Brain Sensing in Neurological and Psychiatric Disorders

Both hyperkinetic and hypokinetic movement disorders are associated with abnormal spatiotemporal activity within the basal ganglia circuitry13,16,17. This can be assessed through measuring local field potentials (LFPs), which represent the product of synchronous neuronal activity at a given site (unsynchronized, random activity being essentially cancelled out)13. Numerous other disorders such as obsessive compulsive disorder, major depression, Tourette's syndrome, epilepsy, eating disorders and cluster headaches are also amenable to successful modulation using DBS6-11. These disorders likely also have unique, disease-specific electrophysiological signatures, the exact nature of which remains to be thoroughly defined. Abnormal electrophysiological activity is useful not only in delineating the pathophysiologic underpinnings of these disorders, but is central to the future development of adaptive DBS systems which respond in real-time to ameliorate pathological brain acticity12. Adaptive DBS may provide further clinical benefit beyond currently employed continuous DBS approaches, with only a fraction of the energy requirements12.

Studies using microelectrode recordings at the time of DBS lead placement as well as recordings from externalised DBS electrodes have identified distinct neurophysiological signatures within different disorders. Examples include excess beta-frequency oscillations in Parkinsonism, alpha and theta frequency oscillations in dystonia and gamma oscillations in dyskinesia. These neurophysiologic biomarkers of disease can be affected by the application of high-frequency DBS. Future closed-loop DBS systems may rely on real-time suppression of such abnormal basal ganglia activity.

LFPs in Parkinson's disease

A significant body of work has confirmed that beta-frequency oscillations, recorded from both the subthalamic nucleus and the internal portion of the globus pallidus, correlate with severity of bradykinesia and rigidity in Parkinson's disease13,16-18. These beta-frequency oscillations are coherent across simultaneous recordings in different nuclei of the same patient, implying that these represent a network-level dysfunction in Parkinson's disease16,19. The amplitude/power of abnormal beta-frequency LFPs correlates with the severity of motor impairment in Parkinson's disease15,18. Moreover, beta-frequency LFPs can be suppressed both by levodopa administration, or by the application of high-frequency DBS20,21; in both scenarios the degree of suppression in beta-oscillations correlates with the degree of clinical motor improvement. Beta oscillations may therefore represent an electrophysiological parkinsonian symptom correlate which can act as a biomarker of the motor state. Hence, they may be useful signals on which to base stimulation using adaptive DBS technologies. However, some observations, such as the suppression of beta frequency oscillations during periods of tremor, have cast doubt on the robustness of this potential biomarker.

Other LFP frequency alterations have also been observed to correlate with clinical symptomatology in PD. For instance, synchronisation at frequencies in the gamma range (60-90Hz) have been correlated with dyskinesia, as have synchronisation at lower frequencies (4-8Hz)22,23. High-frequency oscillations in the 250Hz range have also been found to associate with parkinsonian clinical states and to shift to even higher frequencies (350Hz) following levodopa administration24. In contrast to beta-frequencies, changes in high-frequency LFPs do appear to correlate with tremor25.

Aside from pure frequency characteristics, the temporal distribution of beta-frequency oscillations has also been examined in different disease states. It is suggested that prolonged periods of beta synchronisation (beta bursts) may be responsible for the overall increase in beta power in patients with Parkinson's disease in the OFF state, and that a move to shorter durations of synchronisation may occur in the ON state26.

LFPs in dystonia

Patients with dystonia display a distinct pattern of LFP alterations in the pallidum13,14, particularly an excess of synchronized oscillatory activity in the low frequency 3-12 Hz band13. Not only is pallidal LFP power prominent in the 3-12Hz band in patients with dystonia13,27 but it correlates with28 and may be coherent with (especially contralateral) dystonic muscular activity29. Moreover, there is a direct association between pallidal low-frequency oscillations and dystonia severity30.

Increased LFP power in the low-frequency band is seen across a variety of dystonia phenotypes. In cervical dystonia, inter-hemispheric differences in LFPs have been demonstrated, though clear correlations with directional head movements have not been established31,32. In myoclonus-dystonia patients, increased pallidal LFP in the 3-15Hz range correlate with dystonic muscle activity33. Oscillatory activity may be different in secondary or combined dystonia, and may also differ according to genetic makeup 34-36.

LFPs in essential tremor

LFP recordings from patients undergoing thalamic DBS in ET have identified distinct thalamic LFP frequency characteristics, often coherent with surface EMG recordings in these groups, at frequencies which are either harmonics or subharmonics of the tremor frequency37.

LFPs in other conditions Electrophysiologic recordings from patients with Tourette's syndrome have shown an excess of low-frequency activity(2-13Hz) in both thalamic and pallidal targets38 and that spontaneous tics in Tourette's syndrome are preceded by coherent thalamo-cortical discharges39. Unique, disease specific LFP patterns have also been identified in a number of neuropsychiatric disorders. For example, prominent alpha activity in the basal ganglia has been suggested as an electrophysiologic correlate of major depressive disorders40

Continuous recording from fully implanted DBS systems has only recently become possible, for example using the Medtronic Percept PC battery system. The quality of signals recorded using this system as well as the correlation with clinical features, and coherence with cortical and motor activity however remain largely unexplored. Moreover, the number of patients on which previous LFP studies has been based is small. Hence, their utility in the future development of closed-loop DBS systems remains uncertain.

Our study will address a number of unanswered questions relating to the utility of LFP recordings in neurological disorders including:

1. Is it possible to replicate the previous findings concerning LFP signatures in various movement disorders using currently available DBS systems with LFP sensing capabilities e.g. Medtronic Percept PC?
2. Do LFP signatures persist many years after initiation of deep brain stimulation (most previous studies have evaluated patients only at the time of DBS implantation)?
3. How are LFP signatures related to modulations of wider brain networks as assessed by EEG?
4. Can non-invasive recording methods such as EEG provide alternative or complementary biomarkers for closed-loop DBS?
5. What is the influence of medications, body movements, sensory inputs and differential stimulation parameters on LFPs?
6. How robust are LFP signatures within and between individuals with similar disease states?
7. Do robust LFP signatures exist for non movement disorder basal ganglia network pathologies, and how do these correlate with symptoms severity?