Dystonia is a heterogeneous movement disorder characterized by repetitive movements and/or postures that are caused by sustained or intermittent muscle contractions.1 There are both known and idiopathic causes for this disorder, with classification determined by clinical phenotype and underlying etiology (e.g., inherited, secondary to a structural nervous system abnormality, etc.).2 The classification of childhood movement disorders is further complicated by the frequent coexistence of multiple impairments in the same patient.3 These overlapping features also contribute to difficulties in treatment.
Pharmacotherapy is the mainstay of treatment in children with generalized dystonia, but often provides inconsistent clinical benefit and is associated with significant side effects.4 In instances of refractory disease, there are neurosurgical options such as ablation or deep brain stimulation (DBS).5 The internal globus pallidus (GPi) was proposed as a potential surgical target for dystonia based on the marked reduction in dystonic dyskinesias following pallidal ablation in patients with Parkinson’s disease. In a study including both children and adults, surgical ablation was shown to be significantly less effective in patients with acquired dystonia than with idiopathic or genetic dystonia.6 In addition, ablation is associated with permanent adverse events, especially when performed bilaterally. Although DBS response in acquired dystonia is more variable, therapeutic benefit has been demonstrated even in adult patients with dystonic cerebral palsy.7 In part due to this lack of surgical ablation efficacy, as well as the greater flexibility and reversibility of DBS, GPi-DBS is now the gold standard treatment for medically refractory dystonia. Following DBS surgery, the delivered electrical dose requires individual titration to maximize clinical benefit while minimizing side effects. Our limited understanding of the mechanism of action of DBS enforces an empirical approach to device programing, which can be a challenging and lengthy process. A further challenge is that pediatric programming relies on use of the adult dystonia programming algorithm.8 Because therapeutic benefit in dystonia may take months to become apparent, multiple postoperative DBS programming visits are often required to achieve optimal results.
Methods now exist to accurately localize DBS electrodes within individual patients and coregister them into an average brain template, permitting group-level analysis.9–11 Using computational volume of tissue activated (VTA) modeling, we can approximate the neural substrates being stimulated.12–14 These VTAs can then be clinically weighted and analyzed to allow estimation of brain areas or connectivity profiles that correspond to optimal outcome. Prior investigative efforts have employed VTA modeling to characterize the local region of best response to GPi-DBS in patients with idiopathic and genetic dystonia and clinically effective DBS contacts.15,16
As previously highlighted, pediatric DBS studies for dystonia are currently lacking in the literature.17 To our knowledge, none of those studies have performed group-level voxel-wise analyses to identify neural substrates and/or networks associated with clinical benefits (Table 1). Similarly, a recent meta-analysis reviewed the clinical outcomes in childhood dystonia across subtypes,5 but did not include imaging findings.
Studies that evaluated DBS in pediatric dystonia*
| Mean BFMDRS Motor Score (range) | ||||
|---|---|---|---|---|
| Authors & Year | No. of Pts | Mean Age at Surgery (range), yrs | Preop | Postop |
| Cersosimo et al., 200841 | 7 | 13 (9–19) | 59 (33–96) | 30 (12–66) |
| Ghosh et al., 201218 | 8 | 14 (8–21) | NR | NR |
| Goto et al., 200642 | 2 | 18 (15–21) | 65 (61–68.5) | 1.5 (0–3) |
| Jin et al., 201243 | 1 | 7 | 34 | 3 |
| Keen et al., 201444 | 5 | 11 (8–17) | 61 (38.5–102) | 52 (34–71) |
| Krause et al., 200445 | 4 | 17.5 (13–21) | 80 (66.5–92) | 36 (24.5–49) |
| Krause et al., 201546 | 1 | 20 | 44 | 27 |
| Krause et al., 201647 | 8 | 12.5 (7–17) | 45 (16–74.5) | 23 (1.5–35.5) |
| Kupsch et al., 200348 | 1 | 14 | 34.5 | 27 |
| Mehrkens et al., 201049 | 5 | 13 (8–16) | NR | NR |
| Miyagi & Koike, 201350 | 2 | 10 (9–11) | 60 (42–78) | 76 (63.5–88) |
| Olaya et al., 201351 | 9 | 16 (6–20) | 86 (58–108) | 79 (53.3–108) |
| Oterdoom et al., 201852 | 1 | 9 | 71 | 64 |
| Parr et al., 200753 | 4 | 12 (8–14) | 83 (66–103) | 34 (8–53) |
| Petrossian et al., 201354 | 13 | 14 (9–17) | 40 (12–86) | 22 (0–61.5) |
| Starr et al., 200455 | 7 | 16 (12–18) | 64 (34–90) | 29 (0–71) |
| Tronnier & Fogel, 200056 | 1 | 13 | 77 | 50.5 |
| Vidailhet et al., 200557 | 5 | 17 (15–19) | 38 (25–56) | 17 (0–46) |
| Zorzi et al., 200558 | 9 | 10 (8–15) | 57 (44–91) | 33 (14–83) |
NR = not reported; Pts = patients.
None of these studies included group-level neuroimaging-based inferences.
Adapted from Hale et al., 2020.17
In this paper, we 1) report our experience with GPi-DBS as a treatment for pediatric dystonia, and 2) employ neuroimaging techniques to examine the neural substrates associated with clinical benefits in this patient cohort. This may help guide surgical lead implantation, streamline clinical programming, and inform the pathophysiology of this heterogeneous group of diseases.
Methods
Surgical Procedure
Pediatric patients who met the diagnostic criteria for refractory dystonia underwent GPi-DBS between June 2009 and September 2017. All patients had medically refractory disease and highly disabling symptomatology. Patients were considered for GPi-DBS when 1) dystonia was the main source of disability, 2) a more conservative approach (namely medications and botulinum toxin injection) had failed, and 3) there were no contraindications to surgery. Patients or their guardians provided written consent for the surgical procedure in each case.
All patients underwent unilateral or bilateral electrode implantation of quadripolar DBS electrodes (Medtronic Model 3387) under general anesthesia in a manner similar to that described previously.18 The posteroventral GPi was targeted using direct surgical targeting guided by microelectrode recordings. Intraoperative microelectrode recordings were performed to verify lead placement in the motor GPi territory. Subsequently, implanted electrodes were connected to an implantable pulse generator (Activa PC, Medtronic Model 37601), which was placed in an infraclavicular subcutaneous pocket under general anesthesia.
Pre- and postoperative imaging was acquired on the day of surgery for the respective purposes of surgical planning and electrode location confirmation. Imaging was performed on a 1.5-T system (1.5-T GE Signa Excite MRI, software level HD 12X; GE Healthcare) with a transmit-receive head coil (GE Model 2341973). Three-dimensional T1-weighted fast spoiled gradient echo sequence images were used for analysis (voxel size 1 × 1 × 1 mm, TR = 12 msec, TE = 5 msec, flip angle 20°).
DBS programming was initiated by a movement disorders neurologist approximately 2 months after surgery. Commencement of stimulation programming and adjustment of settings in the subsequent course followed standardized protocols used in the adult population.8
Clinical Outcome Measures and Follow-Up
All patients were assessed both pre- and postoperatively by a neurologist specializing in movement disorders. The patients’ charts and videos were retrospectively reviewed and assessed for clinical improvement (by M.A. and L.M.O.). A 7-point Clinical Global Impression scale was used to assess clinical improvement from the preoperative time point to overcome the absence of Burke-Fahn-Marsden Dystonia Rating Scale (BFMDRS)19 availability in all patients, as has been used previously.20,21 Postoperative improvement was graded as follows: 7 = very much worse, 6 = much worse, 5 = minimally worse, 4 = no improvement, 3 = minimally improved, 2 = much improved, 1 = very much improved. Abnormal movements were also graded with the BFMDRS in patients for whom recorded video data were available (8 preoperatively, 9 postoperatively). When available, BFMDRS scores were incorporated into judgment of global improvement.
Neuroimaging Analysis
Retrospective imaging analysis was performed following research ethics board approval from the University Health Network Research Ethics Board.
DBS Lead Localization
Lead localization largely followed the methodology described by Horn and Kühn.22 In Lead-DBS (www.lead-dbs.org), the pre- and postoperative 3D T1-weighted MR images were rigidly coregistered (http://www.fil.ion.ucl.ac.uk/spm/software/spm12/). Then, the postoperative MRI was nonlinearly warped with “low variance”23 advanced normalization tools (http://stnava.github.io/ANTs/) to Montreal Neurological Institute (MNI) space (MNI152 NLIN 2009b atlas) using the preoperative-to-MNI transforms. Each registration step was visually checked (by A.C.). Following initial prereconstruction, the electrode lead trajectories were manually refined based on the artifact as agreed upon by two trained observers (A.C. and G.J.B.E.), one of whom has experience performing this technique in more than 1000 patients (G.J.B.E.). The post- to pretransforms were corrected for postoperative brain shift using the Lead-DBS subcortical affine transform when required. It has previously been shown that pediatric brains can be successfully normalized to an adult template in children older than 6 years of age.24
VTA Estimation
VTA estimation was conducted in Lead-DBS following the approach described by Horn et al.25 Briefly, the Iso2Mesh toolbox was used to construct a volume conductor model of the DBS electrode and surrounding tissue through creation of a tetrahedral volume mesh. Subcortical gray matter nuclei were defined using the DISTAL atlas,26 and regions devoid of gray matter material were labeled as white matter. Conductivities of 0.33 and 0.14 S/m were used for gray and white matter, respectively. The potential distribution resulting from stimulation as modeled using optimal stimulation settings for each individual patient was then computed using the FieldTrip-SimBio finite element model pipeline and thresholded with a 0.2 V/mm gradient to derive binary VTAs in MNI space.27 Given the small number of patients, all left-sided VTAs were flipped in the midsagittal plane for group-level analysis (n = 21 right-sided VTAs).
Probabilistic Map of Efficacy
Next, a voxel-wise probabilistic map of efficacy was computed in MNI space. A metric of the average clinical improvement at each stimulated voxel was computed by weighting patient VTAs according to improvement (based on the 7-point clinical scale) at the best time point. A frequency map was also computed, which denoted the proportion of VTAs overlapping each voxel. The final probabilistic map of efficacy was calculated by multiplying the average map and the frequency map at each voxel.28 Volume overlap between the positive voxels within this probabilistic map and neighboring structures of interest (as defined by the DISTAL atlas)26 was determined to assess the relationship between clinical outcome and relevant neuroanatomy.
Connectomic Analysis
To investigate patterns of functional and structural connectivity associated with clinical benefits, a normative connectomic analysis was performed.29,30 Specifically, the probabilistic map described above was utilized as a weighted seed region in conjunction with two normative data sets derived from large numbers of healthy subjects: a 1000-subject resting-state functional MRI (rsfMRI) data set31 and a 985-subject diffusion-weighted MRI whole-brain tractography data set. Normative data have been shown to predict clinical outcome in other DBS populations25 and are often able to provide higher spatial resolution and better signal-to-noise ratios than native patient images. While network integration changes with maturation, it has been shown that connectivity patterns do not change drastically during development,32–34 thus validating the use of adult normative data in the context of pediatric patients. With respect to functional connectomic mapping, a whole-brain, voxel-wise connectivity r-map was obtained on the basis of rsfMRI time course–dependent correlations with the seed.35 To compute structural connectivity, the whole-brain tractography data set was seeded with the probabilistic map. Streamlines touching the seed were retained and converted to voxel-wise streamline density maps.
The functional and structural connectivity of the probabilistic map were compared to that of the “traditional” GPi-DBS coordinate-based target used for dystonia. The traditional target volume was generated by placing a theoretical DBS electrode in MNI space such that its second most ventral contact was centered on the right-sided GPi coordinates specified by Horn et al. in their probabilistic atlas of DBS targets.36 The mean voltage settings of the cohort (C+1−, 3 V) were then modeled as described above to approximate a typical GPi-DBS VTA. The functional and structural networks associated with this volume were determined as detailed for the patients above.
Results
Patients
Eleven pediatric patients with refractory dystonia (2 with genetically determined dystonia, 7 with acquired, and 2 with idiopathic) were treated with GPi-DBS between June 2009 and September 2017. Table 2 features patient demographics, disease etiology, and clinical outcomes. Ten of the 11 patients received bilateral lead implantation, while 1 patient with XXX syndrome had only a single implanted lead (21 total implanted leads). The mean (± SD) age at surgery was 12 ± 4 years, with a mean follow-up of 19 ± 21 months until last follow-up (time point of recorded optimal benefit). Ten (91%) of 11 patients showed response to DBS, as defined by a score of 3 or less on the 7-point postoperative rating scale. The patient who showed no clinical improvement (patient 5) developed status dystonicus immediately postoperatively in the absence of electrical stimulation; these symptoms improved over time following treatment with intrathecal baclofen and multiple oral medications. There were no stimulation-induced adverse events in any patient. Patient 2 showed clinical improvement but, due to subsequent infection of the DBS device, required removal of the system after 6 months of stimulation.
Patient demographics, disease duration, medications, duration of follow-up, stimulation settings, and postsurgical outcome
| Pt No. | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | Mean ± SD |
| Sex | M | F | F | F | M | F | M | F | F | M | M | |
| Disease etiology | CP | Near-drowning (HII) | CP | CP | Idiopathic | CP | VPS41 mutation | XXX syndrome | CP | Dystonia myoclonus (SGCE-negative) | Idiopathic | |
| Disease duration (yrs) | 5 | 9 | 8 | 13 | 7 | 8 | 6 | 14 | 17 | 11 | 11 | 10 ± 4 |
| Age at surgery (yrs) | 6 | 11 | 9 | 15 | 10 | 9 | 13 | 17 | 17 | 13 | 13 | 12 ± 4 |
| Length of follow-up (mos) | 5 | 4 | 29 | 4 | 3 | 4 | 41 | 4 | 37 | 7 | 65 | 19 ± 21 |
| Postop complications | None | Hardware infection | None | None | Status dystonicus | None | None | None | None | None | None | |
| No. of antidystonia medications: preop/postop | 3/2 | 1/1 | 1/2 | 1/2 | 5/6 | 2/2 | 3/2 | 1/1 | 1/1 | 0/0 | 4/4 | 2/2 |
| Dominant phenomenology | Limbs > Tr | Gen | UL > LL > Tr | Gen | Gen | Gen | LL > UL | Gen | LL > UL | Tr + UL > LL | Gen | |
| BFMDRS motor score | ||||||||||||
| Preop/postop | NA/34 | 77/NA | 85/53 | NA/69 | 75/35 | NA/52 | 95/82 | 52/29 | 75/NA | 24/06 | 69/59 | |
| % improvement | 38 | 53 | 14 | 44 | 75 | 15 | 40 ± 23 | |||||
| BFMDRS disability preop/postop | NA/25 | NA/NA | 24/23 | NA/24 | NA/27 | NA/28 | NA/26 | 12/13 | 20/NA | NA/NA | NA/24 | |
| Clinical rating score | 3 | 3 | 3 | 3 | 4 | 3 | 3 | 3 | 3 | 2 | 3 | 3 ± 0.4 |
| DBS settings (average of 2 sides) | ||||||||||||
| Amplitude (V) | 4 | 3.3 | 3.6 | 1.9 | 3 | 4 | 2.6 | 4* | 5.1 | 3.4 | 2.5 | 3 ± 1 |
| Mode of stimulation | M | M | M, I | M | M | M | M | M* | M | M | M | |
| Pulse width (μsec) | 60 | 75 | 60 | 60 | 120 | 60 | 90 | 60* | 60 | 60 | 135 | 76 ± 27 |
| Frequency (Hz) | 130 | 130 | 125 | 130 | 130 | 180 | 125 | 60* | 60 | 180 | 130 | 125 ± 38 |
CP = cerebral palsy; Gen = generalized; HII = hypoxic ischemic injury; I = interleaved; LL = lower limbs; M = monopolar; NA = not available; Tr = Trunk; UL = upper limbs.
Unilateral (right).
Electrode Locations and VTA
Once registered to standard MNI space, active electrode contacts were noted to be centered around the posteroventral GPi (Fig. 1). There was overlap in the distribution of active contacts of patients with better clinical outcomes and those experiencing little benefit. The mean distance from the active contact to the nearest voxel in the primary motor GPi26 was 2.8 ± 1.9 mm (Table 3).
Active contact distribution. The distribution of active contacts (red) is shown in sagittal (A) and coronal (B) slices of a 100-micron, 7-T brain in MNI space.59 A 3D representation of the GPi, as defined by the DISTAL atlas,26 is overlaid in green. A total of 29 contacts were stimulated across 21 leads in 11 patients (1 patient had a unilateral lead). The x and y coordinates (sagittal and coronal, respectively) denote the coordinates of the image slice in MNI reference space. Figure is available in color online only.
Active contact coordinates in MNI space
| MNI152 NLIN 2009b Space | Patient-Specific Space (relative to MCP) | |||||||
|---|---|---|---|---|---|---|---|---|
| Pt No. | Contact Side | Active Contacts | X | Y | Z | Lateral (mm) | Anterior (mm) | Inferior (mm) |
| 1 | Rt | 3 | 22.2 | −5.4 | 0.2 | 19.2 | 4.8 | −4.8 |
| Lt | 11 | −19.8 | −8.2 | 0.2 | 17.6 | 2.3 | −3.7 | |
| 2 | Rt | 3 | 20.9 | −7.6 | 1.0 | 16.3 | 3.6 | −4.1 |
| Lt | 10 | −21.6 | −3.3 | 1.7 | 17.8 | 5.9 | −4.8 | |
| Lt | 11 | −21.4 | −1.2 | 4.8 | 17.7 | 7.4 | −7.4 | |
| 3 | Rt | 1 | 25.7 | −11.4 | −5.5 | 20.4 | 0.9 | 1.7 |
| Rt | 3 | 25.2 | −7.9 | 0.7 | 19.8 | 3.4 | −4.0 | |
| Lt | 8 | −24.9 | −13.0 | −12.4 | 19.9 | −0.9 | 6.6 | |
| Lt | 9 | −24.8 | −11.3 | −9.3 | 19.8 | 0.3 | 3.9 | |
| Lt | 11 | −24.7 | −7.9 | −3.2 | 19.5 | 2.8 | −1.7 | |
| 4 | Rt | 0 | 21.7 | −12.9 | −9.3 | 18.3 | −0.2 | 4.7 |
| Lt | 8 | −24.3 | −10.7 | −8.2 | 20.1 | 0.6 | 3.9 | |
| 5 | Rt | 1 | 22.7 | −10.4 | −6.3 | 19.4 | 3.0 | 1.9 |
| Lt | 9 | −21.4 | −7.2 | −7.7 | 18.4 | 3.4 | 3.4 | |
| 6 | Rt | 2 | 25.1 | −9.6 | −3.4 | 20.4 | 2.7 | −1.1 |
| Lt | 10 | −24.1 | −8.3 | −3.1 | 19.5 | 1.8 | −0.4 | |
| 7 | Rt | 1 | 24.1 | −9.9 | −4.9 | 20.7 | 1.7 | 1.1 |
| Rt | 2 | 24.1 | −8.3 | −1.6 | 20.6 | 2.8 | −1.7 | |
| Lt | 8 | −22.8 | −11.5 | −6.9 | 19.9 | 1.0 | 2.5 | |
| Lt | 9 | −23.0 | −9.7 | −3.6 | 20.0 | 2.3 | −0.3 | |
| 8 | Rt | NA | NA | NA | NA | NA | NA | NA |
| Lt | 10 | −25.8 | −6.8 | 0.2 | 22.3 | 3.7 | −3.0 | |
| 9 | Rt | 0 | 23.0 | −13.1 | −12.2 | 18.2 | 0.1 | 6.4 |
| Lt | 8 | −20.5 | −14.1 | −8.6 | 19.4 | −2.6 | 4.5 | |
| 10 | Rt | 2 | 21.3 | −7.1 | −3.4 | 19.3 | 4.6 | −1.8 |
| Lt | 10 | −22.1 | −9.1 | −2.1 | 20.1 | 2.7 | −0.9 | |
| 11 | Rt | 1 | 17.7 | −10.1 | −2.3 | 15.6 | 1.2 | −1.4 |
| Rt | 3 | 18.1 | −7.2 | 4.4 | 15.9 | 3.0 | −7.0 | |
| Lt | 8 | −23.1 | −13.4 | −3.6 | 20.0 | −0.8 | 0.5 | |
| Lt | 9 | −22.9 | −11.8 | −0.2 | 19.9 | 0.2 | −2.5 | |
MCP = midcommissural point.
Active refers to the specific contact (of the 4 available) on the electrode being used to deliver the stimulation. The standard nomenclature for these contacts is 0–3 on the right and 8–11 on the left. For example, patient 2 was using monopolar settings on the right and bipolar settings on the left. Contacts 3, 10, and 11 were therefore considered active.
Probabilistic Mapping and Connectivity Patterns
The probabilistic map of efficacy showed that the voxel cluster most associated with clinical improvement was located at the posterior aspect of the GPi. The top 50% of voxels by voxel value were centered on the following MNI coordinates: x = 23.7, y = −9.4, z = −3.7. These were posterior and somewhat superior to the center-of-gravity coordinates of the “classic” GPi target as defined by Horn et al.36 (x = 22.4, y = −5.6, z = −5.0; Fig. 2). The top 50% of probabilistic map voxels showed considerable overlap with the GPi proper (59.3% overlap); specifically, the greatest overlap was observed with the premotor (12.8%) and primary motor (11.7%) territories. A substantial portion of the map (29.8%) encroached on the globus pallidus pars externa (GPe), while the remainder of the voxels (10.8%) putatively fell within the neighboring white matter (Fig. 3).
Stimulation site associated with clinical benefits after optimized programming. The traditional stereotactic GPi target36 for dystonia (cyan) is projected on axial (A), coronal (B), and sagittal (C) T1-weighted MRI (MNI152 NLIN 2009b). Warm colors depict the probabilistic map of efficacious stimulation for pediatric dystonia by comparison to the traditional stereotactic target. Clinical benefit was seen with activation slightly posterior and superior to the traditional GPi target. Figure is available in color online only.
GPi overlap of the probabilistic map of efficacy. Percentage volume overlap of the probabilistic map of efficacy with GPe and relevant GPi subsections.26 Overall overlap with the GPi was 30%. Figure is available in color online only.
As determined by normative connectomic analysis, the pediatric probabilistic map of efficacy showed broad similarities in terms of functional connectivity with the classic dystonic GPi target as defined by Horn et al.36 Specifically, both seeds demonstrated strong positive correlations to the insular cortex, central and precentral gyri, pallidum, putamen, brainstem, and cerebellum (Fig. 4A and B). Regions functionally negatively correlated with both seeds included the posterior parietal lobe, occipital lobe, and temporal regions. However, there were also differences between the probabilistic map and the classic target. In particular, the middle frontal gyrus as well as the posterior parietal and occipital region were more strongly negatively correlated with the pediatric probabilistic map, while aspects of the temporal lobe and ventromedial prefrontal cortex were more strongly negatively correlated with the conventional GPi target.
Pattern of functional and structural connectivity associated with clinical benefits. The probabilistic map of success was used as a seed into the functional structural template, with resulting connectivity demonstrated in A and B. A and B: Panel A depicts patterns of functional connectivity acquired from rsfMRI correlations for our pediatric cohort, in comparison to panel B, the functional DBS target for dystonia. C and D: Structural connectivity, as defined by streamline density, for the pediatric probabilistic map (C) and the traditional stereotactic target (D). Figure is available in color online only.
With regard to structural connectivity, similarities were again noted between the pediatric probabilistic map and the conventional GPi-DBS target. High structural connectivity, as denoted by streamline density, was noted between both seeds and areas including the precentral and postcentral gyri, parietooccipital region, and brainstem (Fig. 4C and D). Overall, these areas demonstrated greater streamline density with respect to the probabilistic map.
Discussion
In this cohort of pediatric dystonia cases, we showed that VTA-based probabilistic mapping suggests a convergence of DBS-induced clinical benefits within the posterior aspect of the GPi. There was an associated functional and structural connectomic pattern associated with this site, showing both similarities and differences with respect to the conventional coordinate-based target.
A recent meta-analysis found that 88.2% of children with an inherited dystonia showed clinically significant (> 20% improvement in motor score) response to DBS. However, dystonic children with cerebral palsy—one of the most frequent causes of childhood dystonia—responded poorly (median 11.1% motor improvement).5 As such, any insight that could aid in improving treatment response for this often severely debilitated patient population is important. Younger age of onset and acquired dystonia are known predictors of poor treatment response, factors that affected much of our patient population. A systematic review of DBS in pediatric dystonia showed a mean (± SD) 43.7% ± 31% improvement in BFMDRS score, regardless of dystonia type.17 Consistent with the current literature, our 6 patients with available pre- and postoperative BFMDRS motor scores demonstrated a mean 40% ± 23% motor response to DBS. Despite inherent predictors of poor outcome, 10 of our patients showed response to DBS using the 7-point scale.
Inaccurate lead placement is a reported cause of therapeutic failure for DBS in dystonia,37 highlighting the importance of precise surgical targeting. As defined by probabilistic mapping, the optimal stimulation site in our cohort was lateral and superior to the widely described surgical target.36 While in theory the electrodes are surgically inserted in similar places, individualized programming empirically drives the stimulation. This, at least in part, accounts for the differences between our hotspot—based upon clinically weighted VTA locations—and the location of active contacts described by Horn et al.36 In 20 adult patients with DBS-responsive cervical dystonia, an acute stimulation challenge produced maximum response with stimulation around the lamina medullaris, between the posteroventral GPi and GPe.38 In agreement with the adult literature, a stimulation volume encompassing both the ventroposterior pallidal region and adjacent subpallidal white matter appears to result in stronger clinical benefit in our cohort. It has been proposed that DBS may preferentially affect axons, perhaps explaining why our site of probabilistic clinical benefit encroached on the adjacent white matter.39 Connectomic analysis of this probabilistic map, moreover, indicated its dense interconnectedness with primary motor, premotor, and somatosensory cortical regions, underscoring the importance of relationships between subcortical DBS targets and relevant cortical regions.
This work has potential applications for the refinement of surgical targeting and postoperative optimization of DBS therapy for pediatric dystonia. While preliminary, the probabilistic hotspot and connectomic signature described here present an opportunity for guiding future electrode placements and DBS programming. The zone of optimal activation and its associated functional and structural connectivity patterns could be transformed into native patients and used in conjunction with pre- and postoperative imaging. An example of such an experimental method has already been performed on at least 1 adult patient with essential tremor.40 Eventual clinical applications would require larger sample sizes (likely by means of multicenter collaboration) and the adoption of more refined methods, such as robust statistical analysis and perhaps machine learning, to identify the substrates of efficacious outcome. This approach would enable information from retrospective cases to effectively inform future intervention.
The present study has several limitations. Although, to our knowledge, it is the largest study making neuroimaging inferences in pediatric patients with dystonia, it remains exploratory given its small sample size and retrospective character. In addition, we were unable to rigorously quantify or validate these findings due to the small sample size. Furthermore, the etiology of our patients’ underlying disease was heterogeneous. Lastly, we employed large-scale normative data sets for our connectomic analysis; while these may differ somewhat from the native connectivity of pediatric dystonia patients, they have previously been shown to be informative in many other diseases, including Parkinson’s disease and dementia.25,35
Conclusions
DBS for pediatric dystonia is a complex treatment heightened by multiple sources of variability, including heterogeneous disease etiologies and variance in surgical lead placement and postoperative programming. This treatment demands a multidisciplinary team and months of pre- and postinsertion management. Patients affected by this devastating disease have few other treatment options. We reported on 11 cases of pediatric dystonia in which GPi-DBS was used to overall good effect, describing a clinically favorable stimulation site and its structural and functional connectivity signature. These results may provide valuable insights into improving surgical planning, facilitating programming, and informing disease pathophysiology.
Acknowledgments
This work was supported by the German Research Foundation (Deutsche Forshungsgemeinschaft, DFG NE 2276/1-1; C.N.), the RR Tasker Chair in Functional Neurosurgery at University Health Network, a Tier 1 Canada Research Chair in Neuroscience (A.M.L.), and the Chair in Neuromodulation and Multi-Disciplinary Care at UofT/UHN (A.F.). Dr. Coblentz was responsible for the data analysis. The corresponding author confirms that he had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Disclosures
Dr. Oliveira reports receiving funding for travel from Medtronic. Dr. Kalia reports being a consultant to Medtronic. Dr. Lozano reports being a consultant to Boston Scientific, Medtronic, Abbott, and Insightec; and being an employee of Functional Neuromodulation. Dr. Fasano reports being a consultant to, and receiving honoraria from, Abbott, Boston Scientific, and Medtronic.
Author Contributions
Conception and design: Fasano, Coblentz, Elias, Boutet, Widjaja, Lozano. Acquisition of data: Coblentz, Elias, Boutet, Algarni, Oliveira, Ibrahim, Lozano. Analysis and interpretation of data: Coblentz, Elias, Boutet, Germann, Algarni, Oliveira, Neudorfer. Drafting the article: Coblentz. Critically revising the article: Fasano, Elias, Boutet, Germann, Neudorfer, Widjaja, Ibrahim, Kalia, Lozano. Reviewed submitted version of manuscript: Fasano, Elias, Boutet, Kalia. Approved the final version of the manuscript on behalf of all authors: Fasano. Administrative/technical/material support: Jain. Study supervision: Fasano.
Supplemental Information
Previous Presentations
Portions of this work were presented in abstract form on May 3, 2019, at the Society for Pediatric Radiology conference in San Francisco, California.
References
- 1↑
Sanger TD, Chen D, Fehlings DL, et al. Definition and classification of hyperkinetic movements in childhood. Mov Disord. 2010;25(11):1538–1549.
- 2↑
Balint B, Bhatia KP. Dystonia: an update on phenomenology, classification, pathogenesis and treatment. Curr Opin Neurol. 2014;27(4):468–476.
- 4↑
Lumsden DE, Kaminska M, Tomlin S, Lin J-P. Medication use in childhood dystonia. Eur J Paediatr Neurol. 2016;20(4):625–629.
- 5↑
Elkaim LM, Alotaibi NM, Sigal A, et al. Deep brain stimulation for pediatric dystonia: a meta-analysis with individual participant data. Dev Med Child Neurol. 2019;61(1):49–56.
- 6↑
Eltahawy HA, Saint-Cyr J, Giladi N, et al. Primary dystonia is more responsive than secondary dystonia to pallidal interventions: outcome after pallidotomy or pallidal deep brain stimulation. Neurosurgery. 2004;54(3):613–621.
- 7↑
Vidailhet M, Yelnik J, Lagrange C, et al. Bilateral pallidal deep brain stimulation for the treatment of patients with dystonia-choreoathetosis cerebral palsy: a prospective pilot study. Lancet Neurol. 2009;8(8):709–717.
- 8↑
Picillo M, Lozano AM, Kou N, et al. Programming deep brain stimulation for tremor and dystonia: the Toronto Western Hospital algorithms. Brain Stimul. 2016;9(3):438–452.
- 9
Hamani C, Moro E, Zadikoff C, et al. Location of active contacts in patients with primary dystonia treated with globus pallidus deep brain stimulation. Neurosurgery. 2008;62(3 Suppl 1):217–225.
- 10
Horn A, Li N, Dembek TA, et al. Lead-DBS v2: Towards a comprehensive pipeline for deep brain stimulation imaging. Neuroimage. 2019;184:293–316.
- 11
Boutet A, Gramer R, Steele CJ, et al. Neuroimaging technological advancements for targeting in functional neurosurgery. Curr Neurol Neurosci Rep. 2019;19(7):42.
- 12
Johnson MD, McIntyre CC. Quantifying the neural elements activated and inhibited by globus pallidus deep brain stimulation. J Neurophysiol. 2008;100(5):2549–2563.
- 13
Butson CR, Cooper SE, Henderson JM, McIntyre CC. Patient-specific analysis of the volume of tissue activated during deep brain stimulation. Neuroimage. 2007;34(2):661–670.
- 14
Chaturvedi A, Luján JL, McIntyre CC. Artificial neural network based characterization of the volume of tissue activated during deep brain stimulation. J Neural Eng. 2013;10(5):056023.
- 15↑
Cheung T, Noecker AM, Alterman RL, et al. Defining a therapeutic target for pallidal deep brain stimulation for dystonia. Ann Neurol. 2014;76(1):22–30.
- 16↑
Reich MM, Horn A, Lange F, et al. Probabilistic mapping of the antidystonic effect of pallidal neurostimulation: a multicentre imaging study. Brain. 2019;142(5):1386–1398.
- 17↑
Hale AT, Monsour MA, Rolston JD, et al. Deep brain stimulation in pediatric dystonia: a systematic review. Neurosurg Rev. 2020;43(3):873–880.
- 18↑
Ghosh PS, Machado AG, Deogaonkar M, Ghosh D. Deep brain stimulation in children with dystonia: experience from a tertiary care center. Pediatr Neurosurg. 2012;48(3):146–151.
- 19↑
Burke RE, Fahn S, Marsden CD, et al. Validity and reliability of a rating scale for the primary torsion dystonias. Neurology. 1985;35(1):73–77.
- 20↑
Smit M, Kamphuis ASJ, Bartels AL, et al. Fatigue, sleep disturbances, and their influence on quality of life in cervical dystonia patients. Mov Disord Clin Pract (Hoboken). 2016;4(4):517–523.
- 21↑
Guy W. The Clinical Global Impression Scale. In: ECDEU Assessment Manual for Psychopharmacology–Revised. US Department of Health, Education, & Welfare; ADAMHA; NIMH Psychopharmacology Research Branch; 1976:218–222.
- 22↑
Horn A, Kühn AA. Lead-DBS: a toolbox for deep brain stimulation electrode localizations and visualizations. Neuroimage. 2015;107:127–135.
- 23↑
Ewert S, Horn A, Finkel F, et al. Optimization and comparative evaluation of nonlinear deformation algorithms for atlas-based segmentation of DBS target nuclei. Neuroimage. 2019;184:586–598.
- 24↑
Muzik O, Chugani DC, Juhász C, et al. Statistical parametric mapping: assessment of application in children. Neuroimage. 2000;12(5):538–549.
- 25↑
Horn A, Reich M, Vorwerk J, et al. Connectivity predicts deep brain stimulation outcome in Parkinson disease. Ann Neurol. 2017;82(1):67–78.
- 26↑
Ewert S, Plettig P, Li N, et al. Toward defining deep brain stimulation targets in MNI space: a subcortical atlas based on multimodal MRI, histology and structural connectivity. Neuroimage. 2018;170:271–282.
- 27↑
Vorwerk J, Cho J-H, Rampp S, et al. A guideline for head volume conductor modeling in EEG and MEG. Neuroimage. 2014;100:590–607.
- 28↑
Boutet A, Ranjan M, Zhong J, et al. Focused ultrasound thalamotomy location determines clinical benefits in patients with essential tremor. Brain. 2018;141(12):3405–3414.
- 29↑
Yeh F-C, Tseng W-YI. NTU-90: a high angular resolution brain atlas constructed by q-space diffeomorphic reconstruction. Neuroimage. 2011;58(1):91–99.
- 30↑
Horn A, Neumann W-J, Degen K, et al. Toward an electrophysiological “sweet spot” for deep brain stimulation in the subthalamic nucleus. Hum Brain Mapp. 2017;38(7):3377–3390.
- 31↑
Yeo BT, Krienen FM, Sepulcre J, et al. The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J Neurophysiol. 2011;106(3):1125–1165.
- 32
Qin J, Chen S-G, Hu D, et al. Predicting individual brain maturity using dynamic functional connectivity. Front Hum Neurosci. 2015;9:418.
- 33
Faghiri A, Stephen JM, Wang Y-P, et al. Changing brain connectivity dynamics: from early childhood to adulthood. Hum Brain Mapp. 2018;39(3):1108–1117.
- 34
Marek S, Hwang K, Foran W, et al. The contribution of network organization and integration to the development of cognitive control. PLoS Biol. 2015;13(12):e1002328.
- 35↑
Fox MD, Buckner RL, Liu H, et al. Resting-state networks link invasive and noninvasive brain stimulation across diverse psychiatric and neurological diseases. Proc Natl Acad Sci U S A. 2014;111(41):E4367–E4375.
- 36↑
Horn A, Kühn AA, Merkl A, et al. Probabilistic conversion of neurosurgical DBS electrode coordinates into MNI space. Neuroimage. 2017;150:395–404.
- 37↑
Pauls KAM, Krauss JK, Kämpfer CE, et al. Causes of failure of pallidal deep brain stimulation in cases with pre-operative diagnosis of isolated dystonia. Parkinsonism Relat Disord. 2017;43:38–48.
- 38↑
Schönecker T, Gruber D, Kivi A, et al. Postoperative MRI localisation of electrodes and clinical efficacy of pallidal deep brain stimulation in cervical dystonia. J Neurol Neurosurg Psychiatry. 2015;86(8):833–839.
- 39↑
Kiss ZHT, Wilkinson M, Krcek J, et al. Is the target for thalamic deep brain stimulation the same as for thalamotomy? Mov Disord. 2003;18(10):1169–1175.
- 40↑
Al-Fatly B, Ewert S, Kübler D, et al. Connectivity profile of thalamic deep brain stimulation to effectively treat essential tremor. Brain. 2019;142(10):3086–3098.
- 41↑
Cersosimo MG, Raina GB, Piedimonte F, et al. Pallidal surgery for the treatment of primary generalized dystonia: long-term follow-up. Clin Neurol Neurosurg. 2008;110(2):145–150.
- 42↑
Goto S, Yamada K, Shimazu H, et al. Impact of bilateral pallidal stimulation on DYT1-generalized dystonia in Japanese patients. Mov Disord. 2006;21(10):1785–1787.
- 43↑
Jin ST, Lee MK, Ghang JY, Jeon SM. Deep brain stimulation of the globus pallidus in a 7-year-old girl with DYT1 generalized dystonia. J Korean Neurosurg Soc. 2012;52(3):261–263.
- 44↑
Keen JR, Przekop A, Olaya JE, et al. Deep brain stimulation for the treatment of childhood dystonic cerebral palsy. J Neurosurg Pediatr. 2014;14(6):585–593.
- 45↑
Krause M, Fogel W, Kloss M, et al. Pallidal stimulation for dystonia. Neurosurgery. 2004;55(6):1361–1370.
- 46↑
Krause P, Brüggemann N, Völzmann S, et al. Long-term effect on dystonia after pallidal deep brain stimulation (DBS) in three members of a family with a THAP1 mutation. J Neurol. 2015;262(12):2739–2744.
- 47↑
Krause P, Lauritsch K, Lipp A, et al. Long-term results of deep brain stimulation in a cohort of eight children with isolated dystonia. J Neurol. 2016;263(11):2319–2326.
- 48↑
Kupsch A, Klaffke S, Kühn AA, et al. The effects of frequency in pallidal deep brain stimulation for primary dystonia. J Neurol. 2003;250(10):1201–1205.
- 49↑
Mehrkens JH, Borggraefe I, Feddersen B, et al. Early globus pallidus internus stimulation in pediatric patients with generalized primary dystonia: long-term efficacy and safety. J Child Neurol. 2010;25(11):1355–1361.
- 50↑
Miyagi Y, Koike Y. Tolerance of early pallidal stimulation in pediatric generalized dystonia. J Neurosurg Pediatr. 2013;12(5):476–482.
- 51↑
Olaya JE, Christian E, Ferman D, et al. Deep brain stimulation in children and young adults with secondary dystonia: the Children’s Hospital Los Angeles experience. Neurosurg Focus. 2013;35(5):E7.
- 52↑
Oterdoom DLM, van Egmond ME, Ascencao LC, et al. Reversal of status dystonicus after relocation of pallidal electrodes in DYT6 generalized dystonia. Tremor Other Hyperkinet Mov (N Y). 2018;8:530.
- 53↑
Parr JR, Green AL, Joint C, et al. Deep brain stimulation in childhood: an effective treatment for early onset idiopathic generalised dystonia. Arch Dis Child. 2007;92(8):708–711.
- 54↑
Petrossian MT, Paul LR, Multhaupt-Buell TJ, et al. Pallidal deep brain stimulation for dystonia: a case series. J Neurosurg Pediatr. 2013;12(6):582–587.
- 55↑
Starr PA, Turner RS, Rau G, et al. Microelectrode-guided implantation of deep brain stimulators into the globus pallidus internus for dystonia: techniques, electrode locations, and outcomes. Neurosurg Focus. 2004;17(1):E4.
- 56↑
Tronnier VM, Fogel W. Pallidal stimulation for generalized dystonia. Report of three cases. J Neurosurg. 2000;92(3):453–456.
- 57↑
Vidailhet M, Vercueil L, Houeto J-L, et al. Bilateral deep-brain stimulation of the globus pallidus in primary generalized dystonia. N Engl J Med. 2005;352(5):459–467.
- 58↑
Zorzi G, Marras C, Nardocci N, et al. Stimulation of the globus pallidus internus for childhood-onset dystonia. Mov Disord. 2005;20(9):1194–1200.
- 59↑
Edlow BL, Mareyam A, Horn A, et al. 7 Tesla MRI of the ex vivo human brain at 100 micron resolution. Sci Data. 2019;6(1):244.
