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

White Matter Changes and Diffusion MRI

White matter changes are potentially more sensitive to early processes in Parkinson's disease as they represent degeneration of axons and myelin damage, which may occur early in the course of the disease (Burke and Malley, 2013). Diffusion-weighted MRI (DWI) can provide in vivo information about microstructural integrity both in grey and white matter tissue (Le Bihan et al., 2001). Diffusion tensor imaging (DTI) is a technique that can reliably characterize such restriction by modelling the displacement of water molecules as a rotationally invariant tensor. The diffusion tensor is decomposed into a set of primary components that are then recombined as DTI metrics. These include mean diffusivity, which characterizes the overall molecular displacement, and fractional anisotropy, which indirectly captures the spatial coherence of such displacements, thus reflecting the level of restriction imposed by the parenchymal microstructure. Both metrics generally work on the assumption that water molecules are less coherently restricted as a result of disease processes such as axonal loss.

As measured by DTI, white matter alterations in Parkinson's disease increase as cognition worsens (Kamagata et al., 2013; Melzer et al., 2013; Agosta et al., 2014) (Figure 2A). Fractional anisotropy is reduced in PDD (relative to controls) in major white matter tracts (Deng et al., 2013; Kamagata et al., 2013). In cross-sectional DTI studies in Parkinson's disease where white matter and grey matter were analysed concurrently, significant white matter alterations were identified in Parkinson's patients without dementia where signs of grey matter atrophy were yet unremarkable (Hattori et al., 2012; Agosta et al., 2014; Duncan et al., 2015). Such white matter changes included fractional anisotropy reductions (Hattori et al., 2013; Agosta et al., 2014) and mean diffusivity increases (Duncan et al., 2015) in the inferior and superior longitudinal fasciculi and inferior fronto-occipital fasciculus. In these studies, changes in grey matter were only detectable in patients with dementia (Hattori et al., 2012). This suggests DTI might be more sensitive to changes in white matter microstructure as early signs of cognitive involvement in Parkinson's disease when compared to measures of atrophy, but further longitudinal studies will be needed to establish the temporal sequence.

Figure 2.

White matter changes in Parkinson's with cognitive involvement and changes in brain connectivity associated with cognitive changes in Parkinson's disease assessed using graph theoretical approaches. Tract-based spatial statistics results in Parkinson's patients with differing degrees of cognitive involvement. Voxel-wise group differences are shown in red (decreased fractional anisotropy), overlaid on the white matter skeleton (in green). Comparison of white matter integrity in this way reveals decreased fractional anisotropy and increased mean diffusivity in several major white matter tracts in Parkinson's patients with cognitive involvement. (A) Comparison of PDD and cognitively normal Parkinson's disease (PD). Adapted from Kamagata et al. (2013). (B) Associations between mean diffusivity and performance in semantic fluency task. Tract-based spatial statistics map showing areas of increased mean diffusivity (in yellow-red) in the white matter of patients with Parkinson's disease. A significant association is seen between increased mean diffusivity and lower semantic fluency score. Adapted from Duncan et al. (2015). (C) Comparisons between controls and PD-MCI using network-based statistics. Schematic representation of the component consisting of 235 edges considered significantly different between the groups. Brain nodes are scaled according to the number of edges in the significant component to which they are connected. Adapted from Abós et al. (2017). (D) Connectograms comparing patients with Parkinson's disease divided according to cognitive ability into four groups, where Group 1 is cognitively normal, and Group 4 has dementia. As cognitive impairment worsened, functional connectivity decreased. Between-group differences in functional connectivity especially concerned the ventral, prefrontal, temporal and occipital cortices. Links are coloured by connection type: left intrahemispheric (blue), interhemispheric (red) and right intrahemispheric (green). Brain regions are represented symmetrically. FA = fractional anisotropy; Fr = frontal; Ins = insula; Lim = cingular limbic; Par = parietal; Occ = occipital; Sbc = subcortical; Tem = temporal. Adapted from Lopes et al. (2017).

When particular cognitive domains are examined in patients with Parkinson's disease, abnormal tissue diffusivity is seen in specific cortical patterns (measured using fractional anisotropy and mean diffusivity) (Figure 2B). This is found for memory (Carlesimo et al., 2012; Melzer et al., 2013; Zheng et al., 2014), attention (Melzer et al., 2013; Zheng et al., 2014), executive function (Melzer et al., 2013; Theilmann et al., 2013; Zheng et al., 2014), language (Zheng et al., 2014; Duncan et al., 2015), and visuospatial domains (Theilmann et al., 2013). Mean diffusivity of parietal and frontal subcortical tracts is higher in early stage Parkinson's participants with impaired semantic fluency (Duncan et al., 2015), a measure that has been linked with dementia risk. Moreover, increased mean diffusivity is seen prior to reductions in fractional anisotropy or grey matter volume (Melzer et al., 2013).

Despite the sensitivity of fractional anisotropy and mean diffusivity, these measures are relatively non-specific. Recent advances in diffusion MRI technology allow more accurate quantification of tissue microstructure, in particular for neurite morphology. For example, Neurite Orientation Dispersion and Density Imaging (NODDI) is a technique that can better capture the microstructural complexity of axons and dendrites (Zhang et al., 2012). It has been suggested that this technique might be more sensitive to cortical and subcortical changes in Parkinson's disease than traditional voxel-based morphometry or surface-based cortical thickness estimations (Kamagata et al., 2017), but it has not been specifically used to study patients with Parkinson's dementia. More recently, a bi-tensor model has been applied to MRI diffusion data that separates the diffusion properties of water within brain tissue from water in extracellular space. In this way, free water within brain structures can be estimated. This technique may detect higher levels of free water in the posterior substantia nigra for patients with worse cognitive scores (Planetta et al., 2016) and higher levels of free water predicted change in cognitive score after one-year follow-up (Ofori et al., 2015).

processing....