Abstract
Parkinson’s disease (PD) is the second most prevalent neurodegenerative disorder. While diagnostic symptoms are primarily motor-focused, cognitive impairment is a significant, common, and important feature as the disease progresses. Identifying the underlying brain changes associated with cognitive decline and dementia in PD is an important goal; not only does this allow better understanding of disease progression, but it may also motivate and facilitate the assessment of new disease-modifying therapies. Previous work suggests that PD is associated with alterations in functional electroencephalography (EEG) activity, disruptions in connectivity, and compromised network integrity and structure. In order to more accurately interpret neural signals associated with cognitive decline in PD, the current thesis examined the relationship between parameterised EEG spectra, functional connectivity (FC), network topology, and brain structure and cognitive decline in PD.
A total of 164 participants (118 PD and 46 controls) were recruited from the Movement disorders clinic at the New Zealand Brain Research Institute (NZBRI) between 2018-2022, as part of the New Zealand Parkinson’s Progression Programme (NZP3). PD diagnoses followed the UK Brain Bank and Movement Disorders Society (MDS) criteria; comprehensive neuropsychological testing, following MDS criteria, was used to classify participants as PD with normal cognition (PDN), PD with mild cognitive impairment (PDMCI), and PD with dementia (PDD). After quality control and additional exclusions, the final sample comprised 120 individuals: 29 HC, 50 PDN, 32 PDMCI, and nine PDD.
We used EEG to measure spontaneous brain activity during rest (with eyes closed) over 9 minutes. I used the Fitting Oscillations and 1/f (FOOOF) algorithm in an extended alpha band (5-13 Hz) and beta band (13-30Hz) to analyse periodic (rhythmic activity) and aperiodic (arrhythmic, typically 1/f-like activity) components of EEG activity. The metrics included periodic: centre frequency, power and bandwidth, aperiodic: offset, and exponent. Connectivity and network integrity were examined using the debiased weighted phase lag index (dwPLI) and Brain Connectivity Toolbox (BCT) measures of global efficiency, transitivity, assortativity, characteristic path length (global), local efficiency, clustering coefficient, degree, strength, and characteristic path length (local). T1-weighted MRI was used to capture structural information and was investigated using cortical thickness, surface area, and gray matter volume.
I first analysed associations between aperiodic and periodic measures and age using multiple linear regression, in sensor space. I then assessed whether a significant association existed between a continuous measure of global cognitive ability and FOOD-deprived EEG metrics across sensors in the HC and PD group. Lastly, I investigated any differences across the cognitive subgroups, using an ANOVA model, accounting for age and sex. All comparisons were performed across EEG sensors, in sensor space, and were corrected for multiple comparisons using False Discovery Rate (FDR-corrected p<0.05).
In a second series of analyses, source-space derived functional connectivity and BCT measures were investigated using the same three questions: (1) associations with age, (2) a continuous measure of global cognitive ability, and (3) differences among cognitive subgroups.
Lastly, I looked at the same age and global cognitive ability correlations with our structure metrics. Followed by an investigation of the relationship between structural brain measures and summary EEG-derived metrics.
We observed a decrease in extended alpha centre frequency and beta power with cognitive decline in PD, aligning with known EEG patterns. Notable group differences were found between HC and both PDMCI and PDD groups. Unlike previous findings, aperiodic components showed no group differences, but were negatively correlated with cognitive performance in PD, underscoring the importance of EEG analysis methods in understanding PD's neural impact.
Our study found no clear link between dwPLI and overall cognitive function in PD. Differences between PDMCI and PDD were minimal, with a slight increase in connectivity in PDMCI, suggesting possible compensatory brain mechanisms that may falter in PDD, leading to greater cognitive decline. Age was correlated with declines in global efficiency, transitivity and characteristic path length, suggesting reduced information integration and network segregation as age advances, reflecting diminished brain functional integrity.
We identified significant structural brain changes associated with cognitive decline in PD, including changes in cortical thickness and volume, with key findings in the temporal lobe. Correlations between structure and age while not tested statistically, suggest similar patterns in healthy controls and in PD. Structure was associated with functional EEG metrics; cortical thickness was correlated with beta power in HC, while gray matter volume was correlated with extended alpha centre frequency in PD. Furthermore, volume and surface area were correlated with aperiodic components in the temporal lobe, specifically in PD, highlighting unique structural-functional relationships in PD progression.
Our research builds on previous studies by analysing both periodic and aperiodic EEG components in PD using FOOOF within an extended alpha band. This novel approach in the context of PD highlights changes in frequency-specific characteristics linked to disease progression and cognitive status. By separating the aperiodic and periodic components, we can better discern the distinct EEG patterns associated with PD, offering deeper insights into the disease's neurological impact and its progression. Furthermore, linking EEG patterns, including aperiodic activity, to structural alterations, such as temporal lobe atrophy, our research underscores the interconnectedness of brain function and structure in PD, offering new insights into the neurobiological underpinnings of the disease. Moreover, our results broaden the understanding of the dynamic interplay between functional and structural changes accompanying cognitive decline in PD, offering valuable directions for future research.