Assessing White Matter Pathology in Early-Stage Parkinson Disease Using Diffusion MRI: A Systematic Review

Abstract

Structural brain white matter (WM) changes such as axonal caliber, density, myelination, and orientation, along with WM-dependent structural connectivity, may be impacted early in Parkinson disease (PD). Diffusion magnetic resonance imaging (dMRI) has been used extensively to understand such pathological WM changes, and the focus of this systematic review is to understand both the methods utilized and their corresponding results in the context of early-stage PD. Diffusion tensor imaging (DTI) is the most commonly utilized method to probe WM pathological changes. Previous studies have suggested that DTI metrics are sensitive in capturing early disease-associated WM changes in preclinical symptomatic regions such as olfactory regions and the substantia nigra, which is considered to be a hallmark of PD pathology and progression. Postprocessing analytic approaches include region of interest-based analysis, voxel-based analysis, skeletonized approaches, and connectome analysis, each with unique advantages and challenges. While DTI has been used extensively to study WM disorganization in early-stage PD, it has several limitations, including an inability to resolve multiple fiber orientations within each voxel and sensitivity to partial volume effects. Given the subtle changes associated with early-stage PD, these limitations result in inaccuracies that severely impact the reliability of DTI-based metrics as potential biomarkers. To overcome these limitations, advanced dMRI acquisition and analysis methods have been employed, including diffusion kurtosis imaging and q-space diffeomorphic reconstruction. The combination of improved acquisition and analysis in DTI may yield novel and accurate information related to WM-associated changes in early-stage PD. In the current article, we present a systematic and critical review of dMRI studies in early-stage PD, with a focus on recent advances in DTI methodology. Yielding novel metrics, these advanced methods have been shown to detect diffuse WM changes in early-stage PD. These findings support the notion of early axonal damage in PD and suggest that WM pathology may go unrecognized until symptoms appear. Finally, the advantages and disadvantages of different dMRI techniques, analysis methods, and software employed are discussed in the context of PD-related pathology.

Keywords: MRI diffusion; diffusion kurtosis imaging; diffusion tensor imaging; early-stage Parkinson disease; fractional anisotropy; q-space diffeomorphic reconstruction; substantia nigra.

Figure 1

Search strategy based on PRISMA flow diagram.

Figure 2

Histograms for the publication years and dMRI techniques included in this review.

Figure 3

Standard DTI maps, from a healthy volunteer, created by DSI Studio with a DTI diffusion scheme with a total of 67 diffusion sampling directions. The b value was 1,000 s/mm². The in-plane resolution was 2 mm, and the slice thickness was 2 mm. FA, fractional anisotropy; MD, mean diffusivity; AxD, axial diffusivity; RD, radial diffusivity. The range of FA in between 0 (isotropic diffusion) and 1 (anisotropic diffusion).

Figure 4

(A) Fractional anisotropy maps and (B) an example of whole-brain deterministic tractography created by DSI Studio in a healthy volunteer (angular threshold: 60 degrees; step size: 1 mm; anisotropy threshold: 0.20. Tracks with length shorter than 50 or longer than 300 mm were discarded. A total of 500,000 seeds were placed inside the whole brain).

Figure 5

Example of the different dMRI analysis methods. (A) Histogram analysis; (B) ROI analysis; (C) voxel-based analysis (VBA); (D) skeletonized analysis (e.g., TBSS pipeline); (E) connectome analysis.

Figure 6

(A) Q-space diffeomorphic reconstruction using DSI Studio. gFA, QA, ISO (represents background isotropic diffusion contributed from CSF), and the ODF visualization are shown. (B) The performance of the FA-aided, GFA-aided, anatomy-aided, and QA-aided tractographies showing arcuate fasciculus using shell and grid sampling data. The false tracks are colored in black, whereas accurate trajectories are coded by directional color. The tractographies using FA, GFA, and anatomical information show substantially more false tracks than do those using QA. The best performance can be observed in the tractography using QA and the grid dataset. Reproduced with permission from Yeh et al. (128).