Abstract
OBJECTIVE: To investigate the spatial distribution characteristics of alterations in spontaneous brain activity in severe traumatic brain injury (sTBI) patients with disorders of consciousness (DOC), based on the mesocircuit theoretical framework, and to establish models for predicting recovery of consciousness. METHODS: Resting-state functional magnetic resonance imaging was employed to measure the mean fractional amplitude of low-frequency fluctuations (mfALFF) in sTBI patients with DOC and healthy controls, identifying differential brain regions for conducting gene and functional decoding analyses. Patients were classified into wake and DOC groups according to Extended Glasgow Outcome Score at 6 months. Furthermore, predictive models for consciousness recovery were developed using Nomogram and Linear Support Vector Machine (LSVM) based on mfALFF. RESULTS: In total, 28 sTBI patients with DOC and 30 healthy controls were included, with no significant baseline differences between groups (P > 0.05). The results revealed increased mfALFF of subcortical Ascending Reticular Activating System and decreased cortical mfALFF (default mode network) in DOC patients within the framework of the mesocircuit model (FDR_P < 0.001, Clusters > 100). The study identified 2080 differentially expressed genes associated with reduced brain activity regions, indicating mechanisms involving synaptic function, the oxytocin signaling pathway, and GABAergic processes in DOC formation. In addition, significantly higher mfALFF values were observed in the left angular gyrus, supramarginal gyrus, and inferior parietal lobule of DOC group compared to the wake group (AlphaSim_P < 0.01, Cluster > 19). The Nomogram prediction model highlighted the pivotal role of these regions' activity levels in prognosis (AUC = 0.90). Validation using LSVM demonstrated robust predictive performance with an AUC of 0.90 and positive predictive values of 80% for wake and 83% for DOC. CONCLUSIONS: This study offered crucial insights underlying DOC in sTBI patients, demonstrating the dissociation between cortical and subcortical brain activities. The findings supported the use of mfALFF as a robust and non-invasive biomarker for evaluating brain function and predicting recovery outcomes.