OUP user menu

Pathological network activity in Parkinson's disease: from neural activity and connectivity to causality?

Lars Timmermann, Gereon R. Fink
DOI: http://dx.doi.org/10.1093/brain/awq381 332-334 First published online: 29 January 2011

Pathophysiological changes in the basal ganglia thalamocortical loops, first described ∼20 years ago by Alexander et al. (1990), are commonly assumed to underlie key symptoms of Parkinson's disease. In their model, Alexander et al. (1990) described a large network of inhibitory and excitatory connections between different subnuclei of basal ganglia, thalamus and cortex. In the normal (i.e. physiological) state, two distinct loops regulate basal ganglia activity: a direct loop between the striatum and the internal segment of the globus pallidus; and an indirect loop between this structure and the striatum (in this loop, activity is relayed via the subthalamic nucleus). In the healthy state, these two loops are balanced resulting in well-regulated activity of the internal segment of the globus pallidus and the tightly connected pars reticulata of the substantia nigra. Besides its influence on brainstem activity, the ‘output-region’ of this basal ganglia network is assumed to project to the thalamus, thereby modulating thalamo-cortical interactions. Albeit simplistic, the degeneration of dopaminergic neurons in the pars compacta of the substantia nigra is to date considered the key neuropathological change underlying Parkinson's disease. Based upon neuropathological findings, after degeneration of dopaminergic cells in the substantia nigra the Alexander et al.'s (1990) model proposes affection of the indirect loop leading to reduced inhibition of the subthalamic nucleus, and stronger excitation of the internal segment of the globus pallidus and the pars reticulata of the substantia nigra. Conversely, the model predicts that the direct inhibitory projections of the striatum to the internal segment of the globus pallidus and the pars reticulata of the substantia are reduced. The sum effects of these changes are ‘overactivity’ in the internal segment of the globus pallidus and the pars reticulata of the substantia and an inhibition of thalamo-cortical interactions. This model could then explain Parkinsonian symptoms such as bradykinesia and also a number of interesting neurophysiological observations in patients with Parkinson's disease including changes in firing rates. Although the described model most likely oversimplifies the changes in interconnections in the basal ganglia-cortical loops resulting from Parkinson's disease, it does suggest a novel, albeit hypothetical, therapeutic approach: reduction or modulation of pathological activity in the subthalamic nucleus or the internal segment of the globus pallidus might result in rebalancing the direct and indirect loops. Consequently, both areas have been targeted using deep brain stimulation to induce an electrophysiological ‘blockade’ of pathological activity (Bergman et al., 1990), a method now well established in the treatment of patients with Parkinson's disease (Deuschl et al., 2006).

One of the key issues of Parkinson's disease research over the past 20 years has been the nature of pathological activity in the basal ganglia-cortex loops. Experimental evidence from rats and monkeys as well as from human recordings indicates that pathological oscillatory activity and synchronization of neural activity are key mechanisms of basal ganglia pathophysiology. Two different frequencies of pathological activity are associated with the main clinical symptoms of Parkinson's disease. A ∼20 Hz ‘β-band’ activity seems closely related to bradykinesia and rigidity (Brown et al., 2001; Kuhn et al., 2004, 2005, 2006). Especially in the subthalamic nucleus, local field potentials show pronounced oscillatory activity in the β-band (Kuhn et al., 2004); modulation of this activity is closely associated with bradykinesia and levodopa treatment has been shown to suppress the activity (Kuhn et al., 2006). Consistent with these findings, deep brain stimulation also suppresses β-band activity and reduces bradykinesia in a close temporal association (Kuhn et al., 2008). Further evidence for a key role of β-band activity in the generation of bradykinesia stems from the observation that this increases when patients are stimulated with a frequency of 20 Hz (Fogelson et al., 2005; Eusebio et al., 2008). However, the pathological 20 Hz subthalamic nucleus activity does not seem to be directly connected with muscle activity. Coherence measurements have revealed that the 20 Hz β-activity in the subthalamic nucleus is closely coupled with muscle activity, but this coherence does not correlate with bradykinesia (Reck et al., 2009a).

The second well-characterized pathological frequency in patients with Parkinson's disease is ∼10 Hz activity. This is related to tremor (Volkmann, 1998; Timmermann et al., 2003; Reck et al., 2009b, 2010). In patients with tremor-dominant Parkinson's disease, ∼10 Hz activity is found in fine-segregated muscle-specific subloops with strong coupling to the tremor muscle (Reck et al., 2009b). Deep brain stimulation in areas with pronounced pathological activity of ∼10 Hz has been shown to reduce tremor (Reck et al., 2009b), whereas 10 Hz stimulation in the subthalamic nucleus worsens the symptoms of Parkinson's disease (Timmermann et al., 2004). Obviously, pathological ∼20 Hz as well as ∼10 Hz activity play a considerable role in the generation of symptoms. But how do these different pathological frequency bands relate to each other and the brain areas implied by the Alexander and DeLong model?

Using magnetoencephalography and simultaneous electromyography recordings in patients with tremor-dominant Parkinson's disease, a network of cortical and subcortical motor and sensory areas has been identified (Timmermann et al., 2003). These are highly coherent around 10 Hz, double the tremor frequency, and also at 5 and 20 Hz. The network consists of the primary motor cortex, the premotor cortex and the supplementary motor cortex as well as deep diencephalic (presumably thalamic) areas and the cerebellum. In healthy volunteers mimicking Parkinson's disease tremor, an anatomically similar network with a different connectivity pattern is described with cerebral coupling also at around 10 Hz (Pollok et al., 2004). Therefore, one can assume that sensorimotor network activity around 10 Hz is physiologically preformed but pathologically exacerbated in patients with Parkinson's disease resulting in the typical alternating pattern of the tremor. Consistent with this hypothesis, simultaneous magnetoencephalography and subthalamic nucleus local field potential recordings in patients with Parkinson's disease very recently indicated ∼20 Hz network activity as well as ∼10 Hz activity connected to temporal brain areas (Hirschmann et al., 2011). This frequency-dependent distribution implicates a functional separation of the different basal ganglia-cortex loops with different frequency patterns. However, although connectivity between brain areas documents a functional relationship, this does not allow the brain area(s) that primarily ‘drive’ the pathological activity, and which are ‘hooked on’, to be differentiated. Therefore, causality measures such as the partial directed coherence have recently been applied to study the direction of coupling between brain areas, and also between brain activity and peripheral activity measured e.g. by electromyography (Florin et al., 2010a, b).

In the current issue of Brain, Litvak et al. (2011) have investigated Parkinson's disease network activity in the resting state. The authors use combined magnetoencephalographic recordings and local field potentials in the subthalamic nucleus and a methodologically new approach. Based on spectral and spatial cluster analysis, they identify not only the temporal–spatial relationship between different sensorimotor areas, but also the causal relationship between the interconnected brain areas. Interestingly, this new approach reveals two distinct networks that are temporospatially different: one in the previously described 10 Hz α-range comprising temporoparietal-brainstem areas and another in the ‘β-band’ (around 20 Hz), a predominantly frontal network. This separation into two different oscillatory networks resembles recent findings of Hirschmann et al. (2011), who also detected cortical areas in the temporal lobe with high coupling to the subthalamic nucleus in the ∼10 Hz range. The extremely interesting novel approach of Litvak et al. (2011), however, combines a temporospatial network cluster analysis with causality analysis, and is therefore the first approach to show without any prior assumptions that a ∼10 and ∼20 Hz network exists independently in Parkinson's disease patients, and having highly specific and different involvement of cortical and subcortical brain areas. Furthermore, implementation of the causality analysis in this approach allows a specific description of the information flow in the individual networks.

What is the relevance of these novel findings? The currently most interesting and promising therapeutic approach to modulate pathological oscillatory activity is deep brain stimulation. The current techniques used in Parkinson's disease are based upon high-frequency stimulation. Albeit clinically effective (Deuschl et al., 2006), this stimulation has little specificity with respect to the targeted pathological oscillations. The findings of Litvak et al. (2011) suggest that an individual characterization of pathological activity and its specific temporospatial pattern are now feasible. At least two important issues arise from these findings: what is the specific pathophysiological role of these different networks; and how can these be targeted and modulated by ‘intelligent’, i.e. specific deep brain stimulation algorithms? These questions need to be addressed in future studies. But the paper by Litvak et al. (2011) is a further key milestone on the long journey from activity through connectivity to causality and thereby contributes to the future of an individually tailored therapeutic neuromodulation in Parkinson's disease.


‘Manfred und Ursula Müller Foundation’ (T 159/2006); ‘Klüh-Foundation’ to L.T.; German Ministry of Education and Research (BMBF), German Research Foundation (Deutsche Forschungsgemeinschaft, DFG; Clinical Research Group 219) to L.T. and G.R.F.


View Abstract