Introduction

Deep brain stimulation (DBS) in thalamic ventrointermediate (VIM) nucleus is the traditional target for the surgical treatment of pharmacologically refractory essential tremor (ET) or parkinsonian tremor. Studies in recent years on DBS in the posterior subthalamic area (PSA), including the zona incerta (Zi) and the prelemniscal radiation (Raprl), have shown promising results in tremor suppression [1–25], particularly for those difficult to be controlled by VIM DBS, such as the proximal postural tremor, distal intention tremor and some cerebellar outflow tremor in ET, multiple sclerosis (MS), post-traumatic tremor (PTT), cerebellar tremor (CT), Holmes tremor (HT) and spinocerebellar ataxia 2 (SCA2) [8, 12–14, 16]. The adverse effect profile of the PSA DBS is mild and transient, without lasting or striking dysarthria, disequilibrium or tolerance, as would have been seen in the VIM DBS, particularly bilateral DBS [26–30]. However, the studies on PSA are still limited given less than 30 publications in the PubMed so far, with even a handful of studies performed on bilateral PSA, and a short follow up duration compared to VIM. Therefore, a mini-review on DBS in PSA is needed to gain more comprehensive insight into the potential benefits and limits of the PSA DBS compared to that in VIM DBS in the clinical management of various tremors, particularly for those tremors difficult to be well controlled by traditional VIM DBS.

DBS in PSA: evidence on effective tremor control and others

Anatomically, the PSA is bounded anteriorly by the posterior border of the subthalamic nucleus (STN), superiorly by the ventral thalamic nuclei, inferiorly by the dorsal border of the substantia nigra, posteriorly by the medial lemniscus, posteromedially by the anterolateral border of the red nucleus, posterolaterally by the ventrocaudal nucleus, and laterally by the posterior limb of the internal capsule [31]. It consists of Zi and the Raprl. The Zi lies dorsal and posterior to STN, joining both the basal gangalia thalamocortical circuit and the cerebellar thalamocortical circuit. The Zi anatomically also consists of caudal part and rostral part. The Raprl is a fiber bundle that lies posterior to the STN, separated from it by the intervening Zi. It contains fibers from the mesencephalic reticular formation that projects to the thalamus as well as ascending cerebellothalamic fibers. The Forel H1 (thalamic fasciculus) and H2 (lenticular fasciculus) lies dorsal to the STN and immediately anterior to the PSA, and the Forel H lies anterior to the red nucleus.

Whilst the Zi was a target for lesioning since the early sixties, DBS in PSA was first reported by Mundinger in 1977 [1]. The intraoperative stimulation of the Zi and surrounding area obtained good control of torticollis in 7 patients without complications [1]. Successful results were also reported by Brice and McLellan in 1980 and Andy in 1983 [2, 3]. The unsophisticated DBS devices, the successful therapeutic effects of levodopa in Parkinson’s disease (PD), and the successful DBS on VIM by Benabid in 1987 [32] are all possible reasons for the lack of sufficient reports during that period of time. In 2000, a successful attempt on the DBS targeting Zi in controlling intention tremor in ET and dystonic tremor (DT), which were hard to be controlled otherwise by DBS targeting VIM, was reported by Kitagawa et al. [4]. This study reignited the interest of DBS targeting Zi and more broadly PSA. Since then, more than 20 articles have been published, as listed in the Table 1, including 201 cases of ET, 99 cases of PD, 32 cases of MS, 6 cases of DT, and several others (PTT, CT, HT, WC, and SCA) [1–25]. The majority of the DBS was placed unilaterally. Only about 25% of the cases had bilateral DBS (Table 1). They were followed up for 3 months to 6 years. Most of them were targeting caudal Zi (cZi), with different parameters among studies. The therapeutic benefit was prominent, particularly for those tremors difficult to be controlled by VIM, such as proximal postural tremor, distal intension tremor and cerebellar out flow tremor. The complications mostly were mild and transient. There was no lasting or striking dysphasia, dysarthria, or disequilibrium, which could be encountered in VIM DBS, particularly bilateral VIM DBS [26–29]. There was no tolerance either [9, 25], another unfavorable feature often reported in VIM DBS [30]. The effect on axial symptoms, such as vocal tremor, head tremor and swallowing function, was rarely mentioned [9]. A good effect on neck tremor was reported on bilateral cZi DBS [9]. DBS on cZi was not found to have negative effect on the swallowing function [33], but have slightly different effect on voice compared to STN DBS [34].

Table 1 PSA DBS publications: indications, targets, results and side effects
Full size table

The difference in the specific targets of PSA, the number of DBS placed (unilaterally vs bilaterally), the post-surgical follow -up, and the different diseases studied among these articles makes a concise comparison of clinical benefits and limits of PSA with VIM difficult. A site-to-site comparison among different patients could be more helpful to delineate the difference between the targets. One of the good examples was the study by Hamel et al. [13], who found that cZi DBS at the parameters of 12.7 mm lateral, 7.0 mm posterior and 1.5 mm ventral to the mid-commisural point (MCP) provided better outcome than VIM DBS. Though it is a retrospective study with different post-surgical duration, the site-to-site comparison did provide more reliable comparison of these two targets. A similar result was also reported by another group [11, 20, 21], with electrodes below the intercommissural line (ICL) producing better outcome in tremor control than those above the ICL [20], though the better outcome meant to be more efficient for cZi than VIM rather than significant difference in individual or maximal stimulation response in these two targets in some studies [20]. More recently, a prospective study designed for direct comparison of PSA/cZi with VIM in each individual patient was presented to the 16th MDS meeting in Dublin [35]. The electrode in the study was placed across both PSA/cZi and VIM, which would allow precisely comparing these two targets in the same patient, with both targets targeted in the same side of the brain, or with one target in one side of the brain and the other target in the other side of the brain of the same patient. This design avoids the confounding difference in the disease severity, post-surgical duration, and skills/targets in assessing the DBS efficacy, and allows more accurate site specific assessment (Zi vs VIM). A significantly better outcome in general was found in stimulating cZi than VIM, except that some of the patients could not tolerate the adverse effects of paresthesia in cZi despite the best control of the tremor. The vocal tremor and head tremor were also very well controlled by the stimulation. No worsening of gait was observed after the surgery. The study shows very promising tremor control by cZi DBS, though the conclusion is limited by the small sample size of five patients, as acknowledged by the authors [35]. Stimulating bilateral VIM was also reported to improve vocal tremor and head tremor in some studies as reviewed by Lyons and Pahwa [36], however the occurrence and the worsening of dysarthria and disequilibrium were the concerns in choosing bilateral VIM stimulation in some cases [26–29], and made unilateral VIM DBS as an alternative in certain circumstances [37]. The lack of lasting dysarthria and disequilibrium reported even in bilateral cZI DBS is probably because the cZi DBS only overrides tremor oscillations without interrupting patterns of information related to fine movements of vocal cords and proprioceptive sensation [16].

Besides effective tremor control [11, 21, 24], cZi was also found to be better than STN in controlling rigidity and bradykinesia without difference in reducing dyskinesia and levodopa equivalent dose [11]. The rigidity and bradykinesia was not found to be very well controlled by cZi DBS in another study though [24]. Dystonia was also found to be well controlled by cZi DBS [1, 16, 17]. More recently, cZi in combination with pedunculopontine nucleus was found to have positive effect on axial symptoms [38]. cZi DBS could also be a target for some patients with ataxia [39].

DBS in PSA: targeting the targets

As neither cZi nor Raprl can be reliably distinguished on 1.5 Tesla MRI, some used the neighboring structures as reference, such as the STN or VIM, to guide the DBS placement. Most of the studies used posterior and medial to the posterior tail of the STN at the maximal diameter of the red nucleus as the target (Table 1). Some of them used the 2-3mm under the ventral border of the VIM as the target [4, 35]. Only a few studies also used microelectrode recording to guide the electrode placement, as cZi gives silent or low activity neuronal background [35], which differentiates cZi from VIM. Imaging targeting in combination with macrostimulation was applied virtually by all studies.

Specifically, the PSA was identified on trans-axial T2-weighted MRI images slightly posterior medial to the subthalamic nucleus at the level of the maximal diameter of the red nucleus (Table 1), or about 2-3mm below the ventral border of the VIM [4, 35]. A pre-operative MRI fused with a head CT with stereotactic frame and superimposed digitized Schaltenbrand stereotactic atlas was used to plan the trajectory. The electrode was implanted under local anesthesia. EMR was further used in some studies [35]. The final position of the electrode was dictated by the response to the macrostimulation. A postoperative x-ray was performed before removal of the frame. The location of each electrode contact post-surgically was determined on the postoperative CT infused with the pre-surgical MRI and superimposed atlas. The DBS contact location was determined in relation to the anterior commissure (AC) – posterior commissure (PC) line and the coordinates were plotted onto the Schaltenbrand stereotactic atlas. The efficacy of the stimulation was presented as the result on stimulation in relation to the result off stimulation at the same evaluation, based on clinical exam, UPDRS motor scores, tremor rating scores, daily living function, or quality of life assessment.

DBS in PSA: possible mechanism

The mechanism of tremor suppression by stimulation in PSA (mostly in cZi) is not entirely clear. The Zi is a heterogenous nucleus that lies at the base of the dorsal thalamus and is an extension of the reticular /thalamic nucleus [40]. Its rostral component extends over the dorsal and medial surface of the STN, and its caudal or motor component lies posteromedial to the STN [41, 42]. The ZI receives afferents from the globus pallidus internus (GPi), the substantia nigra reticulate (SNr) [41, 43, 44], the ascending reticular activating system [43–45], the interpositus nucleus of the cerebellum, and also the motor, associative and limbic areas of the cerebral cortex [43, 46]. The ZI sends efferents to the centromedian and parafascicular nuclei [47–49], the ventral anterior nucleus and the ventral lateral nucleus of the thalamus [50], the midbrain extrapyramidal area and the medial reticular formation [41], the GPi and SNr [41], the interpositus nucleus of the cerebellum, the inferior olive and cerebral cortex [51–53]. Current hypotheses regarding the mechanisms of tremor generation point to abnormal synchronisation of neuronal firing in the basal ganglia thalamocortical loop (in PD and DT) or the cerebellar thalamocortical loop (in ET, CT and MS tremor) or both loops (HT) [16]. cZi is an effective target for the surgical control of all forms of tremor because of its unique GABAergic connections with both the basal ganglia and cerebellar thalamocortical loops, in addition to the brain stem motor effectors through which tremor oscillation may be transmitted [16]. Stimulation of the Zi is likely to suppress the tremor by overriding the oscillations in these areas [16]. Stimulation of the Raprl could similarly abolish contralateral tremor and reduce rigidity [54]. The Raprl is a fiber bundle that lies posterior to the STN, separated from it by the intervening Zi, and consists of fibers from the mesencephalic reticular formation that project to the thalamus as well as ascending cerebellothalamic fibers. How much of the stimulation of the ZI overflows into the neighboring Raprl is unknown and may vary according to individual electrical conductivity of these structures in individual patients. The exact mechanism of how stimulation of PSA suppresses various tremors still awaits further studies to corroborate.

Conclusion

PSA could potentially be an alternative target for the tremor, particularly for those tremors difficult to be controlled by traditional VIM DBS, including the proximal postural tremor, distal intention tremor, and some cerebellar outflow tremor. The effect of PSA DBS on axial head tremor and vocal tremor also seems to be promising. The adverse effect profile of the PSA appears transient and mild. However, the conclusion is limited by the small numbers of studies so far. More studies, including randomized double-blinded trials comparing the effect of DBS targeting PSA with that targeting VIM, are needed to help us better understand the efficacy and adverse effect profile of the PSA DBS, which could have profound effect on tremor control, particularly for those difficult to be controlled by traditional VIM DBS.