Stereotactic surgery has made a resurgence in the treatment of Parkinson disease (PD), largely because of the long-term complications of levodopa therapy, which result in significant disability despite optimal medical management.  A better understanding of basal ganglia physiology and circuitry (see the image below) and improvements in surgical techniques, neuroimaging, and electrophysiologic recording have allowed surgical procedures to be performed more accurately and with lower morbidity. 
Stereotactic surgery is considered for PD patients who have motor fluctuations and dyskinesia that cannot be adequately managed with pharmacologic manipulation. The principal surgical option is deep brain stimulation (DBS), which has largely replaced neuroablative lesion surgery. Experimental surgical approaches include transplantation and gene therapy.
During stereotactic surgery, imaging data are correlated to a 3-dimensional space, permitting a target deep within the brain to be reached blindly and with minimal trauma. With frame-based techniques (see the image below), application of a reference coordinate system to the skull permits any point in the brain to be described with Cartesian (ie, x, y, z) coordinates. Ventriculography, an important method for target localization before the development of computed tomography (CT) and magnetic resonance imaging (MRI), is now rarely used.
CT-guided stereotaxis provides direct imaging of brain parenchyma without image distortion; however, its gray-white resolution is inferior to that of MRI, and it can provide only axial imaging. MRI provides superior target resolution and triplanar imaging; however, some smaller targets may not be visualizable, and MRI is prone to image distortion. Although these distortions are usually small, they can affect targeting for functional neurosurgery.
The possibility of targeting errors due to image distortion necessitates the use of some form of intraoperative neurophysiologic monitoring to confirm correct targeting during surgery for movement disorders (see the image below).
A comparative study suggests that an image fusion technique making use of stereotactic CT and MRI may be able to record a significantly longer subthalamic nucleus (STN) length through limited microelectrode recording than MRI alone can. [3, 4] Further study is needed to determine whether this technique could be effectively used to improve clinical outcome and reduce morbidity.
Deep Brain Stimulation
Until comparatively recently, surgical treatment of movement disorders primarily involved neuroablative lesion surgeries that destroyed abnormally hyperactive deep brain nuclei. However, the observation that high-frequency electrostimulation in the ventral lateral nucleus (VL) of the thalamus eliminates tremors in patients undergoing thalamotomy led to investigation of long-term deep brain stimulation (DBS) as a reversible alternative to neuroablation.
DBS has become the surgical procedure of choice for Parkinson disease (PD) because it does not involve destruction of brain tissue; it is reversible; it can be adjusted as the disease progresses or adverse events occur; and it allows the performance of bilateral procedures without a significant increase in adverse events.
Continued refinement of the knowledge of basal ganglia circuitry and PD pathophysiology has narrowed the focus of movement disorder surgery to 3 key gray-matter structures (see the image below):
Subthalamic nucleus (STN)Sagittal section, 12 mm lateral to midline, demonstrating subthalamic nucleus (STN; lavender). STN is one of preferred surgical targets for deep brain stimulation to treat symptoms of advanced Parkinson disease.
A randomized controlled trial in 255 patients with advanced PD found that bilateral DBS was more effective than optimal medical therapy in improving on time without troubling dyskinesias, motor function, and quality of life at 6 months; however, DBS was associated with an increased risk of serious adverse events. 
In Australia, guidelines have been developed to help neurologists and general physicians identify PD patients who may benefit from referral to a specialized DBS team; these teams assess the likely benefits and risks of DBS for each referred patient. 
A more extensive discussion of DBS in this setting, including mechanisms of action, advantages and disadvantages, and stages of the procedure, is provided elsewhere (see Deep Brain Stimulation in Parkinson Disease).
Neuroablative Lesion Surgeries
Neuroablative lesion surgeries involve the destruction of targeted areas of the brain to control the symptoms of Parkinson disease (PD); they have largely been replaced by deep brain stimulation (DBS). During neuroablation, a specific deep brain target is destroyed by means of thermocoagulation. A radiofrequency generator is used most commonly to heat the lesioning electrode tip to the prescribed temperature in a controlled fashion.
The 2 most commonly performed neuroablative procedures are thalamotomy and pallidotomy, in which lesions are created in the ventral lateral thalamic nucleus (VL) and the internal segment of the globus pallidus (GPi; also known as the globus pallidus medialis), respectively.
The VL receives afferent innervation from 2 primary sources: the GPi via the ansa lenticularis and thalamic fasciculus and the contralateral cerebellum via the superior cerebellar peduncle. These cerebellar fibers synapse primarily in the ventral intermediate (VIM) and ventral oral posterior (VOP) segments, the most posterior segments of the VL.
Oscillating excitatory input from the cerebellum may be responsible for the tremor observed in PD, in that cellular activity synchronous with the frequency of parkinsonian tremor can be recorded in the VL. These data support the clinical observation that lesions placed within the VL (and specifically within the VIM or VOP) arrest parkinsonian and essential tremors.
VL thalamotomy was the most frequently performed procedure for movement disorders in the prelevodopa era because tremor responds best to thalamotomy and can be monitored more easily in the operating room than gait abnormalities, rigidity, and akinesia. 
In this procedure, a part of the thalamus, generally the VIM, is destroyed to relieve tremor. The VIM is almost unanimously considered the best target for tremor suppression, with excellent short- and long-term results in 80-90% of patients with PD. Thalamotomy has little effect on bradykinesia, rigidity, motor fluctuations, or dyskinesia. When rigidity and akinesia are prominent, other targets, including the GPi and the subthalamic nucleus (STN), are preferred.
Thalamotomy is indicated in patients with PD who are disabled by medically refractory tremor. The anticipated benefit of tremor reduction or elimination must be considered carefully. Rest tremor alone is rarely disabling, and bradykinesia and rigidity can reduce dexterity irrespective of tremor. Most patients with PD who undergo thalamotomy have significant improvement in tremor of the limbs contralateral to the side of the lesion. Bilateral thalamotomy is generally avoided, because complications, especially speech and cognitive impairment, are common.
Morbidity for thalamotomy is reported to range from 9% to 23%. The predominant complication is speech impairment with dysarthria and hypophonia. The risk of speech abnormalities is 30% after unilateral thalamotomy and greater than 60% after bilateral thalamotomy. Other complications include memory loss, contralateral hemiparesis, and, more rarely, hemineglect, dystonia, hemiballismus, athetosis, and dyspraxia.
Preoperative memory and language evaluation can help identify patients who are at greatest risk for postoperative cognitive and language dysfunction. In the largest series, the mortality for thalamotomy ranged from 0.5% to 1%. Death results almost exclusively from intraparenchymal hemorrhage.
Although Svennilson et al described ventral posterior pallidotomy back in the 1960s,  their report was largely overlooked. The original pallidotomy target was in the medial and anterodorsal part of the nucleus. This so-called medial pallidotomy effectively relieved rigidity but inconsistently improved tremor.
Leksell subsequently moved the target to the posteroventral and lateral GPi, achieving sustained improvement in as many as 96% of patients. In 1992, Laitinen et al reported reduced tremor, rigidity, akinesia, and levodopa-induced dyskinesia in 38 patients treated with pallidotomy, prompting a reappraisal of the procedure performed with more modern techniques. 
The negative symptoms of PD (ie, rigidity and bradykinesia) are caused, in part, by excessive inhibitory output from the GPi to the VL thalamic nucleus. Lesioning of the sensorimotor region of the GPi, which lies ventral and posterior in the nucleus, decreases this hyperinhibition of the motor thalamus.
Pallidotomy involves destruction of a part of the GPi. Pallidotomy studies have demonstrated significant improvements in each of the cardinal symptoms of PD (tremor, rigidity, bradykinesia), as well as a significant reduction in dyskinesia. However, the tremor improvement is less consistent than that seen with thalamotomy.
The most serious and frequent (3.6%) adverse effect of pallidotomy is a scotoma in the contralateral lower-central visual field. This complication occurs when the GPi lesion extends into the optic tract, which lies immediately below the GPi. The risk of visual-field deficit is reduced greatly by accurate delineation of the ventral GPi border by microelectrode recording.
Less frequent complications (< 5%) include injury to the internal capsule, facial paresis, and intracerebral hemorrhage (1-2%). Abnormalities of speech, swallowing, and cognition may also be observed.
Bilateral pallidotomy is not recommended, because complications, including speech difficulties, dysphagia, and cognitive impairment, are relatively common.
Hyperactivity of the excitatory projections of the STN to the GPi is a crucial physiologic feature of PD. Subthalamotomy involves destruction of a part of the STN. Although lesioning the STN usually has been avoided because of concerns that hemiballismus might develop, experimental results from animals and humans suggest that subthalamotomy may be performed safely and may reverse parkinsonism dramatically. Studies have shown significant improvements in the cardinal features of PD, as well as the reduction of motor fluctuations and dyskinesia.
Evaluation of Patients for Stereotactic Surgery
Good surgical outcomes from stereotactic surgery for Parkinson disease (PD) begin with careful patient selection and end with attentive, detail-oriented postoperative care. The authors believe that this level of care is best provided by a multidisciplinary team comprising a movement disorder neurologist, a neurosurgeon who is well-versed in stereotactic technique, a neurophysiologist, a psychiatrist, and a neuropsychologist. Additional support from neuroradiology and rehabilitation medicine is essential.
At the authors’ movement disorder center, patients are evaluated for surgery according to the following steps:
Evaluation by a neurologist
Evaluation by a neurosurgeon to identify potential surgical candidates
Evaluation by a psychiatrist
First, a neurologist with expertise in movement disorders evaluates the patient. Patient selection is particularly important for successful subthalamic nucleus deep brain stimulation (STN-DBS) because a number of factors combine to determine positive surgical outcome. [10, 11] These can be summarized as follows:
A diagnosis of idiopathic PD
Positive response to levodopa therapy
Absence of atypical parkinsonian features
Advanced disease, virtually unmanageable with dopaminergic medications
Relatively young age; however, advanced age (> 75 y) is not an absolute contraindication to surgery (if a patient otherwise meets the selection criteria for a procedure and the quality of life is predicted to improve substantially, surgery should be offered)
Absence of active psychiatric disease
Good social support and access to programming
Potential surgical candidates then are evaluated by the neurosurgeon, who determines whether the patient is indeed a good candidate for surgical treatment and decides which procedure(s) would benefit the patient most. Close collaboration between the neurologist and the neurosurgeon aids the decision-making process, thereby minimizing patient confusion and stress. If the neurologist and neurosurgeon agree that the patient is a good surgical candidate, further workup includes the following:
Magnetic resonance imaging (MRI) of the brain to rule out comorbid conditions and to assess the degree of brain atrophy; significant atrophy may increase the risk of perioperative hemorrhage
Detailed neuropsychological testing to rule out subtle cognitive impairment, which can be exacerbated by the surgical procedure
A psychiatrist with expertise in psychiatric complications of movement disorders may be consulted to rule out active psychiatric disease and screen for relevant past psychiatric history that may pose a contraindication to surgery (eg, major depression or suicidality).
Fluorodopa positron emission tomography (PET) may be performed in the unusual circumstance that an alternative diagnosis of multiple-system atrophy cannot be ruled out clinically.
Finally, a medical evaluation is performed to determine the patient’s general fitness for surgery.
Surgery is reserved for patients with medically refractory PD who have disabling problems. Currently, the authors’ center adopts the following surgical recommendations for patients with medically refractory PD:
Unilateral pallidotomy is offered to patients with asymmetric PD who develop fluctuations in their response to levodopa, including disabling dyskinesias and off-state dystonia
Bilateral pallidotomy is avoided, though investigations are under way to evaluate contralateral globus pallidus pars interna deep brain stimulation (GPi-DBS) in patients who have undergone a successful pallidotomy and are experiencing disease progression in the untreated side
Thalamotomy or thalamic DBS is offered to the minority of patients with PD who have predominant and disabling tremor (more commonly, this procedure is performed on patients with disabling essential tremors)
Thalamic DBS is preferred to thalamotomy, particularly in young patients with PD who are disabled solely by tremor early in the course of their disease, because it gives the option of removing the stimulator if more effective therapies are developed or if symptom progression necessitates DBS at another target (eg, the STN)
Bilateral STN-DBS is offered to patients with advanced PD who have bilateral levodopa-induced dyskinesia, significant gait disturbances and axial symptoms, or medically refractory rigidity and akinesia; STN-DBS has a variable effect on speech, but it ultimately results in the deterioration of speech intelligibility 
Before surgery, the patient should be informed that these procedures do not cure PD and that disease progression is still to be expected
Neural transplantation is a potential treatment for Parkinson disease (PD) for the following reasons:
The neuronal degeneration is site- and type-specific (ie, dopaminergic)
The target area is well defined (ie, striatum)
Postsynaptic receptors are relatively intact
The neurons provide tonic stimulation of the receptors and appear to serve a modulatory function
In double-blind studies, neither transplantation of autologous adrenal medullary cells nor transplantation of fetal porcine cells has been found to be effective; both have been abandoned. Although open-label studies of fetal dopaminergic cell transplantation yielded promising results, 3 randomized, double-blind, sham-surgery–controlled studies found no net benefit. In addition, some patients receiving these transplants developed a potentially disabling form of dyskinesia that persisted even after withdrawal of levodopa.
Features such as gait dysfunction, freezing, falling, and dementia are likely to be caused by nondopaminergic pathology and hence are unlikely to respond to dopaminergic grafts. 
Lewy body–like inclusions have been found in grafted nigral neurons in long-term transplant recipients. These inclusions stained positively for alpha-synuclein and ubiquitin and had reduced immunostaining for dopamine transporter, which suggests that PD may affect grafted cells. 
Human retinal pigment epithelial cells produce levodopa, and retinal pigment epithelial cells in gelatin microcarriers have been implanted into the putamen in preliminary studies. A phase II double-blind, randomized, multicenter, sham-surgery–controlled study of this technique is under way.  In one case study, however, postmortem examination of a patient who died 6 months after surgical implantation of 325,000 retinal pigment epithelial cells found only 118 surviving cells.