Grey matter abnormalities are associated only with severe cognitive decline in early stages of Parkinson's disease

Abstract

Cognitive impairment is common in Parkinson's disease (PD), yet with large heterogeneity in the range and course of deficits. In a cross-sectional study, 124 PD patients underwent extensive clinical and neuropsychological assessment as well as a 3T MRI scan of the brain. Our aim was to identify differences in grey matter volume and thickness, as well as cortical folding, across different cognitive profiles as defined through a data-driven exploratory cluster analysis of neuropsychological data. The identified cognitive groups ranged from cognitively intact patients to patients with severe deficits in all cognitive domains, whilst showing comparable levels of motor disability and disease duration. Each group was compared to the cognitively intact PD group using voxel- and vertex-based morphometry. Results revealed widespread age-related grey matter abnormalities associated with progressive worsening of cognitive functions in mild PD. When adjusted for age, significant differences were only seen between cognitively intact and severely affected PD patients and these were restricted to the right posterior cingulate and the right precuneus. Reduced cortical thickness was seen in the right inferior temporal gyrus and reduced folding in the right temporal region. As these differences were not associated with age, we assume that they are associated with underlying pathology of the cognitive decline. Given the limited involvement of grey matter differences, and the absence of differences in vascular changes across the groups, we hypothesize a more important role for white matter tract changes in cognitive decline in PD.

Introduction

Cognitive deficits are common in Parkinson's disease (PD) and are increasingly recognized as an integral part of the disease. Up to 40% of PD patients report mild cognitive deficits already at the time of diagnosis (Yarnall et al., 2014). Approximately 30% of patients convert to dementia during the course of the disease (Svenningsson, Westman, Ballard, & Aarsland, 2012), which increases up to 80% within 20 years after diagnosis (Aarsland et al., 2003, Hely et al., 2008). Yet, a substantial number of PD patients does not show any significant cognitive deficit, even years after being diagnosed (Foltynie et al., 2004, Janvin et al., 2003, Weintraub et al., 2011). Recent studies support the large heterogeneity in the range of cognitive deficits in PD, and several studies have attempted to identify distinctive cognitive profiles (Kehagia, Barker & Robbins, 2010). However, many studies investigated cognitive dysfunction using the diagnostic categories as defined by the Movement Disorders Society (MDS) criteria published by Litvan et al. (2012). These include PD patients without cognitive impairment, patients with mild cognitive impairment (MCI), and patients with Parkinson's disease dementia (PDD). This approach does not allow the study of milder stages of cognitive decline, including cognitive deficits that do not reach the level for a diagnosis of MCI. In addition, although recent neuroimaging studies showed clear evidence for widespread cortical atrophy associated with MCI in PD and PD dementia (Burton et al., 2004, Pereira et al., 2014, Song et al., 2011, Weintraub et al., 2011), using a classification approach may not only have masked neuroanatomical changes associated with more subtle cognitive deficits, but also obscure differences in possible sub-phenotypes that may correspond to different comorbid pathologies in more advanced stages of cognitive decline. This may be important, especially since a substantial number of PD patients appear to have neuropathological changes related to Alzheimer's disease in addition to diffuse Lewy body pathology (Jellinger et al., 2002, Kalaitzakis and Pearce, 2009, Kurosinski et al., 2002).

In a prior study (Dujardin et al., 2013), we conducted a data-driven exploratory cluster analysis using retrospective neuropsychological test data from a large sample (n = 557) of PD patients. We identified five cognitive phenotypes, that were replicated and confirmed in a more recent study by performing a model-based confirmatory cluster analysis on prospective neuropsychological data from an independent sample of mild PD patients (n = 156) (Dujardin et al., 2015). The five phenotypes represent different stages of cognitive decline ranging from cognitively intact patients to patients with severe dysfunction in all cognitive domains. In the present study, we aimed at identifying the neuroanatomical correlates of each of the identified cognitive phenotypes using structural MRI. More specifically, we were looking for regional differences in the rate of grey matter volume, cortical thickness, and cortical folding between the different groups. Considering the heterogeneity of the clinical presentation and course of cognitive deficits in PD, we expect that different patterns of cerebral changes may underlie these different phenotypes. Detection of differences in grey matter may therefore increase our understanding of the underlying neurobiological changes that accompany the spectrum of cognitive disorders in PD. Moreover, if we find alterations in patients with only slight cognitive deficits, this exploratory approach could reveal a group at risk of further cognitive decline, who may be considered for early treatment options. Thirdly, studying grey matter alterations in PD may shed light on possible sub-phenotypes in more advanced stages of cognitive decline other than PD dementia, such as phenotypes with comorbid pathologies, such as Alzheimer's disease (AD).

We hypothesized that all groups would show a decline in grey matter volume, thickness and cortical folding compared with the cognitively intact group, with a progressive severity gradient. We further hypothesized to see profound hippocampal or medial temporal lobe grey matter deviations in those patients with more profound memory deficits, possibly reflecting a sub-phenotype of PD patients with comorbid AD pathology. We followed an explorative whole brain approach, with post hoc region-of-interest analyses.