Dynamic Functional Network Connectivity in Anorexia Nervosa: A Meta‑State Approach and its Implication with Cognitive, Emotional and Severity of the Disease.

Anorexia Nervosa (AN) is a psychiatric disorder associated with high rates of morbidity, mortality and relapse. Most previous neuroimaging studies used static functional connectivity approaches, which cannot show how brain connectivity changes over time. Dynamic Functional Network Connectivity (dFNC) offers a way to investigate these temporal changes. However, only a few dFNC studies exist on AN, and the meta-state approach, which describes brain dynamics in a more continuous manner, has not previously been applied to this disorder. This study aimed to investigate the temporal organization of resting-state functional network connectivity in patients with AN compared with healthy controls (HC), using both a state-based and a meta-state dFNC approach, and to explore possible associations with clinical, cognitive and eating disorder measures. The sample included 35 patients with AN and 39 HC. Preprocessing was performed with CONN and SPM12. dFNC was estimated using the Group ICA of fMRI Toolbox (GIFT), with the RSN28 template and a sliding window approach. Four recurrent connectivity states were identified using K-means clustering. Group comparisons were performed using Welch’s t-tests, with Benjamini-Hochberg FDR correction, and brain-behavior associations were tested with partial Spearman correlations controlling for age and years of education. Results showed that the AN and HC groups had similar connectivity states overall, but differed in the amount of time spend in them. AN participants showed significantly lower Fractional Occupancy in State 3 and significantly higher Fractional Occupancy in State 4 compared with HC, while no significant differences were found for Mean Dwell Time or Number of Transitions. In the meta-state analysis, AN participants showed a significantly lower Total Distance than HC,indicating that they explored a smaller portion of the connectivity space during the scan. This result was consistent across the four decomposition methods used (PCA, temporal ICA, spatial ICA, and K-means), while AN also showed fewer meta-states and meta-state changes in some of these analyses. Several exploratory associations were found between dFNC measures and executive functioning and body dissatisfaction, but none survived correction for multiple comparisons. These findings suggest that AN is characterized by a different temporal organization of large-scale brain connectivity, involving both the amount of time spent in specific connectivity configurations and the extent to which the brain moves through connectivity space, rather than the presence of completely different connectivity patterns. These alterations mainly involve networks related to visual processing and cognitive control, which is in line with previous literature on cognitive rigidity in AN. Given the small sample size and the cross sectional design, these results should be considered exploratory and requires replication in larger and longitudinal studies. Overall, this work supports the use of dFNC, particularly the meta-state approach, as a promising tool to better understand the dynamic brain alterations underlying AN.