Probabilistic mapping of gait changes after STN-DBS for Parkinson’s disease

doi:10.21203/rs.3.rs-2353609/v1

Download PDF

Article

Probabilistic mapping of gait changes after STN-DBS for Parkinson’s disease

https://doi.org/10.21203/rs.3.rs-2353609/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Objective.

Gait disturbances causing impaired mobility are common in Parkinson’s disease after bilateral deep brain stimulation of the subthalamic nucleus. We describe subthalamic subregions where neurostimulation had a positive effect on gait or provoked gait disturbances.

Methods.

Sixty-eight patients were classified according to postoperative gait changes: (1) gait improvement, (2) no change, (3) de novo gait disturbances. We performed a segregation analysis for (1) and (3) by simulating volumes of tissue activated and comparing aggregated spatial data for the two groups and calculated probability maps to forecast gait performance and the parkinsonism control.

Results.

Twenty patients experienced complete remission of presurgical gait problems after stimulation. Nine patients showed de novo gait disturbances one year post-implantation. Active contacts were more ventrally located for de novo gait disturbances versus gait improvement. Strong correlations were found between clinical alterations in gait and the individual stimulation volume within the probabilistic outcome gait map (R² = 0.78; p = 0.01), whereby clinical improvement in parkinsonism correlated with individual stimulation volume within the corresponding probabilistic outcome map (R² = 0.39; p = 0.01). The probabilistic maps predict patients who experience long-term gait benefits based on their volume of tissue activated overlap, which was gait specific and showed no correlation with the global parkinsonism control heatmap.

Interpretation.

Probabilistic mapping showed high correlation for therapy outcomes, especially gait improvement. The concept of sweet- or badspots could not explain individual differences. The thin delineations between close substructures in the subthalamic nucleus correlated with individual gait changes after neurostimulation. Probabilistic mapping may direct future re-programming approaches for greater mobility in parkinsonian patients.

Deep brain stimulation

Freezing of gait

Heat map

Movement disorder surgery

Parkinson’s disease

Sweetspot

Volume of tissue activated

High-frequency deep brain stimulation of the subthalamic nucleus (STN-DBS) has become a mainstay treatment for motor fluctuations, dyskinesia, and intractable tremor in Parkinson’s disease (PD). STN-DBS improves motor symptoms per se and drug-induced motor complications through a reduced need for dopaminergic replacement therapy, thereby enabling improved activities of daily living and quality of life.^1,2 Nevertheless, an approximately 70% variability in motor outcomes has been reported within and between clinical trials, and gait disturbances are a frequent complaint after STN-DBS.^3–5 Gait disturbances are an important clinical problem because they limit mobility and cause falls,^6–10 particularly episodic freezing of gait (FOG), which typically occurs during gait initiation, passing obstacles or turning, and consists of an episodic inability to move forward despite the intention to walk.⁸

Studies have reported that DBS has non-significant or inconsistent benefits regarding gait problems.^3,11,12 Positive effects on axial symptom severity and FOG were reported two years after STN-DBS in the EARLYSTIM trial in the medication-OFF condition.¹³ In contrast, an infrequent but alarming clinical scenario can occur where patients develop de novo FOG in the early adjustment phase of STN stimulation, despite otherwise beneficial effects of neurostimulation.¹⁴ In these situations, the occurrence of gait disturbances seems more likely to be a complication of neurostimulation or due to reduction of dopaminergic medication, rather than a pure progression of the underlying disease.

Recent clinical studies have focused on troubleshooting attempts and biophysical aspects of adjusting the stimulation volume¹⁵ or pulse parameter settings, including interleaving or low-frequency stimulation.^16–19 Yet individualized clinical reprogramming is highly complex and time consuming, especially because the delayed onset of stimulation effects on axial symptoms can take several weeks, particularly if gait is involved.²⁰ With the advent of multiple programming options in DBS, the sole adjustment of parameters using a clinical trial-and-error approach has become nearly impossible.

To overcome the burden of clinically-guided programming, image and model informed approaches have been intensely debated.²¹ Recent publications have used probabilistic mapping of outcome data in large cohorts of DBS-treated patients to identify individually optimal stimulation sites and enable computer-assisted programming approaches.^22,23 We have focused on the effects of DBS on gait and de novo gait disturbances in a cohort of PD patients undergoing STN-DBS. Furthermore, we have tried to disentangle a distinct “optimal efficacy volume” with respect to mobility and gait, by aggregating individual stimulation and imaging data. Our aim was to correlate clinical outcomes with the exact anatomical location of stimulation and calculate a heat map of therapy effects.

Patients and clinical test battery

We retrospectively collected a dataset of 68 patients with idiopathic PD who underwent bilateral STN-DBS (implanted between 2010 and 2016 in Wuerzburg, Germany), which was stratified for motor improvement and gait alterations. Individual datasets were included if they contained: (i) a video-recorded clinical evaluation within 6 months before surgery and at least 12 months after surgery; (ii) pre- and postoperative neuroimaging allowing reconstruction of lead location. The motor examination was performed using the Movement Disorder Society Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) part III after overnight withdrawal of all dopaminergic medication (meds-OFF) preoperative and one- year post-operative. Follow-up routine in our center included at least visits after two weeks, three months and one-year post-surgery for re-assessment.

Patients were divided into two groups according to their change in the “gait” (3.10) and “freezing of gait” (3.11) items of the MDS-UPDRS part III after STN-DBS: 1) a “de novo” group, where gait items worsened by > 1 point one year after surgery (stim-ON/meds-OFF) compared to the preoperative meds-OFF condition (de novo gait disturbances were verified by at least three chart-documented, in- or outpatient presentations due to gait worsening in the first year after implantation and had to be present at the one year post surgery visit); 2) a gait improvement group, where gait items improved by > 1 point. All patients gave their written informed consent to the procedure and the study was approved by the local ethics committee (Schriftzeichen 248/20).

Surgical procedure

The surgical procedure was carried out as previously described.²⁴ On the day of surgery, after induction of anesthesia and additional local anesthesia, the Leksell stereotactic frame was placed (Elekta Instruments AB, Stockholm, Sweden), a contrast-enhanced, full-head computerized tomography (CT) scan was carried out, and the images were fused with preoperative, frameless magnetic resonance imaging (MRI), including the new planning. During an awake phase of 2–3 hours, intraoperative microelectrode recording (MER; FHC, Frederic Haer Company, Boston, USA) and neurological examinations were performed within a range of 5 mm above and 3 mm below the target point. Final lead positioning and choice of lead trajectory were based on anatomical, electrophysiological, and clinical testing results between the implanting neurosurgeon and senior neurologist performing the intraoperative testing. All patients received quadripolar macroelectrodes (model 3389, Medtronic Inc.) or an octapolar macroelectrode (model 2201-A, Boston Scientific Inc.) during one surgery for both hemispheres. The electrode position was verified by image fusion of a postsurgical helical cranial CT and preoperative MRI at least 4 weeks after the implantation to avoid brain shift due to pneumocephalus and acute cerebral spinal fluid loss.

Lead localization and therapy scoring

The electrode position was visualized using SureTune™ (Medtronic Inc.) as previously described²² by fusing individual preoperative MRI with postoperative CT scans. Lead localizations were identified according to previous studies.^25,26 The Yelnik atlas²⁷ was registered onto regions of interest on the STN by a rigid, semi-automatic registration algorithm on T2-weighted, fast spin echo MRI and/or susceptibility weighted imaging MRI sequences for each patient. The volume of tissue activated (VTA) was computed for each lead, based on the applied stimulation parameters at the one year post surgery visit using the Astrom model.²⁸ Images were linearly normalized into Montreal Neurological Institute space (MNI; ICBM 2009b NLIN asymmetric), based on the local atlas registration using a self-programmed MATLAB tool. Patients were grouped based on the clinical change in UPDRS III gait items. The VTAs of patients who experienced improvement in gait items 3.10 & 3.11 were aggregated after mirroring to the left STN. All voxels covering more than six individual VTAs were defined as the good area. The same procedure was carried out for the bad area. The individual centroid distances to those areas were evaluated for association with clinical changes in gait items. In addition, the VTA overlap of all subjects for the overall UPDRS-III improvement was calculated to visualize beneficial stimulation areas for motor improvement.

Heatmap

An essential step was to transform this information for each individual into a detailed local “heatmap”. The detailed process has been reported elsewhere.²² In the common anatomical space (MNI space), every voxel (0.25 x 0.25 x 0.25 mm) was covered by either none, a single, or multiple VTAs. In a further step, each voxel was associated with several individual changes in gait alterations. Subsequently, we analyzed the correlation between the VTA and the individual gait changes. Left and right VTAs were combined, reflecting the bilateral treatment of patients. We created a prediction model for scores of individual VTAs, based on their placement within the heatmap. In a voxel-wise analysis, we compared the distribution of gait scores of the entire population assigned to a single voxel in atlas space to all gait scores not included in that voxel. The resulting two outcome distributions were then compared using the t-test against a null hypothesis of equality. With the T-map for each voxel, we calculated a so-called heat map (or signed p-map) to evaluate and rate the patient-specific VTAs. In the signed p-map, the inverted p-values receive the sign of the t-test, such that the most significant values are near − 1 and 1. To validate the heatmap model, we calculated a leave-one-out cross-validation analysis (LOOCV). In brief, a prediction model is trained using n -1 patients and validated on the lef-tout patient, iterating n times for the full cohort. Finally, we tested the clinical relevance of our heatmap using the individual VTA-related score. Here, a leave-one-out design for the full cohort was used, and the prediction of gait improvement according to the model score was calculated. The same analysis was done with all 68 patients using the total UPDRS III improvement to calculate a global motor improvement heatmap. This was also tested for predictive value in a LOOCV analysis.

Statistical analysis

Data were analyzed using SPSS (version 26, IBM). For gender differences between groups, we used Pearson’s chi-squared test. Other differences between de-novo and remission groups were analyzed by means of Student’s t-test with alpha error adjustment according to Bonferroni for multiple comparisons. The level of statistical significance was set at a threshold of p < 0.05. The heatmap and the LOOCV were calculated with MATLAB Statistics and Machine Learning Toolbox (version 2017a, Mathworks). Continuous variables are presented as mean ± standard deviation.

Data Availability Statement

All supplementary data for VTA-overlap and Heatmap-calculation original data can be reviewed upon request.

Clinical characteristics

Sixty-eight patients with PD were analyzed (Table 1). The mean age was 60.7 ± 8.1 years, the Hoehn & Yahr score was 2.80, and disease duration was 14.1 years. MDS-UPDRS III before surgery was 46.7 ± 11.2 points; after one year, motor improvement following stimulation alone (meds-OFF) was 49.8%.

Twenty patients (13 males) with an average 44.8% motor improvement after stimulation alone (meds-OFF) were classified as having improvement of gait disturbances. Nine patients (six males) with an average 59.6% motor improvement after stimulation alone (meds-OFF) were classified as having de novo gait disturbances within one year after STN-DBS still present at the one-year follow-up assessment. All nine patients had at least three (range 3–5) chart-documented episodes of gait worsening after implantation, with a median onset 13 (range 3–38) weeks after implantation. Most of these patients (8/9) experienced FOG episodes with an increase in the UPDRS-3 item 11. The gait improvement group was significantly older than the de novo gait disturbances group (p = 0.02). However, there were no significant differences in the UPDRS III items pre- and post-surgery, with or without inclusion of items 3.10 and 3.11 (Fig. 1 & Table 1) ,levodopa equivalent dose (LEED) pre- and one year postoperative and stimulation parameters one year postoperative (Table 1). In addition, there were no significant differences in stimulation amplitude, pulse width, or frequency between groups one year after implantation. All patients received high frequency stimulation (> 100Hz).

Active contact locations and VTA

In a standard stereotactic reference space, the mean anterior commissure-posterior commissure-based coordinates of the individual active electrodes were 12.7 mm lateral, 1.1 mm posterior, and 3.1 mm inferior to the midcommissural point in the gait remission group and significantly more ventral at 12.1 mm lateral, 0.7 mm posterior, and 4.2 inferior in the de novo gait disturbances group (p = 0.016).

In a second step, we aggregated the stimulation parameters associated with each active contact by calculating the individual VTAs and transforming them into normative atlas space. By aggregating the distinct, common VTAs of the electrodes in the nine patients with de novo gait disturbances, the calculation of an overlap VTA of gait disturbances was possible. The same process was repeated for the 20 patients in the gait improvement group (Fig. 2). The center of gravity of the overlap VTAs for MNI coordinates was 12.51 mm, -13.27 mm, and − 8.25 mm in the de novo group, and 12.61 mm, -13.75 mm, and − 7.20 mm in the gait remission group. Only the difference for the dorsal/ventral location was significant (p = 0.017). The area stimulated most in patients with de novo gait disturbances after DBS was located more ventral compared to the group with remission of gait disturbances. Overlap of the de novo group with remission group was 0.625 (dice-coefficient according to Soerensen). In MNI space, a typical variety of lead locations in the STN was visible (Fig. 2), with a clustering of the most active electrodes within the dorsolateral aspect of the STN or at the lateral exiting white matter bundles. Interestingly, our calculated mean stimulated areas of both gait groups overlapped within the boundaries of beneficial stimulation areas for global motor improvement for all patients (Fig. 3A, B), and previously reported sweetspots (Fig. 3C, D) were similarly located.²⁹

Outcome prediction models

The distance from the center of gravity of the overlap VTAs to the individual active electrode in each group did not correlate with changes in MDS-UPDRS III items 3.10 and 3.11 (Fig. 2D). Therefore, calculation of a heatmap was performed. Strong association was found between the clinical alterations in gait and the individual stimulation volume within the probabilistic outcome heatmap of gait. The heatmap model was associated with 78% of the variance in gait alterations measured by item 3.10 and 3.11 after STN-DBS (R² = 0.78; p = 0.01). In a further analysis, we validated the predictive value of the VTA heatmap estimation calculated using LOOCV. Here, the model prediction of gait changes significantly correlated with the clinically observed outcome (R² = 0.38; p < 0.01) (Fig. 4). Finally, we found that our STN-DBS heatmap could be used to identify patients with a long-term gait benefit, based solely on their VTA overlap. Patients in our cohort were stratified based on the VTA overlap in three equal groups, each containing 33% of the counted values of the heatmap into high-, medium-, and low-risk cohorts. The PD patients in our high-risk cohort showed significantly better performance in gait than patients in our low-risk cohort (3.9 ± 1.8 points versus 0.6 ± 2.2 points on MDS-UPDRS III items 3.10 and 3.11; p = 0.002) and medium-risk cohort (3.9 ± 1.8 points versus 1.0 ± 3.8 points; p = 0.046). No differences were seen between patients in the low- and medium-risk cohorts (p = 0.78) (Fig. 4).

Interestingly, the overall MDS-UPDRS III improvement on the heatmap model of all subjects was associated with 43% of the variance in outcome (R² = 0.43; p = 0.01). Also, the LOOCV analysis showed that this heatmap was predictive for individual motor improvement (R² = 0.16; p = 0.01) (Fig. 5). Importantly, both heatmaps showed very low spatial correlation (r=-0.23, p = 0.23 Pearson correlation) and were outcome specific. This indicates that the global motor improvement heatmap was not predictive for gait performance (R² = 0.03; p = 0.31; Fig. 5D) and that global motor improvement was unable to be predicted using the gait performance heatmap (R² = 0.003; p = 0.68).

We have found evidence that STN-DBS can induce opposing effects on locomotion in PD, depending on the selected stimulation volume. While one-third of our cohort exhibited positive effects on gait, a minority (9/68; 13.2%) developed de novo gait disturbances. Individual stimulation field analysis demonstrated distinct anatomical clusters in these two groups, with significantly more ventral aspects of the STN being stimulated in patients with de novo gait disturbances. Moreover, probabilistic outcome mapping of our cohort explained 78% of the variance in gait alterations on a group level, whereas probabilistic outcome mapping of global parkinsonism control was unable to explain gait alterations. This supports a strong relationship between individual DBS outcomes and symptom-specific stimulation topographies within the subthalamic area. This finding constitutes an important development towards an image- and computer-assisted programming approach in STN-DBS.

Clinical changes in locomotion after STN-DBS

Gait disorders after STN-DBS are common and unpredictable clinical complaints in up to 20% of all patients, thereby diminishing the impact of the therapy on quality of life despite positive effects on other PD-related motor signs.³⁰ In addition to persistent or aggravated gait problems, a relevant proportion of gait disturbances occurs de novo and typically with a delay after surgery, which is reflected by a median time to onset of 13 weeks after surgery in our cohort. The clinical and pathophysiological interpretation of these DBS-associated gait impairments is complicated, and excessive reduction of dopaminergic therapy is required to distinguish between disease-related progression of axial motor symptoms or unmasking of a parkinsonian gait disorder and a stimulation-related, long-term adverse effect.¹⁴ Several authors have previously reported that STN stimulation could have deleterious effects on axial symptoms, with a delayed onset of postural instability and gait disorders.^31,32 Interestingly, we observed de novo gait problems in a group of patients with an overall excellent response of parkinsonian motor symptoms to STN-DBS, as reflected by an average 59.6% reduction of the total UPDRS III score. The delay between surgery and onset of de novo gait symptoms was too short to assume natural disease progression. According to our internal clinical pathway, an acute and/or chronic levodopa challenge ruled out insufficient dopaminergic stimulation as a cause of gait disturbance in all cases. Thus, the gait symptoms in our group were categorized as long-term, stimulation-related adverse effects, likely resulting from inappropriate stimulation of a gait-related network that could potentially overlap the global parkinsonian motor network targeted by STN-DBS.

This observation contrasted with the documented improvement of typical “off”-period parkinsonian gait impairment in another subgroup of our cohort (37%; 20/68 patients), which interestingly exhibited a less pronounced reduction of the total UPDRS motor score (46.3%). STN-DBS is known to improve levodopa-responsive aspects of an off-period parkinsonian gait disorder, which can be formally assessed and predicted by a preoperative levodopa challenge.¹³ Recent studies have described a positive effect of STN-DBS on gait patterns in advanced PD, with improvements in several biomechanically-assessed patterns of gait, even when the patients were only evaluated in the meds-ON condition.³³ A secondary analysis of the EARLYSTIM trial showed an improvement in gait patterns in the meds-OFF condition compared to best medical treatment within the first two years after STN-DBS.¹³

Optimal stimulation site and prediction of individual gait outcomes

Recent publications have demonstrated sweetspots associated with good global motor outcomes after DBS including FOG, with a center of mass at the dorsal interface between sensorimotor and associative subterritories of the STN.^23,34 In the vertical stereotactic plane, the overlap volume for beneficial effects on gait in our study was more dorsal compared to these published sweetspots for global motor outcomes and to the sweetspot volume for our cohort.^23,29 Of note, the subgroup of patients with de novo gait disturbances had an aggregated stimulation volume that was located even further ventral compared to the other two groups, which could be referred to as a “sour spot” for gait.

When we correlated the individual distances of the active contacts with these calculated “sweet” or “sour” spots for patients belonging to either the gait remission or de novo gait disturbance groups, we were unable to find any association (Fig. 2D). The superiority of voxel-wise statistics comparing each voxel against an average of all other outcomes in the dataset compared to older sweetspot models was recently discussed by Dembek et al.³⁵ The increased rate of variance explained by this heatmap technique was also reported for a model of the probabilistic mapping of antidystonic effects in pallidal neurostimulation.²² Using a similar approach, our heatmap explained 78% of the variance in gait alterations of patients, and could be used to predict patients who might experience a long-term gait benefit based solely on their VTA overlap. This finding is supported by the idea of fiber modulation by DBS, where a therapeutic effect on a certain tract can be achieved in a different location according to the 3D anatomical course of the fiber in the brain.³⁶ The assumption about the role of separate network modulations of gait-related fibers within and around STN warrants future investigation.

Pathophysiological Considerations

Our results offer an anatomic signature of stimulation-associated gait deterioration, observed in 13% of our patients. The stimulation volumes of patients experiencing gait disturbances were located more ventrally within the subthalamic area, compared to the more dorsal region covered by stimulation in patients with gait improvement. Interestingly, all nine patients experiencing de novo gait problems after surgery were otherwise considered excellent motor responders to DBS treatment. Hence, it is conceivable that gait deterioration in this group resulted from excessive stimulation and spillover into a gait-associated brain network, which should be spared by appropriate settings of stimulation. Fleury et al. assumed that high-frequency stimulation of the STN can selectively worsen on-drug akinesia and gait problems, while simultaneously improving tremor and rigidity. They concluded that this paradoxical effect of STN-DBS could be observed by stimulating the ventral contacts of electrodes otherwise correctly placed in the STN. This modulates the pallidothalamocortical pathway to reduce tremor and rigidity, but reinforces akinesia and gait disturbances by recruiting pallidal efferent pathways to the pedunculopontine nucleus (PPN) within the ansa lenticularis.³⁷ In line, our data implies that the negative effects of STN-DBS on gait correlate with a more ventral stimulation site within the STN (Fig. 5). The possible relevance of PPN pathway modulation in gait alterations is supported by positron emission tomography investigations in PD patients after STN-DBS.³⁸ Imaged gait in these patients was associated with regional cerebral blood flow changes in the brainstem locomotor center of the PPN/mesencephalic locomotor region.

An alternative hypothesis emphasizes the impact of STN-DBS on cortical networks responsible for the thalamocortical integration of somatosensory information during gait, via differential antidromic modulation of the two main anatomical constituents of the pallidothalamic tract: the ventrally-located ansa lenticularis and the more dorsal fasciculus lenticularis. Indeed, ventral and dorsal globus pallidus interna (GPi) segments have displayed differences in cortical connectivity profiles, with the ventral GPi being more connected to the primary somatosensory cortex and posterior motor regions, and the dorsal GPi being more connected to anterior motor and premotor regions.³⁹ Corresponding to these findings in the GPi, visualization of both tracts with the Petersen-Connectom for the STN displayed a clear ventro-dorsal arrangement; the dorsal STN was more connected to the primary somatosensory cortex and posterior motor regions, while the ventral STN was more connected to the anterior motor and premotor regions.⁴⁰

Limitations

There are certain limitations to our study. The retrospective utilization of items 3.10 and 3.11 of the UPDRS III score for the groups did not allow us to specify different clinical types of gait alteration, such as the occurrence of episodic gait problems such as FOG or permanent balance problems like ataxia. For a prospective cohort, the scoring of therapy should consider specific kinematic motor measurements similar to recent studies of gait parameters,³³ to correlate motor states with specific neurostimulation settings. Reported beneficial effects of low frequency stimulation where not addressed since all patients received high frequency stimulation (> 100Hz). ^41,42

The predictive value of our model was not validated on an external cohort. However, to assess the validity we conducted a LOOCV analysis – a widely accepted method^22,43 that showed robust estimation of the therapy effect. Recent publications have revealed the concept of VTA-related characteristic connectivity profiles for clinical effects.⁴⁴ Future research efforts could uncover networks of stimulation-induced gait abnormalities using this method of structural and functional connectivity profiles of sweet- and badspots, and to increase the predictive value of stimulation parameters and locations.

Delayed-onset gait disturbances after STN-DBS constitute a major clinical challenge in postoperative management. Our study suggests that a ventrocaudal gradient of stimulation sites within the STN has beneficial or detrimental effects on gait, which are not adequately represented by the classical sweetspot volumes derived from changes in global motor scores. We recognized that overlapping areas within the ventral, sensory-motor part of the STN were more likely to induce stimulation-related gait alterations, while otherwise providing good control of appendicular motor symptoms of PD. We assume that these divergent effects are the result of differential neuromodulation of symptom- or task-specific motor networks by STN-DBS, which should be further characterized by a connectomic approach. However, the regional stimulation topography reflected by our heatmap model already allows individual therapy outcomes to be predicted with high confidence, and may in future assist troubleshooting for mobility problems in the postoperative management of STN-DBS in PD.

Acknowledgements

The study was sponsored by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 424778381 – TRR 295, and the Deutsche Gesellschaft für Parkinson und Bewegungsstörungen -Nachwuchsförderung to MR. The draft manuscript was reviewed by a native English speaker (Deborah Nock, Medical WriteAway, Norwich, UK).

Author contribution statement

Conception and design of the study: RN, AG, HE, JR, JV, MR

Acquisition and Analysis of data: RN, AG, HE, JR, PF, PN, PC, FL, MR

Drafting a significant portion of the manuscript: RN, HE, JR, II, CM, JV, MR

Potential conflicts of interest

RN received grant support & honorary speaking from Boston Scientific not related to this work.

AG, HE, JR and PF report no conflicts of interest

II received grant support & honorary speaking from Medtronic & Newronika Spa not related to this work

PC reports grant support and honoraria payment from Boston Scientific & BrainLab not related to this work

CM revieved honorary speaking from Boston Scientific not related to this work

JV received grant support & honorary speaking from Boston Scientific, Medtronic & Abbot not related to this work

MR received grant support & honorary speaking from Medtronic & Boston Scientific not related to this work

Deuschl, G. et al. A randomized trial of deep-brain stimulation for Parkinson's disease. N Engl J Med 355, 896-908, doi:10.1056/NEJMoa060281 (2006).
Schuepbach, W. M. et al. Neurostimulation for Parkinson's disease with early motor complications. N Engl J Med 368, 610-622, doi:10.1056/NEJMoa1205158 (2013).
St George, R. J. et al. The effect of deep brain stimulation randomized by site on balance in Parkinson's disease. Mov Disord 29, 949-953, doi:10.1002/mds.25831 (2014).
Kleiner-Fisman, G. et al. Subthalamic nucleus deep brain stimulation: summary and meta-analysis of outcomes. Mov Disord 21 Suppl 14, S290-304, doi:10.1002/mds.20962 (2006).
Okun, M. S. et al. Management of referred deep brain stimulation failures: a retrospective analysis from 2 movement disorders centers. Arch Neurol 62, 1250-1255, doi:10.1001/archneur.62.8.noc40425 (2005).
Pozzi, N. G. et al. Freezing of gait in Parkinson's disease reflects a sudden derangement of locomotor network dynamics. Brain 142, 2037-2050, doi:10.1093/brain/awz141 (2019).
Kerr, G. K. et al. Predictors of future falls in Parkinson disease. Neurology 75, 116-124, doi:10.1212/WNL.0b013e3181e7b688 (2010).
Nutt, J. G. et al. Freezing of gait: moving forward on a mysterious clinical phenomenon. Lancet Neurol 10, 734-744, doi:10.1016/S1474-4422(11)70143-0 (2011).
Palmisano, C. et al. Gait Initiation in Parkinson's Disease: Impact of Dopamine Depletion and Initial Stance Condition. Front Bioeng Biotechnol 8, 137, doi:10.3389/fbioe.2020.00137 (2020).
Palmisano, C. et al. Sit-to-walk performance in Parkinson's disease: A comparison between faller and non-faller patients. Clin Biomech (Bristol, Avon) 63, 140-146, doi:10.1016/j.clinbiomech.2019.03.002 (2019).
Potter-Nerger, M. & Volkmann, J. Deep brain stimulation for gait and postural symptoms in Parkinson's disease. Mov Disord 28, 1609-1615, doi:10.1002/mds.25677 (2013).
Collomb-Clerc, A. & Welter, M. L. Effects of deep brain stimulation on balance and gait in patients with Parkinson's disease: A systematic neurophysiological review. Neurophysiol Clin 45, 371-388, doi:10.1016/j.neucli.2015.07.001 (2015).
Barbe, M. T. et al. Deep Brain Stimulation for Freezing of Gait in Parkinson's Disease With Early Motor Complications. Mov Disord 35, 82-90, doi:10.1002/mds.27892 (2020).
Reich, M. M. et al. The Pirouette Test to Evaluate Asymmetry in Parkinsonian Gait Freezing. Mov Disord Clin Pract 1, 136-138, doi:10.1002/mdc3.12018 (2014).
Wodarg, F. et al. Stimulation site within the MRI-defined STN predicts postoperative motor outcome. Mov Disord 27, 874-879, doi:10.1002/mds.25006 (2012).
Moreau, C. et al. STN-DBS frequency effects on freezing of gait in advanced Parkinson disease. Neurology 71, 80-84, doi:10.1212/01.wnl.0000303972.16279.46 (2008).
Fasano, A. et al. Modulation of gait coordination by subthalamic stimulation improves freezing of gait. Mov Disord 26, 844-851, doi:10.1002/mds.23583 (2011).
Weiss, D. et al. Nigral stimulation for resistant axial motor impairment in Parkinson's disease? A randomized controlled trial. Brain 136, 2098-2108, doi:10.1093/brain/awt122 (2013).
Meoni, S. et al. Asymmetric STN DBS for FOG in Parkinson's disease: A pilot trial. Parkinsonism Relat Disord 63, 94-99, doi:10.1016/j.parkreldis.2019.02.032 (2019).
Reich, M. M. et al. Progressive gait ataxia following deep brain stimulation for essential tremor: adverse effect or lack of efficacy? Brain 139, 2948-2956, doi:10.1093/brain/aww223 (2016).
Merola, A. et al. Current Directions in Deep Brain Stimulation for Parkinson's Disease-Directing Current to Maximize Clinical Benefit. Neurol Ther 9, 25-41, doi:10.1007/s40120-020-00181-9 (2020).
Reich, M. M. et al. Probabilistic mapping of the antidystonic effect of pallidal neurostimulation: a multicentre imaging study. Brain 142, 1386-1398, doi:10.1093/brain/awz046 (2019).
Dembek, T. A. et al. Probabilistic sweet spots predict motor outcome for deep brain stimulation in Parkinson disease. Ann Neurol 86, 527-538, doi:10.1002/ana.25567 (2019).
Nickl, R. C. et al. Rescuing Suboptimal Outcomes of Subthalamic Deep Brain Stimulation in Parkinson Disease by Surgical Lead Revision. Neurosurgery 85, E314-E321, doi:10.1093/neuros/nyz018 (2019).
Hemm, S. et al. Contact position analysis of deep brain stimulation electrodes on post-operative CT images. Acta Neurochir (Wien) 151, 823-829; discussion 829, doi:10.1007/s00701-009-0393-3 (2009).
Pollo, C. et al. Magnetic resonance artifact induced by the electrode Activa 3389: an in vitro and in vivo study. Acta Neurochir (Wien) 146, 161-164, doi:10.1007/s00701-003-0181-4 (2004).
Yelnik, J. et al. A three-dimensional, histological and deformable atlas of the human basal ganglia. I. Atlas construction based on immunohistochemical and MRI data. Neuroimage 34, 618-638, doi:10.1016/j.neuroimage.2006.09.026 (2007).
Astrom, M., Diczfalusy, E., Martens, H. & Wardell, K. Relationship between Neural Activation and Electric Field Distribution during Deep Brain Stimulation. Ieee T Bio-Med Eng 62, 664-672, doi:10.1109/Tbme.2014.2363494 (2015).
Todt, I. et al. The Contribution of Subthalamic Nucleus Deep Brain Stimulation to the Improvement in Motor Functions and Quality of Life. Mov Disord 37, 291-301, doi:10.1002/mds.28952 (2022).
Castrioto, A., Volkmann, J. & Krack, P. Postoperative management of deep brain stimulation in Parkinson's disease. Handb Clin Neurol 116, 129-146, doi:10.1016/B978-0-444-53497-2.00011-5 (2013).
Ferraye, M. U. et al. Effects of subthalamic nucleus stimulation and levodopa on freezing of gait in Parkinson disease. Neurology 70, 1431-1437, doi:10.1212/01.wnl.0000310416.90757.85 (2008).
Guehl, D. et al. Evolution of postural stability after subthalamic nucleus stimulation in Parkinson's disease: a combined clinical and posturometric study. Exp Brain Res 170, 206-215, doi:10.1007/s00221-005-0202-z (2006).
Navratilova, D. et al. Deep Brain Stimulation Effects on Gait Pattern in Advanced Parkinson's Disease Patients. Front Neurosci 14, 814, doi:10.3389/fnins.2020.00814 (2020).
Karachi, C. et al. Clinical and anatomical predictors for freezing of gait and falls after subthalamic deep brain stimulation in Parkinson's disease patients. Parkinsonism Relat Disord 62, 91-97, doi:10.1016/j.parkreldis.2019.01.021 (2019).
Dembek, T. A. et al. Sweetspot Mapping in Deep Brain Stimulation: Strengths and Limitations of Current Approaches. Neuromodulation, doi:10.1111/ner.13356 (2021).
Elias, G. J. B. et al. Probabilistic Mapping of Deep Brain Stimulation: Insights from 15 Years of Therapy. Ann Neurol, doi:10.1002/ana.25975 (2020).
Fleury, V. et al. Subthalamic stimulation may inhibit the beneficial effects of levodopa on akinesia and gait. Mov Disord 31, 1389-1397, doi:10.1002/mds.26545 (2016).
Weiss, P. H. et al. Subthalamic nucleus stimulation improves Parkinsonian gait via brainstem locomotor centers. Mov Disord 30, 1121-1125, doi:10.1002/mds.26229 (2015).
Hill, K. K. et al. Cerebral blood flow responses to dorsal and ventral STN DBS correlate with gait and balance responses in Parkinson's disease. Exp Neurol 241, 105-112, doi:10.1016/j.expneurol.2012.12.003 (2013).
Petersen, M. V. et al. Holographic Reconstruction of Axonal Pathways in the Human Brain. Neuron 104, 1056-1064 e1053, doi:10.1016/j.neuron.2019.09.030 (2019).
Ramdhani, R. A., Patel, A., Swope, D. & Kopell, B. H. Early Use of 60 Hz Frequency Subthalamic Stimulation in Parkinson's Disease: A Case Series and Review. Neuromodulation 18, 664-669, doi:10.1111/ner.12288 (2015).
Xie, T. et al. Long-term effect of low frequency stimulation of STN on dysphagia, freezing of gait and other motor symptoms in PD. J Neurol Neurosurg Psychiatry 89, 989-994, doi:10.1136/jnnp-2018-318060 (2018).
Arlot S, C. A. A survey of cross-validation procedures for model selection. Statist Surv 4: 40–79 (2010).
Horn, A. et al. Connectivity Predicts deep brain stimulation outcome in Parkinson disease. Ann Neurol82, 67-78, doi:10.1002/ana.24974 (2017).

Table 1. Patient characteristics.

	De novo gait disturbances (n=9)	Gait remission (n=20)	p-value
Study age (years)	57.8 (6.6)	64.9 (7.2)	0.02*
Disease duration (years)	15.2 (5.9)	17.7 (4.5)	0.24
Hoehn & Yahr score, pre-DBS	3.2 (0.7)	2.9 (0.6)	0.29
UPDRS III
Preop, meds-OFF	49.1 (13.6)	50.0 (11.8)	0.91
Meds-OFF, stim-ON	27.1 (8.6)	20.2 (8.1)	0.051
Preop, minus items 3.10 & 3.11	47.1 (13.1)	45.5 (11.3)	0.73
Postop, stim-ON, minus items 3.10 & 3.11	23.3 (8.7)	19.1 (8.0)	0.21
LEED (mg) pre-OP	1337 (794)	1289 (437)	0.83
LEED (mg) post-OP	616 (384)	506 (246)	0.36
Stimulation parameters
Frequency (1/s)	158 (24)	148 (28)	0.19
Amplitude (mA)	3,07 (0,85)	3,29 (1,06)	0.44
Pulse width (µs)	57,2 (8,3)	59 (4,4)	0.29
Clinical/reported gait problem	3.9 (0.8)	—	—
First reported problem in weeks after DBS	16.6 (12.6)	—	—

* p-values <0.05 considered statistically significant (paired t-test). Values shown are mean (standard deviation). de novo = new onset or aggravation of gait disturbances. DBS, deep brain stimulation; UPDRS, Unified Parkinson’s Disease Rating Scale. LEED, levodopa equivalent dose. Stimulation parameters one year post-operative.

There is a conflict of interest RN received grant support & honorary speaking from Boston Scientific not related to this work. AG, HE, JR and PF report no conflicts of interest II received grant support & honorary speaking from Medtronic & Newronika Spa not related to this work PC reports grant support and honoraria payment from Boston Scientific & BrainLab not related to this work CM revieved honorary speaking from Boston Scientific not related to this work JV received grant support & honorary speaking from Boston Scientific, Medtronic & Abbot not related to this work MR received grant support & honorary speaking from Medtronic & Boston Scientific not related to this work

Download PDF

Version 1

posted

You are reading this latest preprint version

Probabilistic mapping of gait changes after STN-DBS for Parkinson’s disease

Status:

Version 1

Abstract

Objective.

Methods.

Results.

Interpretation.

Figures

Introduction

Methods

Patients and clinical test battery

Surgical procedure

Lead localization and therapy scoring

Heatmap

Statistical analysis

Data Availability Statement

Results

Clinical characteristics

Active contact locations and VTA

Outcome prediction models

Discussion

Clinical changes in locomotion after STN-DBS

Optimal stimulation site and prediction of individual gait outcomes

Pathophysiological Considerations

Limitations

Conclusion

Declarations

References

Table 1

Additional Declarations

Status:

Version 1