Long-Term Follow-Up of Unilateral Deep Brain Stimulation Targeting the Caudal Zona Incerta in 13 Patients with Parkinsonian Tremor

Introduction: Deep brain stimulation (DBS) is an established treatment for Parkinson’s disease (PD) and other movement disorders. The ventral intermediate nucleus of the thalamus is considered as the target of choice for tremor disorders, including tremor-dominant PD not suitable for DBS in the subthalamic nucleus (STN). In the last decade, several studies have shown promising results on tremor from DBS in the posterior subthalamic area (PSA), including the caudal zona incerta (cZi) located posteromedial to the STN. The aim of this study was to evaluate the long-term effect of unilateral cZi/PSA-DBS in patients with tremor-dominant PD. Methods: Thirteen patients with PD with medically refractory tremor were included. The patients were evaluated using the motor part of the Unified Parkinson Disease Rating Scale (UPDRS) off/on medication before surgery and off/on medication and stimulation 1–2 years (short-term) after surgery and at a minimum of 3 years after surgery (long-term). Results: At short-term follow-up, DBS improved contralateral tremor by 88% in the off-medication state. This improvement persisted after a mean of 62 months. Contralateral bradykinesia was improved by 40% at short-term and 20% at long-term follow-up, and the total UPDRS-III by 33% at short-term and by 22% at long-term follow-up with stimulation alone. Conclusions: Unilateral cZi/PSA-DBS seems to remain an effective treatment for patients with severe Parkinsonian tremor several years after surgery. There was also a modest improvement on bradykinesia.


Introduction
Parkinson's disease (PD) is a neurodegenerative disorder of the basal ganglia, resulting in motor and non-motor symptoms.One of the hallmark features of PD is tremor, typically resting tremor, which can debilitate the patients' autonomy and have a significant negative impact on their quality of life [1].Symptomatic relief can be achieved with oral medication; however, with disease progression comes a decrease in efficacy as well as drug-related side effects [2].Between a fourth and up to half of all PD patients have tremor as the dominant symptom, and tremor is especially difficult to control with dopaminergic drugs [3][4][5].Functional stereotactic neurosurgery with lesions, or deep brain stimulation (DBS), is an established treatment for PD when pharmacological therapy alone fails to provide sufficient relief [2].
During the lesional era, thalamotomy became the procedure of choice for PD, mainly due to its effect on tremor [6].Thalamotomies naturally led to an exploration of the underlying subthalamic area, and by the late 1970s, thousands of patients had undergone lesional surgery in an area, distinct from the subthalamic nucleus (STN), known as the posterior subthalamic area (PSA) [7].The PSA contain structures such as the caudal zona incerta (cZi) and cerebello-and pallidothalamic fiber tracts (CTT and PTT).Today, lesional surgery has mostly been replaced with DBS and the STN is the most common target for PD.Other brain regions, such as the globus pallidus internus or the ventral intermediate nucleus (Vim) of the thalamus are preferred targets in some patients due to their symptomatic profile of PD or other patient-specific considerations [8,9].
The Vim is the most common target for tremor, but the PSA has gained increasing interest as an alternative for various tremor disorders, including PD tremor [10].DBS in the cZi/PSA has been found to be more efficient than Vim-DBS for essential tremor, and contrary to Vim-DBS, cZi/PSA-DBS seems to have an effect on other parkinsonian symptoms such as akinesia or rigidity [11][12][13].

Materials and Methods
The present study is a retrospective review of patients with idiopathic PD with tremor as a dominant symptom, treated with unilateral cZi/PSA-DBS and followed up for at least 3 years.The cZi/PSA has in our department replaced the Vim and is our preferred target for patients with tremor-dominant PD deemed not suitable for bilateral STN-DBS due to old age, symptoms nonresponsive to levodopa (L-Dopa), moderate cognitive impairment, and other risk factors for poor outcome.In total, unilateral cZi/PSA-DBS was chosen in 35% of the patients undergoing DBS for PD at our department during the period in question.Eighteen patients were identified, 14 of whom had had the short-term follow-up results published previously [18].Five of the 18 patients were excluded from the current study: two had died of causes not related to surgery (pneumonia and traumatic intracerebral hemorrhage) before undergoing long-term follow-up, two had developed dementia and were unable to comply during evaluation, and one had failed to complete the short-term followup.The remaining 13 patients were included in the study.The diagnosis of idiopathic PD was established by a senior movement disorders neurologist according to criteria of the UK Parkinson's Disease Brain Bank [25].

Patient Characteristics
Patient characteristics are displayed in detail in Table 1.Ten patients had L-Dopa responsive symptoms, defined as >30% improvement of the Unified Parkinson Disease Scale, Part III (UPDRS-III) total score on 1.5× their morning dose (maximum 300 mg L-Dopa equivalent dose).One patient who underwent bilateral cZi/PSA-DBS was included in the current study after one electrode was removed due to a suspected infection 3 weeks after surgery, as described in a previous paper [26].

Evaluation
Patients were evaluated at baseline before surgery, at short-term follow-up 1-2 years after surgery, and at long-term follow-up at least 3 years after surgery.The specific follow-up times of each patient with corresponding improvement are shown in Table 2. Patients were evaluated on/off medication and on/off stimulation using the motor part of the Unified Parkinson's Disease Rating Scale (UPDRS-III).Medication was withheld 12 h prior to evaluation, and evaluation off-stimulation was conducted after the pulse generator had been switched off for at least 60 min.Adjustments to the stimulation parameters were made after evaluation.The daily levodopa equivalent dose (LEDD) was calculated at baseline and at the time-point for each follow-up [27].

Surgical Technique
The surgical technique has been described in detail previously [18].The procedures were frame-based stereotactic implantations of quadripolar electrodes (DBS 3387 or 3389, Medtronic, Minneapolis, MN, USA) in the cZi/PSA using the Leksell frame model G (Elekta Instruments, Linköping, Sweden).Targeting was performed with a visual anatomical approach using stereotactic T2weighted axial MR images and guidelines, as shown in Figure 1.The target cZi was defined anatomically as lying posteromedial to the visualized posterior tail of the STN on the axial scan showing the maximal diameter of the red nucleus (RN) (Fig. 1).A detailed step-by-step explanation of the targeting method can be found online [28].Ten procedures were done with the patient awake with intraoperative evaluation using macrostimulation, and three under general anesthesia.Microelectrode recording was not used.Expressed in means ± SD (range).

Contact Location and Electric Field Simulations
A postoperative CT scan was merged with the preoperative T1and T2-weighted MR images using StealthStation™ Cranial software (©Medtronic) for each patient.The merged images were transferred to SureTune™ software (©Medtronic), where the lead was modeled into each merge based on the CT artifact.The simulated fields of electric stimulation, called volume of tissue activation (VTAs), were created for each patient using SureTune™, based on a finite element method as described by Astrom et al. [29].Clinical chronic stimulation settings at the time of short-term and long-term follow-up, respectively, were used for the creation of each VTA.Homogeneous tissue in the form of randomly oriented straight long axons, 2.5 µm in diameter, was assumed (isotropic voxels).The VTAs were converted into high-resolution binary images.The Yelnik-Bardinet atlas was transformed to match the MRI-visualized RN and the STN in each patient, with a specific focus on the posteromedial border of the STN and the lateral border of the RN [30].The merge and atlas-fit were meticulously examined and approved by two examiners (RSP and MR).
The Yelnik-Bardinet atlas space was used as a common 3D anatomical domain.The atlas-to-patient transforms were reversed and applied to the anatomical T2 MR images and synthetic VTA images.To verify a satisfactory transform, the individual STN and RN delineations were visually compared with the domain images.The active contact (cathode) used in each patient were inserted into the domain via the lead models.
The VTAs were aggregated in the anatomical domain as stimulation maps.Visualization of the stimulation maps was done with a threshold of 0.7, which is at least 4 overlapping VTAs, to account for outliers.The result was visually analyzed in the 3D space   Long-Term Follow-Up of Unilateral cZi-DBS for Parkinsonian Tremor in relation to STN and the RN (Fig. 2) as well as on axial T2-slices with the histological borders of the STN and RN superimposed and then compared to two commonly known stereotactic atlases: Schaltenbrand-Wahren's atlas and Morel's atlas (Figure 3) [31,32].

Statistical Analysis
Friedman's nonparametric test was used for statistical evaluation of discrete values and the Wilcoxon signed rank test was used as a post hoc analysis.Analysis of variance for repeated measurements was used for continuous variables with the Bonferroni correction method as a post hoc test.A p value ≤0.05 was considered statistically significant.Data are presented as mean ± SD (min-max) for continuous variables and median (interquartile range) otherwise.

Motor Symptoms
Patients were evaluated at short-term follow-up at a mean ± SD of 14 ± 5 (12-28) months after surgery and at long-term follow-up at a mean of 62 ± 17 (36-99) months after surgery.UPDRS-III sub-scores on-/off-medication/ stimulation are presented in Table 3.There were no significant changes in off-medication/off-stimulation scores between short-and long-term follow-up.Scores below are presented as off-medication unless otherwise stated.
Median total UPDRS-III score was reduced from 42 off stimulation to 28 (33%) with unilateral stimulation at short-term and from 45 to 35 (22%) at long-term follow-up in comparison to off stimulation at the same follow-ups (p ≤ 0.001).The corresponding improvement for contralateral UPDRS-III were 50% and 43%, respectively (p ≤ 0.001).Contralateral tremor (CL.tremor, UPDRS-III items 20-21) was improved by 88% with stimulation alone both at short-term and long-term follow-up (p ≤ 0.001).Stimulation and medication combined completely abolished CL. tremor in 7 out of 13 patients at long-term follow-up, and only 1 out of 13 patients had a slight residual resting tremor (1 point on UPDRS-III item 20b).
CL. bradykinesia was reduced by 40% from 10 to 6 at short term and by 20% from 10 to 8 at long term (p ≤ 0.005).CL. rigidity was reduced by 33% at short-term evaluation but did not reach statistical significance after Bonferroni correction (p ≤ 0.05) and was not reduced at long-term follow-up.

Stimulation Parameters
Stimulation parameters and mean contact location according to the midcommissural point are presented in Table 4. Individual coordinates of the active contacts can be found as online supplementary material S1 (for all online suppl.material, see https://doi.org/10.1159/000533793).location over time.Three patients had bipolar settings at short-term follow-up, and 6 patients had bipolar settings at long-term follow-up.

Electric Field Simulations
The location of the active cathodes and 3D electric field simulations are visualized in Figure 2. Visualization of the VTAs showed that most of the electric field is contained within the subthalamic area in between the STN and the RN (Fig. 3).The field extends along the common trajectory of the lead and is concentrated just inferior to the thalamus.The contacts and simulation fields were virtually completely overlapping between the short-term and long-term follow-ups.

Medication and Complications
The mean LEDD did not significantly change from baseline to short-term follow-up but was increased by a mean of 337 mg ± 88 mg from short-term to long-term follow-up (p ≤ 0.01).No hardware-related complications occurred during the long-term follow-up.In 4 patients, the stimulation parameters were adjusted during the long-term follow-up due to affection of gait.An example of gait disturbance was a slight drag of the contralateral foot.Adjustments were different for each patient but could involve changing to a bipolar setting.The implantable pulse generator was replaced a total of 5 times in two patients after a mean time of 38 ± 21 months due to battery depletion.

Discussion
This study evaluated 13 patients treated with unilateral DBS targeting the cZi for medically refractory PD tremor after a mean of 5 years, with the longest follow-up being 8 years after surgery.Contralateral tremor was significantly improved, and the effect was sustained at the final evaluation.There was a modest improvement on contralateral rigidity and bradykinesia 1 year after surgery, but the improvement only remained significant for bradykinesia over time.Visualization of the electric field simulations showed that these effects are probably achieved by stimulation of structures, such as cZi and CTT, in the PSA, not the STN, and that the stimulation did not need to be significantly adjusted over time.

The Target
The PSA is a heterogeneous area and has been described in detail previously [7].The PSA is located just inferior to the ventral thalamic nuclei (VL) with its principal components being the zona incerta and prelemniscal radiations (Raprl) neighboring the fields of Forel (Fig. 3).This small region contains CTT and PTT involved in the cerebellothalamicand the basal ganglia-thalamo-cortical networks [34].The nomenclature varies between studies and atlases but in essence the CTT, also called dentato-rubro-thalamictract, corresponds to fasciculus cerebellothalamicus in Morel's atlas and in large part to the Raprl in Schaltenbrand-Wahren atlas.The PTT is more complex in its course and is divided into smaller bundles; fasciculus lenticularis (Morel) corresponds to H1 and ansa lenticularis to H2 and they merge to form the fasciculus thalamicus.The zona incerta itself is an extension of the reticular nucleus running above and then medial to the medial border of the STN and terminating in the socalled Q area, where it is named cZi (Fig. 3, far right panel).It has wide-spread connections throughout the central nervous system including the sensorimotor part of the cortex, globus pallidus internus, cerebellum, and the posterior VL thalamus (corresponding to Vim) [35].
While the traditional pathophysiological model of PD is mainly based on nigrostriatal dopamine depletion and its effects on the basal ganglia, PD tremor as a symptom seems to have a different basis [36].Although PD tremor is alleviated by dopaminergic therapy, possibly by acting on the cerebellar thalamus (posterior VL thalamus) and the pallidum [37], it has a more unpredictable response than rigidity and bradykinesia.There is also increasing evidence that other neurotransmitters, serotonin and noradrenaline, plays an important role in PD tremor.During the last decade, new models of PD tremor have been proposed with evidence mostly supporting the involvement of two circuits: the basal ganglia and the cerebello-thalamo-cortical motor loop (CTC).Helmich et al. [37] have proposed a dimmer-switch hypothesis with basal ganglia as a switch to initiate PD tremor and the CTC as a dimmer to modulate tremor amplitude, which would explain why different DBS targets in these two separate networks, STN/Vim/PSA, are able to treat PD tremor.
The optimal target in the PSA has not been established but the mean coordinates used for chronic stimulation in this study were comparable to those of other groups [10] and to the area with optimal tremor reduction identified by Fytagoridis et al. [38] in 50 patients with ET.As shown in Figure 2, the contacts used for stimulation are distinctly separated from the STN and concentrate below the VL thalamus.
Terminology and anatomy are possible confounding factors regarding PSA-DBS.In this study, the treatment is called cZi-DBS because it reflects the anatomical structure used during targeting and corresponds to an area with clear anatomical reference.Simply advancing through a Long-Term Follow-Up of Unilateral cZi-DBS for Parkinsonian Tremor given trajectory, the electrode targeting the Vim a few mm deeper during surgery may often place it in the PSA.Several studies have found that what is called Vim-DBS is often PSA-DBS [39][40][41][42][43].It is likely that the major effect of Vim-DBS is achieved through stimulation of the cerebellothalamic fibers.These will pass through a "funnel" in the PSA, before entering and dispersing into the VL thalamus.Hence, more of these fibers might be encompassed by the field of stimulation in the PSA, than after they have dispersed in the Vim [44,45].
Prior to the advent of modern DBS, lesioning in the PSA with the involvement of the CTT was known to alleviate tremor [7], and the CTT itself has during the last decade gained increased interest as a specific target for DBS with the advancements of tractography imaging [46].In our study, the simulated electric fields are plausibly concentrated around and extend along these fibers and would thereby modulate pathological activity in the cerebello-thalamocortical-motor loop.On the other hand, Plaha et al. [19] have proposed that the zona incerta itself acts as the transmitter and modulator of oscillatory activity through its connections with every part of the circuitry.Be it as it may, the PSA is an important anatomical node where modulation of pathological neuronal activity, either through acting on zona incerta or white matter tracts, can be achieved.
Concerning the STN, a systematic review as well as recent studies evaluating its spatial topography have shown that contacts used for chronic stimulation in the "STN" are often located in the border zone, or outside the STN, in the zona incerta [47][48][49].However, these are predominantly in the antero-dorsal part overlying the STN and not in the PSA itself.

Efficacy of PSA-DBS
The effect on tremor compares well with other studies on PSA-DBS.However, there is a discrepancy concerning the effects on other PD symptoms.Other groups have reported alleviation of tremor by 75-93%, rigidity by 45-94%, and bradykinesia by 46-75% at short-term follow-up ranging from 6 to 24 months after surgery [16,[19][20][21][22][23][24].To our knowledge, only one previous study has evaluated the long-term effect of PSA-DBS for PD [15].Eight patients with unilateral Raprl-DBS were followed for 48 months.The median CL.Tremor score was reduced from 5.9 at baseline to 2.2 at last follow-up, which equals 63% tremor reduction.Half of the long-term patients were classified as having excellent effect on tremor (90-100%), while half had sub-optimal effect on tremor (33-75%) [15].One reason behind the modest effect on non-tremulous symptoms in our material would seem to be due to a more concentrated stimulation on the CTT instead of the PTT which transmits output from the basal ganglia to the thalamus (Fig. 3).However, it is important to note that the patient-specific VTAs are a rough estimation of the electric field and its effect on the surrounding tissue.The simulations are based on probability of activation of an axon and visualizations are often based on an isolevel corresponding to a small percentage of neurons activated at that border [50].
Therefore, other factors such as patient selection, interindividual location of fiber tracts, or stimulation parameters might also account for the difference compared to previous studies.Notably, contralateral rigidity was not significantly reduced by L-Dopa alone at long-term follow-up in our cohort (Table 3).
While there has been no comparative study between PSA-DBS and VIM-DBS for PD, there are some studies on the long-term effect (4-6.6 years) of unilateral Vim-DBS on PD tremor.These found a tremor reduction between 55 and 82%, but with mostly no significant effect on akinesia and rigidity [33,[51][52][53][54].The largest study by Hariz et al. [51] followed 30 PD patients with Vim-DBS, 6.6 years after surgery.The mean improvement of CL. tremor was 82%, CL. rigidity 20%, and CL.akinesia 17%.The authors speculated that the modest effect on nontremulous items could be an evaluation artefact.Further, it is of interest to note that the patients in that multicenter study showed a significant spontaneous reduction of tremor over time, as is sometimes the case in patients with PD.The tremor score in the off condition was 8.4 at baseline before surgery and 3.3 at last follow-up.In contrast, the tremor scores in our study remained high in the off condition even at long-term follow-up.It is possible that comparing the effect of DBS with the offstimulation condition at the same time, rather than with a preoperative baseline, might be more representative of Long-Term Follow-Up of Unilateral cZi-DBS for Parkinsonian Tremor the actual effect of stimulation.Regarding unilateral STN-DBS for PD tremor, there is a scarcity of data in the literature.Furthermore, only a few studies have specified the improvement of the different motor symptoms following unilateral STN-DBS.The follow-up was usually short, between 3 and 23 months, and the preoperative tremor score was modest (mean 2.3 points) [55][56][57][58].Nevertheless, the improvement was 72-86%, 22-46%, and 39-66% on tremor, bradykinesia, and rigidity, respectively.Recently, Liang et al. [59] reported on the long-term effect of unilateral STN-DBS in 23 patients with asymmetrical PD.Five years after surgery, they noted a decrease of CL. tremor by 82%, with a nonsignificant effect on action tremor, rigidity by 81%, and bradykinesia by 58%.As the current study is limited to cZi/PSA-DBS, any comparisons of the results to other targets such as VIM and STN are of course speculative.A prospective comparative trial like those conducted regarding VIM versus PSA-DBS for ET, would be needed to distinguish superiority between targets, especially with regards to possible loss of efficacy on tremor over time as well as effects on non-tremor symptoms such as bradykinesia [11,12].

Medication
There was no change of LEDD at short term, but an increase at long term, when compared to the preoperative baseline.During the same time, total off-medication UPDRS-III worsened, mainly concerning ipsilateral symptoms.Therefore, the increase in medication can at least partly be attributed to worsening of ipsilateral symptoms.The change in this study was in the same range as seen in most of the long-term studies on unilateral Vim-DBS for PD tremor, apart from the multicenter study by Hariz et al. [51] which reported no change [33,53,54].

Limitations
The retrospective, non-blinded manner as well as the small sample size of this study are the main limitations.On the other hand, as has been previously shown [60], results of a non-blinded evaluation are not necessarily different from those of a blinded evaluation, and the present patients were evaluated at all time-points by the same experienced rater [60].The small sample size increases the risk of type 2 errors, which could be contributing to, but probably not fully explaining, the modest effects on rigidity and bradykinesia compared to other PSA-DBS studies due to the difference in effect size.Another limitation was the variation in follow-up time after surgery.While almost all patients were evaluated around 1 year after surgery for the short-term follow-up, the long-term follow-up varied between 3 and 8 years after surgery.However, as seen in Table 2, there was no trend toward an inferior result among the patients evaluated later than 5 years compared to earlier than 5 years.The simulations used in SureTune™ have inherent limitations such as not accounting for differences in conductivity of different biological tissues (e.g., white matter vs. gray matter).However, it has been shown that for constant voltage stimulations, this difference does not affect the VTA to a significant degree [61].The small sample size and lack of monopolar review data further limit the possibility to reliably test the simulated fields for any significant correlations between specific voxels and differences in clinical outcome between the patients [62].The small sample size also increases the probability of significant outliers which was why a thresholding was done.In addition, the axon diameter used for the visualization of the VTAs are 2.5 µm, while the majority of axons in the brain are around 1 µm, which would translate into an overestimation of the volume of activated tissue [50].Further, the merge and modeling and transform of patient-specific imaging into a common domain comes with the risk of incremental deviation of contact location in relation to the anatomy.To mitigate this, we compared the contact location in the 3D domain to each patient's CT-MRI fusion.Finally, when comparing to commonly known stereotactic atlases, Morel's and Schaltenbrand-Wahren, there is a difference between slice thicknesses, angles, and distances to the AC-PC plane as well as nomenclature.For this reason, a side-by-side comparison at different depths highlighting the STN and RN was done as these are the two clearly delineated structures on standard T2-imaging commonly used for DBS targeting of the cZi/PSA.Unilateral cZi/PSA-DBS seems to be a safe and effective treatment for patients with severe Parkinsonian tremor even several years after surgery.The effect is pronounced regarding tremor, even if modest improvements are also seen regarding bradykinesia.Future comparative trials would be helpful in evaluating DBS of the cZi/PSA versus DBS in more established targets such as the VIM.
tremor LT ON-med ON-stim (score) ) = off versus on stimulation in the off-medication state in each patient.ON-med ON-stim = individual UPDRS-III rated score with both medication and stimulation at the latest follow-up.CL., contralateral; ST, short-term follow-up; LT, long-term follow-up.

Fig. 1 .
Fig.1.The target (orange marker) based on axial slices with guidelines.As the relation between the red nucleus (RN) and the posterior tail of the subthalamic nucleus (STN) are different between the atlases, as well as among patients, the target marker has been placed in a position which would be deemed as adequate lead placement.To the left is an T2 MR image, and to the right are slices from Schaltenbrand-Wahren's atlas and Morel's atlas.
Fig. 2. a-c Three-dimensional model of the contact location and aggregated stimulation maps in relation to the anatomy.Active cathodes/stimulation maps at short-term = light blue.Active cathodes/stimulation maps at long-term = dark blue.STN, subthalamic nucleus; RN, red nucleus.

Table 1 .
Patient characteristics

Table 2 .
Individual follow-up time after surgery with corresponding change in tremor score according to UPDRS-III

Table 3 .
Absolute scores for UPDRS-III itemsExpressed in medians (IQR, min-max).Off/off, off medication/off stimulation; off/on, off medication/on stimulation; on/off, on medication/off stimulation; on/on, on medication/on stimulation; CL, contralateral; IQR, interquartile range.a p ≤ 0.005 versus off/off at the same follow-up.b n = 11.

Table 4 .
Stimulation parameters and mean location of the contact acting as active cathode Expressed as mean ± SD.X = lateral to the midline.Y = posterior to the midcommissural point (MCP).Z = inferior to the intercommissural line (ICL).PEV, pulse effective voltage.# Number of patients with corresponding settings.