Can Neuroimaging Predict Dementia in Parkinson's Disease?

Juliette H. Lanskey; Peter McColgan; Anette E. Schrag; Julio Acosta-Cabronero; Geraint Rees; Huw R. Morris; Rimona S. Weil

Disclosures

Brain. 2018;141(9):2545-2560. 

In This Article

Abstract and Introduction

Abstract

Dementia in Parkinson's disease affects 50% of patients within 10 years of diagnosis but there is wide variation in severity and timing. Thus, robust neuroimaging prediction of cognitive involvement in Parkinson's disease is important: (i) to identify at-risk individuals for clinical trials of potential new treatments; (ii) to provide reliable prognostic information for individuals and populations; and (iii) to shed light on the pathophysiological processes underpinning Parkinson's disease dementia. To date, neuroimaging has not made major contributions to predicting cognitive involvement in Parkinson's disease. This is perhaps unsurprising considering conventional methods rely on macroscopic measures of topographically distributed neurodegeneration, a relatively late event in Parkinson's dementia. However, new technologies are now emerging that could provide important insights through detection of other potentially relevant processes. For example, novel MRI approaches can quantify magnetic susceptibility as a surrogate for tissue iron content, and increasingly powerful mathematical approaches can characterize the topology of brain networks at the systems level. Here, we present an up-to-date overview of the growing role of neuroimaging in predicting dementia in Parkinson's disease. We discuss the most relevant findings to date, and consider the potential of emerging technologies to detect the earliest signs of cognitive involvement in Parkinson's disease.

Introduction

Although traditionally considered a movement disorder, Parkinson's disease is often characterized by cognitive impairment, with dementia affecting 50% of patients within 10 years of diagnosis (Williams-Gray et al., 2013). Subtle cognitive changes are found in some patients with Parkinson's disease, even early in the disease. Where cognitive deficits do not impact on day-to-day functioning, the term Parkinson's disease with mild cognitive impairment (PD-MCI) is used (Litvan et al., 2012), with prevalence estimates of 19–42% in newly diagnosed patients (Aarsland et al., 2009, 2010; Yarnall et al., 2014). This is in contrast with Parkinson's disease dementia (PDD), where cognitive changes are seen in more than one domain and affect daily activities (Emre et al., 2007). Patients vary in the timing and severity of Parkinson's dementia. Generally, the presence of PD-MCI predicts the development of dementia (Domellöf et al., 2015). However, PD-MCI does not reliably predict transition to dementia since ~10% of individuals with PD-MCI revert to normal cognition during follow-up (Pedersen et al., 2013; Domellöf et al., 2015).

Robust neuroimaging measures to identify patients with Parkinson's disease at highest risk for cognitive decline is important for three key reasons: (i) to identify at-risk patients for clinical trials of novel disease-modifying treatments (Athauda et al., 2017); (ii) to provide prognostic information for individuals to plan their future, and enable healthcare providers to plan population health and social needs; and (iii) to uncover mechanistic explanations for underlying disease processes.

Parkinson's disease is classically associated with Lewy bodies, intracellular inclusions composed of α-synuclein (Spillantini et al., 1997). However, cognitive involvement in Parkinson's disease is most strongly related to the combination of Lewy bodies with Alzheimer's pathology, in particular fibrillary amyloid-β and intraneuronal hyperphosphorylated tau tangles (Compta et al., 2011). Evidence is also emerging for a synergistic relationship between α-synuclein and amyloid-β. For example, in a large retrospective study, a strong correlation was shown between extent of neurofibrillary tangles, neuritic plaques and α-synuclein (Irwin et al., 2017). In vitro models demonstrate that this relationship is causative, with amyloid-β inducing conformational changes in α-synuclein (Swirski et al., 2014). Intriguingly, certain distribution patterns of pathological inclusion across the cortex at post-mortem are strongly linked with more rapid progression of dementia in life in patients with Parkinson's disease. Specifically, patients with a high burden of Lewy-related pathology in occipital regions showed more rapid progression to dementia (Toledo et al., 2016).

Axonal involvement appears to be critical in the pathophysiology of Parkinson's disease and associated cognitive involvement. α-Synuclein accumulation may begin in the axonal compartment (Chung et al., 2009) with dystrophic changes in axons occurring before neuronal loss. Notably, cells that are especially vulnerable in Parkinson's dementia, including cholinergic cells of the nucleus basalis of Meynert and serotonergic cells of the raphe nucleus share the common morphological phenotype of long axonal projections (Perry et al., 1985; Hale and Lowry, 2011; Wu et al., 2014). Therefore, neuroimaging techniques sensitive to specific pathological accumulation, particularly in occipital regions, and those that detect axonal damage or alterations in neurotransmitter levels are most likely to detect the earliest stages of Parkinson's dementia.

Until recently, neuroimaging has not had a large role in predicting cognitive involvement in Parkinson's disease. This is unsurprising, given that conventional methods rely on loss of volume caused by neuronal death, a relatively late event in Parkinson's dementia (Schulz-Schaeffer, 2010; Hattori et al., 2012). Furthermore, there are few longitudinal neuroimaging studies of the early signs of dementia in Parkinson's disease. Where these are lacking, cross-sectional studies that detect differences between Parkinson's disease patients with and without early cognitive involvement can provide insights into the power of these techniques to identify patients likely to progress to Parkinson's dementia.

A further, important consideration is that cognitive dysfunction in Parkinson's disease is a heterogeneous entity, especially at the very earliest stages. Two distinct phenotypes are now recognized (Williams-Gray et al., 2009): a fronto-striatal/executive pattern, which is related to dysfunction in dopaminergic fronto-striatal networks; and a posterior cortical/visuospatial phenotype, with a non-dopaminergic substrate, that may involve changes in cholinergic transmission (Klein et al., 2010), or excess cortical protein aggregation, as implicated by association with the MAPT genotype (Nombela et al., 2014). In longitudinal population studies, the fronto-striatal phenotype does not always progress to Parkinson's dementia (Williams-Gray et al., 2013) and therefore imaging techniques sensitive to executive dysfunction may have less value in predicting the earliest stages of Parkinson's dementia than those sensitive to visuospatial and cholinergic dysfunction.

New emerging technologies show potential for detecting even subtle cognitive involvement in Parkinson's disease that will need confirmation in longitudinal progression studies. Here we provide an up-to-date overview of the potential role of neuroimaging in predicting dementia in Parkinson's disease. We consider conventional methods and then examine the sensitivity of emerging technologies to detect the earliest signs of cognitive involvement in Parkinson's disease.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....