Trapped in time: a collapse of temporal irreversibility and remodeling of causal connectivity in major depressive disorder

Healthy brain dynamics evolve in the direction far from thermodynamic equilibrium. The temporal asymmetry nature recently has been described in several normal and pathological brain states. However, whether the temporal asymmetry of spontaneous brain activity is affected by Major Depression Disorders (MDD) remains unknown. Resting-state fMRI data from 178 MDD patients and 205 healthy controls (HCs) were analyzed. We quantified temporal irreversibility (IR) by comparing forward and backward time series through time-shifted correlations using a sliding-windowed approach, constructed IR-based functional hierarchy, identified IR substates through k-means clustering, and estimated effective connectivity among key IR impaired regions using spectral Dynamical Causal Modeling (DCM). Associations with Hamilton Depression Rating Scale (HAMD) scores were further examined. MDD patients exhibited globally reduced IR, particularly within the default mode (DMN), attention (ATN), and sensorimotor (SM) networks, and across eight key regions. Functional hierarchy were significantly flattened in these systems. Clustering revealed two IR substates, characterized by globally high IR (State1) or low IR (State2). MDD patients occupied State2 more dominantly, exhibited longer dwell time, and transitioned into it more frequently. DCM revealed strengthened excitatory drive within the DMN and altered cross-network influences in MDD. DCM for subgroups further demonstrated progressive remodeling of causal connections across different severity levels. We demonstrate a pervasive shift for MDD toward equilibrium dynamics and network-specific causal pathologies. Our study provides a multi-faceted characterization of MDD-related brain dynamic alterations.