Microstructure predicts non-motor outcomes following Deep Brain Stimulation in Parkinson’s disease

1

Deep brain stimulation of the subthalamic nucleus (STN-DBS) is an effective 2 treatment for motor and non-motor symptoms in advanced Parkinson’s disease (PD). 3 However, considerable interindividual variability of outcomes exists. Neuroimaging based 4 biomarkers, such as neurite orientation dispersion and density imaging (NODDI), a 5 biophysical model based MRI-technique, have been proposed to predict clinical outcomes and 6 therefore inform preoperative patient counselling. 7

To detect microstructural properties of brain areas associated with short-term non-8 motor outcomes following STN-DBS in PD. 9

In this prospective open-label study, 37 PD patients underwent diffusion MRI and 10 comprehensive clinical assessments at preoperative baseline and 6-month follow-up. Neurite 11 density index (NDI), orientation dispersion index (ODI), and fractional anisotropy (FA) were 12 derived. Whole brain voxel-wise analysis assessed associations between microstructural 13 metrics and non-motor outcomes corrected for multiple comparisons using a permutation-14 based approach. 15

Intact microstructure within specific areas including right insular cortex, right 16 putamen, right cingulum, and bilateral corticospinal tract were associated with greater 17 postoperative improvement of non-motor symptom burden. Furthermore, microstructural 18 properties of distinct brain regions were associated with postoperative changes in sleep, 19 attention/memory, and urinary symptoms. 20

Microstructural properties of distinct brain areas predict non-motor outcomes in 21 DBS for PD. Therefore, diffusion MRI can support preoperative patient counselling and 22 treatment selection by identifying patients with above- or below-average non-motor 23 responses. 24 25 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 3

1 Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is an established therapy for 2 advanced Parkinson’s disease (PD), improving motor- and non-motor symptoms. 1-3 Despite 3 significant therapeutic benefits at the group level, there is large variability in outcomes at the 4 individual subject-level, with some patients even experiencing persistent symptoms. 4 To 5 predict postoperative outcomes and thereby improve preoperative patient counselling, the use 6 of neuroimaging-based markers has been proposed. 4 In this regard, quantitative MRI 7 techniques such as neurite orientation dispersion and density imaging (NODDI) have gained 8 increasing attention in recent years to track disease progression and treatment response. 5 9 NODDI is a multi-compartmental diffusion-weighted MRI technique which facilitates the 10 assessment of specific microstructural properties directly related to neurite morphology. 6 The 11 model provides two voxelwise metrics of neurite morphology: the neurite density index 12 (NDI), describing the density of axons and dendrites within a voxel, and the neurite 13 orientation dispersion index (ODI), characterizing the variability of neurite orientations, i.e. 14 how parallel they are. Importantly, the relationship between NODDI metrics and underlying 15 tissue properties has recently been confirmed histologically. 7 Despite its sensitivity and 16 specificity, NODDI protocols have a clinically feasible data acquisition time, making them an 17 important tool for use in clinical research. 6 Previous studies employing NODDI in PD have 18 shown that the model is capable of differentiating PD patients from healthy controls 11 and 19 patients with atypical Parkinsonism. 12 Furthermore, NODDI characterized disease related 20 pathology such as retrograde degeneration of the nigrostriatal pathway,13 and its metrics were 21 associated with bimanual motor control, 14 disease severity, and duration. 6,11 Combining 22 NODDI with conventional DTI metrics in a multi-parametric analysis therefore provides 23 complementary information on microstructural properties and may serve as an imaging-based 24 marker to facilitate treatment prediction and inform patient counselling. Thus, we sought to 25 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 4 demonstrate that a multi-parametric quantitative MRI approach can detect microstructural 1 properties of brain areas that predict non-motor outcomes following STN-DBS in PD. 2 Specifically, we aimed to identify regions whose microstructural metrics, i.e. NDI, ODI, and 3 fractional anisotropy (FA), were associated with (1) changes in overall non-motor symptom 4 burden and (2) changes in non-motor symptom domains, showing significant improvements 5 after STN-DBS. The results of the present study should help to guide preoperative patient 6 counselling by identifying microstructure that predicts above- or below-average non-motor 7 response to STN-DBS. 8

9 The study was approved by the local ethics committee (study-number: 155/17) and carried out 10 in accordance with the Declaration of Helsinki. 11 Participants. 12 Thirty-seven PD patients (9 female, mean age 58.8 ± 7.3 years) were enrolled in this 13 prospective, observational, ongoing study upon written informed consent (for demographics 14 cf. Table 1). Inclusion criteria comprised indication for DBS lead surgery because of 15 advanced PD according to modern criteria. 8 Patients were excluded if they had pathological 16 MR imaging, a concomitant neurological or psychiatric disease, or impaired visual or auditory 17 function. 18 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 5 Clinical Assessment 1 Patients were assessed preoperatively and six months after DBS lead surgery, with medication 2 at both times and DBS switched on postoperatively. All subjects underwent a 3 neuropsychological assessment including the Non-Motor Symptom Scale (NMSS), the 4 Parkinson’s Disease Questionnaire (PDQ)-8 and the Scales for Outcomes in PD - Motor 5 Function (SCOPA-M) using standardized case report forms. Detailed information on the 6 clinical assessment is provided in the supplementary material. 7 MRI Data Acquisition and Processing 8 PD patients were scanned at baseline on a 3-Tesla Trio scanner (Siemens, Erlangen, 9 Germany) at the Core Unit Brain Imaging of the University of Marburg. The acquisition 10 protocol is reported in the supplementary material. All images were investigated to be free of 11 motion or ghosting and high frequency and/or wrap-around artefacts at the time of image 12 acquisition. 13 Image Processing 14 Image analysis was performed within the FreeSurfer image analysis suite 7.1.1 15 (http://surfer.nmr.mgh.harvard.edu) as reported previously by our group. 9 The processing of 16 T1-weighted scans included skull stripping, automated Talairach transformation, cortical and 17 subcortical segmentation, intensity normalisation, tessellation of the grey/white matter 18 boundary, automated topology correction, and surface deformation following intensity 19 gradients.10 20 DTI data were processed using FMRIB Software Library (FSL) 6.0.5.2 21 (https://fsl.fmrib.ox.ac.uk/fsl). To correct for eddy-current distortions and involuntary 22 movements, raw DTI volumes were linearly registered and resampled to the first b0 volume.11 23 Subsequently, the diffusion tens or for each voxel was fit to the data using linear regression 24 and FA was derived from the diffusion tensor. 12 Furthermore, NODDI-DTI,13 a modification 25 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 6 of NODDI,6/i1 was used to obtain NDI and ODI from the DTI data. Visual inspection of the 1 b0-images confirmed that no changes beyond those in the tissue structure contributed to the 2 observed effects. To perform regional analyses, the first b0 image of each scan was linearly 3 registered to the structural T1-weighted image using a boundary based method, yielding an 4 affine matrix.14 Subsequently, T1-derived segmentations and brain masks were transformed to 5 the diffusion space via the inverse of the affine matrix. 6 Statistical analysis 7 Statistical analysis of clinical outcomes was performed in MATLAB (The MathWorks, Inc., 8 R2018a). Changes between baseline and follow-up were analysed using the Wilcoxon signed-9 rank or t-test when parametric test criteria were met. The type-I error was controlled using the 10 Benjamini-Hochberg method and effect sizes were calculated according to Cohen. 11 Relationships between change scores of clinical data and NMSS total score (NMSS-T) were 12 explored using Spearman correlations as reported previously.15,16 13 Statistical voxelwise analysis of image data was performed using a generalized linear model. 14 First, FA-maps were co-registered to the MNI152 space using linear and nonlinear 15 transformation.17 Masked FA-, NDI-, and ODI-maps were subsequently registered to the 16 MNI152 space using the transformation from the previous step. Only voxels of brain tissue 17 existing in every subject were included in the analysis. Significant associations between 18 metrics of microstructure and change of non-motor symptoms were carried out for the whole 19 brain as described previously.18 Here, percentage differences between baseline and follow-up 20 values were calculated for NMSS-T and NMSS domains with significant postoperative 21 improvement ((2) sleep/fatigue, (5) attention/memory, and (7) urinary). A permutation-based 22 approach based on the Analysis of Functional NeuroImages (AFNI) null-z simulator was used 23 to corrected for multiple comparisons employing 12,000 simulations under the null 24 hypothesis.19 Clusters were formed using a threshold of p<.01 and a clusterwise p-value was 25 calculated. Results were accepted as significant with clusterwise p<.05. 26 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 7

1 Clinical outcomes 2 Longitudinal changes of clinical outcomes are reported in Table 1. At the 6-month follow-up, 3 we observed improvements of NMSS-T (z=-2.95, p=.007, r=-.34), PDQ-8 SI (t(36)=4.22, 4 p<.001, Cohen’s d: .67), and SCOPA-M total score (z=-4.01, p<.001, r=-.47), as well as 5 reductions of LEDD (t(36)=7.63, p<.001, Cohen’s d: 1.2) and LEDD-DA (t(36)=4.98, 6 p<.001, Cohen’s d: .7). Analysis of NMSS domains revealed beneficial effects of STN DBS 7 on sleep/fatigue (domain 2; z=-3.79, p<.001, r=-.44), attention/memory (domain 5; z=-2.49, 8 p=.022, r=-.29), and urinary symptoms (domain 7; z=-2.39, p=.026, r=-.28). Analysis of 9 SCOPA-M domains showed improvements in motor examination (z=-2.87, p=.008, r=-.33), 10 activities of daily living (z=-3.27, p=.003, r=-.38), and motor complications (z=-3.76, p<.001, 11 r=-.44) at the 6-month follow-up. Results of the correlation analysis are reported in the 12 supplementary materials (Table e-1). 13 Interaction between Fractional Anisotropy and postoperative non-motor 14 symptom change 15 Higher FA-values in the right insular cortex were associated with greater postoperative 16 NMSS-T reduction (positive cluster P1, cluster wise p-value (CWP): .032). Furthermore, 17 lower FA values were found in clusters including the bilateral cingulum and the left inferior 18 longitudinal fasciculus (ILF), which were related to greater postoperative NMSS-T reduction 19 (negative clusters N1-3, CWP: <.001-.014). Regional mean fractional anisotropy values 20 associated with postoperative change in non-motor symptom burden are detailed in Table e-2 21 and Figures e-1 and e-2. 22 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 8 Interaction between NODDI-parameters and postoperative non-motor symptom 1 change 2 Whole brain analysis of NODDI parameters showed that both, ODI and NDI, were associated 3 with changes in postoperative NMSS-T. Higher ODI-values in the right putamen, the right 4 cingulate cortex, and the left forceps major were related to a higher postoperative NMSS-T 5 reduction (P1-4, CWP: <.001-.034, Figure 1, Table e-3). Furthermore, lower ODI-values in 6 regions of both corticospinal tracts as well as left occipital fusiform gyrus were related to a 7 higher postoperative NMSS-T reduction (N1-4, CWP: <.001-.047, Figure 2, Table e-3). 8 Lower NDI-values in the left postcentral gyrus, left cingulum, and right forceps minor were 9 associated with a higher postoperative NMSS-T reduction (N1-4, CWP: <.001-.047, Figure e-10 4, Table e-4). No positive associations between NDI-values and postoperative NMSS-T 11 change were detected. 12 Interaction between microstructure and postoperative change in Non-Motor 13 Symptoms Scale domains 14 Regional mean values of microstructural metrics associated with significant postoperative 15 changes in non-motor symptoms scale domains are detailed in Table e-5 (sleep/fatigue), Table 16 e-6 (attention/memory), and Table e-7 (urinary symptoms). Higher values of microstructural 17 metrics within vast regions of bilateral corticospinal tract (CST) were related to a higher 18 symptom reduction in the sleep and fatigue domain (FA: P1-2, CWP: .001-.005; ODI: P1-4, 19 CWP: <.001; NDI: P1, CWP: .004; Table e-5). With regard to attention and memory, higher 20 FA-values in bilateral cingulum, left insular cortex, and left anterior thalamic radiation 21 (AThR) were associated with higher postoperative symptom reduction (P1-4, CWP: <.001-22 .031, Table e-6). Furthermore, higher ODI-values in left parahippocampal gyrus and within 23 the right frontal pole were related to positive postoperative outcomes in attention and memory 24 (P1-2, CWP: <.001-.049), whereas lower ODI-values in bilateral AThR and left cingulum 25 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 9 were associated with higher symptom reduction (N3, N5, N8-10, CWP: <.001-.039). Also, 1 vast areas of bilateral CST and putamen that had lower ODI- and NDI-values were associated 2 with higher symptom reduction in the attention/memory domain (ODI: N1, N2, N4 and N7-8, 3 CWP: <.001-.011; NDI: N1-6 CWP: <.001-.023). Concerning urinary symptoms, higher ODI- 4 and NDI-values in left putamen, pallidum, and AThR as well as left cerebellar lobule V and 5 VI were related to higher postoperative symptom reduction in the urinary domain (ODI: P1-3, 6 CWP: <.001-.031; NDI: P1-2, P4, CWP: <.001-.002, Table e-7). This association was also 7 present between NDI-values and right pallidum and putamen (P3, CWP: .002). Furthermore, 8 higher FA-values in right cingulate gyrus and left superior longitudinal fasciculus (SLF) were 9 associated with higher postoperative reduction of urinary symptoms (P1-2, CWP: <.001-.01). 10

11 In the present study, NODDI-DTI, a novel method for analysing DWI data, was applied to 12 preoperative imaging to investigate cerebral microstructure associated with non-motor 13 symptom changes following neurostimulation in PD. There are two key findings. First, we 14 demonstrate that intact microstructure within specific areas including right insular cortex, 15 right putamen, right cingulum, and bilateral CST were associated with higher reduction of 16 postoperative non-motor symptom burden. Second, we delineate the structures and their 17 microstructural properties which are associated with postoperative improvements in specific 18 non-motor domains. 19 Whole brain analysis of NODDI parameters identified an association between higher ODI in 20 right putamen as well as right cingulum and a higher reduction of postoperative non-motor 21 symptom burden. As ODI is high in gray matter, 6 this finding supports the notion that intact 22 sprawling of dendritic processes in these areas is important for beneficial postoperative non-23 motor outcomes. In PD, the pathological deposition of alpha-synuclein in intraneuronal Lewy 24 inclusions is accompanied by a degeneration of neurons and severe morphological changes of 25 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 10 dendrites.20,21 These tissue changes cannot be seen in conventional MRI, whereas NODDI is 1 sensitive to neurite morphology. 6 Indeed, previous work demonstrated reduced putaminal 2 ODI-values in patients with PD compared to healthy controls. 21 This finding was interpreted 3 as decreased dendrite length and loss of spines of striatal medium spiny neurons, which are 4 the primary target of dopaminergic nigrostriatal projections. 21 Considering the close 5 topographical relationship between STN and putamen as well as insular cortex and the 6 integration of STN in basal ganglia-thalamo-cortical loops with its motor, associative, and 7 limbic projections, it is important to examine the assumed mechanisms of action of DBS. 4,22 8 Besides effects on the micro- and mesoscale, the high-frequency pulses of electrical current 9 emitted by DBS electrodes affect interregional networks on the macroscale. 4 Here, 10 modulation of networks has been able to predict postoperative outcomes across several motor 11 and non-motor symptoms. 4 Furthermore, previous work has shown that compromised 12 putaminal microstructure in PD patients can be associated with higher non-motor symptom 13 burden independent of motor symptoms. 23 Therefore, it can be hypothesised that the positive 14 association between ODI and postoperative non-motor outcomes in the present study 15 represents the dependency of DBS on intact tissue structure to exert its network effects. 16 Besides associations with gray matter areas, ODI showed a negative association with 17 postoperative non-motor symptom burden in bilateral CST. In healthy white matter tissue, 18 ODI is usually low, as fibers are highly coherent to another whereby high values of ODI 19 represent axonal disorganisation and degeneration. 6 Therefore, the observed negative 20 association might reflect the dependence of DBS on intact tissue structure in CST, as low 21 ODI, i.e. intact white matter microstructure, was associated with beneficial postoperative 22 outcomes. Previous DTI-studies have repeatedly demonstrated microstructural alterations of 23 CST in PD. In particular, increased FA has consistently been reported and hypothesised to 24 demonstrate a compensatory mechanism, reflecting axonal sprouting secondary to a reduced 25 input from striatum and thalamus. 24 When, however, microstructure in CST deteriorates over 26 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 11 the course of the disease, an association with motor dysfunction was demonstrated. 25 1 Associations between altered CST-microstructure and non-motor symptoms, on the other 2 hand, have received little attention. Employing a connectometry analysis in 85 patients, 3 Ashraf-Ganjouei and colleagues could demonstrate that lower axonal density in CST was 4 associated with a higher burden of gastrointestinal symptoms. 26 Furthermore, lower FA-5 values in right CST were observed in PD patients with depression compared to non-depressed 6 patients.27 Extending these findings, the results of the present study demonstrate that not only 7 non-motor symptoms but also beneficial non-motor outcomes following STN-DBS depend on 8 intact CST microstructure. Considering the role of the CST as major effector of motor control, 9 motor outcomes and the postoperative reduction in dopaminergic medication could contribute 10 to the beneficial non-motor effects in the present study. Importantly, however, postoperative 11 improvements in non-motor symptom burden were not related to improvements in LEDD, 12 LEDD-DA, and SCOPA-motor examination scores, which is in accordance with the 13 literature.28,29 Therefore, the association between intact CST microstructure and beneficial 14 postoperative non-motor outcomes seems to be independent of the motor effects of DBS and 15 suggests that CST microstructure is inherently relevant for non-motor symptoms in PD. 16 Microstructure is associated with beneficial sleep outcomes 17 Sleep disturbances affect the majority of PD patients and result in poor quality of life. 30 18 Associated disorders encompass both, disturbances of sleep-wake transition, as well as 19 parasomnias.30 Although the exact neural mechanisms remain to be established, a disrupted 20 interaction of neuronal circuits and different neurotransmitter systems has been suggested to 21 underlie the sleep disturbances in PD. 31,32 STN-DBS has been suggested to improve sleep by 22 alleviating motor symptoms and directly altering sleep physiology resulting in increased total 23 sleep time, sleep efficiency, and quality of sleep as well as reduced wakefulness after sleep 24 onset and insomnia.30 In the present study, STN-DBS improved symptoms of the sleep/fatigue 25 domain and beneficial outcomes were associated with higher FA-, ODI-, and NDI-values in 26 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 12 vast regions of the CST. On the other hand, lower ODI- and NDI-values in these areas were 1 associated with detrimental or below average response to STN-DBS. As most of the clusters 2 were overlapping and within regions of high fiber crossing and dispersion, the selective 3 degeneration of crossing fibers might underlie the observed relationship. The results, 4 therefore, support the hypothesis that intact microstructure of the CST and its crossing fibers 5 is important for beneficial postoperative changes in the sleep and fatigue domain. 6 Additionally, high ODI-values of cortical structures including right superior and middle 7 temporal gyrus and right parietal operculum cortex were associated with beneficial 8 postoperative outcomes. Previous studies of healthy controls using simultaneous recordings of 9 electroencephalography and functional MRI during sleep reported an association of activity 10 increases within these areas with sleep spindles during early stages of non-rapid eye 11 movement (NREM) sleep.33 In PD, spindle density and amplitude seems to be reduced during 12 NREM and modulated by dopaminergic therapy. 30 Integrating these findings with results of 13 the present study one could speculate that intact microstructure, i.e. dendritic arborisation, in 14 these areas is important for STN-DBS to modulate sleep physiology. 15 Microstructure is associated with beneficial attention and memory outcomes 16 Higher FA (left hemisphere) and lower ODI in bilateral AThR were associated with beneficial 17 postoperative outcomes in the attention and memory domain. These results suggest, that 18 degenerative changes in axonal structure underlying FA alterations in AThR are attributable 19 to changes in axonal fanning and dispersion. AThR connects the frontal lobe, the dorsolateral 20 prefrontal cortex (DLPFC) in particular, with the anterior and midline nuclei of the 21 thalamus.22 As these nuclei are integrated in functional loops with the cingulum and the 22 pallidum, the nuclei and their fiber connections are associated with the limbic system and 23 thought to be involved in executive functions and planning of complex behaviour. 22,34 The 24 DLPFC is involved in various higher-level cognitive functions including attention, working 25 memory, and executive control. 35,36 Considering the functional association of AThR with 26 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 13 STN, as well as its role in connecting the structures named above, it seems reasonable that 1 intact microstructure in AThR is important for the conveyance of beneficial effects on 2 memory function and attention. 3 Higher FA and lower ODI in overlapping clusters in bilateral cingulum as well as associated 4 structures were related to increases in postoperative attention and memory function, indicating 5 that axonal fanning and dispersion underlies alterations in FA in the cingulum. The cingulum 6 is a group of nerve fibers connecting the hippocampus, prefrontal, parietal, and anterior 7 cingulate cortex. 37 This allows for the integration of information from these structures and 8 explains the involvement of the cingulum in attention, memory, and emotion regulation. 37 9 Previous studies in PD have shown that compromised microstructure of the cingulum was 10 related to reduced scores in cognitive assessments, impaired visuospatial memory, and 11 dementia.38 Taken together, the association between microstructural properties and beneficial 12 attention and memory outcomes in bilateral AThR and cingulum might represent the 13 dependency of DBS on intact tissue structure to exert its network effects. 14 Microstructure is associated with beneficial outcomes of urinary symptoms 15 Deficient perception of multimodal sensory information is a characteristic of PD leading to 16 debilitating non-motor symptoms.39 Sensory deficiencies in PD have been described in both, 17 somatosensory pathways associated with proprioception as well as visceral pathways involved 18 in monitoring of urinary bladder filling. 40 STN-DBS in PD was shown to improve the 19 perception of urinary bladder filling, resulting in a delayed desire to void and increased 20 bladder capacity.39 Improved urinary function was attributed to a beneficial influence of STN-21 DBS on a gain of afferent bladder information due to an increase or decrease of activation of 22 primary sensory areas. 39 In particular, previous studies showed an activation of anterior 23 cingulate gyrus (ACC) and left lateral frontal cortex during monitoring and controlling the 24 storage phase of the urinary cycle and these structures are thought to be involved in the urge 25 to void, withholding urine, and the onset of micturition. 39 In the present study, intact 26 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 14 microstructure, i.e. high FA, in right ACC was associated with postoperative improvements in 1 urinary symptoms suggesting that sound microstructure in ACC is necessary for STN-DBS to 2 have beneficial effects on controlling the storage phase of the urinary cycle. Furthermore, 3 high NDI and ODI, i.e. high dispersion without a decrease in axonal density, in left AThR 4 was associated with positive outcomes of urinary symptoms. AThR connects the frontal 5 cortex with the thalamus and the pallidum and previous studies suggested that modulation of 6 left frontal cortex by STN-DBS is important for urge control. 39 Therefore, intact 7 microstructure in left AThR might be relevant for STN-DBS to modulate left frontal cortex 8 during urge control. 39 Further important structures implicated in processing afferent urinary 9 bladder information are posterior thalamus, which is activated during bladder filling and 10 micturition, and ventrolateral as well as reticular thalamus, which receive input from the 11 striatum to modulate the flow of visceral information between posterior thalamus and the 12 cortex. Previous studies speculated that STN-DBS may recondition the interaction between 13 pallidal output and the modulatory effect of the thalamus, resulting in improved gating of 14 sensory information.39 One might speculate, that the association of intact microstructure in 15 left pallidum and putamen with better outcomes in the urinary domain represents the 16 importance of these structures for the effect of STN-DBS on gating sensory information. 17

18 Three main limitations of our study have to be addressed. First, despite histopathological 19 validation of the NODDI model and its frequent use in PD, no studies validating the model in 20 post-mortem brain tissue of PD patients exist. Second, the underlying assumptions of the 21 NODDI model may represent an oversimplification and might therefore result in reduced 22 specificity.6 Third, the resolution of the DTI scan is limited to 2.0x2.0x2.0mm, which might 23 be too coarse for valid assessments of small fiber bundles. 24 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 15

1 In conclusion, we describe a spatially distinct profile of microstructural alterations associated 2 with beneficial non-motor outcomes following neurostimulation in PD, including right insular 3 cortex, putamen, cingulum, and bilateral corticospinal tract. Furthermore, we delineate the 4 structures and their microstructural properties which are associated with postoperative 5 improvements in specific non-motor domains. Therefore, we suggest that diffusion MRI can 6 support preoperative patient counselling by identifying patients with above- or below-average 7 non-motor responses. 8

9 The authors would like to thank the participants for their active engagement in this study. 10 Data and Code Availability 11 The data that support the findings of this study are available on request from the 12 corresponding author (PAL). The data are not publicly available due to privacy or ethical 13 restrictions. All tools used for the analysis of MRI data are based on FreeSurfer Version 7.1 14 (http://surfer.nmr.mgh.harvard.edu/) and FSL 6.0.5.2 (http://www.fmrib.ox.ac.uk/fsl) 15 packages, which are freely available. Scripts for automation were written in tcshell and parts 16 of the statistics were written in Python using the packages numpy, pandas, seaborn, 17 matplotlib, nibabel and scipy, which are also freely available. Python program code for the 18 analysis of NODDI-DTI is available from https://github.com/dicemt/DTI-NODDI. 19 Contributorship 20 PAL: study concept and design, data acquisition, data analysis, drafting of the manuscript 21 MBo: data acquisition, surgical intervention, critical revision of the manuscript 22 HSD: study design, critical revision of the manuscript 23 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 16 SS: data acquisition, critical revision of the manuscript 1 SK: critical revision of the manuscript 2 CN: data acquisition, surgical intervention, critical revision of the manuscript 3 LT: study design, critical revision of the manuscript 4 DJP: study concept and design, data acquisition, data analysis, drafting of the manuscript 5 MBe: study concept and design, data acquisition, data analysis, drafting of the manuscript 6 Financial disclosure/Conflicts of Interest 7 PAL was supported by the SUCCESS-Program of the Philipps-University of Marburg and the 8 ‘Stiftung zur Förderung junger Neurowissenschaftler’. MBo is a scientific consultant for 9 Brainlab. HSD was funded by the EU Joint Programme – Neurodegenerative Disease 10 Research (JPND), the Prof. Klaus Thiemann Foundation in the German Society of Neurology, 11 the Felgenhauer Foundation, the KoelnFortune program of the Medical Faculty of the 12 University of Cologne and has received honoraria by Everpharma, Kyowa Kirin, Bial, Oruen, 13 and Stadapharm. SS reports no financial disclosures. SK reports no financial disclosures. CN 14 is a scientific consultant for Brainlab. LT received payments as a consultant for Medtronic 15 Inc. and Boston Scientific and received honoraria as a speaker on symposia sponsored by 16 Bial, Zambon Pharma, UCB Schwarz Pharma, Desitin Pharma, Medtronic, Boston Scientific, 17 and Abbott. The institution of LT, not LT personally, received funding by the German 18 Research Foundation, the German Ministry of Education and Research, and Deutsche 19 Parkinson Vereinigung. DJP has received honoraria for speaking at symposia sponsored by 20 Boston Scientific Corp, Medtronic, AbbVie Inc, Zambon and Esteve Pharmaceuticals GmbH. 21 He has received honoraria as a consultant for Boston Scientific Corp and Bayer, and he has 22 received a grant from Boston Scientific Corp for a project entitled "Sensor-based optimisation 23 of Deep Brain Stimulation settings in Parkinson's disease" (COMPARE-DBS). The institution 24 of DJP, not DJP personally, has received funding from the German Research Foundation, the 25 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 17 German Ministry of Education and Research, the International Parkinson Foundation, the 1 Horizon 2020 programme of the EU Commission and the Pohl Foundation in Marburg. 2 Finally, DJP has received travel grants to attend congresses from Esteve Pharmaceuticals 3 GmbH and Boston Scientific Corp. MBe reports no financial disclosures. 4

5 1 . Jo st ST, Sau erbie r A, Vi s s er - V a ndewa ll e V , et al. A pro spec tiv e, c ont r ol le d s t u dy of non-mo tor 6 e ffe ct s o f s ubth alamic stimula tion in P a rk inson' s dise a s e : re s ult s at th e 3 6-mon th fol l ow-u p . 7 Jou rna l o f ne urol o gy, ne uro surge r y , an d psyc hiatry. 20 20;91(7 ):687 - 6 94. 8 2 . Jo st ST, Vi ss er -Va ndewa ll e V, Rizo s A, e t a l. Non -mot or pred icto rs o f 36 -month qu ali t y of li fe 9 a fte r subt hala mic stimula tio n in P a r k in s o n dis e a s e . NP J Parki ns on's dis ea se. 202 1; 7(1):48 . 10 3 . S auerbi er A, B a chon P , Ambro s io L, e t a l. The N e w Sa t i sfac ti on with Life a nd Trea t ment S cale 11 ( SL T S- 7 ) i n P a t ien t s wi t h P a r k i ns o n 's D is e a s e. J P a r k i ns ons D is . 2021. 12 4 . Hol lunder B , Raja mani N, S i ddiqi SH, e t al. Toward pe rson aliz ed me dic ine in conne c tomic 13 de ep bra in stimula ti on. Prog Neuro biol . 2022;2 10:10221 1. 14 5 . Ka miya K, Hori M, Aoki S. N O D DI in clini c al r e sea rc h. J ourn al o f n e uro s c ienc e me th od s . 15 2 020;346:1 08908. 16 6 . Z hang H , S chne ider T, Wh ee le r-King sho tt C A , Al exa nder DC . N O D DI: Prac tic al in vivo neurite 17 orie n t a t i o n dis per s io n and de n s i ty imagin g of the hu ma n brain . N eu roI m age . 18 2 012;61(4) :10 00-1016. 19 7 . Gru ssu F , Sc hneid er T, Tur C , et a l. Ne urit e disp er s ion : a new marke r o f mul tiple scl e r o si s 20 spina l co rd pa th ology? An n Cli n Tran sl N eurol. 20 17;4 (9) :663-67 9 . 21 8 . P o s tuma R B, Be rg D, Adler CH , e t a l. The new d efini tion a nd di ag no s tic crit eria o f P arkin son' s 22 di s e a s e. Th e L anc et N e ur o lo gy. 2016;15( 6):5 46-548 . 23 9 . L oehrer PA, Webe r I, O ehrn CR , e t al. Mi c rostruct ural a l ter ation s p r e dict imp ai re d bima nual 24 c ontrol in P a r k in son' s dise a s e . Brai n c om munic at ion s. 2022 ;4 ( 3 ) : fc ac 137. 25 1 0. F ischl B. F reeS ur fer . Neur oIm ag e . 2012;6 2(2):774 -781. 26 1 1. Jen kin son M, Sm i t h S. A g lobal o p t i mi s a t i o n method for r obu st a f f i ne r egis tr a t i on of brain 27 i mages . Medi cal I mag e An al ysi s . 2 001;5: 143 -156. 28 1 2. P ierp aoli C , Ba s s er PJ . T oward a quan tit a tive ass e s s me nt o f di ffu s i o n ani sot r o py. Magn e t i c 29 R es onan c e i n M e d ic i n e. 1996;36 :893 -906 . 30 13 . DTI- N ODDI . I m pleme n tati o n o f di ff u sion ten sor im ag e ba sed ne uri t e orie nta tio n d is pe rsio n 31 an d de ns i t y im ag i ng ( D TI -NO D D I) wr i t t en in P y t h on. A va i l a b l e f ro m 32 h t t p s: / / git hub .c om /dice mt /DTI -N ODD I [ co m pu t er pr o gr am] . h t tp s : / /g i th u b . c o m / d i c e m t / D T I -33 NODDI20 20 . 34 1 4. Greve D N, F i s c hl B . Ac cura te a nd r obu st brai n imag e ali gnmen t usi ng b oundary -b as ed 35 reg istra tion . Neur oIm a ge . 200 9;48:63 -72 . 36 1 5. Jo st ST, Konit s i o t i A, Lo e hr e r P A, et al. No n-mot or e ff ec ts o f dee p b rain stimul a tio n in 37 P arkin son' s dis ea se mo tor s ub ty pe s . Park inso ni sm Rel at Di s ord. 202 3:105318 . 38 1 6. Jo st ST, St r o bel L , Rizo s A, e t al. G ende r g ap in d eep br ain stimula t i on f or P arkin so n’s di se a se. 39 NPJ Park in son 's di s e a s e . 2022;8 (1 ):4 7. 40 1 7. Ande rs son JLR, Je nk ins on M, Smi t h S . No n-line a r reg i stra tion, a ka spa tia l no rmali s ation . - 41 F MRIB t echni cal re port T R07JA2 fr o m www fmriboxa cuk/ana ly si s / techr ep. 2007. 42 1 8. Be lke M, U ng e r M, Ha ttem er K, e t al. Di ff usion Ten so r I maging ( D T I) in idi opath ic REM s l e ep 43 be havi our di s or der (iRB D) . Kli nische Neu roph ysi ologi e . 2010;4 1. 44 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 18 1 9. Nic hols TE, Hol me s AP . N onp a r a me tr i c p ermuta tion te st s for func tion al n eur oima gin g: A 1 prime r with ex ample s . Hum an bra in m ap pi ng . 2002 ;15:1 -25. 2 2 0. Hal liday G M, Lev ere nz J B, S chne id er JS , A dle r CH . T he neu robiol ogic al ba si s o f co g nitive 3 i mpairment in Parkin son' s di sea s e. Mo ve ment dis order s : o f ficia l journ al of t h e M ove me nt 4 Dis order So ciety . 2014;2 9(5 ):634 -650. 5 2 1. Ka maga t a K, Hat ano T , Okuz u mi A, et al. N e ur i t e ori ent atio n disp er sion and den s i t y imag ing 6 i n the sub s ta ntia nigr a i n idiopa t hi c Pa r k i ns o n di s ea se . Europ ea n r adiol og y. 201 6; 26(8):2567 -7 2 577. 8 22 . Bähr M , Frot s c he r M. N eurologisch -t opi s che D i agno st ik Anat omie - Funk tion - Kl in i k. 201 4. 9 2 3. L enfeld t N, Ha ns so n W, Lar s s on A, N y ber g L , Birgande r R, For s g re n L. D i f fu sion t en sor 10 i maging and c orr e la tion s to Parkin son rat ing s c ale s. Jo urnal o f Ne ur o l ogy. 11 2 013;260(11 ):2823 -2830 . 12 2 4. Atki n s on -Cle ment C, Pi nto S, Eu se bio A , Coul on O. Di f fu s i on t en sor imagin g in Pa rk inson' s 13 di s e a s e: R eview and meta -a naly s i s . Ne uroIma g e: Cli n ical. 2017;16 :98 -110. 14 2 5. Z han W, Ka ng GA, Gla s s GA , et a l . Regi on al a lt e r a t i on s of bra in mic r o st ructur e in P a r k ins on's 15 di s e a s e u s ing dif fu sion ten sor imagin g. M o v eme n t D is o r d e r s . 20 12;27(1 ):90 -97 . 16 2 6. Ash ra f- G a njou ei A, M a jd A, Jav inani A, Aa r a bi MH . Auto nomic dys function a nd whi t e mat te r 17 mi crostruc t u ral chan ge s in drug -naï ve pa t i ent s w ith Parkin so n's di sea s e. Pe erJ. 2018;6:e 5539. 18 2 7. Ans ari M, A dib Mor a di S, G ha z i S herb af F , H e daya t ni a A, A arabi M H . Co mp ari s on of s truc tural 19 c onnec tivity i n Pa rkins on's di se a se w ith depre s s iv e symptom s ver su s non -de p r e ssed : a 20 di ffu s i on M RI conn e ctom etry st u dy. In ter nati on a l p sycho geria trics. 2019 ;31 (1):5 -1 2. 21 2 8. P etry - S chme lzer J N, Krau s e M, D embek TA, et a l . Non -mo tor ou tcome s dep e nd on loc ation 22 of neu ro s timula tion in Pa r k in son' s dise a se . Brai n : a jour nal o f n e urolo gy . 2019;1 4 2(11):3592 -23 3 604. 24 2 9. S auerbi er A, L o ehre r P, Jo st ST, e t al. Pre di ctor s o f sh ort -term i mpul sive a nd com p uls i ve 25 be havi our a fte r subt hala mic stimula t i on in Pa r k in s o n di se a se. Jour nal o f n eurol og y, 26 n euro surge r y , an d p s yc hiatry. 2 021;92(1 2):1 313-1318 . 27 3 0. Z ahed H, Z uz ua r r eg ui JR P, Gilron R, D eni son T , Starr PA, L it t l e S . T he Neu rophy siology of 28 S leep in P arkin s on's Di sea s e. M ove me nt dis order s : o ffici al jo urnal of t he Mo ve me nt Di sord er 29 So c i et y . 202 1;36(7):1526 -1 542. 30 3 1. Qama r MA, S auerbi er A , P oli ti s M , C arr H, Loehrer P, C haudhu ri KR. Pre s yn aptic 31 do paminergi c t erminal imagi ng and non - motor s y mpt o m s a sse ssmen t of Pa r k in so n's dise a s e : 32 e videnc e f or dopa min ergic ba si s ? NP J Pa rkinso n's di se as e. 201 7;3:5 . 33 3 2. Jo st ST, Ray Ch audhuri K, A shka n K, et a l. Subthalamic Stimul a tio n Imp rove s Qual i ty o f Sl e ep 34 i n Parkin son Di sea se : A 36 -Mont h Co n t ro lle d Study. J Parkin s on s Di s . 2021;11 :3 23- 335. 35 3 3. S chab us M , Da ng - Vu TT, A l bouy G, e t a l. Hemody namic cerebr al corr elat e s of sl ee p spindl e s 36 du r in g human non -rapid ey e mov ement sle ep. Proc eedi ng s o f the Na tion al Ac ade my of 37 Sc i en c e s . 20 07;104(32 ):1 3164-1316 9 . 38 3 4. Net ter s h e im FS , Loe hrer PA, We be r I, e t al. Dop amine s ub sti tut i on al t er s e ff ectiv e 39 c onnec tivity o f co rtical pr ef r on ta l, pr em otor, and mo t o r regi on s d ur i n g com plex bimanua l 40 fin ger move men t s i n Pa rkins on's di se a se. NeuroI ma g e. 201 9;190 :118-1 32. 41 3 5. L oehrer PA, Ne tte rshei m FS, Ju ng F, et al. Age ing c hange s e ffe c tive connec tivi ty of motor 42 ne t w ork s during bi manu al f i nge r coo r di n ation . Neur oIm age . 2016; 143:325 -34 2. 43 3 6. L oehrer PA, Ne tte rshei m FS, Oehrn CR, e t al. Incr ea se d pre fron tal t op- dow n contr ol in older 44 a dult s pr e d ict s moto r perform an ce a nd a ge -group a s s oc i ation . Ne uroI m ag e. 45 2 021;240:1 18383. 46 3 7. Wu Y , Sun D , W a ng Y, Wan g Y, Ou S . Seg menta tion o f the C ingulum Bundle i n the Human 47 Brai n: A New P e rspec tiv e Ba sed on D S I T rac t og r a phy a nd Fibe r D i s sectio n S tudy. Fr o nt 48 Neuro an at . 2016;10 :84. 49 3 8. Ch en B, F an G G , Liu H, Wan g S. Change s i n anatomic al a nd functi ona l con n ectivi t y of 50 P arkin son' s dis ea se p ati ent s ac cording t o c ogni t iv e st atu s. E ur o p ean Jour nal o f Ra diol o gy. 51 2 015;84(7) :13 18-1324. 52 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 19 3 9. Herzog J, Wei s s PH, Assmu s A, et al. I mprov ed se ns or y gating of u rinary bl add er a f fer ent s in 1 P arkin son' s dis ea se follow ing subth alam i c stimula tion. Br ai n : a jo urn al o f ne uro lo gy. 2 2 008;131(P t 1 ) :1 32 -145. 3 4 0. S aka kibara R, Ha ttori T , Uchiy ama T, Yama nis hi T. Vide our o dynamic and s phi nc ter motor unit 4 po t e n t i al an a lys es in Parki n s on' s di sea se a nd multipl e s y s tem a tr oph y . Jo urnal o f neurol ogy , 5 n euro surge r y , an d p s yc hiatry. 2 001;71(5) :60 0-606. 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 20 Tables 1 Table 1: Baseline characteristics and outcomes at baseline and 6-month follow-up 2 n M SD Age [y] 37 58.8 7.3 Disease duration [y] 37 8.9 4.3 Sex (female/male) [%] 37 9/28 [24.3/75.7] Baseline 6-MFU Baseline vs. 6-MFU n M SD n M SD p effect size NMSS total score 37 71.9 39.1 37 48.4 29.1 .007 -.34 Cardiovascular 37 1.6 2.7 37 1.1 2.5 .225 -.15 Sleep/fatigue 37 16.9 11.5 37 7.9 7.6 <.001 -.44 Mood/ apathy 37 11.6 13.3 37 6.0 7.6 .109 -.21 Perceptual problems/ hallucinations 37 .9 2.5 37 1.1 3.0 .865 -.03 Attention/ memory 37 7.0 6.5 37 5.0 5.9 .022 -.29 Gastrointestinal 37 5.6 5.0 37 4.4 4.9 .223 -.16 Urinary 37 13.2 10.4 37 9.1 10.2 .026 -.28 Sexual function 37 3.8 5.0 37 3.0 4.3 .296 -.13 Miscellaneous 37 11.2 8.9 37 10.7 7.8 .865 -.02 PDQ-8 SI 37 32.3 14.8 37 22.6 14.1 <.001 .67 SCOPA-M total score 37 19.8 6.2 37 12.9 6.4 <.001 -.47 SCOPA-M-motor examination 37 9.4 4.2 37 6.7 4.1 .008 -.33 SCOPA-M-activities of daily living 37 6.6 2.7 37 4.1 3.1 .003 -.38 SCOPA-M-motor complications 37 3.8 2.3 37 2.1 2.4 <.001 -.44 LEDD [mg] 37 962.3 393.2 37 537.6 269.0 <.001 1.2 LEDD DA [mg] 37 261.8 133.6 37 170.6 123.8 <.001 .7 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 21 Table 1: Demographic characteristics and outcome parameters at baseline and 6-months 1 follow-up. Reported p-values are corrected for multiple comparisons using Benjamini-2 Hochberg’s method. Bold font highlights significant results, p<.05. 3 Abbreviations: 6-MFU = 6-month follow-up; LEDD = Levodopa equivalent daily dose; 4 LEDD-DA: LEDD of Dopamine Agonists; NMSS = Non-Motor Symptom Scale; PDQ-8 SI = 5 8-item Parkinson’s Disease Questionnaire summary index; SCOPA = Scales for Outcomes in 6 Parkinson’s disease; 7 8 9 10 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 22 Figures 1 2 Figure 1. Clusters with a positive association between PD patients’ ODI-values and 3 postoperative change in NMSS-T (yellow), Domain 2 (sleep/fatigue, red), Domain 5 4 (attention/memory, green), and Domain 7 (urinary, blue), as revealed by the whole brain 5 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 23 analysis. P-Values were corrected for multiple comparisons using a permutation-based 1 approach. 2 3 4 5 Figure 2. Clusters with a negative association between PD patients’ ODI-values and 6 postoperative change in NMSS-T (yellow), Domain 2 (sleep/fatigue, red), Domain 5 7 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint 24 (attention/memory, green), and Domain 7 (urinary, blue), as revealed by the whole brain 1 analysis. P-Values were corrected for multiple comparisons using a permutation-based 2 approach. 3 4 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 25, 2023. ; https://doi.org/10.1101/2023.04.25.23289088doi: medRxiv preprint