Cerebral—deep

The thalamus

Following the pioneering work of Hassler in Germany [ ], Cooper in the USA [ ], and Narabayashi in Japan [ ], the thalamus was the favored target of functional neurosurgeons in the precomputed tomography (CT), premicroelectrode, prelevodopa era. The reasons are obvious:

• 1.

  in structures such as the ventrocaudal (Vc) and ventrolateral (VL) nuclei, neurons are arranged in a clear topographic manner, simplifying electro-physiological mapping with the cruder macroelectrode techniques of the day

• 2.

  the effects of stimulation at these targets typically are immediate, allowing the surgeon to feel comfortable about electrode position prior to performing an irreversible ablation

• 3.

  functions are relatively compartmentalized in the thalamus so that one may treat movement, for example, without affecting sensation.

Today, the thalamus is targeted less frequently than other deep cerebral structures, but a working knowledge of thalamic anatomy and physiology remains essential. The interested reader is directed to Dr. Ronald Tasker's classic work on thalamic physiology [ ] and Dr. Patrick Kelly's detailed description of his ventrolateral thalamotomy technique, employing semimicroelectrode recording (MER) [ ].

Pain

DBS-derived analgesia was first observed and reported by Pool [ ] and Heath [ ] who found that stimulating the septal nuclei, including the diagonal band of Broca anterolateral to the forniceal columns, resulted in significant pain relief in psychiatric patients. Mazars reported that thalamic stimulation produces paresthesias with simultaneous long-lasting relief of deafferentation pain [ ]. As a direct extension of Melzack and Wall's “Gate theory,” [ ] Reynolds reported on the analgesic effect of aqueductal stimulation in rats [ ]. Analogous work by Hosobuchi [ ] and Richardson [ , ] first demonstrated the efficacy of thalamic and periventricular/periaqueductal gray (PVG/PAG) stimulation for the relief of pain. Since then, the Vc and PVG/PAG have been the most studied sites of DBS for pain in humans.

The mechanisms underlying pain relief via stimulation at these sites appear to be different though they are still not completely understood. Hosobuchi proposed that pain relief derived from PVG/PAG stimulation is mediated by opioid release following the observation that stimulation-induced analgesia at this site is blocked with naloxone [ ]. Current thought maintains that the analgesic effect of PVG/PAG stimulation is mediated by multiple opioid- and biogenic amine-dependent supraspinal descending pain modulatory systems. In addition, ascending pathways from the PVG to the medial dorsal nucleus of the thalamus, an area associated with the limbic system and with extensive connections to the amygdala and cingulate cortex, have been identified, raising the possibility that stimulation of the PVG may also modify the patient's emotional response to pain [ ]. Consequently, the majority of PVG stimulation studies have concentrated on its utility in treating intractable nociceptive rather than neuropathic pain. The results of many individual studies and pooled meta-analyses of PVG/PAG DBS for nociceptive pain have demonstrated success rates as high as 63%, depending on the etiology [ ].

In contrast, pain relief from Vc thalamic stimulation is thought to be mediated by activation of the nucleus raphe magnus of the rostro-ventral medulla as well as descending inhibitory pain pathways [ ]. Ventrocaudal thalamic stimulation has been applied most frequently in the setting of neuropathic/deafferentation pain syndromes, including anesthesia dolorosa, poststroke pain, brachial plexus avulsion, postherpetic neuralgia, and postcordotomy dysesthesia. In general, deafferentation pain syndromes respond less well to stimulation than do nociceptive syndromes, with relief in a mean of 47% of patients [ ]. Of these, 31% of patients with a central pain etiology (e.g., thalamic poststroke pain) respond to thalamic DBS, while 51% of those with a peripheral etiology (e.g., postherpetic neuralgia) experience a meaningful response. Interestingly, the rate of long-term pain alleviation is highest in those patients undergoing DBS of the PVG/PAG alone (79%), or the PVG/PAG plus the thalamus (87%). Stimulation of the thalamus alone is less effective (58%) than stimulation of the PVG/PAG ± thalamus ( P < .05) [ ]. Many studies have thus concluded that DBS is more effective in treating nociceptive pain syndromes and that stimulation at both the PVG/PAG and thalamus may be most effective [ ].

Expanding upon the gate-control theory of pain, the neuromatrix theory conceives of “pain” as a unified experience amalgamating three components: sensory-discriminative, affective-emotional, and evaluate-cognitive [ ]. Whereas PVG/PAG and Vc stimulation target the sensory or nociceptive aspects of pain, the anterior cingulate cortex of the limbic system has been proposed as modulating the patient's subjective evaluation of and response to pain [ ]. To this end, Boccard et al. explored DBS of the ACC in an unblinded, nonrandomized series for treatment of neuropathic pain resistant to pharmacotherapy. Twenty-two of the 24 patients enrolled had an initial response to stimulation at 1 week and were permanently implanted. Among 12 patients with 2-year follow-up, 60.3% improvement in pain as measured by the Numerical Rating Scale was observed at 6 months ( P < .001), but efficacy declined over time [ ]. Lempka et al. targeted an alternative structure in the emotional pathway, the ventral striatum/anterior limb of the internal capsule (VS/ALIC), in a double-blinded, randomized, cross-over trial of nine patients with poststroke pain syndrome. Although their primary endpoint of ≥50% pain disability index improvement in >50% of patients was not achieved, they reported the affective component of chronic pain demonstrated >50% improvement in a third of patients at 2 years [ ]. Of significant concern, stimulation-induced seizures were observed in both trials. Lempka et al. reported isolated, stimulation-induced seizure in one of nine patients [ ]; while Boccard et al. noted seizures in four patients, two of whom resisted reducing stimulation because of pain relief and developed recurrent seizure [ ]. Presently, DBS is not approved in the USA at any target for the treatment of refractory pain.

Tinnitus is a disorder of perception in which a patient experiences sound in the absence of auditory input. As tinnitus is an uncomfortable response to absent external stimuli, the affective experience of tinnitus has been analogized to that of neuropathic pain [ ]. The disorder appears to implicate both the auditory and limbic pathways: neuroimaging reveals aberrant activity of the prefrontal cortex, anterior cingulate, insula, and amygdala in addition to the central auditory system [ ]. However, to date, neuromodulation has primarily targeted structures associated with motor dysfunction. This is because tinnitus has been incidentally noted to improve during lead placement for motor disorders when targeting in the Vim [ ], STN [ ], and during traversal of the area LC [ ]. As a result, Cheung et al. targeted the bilateral caudate nucleus in a Phase I trial of DBS for tinnitus resistant to conventional therapy. After 6 months of optimal stimulation, three of five patients were deemed “responders” by the primary endpoint (Tinnitus Functional Index) and four were responders by the secondary outcome (Tinnitus Handicap Inventory). Although no serious adverse effects were observed, optimization required an average of 9.2 months [ ].

Cluster headaches are a subtype of trigeminal autonomic cephalalgias characterized by unilateral episodes of intense pain secondary to activation of CN V1 in conjunction with parasympathetic dysautonomia from CN VII. The cephalalgia may persist up to 3 h and be accompanied by ipsilateral miosis, lacrimation, and rhinorrhea [ ]. May et al. induced cluster headaches in nine susceptible patients and noted increased activation of the ipsilateral hypothalamic gray matter [ ]. Subsequent work detected increased functional connectivity between the posterior hypothalamus ipsilateral to the headache and the ventral tegmental area (VTA), dorsal raphe nuclei, bilateral substantia nigra, STN, and red nucleus [ ]. DBS for cluster headaches has primarily targeted the VTA and posterior hypothalamus [ ]. Leone et al. detailed the results of 16 patients receiving hypothalamic stimulation after 23 months' follow-up, noting that 13 patients achieved major pain relief and the remaining three saw abatement in symptoms [ ]. In the largest series of patients undergoing VTA stimulation, Akram et al. reported 60% reduction in headache frequency and 30% reduction in severity with 17/21 patients classified as responders [ ]. The most common side effects of targeting either the VTA or hypothalamus are diplopia, vertigo, and emotional but rarely persist after adjusting stimulation [ ].

Tremor

Tremor is a rhythmic, involuntary oscillation of the musculature that can affect the head, extremities, and/or trunk. Tremor is characterized by its clinical manifestations (i.e., resting, postural, action, and/or intention) and may be caused by multiple neurological disorders including PD, ET, traumatic brain injury, stroke, and multiple sclerosis. In the 1950s, Cooper serendipitously discovered that ligation of the anterior choroidal artery ameliorated tremor, though paresis could also result [ ]. Further research by Narabayashi [ ], Hassler [ ], Cooper [ ], and others identified the ventrolateral nucleus of the thalamus as the primary target for eliminating tremor and employed ventriculography-based stereotaxis to ablate this site directly. Thereafter, thalamotomy remained the most commonly performed procedure for involuntary movement disorders until the late 1980s when Benabid developed Vim DBS as an alternative [ ].

The junction of the ventral intermediate (Vim) and ventral oralis posterior (Vop) subnuclei of the VL thalamus is the most commonly targeted site for treating disabling parkinsonian and ET with DBS [ , ]. The Vim and Vop are histologically distinct subnuclei located posteriorly in the VL nucleus. The Vim receives excitatory cerebellar input and projects to the motor cortex.

Multicenter trials in North America [ , ] and Europe [ , ] as well as smaller case series report excellent results with unilateral and bilateral thalamic DBS for tremor. Taken together, these studies report significant improvement of hand tremor in up to 75% of unilaterally and 95% of bilaterally stimulated patients, respectively [ ]. Axial tremor (head, voice) is improved in up to 50% and 100% of unilaterally and bilaterally stimulated patients, respectively [ ]. These effects appear to be long lasting, though tremor recurrence due to stimulation tolerance has been reported. Studies comparing DBS to radiofrequency thalamotomy demonstrate equivalent tremor suppression but a lower risk of neurological complications in patients treated with DBS [ ]. The most common deficits related to thalamic interventions are hemiparesis, dysarthria, ataxia, and sensory deficits, most of which abate over time. Suppression of tremor results in significant reductions in functional disability in patients with ET [ ]. In contrast, patients with advanced PD do not realize significant functional improvements following thalamic DBS because their other more disabling symptoms (e.g., rigidity, bradykinesia, motor fluctuations, levodopa-induced dyskinesia, and gait disturbance) are not improved. Consequently, DBS at other targets is more commonly employed for patients with advanced PD (see below).

An additional target for tremor suppression is the posterior subthalamic area (PSA). The PSA is a bottleneck of traversing fibers and sparse somata critical to motor control. It is circumscribed anteriorly by the subthalamic nucleus, laterally by the internal capsule, medially by the red nucleus, posteriorly by the medial lemniscus, inferiorly by the substantia nigra, and superiorly by the ventral thalamic nulcei. It primarily contains the caudal portion of the zona incerta (cZi) and prelemniscal radiations (Raprl) while in close promixity to Forel's fields H, H1, and H2. Within the PSA, the cZi is lateral to the Raprl, lateral in turn to the red nucleus. Forel's fields H1 and H2 travel medial to the STN as field H is prerubral [ ]. The cZi contains GABAergic neurons with reciprocal connections to the thalamus, basal ganglia, cerebellum, and mesencephalon [ ], whereas the cerebellothalamic and mesencephalic reticular fibers travel through the Raprl [ ]. Numerous motor, sensory, and visceral tracts have been implicated in the fields of Forel but most clinically pertinent are the cerebellothalamic tracts of fields H and H1 [ ].

Despite recognition of the PSA as a target for lesioning and DBS since the 1960s [ ], Kitagawa et al. renewed interest in the target after achieving tremor suppression by stimulating the target in two patients, one with dystonic tremor uncontrolled by prior thalamotomy and one with ET uncontrolled by intraoperative VIM stimulation [ ]. Subsequently, PSA has been targeted for effective tremor relief in PD [ ], ET [ ], and multiple sclerosis [ ], as, less commonly, dystonia [ ], posttraumatic tremor [ ], cerebellar tremor [ ], Holmes tremor [ ], writer's cramp [ ], spinocerebellar ataxia [ ], and orthostatic tremor [ ]. Improvement in tremor has been reported at 58.2%–80% in ET [ , , ], 78.3%–94.8% in Parkinsonian tremor [ , , ], and 50.4%–57.2% in MS-associated tremor [ , ]. Unlike the VIM, stimulation of the PSA has provided improvement in other PD symptoms, such as rigidity (34.3%–94% [ , ]), akinesia (32%–65.7% [ , , ])/bradykinesia (26.7%–75% [ , , ]), and overall UPDRS-III score (41%–76%) [ , , , ] though not postural instability or gait [ , ].

Comparison between PSA and VIM for tremor relief is ongoing. Early evidence suggests that PSA stimulation may provide similar-to-slightly-superior tremor control compared to VIM stimulation but at a more efficient stimulation setting. In a randomized, double-blind crossover trial of 13 patients with electrode contacts in both the VIM and the PSA, Barbe et al. reported that stimulation of the PSA achieved 64% tremor relief from baseline while VIM contacts achieved 50% improvement ( P = .086) at lower current delivery than VIM ( P = .006) [ ]. In a review of 36 patients with VIM or PSA DBS, Sandvik et al. noted that in the “VIM” cohort placement, a plurality (47%) of the most effective contacts were actually located in the Zi/Raprl [ ]. A retrospective comparison of 34 patients who underwent VIM and 34 patients who underwent PSA documented that overall and hand-specific tremor at final follow-up improved 62% and 89% with PSA compared to 49% and 70% with VIM stimulation. However, dissimilar follow-up between the VIM and PSA cohorts (28 vs. 12 months) limited conclusions out of concern that target efficacy could vary with time [ ].

An additional advantage of PSA stimulation over VIM may be tolerability. Namely, ongoing dysphasia, dysarthria, and disequilibrium are known risks of VIM DBS (especially when bilateral) as well as waning tremor relief [ , , , ]. In PSA, these side effects are transient or uncommonly described [ ]. For example, Fytagoridis and Blomstedt described dysphasia in 22.5% of their PSA patients, but it was mild and disappeared by 5 weeks [ ]. Although PSA efficacy may wane slightly as microlesioning effect subsides and underlying tremor progresses [ ], stimulation parameters may not need to be significantly increased [ ]. Maintenance of lower stimulation may be particularly important when targeting the PSA since higher voltages can risk ataxia from nearby fiber recruitment [ ] and dystonic phenomena have been noted to mark the upper limit of PSA stimulation [ ].

In 2016 the FDA also approved MRgFUS thalamotomy for the treatment of refractory tremor based on the results of a multicenter randomized trial, which demonstrated significant tremor control and an acceptable safety profile. Though incisionless, MRgFUS thalamotomy still entails the creation of an irreversible lesion within the thalamus and so the risks of lesioning delineated above remain. While short-term results and patient acceptance are promising, long-term efficacy and safety data are still lacking [ ].

Epilepsy

By its very nature, epilepsy would appear to be the ideal disorder to treat with electrical neurostimulation and, in particular, intermittent responsive stimulation. Toward that end, neurosurgeons have targeted a number of deep cerebral, cerebellar, and brainstem sites with the hope of controlling seizure disorders. These include the corpus callosum, caudate nucleus, centromedian thalamus, posterior hypothalamus, subthalamic nucleus, and the hippocampus. Stimulation at these targets has often appeared efficacious in small open-label studies, but failed to achieve significant seizure control when tested in a controlled fashion [ ]. Consequently, most of these DBS strategies have been abandoned and the substantial population of medically refractory epilepsy patients who are not candidates for resective/ablative surgery are currently treated with vagus nerve stimulation. Two neurostimulation strategies, one “open-loop” and one responsive have been approved for use.

Tourette's syndrome

Tourette's syndrome (TS) is a chronic complex neuropsychiatric disorder characterized by sudden, repetitive, stereotyped motor, or vocal tics. TS is often comorbid with OCD, attention deficit hyperactivity disorder, and/or selfinjurious behavior [ , ]. Similar to PD (see below), disordered cortico-striato-pallido-thalamo-cortical circuitry may be responsible for the motor and nonmotor manifestations of TS. Hyperactivity within the dopaminergic system may lead to excessive thalamocortical drive, resulting in hyperexcitability of cortical motor areas and the release of tics. Hyperactivity in Broca's area, the frontal operculum, and the caudate nucleus may underlie vocal tics, while abnormal activation of the orbitofrontal region (as is observed in OCD) may underlie the compulsions that patients with TS experience [ , ].

Based on Hassler and Dieckmann's success with thalamic lesioning for TS [ ], Visser-Vandewalle et al. performed the first thalamic DBS for TS in a 42-year-old male, achieving complete resolution of his tics 1 year postoperatively [ ]. Since then, several small series have reported success with DBS for TS at several different targets:

• 1.

  the centromedian nucleus including either the substantia periventricularis and nucleus ventro-oralis internus (CM–SPv–Voi) [ ] or the parafascicular nucleus (CM-Pf) [ ] or dorsomedial nucleus [ ].

• 2.

  the posteroventral globus pallidus pars interna (GPi) [ ].

• 3.

  the anteromedial GPi [ ].

• 4.

  the nucleus accumbens (NAc) and anterior limb of internal capsule (ALIC) [ , , ].

• 5.

  Subthalamic nucleus [ ].

Preliminary data suggest that the efficacy of thalamic versus pallidal DBS in TS is similar, though pallidal stimulation may attenuate tics more abruptly and thalamic stimulation may yield better effects on mood and impulsivity [ , ]. According to a review by Porta et al., tic reduction ranging from 25% to 100% has been reported in a total of 39 TS patients with follow-up periods of 3–60 months [ ]. A pooled analysis of 156 cases across heterogenous targets by Balderman et al. estimated a median reduction of 53% on Yale Global Tic Severity Scale; while both vocal and motor tics improved, greater tic relief was observed in the vocal aspect. No optimal target was identified, but ALIC/NAc seemed to offer less clinical benefit as a target [ ]. Additional research will be necessary to determine whether an optimal target for TS exists and whether efficacy can be demonstrated in larger case series.