Structural network topology correlates with cognitive impairment in Parkinson’s disease

Zhichun Chen, Guanglu Li, Liche Zhou, Lina Zhang, Yang Li, Jun LiumedRxiv 2023 · DOI

TL;DR

This study demonstrates that cognitive impairment in Parkinson’s Disease (PD) is not just a side effect, but a core indicator of the brain's structural integrity. By analyzing white matter "wiring" through graph theory, researchers found that lower cognitive scores—specifically on the Symbol Digit Modalities Test (SDMT)—correlate with a breakdown in the brain's ability to efficiently transfer information across both global and local networks.

What Problem Does This Paper Address?

While Parkinson’s Disease is primarily known for motor symptoms like tremors and rigidity, cognitive decline is a frequent and debilitating non-motor feature that often predicts a faster disease progression. We know that PD involves the disruption of white matter (the brain's long-range "cables"), but the specific relationship between the topology—the mathematical organization—of these networks and cognitive health has remained fuzzy.

The authors sought to determine if the brain's structural network architecture could serve as a biomarker for cognitive impairment and whether specific "hubs" in the brain mediate the relationship between aging and cognitive decline in PD patients.

Methods

The researchers employed a combination of clinical assessment and advanced neuroimaging:

  • Grouping by Cognition: 146 PD patients were divided into quartiles based on their Symbol Digit Modalities Test (SDMT) scores. This test measures processing speed and attention, serving as the primary lens for the study.
  • Structural Connectomics: Using Diffusion Tensor Imaging (DTI) and deterministic fiber tractography, the team mapped the white matter connections between 90 different brain regions (based on the AAL atlas).
  • Graph Theory: They calculated "topological metrics" to describe the network. This includes Global Efficiency (how well the whole brain communicates), Local Efficiency (how well neighboring regions communicate), and Nodal Metrics (the importance of specific "hubs").
  • Mediation Analysis: A statistical technique used to see if the structural changes in the brain actually "carry" the effect of age onto cognitive performance.

Datasets

The study utilized data from the Parkinson’s Progression Markers Initiative (PPMI), a large-scale, international multicenter cohort study. This provided a standardized dataset of clinical evaluations, SPECT scans (to measure dopamine depletion), and high-resolution MRI/DTI scans.

Key Insights and Ideas

  • SDMT as a Structural Proxy: The SDMT isn't just a cognitive test; it's a powerful reflection of the brain's information transfer efficiency. Patients with the lowest SDMT scores showed significantly "leakier" or less efficient brain networks.
  • The Cognition-Motor Link: Cognitive decline isn't an isolated symptom. Patients in the lowest cognitive quartile also exhibited more severe motor symptoms and greater dopamine depletion in the striatum. This suggests that cognitive health is a barometer for overall disease severity.
  • Hub Vulnerability: The breakdown isn't uniform. Specific regions, such as the Right Middle Temporal Gyrus and the Right Anterior Cingulate Cortex, emerged as critical nodes. When these specific "hubs" lose their connectivity, cognitive scores drop.
  • Small-World Disruption: The PD brain typically maintains a "small-world" architecture (a balance of local clusters and long-range shortcuts). This study found that as cognition fails, the "path length" between regions increases, making the network less "small-world" and more difficult for signals to traverse.

Key Results

  • Global Network Decay: The lower quartile group (Q1) showed significantly lower global and local efficiency across multiple network density thresholds compared to those with better cognition (Q2-Q4).
  • Dose-Dependent Changes: Many nodal metrics showed a "dose-dependent" relationship with SDMT scores; as cognitive performance improved across quartiles, network efficiency metrics rose in tandem.
  • Mediation Findings: Mediation analysis revealed that the effect of age on cognitive scores (like the LNS and SDMT) was statistically "mediated" by the efficiency of specific nodes, particularly in the right anterior cingulate cortex.
  • Motor Correlation: Nodal efficiency in the right anterior cingulate cortex was negatively associated with UPDRS-III (motor) scores, suggesting this region is a bridge between cognitive and motor decline.

Limitations and Future Work

  • Cross-Sectional Design: The study captures a snapshot in time. While it shows strong correlations, it cannot definitively prove that network changes cause cognitive decline over time. Longitudinal studies are needed.
  • Control Group Data: While the study compared PD subgroups, a lack of consistent MRI data for healthy controls in the PPMI database limited the ability to compare these networks against a non-PD baseline in this specific analysis.
  • Fiber Tractography Limits: The use of deterministic tractography is standard but can struggle with "crossing fibers" in the brain; future work using more complex models (like HARDI) might provide even higher resolution.

Reference: Chen, Z., et al. (2023). Structural network topology correlates with cognitive impairment in Parkinson’s disease. medRxiv preprint. DOI: 10.1101/2023.09.18.23295719.

Summary updated

Abstract

ABSTRACT In patients with Parkinson’s disease (PD), cognitive impairment is common and substantially exacerbates disease severity. The objective of this study is to investigate whether cognitive decline affects disease severity and whether structural network disruption contributes to cognitive impairment in PD patients. PD patients (n = 146) receiving structural and functional magnetic resonance imaging from Parkinson’s Progression Markers Initiative (PPMI) database were included. Based on the scores of Symbol Digit Modalities Test (SDMT), patients were classified into lower quartile group (SDMT score rank: 0%∼25%), interquartile group (SDMT score rank: 26%∼75%), and upper quartile group (SDMT score rank: 76%∼100%) according to their SDMT score quartiles to assess how cognitive function shapes disease manifestations and brain network topology. Patients in the lower quartile group exhibited more severe motor symptoms, non-motor symptoms, and striatal dopamine depletion compared to interquartile group and upper quartile group. In addition, they also showed impaired structural network topology both at global and local levels compared to interquartile group and upper quartile group. Furthermore, different nodal network metrics in structural network contributed to cognitive and motor impairment of PD patients. Taken together, PD patients with worse cognitive function exhibited more severe motor and non-motor symptoms and aberrant structural network metrics were associated with both cognitive and motor decline of PD patients.
BibTeX
@misc{chen2023structural,
  author = {Zhichun Chen and Guanglu Li and Liche Zhou and Lina Zhang and Yang Li and Jun Liu},
  title = {{Structural network topology correlates with cognitive impairment in Parkinson’s disease}},
  year = {2023},
  doi = {10.1101/2023.09.18.23295719},
  howpublished = {medRxiv},
}
Figure