Sulcal depth-based individualized structural covariance networks reveal multi-level brain reorganization associated with cognitive performance in COPD

Abstract

Background Chronic obstructive pulmonary disease (COPD) is frequently complicated by cognitive impairment, yet the underlying large-scale structural network alterations remain poorly understood. This study aimed to investigate individualized structural covariance networks (ISCNs) based on sulcal depth in COPD patients and to explore their relationship with cognitive performance. Methods Seventy-two patients with stable COPD and 68 age- and sex-matched healthy controls underwent 3T T1-weighted MRI. ISCNs were constructed using Jensen–Shannon divergence-based similarity of sulcal depth across 148 cortical regions. Graph-theoretical metrics, network-based statistics, and partial least squares regression were employed to characterize network topology, identify altered connectivity, and examine network associations with cognition assessed via Montreal Cognitive Assessment (MoCA) scores. Results COPD patients exhibited region-specific, bidirectional sulcal depth changes. At the network level, morphological similarity was reduced between the cingulo-opercular network and both the ventral multimodal and frontoparietal networks, whereas default mode–auditory network similarity was increased. Graph analysis revealed significantly higher gamma and lambda in COPD patients, indicating excessive local modularity alongside impaired global integration. Nodal efficiency was widely reduced in bilateral insular and frontal regions, but paradoxically increased in the left superior circular sulcus of the insula and right anterior cingulate gyrus. Partial least squares regression identified 14 interregional connections that collectively explained 28.3% of the variance in MoCA scores, with both positively and negatively weighted connections. Conclusion COPD is associated with a multi-level reorganization of brain structural networks, spanning from local sulcal depth abnormalities to global topological imbalance and bidirectional connectivity changes. These bidirectional network shifts may be associated with the cognitive impairment in this population, offering novel network-based biomarkers for early detection and intervention.