CHAPTER 80 Subthalamic Deep Brain Stimulation for Parkinson’s Disease
Surgery was the only effective treatment of PD until the 1960s, when Cotzias introduced treatment with levodopa based on the pioneering work of Arvid Carlson. PD has been known to be related to a deficit of dopamine in the basal ganglia since the report of Ehringer and Hornykiewicz.1 The drawbacks of ablative surgery, when contrasted with the strikingly beneficial effects of levodopa, were responsible for the almost total disappearance of ablative lesions until recognition of the long-term side effects of levodopa therapy, mainly motor fluctuations and dyskinesias, triggered renewed interest in surgical methods, but with no or little tolerance for complications. The need for less invasive techniques generated a series of basic research–based approaches. This was the main motivation for neural transplantation, which is still not considered routine treatment. In this context of a quest for a new procedure, the unexpected observation during intraoperative exploratory electrophysiology that high-frequency stimulation (HFS) was able to mimic in a reversible and adjustable manner the effects of local destruction of functional targets2 serendipitously led to the development of HFS of functional targets as a specific therapeutic tool. VIM-HFS was demonstrated to alleviate the most spectacular symptom of PD—tremor. However, the therapeutic effect was soon clearly recognized as being almost strictly limited to improvement of tremor with no effect on bradykinesia and rigidity. In actuality, although tremor is the most visible of the symptoms of PD, it is not the most disabling; the difficulties of patients with advanced parkinsonism are essentially related to akinesia and rigidity. Reports in the early 1990s by groups led by Mahlon DeLong3 and Crossman4 of the prominent role of the subthalamic nucleus (STN) in the control of motor function and the ability of STN destruction to improve bradykinesia and rigidity in 1-methyl-4-phenyl-1,2,3,6-tetrahydorpyridine (MPTP)-treated monkeys suggested that these symptoms could be also controlled. However, because of its reputation as a source of hemiballismus when destroyed by hemorrhage, the STN was not a very attractive surgical target.5 For this reason, subthalamotomy was not a good procedure to exploit this basic science discovery.
The experience acquired during HFS of the thalamus suggested that the STN could be a target for neuroinhibition-type methods as provided by HFS. This was supported by experiments in MPTP-treated monkeys, but with the use of HFS instead of lesioning.6 The ultimate validation of this target occurred when the first patients with advanced PD underwent implantation; the tremor, rigidity, and bradykinesia were very significantly improved by this method,7 which allowed the dosage of levodopa to be decreased by 60% on average8 and therefore alleviated the levodopa-induced motor fluctuations and dyskinesias.9 Since that time, several thousands of patients have been operated on all over the world and demonstrated improvement, thus making this method the reference surgical procedure for advanced PD.
What is the Subthalamic Nucleus? Anatomy and Physiology
Anatomy
The STN was discovered and described by Luys in 1865.
Shape and Situation
The STN, or corpus Luysi, is oval shaped and lies on the inner surface of the peduncular portion of the internal capsule. Caudally, the medial part of the nucleus overlies the rostral portion of the substantia nigra. Lesions of the STN in primates, in whom it is particularly well developed, generate very distinct clinical symptoms,10–12 and it is extraordinary that the dysfunction of such a small collection of cells, with apparent neurochemical simplicity and connectivity, can disturb normal neurological functions to such an extent.
Texture: Cytoarchitectonics, Histology, Histochemistry
The STN has a high density of dark cells on Nissl-stained sections; its boundaries stand out clearly from the surrounding diencephalic structures. The cells of the STN have been described as having either spindle-shaped, pyramidal, or rounded somata. Each gives rise to several spiny dendritic processes aligned parallel with the rostrocaudal axis of the nucleus. These processes may be very long (up to nearly 1 mm), and they remain within the boundaries of the STN in primates, whereas in nonprimates, they tend to “trespass” onto the territories of adjacent structures, such as the zona incerta. The majority of STN cells project to other neural centers, with only a small population of interneurons. The dominant neurotransmitter within STN cells is glutamate, but other neurochemicals (parvalbumin [calcium-buffering protein] and nitric oxide synthase) have also been reported. Although classically the STN was thought be GABAergic (secreting γ-aminobutyric acid) and inhibitory, most, if not all STN projections have now been revealed, in contradistinction, to be glutamatergic, and they exert extremely powerful excitatory effects on their target structures. For this reason, among others, the STN has been suggested to be a major driving force and central feature of the basal ganglia circuitry.10–13
Connections
The STN connections with other parts of the brain have been well described in nonprimates, as well as in primates. In general, its major connections are with the primary motor cortex, the globus pallidus pars externa (GPe), the substantia nigra pars reticulata (SNr), and the pedunculopontine tegmental nucleus (PPN); its minor connections are with the striatum, intralaminar nuclei of the thalamus, and various brainstem nuclei.10–13 The STN has been reported to have other, generally more sparse connections with various neural centers, including the intralaminar nuclei of the thalamus and the dorsal raphe, as well as a direct projection to the striatum.11–17
Afferents
Cortex
The cortex provides major input to the STN that is rather selective and arises from layer V, mainly from the primary motor cortex and, to a lesser extent, from the prefrontal cortex. The input is primarily via collaterals of axons terminating elsewhere.11 These cortical axons terminate in small boutons on spines and fine dendrites of the STN cells. Through such input the STN is thought to be the pivotal nucleus of the basal ganglia through which the cortex influences the output of the basal ganglia.10–13
Efferents
Substantia Nigra Complex
The STN also projects to the other basal ganglia output nucleus, the SNr. The STN axonal terminals in the SNr have again been shown to be glutamatergic and form asymmetric synapses with dendritic shifts and, to a lesser extent, somata. There have also been reports of a direct but sparse STN projection to the substantia nigra pars compacta (SNc),11,19 which could still generate glutamate excitotoxicity in the parkinsonian condition. In this context, it should be noted that the STN projections to the SNr are thought to be strongly related to the SNr projection to the SNc.11
Pedunculopontine Tegmental Nucleus
In view of recent experimental evidence, this brainstem nucleus has become included by most authors in the basal ganglia scheme and might be a contributing factor in generating the overactivity of the STN in patients with PD.20 Reciprocal connections between the PPN and STN have been documented in both primates and nonprimates. Individual STN cells project directly to the PPN (pars compacta portion) and not elsewhere, whereas a distinct population of STN cells project to the basal ganglia output nuclei, the globus pallidus and SNr.
Physiology
Electrophysiology
STN cells have high spontaneous tonic activity with irregular bursts of activity.21 During a typical electrode track through the brain, the STN is characterized by a sudden increase in background noise (from the relatively quiet of the zona incerta) as a result of the high density of bursting cells. Location of the electrode within the sensorimotor sector of the STN is often verified by the response of cells (increased firing audible through microrecording) to passive movement of the contralateral limbs.
Functions
As with other major nuclei of the basal ganglia, the STN can be divided into distinct functional sectors: (1) a large sensorimotor sector that occupies the dorsolateral regions of the nucleus—STN cells in this sector respond to somatosensory stimulation by changing discharge rates during movement of the contralateral limbs (in fact, there is a segregated map of the body found within the STN); (2) a small associative territory that occupies the ventromedial region of the nucleus—cells here are activated during visual oculomotor tasks; and (3) a limbic sector that occupies the medial tip of the nucleus—this sector receives input from the limbic cortex and ventral regions of the globus pallidus. In primates, these sectors involve distinct cell groups that respond to the different types of stimuli; in nonprimates, however, the sectors are less clear, perhaps because individual STN cells tend to have multiple projection sites and input.10–13
Dysfunction of the Subthalamic Nucleus Leads to Motor Disorders
Overexcitation
Conversely, overactivity of the STN (oscillatory bursts and synchronicity) induces PD symptoms (akinesia, rigidity, and tremor).11 In PD patients, as well as in two well-known animal models of PD—6-hydroxydopamine–lesioned rats and MPTP-treated monkeys—there is an increase in the firing rate of STN cells. Simultaneously, STN cells change their pattern of firing and exhibit an increase in oscillatory bursts and synchronicity,11,12,14,15,22 and MPTP-treated monkeys have an increase in glucose metabolism and mitochondrial enzyme activity in the STN. In fact, STN cells enhance their activity during the course of MPTP treatment, even before the first appearance of clinical signs.24 Furthermore, in PD patients and MPTP-treated monkeys, both lesioning and high-frequency deep brain stimulation (DBS) of the STN alleviate the parkinsonian symptoms, presumably by reducing the activity of the basal ganglia output streams. For these reasons, the STN is viewed as a centerpiece of pathophysiology in PD and has become a popular surgical target for the relief of motor symptoms.25,26
Chemical Modulation
Because the SNr sends projections to the SNc, the STN may “switch off” the dopaminergic SNc cells via its heavy projection to the SNr. Hence, during DBS of the STN, which inhibits the abnormally overactive STN cells in PD, there is an increase in the activity of SNc dopaminergic neurons27 and dopaminergic levels in the striatum.28 This may counteract the direct STN projection to the SNc, which is weaker than the SNr projection,19 but may not necessarily prevent the release of glutamate onto SNc cells by the overactive STN in patients with PD (and hence not prevent glutamate excitotoxicity).
Rationale for Using the Subthalamic Nucleus as a Target
Animal Experiments
Improvement of MPTP-induced parkinsonism has been achieved by inducing lesions of the STN electrolytically3,4 and by STN-HFS in the same model.6
Strategic Situation
The STN is functionally situated at a crossroad on the output of the basal ganglia, which may explain its major role in the control of motricity; the basal ganglia are closely involved in the control of movements and subsequently in the genesis of movement disorders when a component of this network is altered by a pathologic process, such as the nigrostriatal degeneration that is at the basis of PD. This results in dysregulation of the function of these networks by disrupting the various nodes of the networks and most often results in an abnormal firing pattern consisting of a combination of hyperactivity and bursting, thereby generating irregular activity, the most demonstrative being associated with tremor and observable in several nuclei of the network, such as the thalamus, the STN, and the GPi. A simplistic approach toward therapeutic application is to consider that alleviation of symptoms could be achieved by altering these abnormal activities, either by suppression (via ablative surgery) or by functional inhibition, such as by delivering a sort of “blank noise” via HFS that would jam, mask, or disrupt the abnormal burst-like hyperactivity. Because the STN projects to both the globus pallidus and SNr, it is thought to influence the activity of cells that constitute the output systems of the basal ganglia. These STN efferents may control the initiation of movement by setting the physiologic conditions (e.g., membrane potential) of the pallidal and SNr cells to appropriate levels before arrival of the striatal signals.11 Thus, the STN is in a position to influence the whole net output of the basal ganglia. In particular, the cortex is thought to drive the basal ganglia circuitry, not only through its classic input to the striatum but also through the STN.
How to Interfere with the Target
Lesioning
Lesioning is the most classic approach, as suggested by clinical observation and the aforementioned animal experiments, but lesioning of the STN in humans induces hemiballismus.29
Chemical Modulation
Chemical modulation by glutamate antagonists such as MK-801 has been used to induce temporary inactivation,30 and neurotoxins (kainic acid, ibotenic acid) are used to generate a permanent lesion in neurons or fibers, or both.
Electrical Modulation
Rationale
The rationale for using STN-HFS to treat PD comes from the application of HFS for the treatment of tremor, both parkinsonian tremor, which has been performed since 1987,31 and essential tremor.32 The fact that replication of the effects of lesions in the STN with HFS for the treatment of MPTP-induced parkinsonism in monkeys3,4 proved to be reversible, titratable, and safe in treating parkinsonian and essential tremor provided experimental support to attempt it in humans, which was done in 1993.7,33
Mode of Action
HFS mimics the effects of ablative lesions in all the targets that have been used thus far (thalamus, pallidum, STN, accumbens nucleus, hypothalamus, and others). However, HFS does not create a lesion; all effects are created by stimulation, and all improvements are only temporary and last as long as the duration of the stimulation. Another difference from lesioning is that HFS probably induces specific effects by excitation or inhibition of structures at a distance, such as the motor or sensory side effects obtained by diffusion to the motor pyramidal tract or to the lemniscus medialis fibers. The mechanism of action of DBS at high frequency is still completely unknown 22 years after its introduction. It is independent of the target because DBS mimics the effects of ablation in all targets used thus far, but it does not create a lesion and all of its effects occur only during stimulation. A large series of papers have been published on the mechanism of action of HFS, and there is an important trend in basic research aimed at this goal. One can surmise that several submechanisms are probably involved in combination to produce functional inhibition34: (1) jamming of the neuronal message transiting through the stimulated structure32 is highly probable, and we have suggested that it might take place in the very early stages of the procedure2,32,34 and could be responsible for desynchronizing abnormal oscillations expressed by the bursting patterns.14,35 This could be consistent with the observation that pathologic states such as PD create synchronization of neuronal activities, which are expressed by bursting patterns visible during microrecording, as shown in the pallidum in monkeys.14 This is supported by experimental data obtained in monkeys showing that STN-HFS resets the subthalamic firing and reduces the abnormal oscillations of STN neurons.35,36 (2) The hypothesis of extinction or strong inhibition of neuronal firing is also supported by direct observation of a decrease in the discharge rate during stimulation.37–39 (3) Dual effects could also be the mechanism, such as associated excitation and induction of high-frequency bursts.40 (4) Finally, it has been shown in our laboratory that HFS inhibits the production or release, or both, of certain neurotransmitters and hormones in culture,41 thus suggesting that this phenomenon could participate in the functional inhibition induced by HFS. Neuronal mechanisms that could be at the origin of the paradoxical inhibitory-like effect of HFS have recently been reviewed.40
Materials and Methods
Surgery
The procedure varies among teams, depending on their equipment and usual practices.
Operative Techniques
The procedure detailed in this chapter is the one followed in Grenoble and describes the state of the art of the method by our team, as applied to all targets (STN, GPi, thalamic VIM, PPN, subgenual cortex CG25). There are obviously other ways to perform it, as reported by other teams, and depending on future technical improvements, such methods might be undertaken by our own team. For instance, the implantation session is generally performed with the patient under local anesthesia, whereas some teams are using general anesthesia to decrease the stress and pain42 but lose intraoperative observation of the clinical benefits of DBS.
Stereotactic Frame and Robotized Arm
At our institution we use a robotized arm that allows high-precision positioning of the guide tube along the trajectory. The number and types of these devices are slowly increasing, but until recently the NeuroMate (Schaerer-Mayfield, Lyon, France) and more recently the Rosa (Medtech, Montpellier, France) were the only ones available (Fig. 80-1).
Preoperative Imaging and Planning
Stereotactic x-Ray Ventriculography
Stereotactic ventriculography (Fig. 80-2) is still performed by some teams under general anesthesia,43 as by ours, although a large number of centers do not use it for fear of complications or because they consider MRI localization to be satisfactory. The problem of MRI distortion, which is the major reason to continue performing ventriculography, is still rarely addressed. After a small skin incision, a hole through the skull, 9 cm from the nasion and 2.5 cm from the midline, is created with a twist drill. The right frontal horn of the ventricle is tapped with a Cushing cannula at a depth of 6.5 cm from the skin surface. A 2-mL air bubble is injected to check for correct placement of the tip of the cannula. During the injection of 6.5 mL of contrast medium (Iopamiro, Schering), a 12-second sequence of x-ray images is recorded in the lateral direction, immediately followed by radiographs taken in the anteroposterior direction. These x-ray images provide internal landmarks for the third ventricle, to which various atlases and coordinates of targets can be related.
Magnetic Resonance Imaging and Atlas Fusion
Stereotactic MRI is performed on awake patients with an MRI-compatible stereotactic head holder and localizer. This provides direct visualization of the STN target, which is visible on T2-weighted sequences as a hypointense signal surrounded by white matter (zona incerta above and fields of Forel bundles below) separating the STN from the SNr (Fig. 80-3). MRI, eventually merged to the ventriculographic images, is used to plan the procedure. The STN stereotactic target can be also constructed with graphic tools included in the navigation software.
Coordinates of the Target and Entry Point
Construction of the target is based on ventriculographic landmarks—the anterior commissure (AC), posterior commissure (PC), height of the thalamus (floor of the lateral ventricle), and midline of the third ventricle (Fig. 80-4, Table 80-1)—which allows automatic drawing of the intended target (in the present situation the STN, but the coordinates of any other stereotactic target can be introduced in the software). This x-ray target is fused with the MRI scans (three-dimensional axial T1- and coronal T2-weighted, with injection of contrast material) that are imported into the Voxim software of the robotized arm (NeuroMate) through the hospital image network. Importation of both the MRI and x-ray images obtained stereotactically in the same patient on the same day allows one to check the coherence of x-ray and MRI data by matching two anatomic structures, AC and PC, which are clearly visible with both modalities (Fig. 80-5). They usually match, but there is often a non-negligible and unpredictable discrepancy of up to 2 to 3 mm between AC images, for instance, thus stressing the still unresolved problem of MRI-based localization. If it is satisfactory, one checks for matching between the “theoretical” ventriculography-based STN target and the T2-weighted MRI actual scan of the STN. If there is a significant mismatch, particularly in the lateral direction, the laterality is corrected, which is not helped by the use of atlases because they are still not always coherent in the three spatial dimensions (Fig. 80-6). When MRI/ventriculography coherence is not satisfactory, the theoretical target is used alone. Planning of the entry point with MRI must avoid the cortical vessels, deep vessels in the sulci, the ventricle if possible, and the caudate nucleus. The planning data are then exported to the NeuroMate robotized arm controller through the neuronavigation software (see Figs. 80-1 and 80-5).
° the height of the thalamus. B, Lateral view. The horizontal axis represents the distance from the posterior commissure (PC) and is 
the length of anterior commissure (AC)-PC; the vertical axis units are in 
° the height of the thalamus. The STN target extends over the middle third of AC-PC, below the line. On the AP view, the average laterality is 12 mm and the STN extends from 11 to 13 mm. On this scheme are projected the image of the average target (black square) and the limits of the standard deviations around the average target (rectangle). On the lateral view is represented the trajectory whose theoretical angle with the AC-PC plane is 63.57 ± 2.34 degrees according to the Guiot scheme; it is actually 63.11 ± 3.87 degrees on the right side and 62.81 ± 4.59 degrees on the left side. The globus pallidus interna target is situated in the anterior third of the AC-PC, from 
+ the height of the thalmus (HT) above to 
+ of HT below the AC-PC plane. Laterality extends 15 to 22 mm from the midline. The ventral intermediate nucleus is 14 mm from the midline (ML) and is about 
° to 
° of the AC-PC distance ahead of the PC.
Targeting and Electrode Implantation
The implantation session is performed 3 days after pretargeting with the patient under local anesthesia. The patient is reinstalled on the frame and the pins reinserted into the hollow screws in accordance with the previous four vernier readings. Correct placement is checked on radiographs. Using the preplanning data, which have been stored in the neuronavigation software, the NeuroMate is launched and reaches the preplanned position on the first side to be operated (in general, the side contralateral to the worst clinical side). An arciform skin incision centered on the entry point is made. The skin, subcutaneous tissue, and periosteum are retracted en bloc from the skull by a rugination, and a 6- or 9-mm burr hole is made through the NeuroMate tool holder. The electrode guide tool (Ben Gun with five parallel channels, distant 2-mm axis-to-axis accommodating guide tubes, and long-term DBS electrodes 1.27 mm in diameter) is inserted down to the dura, which is not opened. The microelectrode guide tubes are introduced by perforating the dura matter with sharp stylets and then lowered into the brain with blunt stylets. When the stylets are removed, they are replaced by microelectrodes (tip diameter, 1 µm; impedance, 1 to 10 MΩ; FHC Bowdoinham, ME) in their own guide tubes and inserted 15 mm above the target point under radiographic guidance (Fig. 80-7). They are connected to the electronic stages of the data acquisition and processing system (Alpha Omega NeuroTrek) and moved toward the target with a micromanipulator.
Electrophysiology
Electrophysiologic Exploration
Two types of electrophysiologic investigation are performed:
The electrophysiologic signature of the STN units, as reported in literature,44,45 is typically composed of asymmetric spikes at rather high frequency, and it exhibits burst patterns in PD (Fig. 80-8). They respond to passive contralateral limb movements and proprioceptive input, such as muscle pressure, and exhibit synchronous contralateral tremor activity. Recording length in the STN varies from 5 to 6 mm between two silent zones corresponding to white matter, the first one between 0 and 2 to 3 mm below the AC-PC plane (subthalamic area and anterior zona incerta) and the other one corresponding to the white matter between 9 and 11 mm just above the SNr. SNr neurons fire in regular, symmetrical, large-amplitude spikes that are generally unresponsive to external stimuli (see Fig. 80-8).
Microstimulation: Looking for Beneficial Effects and Side Effects
Microstimulation can be performed with the microelectrode that has been used to record neuronal activity at current intensities of up to 10 mA for short (10 to 30 seconds) periods. Microstimulation is essential because it allows observation of the beneficial effects (improvement of PD symptoms) inside the target and the side effects (limiting factors for efficient stimulation) outside it.46
Final Choice: Synthesis of Benefits and Side Effects
When the best track (best beneficial effects, least side effects, largest security margin between the threshold for improvement and side effects) has been identified, the corresponding microelectrode is removed and replaced with a final lead (DBS 3389, 1.5-mm contact length, 0.5-mm spacing, 1.27-mm diameter) sutured to the rim of the bur hole (made with a small oblique twist drill) and, after removing the other microelectrodes, embedded in dental cement to prevent leakage of cerebrospinal fluid (CSF) and infection back along the track. The external part of the electrode is folded under the periosteum and the skin is sutured (Fig. 80-9). The NeuroMate is then aimed at the other side for a similarly preplanned procedure. At the end of the procedure, final x-ray images are taken for use in determination of electrode position and statistical evaluation of the most effective contacts (Fig. 80-10). The question of unilateral STN-DBS, which is often raised because of the alledgedly lower morbidity47 in symptomatically asymmetric patients and the observation of ipsilateral effects of unilateral STN stimulation,48 still needs definitive resolution through a large and well-conducted and controlled clinical trial.
