorff JM (2011) Spildooren J, Vercruysse S, Desloovere K, Vandenberghe W, 785
715 Effects of cognitive function on gait and dual tasking abilities in Kerckhofs E, Nieuwboer A (2010) Freezing of gait in 786

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056
 NSC 16813 No. of Pages 11
8 January 2016

G. Vervoort et al. / Neuroscience xxx (2016) xxx–xxx 11

787 Parkinson’s disease: the impact of dual-tasking and turning. Mov Wu T, Hallett M (2008) Neural correlates of dual task performance in 817
788 Disord 25:2563–2570. patients with Parkinson’s disease. J Neurol Neurosurg Psychiatry 818
789 Strouwen C, Molenaar EA, Münks L, Keus SH, Bloem BR, Rochester 79:760–766. 819
790 L, Nieuwboer A (2015a) Dual tasking in Parkinson’s disease: Wu T, Chan P, Hallett M (2008) Modifications of the interactions in the 820
791 should we train hazardous behavior? Expert Rev Neurother:1–9. motor networks when a movement becomes automatic. J Physiol 821
792 Strouwen C, Molenaar EA, Keus SH, Münks L, Bloem BR, Nieuwboer 586:4295–4304. 822
793 A (2015) Test-retest reliability of dual-task outcome measures in Wu T, Wang L, Hallett M, Chen Y, Li K, Chan P (2011) Effective 823
794 people with Parkinson’s disease, Phys Ther, in press connectivity of brain networks during self-initiated movement in 824
795 Tessitore A, Amboni M, Esposito F, Russo A, Picillo M, Marcuccio L, Parkinson’s disease. NeuroImage 55:204–215. 825
796 Pellecchia MT, Vitale C, Cirillo M, Tedeschi G, Barone P (2012) Wu T, Wang J, Wang C, Hallett M, Zang Y, Wu X, Chan P (2012) 826
797 Resting-state brain connectivity in patients with Parkinson’s disease Basal ganglia circuits changes in Parkinson’s disease patients. 827
798 and freezing of gait. Parkinsonism Relat Disord 18:781–787. Neurosci Lett 524:55–59. 828
799 Tessitore A, Giordano A, De Micco R, Russo A, Tedeschi G (2014) Wu T, Liu J, Hallett M, Zheng Z, Chan P (2013) Cerebellum and 829
800 Sensorimotor connectivity in Parkinson’s disease: the role of integration of neural networks in dual-task processing. 830
801 functional neuroimaging. Front Neurol 5:180. NeuroImage 65:466–475. 831
802 Tombu M, Jolicoeur P (2003) A central capacity sharing model of Wu T, Liu J, Zhang H, Hallett M, Zheng Z, Chan P (2014) Attention to 832
803 dual-task performance. J Exp Psychol Hum Percept Perform automatic movements in Parkinson’s disease: modified automatic 833
804 29:3–18. mode in the striatum. Cereb Cortex. 834
805 Vercruysse S, Spildooren J, Heremans E, Wenderoth N, Swinnen Wu T, Hallett M, Chan P (2015) Motor automaticity in Parkinson’s 835
806 SP, Vandenberghe W, Nieuwboer A (2013) The neural correlates disease. Neurobiol Dis 82:226–234. 836
807 of upper limb motor blocks in parkinson’s disease and their Yogev G, Giladi N, Peretz C, Springer S, Simon ES, Hausdorff JM 837
808 relation to freezing of gait. Cereb Cortex. (2005) Dual tasking, gait rhythmicity, and Parkinson’s disease: 838
809 Wenderoth N, Debaere F, Sunaert S, Swinnen SP (2005) The role of which aspects of gait are attention demanding? Eur J Neurosci 839
810 anterior cingulate cortex and precuneus in the coordination of 22:1248–1256. 840
811 motor behaviour. Eur J Neurosci 22:235–246. Yu H, Sternad D, Corcos DM, Vaillancourt DE (2007) Role of 841
812 Wickens CD (2008) Multiple resources and mental workload. Hum hyperactive cerebellum and motor cortex in Parkinson’s disease. 842
813 Factors 50:449–455. NeuroImage 35:222–233. 843
814 Wu T, Hallett M (2005) A functional MRI study of automatic
815 movements in patients with Parkinson’s disease. Brain
816 128:2250–2259.
844
845 (Accepted 30 December 2015)
846 (Available online xxxx)

Please cite this article in press as: Vervoort G et al. Dual-task-related neural connectivity changes in patients with Parkinson’ disease. Neuroscience
(2016), http://dx.doi.org/10.1016/j.neuroscience.2015.12.056