CHAPTER 83 Deep Brain Stimulation
Mechanisms of Action
Key Developments
The practical application of DBS has benefited from several important neurosurgical innovations and discoveries since the late 1800s that were motivated by the need to improve the efficacy of subcortical ablative procedures for treating patients with neurological disorders (Fig. 83-1). Perhaps the most noteworthy of these advances was reported by Spiegel and associates1 in 1947, when they described the use of a human stereotactic frame derived from and similar in concept to that reported by Horsley and Clarke2 for animal work. The stereotactic frame was developed to allow subcortical lesions to be made with more precision and accuracy. Using a variation of this apparatus, Albe-Fessard and colleagues3 demonstrated the practical application of a neurosurgical targeting technique known as intraoperative microelectrode recording. This approach involved sampling neuronal spike activity from a microelectrode inserted in the brain as a means of identifying regions with pathologic activity and then using that information to refine anatomic maps before ablative procedures. Neurosurgical targeting also progressed with the advent of computed tomography (CT) and magnetic resonance imaging (MRI), which allow one to determine the coordinates for a target based on the specific anatomy of the patient’s brain.
The seminal study of the potential application of DBS in humans was performed by Hassler and colleagues4 in 1960. They described a series of parkinsonian patients who received electrical stimulation via electrodes introduced into the globus pallidus for the purpose of creating therapeutic ablations. They observed that low-frequency stimulation (<25 Hz) in the globus pallidus could exacerbate contralateral tremor, whereas high-frequency stimulation (25 to 100 Hz) through the same electrode could alleviate or abolish tremor entirely. They suggested that electrical stimulation could be used to pinpoint regions of the brain where lesions would reduce the motor symptoms of movement disorders. Following the reported success of intraoperative stimulation for mitigating pain and reducing the motor signs of movement disorders,5–8 the next step was to deliver this therapy continuously through an implanted system. Bechtereva and colleagues9 were the first to describe a semichronic DBS system, but not until the maturation of battery-powered implantable pulse generators (IPGs) in the 1980s and the work of Benabid and colleagues10 were chronic DBS systems fully realized.11 Since then, DBS systems have been implanted in tens of thousands of patients with medication-refractory neurological and psychiatric disorders. For many of these disorders, electrical stimulation has now supplanted ablation as the surgical treatment of choice because DBS is reversible and can be titrated to maximize therapeutic benefit.
General Principles
Implantation Surgery
Once the final implantation coordinates are determined physiologically, the mapping electrode is removed, and the DBS lead is implanted. The Medtronic Model 3387 and 3389 DBS leads (Fridley, MN) are the only leads currently approved by the U.S. Food and Drug Administration. They consist of four cylindrical electrode contacts (1.5 mm high, 1.27 mm in diameter) spaced either 1.5 mm (Model 3387) or 0.5 mm (Model 3389) apart from one another (Fig. 83-2A). During insertion of the DBS lead, fluoroscopy is often used to confirm that the lead has reached the correct depth and that it has not moved during the process of securing it at the cranium. Stimulation through the DBS lead is then performed using a handheld screener, and symptoms and side effects are evaluated in real time to verify that the threshold intensities for generating adverse side effects are sufficiently greater than those necessary to generate therapeutic effects.
Once implanted, the DBS lead is secured by means of a bur-hole cap, and any extra length of lead wire is carefully placed beneath the galea. Postoperative CT and MRI scans are generally obtained to verify lead location and to confirm that hemorrhage has not occurred (Fig. 83-2B). The remainder of the DBS system, which consists of an IPG placed within a subcutaneous infraclavicular pocket and an extension cable that is tunneled percutaneously along the neck to connect the distal end of the DBS lead to the IPG, is implanted during a second procedure performed with the patient under general anesthesia.
Stimulation Parameters
After the patient has recuperated from surgery, the IPG is programmed in the outpatient clinic. At the initial programming session, the neurologist or psychiatrist seeks to determine the settings that provide the most therapeutic benefit, minimize undesirable side effects, and limit the stimulator’s power consumption to extend battery life.12 When treating PD, for example, most experienced programmers first stimulate in a monopolar configuration using an electrode contact as the cathode (negative voltage) and the IPG’s outer casing as the anode (positive voltage). The applied stimulation pulse train consists of a pulse width of 60 to 90 µsec, a pulse frequency of 130 Hz, and an amplitude of 0.5 to 4 V or more. Adverse effects may occur at higher voltages, but if they occur at voltages within or before the therapeutic range, one can use a bipolar stimulation configuration (i.e., one of the contacts acts as the cathode, another as the anode) to restrict the spread of current into adjacent regions of the brain responsible for inducing the side effects.
Anatomic Targets
Identifying anatomic targets for DBS initially relied on the experience and knowledge gained from surgical ablative procedures. Although the therapeutic correlation between the lesion target and the DBS target is true in most cases, it is not able to be generalized. For example, DBS of the external globus pallidus (GPe) can improve PD motor symptoms,13 but lesioning the GPe can worsen PD motor symptoms.14 Positron emission tomography (PET) has proved useful for establishing which brain regions show abnormal activity preoperatively15 and whether DBS reverses these abnormalities postoperatively.16 Intraoperative microelectrode recordings also support the concept that brain regions with abnormal neuronal activity coherent with the patient’s symptoms are potential target areas for DBS. Other investigators have used computational models to examine the inhomogeneity of voltage fields produced by DBS and how these distributions can lead to suprathreshold currents in multiple neuronal and non-neuronal elements, some of which may be adjacent to the presumed target nucleus or fiber pathway.17,18 Correlation analysis between model predictions and clinical outcomes of DBS can then be used to determine which stimulated neural elements improve symptoms or generate undesirable side effects. These considerations are addressed in more detail in the following sections on the clinical indications for DBS. Remarkably, DBS can produce functional benefit by targeting any of a number of different interconnected brain regions, which emphasizes the role of a common dysfunctional network in many neurological disorders (Fig. 83-3).
Tremor
The most common surgical target for treating medication-refractory ET with DBS is the ventral intermediate (VIM) nucleus of the thalamus.19 The VIM nucleus rests in the posterior third of the thalamus, 0 to 15 mm dorsal to the anterior commissure–posterior commissure (AC-PC) line and approximately 10 to 20 mm from the midline. Intraoperative microelectrode mapping of the VIM nucleus is often necessary to define the implantation site because the minimal current intensity necessary to relieve tremor usually depends on the electrode’s precise location within the nucleus. Correct placement of the lead can be ascertained not only by suppression of tremor in the upper limb but also by transient paresthesias in the fingers or face.20 The lead is generally placed at the border of the VIM nucleus and the posterior ventral oral (VOP) nucleus of the thalamus to prevent current from spreading too posteriorly into the sensory nucleus (ventral caudal nucleus [VC]). If paresthesias persist with stimulation, current is likely spreading into the VC nucleus and the lead is too posterior (Table 83-1). Leads placed too medially elicit intraoral paresthesias with stimulation, and paresthesias in the leg usually indicate that the lead is too lateral. Placing the lead in the anterior portion of the VIM nucleus allows one to avoid continued activation of the VC nucleus; however, it is likely that the VOP nucleus is also affected by stimulation under these conditions. Given the reported relationships among the VOP nucleus, neuronal activity, and tremor,21 it is possible that some of the benefit obtained with stimulation in this area occurs because of the effects of stimulation on both the VIM nucleus and the VOP nucleus.
Because efferents from the VIM nucleus project directly to the primary motor cortex and, to a lesser extent, the premotor and supplementary motor cortices,22,23 the VIM nucleus is in a critical position to affect motor output. In patients with ET, pathologic spike activity in the VIM nucleus is characterized by increased “burstiness” (with 5 to 10 spikes/burst), which for many cells is coherent with the frequency of tremor.24 Whether this activity originates from within the VIM nucleus, arises from recurrent thalamocortical loops, or derives from tremor-related activity occurring in the cerebellum,25 or indirectly from the inferior olivary nucleus26 is unclear. However, it is clear that high-frequency stimulation in regions of the VIM nucleus where burst cells tuned to the tremor frequency are found can have dramatic suppressive effects on tremor. VIM-DBS is typically indicated for suppressing hand tremor, but improvement in vocalization and head tremor has also been observed.27 Leg tremor can be more difficult to treat with thalamic DBS because the relevant region is found laterally in the thalamus near the internal capsule. Patients with PD exhibiting considerable resting tremor have also benefited from VIM-DBS, but the therapy appears to have little beneficial effect on other motor signs.28 Stimulation in regions anterior to the VIM nucleus, including the VOP nucleus and the anterior ventral oral (VOA) nucleus, can improve rigidity and drug-induced dyskinesia similar to that reported for lesions in this region; however, it is unlikely to improve bradykinesia. VIM-DBS can also produce some benefit for patients whose tremor is associated with cerebellar dysfunction or multiple sclerosis, but the magnitude of the effect is often less than that obtained by patients with ET or PD.29 This may be due to the proximal nature of such tremor and the requirement that a greater area of the VIM nucleus be stimulated to cover the somatotopic region contributing to the tremor.30 Using two leads in such cases may provide greater therapeutic benefit by stimulating a greater area of the thalamus.31,32
Parkinson’s Disease
DBS is an effective therapy for PD patients who have responded well to levodopa therapy in the past but now exhibit motor fluctuations, drug-induced dyskinesias, and an unpredictable response to antiparkinsonian medication. In these cases, the therapeutic effects of DBS can be produced by targeting one of a number of different nuclei or fiber pathways (see Fig. 83-3). The most common DBS target for PD patients is the STN, which is positioned 0 to 6 mm below the AC-PC line, in the middle third of the intercommissural distance, and 9 to 15 mm lateral to the midline. To avoid implanting through motor cortical areas and passing through the lateral ventricle, the implantation trajectory is usually oblique in both the coronal (about 15 degrees off the vertical axis) and sagittal (about 20 degrees anterior to the vertical axis) planes. Compared with the STN of normal nonhuman primates, human STN neurons in the parkinsonian state have several atypical characteristics, including increased burstiness (30% to 40% versus 4% to 5%), higher firing rates (25 to 40 versus 15 to 25 spikes/sec), and augmented oscillatory activity in the beta frequencies (15 to 30 Hz).33 During microelectrode mapping, STN neuronal activity is easily identified owing to the density of neurons in the STN compared with the surrounding structures. Neurons in the dorsolateral STN are responsive to passive sensorimotor manipulation and are thought to be one of the primary cellular targets for DBS. However, recent studies suggest that the therapeutic target for STN-DBS also includes regions dorsomedial to the STN,34,35 an area that encompasses the zona incerta,36 pallidofugal fibers passing through the fields of Forel en route to the thalamus and brainstem,18 cerebellothalamic fibers,37 and dopaminergic nigrostriatal fibers projecting to the striatum, pallidum, and STN.38 When the spread of suprathreshold currents envelops nonmotor regions of the STN and other adjacent structures, mood alterations and difficulty performing working memory tasks have been reported (see Table 83-1).39,40
The sensorimotor (posteroventrolateral) internal globus pallidus (GPi) is another therapeutic target for DBS.40,41 The GPi resides in the anterior two thirds of the intercommissural distance, 4 mm below to 8 mm above the intercommissural plane and 7 to 23 mm lateral to the midline. The GPi is a much larger target than the STN (478 versus 158 mm3),42 which may necessitate the use of higher stimulation voltages to produce the same therapeutic benefit; however, it may also provide more opportunities to limit current spread to adjacent regions implicated in the generation of adverse side effects (see Table 83-1). In PD patients, microelectrode mapping studies have demonstrated that spike activity in the GPi is characterized by a higher than normal firing rate (80 to 100 Hz), increased burstiness, and elevated power in the beta oscillation frequency.33,43
The external globus pallidus (GPe) is also an effective target for managing parkinsonian motor symptoms,13 and it is an even larger structure than the GPi (808 mm3).42 Neurons in the GPe exhibit increased burstiness with long pauses in between,43 whereas regions between the pallidal segments are identifiable by the presence of large, regularly firing neurons called border cells. Studies suggest that contact electrodes near the border of the GPe-GPi or entirely within the GPe may be more effective at relieving bradykinesia than stimulation with contacts entirely within the GPi, a site considered optimal for mitigating levodopa-induced dyskinesias.44–46 These studies found that in contrast to bradykinesia, rigidity was reduced no matter which pallidal segment was stimulated, suggesting that the underlying therapeutic mechanism for rigidity may be different from that for bradykinesia. It is possible, however, that the lack of benefit on akinesia observed with GPi-DBS resulted from stimulation current spreading to the internal capsule. Indeed, studies in the nonhuman primate model of PD have demonstrated worsening bradykinesia in the face of improving rigidity during stimulation at voltages near the threshold for activation of the corticospinal tract.47 Another possibility, although less likely, is that these effects stem from the modulation of different motor subcircuits within the pallidum. Hoover and Strick22,48 showed through transsynaptic anatomic tracers that the motor cortex and supplementary motor areas target different subregions of the sensorimotor pallidum. Along these lines, electrical stimulation of the motor cortex modulated neural activity in the posteroventral globus pallidus, which is related to the execution of movement, whereas stimulation of the supplementary motor areas affected neurons in the anterodorsal pallidum, which is related to the planning of movement.49,50
Gait initiation and balance disturbances are often observed in patients with PD, but neither STN-DBS nor GPi-DBS provides uniform benefits for all patients, particularly for medically refractory gait and balance disorders. Recent reports suggest that the pedunculopontine nucleus (PPN) may be a potential DBS target for improving drug-resistant gait and balance problems.51 The PPN has a central role in the mesencephalic locomotor region, and during continuous electrical stimulation in animal models (20 to 60 Hz), locomotion and muscle tone are increased.52,53 A few patients have now been implanted with PPN-DBS systems, but improvement in gait has not been consistent.54 This variability may stem from the fact that in humans we do not know what pathologic activity exists in the PPN, which regions of the PPN when stimulated provide a therapeutic benefit, and what stereotactic coordinates and stimulation parameters to use to minimize paresthesias and other side effects.55,56 Moreover, the capacity for DBS to modulate PPN output in some patients may be confounded by progressive PPN degeneration over time. Histologic analysis, for example, indicates that approximately 50% of the large cholinergic neurons in the PPN pars compacta degenerate in PD.51
Dystonia
In patients with medication-resistant primary generalized dystonia, segmental dystonia, and complex cervical dystonia, DBS of the posteroventrolateral (sensorimotor) GPi is the preferred surgical treatment, largely displacing pallidotomy.57 At present, it is unclear what neural elements underlie the therapeutic benefit for either phasic or tonic dystonia, especially because DBS alleviates phasic dystonia more readily and more quickly than tonic dystonia. One likely hypothesis is that DBS affects the output of GPi efferents projecting to the thalamus and brainstem. Indeed, these cells have a significant degree of abnormal firing patterns in patients with dystonia, which is also observed in the GPi of PD patients. In both cases, there is increased burstiness, but in dystonia, GPi neurons exhibit, on average, lower than normal firing rates and increased oscillatory activity in the 4- to 10-Hz band.58,59
Other anatomic targets for treating dystonia with DBS include the VIM nucleus, the VOP nucleus, and the VOA nucleus. Given the success of GPi-DBS for treating primary dystonia, however, there has been little movement toward thalamic targets, despite the report of some success with thalamotomy for dystonia.60,61 Moreover, when thalamic DBS has been attempted, the results have been mixed. This may be due to the fact that most of these patients had secondary dystonia,62 which, other than tardive dystonia,63 does not respond well to DBS.57,64 For focal dystonia, one group in Italy reported that extradural motor cortex stimulation can provide some therapeutic benefit.65 Sun and colleagues66 reported that STN-DBS can be effective for managing dystonia, arguing that in their patients, STN-DBS provided faster improvement of dystonic symptoms, with lower voltages necessary to elicit the improvement, compared with GPi-DBS. Although control of symptoms was no better with STN-DBS than with GPi-DBS, their GPi-DBS patients did not do as well as others reported in the literature. Double-blind randomized clinical trials comparing the long-term effects of STN- and GPi-DBS in dystonia patients may be necessary to determine the validity of their assertion.
Neuropsychiatric Indications
The pathophysiology of Tourette’s syndrome is thought to include dysfunction of both associative-limbic and sensorimotor circuits. Not surprisingly, anatomic targets for DBS have followed this multimodal composition. Vandewalle and colleagues67 first reported that bilateral stimulation of the medial thalamus reduced tic severity in a patient with Tourette’s syndrome. Their coordinates (5 mm lateral of the AC-PC line and 4 mm posterior of the midcommissural point, at the level of the AC-PC plane) were based on those described by Hassler and Dieckmann68 for partially suppressing tics with bilateral lesions of the median and rostral intralaminar thalamic nuclei and the internal VOA nucleus.68 Other regions in the associative-limbic network have also proved effective, including the centromedian-parafascicular complex of the thalamus and the associative-limbic ventromedial GPi.69,70 The motor territory of the GPi and GPe is reportedly an effective stimulation target as well,71,72 which parallels the improvement seen with sensorimotor GPi-DBS in patients with hyperkinetic and hypokinetic movement disorders.
Two other emerging indications for DBS are medication-refractory obsessive-compulsive disorder and depression—both of which have targets in the anterior limb of the internal capsule. Nuttin and colleagues73 were the first to report the use of DBS at the anterior limb of the internal capsule for treating obsessive-compulsive disorder in 1999. This fiber tract contains numerous projections to and from frontal cortical areas. The limbic basal ganglia (caudate nucleus, nucleus accumbens, GPi, STN) are also thought to be involved in obsessive-compulsive disorder, and stimulation in these regions has produced therapeutic effects in some patients. DBS of the anterior limb of the internal capsule as well as the anterior cingulate cortex (Cg25) reportedly improves the symptoms of depression.16 The trajectory used for Cg25 places the distal and proximal contacts in cortical gray matter, with the remaining two middle contacts in white matter. Interestingly, the contacts used in these studies tended to be the two middle contacts, which, when activated, would be expected to affect both afferent and efferent projections of Cg25.
Neural Responses
The similarity in therapeutic outcomes derived from surgical ablation and DBS at the same targets led to the hypothesis that high-frequency stimulation inhibits output from the stimulated nucleus.8,10 Recent studies have challenged this view, however, suggesting that instead of suppressing output, DBS overdrives the output of neurons and fibers of passage near the active electrode.74 This process occurs because the threshold for eliciting action potentials is significantly lower in the axon than in the cell body,75 which means that even if somatic and initial segment processes are suppressed by electrical stimulation, action potentials can still be evoked in axonal efferents downstream.76 The dissociation between somatic and axonal responses to DBS has been confirmed experimentally for a number of different DBS target structures.
Changes in Somatic Activity in the Stimulated Nucleus
Inhibition of somatic activity during DBS has been observed via microelectrode recordings in the STN77,78 and GPi79,80 of human patients. Several hypotheses have been proposed to explain this inhibition, including (1) depolarization block resulting from the inactivation of sodium channels81,82 or an increase in potassium current,83 (2) presynaptic depression of excitatory afferents,84 and (3) stimulation-induced activation of inhibitory afferents.85 Evidence to support the depolarization block hypothesis is derived primarily from brain slice recordings. Following the onset of high-frequency stimulation in the STN of rat brain slices, for example, neurons in the STN initially show an elevated firing rate, after which these cells fail to be driven by the stimulation.86 Brain slices, however, inevitably sever connections from afferent input structures. Several in vivo studies have noted that although STN or GPi neurons are directly inhibited by high-frequency stimulation, their firing probabilities are not completely suppressed.78,79,87 Other studies have indicated that somatic inhibition can occur even after a single stimulation pulse,77 with both inhibition and recovery from inhibition occurring at latencies consistent with the kinetics of GABAergic synapses.87 If one considers the hypothesis of driving afferent input with DBS, the observation of decreased activity in the stimulated nucleus (STN or GPi) is not surprising. The majority of afferents to these structures are GABAergic (ratio of 9 : 1 for the GPi),88 forming dense synaptic baskets around the soma and proximal dendrites. In the VIM nucleus, which contains mostly excitatory input, it was reported that approximately half of all recorded neurons had higher than normal firing rates during stimulation.89
Yet, simply driving synaptic input does not explain the variety of somatic responses elicited by DBS. For instance, a small fraction of cells in the STN or GPi shows an excitatory response to local high-frequency stimulation.90,91
