ABSTRACT Objective Patients with anti-NMDA receptor encephalitis suffer from a severe neuropsychiatric syndrome, yet most patients show no abnormalities in routine magnetic resonance imaging. In contrast, advanced neuroimaging studies have consistently identified disrupted functional connectivity in these patients, with recent work suggesting increased volatility of functional state dynamics. Here, we investigate these network dynamics through the spatiotemporal trajectory of meta-state transitions, yielding a time-resolved account of brain state exploration in anti-NMDA receptor encephalitis. Methods Resting-state functional magnetic resonance imaging data were acquired in 73 patients with NMDAR encephalitis and 73 age- and sex-matched healthy controls. Time-resolved functional connectivity was clustered into brain meta-states, giving rise to a time-resolved transition network graph with states as nodes and transitions between brain meta-states as weighted, directed edges. Network topology, robustness, and transition cost of these transition networks were compared between groups. Results Transition networks of patients showed significantly lower local efficiency ( t = -2.54, p FDR = 0.026), lower robustness ( t = -2.01, p FDR = 0.048) and higher leap size ( t = 2.33, p FDR = 0.026) compared to controls. Furthermore, the ratio of within-to-between module transitions and state similarity was significantly lower in patients. Importantly, alterations of brain state transitions correlated with disease severity. Interpretation These findings reveal systematic alterations of transition networks in patients, suggesting that anti-NMDA receptor encephalitis is characterized by reduced stability of brain state transitions and that this reduced resilience of transition networks plays a clinically relevant role in the manifestation of the disease.

Reduced resilience of brain state transitions in anti-N-Methyl-D-Aspartate receptor encephalitis