Local-Global breakdown of Cortical Similarity Networks in Anorexia Nervosa

Anorexia nervosa (AN) profoundly alters brain structure, yet the large-scale organization of the cortex under severe nutritional stress remains poorly understood. In this study, we map cortical architecture in 50 patients with AN and 40 healthy controls using Morphometric INverse Divergence (MIND), a multivariate framework integrating multiple morphometric features into cortical similarity networks. Patients showed a marked global reduction in morphometric similarity that scaled with nutritional status, with lower body-mass index associated with weaker global and network-level connectivity. Disruptions selectively involved transmodal systems including limbic, default mode, frontoparietal and dorsal attention, whereas primary sensorimotor and visual cortices were relatively preserved. Region-level effects followed the topology of the structural connectome, and network-based statistics revealed weakened long-range coupling between attentional and limbic-prefrontal circuits. Spatial correspondence with cortical gradients of serotonergic, dopaminergic, mu-opioid and CB1 receptor densities suggests that malnutrition compromises cortical integrity along neurochemically defined axes of vulnerability.