Rumination persists in remitted depression (RD) and increases relapse risk. Mindfulness-based strategies may interrupt rumination, but network reorganization during the transition from rumination to mindfulness is poorly understood.

Fifty-one patients with RD and 41 healthy controls (HCs) were recruited and completed a six-state fMRI paradigm (resting state I, negative autobiographical recall, rumination, mindfulness, distraction, and resting state II). We used network-based statistics (NBS) within each state and graph-theoretical analyses across sparsity thresholds. Rumination severity was assessed with the Ruminative Responses Scale (RRS). An exploratory analysis tested rumination to the mindfulness similarity of the left inferior parietal lobule (IPL) connectivity profile.

During mindfulness, RD showed a hyperconnected NBS component linking default mode, dorsal and ventral attention, and frontoparietal and somatomotor networks, anchored in a left IPL hub. A more widespread hyperconnected component was observed during resting state II in RD. Graph theoretical analyses showed that, during mindfulness, RD exhibited reduced degree centrality, betweenness centrality, and nodal efficiency in the left IPL (all FDR-corrected p < 0.05). IPL connectivity profile similarity from rumination to mindfulness was significantly lower in RD compared with HCs. Within RD, IPL degree and nodal efficiency were negatively associated with RRS.

RD is characterized by an IPL-centered hub configuration during mindfulness following rumination, marked by broad coupling but reduced hub efficiency and reduced rumination-to-mindfulness reconfiguration. Clinically, inefficient hub reconfiguration during mindfulness attempts may index a residual vulnerability and may represent a candidate target for monitoring and intervention.