Deep brain stimulation for the heterogeneous pathophysiology of Parkinson’s disease

Firing rate model

The basal ganglia are central to the motor circuit. Various factors determine the output of the basal ganglia.^[1] Based on the firing rate model, in the normal state, the basal ganglia are controlled by a direct and indirect pathway. The indirect pathway inhibits the thalamus from sending messages to the cortex. This occurs by inhibitory signals from the striatum and decreased activity from the globus pallidus externus (GPe), which, in turn, increases STN activity. The STN then sends excitatory messages to the GPi and increases the GPi inhibitory signals, which controls the thalamus.^[2] On the other hand, the direct pathway disinhibits the thalamus from stimulating the motor cortex through excitation of the striatum, which inhibits the GPi inhibitory signals. This will allow for the initiation of movement.^[1] The substantia nigra influences the pathways with an excitatory and inhibitory signal to the striatum in the direct and indirect pathway, respectively.

In the Parkinsonian model, the dopamine degeneration in the substantia nigra shifts the activity of the indirect motor pathway described above. The loss of the cells in the substantia nigra changes the indirect pathway and there is less inhibition of the striatum inhibitory activity.^[1] This leads to the suppression of the thalamus-motor cortex network through the increased activity of the STN-GPi projection. This is thought to be a central component from which Parkinsonian symptoms arise.^[3] This is supported by the fact that STN firing rates increase as the disease progresses and dopaminergic loss in the substantia nigra increases.^[4] With more data, the classic model for Parkinsonian pathogenesis has been modified significantly. For DBS, it is not only important to understand this overall model but also to understand the different symptoms that arise from stimulation in different sites within the STN, GPi or Vim.

The classic model of Parkinsonian pathophysiology seems only to explain akinetic features of the disease. In the classic model, the substantia nigra dopaminergic loss leads to a loss of input to the putamen. This translates to clinical akinetic symptoms, but it does not always render tremors.^[5] Several studies have shown that Parkinsonian tremor seems to have a distinct pathophysiology from the other cardinal symptoms of PD.^[6,7] Dopaminergic therapy shows a consistent effect on the akinetic and bradykinesia symptoms in patients but is not consistent with tremors. Imaging of dopaminergic cells does not consistently correlate with PD tremor severity either.^[8-10] Therefore, the classic model and the dopaminergic loss cannot fully explain tremor. Several studies suggest that both the cerebello-thalamo-cortical and spino-thalamo-cortical pathways are central to tremor.^[11] Other criticisms of the firing rate theory pose issues as well. According to the model, the role of the GPi is a central component of the overall circuit, but inactivation of the GPi in animal models does not result in motor dysfunction.^[12] Furthermore, pallidotomy has also been used to alleviate L-dopa-induced dyskinesias, which does not follow the expected effect of decreased GP activity.^[13]

Firing pattern model

Newer models suggest that the firing patterns rather than firing rates play a role in the pathophysiology of PD. In the PD state, the central components of the basal ganglia have an abnormal synchronisation of neural firings.^[14] PD medication and DBS and have been shown to disrupt this synchronisation, leading to motor improvements.^[15,16] Single unit activity and local field potential analysis verify this model and show the abnormal synchronised activity.^[17] In the MPTP primate model, the STN, GPi and GPe have irregular synchronised neural firings.^[18] Several studies have demonstrated both a low and high frequency oscillation relating to PD tremor. In PD patients, the thalamus and basal ganglia structures have been shown to have prominent oscillatory neuron activation in the muscular activity and tremor frequency band of 4–6 Hz and the beta band frequency of 13–30 Hz, respectively, indicating the central involvement of these structures in the pathophysiology of PD.^[19] The STN and GPi were seen to have greater oscillation activity in the beta band, which seems to be suppressed with dopaminergic medication and treatment. There is also a correlation between increased beta activity and UPDRS scores.^[20] LFP recordings confirm the presence of a beta band in PD patients in the off-state.^[20]

The exact mechanism that the beta band controls is still unclear. Beta bursts are suppressed during continuous motion, especially before and during movement.^[21] The suppression of beta is diminished over time, which indicates that the worsening of the performance is correlated with increased beta oscillation. Furthermore, to cancel a prepared movement in a Go and No-Go model, the beta band activity is increased for the antikinetic movement. Further studies have found that tremors in different muscle groups or stopping motions can cause oscillatory activity in distinct frequency bands.^[22-24]

Gamma activity in the 50–200 Hz range was also shown to have coupling synchrony with the beta band in PD. Dystonic patients do not express this coupling. However, this does not seem to be caused by altered neural activity and may be caused by the beta band itself.^[25] With the loss of dopamine, the GPeSTN projection could cause oscillator activity by controlling the GPi-thalamus-cortex projections.^[26] The synchronisation is not limited to the basal ganglia. Tremor and akinetic PD dominant patients have shown synchronisation in the putamen, which supports the firing rates theory on striatal dopamine depletion affecting the indirect pathways of neural activity.^[27] There are indications that the striatum and cortical input to these structures drive oscillatory activity.

Studies have not shown any consistent homology for PD tremor, so the model of firing patterns may never encompass all patients.^[28] This is especially true with the differing nature of oscillations between patients. There is a minority of PD patients who do not express the beta band or gamma band synchronisation, conflicting with the idea that this concept is central to the symptomology of the patient. Therefore, it is necessary to consider both these models as essential to understanding the basic ideas behind PD pathophysiology.

Vim

The Vim thalamic nucleus has been an effective location for stimulation or lesioning to control PD tremor and ET. PD patients will typically be affected by resting tremors, while ET patients will be affected by postural or kinetic tremors. The Vim, along with the cerebellum and pons, is central to the tremor pathway. The cerebellum sends glutamatergic projections to the Vim.^[29] Thalamic inhibition of the overactive projections within the cerebello-thalamo-cortical network through Vim DBS seems to be essential to help reduce tremors. It has been suggested that high-frequency DBS will uncouple several common neural firing pathways.^[30] In ET patients with Vim DBS, regional cerebral blood flow of the cortex showed both excitatory and inhibitory effects, demonstrating the different fibre tracts that the current spreads through.^[31] Vim DBS has been shown to reduce the neural activity of the motor cortex and anterior cerebellum.^[32] Regarding the firing pattern alteration, alpha and theta activity in the cortex were desynchronised with Vim stimulation.^[33] However, thalamotomy results in similar tremor control with very different effects on the alpha and theta waves. The firing pattern rate model still needs a lot of work to fully understand due to the variability between subjects and differing literature. PD with tremor-dominated symptoms has been shown to have a larger effect with DBS treatment in the Vim over the STN. Targeting both these structures has been shown to be very effective in reducing tremors for PD patients as well.^[34] While the differing baselines of patients lead to an unclear conclusion from these results, Vim is a less optimal target for akinetic symptoms in PD. Due to this reason, it is not an effective site for other symptoms of PD compared to the STN and GPi, and it is not used often. Vim DBS is a much better option for ET patients, which is emphasised by the fact that kinaesthetic cells in the Vim are more active in ET patients than PD patients.^[35] However, there is a discussion about the fact that STN stimulation does not alter the cerebellum activity to the extent that Vim stimulation does, so it is still used in tremor-dominated PD patients.^[36] In an intraoperative study, the ventroposterior site of the Vim was shown to be the best site for implantation for tremors.^[30] Due to substantial variability in patient-specific anatomy, ongoing research with microelectrode recordings is being utilised to improve the accuracy of implantation in the ventroposterior border.^[37]

STN

The STN is the excitatory node of the basal ganglia and is central to the control of movement. The STN has been shown to help with various PD symptoms, demonstrating its importance to the pathogenesis of the disease. In animal models, it has been demonstrated that the STN in the PD state has increased firing rates and synchrony.^[38-46] STN DBS clinical improvements are correlated with the reduction in the beta band synchrony.^[24,46] The wide window of therapeutic options from STN DBS may be due to an antidromic effect of stimulation. Although not directly supported by some animal studies, one theory suggests that STN activity in the 4–6 Hz tremor range is perpetuated to the GPi and then the cortex.^[26,43] STN stimulation can disrupt this loop and can result in spontaneous activity in the basal ganglia. The tissue activated around the STN, such as the cerebellothalamic fibres or the zona incerta, can also contribute to its effects on tremors.^[44,45] However, STN stimulation seems to affect the rigidity pathway much more significantly than the tremor pathway.^[36] There is a large variability of 37–89% tremor reduction rates with STN DBS, which could be due to variable lead location and patient-specific anatomy.^[47] A recent study demonstrated the greatest motor benefits and levodopa reduction rates for PD were seen in patients with a posterolateral portion of the STN where the beta band activity is the greatest and, thereby, a potential high therapeutic region.^[48] However, this is an area in need of further research as other studies have shown conflicting outcomes based on lead location.^[48]

GPi

GPi DBS has several benefits for PD and dystonia, a neurological disorder characterised by uncontrollable muscle contractions. GPi DBS has a somatotopic organisation that affects different parts of the body at varying levels.^[49] Like the STN, GPi in the DBS reduces the beta band. However, there is conflicting literature on the effect that GPi DBS has on cortical synchronisation. Both the STN and GPi DBS are effective in reducing PD tremors.^[50,51] GPi DBS can affect the STN through the GPe cell projections from the GPi through the STN.

GPi DBS has been shown to improve dystonic symptoms better than STN DBS. However, in several PD GPi DBS cases, stimulation-induced dystonic (SIDs) symptoms were observed during the post-programming phase of DBS implantation. These symptoms may also be due to the lead placement. The lack of literature on these side-effects can be a product of the large overall motor improvements, leading to the SIDs going unnoticed. One such study noted dystonia following stimulation; yet, it did not record the precise location of contacts within the GPi, thus hindering the ability to ascertain the underlying causes of SIDs.^[52] On the other hand, GPi stimulation has been shown to improve dystonia and heighten Parkinsonism in Huntington’s Disease and craniocervical dystonia patients.^[53,54] This demonstrates the known likelihood in the connection of physiology between the movement disorders.

Dystonia electrophysiological data show a decrease in the mean discharge rates of neurons in the GPi.^[55,56] This could be a result of increased inhibitory pathways from the GPe and the striatal projection to GPi.^[57] On the other hand, PD motor symptoms have been correlated with increased GPi neuronal activity.^[58,59] PD patients who developed dyskinesias observed both reduced GPi neuronal firings and a non-continuous burst-like discharge rate.^[59] The altered synchronisation and spontaneity of the GPi neural firings will ultimately lead to distorted cortical neuron firings and changes throughout the motor cortex.

These abnormal neuronal firings from the GPi could be spread through the pallido-thalamo-cortical circuit and this pathway could explain how stimulation induces the DSIs.^[59] The ventral and dorsal GPi travel to different cortical areas, which could elucidate the varying effects of stimulations. The posteroventral GPi improves rigidity and dyskinesia, while dorsal GPi improves akinesia in PD patients.^[52] Stimulation in the dorsal GPi can also antagonise the L-dopa dyskinesia in PD patients. The connectivity of these structures could determine their respective motor effects and the pathophysiology of certain movement disorders during stimulation.