# The alterations in brain network functional gradients and dynamic functional connectivity in Alzheimer’s disease: a resting-state fMRI study

## TL;DR

This study explores brain network changes in Alzheimer's disease using resting-state fMRI, revealing altered functional gradients and dynamic connectivity that could serve as imaging biomarkers.

## Contribution

The study introduces a novel combined analysis of functional gradients and dynamic functional connectivity to uncover mechanistic imaging biomarkers for Alzheimer's disease.

## Key findings

- AD patients show widespread functional gradient alterations between the Default Mode Network and Sensorimotor Network.
- Dynamic functional connectivity reveals four recurrent states with altered flexibility and connectivity patterns in AD.
- The FG-dFC analysis identifies imaging features linked to T-Tau levels and age, offering insights into AD neuropathology.

## Abstract

Alzheimer’s disease (AD), the most common form of dementia worldwide, is characterized by progressive cognitive decline. Extensive evidence from dynamic functional connectivity (dFC) studies has demonstrated unstable functional states, reduced network flexibility, and impaired transitions between large-scale neurocognitive networks across the AD continuum. However, how these temporal abnormalities are embedded within the hierarchical spatial organization of brain networks, as captured by functional gradients (FG), and whether combined FG-dFC metrics can provide mechanistically interpretable and potentially sensitive imaging biomarkers, remain to be elucidated.

This study enrolled 46 AD patients who were diagnosed according to the Amyloid/Tau/Neurodegeneration (ATN) biological diagnostic framework and 37 age- and sex-matched healthy controls (HC). All participants underwent resting-state fMRI. Functional gradients were derived using connectivity similarity matrices and diffusion embedding (aligned and standardized), while dFC was estimated with a sliding window approach and clustered into four recurrent states. Group differences were assessed with two-sample t-tests with Gaussian Random Field (GRF) correction. Correlation analyses included ATN biomarkers and cognitive scores. A linear support vector machine (SVM) with leave-one-out cross-validation evaluated classification performance based on significant FG features.

Compared to the healthy controls, AD patients exhibited widespread FG alterations between regions of the Default Mode Network (DMN) and the Sensorimotor Network (SMN). In the first gradient DMN, the left precuneus showed reduced gradient scores, whereas the right medial superior frontal gyrus and bilateral angular gyri were increased. In the first gradient of the SMN, the right supplementary motor area increased while bilateral superior temporal gyri decreased. Second-gradient reductions were confined to two regions: the left postcentral gyrus (SMN) and left middle occipital gyrus (visual network, VIS). The right medial superior frontal gyrus first-gradient score correlated negatively with T-Tau (r = −0.50, P = 0.006) and age (r = −0.36, P = 0.02); the right angular gyrus correlated negatively with age (r = −0.29, P = 0.04); the left precuneus correlated positively with age (r = 0.38, P = 0.009). dFC revealed four recurrent states (27.59, 17.67, 28.27, 26.47% of total occurrences). Relative to HC, AD showed higher FT and MDT in states 1–2 and lower scores in state 3, with NT unchanged, alongside state-dependent bidirectional connectivity changes (fronto-insular-sensorimotor increases; DMN-temporal and visuo-auditory decreases). The SVM achieved an AUC of 0.776, sensitivity 78.26%, specificity 67.57%, and accuracy 73.49%, with the right superior temporal gyrus within SMN first-gradient contributing most.

AD is characterized by macro-scale hierarchical disorganization centered on the principal functional gradient, accompanied by reduced cross-state flexibility and state-dependent connectivity abnormalities. The combined functional gradient-dynamic functional connectivity (FG-dFC) analysis provides complementary spatiotemporal insights and reveals imaging features associated with T-Tau levels and age, offering new perspectives on the neuropathological mechanisms of AD and potential imaging biomarkers. Moreover, these network topology and dynamic connectivity metrics may prove useful for monitoring disease progression, evaluating treatment effects, and stratifying patients in future clinical and interventional studies.