Globus Pallidus Internus (GPi) Deep Brain Stimulation for Parkinson's Disease: Expert Review and Commentary.

INTRODUCTION: The globus pallidus internus (GPi) region has evolved as a potential target for deep brain stimulation (DBS) in Parkinson's disease (PD). DBS of the GPi (GPi DBS) is an established, safe and effective method for addressing many of the motor symptoms associated with advanced PD. It is important that clinicians fully understand this target when considering GPi DBS for individual patients.
METHODS: The literature on GPi DBS in PD has been comprehensively reviewed, including the anatomy, physiology and potential pitfalls that may be encountered during surgical targeting and post-operative management. Here, we review and address the implications of lead location on GPi DBS outcomes. Additionally, we provide a summary of randomized controlled clinical trials conducted on DBS in PD, together with expert commentary on potential applications of the GPi as target. Finally, we highlight future technologies that will likely impact GPi DBS, including closed-loop adaptive approaches (e.g. sensing-stimulating capabilities), advanced methods for image-based targeting and advances in DBS programming, including directional leads and pulse shaping.
RESULTS: There are important disease characteristics and factors to consider prior to selecting the GPi as the DBS target of PD surgery. Prior to and during implantation of the leads it is critical to consider the neuroanatomy, which can be defined through the combination of image-based targeting and intraoperative microelectrode recording strategies. There is an increasing body of literature on GPi DBS in patients with PD suggesting both short- and long-term benefits. Understanding the GPi target can be useful in choosing between the subthalamic (STN), GPi and ventralis intermedius nucleus as lead locations to address the motor symptoms and complications of PD.
CONCLUSION: GPi DBS can be effectively used in select cases of PD. As the ongoing DBS target debate continues (GPi vs. STN as DBS target), clinicians should keep in mind that GPi DBS has been shown to be an effective treatment strategy for a variety of symptoms, including bradykinesia, rigidity and tremor control. GPi DBS also has an important, direct anti-dyskinetic effect. GPi DBS is easier to program in the outpatient setting and will allow for more flexibility in medication adjustments (e.g. levodopa). Emerging technologies, including GPi closed-loop systems, advanced tractography-based targeting and enhanced programming strategies, will likely be future areas of GPi DBS expansion. We conclude that although the GPi as DBS target may not be appropriate for all PD patients, it has specific clinical advantages.

Key Summary Points

Digital Features

This article is published with digital features, including a summary slide, to facilitate understanding of the article. To view digital features for this article go to 10.6084/m9.figshare.13061327.

Introduction

Parkinson’s disease (PD) is a neurodegenerative disorder resulting from progressive loss of nigrostriatal neurons and also from widespread degeneration and deposition of alpha synuclein across multiple basal ganglia networks [1]. Patients may manifest various combinations of the typical motor symptoms, such as tremor, bradykinesia, rigidity, postural instability and gait and balance difficulties, and also non-motor symptoms, such as mood and cognitive difficulties, autonomic dysfunction, speech and swallowing difficulties and sexual dysfunction [1]. Although levodopa remains the gold standard pharmacological treatment for PD, long-term use coupled with disease progression often lead to complications, including dyskinesias and motor fluctuations in approximately 50% of patients at 5 years [2, 3]. These complications significantly impair quality of life and can transform into one of the major sources of disability in PD patients [4]. Invasive approaches, such as ablative procedures [5, 6], carbidopa/levodopa intestinal gel [7], subcutaneous apomorphine infusions and deep brain stimulation (DBS) [8], have been utilized over the past 3 decades to address these PD-related symptoms. Evolving from lesional therapies, DBS is a modality that can be applied to various brain regions, with the globus pallidus interna (GPi) and subthalamic nucleus (STN) being the most common targets of DBS for the management of PD. Occasionally, the ventralis intermedius nucleus (VIM) is used as the target, mainly in tremor-predominant cases where tremor is the sole disability.

The focus of this comprehensive review and expert commentary is on multiple aspects of pallidal stimulation for PD, including the history, anatomical and physiological characteristics of DBS of the GPi (GPi DBS), target selection, surgical planning and patient programming strategies. We also present the evidence on clinical effectiveness and on the potential for next-generation neuromodulatory approaches which may possibly utilize GPi DBS.

This article is based on previously conducted studies and does not contain any studies with human participants or animals performed by any of the authors.

GPi in the History of Deep Brain Stimulation

The first evidence of surgical interventions using the basal ganglia as a potential treatment modality to mitigate symptoms of PD dates back to 1939 with Meyers resecting the anterior two-thirds of the caudate nucleus. The procedure led to improvement of postencephalitic parkinsonian tremor [9, 10]. The subsequent advent of a functional neurosurgery apparatus in 1947 by Spiegel and Wycis (e.g. the head frame) opened the door to the era of stereotactic ablative therapies [11, 12]. Surgeons began to explore lesions of the pallidum and its connections (pallidoanostomies) for various disorders, including PD [11]. Subsequently, additional targets for surgical treatment of PD were also explored, including ligation of the anterior choroidal artery by Cooper [13], chemopallidotomy by Narabayashi and Okuma [14] and pallidotomy via electrical coagulation by Guiot and Brion [15]. These studies were all reported in the early 1950s. Also in the 1950s, Leksell and colleagues began performing stereotactic thermocoagulation-induced pallidotomies, shifting from the initial anterodorsal pallidum approach to posteroventral interventions. These surgeons felt that this shift yielded improved antiparkinsonian benefits [16]. In the same decade, Hassler and Riechert explored the thalamus as an additional surgical target for PD, ultimately resulting in 1954 with successful ablations of the VIM nucleus of the thalamus that improved parkinsonian tremor [17].

Lesional therapies thus became an important treatment for advanced PD and were considered the best treatment until the introduction of levodopa in 1968 [18]. With the initiation of a safe and effective medication therapy for PD, enthusiasm for surgical interventions dwindled. It was only with the emergence of complications resulting from prolonged dopaminergic therapy combined with disease progression (disabling peak-dose levodopa-induced dyskinesias, motor fluctuations including severe off-freezing) that led to the revisiting of lesion therapies. Posteroventral pallidotomy would reemerge as an ablative brain surgery for PD in the late 1980s [19].

In 1987, Benabid, Pollack and colleagues observed, while using intraoperative electrophysiological recordings to ensure accuracy of the target (prior to ablation), that high-frequency electrical stimulation (≥ 100 Hz) could reversibly suppress both parkinsonian rest tremor and postural tremor [20]. The notion of combining electrical stimulation (utilized for brain mapping prior to permanent lesioning therapy) with the existing implantable pulse generator technology (adapted from cardiac pacemakers) birthed a new chronic neuromodulatory approach [11]. This approach would largely replace bilateral thalamotomy as the treatment for PD tremor. In the early 1990s, additional targets for DBS emerged, including the STN [21] and the GPi [22], which have since been carefully evaluated in multiple clinical trials [23-31]. The historical aspects of surgical therapies which later contributed to the rise of DBS as a means to address the motor symptoms and complications in PD are summarized in Fig. 1.

Fig. 1

Timeline of the development of pallidal deep brain stimulation (DBS). FDA US Food and Drug Administration, ET essential tremor, PD Parkinson’s disease, STN subthalamic nucleus

GPi: PD Pathophysiology, Anatomy and Somatotopy

The basal ganglia rate model, which postulated that an abnormal increase in GPi activity would result in thalamic inhibition and decreased prokinetic cortical activity, was one of the initial pathophysiological hypotheses potentially explaining PD symptoms [32, 33]. However, this model failed to elucidate the subsequent observations of GPi lesions not resulting in akinesia nor did it explain tremor control, dyskinesia reduction or the therapeutic benefit of dystonic symptoms. Much work has since been done, ultimately demonstrating that an aberrant network oscillatory activity is related to the pathophysiology of PD [34]. Exaggerated pallidal oscillations in the beta range (12–30 Hz) have been associated with bradykinesia severity and have been observed to attenuate with levodopa administration and with therapeutic DBS [35-39]. Beta power in the GPi in PD has also been shown to decrease with increased volitional movement, a phenomenon not exclusive to PD patients [38-43]. The results of subsequent DBS studies have suggested that neuromodulation may affect the oscillatory activity at a network circuitry level rather than by simply increasing or decreasing the firing rate of a single basal ganglia structure [34]. These findings have provided an opportunity for the further understanding of the network oscillations and physiology inclusive of multiple PD brain regions, including the GPi. Such information could facilitate the development of more targeted therapies using next-generation neuromodulatory devices.

Understanding the anatomy of the pallidum and its neighboring structures is of crucial importance to improve clinical DBS outcomes. The globus pallidus consists of internal (GPi) and external (GPe) segments. Along with the substantia nigra pars reticulata, the GPi relays information from the striatum, GPe and STN to the thalamus, although there are many other connections [44]. Post-mortem evaluations have reported GPi volumes ranging from 263.5 to 494.0 mm3; however, the mean volume values usually range between 400 and 500 mm3 [45]. A recent magnetic resonance imaging (MRI) evaluation of ten healthy, right-handed controls reported a mean GPi volume of 541.4 ± 81.9 mm3 [46], revealing that the GPi has roughly threefold the volume of the STN [45], which typically ranges in size from 150 to 300 mm3. Important to understanding the volume of targets is the continuous atrophy which occurs faster in PD than in control subjects and that this atrophy could affect DBS outcomes [47].

A dissected view of the GPi and its surrounding structures is shown in Fig. 2. Within the GPi, there are three distinct functional regions: the postero-ventro-lateral sensorimotor territory [48], the anterior associative territory, and the anteroventral limbic territory [49]. Alternative atlases using multimodal MRI, histology and structural connectivity have defined primary motor, sensorimotor and sensory regions of the GPi [50]. These distinct functional regions, along with a potentially optimal DBS lead placement for the symptomatic control of PD patients, are illustrated in Fig. 3. The sensorimotor region of the GPi is usually the DBS target for both for PD and dystonia [51, 52]. In this region, upper extremity somatotopy is usually located ventral and lateral to the area representing the lower limb, and orofacial movement is located even further ventral to that of the upper extremities (Fig. 4) [53]. Although many of these representations have been shown mainly in primate models of PD and in dystonia [54], they have been replicated by imaging studies and human intraoperative studies [49, 55].

Fig. 2

Dissection of white matter of the internal globus pallidus (GPi). a Lateral to medial dissections, b anterior to posterior dissections, c medial to lateral dissections, d inferior to superior dissections showing relationship of GPi to the optic tract, e inferior to superior dissections showing relationship of GPi to external globus pallidus (GPe), putamen and internal capsule (Int. Caps.), f deep brain stimulation of the GPi lead trajectory dissections until the GPi is reached. Accumb Accumbens, Ant. anterior, Comm. commissure, Caud. caudate, CN III third cranial nerve, Cor. Rad. corona radiata, Gl. gland, Innom. innominata, Nucl. nucleus, Olf. olfactory, Post. posterior, Subst. substantia, Tr. tract

Fig. 3

Functional regions of the GPi

Fig. 4

Somatotopic organization of the GPi. IC Internal capsule, OT optic tract, Th thalamus, Pt Putamen

Targeting and Intraoperative Physiology of the GPi: Pearls and Pitfalls

The GPi region includes the GPe dorsally and laterally, the optic tract ventrally and the internal capsule medially and posteriorly (Fig. 5). As previously mentioned, proper identification of these neighboring landmarks is important during intraoperative mapping and for selecting the final DBS lead position. Differences in targeting protocols, quality of imaging and utilization of microelectrode recordings together with macrostimulation may contribute significantly to the determination of final DBS lead localization. Many of these techniques have evolved over many years.

Fig. 5

The GPi and surrounding structures

In the early years of DBS surgery, indirect stereotactic targeting methods were used, as first described by Laitinen and colleagues, [19, 56]. However, the substantial variability in GPi size, shape and position relative to the midcommissural point among patients [57], along with some variability in the technique used by neurosurgeons for identifying the midcommissural point [58], has rendered such indirect GPi targeting methods unreliable. As a result, indirect targeting has been largely abandoned as a stand-alone technique and is currently used primarily as a starting point for direct anatomical approaches using high-quality imaging [59].

Improvements in imaging technology have lead to the utilization of direct targeting methods, which better account for patient-specific anatomical variation and can be used to improve the accuracy of the final GPi DBS lead position [57]. The fast gray matter acquisition T1 inverse recovery (FGATIR) 3T MRI sequence has been used by many groups to achieve direct visualization and a sharper delineation of the boundaries of subcortical structures, such as the STN, GPe, GPi, and the internal capsule, thereby improving GPi DBS targeting [60]. The use of this sequence has resulted in up to a threefold improvement in contrast-to-noise ratio when compared to T1 and T2/fluid-attenuated inversion recovery (FLAIR) MRI sequences [60]. Hybrid quantitative susceptibility mapping has also been used as a method to visualize the GPi and, similar to FGATIR MRI, it has shown reproducibility across centers [61]. When utilizing imaging for direct targeting, attention should be directed not only to a target point, but also to the position of the DBS electrode array relative to the optimal target volume. For example, in a situation where the entry angle is adjusted to a more lateral position to avoid a cortical vein, the target point should be revised medially to accommodate the adjustment and to avoid a lateral placement of the DBS lead contacts in relation to the target. Subsequently, entry angles may significantly impact GPi outcomes. A GPi DBS lead with a more vertical trajectory produces a more favorable configuration, with more contacts positioned within an optimal target volume as compared to a GPi lead with an entry point that is excessively anterior or lateral.

In addition to the advances in imaging modalities and software that have facilitated direct targeting, neurophysiological data from microelectrode recording (MER) and macrostimulation via the implanted DBS lead can also be used to further refine the final DBS lead placement. A typical GPi DBS entry angle trajectory (60–70° sagittal angle relative to the intercommissural plane, 5–15° coronal angle relative to the median plane) will traverse the striatum, corona radiata, GPe, GPi and the optic tract. MER typically shows characteristic cell firing frequencies and patterns for each of these structures, and these data can aid in the localization of the anatomical position of the individual MER tracts. Striatal cells tend to fire irregularly, at very low frequency (4–6 Hz), and are commonly activated by mechanical or electrical stimulation. Neuronal activity in the GPe is characterized by a low firing frequency (19–34 Hz) with occasional bursting cells. GPi cells in PD typically exhibit high tonic frequency, often with bursts (24–82 Hz). In both GPe and GPi, both short and long pauses can occur during MER recordings, with fewer pauses occurring in the GPi which possesses a more tonic pattern. The GPe can also manifest intermittent and characteristic “bursting” cells. Border cells are commonly encountered between the GPe and GPi, inside of the GPi (lamina), or on the posterior or ventral border of the GPi. The position of the optic tract can also provide useful information to optimize the ventral GPi DBS lead position. During MER, visual evoked potentials can be elicited by shining a flashlight into the eye. In the majority of cases, a soft audible evoked potential can be elicited through the audio speakers. An evoked potential that is “loud or prominent” may suggest a medial position of the DBS lead, but it should be noted that the trajectory can impact this finding.

When optimally exploited, MER “mapping” can be used to more precisely localize the target in anatomical space. Increasing the number of microelectrode passes can improve localization of the target, but it can also increase the risk of the procedure (and the intraoperative time) [62]. The sophistication of MER data interpretation has improved, and in most centers the number of MER passes per case has decreased. For GPi DBS procedures performed in the future, it is conceivable that less MER exploiting macrostimulation will be required through the use of the MER guidetube sheath, which may generate sufficient information to inform optimal lead positioning. More data will be required to refine such an approach. These techniques, however, should each be adapted and be based on user experience and degree of comfort.

Once MER mapping has been performed and a potential tract for the final DBS lead has been selected, macrostimulation via the implanted DBS lead can serve as a final test to confirm appropriate positioning. Macrostimulation is often performed at a pulse width of 60–90 μs with a frequency of 130–180 Hz (parameters commonly used in DBS in chronic PD). The amplitude (voltage) is gradually increased at each contact to determine thresholds for stimulation-induced optic, sensory and motor side effects. During macrostimulation, the distance of the most ventral DBS contact from the optic tract can be estimated. This is usually accomplished by the clinician darkening the operating suite, asking the patient to close her/his eyes and then determining the threshold voltage for perceptible light response in the contralateral visual field (e.g. phosphenes). Combined with capsular thresholds, optic tract information can be used to confirm or to adjust the lead position. If the optic tract threshold cannot be elicited by this technique and capsular thresholds are appropriate (2–4 V), then it is possible the lead is located in too shallow of a position. If the optic tract threshold is absent and capsular thresholds are high (4–6 V), the lead may be lateral or anterior. If the optic tract threshold is low and capsular thresholds are also low (< 2 V), the lead may possibly be medial. While intraoperative side effect thresholds do not always predict postoperative effects, excessively low side effect thresholds should prompt strong consideration of DBS lead repositioning.

While the DBS lead position can be estimated based on high-quality imaging, it is the physiological effect, and not radiographic imaging, that will ultimately determine the success or failure of the DBS surgery. A well-positioned DBS lead produces appropriately wide therapeutic windows for stimulation and facilitates easy DBS programming at the bedside. While directional stimulation and/or complex programming strategies can sometimes salvage a reasonable DBS outcome in a suboptimally placed lead, it is important to try to avoid the need for salvage. Optimizing the intraoperative testing protocol can ensure a DBS lead is well placed before exiting the operating room. Table 1 shows the typical lead location based on MER and macrostimulation effects. Though many experienced centers have abandoned intraoperative physiology in favor of asleep, image-guided DBS implantation, there are advantages to an awake approach. Interacting with an awake patient during DBS can help to determine the precise thresholds for adverse effects during macrostimulation; this approach may not be feasible with an asleep patient.

Table 1

Globus pallidus internus microelectrode recording lead location based on recordings and stimulation

Position of lead	Recordings and effects of stimulation
Anterior	MER often reveals reduction/absence of the GPi or kinesthestic cells (lack of somatotopy/evoked potentials)
	Long GPi MER run lengths in addition to long runs of striatum
	Absence of capsular side effects of fosphenes during macrostimulation
Posterior	Posterior tracks often yield motor side effects (i.e. tonic muscle contractions) due to stimulation of capsular fibers. Thresholds for such side effects are often narrow and may widen with dorsal contacts (depending on entry point/angle)
	Visual evoked potentials might be more clearly identified with posterior tracks; however, there may be a gap between the inferior border and optic tract (due to shape of GPi tapering superiorly on the posterior aspect)
Lateral	MER demonstrating a large segment of GPe and a small segment of GPi, a large lamina between them (border cells)
	Absence of optic tract/phosphenes and/or capsular side effects during MER and macrostimulation
Medial	MER typically reveals large segment of GPi and a small segment of GPe
	Loud visual evoked potentials in optic tract testing
	Low threshold for capsular side effects (i.e. muscle contractions) across all contacts (depending on angles)
Dorsal	Lack of side effects during macrostimulation
Ventral	Low thresholds for phosphenes during optic tract testing
	Lower thresholds for capsular side effects (i.e. muscle contractions)

GPe Globus pallidus externus, GPi globus pallidus internus, MER microelectrode recording

Our DBS team has implanted over 850 GPi DBS leads over the past 18 years, and many of the previously mentioned aspects of this section of the commentary pertain to observations and insights based on our cumulative experience [59, 61, 63]. Within the sensorimotor territory, positioning the DBS electrode 3 mm lateral to the medial internal capsule–GPi border, and 3–5 mm anterior to the posterior internal capsule–GPi border seems to be, in our experience, the optimal site for chronic therapeutic stimulation when using omnidirectional ring electrodes. The deepest contact is ideally situated immediately superior to the lateral aspect of the ipsilateral optic tract. With this lead position, there is a greater likelihood of maximum efficacious stimulation to the sensorimotor region of GPi without intolerable stimulation-induced side effects that may inadvertently occur from the spread of current into the internal capsule.

It is our group’s opinion that postoperative imaging should be obtained to carefully assess the position of every implanted DBS lead and that this information be used to facilitate programming. Appropriate DBS quality control should include neurosurgeons receiving feedback from the clinicians programming the implanted devices. Holanda et al. [63] performed a GPi DBS observational series of 299 patients over a 15-year period. These authors demonstrated a gradual shift over time toward a more lateral and ventral placement of the GPi DBS leads and hypothesized that the move was primarily driven by clinician feedback and based largely on device programmability within the clinic setting [63]. Following this overall strategy of utilizing imaging and direct targeting, intraoperative techniques with MER mapping and macrostimulation and with the implementation of postoperative imaging to provide feedback and adjustments a DBS team can usually increase the likelihood of achieving a successful outcome [64].

GPi DBS Versus STN DBS in PD: Effectiveness, Side Effect Profile and Long-Term Outcomes

Effectiveness and Side Effect Profile

The “best target for DBS in PD, whether STN versus GPi” has been the subject of a large body of recent literature [65, 66]. However, even though experts agree that targets should be chosen based on a patient’s unique characteristics [65, 66], worldwide STN remains the preferred target for DBS in patients with advanced PD. This was pointed out by the DBS for PD Study Group (2001), commenting on the bias towards STN DBS (n = 96) over GPi DBS (n = 38) [24, 67]. As a consequence of this bias, a greater amount of data on STN outcomes has been generated over the subsequent years when compared to data on GPi DBS outcomes. However, since the initial experience of GPi DBS in PD shared by Siegfried and Lippitz [22], several other case studies have replicated the positive GPi DBS results [68-84] and, more importantly, the efficacy of GPi as a target for PD has been confirmed in many randomized clinical trials.

The NIH COMPARE trial [29] and the VA study by Weaver et al. [85] both addressed the role of GPi in managing the motor symptoms of PD. Several more recent randomized, double-blind clinical trials have demonstrated that GPi and STN are both viable targets for the motor symptoms and complications of PD [23, 25, 29]. There was however one study (NSTAPS) where the secondary outcome of reducing akinesia favored STN [30]. Collectively, however, the current literature has revealed that stimulation of either STN or GPi can improve the motor features of parkinsonism and positively impact the quality of life.

While STN DBS has been the preferred target for medically refractory tremor in PD, a recent meta-analysis has shown that GPi DBS is effective for tremor control [86]. None of the randomized GPi versus STN DBS comparison trials have revealed an advantage of one target over the other in terms of addressing tremor. They have, however, collectively utilized the Unified Parkinson’s Disease Rating Scale (UPDRS) motor subscores to assess tremor outcomes, which was limited to single items without the ability to assess constancy, duration or associated disability related to the actual tremor.

Data from randomized clinical trials have shown that GPi DBS improves baseline UPDRS motor off-medication scores by 27–54% [68, 70–72], which is similar to the improvement in UPDRS motor off-medication scores of 30–67% obtained with STN DBS [28, 30, 68]. The activities of daily living scores have been found to improve on average by 30–39% in GPi DBS and by 6–56% in STN DBS [24, 70–72, 87]. The main findings reported for GPi DBS in randomized clinical trials are summarized in Table 2 (case reports in Electronic Supplementary Material Table 1) [23–31, 87–93]. Overall, these results reveal sustained benefits for up to 3 years on motor function, motor fluctuations, dyskinesia and quality of life in both the STN DBS and GPi DBS [28, 94, 95].

Table 2

Overview of randomized clinical trials on deep brain stimulation of the globus pallidus internus

Study	Design	n	Age, years (mean ± SD)	Duration of disease, years (mean ± SD)	Stage of disease (H&Y scale)	Unilateral (U) or bilateral (B)
Burchiel et al. [88]	Randomized, blinded pilot study	4	46.5 ± 11	10.6 ± 2	4.0 (4.0–4.8)	B
Katayama et al. [89]	Double blinded evaluation	7				5U, 2B
Anderson et al. [23]	Randomized, blinded, parallel- group study	10	54 ± 12	10.3 ± 2	4–4.5	B
DBS for Parkinson’s Disease Study Group et al. [24]	Prospective, double-blinded, randomized crossover trial	38	55.7 ± 9.8	14.5		B
Rodriguez-Oroz et al. [26]	Multicenter, double blinded, non-randomized trial	20	55.8 ± 9.4	15.4 ± 6.2	3.0–5.0	B
Follett et al. [25]	Multicenter, randomized, blinded trial	152	61.8 ± 8.7	11.5 ± 5.4	3.3 ± 0.9	B
Weaver et al.a[27]	RCT	61	62.4 ± 8.8	10.8 ± 5.4	≥ 2	B
Robertson et al. [90]	Randomized double-blinded controlled trial	13	65.5 ± 8.6	15.1 ± 10.2	3.5 ± 0.8	B
Weaver et al. [28]	Multicenter RCT	89	60.4 ± 8.3	11.4 ± 4.9	3.3 ± 0.8	B
Okun et al. [29]	Prospective, blinded RCT	23	60.2 ± 6.2	12.5 ± 3.6	2.0–5.0	U
Rocchi et al. [91]	Randomized	14	61.1 ± 8.4	12.9 ± 10.17	3.5 ± 0.9	B
Odekerken et al. [30]	RCT	65	59.1 ± 7.8	10.8 ± 4.2	2.5 (0–4)	B
Moro et al. [92]	Nonrandomized, multicenter study	16	56.0 ± 2.1	15.1 ± 1.5	3.9 ± 0.2	B
Rothlind et al. [31]	Prospective, randomized, controlled study	80	61.3 ± 8.9	11.0 ± 4.7	3.2 ± 0.8	B
Zahodne et al. [87]	RCT	22	61.3 ± 5.5	12.4 ± 3.6		U
St George et al. [93]	RCT	10	62.8 ± 8.2	15.4 ± 8.4		B

DBS Deep brain stimulation, H&Y Hoehn and Yahr, LEDD levodopa equivalent daily dose, MED OFF, off-medication, PD Parkinson’s disease, PQD-39 39-item Parkinson's Disease Questionnaire, RCT randomized controlled trial, SD standard deviation, STN subthalamic nucleus, UPDRS Unified Parkinson’s Disease Rating Scale

aCombined GPi and STN result

In the domain of postoperative medication reduction, data support that STN DBS has a greater impact on reducing the total daily doses of dopaminergic drugs [92] as compared to GPi DBS [93]. However, GPi DBS has shown a powerful direct, anti-dyskinetic effect, with rates of improvement in levodopa-induced dyskinesia ranging from 47 to 88%, in comparison to STN DBS with rates of 20–83% [96]. GPi DBS has also been shown to have a greater anti-dyskinetic effect than STN DBS, a phenomenon that may be largely driven by the need for medication reduction in the STN (e.g. medication reduction required to avoid dyskinesia) [97, 98]. However, there is a subset of patients with STN DBS that have been shown to achieve a 12–16% reduction in dyskinesias, suggesting some possible direct anti-dyskinetic effects in some STN DBS patients [84, 89].

There may be DBS patients who are more susceptible to the occurrence of stimulation-induced dyskinesias (SID). Some authors have referred to this phenomenon as brittle dyskinesia [99, 100]. For example,, a patient who takes a low dose of levodopa (less than a tablet of carbidopa/levodopa 25/100 mg per dose) and experiences a brittle dyskinetic response may be more amenable to GPi DBS over STN DBS. However, these data are supported only by anecdotal experience and a few case reports. Direct comparative studies of this population have not been performed [100]. It is important to consider that following DBS surgery it may be necessary to escalate levodopa dosages to maintain symptom control resulting from disease progression, and GPi may be a more viable target in this scenario due mainly to its direct anti-dyskinetic effects [66]. GPi DBS may have an advantage in providing long-term flexibility for medication adjustments. Patients implanted in the STN have an increased likelihood of dyskinesia if they increase their doses of levodopa post-surgery, an effect that does not seem to occur in most patients implanted in the GPi [25, 28–30, 101]. Finally, SID are less commonly observed in GPi DBS when compared to STN DBS, but they can occur by stimulating the most dorsal DBS contact in the GPi/GPe border zone or in the GPe [102].

As neuropsychiatric comorbidities pose one of the main potential contraindications for DBS implantation in PD, postoperative complications and/or stimulation-induced limbic effects have been studied extensively. Mood and cognitive changes have been reported following both STN DBS and GPi DBS [25, 28, 29, 31]. In the 36-month outcome data drawn from the VA study, patients undergoing GPi DBS implantation experienced no change in the Mattis Dementia Rating Scale and Hopkins Verbal Learning Test. Patients in the STN arm of the study experienced worsening in these domains at 3 years postsurgery (p = 0.01) [28]. The COMPARE study did not reveal a difference between STN DBS or GPi DBS in its primary outcome of mood as measured by the eight subscales of the Visual Analogue Mood Scale; however, there were more cognitive adverse events reported in STN DBS [29].

Evaluation of the overall side effect profile has shown that the GPi is similar to the STN as DBS target. A few studies suggest that the more commonly encountered adverse effects on gait, speech and swallowing may possibly be less frequently encountered in GPi DBS [65]. A possible explanation is that the GPi size dimensions are larger and there is thus a more expansive sensorimotor territory when compared to the STN. This expanded territory may allow clinicians to deliver effective stimulation that is more contained within the region of interest, yielding wider therapeutic windows and theoretically fewer adverse effects.

Adverse effects in speech have been more commonly reported in STN DBS than in GPi DBS [103], with a greater proportion of irreversible speech side effects reported (STN: 16.7% irreversible and 5.1% reversible speech adverse effects; GPi: 22.2% reversible speech adverse effects) [104]. There is, however, a paucity of studies directly comparing swallowing function between the two targets [105]. In a retrospective study, Troche et al. demonstrated dysphagia, as measured by the penetration–aspiration scores, to be worse in unilateral STN DBS than in unilateral GPi DBS [106]. Gait disturbances are more commonly reported with STN DBS (16.7%) than with GPi DBS (5.6%); however, the distinct characteristics and elements of gait disturbance in the short and long term have yet to be evaluated [104].

Long-Term Benefit

Few studies have been able to document data for clinical efficacy in the long-term when comparing GPi and STN as targets for DBS. Three studies [92, 107, 108] have reported outcomes of GPi DBS for PD patients with a minimum of 5 years of follow-up. Moro et al. reported sustained improvement in tremor, rigidity and dyskinesia at 5–6 years after surgery in both STN DBS and GPi DBS for advanced PD [92]. Volkmann et al. reported continued improvement in rigidity and dyskinesia at 5 years following bilateral GPi DBS (n = 6) [107]. In two studies in STN DBS, UPDRS motor off-medication scores were reported to be maintained at 5 years, but deterioration at between 5 and 8 years was reported [109, 110], with cognitive dysfunction being present in 17.1 and 16.7% of cases at 5 and 8 years, respectively [109]. Another report on the comparison of GPi DBS and STN DBS showed a maintained benefit from bilateral STN DBS in tremor, bradykinesia and motor fluctuations (n = 18) at 10 years [111]. In a recent study on 16 patients who underwent either unilateral or bilateral GPi DBS with a mean follow-up period of 6 years, Lachenmayer et al. reported sustained improvement in tremor and dyskinesia for 5.5 years [108]. Similarly, in a recent systematic review, Limousin et al. [112] pooled the 5-year outcomes of STN DBS across 551 PD patients from 15 independent studies and demonstrated sustained improvements in tremor, rigidity and dyskinesia at 5 years, while benefits in bradykinesia and axial symptoms decreased between the 1- and 5-year assessments [112]. The evaluation of stimulation-induced dyskinesias has not been systematically reviewed in the long term when comparing GPi DBS and STN DBS for PD.

The overall long-term GPi DBS outcomes appear to be similar to those of STN DBS, including the waning therapeutic effect, specifically on axial symptoms [113]. However, there are considerable differences in clinical profiles and study designs, making a direct comparison difficult. Although randomized controlled trials of GPi DBS and STN DBS have revealed comparable benefits on motor symptoms, dyskinesia and quality of life for up to 3 years of follow-up [28, 94], longer follow-up studies are needed. The differences between GPi DBS and STN DBS use in PD are summarized in Table 3.

Table 3

Comparison of outcomes of deep brain stimulation in the subthalamic nucleus versus globus pallidus internus

Outcomes	GPi	STN
Bradykinesia	+ +	+ +
Rigidity	+ +	+ +
Tremor	+ +	+ +
Quality of life	+ +	+ +
Dyskinesia reduction	+ +	+
Stimulation-induced dyskinesia	−	−−
Medication reduction		+ +
Flexibility of long-term medication adjustments	+ +
Cognitive adverse effects	−	−−
Mood adverse effects	−	−−
Gait adverse effects	−	−−
Speech and swallowing adverse effects	−	−−
Ease of programming	+ +
Reduced battery consumption		+ +
Ability to use one lead	+ +

+ slight benefit; ++ clear benefit; − rarely occurring; −− more commonly occurring

Several studies have focused on comparing the STN and GPi as the target of DBS, reporting that this exercise should be patient tailored, factoring in a variety of elements, such as the preoperative presence of dyskinesias, the need for medication reduction and the need for a pre-surgical risk assessment (co-morbid cognitive, psychiatric, balance, speech or swallow disorders). A full multidisciplinary screening is recommended for any patient seeking potential DBS therapy [66]. Many experts cite GPi as an easier target for long-term programming as compared to STN, especially in communities where expert clinicians are not available or are scarce; however, direct comparisons on ease of programming have not been performed. It is possible that GPi DBS gives more flexibility in the use of levodopa in the long term (e.g. less risk of dyskinesia as the need for more levodopa emerges), but again, there is lack of direct comparative data between STN and GPi [66].

Pearls in Programming GPi DBS for PD

Several algorithms have been developed to guide clinicians in GPi DBS programming for PD [8, 114–116]. Regarding initial programming, a monopolar review has classically been employed, with a careful assessment of the clinical benefit and therapeutic window for each individual DBS lead contact by using gradual increments in stimulation amplitude and by maintaining a constant pulse width and frequency of stimulation [114, 116]. Although there are slight differences in protocol across institutions, a monopolar survey is usually performed at pulse widths ranging between 60 and 120 μs and at frequencies of 130–185 Hz in the GPi target. These values reflect commonly used initial programming settings following DBS lead implantation [114, 116]. Corticospinal and corticobulbar side effects can be identified by examining for contralateral muscle contractions and dysarthria, respectively, when performing the monopolar review. For example, speech difficulties can be characterized by slurred, strained/strangled speech, imprecise voice quality with occasional bursts of loudness and unanticipated arrests. If the DBS lead is placed excessively medial within the GPi target, corticospinal or corticobulbar side effects can appear at low values of stimulation in all contacts (e.g. low voltage with a constant pulse width and frequency). If the lead is located too posteriorly in the GPi, speech difficulties or muscle contractions may occur with low voltages when activating the most posterior/ventral contacts. However, the threshold to induce side effects will increase (e.g. higher voltage) with more dorsally and anteriorly located contacts based largely on the chosen targeting approach/angle. If the DBS lead is placed excessively lateral, dorsal or anterior to the sensorimotor GPi, the threshold voltages for corticospinal side effects will be wider and, for example, no adverse effects may occur with the dorsal contacts. For leads placed ventrally, given the proximity to the optic tract, frequently there is appreciable visual phenomena (phosphenes), which can possibly be avoided by choosing the more dorsal DBS contacts on the lead. Programming considerations of the final GPi DBS settings should take into account the anatomy and the specific location of each DBS lead for contact.

Impact of GPi in Future Directions for DBS in PD

Is GPi a Viable Target for Closed-loop DBS Systems?

The expanding understanding of the physiology and network effects underpinning pallidal stimulation has led to a more comprehensive view of the pathophysiology underpinning human PD. Beta oscillations (12–30 Hz) have emerged as a potential marker for some PD-related symptoms, making them a potentially viable target for closed-loop neuromodulatory therapy. Commercially available DBS devices are becoming increasingly capable of measuring local field potentials. These devices thus have the potential for patient-tailored adaptive DBS paradigms designed to treat pathological oscillations in an on-demand, closed-loop manner. Sensing–stimulating technology offers the prospect for more efficient neuromodulatory therapy that can potentially prolong battery longevity, be more responsive to individual physiological demands and be associated with fewer side effects. In the first study of its kind in the GPi region, adaptive DBS was well-tolerated when applied to a single PD patient; however, long-term efficacy data was not assessed [37]. In a recent study, Eisinger et al. [34] demonstrated that a greater beta power was present in PD GPi as compared to PD STN, suggesting GPi to be a candidate for closed-loop DBS systems of the future [39]. Although preliminary work for the safety, feasibility and effectiveness of closed-loop systems in PD has been mostly restricted to targeting the STN [37, 117–124], more studies in the future will likely explore the GPi, especially given the robustness of the physiological signal [37].

Understanding Pallidal Sub-Regions and Their Role in Network Modeling

Going beyond a better understanding of the anatomy and physiology, it is likely that a comprehensive exploration of the relationship between the pallidum and other brain circuitries will be pursued. This information will shine a light on different diseases and techniques, such as MRI diffusion tractography. In 2018, Middlebrooks et al. [59] employed structural connectivity profiling techniques to parcellate the GPi based on axonal connectivity to other neural structures. This approach provided a method to visualize the GPi based on histology and on functional anatomy [59, 125, 126]. Eleven unilateral PD GPi DBS patients underwent modeling of the volume of tissue activation based on clinic programming parameters which were correlated with the connectivity-based segmentation. Across the ten pre-defined connectivity-based targets (caudate, GPe, primary motor cortex, pedunculopontine nucleus, prefrontal cortex, putamen, supplemental motor area, STN, substantia nigra and thalamus), the authors found greater improvement in the UPDRS part III (motor examination) with a greater activation of GPi regions. There was a strong connectivity to the primary motor cortex and supplemental motor area. This work is just one example of the importance of understanding GPi subregions and connectivity profiles as we develop new treatment approaches. As technology evolves, new methods and techniques will emerge and refine presurgical planning. This will hopefully improve the procedure, refine lead placement, expedite the DBS programming and ultimately lead to improved outcomes.

Conclusion

Surgery within the GPi region alleviates many of the symptoms of PD, and this approach has been refined from clinical experience using lesioning protocols. As the ongoing target debate continues (GPi vs. STN), clinicians should keep in mind that GPi DBS has been shown to be effective for a variety of symptoms, including bradykinesia, rigidity and tremor control. GPi DBS also has an important, direct anti-dyskinetic effect. Long-term outcomes of pallidal DBS appear to show sustained benefit for up to 6 years following surgery. The adverse effect profile of GPi DBS is similar to that of STN DBS, although it may be associated with less speech, swallow and gait difficulties. In addition to these potential advantages, GPi DBS might also be an easier target to program and allow more long-term flexibility for medication adjustments (e.g. levodopa) that are commonly required with PD progression. Emerging technologies, including GPi closed-loop systems, advanced tractography-based targeting and enhanced programming strategies, warrant further research. We conclude that although the GPi target may not be appropriate for all PD DBS patients, it has specific advantages.

Below is the link to the electronic supplementary material.

Supplementary file1 (XLSX 12 kb)

We review the history of the globus pallidus internus (GPi) as target for deep brain stimulation (DBS), which is now an established, safe and effective method of treating the motor complications of advanced Parkinson’s disease (PD).

We comprehensively review the literature on GPi DBS for PD, including anatomy, physiology, somatotopy, surgical targeting and management, potential pitfalls and optimal location for lead placement.

We outline the evidence underlying the effectiveness of GPi DBS in PD for managing PD symptoms.

We present common patient programming strategies in GPi DBS and strategies to avoid adverse effects

We discuss new emerging technologies that will modify application of GPi DBS in PD in the future.