Globus pallidus interna

The lateral position of the GPi relative to the STN minimizes the risk that a parasagittal approach will violate the medially located vascular structures. At the same time, this approach is able to capitalize on the posterolateral position of the motor region of the nucleus [4]. The trajectory is planned using an angle of 0–15° in the coronal plane and 30–35° anteriorly in the sagittal plane.

Posteromedial to the GPi is the posterior limb of the internal capsule. Suboptimal electrode placement in this location will induce tetanic contractions. The globus pallidus externus (GPe) lies laterally to the GPi. The internal medullary lamina partitions the internal and external segments. Below the GPi runs the ansa lenticularis anteromedially and the optic tract posteriorly and laterally. The optic tract serves as a landmark to guide electrode depth. Anteriorly, the nucleus basalis lies deep to the GPi.

On entering the GPe, there is an increase in firing activity. The GPe contains neurons that fire tonically at high frequency (50 Hz), interspersed with short pauses. Another population of neurons is identified by low-frequency activity (18 Hz) intermixed with periods of high frequency [5]. The internal medullary lamina is identified by a decrease in firing for 1–2 mm and border cells with regular tonic firing at 5–30 Hz [6]. Parkinson’s disease patients have high-frequency firing in the GPi with fewer pauses than recorded from the GPe. The sensorimotor GPi demonstrates driving activity with limb movement. The optic tract, located approximately 1.5 mm below the GPi, will fire when a flashlight is directed at the eyes. The optimal electrode location is far from the internal capsule medially and optic tract inferiorly to prevent side effects at the high voltages sometimes needed to improve motor symptoms [5]. The use of GPi-DBS may have a local inhibitory effect that decreases pathological overactivity or modulates firing patterns [7, 8].

Subthalamic nucleus

Bordering the STN anterolaterally is the internal capsule. Medial and deep to the STN is the red nucleus and third nerve nucleus. Posteriorly run the prelemniscal radiations. The motor units of the STN are arranged within the dorsal, posterolateral sector. The proximity of the internal capsule to the optimal target location complicates electrode insertion. A “routine” MER track will cross the thalamus, recording bursting (15–19 Hz) or tonically active (28 Hz) cells. After passing through the zona incerta, the STN is entered. An MER of the STN is characterized by an increase in background noise. A tract length of 4–5 mm is generally sought before implantation.

Due to the more medial position of the STN compared with the GPi, the trajectory is often angled to avoid the deep vascular structures and lateral ventricle. Enlarged ventricles are especially difficult to avoid. When an extreme lateral trajectory (greater than 30°) is selected, the path of the microelectrode will often miss the thalamus entirely and the STN units recorded may span only 2–3 mm. Imaging may dictate the GPi as a better target in such cases.

Stimulation of the globus pallidus interna and the subthalamic nucleus

Intraoperative stimulation is used to determine voltage requirements for therapeutic effect and voltage-limiting side effects [4]. Reduction of PD symptoms with intraoperative stimulation is less frequent with the GPi. However, intraoperative stimulation of both targets determines the threshold for adverse effects. Stimulation-related side effects emerge when the current spreads to corticobulbar, corticospinal or optic tracts. Stimulation of the deep contacts on the lead may cause visual symptoms during GPi stimulation [4], and STN stimulation of deep contacts may activate the third nerve nucleus or red nucleus, triggering dysconjugate or conjugate gaze paresis.

Stimulation spread to the internal capsule induces contralateral motor contractions. Specific to GPi-DBS, a decreased blood oxygen level-dependent (BOLD) signal is seen in the contralateral motor cortex and subcortical structures such as the striatal and thalamic regions when the electrode location is close to the dorsomedial border of the GPi and the neighboring internal capsule [9]. This may be due to effects on internal capsule fibers and inhibitory corticocortical projection neurons.

The volume of the GPi is approximately three times that of the STN. It has been suggested that stimulation efficacy may be related to the volume of the area stimulated. As the STN is significantly smaller than the GPi, it follows that a larger proportion of the motor region is likely to be stimulated at lower voltages [10]. Stimulation of the smaller STN is also more likely to spread to nearby structures and results in adverse effects [11]. Due to the large size of the GPi, stimulation may not elicit intraoperative side effects during testing [10]. Furthermore, the clinical effect of GPi stimulation is often delayed. Taken together, this limits the predictive value of intraoperative stimulation when the GPi is the target [10]. The microlesional effect that often presages symptom control in the STN even before programming is less evident following MER and lead placement in the GPi. This may also be related to the large nucleus size [12].

Tremor

Tremor-predominant PD may respond better to STN or thalamic DBS than GPi-DBS. After STN-DBS, tremor improvement is estimated to be 90% at 1 year and 75% at 5 years [16, 17]. The use of GPi-DBS results in 80% improvement after 1 year. This disparity in long-term tremor control between the two targets may be related to the size of the GPi. If sufficient current density is unable to spread to the posterior portion of the nucleus, tremor control may be suboptimal [11]. When hand dexterity is studied, it is noted to be dependent on the medication state. Thus, GPi-DBS improves hand dexterity during the on-medications state; conversely, STN-DBS improves hand dexterity in the off-medications state [18].

Bradykinesia

Bradykinesia, when the predominant symptom, may also respond better to STN-DBS [16]. An improvement of 70–80% is expected after STN-DBS compared with a 30–40% improvement after GPi-DBS [19]. Bradykinesia affects almost any motor function. Parkinson’s disease patients may have poor oropharyngeal functions necessary to speak and/or swallow. Jaw velocity during planned movements can be used as a surrogate marker of oropharyngeal function. The use of STN-DBS may result in slowed jaw velocity, while GPi-DBS maintains or improves it [20]. Rigidity improves equally with STN- and GPi-DBS.

Gait and posture

While not cardinal symptoms of PD, postural instability and gait dysfunction affect a significant percentage of patients and respond to STN- and GPi-DBS. In the immediate postoperative period, the improvement is similar, but over time patients with STN-DBS decline. There may be a more durable effect on postural stability with GPi-DBS [21]. Furthermore, control of axial motor symptoms, especially in the off-medication state, is better after GPi-DBS [22]. One study found a significantly greater improvement in the stand–walk–sit test in patients with GPi-DBS compared with those with STN-DBS. Interestingly, this difference in mobility was seen in the off-medication, off-stimulation state [23].

Dyskinesias

Dyskinesias, involuntary movements induced by dopaminergic medications, affect a significant proportion of patients with PD. Dyskinesias may be caused by abnormal patterns of activity within the ventral pallidum. There may be a decrease in dyskinesias following GPi-DBS resulting from interference with such maladaptive patterns of activity or by suppressing GPi output via activation of nearby axons [24]. Deep-brain stimulation of the ventral pallidum may reduce dyskinesias but increase akinesias, whereas dorsal GPi stimulation has been suggested to have the opposite effect [25].

The incidence of levodopa (l-DOPA)-induced dyskinesias varies by age, affecting 50% of young-onset PD patients, compared with 16% of those over the age of 70 [26]. There is no standard levodopa dose that induces dyskinesias; the effect varies from patient to patient. The use of GPi-DBS has a direct antidyskinetic effect [25, 27], and some studies suggest it may be more effective for dyskinesia reduction than STN-DBS [24, 28]. In one study of patients implanted with both STN and GPi electrodes bilaterally, within-subject comparison was possible. The authors reported that, after prolonged medication withdrawal, GPi-DBS significantly reduced apomorphine-induced abnormal involuntary movements, while STN-DBS had no effect [29]. Another study found that GPi-DBS reduced dyskinesias compared with STN-DBS, despite greater medication reduction when the STN was implanted [30]. Other studies have demonstrated a significant reduction in dyskinesias in patients with STN-DBS in both the on- and off-stimulation states. Patients with GPi-DBS require stimulation to be on to experience a reduction of dyskinesias [16].

Long-term symptom control

Some patients with GPi-DBS may show a response reduction with long-term follow-up [25]. This may be related to the variability in clinical outcomes with small differences in lead position [31]. This difference, observed in early studies, has not been replicated and may only be an artifact from early management style [11, 32]. Nevertheless, in a young patient, with a long life expectancy, a potential for a decline in the response rate should be addressed.

Unified Parkinson’s Disease Rating Scale (UPDRS) III

The effects of STN- and GPi-DBS are equivalent on general measures of motor function such as the UPDRS-III [23, 33]. However, the degree of motor symptom improvement is target dependent. This effect is evident when comparing on- versus off-stimulation state. A large randomized, controlled, blinded trial evaluating STN- versus GPi-DBS in 299 patients with PD did not find a significant difference in UPDRS-III scores between groups at 24 months when patients were on stimulation but off medication [23]. It did discover a difference in the off-stimulation, off-medication state. Without stimulation or medication, GPi patients had a small improvement in UPDRS-III. This was significantly different from the small decline found in STN patients [23]. Others have reported a similar effect [32]. This may be related to a longer duration of action and washout time for GPi-DBS. A similar effect has been described in the on-stimulation, on-medication state [32]. Additionally, in this state, patients with GPi-DBS demonstrated an improvement in UPDRS-II (activities of daily living) scores, while those with STN-DBS did not [33].

Cognition

Cognitive decline occurs in many patients with advanced PD. For this reason, patients submit to a neuropsychological evaluation during the evaluation for surgical treatment. Following STN-DBS, those with minimal or no measurable cognitive changes preoperatively may worsen in the postoperative period. This effect is not as robust after GPi surgery [10]. In a randomized study of patients with GPi- or STN-DBS, slight declines in performance on the Mattis Dementia Scale at 36 months were detected in both groups. However, the decline was significantly greater for STN-DBS [32]. Similarly, at 36 months, performance on the Hopkins Verbal Learning Test was worse after STN-DBS [32]. In a small study that randomized patients to STN- or GPi-DBS, a higher rate of cognitive decline was noted in STN-DBS patients [16]. This effect, it seems, is more pronounced in older patients and in those with moderate baseline cognitive impairment preoperatively. For patients in whom STN-DBS is favored to treat motor symptoms, a staged bilateral procedure may be desirable [35]. Repeat neuropsychological evaluation may be warranted prior to placing the second electrode. At least one study has reported that, following GPi-DBS, confusion scores increased postoperatively. However, not all patients in the study were randomized to a surgical target. Because GPi-DBS is often selected in patients with worse preoperative cognitive function, this finding is most likely related to selection bias [36].

The use of STN-DBS appears to affect verbal fluency negatively. In patients randomized to unilateral STN- or GPi-DBS, STN-DBS decreased letter verbal fluency independent of the stimulation state being on or off. This suggests that inserting an electrode into STN displaces or damages the neural elements in a clinically relevant manner [37]. Thus, STN-DBS may be associated with improved executive function but worse performance on tasks that test behavioral switching [38]. This higher-order processing is associated with increased activity in the dorsolateral prefrontal cortex (DLPFC): STN-DBS may augment DLPFC activity indirectly via thalamic projections from the substantia nigra reticulata. This increased DLPFC activity differentiates STN from GPi stimulation [39].

Anger

It has been shown that STN-DBS may interfere with the emotional state of PD patients. One study that randomized patients to unilateral STN- or GPi-DBS discovered that patients with STN-DBS had higher scores on the anger subscale of a visual analog mood scale [37]. However, it has also been reported that both STN- and GPi-DBS increase anger scores. This increase was independent of device activity, suggesting an effect related to local effects of the implant or due to events during surgery. Interestingly, in both STN- and GPi-DBS patients, microelectrode passes were directly correlated with anger scores. This further strengthens the argument that there is a surgical effect underlying the emotional change [36].

Mood

Mood is also more likely to be affected by STN-DBS than GPi-DBS [10]. A large randomized, controlled, blinded trial comparing STN- and GPi-DBS found that neither target resulted in a significant change in Beck Depression Inventory (BDI) scores between preoperative evaluation and 24-month follow-up. However, at the 24-month follow-up, BDI scores were slightly improved in the GPi group and worsened in the STN group, with a significant difference between groups [23].

Some negative effects of STN-DBS including depression, anhedonia and abulia may be related to medication withdrawal [40]. Maintaining a constant dose while still reducing dyskinesias may make the GPi a better target for patients dependent on the beneficial effects of their medications [24]. However, in one study, adverse events were less frequent in GPi patients, but those that did develop psychiatric, behavioral or speech adverse events were on higher doses of dopaminergic medication [41]. This suggests that medication dosing postoperatively requires a careful balance.

Quality of life

There may be greater improvement in quality of life with GPi-DBS than with STN-DBS. In one report, quality of life improved by 38% for GPi-DBS versus 15% for STN-DBS at 6 months postoperatively [42]. The GPi patients had a significant improvement in all four subscales of the 39-Item Parkinson’s Disease Questionnaire (PDQ-39) (mobility, activities of daily living, stigma and social support), while the STN patients improved only in the stigma subscale.