Deep Brain Stimulation in Essential Tremor: Targets, Technology, and a Comprehensive Review of Clinical Outcomes

Abstract

Introduction:

Essential tremor (ET) is a common movement disorder with an estimated prevalence of 0.9% worldwide. Deep brain stimulation (DBS) is an established therapy for medication refractory and debilitating tremor. With the arrival of next generation technology, the implementation and delivery of DBS has been rapidly evolving. This review will highlight the current applications and constraints for DBS in ET.

Areas covered:

The mechanism of action, targets for neuromodulation, next generation guidance techniques, symptom specific applications, and long-term efficacy will be reviewed.

Expert opinion:

The posterior subthalamic area and zona incerta are alternative targets to thalamic DBS in ET. However, they may be associated with additional stimulation induced side effects. Novel stimulation paradigms and segmented electrodes provide innovative approaches to DBS programming and stimulation induced side effects.

1. Introduction

Essential tremor (ET) is one of the most common movement disorders with an estimated prevalence of 0.9% worldwide [1]. It is classically characterized as a 4 – 12 Hz kinetic tremor of the upper extremities but can also spread to other body regions such as the head, jaw, and lower extremities [2]. The most recent consensus statement from the Movement Disorder Society (MDS) defines ET as an isolated tremor syndrome consisting of bilateral upper extremity action tremor of at least 3 years duration, with or without tremor in other locations and without other neurological signs [3]. ET is typically insidiously progressive, with gradual worsening of tremor severity over time. Natural history studies suggest that arm tremor worsens anywhere from 2-5% per year [4].

Pharmacological therapy has long been the mainstay of treatment for ET [5]. The two medications with the highest level of evidence to treat ET are propranolol and primidone [6]. Propranolol and primidone provide approximately a 55% and 60% mean reduction in tremor amplitude, respectively, when used as monotherapy [5]. When used together, the combination allowed patients to achieve up to 70% tremor reduction [7]. Unfortunately, pharmacologic therapy is only effective in 50% of ET patients [8]. Furthermore, nearly half of the patients who do experience benefit from medications eventually stop therapy due to limited efficacy or dose dependent side effects [9]. These obstacles have driven a search for alternative therapies for debilitating tremor.

Neurosurgical ablative procedures have provided successful symptomatic relief from medically refractory tremor; however, bilateral thalamotomies were associated with significant side effects, such as dysarthria, dysphagia, cognitive impairment and gait and balance difficulties [10–12]. Deep brain stimulation (DBS) emerged as a promising alternative in 1991 for ET. It offered potent tremor suppression while avoiding the common complications of thalamotomies [13]. DBS was also found to have a favorable side effect profile compared to unilateral thermolesion thalamotomies (8% vs 37% respectively) and unilateral focused ultrasound thalamotomies (14% vs 34% respectively) [14,15].

The patient selection process is a critically important step in ensuring optimal post-surgical outcomes. Although DBS has been applied in complex tremors such as multiple sclerosis and dystonic tremor, post-operative tremor suppression can be more variable in these cases [16–18]. Additionally, proximal tremors can also be difficult to address with neuromodulation [19,20]. As such, differentiation of ET from other tremor syndromes is a crucial component in the pre-operative assessment of DBS candidacy. The MDS consensus criteria, as mentioned previously, can be a helpful guideline in directing a patient through the DBS evaluation process. Since the adoption of DBS for ET, research into ET neuromodulation has rapidly expanded as summarized in Figure 1. We highlight the current benefits and limitations of DBS for ET in this comprehensive review, and we also discuss new technologies.

Figure 1. Publications of DBS for ET over time.

Publication metrics obtained from PubMed utilizing the search strategy previously described.

1.1. Search strategy

Our search strategy for this review is provided as follows:

1. PubMed search terms:

   • (“essential tremor”[MeSH Terms] OR (“essential”[All Fields] AND “tremor”[All Fields]) OR “essential tremor”[All Fields]) AND (“deep brain stimulation”[MeSH Terms] OR (“deep”[All Fields] AND “brain”[All Fields] AND “stimulation”[All Fields]) OR “deep brain stimulation”[All Fields])
   • (“tremor”[MeSH Terms] OR “tremor”[All Fields]) AND (“deep brain stimulation”[MeSH Terms] OR (“deep”[All Fields] AND “brain”[All Fields] AND “stimulation”[All Fields]) OR “deep brain stimulation”[All Fields])

2. Inclusion criteria: articles must focus primarily on deep brain stimulation for essential tremor

3. Additional search methods: Further relevant articles were identified from the reference list of reviewed articles.

2. Mechanism of action

Our understanding of the principles of neuromodulation are constantly changing. We highlight the major hypotheses that cover the mechanism of action in DBS.

Although the mechanism of action of DBS remains incompletely understood, several hypotheses have been proposed [21,22]. Initially, DBS was thought to have an inhibitory effect based on high frequency stimulation (HFS) of the subthalamic nucleus (STN) in Parkinson’s disease (PD) animal models [23]. With HFS, scientists were able to reduce rigidity and bradykinesia of PD animal models similar to STN lesioning and to gene therapy experiments [24,25]. This led to the hypothesis that DBS induced a jamming of the sensory-motor circuitry in the thalamus and inhibited neurons in the STN [26]. Subsequent electrophysiology studies found that HFS in STN and globus pallidus interna (GPi) DBS inhibited neighboring neurons in addition to suppressing a single target [27].

As technology developed, DBS research began to incorporate computational modeling to explore mechanistic hypotheses. Biophysically accurate models of individual neurons (e.g., GPi or globus pallidus externa (GPe) projection neurons, thalamocortical relay neurons) coupled with simulation of external electric fields produced by DBS were used to assess excitation or inhibition [28–31]. Grill et al. found that suprathreshold stimulus levels could activate neuron somas, dendrites, and axons and elicit action potentials at the stimulation frequency [31]. When high frequency stimulation was used, underlying burst or tonic firing that mimicked the pathology could be suppressed. This led to the theory that DBS regulates the malfunctioning basal ganglia by minimizing pathologic communication, a concept described as an “informational lesion.” Results from McIntyre et al. support those results and also showed that somal suppression could be achieved through DBS with suprathreshold and subthreshold stimulation and was dependent on the extent of afferent and efferent activation [28–30]. Model results suggest complex interactions at the neuron level and DBS effects at the neural circuit level are still unclear. ET is largely underpinned by an abnormal physiological signal in the cerebello-thalamocortical loop, and in some positive way DBS modulates this dysfunction.

The complex interactions within the basal ganglia have been digitized into an electronic model and tested under various DBS scenarios [28–30]. These simulations suggested that GPi neuromodulation was associated with local neuronal activation. Additionally, stimulation at subthreshold parameters led to neuronal suppression, but suprathreshold stimulation produced action potentials that were detected in downstream axons, while maintaining suppression of the neuron cell bodies.

According to a recent review of the mechanisms of action of DBS, there is a combination of factors that likely contribute to the clinical outcome [32]. One salient difficulty has been distinguishing biological changes from actual mechanism of action. DBS induces changes in neurophysiology, neurochemistry, neurovascular compartments, and even local and upstream neurogenic changes. Additionally, DBS is known to influence neural oscillations, and this change may underpin much of its effect [33,34]. Finally, DBS for ET, in particular, has been potentially linked to induced changes in the neurotransmitter adenosine [35].

3. Existing targets of neuromodulation

DBS for ET began with stimulation of the ventralis intermedius nucleus (VIM) of the thalamus in 6 patients [13]. The ability to obtain robust tremor suppression with relatively small lesions and the ability to avoid the complications seen in bilateral thalamotomies fostered strong support as an alternative therapeutic option [10–12,36]. Studies examining long term clinical efficacy of VIM DBS for ET are summarized in Table 1. As physicians adopted VIM DBS for use in tremor, localization studies began emerging to find the ideal stimulation site. Two postmortem case reports observed that significant tremor suppression had been obtained with DBS leads positioned in the ventral border of the VIM [37,38]. This led to the hypothesis that perhaps stimulation of the cerebello-thalamic fibers led to tremor relief. A subsequent study of VIM DBS analyzed the stereotactic coordinates of the DBS lead in a retrospective fashion and found that distal contacts were more effective than proximal contacts [39]. The authors proposed that the distal contacts were stimulating the subthalamic area (STA), which contained prelemniscal radiations, and the zona incerta, and modulation of these fibers provided more tremor suppression than the ventro-lateral thalamus. Thereafter, a study with dedicated targeting of the posterior subthalamic area observed that patients experienced 60% improvement in tremor at 1 year post DBS implantation [40]. This finding spurred several endeavors to investigate alternative targets for ET DBS to identify the optimal site of neuromodulation.

Table 1.

Long term studies of VIM DBS for ET

Study	Patients (n*)	Age (years)	Follow up (months)	Tremor improvement (%)
Koller et al. [106]	25 / 0	72 ± 9	40 ± 15	50
Sydow et al. [107]	12 / 7	62 ± 11	80 ± 7	41
Putzke et al. [108]	29 / 23	72 ± 8	36	87
Pahwa et al. [89]	15 / 7	71 ± 5	60	54
Blomstedt et al. [109]	18 / 1	68 ± 7	86	32
Zhang et al. [76]	23 / 11	58 ± 13	57	80
Rodriguez Cruz et al. [82]	3 / 11	61 ± 3	92 ± 46	50
Cury et al. [110]	0 / 12	64 ± 11	156 ± 33	48
Paschen et al. [78]	0 / 20	67 ± 2	72 ± 7	23

^*Unilateral DBS / bilateral DBS. Age and follow up duration are reported as mean ± SD if the data was available. Tremor improvement is reported as the percent change in the respective clinical scale used in each study.

4. Future targets of neuromodulation

The search for the optimal target for neuromodulation in ET is an extensively studied topic. Here we review the latest studies that have compared the emerging regions of interest for DBS. A summary of ET DBS targets is illustrated in Figure 2.

Figure 2. Evolution of DBS targeting in essential tremor.

Row A illustrates oblique axial, coronal and sagittal views of MRI based targeting for DBS. Row B demonstrates atlas-based targeting strategies also in oblique axial, coronal and sagittal views. The white circle represents an approximation for the DBS lead tip. Row C shows 3D-reconstruction-based targeting methods centered around the thalamus.

4.1. Posterior subthalamic area

Comparisons between the VIM and STA for ET has been an ongoing deliberation. One study sought to characterize the side effect profile in eight patients who received STA DBS for ET [41]. By using suprathreshold stimulation, defined as input current of 0.5 mA above stimulation induced ataxia, the authors found that the most ventral contacts were generating the strongest stimulation induced ataxia. They found a linear correlation of ataxia with ventral electrode positions 0–4 mm below the middle commissural point (MCP). They also found that stimulation of more ventral positions (4–6mm) did not further increase the ataxia. Stimulation of the antero-dorso-medial STA region was associated with less tremor suppression but also less ataxia. The authors proposed that these findings suggested that stimulation induced ataxia involves ventrally located cerebello-rubro-spinal fiber tracts.

A similar study conducted a double-blind crossover trial comparing VIM DBS with the posterior subthalamic area (PSA) DBS [42]. The study utilized a lead trajectory that traversed both the VIM and PSA. Each lead would have one dorsal contact in the VIM, one neutral contact in the intercommissural line (ICL), and one ventral contact in the PSA. Previous studies had observed that for similar programming, contacts below the ICL provided better tremor reduction that those above the ICL [39,43]. Patients were programmed with monopolar stimulation at the neutral contact for the first 3 months post DBS implantation. Patients were then randomized to either the VIM contact or PSA contact with monopolar stimulation for 2 months. After 2 months, the patients were crossed over to the other contact for another 2 months. Then, the patients and physicians could freely program based on clinical judgement until the one-year post DBS implantation follow-up visit. This study found that there was no statistical difference in tremor suppression between the VIM and PSA, however there was a trend towards better tremor suppression with PSA stimulation. The stimulation induced side effect profile between the 2 targets were comparable. At the end of the 12-month follow-up, 64% of the patients were using the active contact located in the PSA, 21% of patients were using active contact in the VIM, 7% of patients were using active contact in the ICL and the remaining patients had complex programming settings. Analysis of the programming settings revealed that PSA DBS required less current than VIM DBS for tremor suppression (4.35 mA vs 5.88 mA, p = 0.006). One notable issue with this study was the high current utilized, raising the question of the actual location of the DBS leads relative to the ventralis caudalis nucleus of the thalamus. This may have been a slightly different target than other groups have commonly used.

Another study comparing VIM DBS to PSA DBS for ET tremor suppression employed a retrospective analysis of 19 patients based on the active contact used [44]. Similar to previous studies, the authors defined VIM stimulation as having an active contact greater than 3 mm above the mid commissural point (MCP), intermediate stimulation as 0–3 mm above the MCP, and PSA stimulation as below the MCP. After a mean follow-up time of 19 months, patients with VIM DBS experienced a 63% tremor reduction. Patients with PSA stimulation achieved 47% tremor reduction and intermediate stimulation achieved 67% tremor reduction (p = 0.059). After 70 months of follow-up, the degree of tremor reduction compared to preoperative baseline was 50%, 34%, and 45% for the same three patient cohorts respectively. When examining the programming settings, the authors did not find any significant differences between the mean voltage or pulse width. Analysis of the side effect profile revealed 75% of patients experienced dysarthria, 37% ataxia and disequilibrium, and 11% paresthesias. Aside from dysarthria, this was similar to side effect profiles seen in previous studies [45–47]. The authors observed a significantly higher rate of dysarthria but also acknowledged that their testing protocol did not allow them to test with the DBS turned off. Therefore, they could not confirm if the dysarthria was related to disease progression or to a stimulation side effect.

4.2. Zona incerta

The zona incerta (ZI) is another alternative target that has been used for ET DBS. One study sought to compare differences in zona incerta (ZI) DBS versus VIM DBS in a retrospective review of 47 DBS patients [48]. After DBS implantation, a delayed post-operative head CT was fused to the pre-operative targeting MRI for 3-D lead localization. These images were then further fused with a patient-specific, deformed 3-D anatomic atlas via the Schaltenbrand-Bailey Sudhyadhom technique [49]. The active DBS contact was identified in 3-D space with respect to the anterior commissure - posterior commissure (AC-PC) line. Active contacts above the AC-PC line were categorized as VIM whereas contacts below the AC-PC line were categorized as caudal zona incerta (cZI). The authors found that short-term benefits (at 6 months and 2 years) compared to baseline were similar between the VIM and cZI. However, at 3- and 4-years post DBS implantation, VIM DBS provided better tremor suppression (p < 0.01). Analysis of the trend over time also revealed that Fahn-Tolosa-Marin Tremor Rating Scale (TRS) scores in VIM DBS trended down over time whereas TRS scores trended up in cZI DBS (tau = 0.26, p < 0.01) [50]. The authors also observed that VIM DBS required lower voltage settings to obtain tremor suppression at 3 years post DBS implantation (2.55v vs 3.17v, p < 0.001). There were no significant differences between the other programming parameters.

Another retrospective study comparing the zona incerta (ZI) versus the VIM for ET DBS contradicted these findings [51]. The study analyzed 18 patients who underwent ZI DBS and 10 patients who underwent VIM DBS. Target localization was categorized per the local medical center DBS protocol. Specifically, all ET DBS cases performed before 2004 were designated as targeting the VIM whereas all cases after 2004 were designated as ZI. Primary outcome was the patient reported Washington Heights Inwood Genetic study of Essential Tremor (WHIGET) scale. This study found VIM DBS provided 32% reduction of tremor compared to 59% reduction with ZI DBS. The authors also observed that stimulation induced dysarthria was present in 75% of VIM DBS patients versus 39% of ZI DBS patients (p = 0.02). These results should be interpreted cautiously as there were several limitations acknowledged by the authors in this retrospective study. Firstly, the primary outcome was a patient reported metric rather than a clinical assessment. Additionally, the surgical targets were defined based on date of surgery and essentially represented the DBS protocol at that given time. This made confounding variables difficult to control for such as surgical technique and pre-operative planning.

5. Image guided targeting and programming

Advances in MRI technology have transformed its application and utility in DBS. We present several software and hardware-based concepts that have emerged to assist DBS planning and stereotactic neurosurgery.

Connectivity profiling has been increasingly used to assist with DBS targeting [52]. One study created connectivity profiles using structural and functional connectivity data of a normative connectome database to predict “optimal connectivity.” Structural connectivity profiles were created by utilizing MRI diffusion tensor imaging (DTI) data and reconstructing approximations of axonal connections using computational tractography algorithms. Functional connectivity profiles were created by analyzing the real-time fluctuations of oxygen simultaneously throughout the brain in the absence of external stimuli via a technique known as blood-oxygen-level dependent resting state fMRI (BOLD rs-fMRI) [53]. Connectivity data from the entire brain network was mapped into a statistical model and correlated with clinical outcomes. The statistical model was combined with volume of tissue activation (VTA) analysis to map tremor suppression with brain tissue and somatotopy between the hand and head. VTA analysis involves approximating the electric field from the DBS electrode based on the specific input of programming parameters [54]. This technique, when combined with 3D imaging data of the brain, can estimate the region of neuronal tissue stimulated by the DBS system. The authors identified optimal connectivity for tremor suppression at the region of the inferior-posterior border of the VIM and the dorsal border of the ZI.

While the use of normative connectome data is a widely utilized method in DBS research, it is yet to be fully determined whether these inferences represent the true network altered by DBS stimulation. Many factors, such as frequency and pulse width, may dictate the areas of the brain truly affected amongst those with estimated functional connectivity to the VTA. Gibson et al. explored the effects of active stimulation after VIM DBS using BOLD fMRI [55]. Numerous brain regions were shown to be activated with stimulation of the VIM that correlated with tremor reduction, including the contralateral cerebellum and deep cerebellar nuclei, supplementary motor area, sensorimotor cortex, thalamus, brainstem, and inferior frontal gyrus.

Structural connectivity has also been utilized to estimate the connectivity profile of the VTA. One study parcellated the thalamus into divisions based on the outgoing connections to various cortical regions [56]. This technique provided a novel method to visualize the thalamic sub-nuclei that are difficult to distinguish on conventional MRI sequences. Connectivity was analyzing into the following cortical targets: primary motor cortex, primary sensory cortex, supplemental motor area/premotor cortex, prefrontal cortex, occipital lobe, temporal lobe and parietal lobe. These structural connectivity profiles were combined with VTA analysis to identify the optimal DBS lead location with respect to clinical outcomes. The authors found that stimulation of the thalamic region with the strongest degree of connectivity to the supplemental motor area/premotor cortex was associated with the greatest degree of tremor suppression.

6. Technological advances

Innovations in commercially available DBS hardware technology are reshaping the field of movement disorders. The emerging developments are discussed below.

6.1. DBS programming

As biomedical technology evolves, the field of neuromodulation is rapidly becoming more complex. With the arrival of next generation DBS leads, electrodes and implantable pulse generators (IPG), the possible combinations for programming parameters are exponentially increasing. To address this issue, one study created customized software to predict the most effective DBS contact [57]. This study analyzed 33 patients who underwent ZI DBS. The software created probabilistic stimulation maps (PSM) that were correlated with clinical outcome data to estimate the most effective region of stimulation. This technique involved modeling of the VTA and estimating the final location of the DBS lead by fusing the post-operative head CT with a pre-operative MRI. The custom software created a prediction algorithm that ranked the contacts from 1 to 4, assigning 1 to the contact predicted to be most effective and 4 to the least effective. Rank 1 contacts matched with the empirically selected clinical contact 60% of the time. The top 2 ranked contacts matched the clinical contact 83% of the time. Although the prediction algorithm did not entirely agree with the clinical findings, such efforts to simplify DBS programming will be necessary to manage the increasing complexity of DBS systems and avoid intolerably long programming times. Manual monopolar review will become much more time-consuming as the complexity of DBS leads and contacts continues to increase.

As DBS programming configurations evolve alongside advancing hardware technology, a drive to explore novel stimulation waveforms has also surfaced. One study investigated the use of novel stimulation parameters for VIM DBS in ET [58]. Novel stimulation paradigms are thought to offer potentially improved battery utilization or to trigger different neuronal excitation patterns [59–62]. The trial was conducted as a safety, tolerability, and feasibility trial. The authors employed square biphasic pulses with an active symmetric charge balancing phase to 11 VIM DBS patients. Conventional DBS stimulation delivers an asymmetric biphasic pulse with a passive charge balancing phase [63]. Although the trial was designed as a pilot study, the authors observed effective tremor suppression with active biphasic stimulation. Previous studies have suggested that a combination of anodic and cathodic stimulation can have complex neuronal excitatory and inhibitory patterns [64–66]. Larger trials are needed to fully characterize the different stimulation patterns and their effects on battery usage, side effect profile and therapeutic window.

6.2. Modern hardware

Another study investigating novel stimulation paradigms sought to optimize programming efficiency and minimize the stimulation induced side effect profile in next generation DBS systems. This study compared the effects of directional current steering and short pulse stimulation on tremor severity and side effect profile [67]. 8 VIM DBS patients were tested under 3 stimulation conditions: 1) Off stimulation 2) suprathreshold stimulation with a pulse width of 60 μs 3) equivalent energy stimulation with a pulse width of 30 μs. The equivalent energy stimulation was calculated via the total energy delivered formula TED = [(voltage² × frequency × pulse width)/impedance] × 1 second [68]. Stimulation was applied to the most ventral contact to compare tremor suppression and stimulation induced ataxia. The authors found that at dose equivalent stimulation levels, both stimulation paradigms provided an equal degree of tremor suppression. The short pulse stimulation, however, induced less ataxia and paresthesia compared to conventional stimulation at 60 μs. At suprathreshold stimulation parameters, there were no differences in the rate of stimulation induced ataxia and paresthesia when providing stimulation through a segmented contact. This finding was also independent of segment orientation. The authors concluded that short pulse stimulation can provide a programming option to provide tremor suppression while decreasing stimulation induced side effects.

The rise of next generation DBS hardware has also introduced novel energy delivery paradigms. One new feature is the utilization of constant current stimulation. Traditional DBS systems utilize a constant voltage delivery method in which the delivered current can vary over time as a reflection of the changing impedance between the DBS lead and the surrounding brain tissue. Studies have investigated long-term changes of impedance over time, and while impedance variability is minimal it is not non-zero [69]. In one study, 10 patients with VIM DBS were implanted with the St. Jude Medical Infinity DBS system (St. Jude Medical Neuromodulation, Plano, TX, USA) and programmed with constant current stimulation [70]. Previous studies have investigated the use of constant current stimulation in Parkinson’s disease and dystonia but not ET [32,71]. This study sought to fill that gap in the literature. Patients were followed over a mean of 49.7 months. The authors shared their programming experiences and reported the following mean DBS parameters: current 2.5 mA (range 1.2-3.7 mA), pulse width 68 μs (range 65-91 μs), frequency 134 Hz (range 130-160 Hz). Clinically, they found a mean 77% improvement in TRS and 65.3% improvement in the TRS ADL total score. There were no comparisons made with constant voltage systems and the authors acknowledge that the follow-up time in this study was much shorter than many of the constant voltage system studies. Despite these weaknesses, however, the preliminary data in this study do suggest that constant current DBS systems provide effective tremor suppression in patients with ET.

With multiple DBS hardware systems now commercially available, movement disorders centers are also beginning to conduct comparison studies of the various technologies on the market. One study evaluated 7 VIM DBS patients using the St. Jude Medical Infinity DBS system (St. Jude Medical Neuromodulation, Plano, TX, USA) versus 7 VIM DBS patients using the Medtronic Neurostimulator system (Medtronic, Minneapolis, MN, USA) [72]. All patients in this study were evaluated 6 months post DBS implantation. The St. Jude system provided stimulation using constant current as opposed to constant voltage, meaning that the theoretical delivered electrical stimulation remains constant regardless of the electrophysiologic changes at the brain–tissue interface. Conventional Medtronic DBS systems provided constant voltage electrical stimulation, which is subject to theoretical “drift” as impedance of the brain tissue immediately adjacent to the DBS lead changes over time [73]. Long-term studies have found this is not the case [69]. The authors observed that the St. Jude system offered a wider therapeutic window but required higher equivalent voltages (3.06v vs 1.85v, p = 0.035). Among the 7 St. Jude patients with directional lead capability, 2 patients utilized the directional leads while the remaining 5 utilized conventional full ring stimulation. The study also found that both DBS systems provided a similar degree of tremor suppression. It should be noted, however, that the study was too small to draw firm conclusions.

A previous study evaluated 127 ET patients treated with VIM DBS using the St. Jude Libra constant current (non-directional) DBS system [74]. This was the largest ever prospective trial evaluating the safety and efficacy of VIM DBS for ET, while also investigating the safety and efficacy of constant current stimulation in ET. At 180 days post DBS implantation, comparison of off DBS versus on DBS revealed a mean improvement of 60% in TRS scores. When comparing on DBS to baseline, there was a 65% improvement from baseline in TRS scores. Characterization of the DBS related side effect profile revealed 77 adverse events occurred in 39 patients who underwent staged bilateral DBS. 14 adverse events were reported in 8 patients who underwent simultaneous bilateral DBS implantation. The authors acknowledged that this study was not designed to compare adverse event profiles and proposed follow-up studies to better characterize this difference.

7. Long term efficacy

Several studies have evaluated long-term outcomes of DBS for ET, and multiple potential explanations have been given for the diminished tremor suppression that is commonly observed over time following initially successful DBS for ET [75–77]. Two primary hypotheses have been proposed to explain a gradual loss of efficacy in ET DBS: disease progression versus habituation. One study evaluated 20 patients with bilateral VIM DBS with long-term follow-up [78]. Long-term follow-up time ranged from 32 months to 120 months with a mean follow-up of 71.9 months. TRS scores were evaluated at baseline prior to DBS surgery and in the on DBS state at approximately one-year post-op (“short-term”). These scores were compared to TRS scores measured at long-term follow up in the on and off DBS states. Off DBS scores were measured 1 hour after turning the DBS off to avoid a tremor rebound phenomenon. The authors observed progressive worsening of TRS over time in both the on and off DBS states. They also noted that the tremor worsening was more pronounced in the on DBS state, which they attributed to habituation to stimulation. In an attempt to quantify both disease progression and habituation, they calculated the change in TRS per month in both on and off DBS states over the course of their follow-up time. They found that compared to the pre-op off DBS state, the long-term off DBS TRS had worsened by 0.32 points per month. When compared to the 1-year post-op on DBS state the long-term DBS on TRS worsened by 0.37 points per month. The authors concluded that the 0.32 decrease per month was due to disease progression and the 0.05 difference seen while on DBS was due to habituation to DBS, concluding that 13% of the on DBS TRS score was attributable to habituation while 87% could be attributed to disease progression. It is noteworthy, however, that this study failed to account for the thalamotomy effect of the DBS implantation procedure that has repetitively been shown to result in tremor reduction and improvement in the pre-op baseline TRS score without DBS activation [79,80]. This thalamotomy effect might well explain the discrepancy in the measured rate of TRS worsening between the off and on DBS states without invoking habituation. A more valid comparison would have been to compare the long-term off DBS TRS score to a 1-year post-op “short-term” off DBS TRS score, as was done for the measurements in the on DBS state, but no short-term post-op off DBS TRS scores were obtained. The study also investigated the tremor rebound phenomenon by using a spiral score rating and electrophysiology. With regards to tremor rebound, the authors found that tremor amplitude reached its maximum value 30-60 minutes after turning the DBS off.

Favilla et. al. compared 28 patients who underwent VIM DBS for ET (19 unilateral, 9 bilateral) versus 21 age-matched controls with ET without DBS [75] Patients were followed for 36 months with serial examinations in the off DBS states. TRS scores were obtained after the DBS had been turned off for 30 minutes to minimize any rebound phenomenon. 7 DBS patients demonstrated worsening of TRS scores in the off DBS state. Within this subgroup, 6 DBS patients continued to experience motor benefit from neuromodulation. One patient experienced both progression of tremor in the off DBS state as well as loss of benefit from stimulation after 24 months follow up. Upon further analysis, the authors attributed this patient’s experience to suboptimal lead location. Overall, the authors concluded that these observations suggest disease progression was the most likely explanation for loss of DBS efficacy over time.

Another study conducted a randomized controlled trial of alternating stimulation versus conventional stimulation in 16 chronically implanted VIM DBS patients to further explore this topic [81]. The patients were randomized to 2 groups that each received 2 programming group settings (A and B). In the alternating stimulation group, A and B represented stimulation settings that differed by at least 2 of the following: electrode configuration, voltage, current, pulse width or frequency. In the conventional stimulation group, both programming groups A and B represented identical settings. The patients were instructed to turn off the DBS every night and to change the programming group (A to B or B to A) on a weekly basis. Outcome measures were reported as changes in the TETRAS score. After 12 weeks of testing, the authors observed that the standard stimulation group experienced a 6.7-point increase in TETRAS whereas the alternating stimulation group experienced a 0.6-point decrease in TETRAS. The authors concluded that the clinical worsening over the 12 weeks was due to DBS habituation, though the experiment was not designed in such a way to reach this conclusion.

A related study approached the same question from different perspective by using the clinical outcomes and programming parameters of 14 patients with an average of 7.7 years of follow-up to build a statistical model that characterized the long-term effects of VIM DBS in ET [82]. The authors found a mean reduction in TRS of 73.4% at 6 months post DBS implantation. The study tracked the change in TRS over time and noted that the progressive worsening in TRS scores fit a linear regression model. The authors then analyzed the programming parameters required to optimize tremor over time and calculated the total electrical energy delivered (TEED) over time [68]. The TEED is equivalent to the TED referenced earlier in this review. Visualization of the TEED revealed that programming parameters fit a third order polynomial curve corresponding to an initial exponential increase in TEED within the first four years of DBS implantation followed by a plateau followed by a second increase starting at the seventh-year post DBS implantation. The mean voltage at 6 months post DBS implantation was 2.1v whereas last follow-up it was 3.5v. The mean TEED at 6 months follow-up was 43 joules whereas at last follow-up it was 163.5 joules.

One study investigated the phenomenon of progressive ataxia after DBS and whether or not this was due to disease progression or a stimulation induced phenomenon [83]. Five ET DBS patients with chronic progressive ataxia and 5 ET DBS patients without chronic ataxia were studied using VTA analysis and PET CT imaging. PET analysis found that patients with ET and chronic ataxia had increased metabolic activity in the cerebellum compared to ET patients without ataxia. As part of the study protocol, all patients had their DBS system turned off for 72 hours. There was an initial severe rebound of ataxia and tremor in all patients. However, after 72 hours, all patients experienced near complete resolution of ataxia symptoms. The authors proposed the possibility that ET DBS patients experience ataxia as a result of antidromic stimulation of the vestibulo-cerebellar-thalamic afferents within the STA region.

Taken collectively these studies suggest, as the Favilla study found, disease progression may play a contributory role to the progressive worsening of ET post-DBS and that the gradual worsening of tremor in ET DBS patients is similar to that seen in ET patients without DBS [75]. The clinical importance of this point is that patients undergoing DBS surgery for essential tremor should be pre-operatively informed that their underlying tremor will likely progress slowly and may re-emerge with time despite DBS therapy.

8. Axial tremors

Symptom specific targeting is also a frequently studied topic in ET DBS research. Axial tremors (involving the head, voice, face, tongue, or trunk) can be particularly debilitating and are more difficult to treat than appendicular tremors. Early DBS studies suggested that unilateral VIM DBS was inferior to bilateral VIM DBS for suppression of axial tremors [84–87], and many centers have preferentially used bilateral DBS to address axial symptoms in ET. These observations were not invalid, but these studies were typically limited by a small sample size. It is also well known that bilateral procedures are associated with greater risk of adverse events [10–12] and that in many cases, head tremor has been improved with unilateral stimulation.

One study sought to clarify the unilateral versus bilateral relationship and collected data from 12 movement disorders centers accumulating a total of 158 patients for analysis [88]. The authors found that unilateral VIM DBS provided a 58% improvement of axial symptoms at 90 days post DBS implantation and 65% improvement at 180 days post DBS implantation. DBS related dysarthria and incoordination/gait impairment was seen in 8% and 8% of patients, respectively, in the unilateral cohort. Bilateral VIM DBS provided an additional 63% improvement on midline tremor at the expense of 10% and 26% of patients experiencing DBS related dysarthria and incoordination/gait impairment respectively. For comparison, a previous long term bilateral VIM DBS study with 5 year follow-up data reported that 63% of patients experienced dysarthria, 38% experienced incoordination and 25% experienced abnormal gait [89]. This was in comparison to 17% who experienced dysarthria and incoordination and 0% who experienced abnormal gait in unilateral VIM DBS. Another study found that 25% of patients who proceeded from unilateral to bilateral VIM DBS developed new onset dysarthria [90].

8.1. Head tremor

Targeting strategies for specific axial symptoms have also been explored in ET. One retrospective study analyzed surgical parameters for 23 unilateral and 6 bilateral VIM DBS patients with problematic head tremor prior to surgery, in an effort to determine predictors of effective suppression of head tremor [91]. Clinically, at both 6- and 12-months post DBS implantation the authors found similar improvement in head tremor relative to pre-operative baseline with both unilateral and bilateral VIM stimulation.

A logistic regression was constructed to identify predictors of head tremor improvement. The study found that the AC-PC angle (the angle between the sagittal projection of the DBS lead trajectory and the horizontal AC-PC line) was negatively associated with tremor outcome. In other words, larger AC-PC angles (more vertical entries in the sagittal projection) were associated with greater tremor suppression. Further sub-group analysis found the greatest degree of head tremor improvement was associated with AC-PC angles greater than 62 degrees. Although the authors acknowledge a larger study is needed to confirm these results, these findings suggest that the commonly held belief that bilateral stimulation is necessary to control axial tremor is likely untrue, and that more vertical AP entry angles (more parallel to the anterior border of the VIM thalamic nucleus) may offer better somatotopic targeting. Such vertically oriented entry angles also enable stimulation of both VIM and ZI with the same DBS lead.

8.2. Voice tremor

Another study examined the effect of unilateral versus bilateral VIM DBS on voice tremor [92]. Seventy-seven patients were reviewed in a retrospective fashion for the DBS effect on voice tremor as graded by the TRS. The authors found that as a cohort, voice tremor amplitude decreased by 80% in the DBS ON state but there was no difference in efficacy between unilateral and bilateral DBS. The authors then mapped DBS stimulation using a centroid of stimulation model such that the active area of stimulation was defined as the geometric mean of the locations of the stimulated contacts [93]. By using this method, the authors found that more anterior VIM stimulation was associated with better voice tremor suppression in the left hemisphere. This finding was not present in the right hemisphere. When all contact coordinates were analyzed as a group by using absolute values, a cluster of centroids were located more anteriorly and associated with overall better voice tremor reduction.

9. Adverse effects

All DBS procedures carry similar known risks of transient altered mental status, infection, hemorrhage and stroke [94]. The risk of transient delirium has been shown to be higher when the trajectory traverses the ventricle [95]. Adverse effects are common in ET DBS, both as sequelae (transient and persistent) of the surgical procedure and as induced effects of stimulation that may be diminished with optimal programming.

9.1. Dysarthria

Studies have demonstrated that dysarthria is the most common adverse effect of VIM DBS in ET [96–98]. It is more common in bilateral DBS but can occur in unilateral DBS cases. One study examining the effect of VIM DBS on speech in ET patients evaluated 12 ET patients and 12 healthy controls using electromagnetic articulography (EMA) [99]. Study participants were evaluated with a fast syllable repetition task and graded on a voice handicap index (VHI). The authors observed that ET DBS patients had coordination problems of the labial and lingual system in terms of articulatory imprecision and slowness in the DBS OFF state. In the DBS ON state, the ET patients showed prolonged syllable duration, worsened articulation rate and frication—common features in dysarthria [100]. This study did not include non-DBS ET patients in the analysis, so the authors were unable to specify whether the speech findings were due to a lesional effect from DBS or were intrinsic to the ET disease process.

9.2. Gait and ataxia

One study evaluated the effect of unilateral VIM DBS on gait in essential tremor patients [101]. Gait abnormalities can be challenging to study in this population as both ET and DBS can contribute to worsening gait [102,103]. In this study, gait evaluation was performed prior to surgery, and at 6 months post DBS implantation in the off and on DBS states. The study found a minimal effect of VIM DBS on gait at the cohort level, but 25% of patients had a clinically meaningful worsening of gait regardless of whether the DBS was turned on or off. The authors proposed that this may be due to a durable microthalamotomy effect of the lead implantation procedure, rather than a stimulation induced phenomenon. The authors also observed that despite worsening of gait, there was substantial improvement of tremor. Analysis of patient characteristics found that tandem walking performance may be the best predictor of gait worsening following DBS surgery. Another study examined 4 ET patients who reported gait/balance difficulties 6 to 19 months after VIM DBS (2 unilateral, 2 bilateral) [104]. Review of the settings utilized for optimal tremor suppression found a frequency range of 170 to 185 Hz. The patients were reprogrammed by reducing the frequency at 10-20 Hz intervals to identify the lowest effective setting for tremor control and were able to achieve comparable tremor suppression with 130 Hz stimulation. Re-evaluation 4 weeks after implementing lower frequency settings revealed no significant differences in tremor suppression with improvement in gait/balance. Re-activation of previous high frequency settings caused recurrence of gait/balance difficulties in 3 of the 4 patients. The authors concluded that higher frequency stimulation was associated with worsening balance, but given the heterogeneity of ET, the pathophysiology behind this relationship remains unclear.

Ataxia is an expected adverse effect of ET DBS, in every case to some degree [83]. Although the mechanism for this phenomenon has not been fully elucidated, it is thought that the tremor circuit is the same circuit largely responsible for cerebellar modulation of coordinated movement [43,105]. Reich et al. investigated the effects of stimulation on ataxia in 10 ET patients with optimized bilateral thalamic DBS [83]. 5 patients with complaints of chronic progressive gait ataxia after DBS and 5 patients without ataxia complaints after DBS were recruited for this study. Observation in the DBS on versus off states combined with advanced imaging and VTA analysis revealed that ataxia is likely a stimulation induced phenomenon associated with modulation of the ventrocaudal subthalamic region with inadvertent stimulation of the uncinate tract. Further analysis by FDG PET of the 5 patients with progressive ataxia found increased metabolic activity in the cerebellum compared to the 5 patients without progressive ataxia. After turning off the DBS and allowing a 72-hour stimulation wash out, there was no difference in cerebellar metabolic activity of the DBS compared to age/gender matched healthy controls.

Evidence suggests that stimulation deep to the thalamus (ZI, STA, etc) can in selected cases be very effective for tremor suppression, and we certainly have several tremor patients that are successfully managed with this strategy. Predictably, however, there will be a higher incidence of stimulation induced ataxia when targets deep to the thalamus are stimulated for tremor suppression. As Reich et al. noted in their study, precise electrode targeting, and diligent post-operative programming adjustments can maximize tremor suppression while minimizing cerebellar symptoms.

10. Conclusion

DBS for ET has evolved significantly since its inception in 1991. Advances in biomedical technology and electrophysiology have added new layers to the methodological approach towards neuromodulation in ET and the understanding of the pathophysiology of tremor. Although DBS typically provides substantial tremor suppression for most ET patients, long term efficacy has been variable. Several advances in targeting, programming, and hardware have resulted in improved outcomes, but future studies are needed to better understand the entire disease process and to provide a more personalized neuromodulatory therapy.

11. Expert opinion

The technological advances in DBS for ET covered in this review will likely have noteworthy impact on real world experiences in movement disorders centers around the world. The optimal target for neuromodulation remains unclear as some centers have adopted a transition to PSA DBS whereas others still favor traditional VIM DBS. Clinical efficacy, battery life, and side effect profile are all key components that should be taken into consideration when deciding on a DBS target. As new hardware becomes commercially available, the importance of the key components will shift over time. For example, as rechargeable batteries become commonplace, battery life may become less of a concern. This might also be relevant when increasing DBS energy is required to address decreasing efficacy over time, as it would allow physicians to aggressively approach debilitating tremor without having to consider battery life and current drainage.

Stimulation induced side effect profiles can also be approached with contemporary DBS system designs. Future technological advances may allow physicians to refine and contour VTAs to minimize unintentional neuronal activation. Alongside new hardware, novel programming paradigms ranging from segmented contacts to current steering to novel waveforms are already changing clinic visit protocols for DBS programming. This will give physicians a wealth of new options to program around stimulation induced side effects; however, the abundance of settings can also exponentially increase clinic visit time and decrease programming efficiency. The new configurations may also pave way to the identification of new stimulation strategies and targets for symptoms that have traditionally been more difficult to capture with conventional DBS. Novel programming formulas may provide better relief in axial or midline tremor symptoms when directionality and current control techniques can be employed. New strategies are needed to maximize the benefits of next generation devices while evolving from time-consuming conventional practices. Perhaps the answer lies in machine learning or neural network assisted decision making.

Progression of ET DBS research will parallel our exploration into the pathophysiology of ET. As the field of computational modeling and bioinformatics matures, our view of neurologic diseases will start to incorporate network-based representations and hypotheses. The growth of MRI DTI and rs-fMRI technology has provided novel in vivo illustrations of brain circuitry and connectivity to the field of neuroscience. Analysis of these connections will hopefully allow us to identify reliable patterns of abnormal networks that drive tremor generation in ET. Utilizing network-based strategies may also facilitate widening of therapeutic window as we catalogue circuit profiles that correspond to stimulation induced side effects. This could then lead to targeting strategies that focus on neuromodulation of networks rather than focal regions of interest within the brain. As we foster a global, whole brain network approach to neurological diseases, we may discover that while principle connectivity may be a universally common feature, each person may have distinctive variations that lead to a unique connectome profile. By approaching the connectome in this fashion, we can create patient specific neuromodulation therapies for each individual and offer patients meaningful changes in quality of life.

Article highlights.

• The posterior subthalamic area and zona incerta are novel, effective targets for ET DBS, but may be associated with more stimulation induced ataxia

• Connectivity profiling, constant current stimulation, and segmented DBS electrodes are emerging features that will assist in DBS optimization

• Unilateral VIM DBS provides substantial contralateral and modest ipsilateral tremor suppression, and has been shown in select cases to be sufficient for management of axial tremors in ET

• Dysarthria, ataxia and gait instability are common adverse effects of ET DBS that are more frequent after bilateral than unilateral procedures, and may result from both microthalamotomy effects and from stimulation-induced side effects

• Loss of long-term efficacy in ET DBS is not uncommon, and is due primarily to disease progression rather than to tolerance and/or habituation