Longitudinal magnetic resonance imaging of cognitive impairment in Parkinson’s disease

Parkinson’s disease (PD) is a neurodegenerative movement disorder characterized by slowness of movement, rigidity, and tremor. However, most patients additionally develop cognitive impairment and eventual dementia (PDD), which becomes the most burdensome aspect of the disease. Pathological processes associated with Parkinson’s extend beyond the classic neurodegenerative changes of neuronal damage in the substantia nigra and the aggregation of misfolded alpha-synuclein protein, leading to the relatively recent understanding of Parkinson’s as a multi-system disorder. Cognitive impairment in PD can vary in the timing of presentation, but dementia eventuates in about 80% of patients. A more mild manifestation of cognitive impairment, also known as the “Mild Cognitive Impairment” or “PD-MCI”, is found in over a third of newly diagnosed Parkinson’s disease patients. Identifying individuals with PD-MCI early in the disease process may eventually facilitate the implementation of novel therapeutic options prior to development of the debilitating stage, dementia. Currently, there are no objective or clinically useful markers for cognitive impairment in PD. However, recent neuroimaging techniques have shown promise in this regard. Magnetic resonance imaging (MRI) is a non-invasive medical imaging technique that may potentially be used to objectively characterize the structural and functional changes in the brain in relation to cognitive impairment in PD. In this thesis, 138 participants meeting the UK Parkinson’s Disease Society’s criteria for idiopathic PD and 50 matched healthy controls completed extensive neuropsychological testing. On the basis of this testing, participants were classified as having normal cognition (PDN=79), mild cognitive impairment (PD-MCI=36), or dementia (PDD=23). Participants also completed an MRI scanning session. These participants were then followed up with the same neuropsychological battery and MRI scanning approximately every two years, with some completed assessments up to six years after baseline. Using a three tesla MRI scanner, three types of MRI data were acquired for each participant: (1) structural T1-weighted images to assess cortical thickness and surface area, (2) MR spectroscopy (MRS) to explore the metabolic changes of the posterior cingulate cortex, and (3) resting-state functional MRI to evaluate functional connectivity of the default mode network. In order to properly model the longitudinal nature of the study, I used Bayesian generalized linear multilevel models to analyse the three MRI data types. The analysis was aimed at evaluating the within- and between-subject association of the MRI-derived metrics and participants’ cognitive impairment. Analysis of structural MRI scans (cortical thickness “CTh” and surface area “SA”) showed strong association with cognition and cognitive decline over time. Baseline cognitive ability was associated significantly with cortical thinning and surface area reduction. However, most importantly, longitudinal assessment showed that cognitive deterioration of PD patients was associated with reduced cortical thickness and surface area in several brain regions. These structural findings, particularly the longitudinal ones, indicate the potential role of both CTh and SA as predictive markers for cognitive impairment in PD. After accounting for age, sex, and motor impairments, none of the MRS-derived metabolites extracted from the posterior cingulate cortex (PCC) showed significant group differences at baseline. Similarly, metabolite changes overtime did not significantly associate with declining cognitive ability of the study participants. These findings indicate that MRS of the PCC is not a clinically useful marker of cognitive impairment in PD. Resting state functional connectivity (RS-fMRI) of the default mode network (DMN) revealed no significant relationship between baseline nor decline in cognitive ability over time and DMN functional connectivity. While DMN dysfunction is strongly related to cognitive impairment and decline in Alzheimer’s disease, the current findings suggest that DMN functional connectivity does not hold the same promise in PD. Hence, it also appears that DMN connectivity does not provide clinically useful information about cognitive status or decline over time in PD. In this thesis, posterior cingulate MRS and DMN connectivity did not provide clinically reliable information about cognitive impairment in PD. However, both cortical thickness and surface area showed reliable and robust association with cognitive ability in PD, at cross section and over time. These results suggest that longitudinal structural MRI measurements may hold promise as outcome measures, along with complimentary clinical and cognitive assessments, in future PD-modifying therapeutic trials