Parkinson's disease (PD) is a common neurodegenerative disease characterised by a gradual accumulation of Lewy bodies and death of dopaminergic neurons.1,2 The Lewy body pathology of PD begins long before the manifestation of motor symptoms. Accumulation of Lewy bodies is initially found in the olfactory bulb and brain stem and then spreads to the substantia nigra pars compacta, followed by several brain regions, including the basal ganglia and the neocortex.3 The progressive structural and neurochemical changes in PD are accompanied by widespread functional changes in neuronal activity, which in turn lead to worsening clinical signs and symptoms such as tremor, rigidity, and bradykinesia and co-occurring non-motor symptoms like sleep disorders, depression, fatigue, and cognitive deficits.1
The changes in brain function in PD are particularly prominent in the oscillatory activity of neurons.4 In PD, spontaneous oscillatory beta band (13–30 Hz) activity in the sub-thalamic nucleus (STN) exhibits a systematic disease-related increase in synchronicity that is related to the dopamine level5–8, and correlates with the severity of bradykinesia and rigidity symptoms.9,10 Changes in the beta band extend beyond the STN through the basal ganglia-thalamic cortical sensorimotor network. The cortical manifestation of the disease-related changes in the sensorimotor network can be measured non-invasively from the cortex, using electro- or magnetoencephalography (EEG/MEG). Such non-invasive neural recordings can potentially provide easily available prospective biomarkers of disease or symptom-related neural changes in PD. Increased oscillatory beta-band activity in the sensorimotor cortex has been linked to increased symptom severity, such as rigidity and bradykinesia.11,12 The role of dopamine on the cortical beta band is, however, still unclear. There is no consensus on how dopaminergic medication affects cortical beta-band power, with some studies reporting no effects11,13−15 and others an increase in beta-band power.16–18 Deep brain stimulation of the STN in PD patients has been shown to lead to a decrease in the power of spontaneous activity in the cortical sensorimotor beta and alpha (8-12Hz) bands19,20 (but see also16,21).
Importantly, there is evidence that the beta-band changes are not in the same direction across the different stages of PD. For example, there are reports of increased cortical beta-band power in the early stages of PD22, whereas the later stages are associated with decreased beta-band power.23 Further, the beta-band power is not the only feature of the sensorimotor rhythms that is altered in PD. Several studies have found a shift in the beta-band centre frequency (the frequency at which the power spectrum density peaks in the beta-band) towards a lower frequency in PD patients compared to healthy controls.24–26 The shift towards lower beta-band centre frequency is more pronounced in PD patients with dementia27–30 and correlates with reduced cognitive ability.26,31 Notably, the centre frequency shift is detectable already in the early stages of PD25 and does not seem to be affected by dopaminergic medication.32 The changes in beta-band power and centre frequency in PD could indicate that different features of the oscillatory beta-band activity reflect different underlying neural functions expressed in the measured sensorimotor signals. Changes in beta-band power could be functionally related to sensorimotor disturbances, and changes in centre frequency could be related to cognitive function.
The characteristics of neuronal oscillatory activity may hold additional information of disease-related changes in PD.33 Both beta-band power and centre frequency reflect a quantification of power spectral density (PSD). While these features can provide valuable information about disease-related changes in PD, the PSD quantification of a neural time series provides a static summary of the oscillatory activity across the entire time series. PSD does not account for inherent dynamics in this activity or changes in the time series on shorter time scales—as is prevalent in neural time series. The beta-band exhibits a great degree of variation over time and contains characteristic high-amplitude "bursts" that last about 50–200 ms, both in the cortical and sub-cortical beta-bands.34–37 Functionally, the transient bursts appear to play a pivotal role in sensorimotor processing through the basal ganglia-thalamic-cortical network. For instance, the presence of a beta burst in the sensorimotor cortex close to a tactile stimulation decreased the likelihood of tactile detection38, and the rate of beta bursts is shown to decrease in the time leading up to a movement both in STN39–41 and in the sensorimotor cortex.42,43
In PD, quantification of beta-band burst activity from recordings in the STN has shown that beta-burst rate and duration are reduced by dopaminergic medication44,45 and deep brain stimulation.37 Furthermore, PD patients exhibit a decrease in the rate of beta burst at the cortical level compared to healthy controls.14 This decrease in beta burst rate is inversely related with increased severity of motor symptoms;46 particulary bradykinesia and postural-kinetic tremor symptoms, but there is not evidence pointing to an effect of dopaminergic medication on cortical bursting properties.14 Notably, the burst rate showed a higher sensitivity than PSD beta power for discriminating PD patients from healthy controls, demonstrating that the choice of method for analysing beta-band features influences the sensitivity of subsequent analyses. This is further complicated by the fact that in addition to disease-related changes, these features likely differ with age,43,47 and the fact that most studies on oscillatory changes in PD come from studies with small sizes.48 The central challenge is quantifying the measured neural signals to extract the disease's relevant features from the signals, be it the spectral power, centre frequencies, or burst-like features.
In the current study, we aimed to compare how different oscillatory features of cortical sensorimotor activity change in PD to elucidate what oscillatory features in the neural time-series differ between PD patients and healthy controls and how these features are associated with different motor symptoms in PD. We extracted the sensorimotor neural resting-state activity from source reconstructed resting-state MEG signals in the sensorimotor cortex (Fig. 1) and quantified the time-series in terms of the PSD in the canonical mu-band (8–30 Hz).49,50 In addition to the band-specific analysis, we compared the 1/f broadband characteristics of the PSD.51,52 Finally, we compared features of the sensorimotor rhythm in terms of time-domain analysis of spontaneous transient bursts.14,38 We tested the hypotheses of altered functional changes in PD by analysing how these features differed between PD patients and healthy controls and further investigated the interactions with age and sex. As ageing is associated with structural and functional changes in the sensorimotor cortex53,54, we investigated if the potential changes in sensorimotor activity in PD differed across age. Since both healthy ageing and PD disease progression are linked to thinning of the cortex55,56, we further included thickness of the sensorimotor cortex in the analysis as a potential mediating factor on the sensorimotor activity that potentially also interacts with disease state.
The central hypothesis was that there would be differences between healthy controls and PD in features of the sensorimotor signals, but also that different features may be related to different functional changes. We hypothesised that individual oscillatory features would reflect different underlying neural functions in the sensorimotor system and thereby show different relationships to the clinical manifestations of specific motor symptoms in PD. We tested this hypothesis in two steps: first, examining the inter-relationship between all different measures, and subsequently, examining what feature—or combination of features—best explained the variation in severity within each motor symptom.