Degeneration of structural brain networks is associated with cognitive decline after ischaemic stroke

Michele Veldsman, Hsiao-Ju Cheng, Fang Ji, Emilio Werden, Mohamed Salah Khlif, Kwun Kei Ng, Joseph Lim, Xing Qian, Haoyong Yu, Juan Zhou, Amy BrodtmannResearch Square 2020 · DOI

Introduction

Approximately one-third of ischemic stroke survivors develop cognitive impairment. Traditionally, clinicians have looked to the size and location of the initial brain lesion (the infarct) to predict recovery. However, these factors are often poor predictors of long-term cognitive outcomes. This discrepancy suggests that stroke is not merely a localized injury but a disruption to a complex, distributed system.

Recent research has shifted focus toward "secondary neurodegeneration"—the gradual wasting away of brain tissue in regions remote from, but connected to, the initial injury. This study investigates whether the topographical degeneration within distributed Structural Covariance Networks (SCNs)—patterns of gray matter volume that change in tandem across the brain—underlies cognitive decline in the year following an ischemic stroke.

Methodology

The researchers conducted a longitudinal study involving 73 ischemic stroke patients (Discovery Dataset) and an independent validation cohort of 22 patients. Participants underwent 3T MRI scans and a comprehensive battery of 17 neuropsychological tests at two critical timepoints: subacute (3 months post-stroke) and chronic (1 year post-stroke).

Overview of the study workflow, including network derivation via Seed PLS and behavioral linking via Behavioral PLS.

To map the brain's structural architecture, the team used Voxel-Based Morphometry (VBM) and a technique called Seed Partial Least Squares (PLS). They derived six canonical higher-order cognitive networks—Dorsal Attention, Executive Control, Salience, Default Mode, Language, and Memory—based on established functional seeds. They then employed Behavioral PLS to identify multivariate relationships between the integrity of these structural networks and cognitive performance across five domains: attention, executive function, visuospatial ability, language, and memory.

Key Results

The "Mirror" Effect and Subacute Impairment

The study found that the structural networks in stroke patients closely mirror the functional networks observed in healthy populations. At the 3-month subacute mark, lower network integrity was already significantly associated with poorer cognitive performance across all tested domains ($r = 0.194, p < 0.001$). This suggests that the initial stroke triggers a rapid reorganization or degradation of these large-scale networks.

Trajectory of Decline

The most critical finding was that the rate of network degradation predicted the severity of cognitive decline. Patients whose SCNs degraded more rapidly between the 3-month and 1-year marks showed significantly steeper declines in attention, memory, and language. The longitudinal model explained 64.91% of the variance in the relationship between structural brain changes and cognitive shifts ($r = 0.287, p < 0.001$).

Domain-Specific Vulnerabilities

While all networks showed some degree of change, attention-related tasks (such as the Choice Reaction Time task) showed the strongest correlation with network integrity. This suggests the attention network may be a "canary in the coal mine" for distributed structural damage. Other specific deficits linked to faster SCN degradation included declines in verbal fluency (COWAT-FAS) and delayed memory recall (RCF-delay).

Discussion

These findings confirm that stroke triggers a "cascade" of neurodegeneration that extends far beyond the initial lesion. This process, often referred to as structural "rusting," may be driven by Wallerian degeneration—where neurons die because their axonal connections are severed—or by the chronic underutilization of disconnected brain regions.

Crucially, these results held true even after controlling for age, sex, and the volume of the initial stroke lesion. This reinforces the idea that the health of the global network is more predictive of long-term cognitive health than the focal damage itself.

Limitations and Future Directions

The study focused on patients with relatively mild strokes (low NIHSS scores), so it remains to be seen if these patterns hold for more severe cases. Additionally, while the study identifies that networks degenerate, the exact biological mechanisms—such as neuroinflammation or vascular burden—require further investigation. Future research should extend the observation window to five years to determine if this degradation eventually plateaus or serves as a precursor to post-stroke dementia.

Summary updated

Abstract

<title>Abstract</title> One third of ischemic stroke patients develop cognitive impairment. It is not known whether topographical secondary neurodegeneration within distributed brain structural covariance networks (SCNs) underlies this cognitive decline. We examined longitudinal changes in SCNs and their relationship to domain-specific cognitive decline in 73 ischemic stroke patients. Patients were scanned with magnetic resonance imaging (MRI) and assessed on five cognitive domains at subacute (3-months) and chronic (1-year) timepoints. Individual-level SCN scores of major cognitive networks were derived from MRI data at each timepoint. We found that distributed degeneration in higher-order cognitive networks was associated with cognitive impairment in subacute stroke. Importantly, faster degradation in these major cognitive SCNs over time was associated with greater decline in attention, memory, and language domains. Our findings suggest that subacute ischemic stroke is associated with degeneration of higher-order structural brain networks and degradation of these networks contribute to individual trajectories of longitudinal domain-specific cognitive dysfunction.
BibTeX
@inproceedings{veldsman2020degeneration,
  author = {Michele Veldsman and Hsiao-Ju Cheng and Fang Ji and Emilio Werden and Mohamed Salah Khlif and Kwun Kei Ng and Joseph Lim and Xing Qian and Haoyong Yu and Juan Zhou and Amy Brodtmann},
  title = {{Degeneration of structural brain networks is associated with cognitive decline after ischaemic stroke}},
  booktitle = {Research Square},
  year = {2020},
  doi = {10.21203/rs.3.rs-36274/v1},
}
Figure