Altered static and dynamic effective connectivity in patients with type 2 diabetes mellitus with and without microvascular complications.

objectiveTo address critical knowledge gaps in static connectivity studies by characterizing how directional, dynamic effective connectivity reconfigures across brain states in T2DM with and without microvascular complications from a causal perspective, and how these network reorganization patterns relate to cognitive dysfunction and clinical measures.

methodsThe study included 112 T2DM patients without microvascular complications (T2DM-MA -), 141 T2DM patients with microvascular complications (T2DM-MA +), and 154 age-matched healthy controls (HC), collecting resting-state functional MRI data. Granger causality analysis was employed to evaluate static and dynamic effective connectivity among the three groups, combined with k-means clustering to identify dynamic connectivity states, and network topological efficiency was analyzed using graph theory. Furthermore, connectivity features were correlated with clinical indicators and neuropsychological test results.

resultsUsing k-means clustering analysis, we precisely identified three recurring connectivity states. Compared with HC, T2DM patients show a dynamic evolution from compensatory high connectivity to functional degeneration in low-connectivity states. Dynamic functional connectivity analysis revealed significant and extensive connectivity abnormalities in the low-connectivity state for the T2DM-MA + group, and captured more subtle connectivity differences between the T2DM-MA + and T2DM-MA - groups in mid-to-low connectivity states, demonstrating the high sensitivity of dynamic analysis methods. When cognitive function declines, functional connectivity in the strong connectivity state increases, while that in medium and low connectivity states falls. Furthermore, correlation analysis shows that time features in medium connectivity states are negatively related to blood glucose levels. Conversely, in low-connectivity states, they show an inverse correlation pattern and are negatively linked with cognitive scores.

conclusionOur analysis of dynamic effective connectivity reveals that T2DM patients, particularly those with microvascular complications, undergo a state-dependent reconfiguration from compensatory high connectivity to inefficient low-connectivity states. The increased time spent in these inefficient states was associated with poorer glycemic control. Thus, this work provides a novel dynamic and directional framework for understanding the neural mechanisms of cognitive dysfunction in T2DM.