BackgroundCocaine use disorder (CUD) is a chronic, relapsing condition with substantial clinical burden, yet its neurobiological basis remains incompletely understood, particularly in the socially vulnerable populations most affected in real-world clinical settings. Existing neuroimaging evidence is fragmented across imaging modalities and derived largely from samples that underrepresent sustained adversity and polysubstance use. We used multimodal MRI to characterize structural and functional brain alterations in a clinically representative CUD sample.MethodsSeventy-two individuals with chronic cocaine use and 53 healthy controls underwent structural MRI and resting-state fMRI on a 3T scanner. Surface-based morphometry included cortical thickness, surface-based volume, and local gyrification index (a measure of the complexity of cortical folding), together with subfield-level volumetry of the hippocampus and amygdala. Functional connectivity was assessed using edge-wise general linear models. Analyses were adjusted for age, sex, and relevant covariates, with correction for multiple comparisons using cluster-wise and false discovery rate procedures.ResultsCUD was associated with widespread cortical abnormalities, marked by reduced thickness and surface-based volume predominantly in prefrontal, temporal, and parietal cortices, with the strongest effects in the lateral orbitofrontal cortex, superior frontal gyrus, and inferior parietal lobule. Local gyrification index alterations were more focal and confined to frontal regions. No total hippocampal or amygdalar volume differences survived false discovery rate correction, although exploratory subfield analyses suggested localized increases in the hippocampal head, hippocampus-amygdala transition area, and corticoamygdaloid transition region. Functional connectome analysis identified four significant inter-network alterations: one hypoconnectivity between the left somatomotor network and dorsolateral prefrontal cortex and three hyperconnectivities between right somatomotor regions and medial frontal salience/ventral attention nodes.ConclusionsCUD is characterized by a predominantly prefrontal-centered cortical structural phenotype accompanied by selective frontal network dysconnectivity. The functional findings suggest a reweighting of large-scale networks favoring salience-driven action over prefrontal regulatory control, consistent with incentive-sensitization models of addiction. Observed in a treatment-seeking sample representative of real-world clinical populations, these results support frontal systems as a core locus of vulnerability in CUD and point to candidate targets for circuit-based intervention.