Symptom network connectivity indices as predictors of relapse in major depressive disorder

Major Depressive Disorder (MDD) demonstrates heterogeneous symptom profiles and a high relapse risk. Understanding how symptom interactions relate to relapse in MDD may enhance maintenance strategies. We thus investigated how connectivity in depressive symptom networks relates to relapse in MDD. We analyzed longitudinal data from 87 patients with remitted MDD who were followed for an average of 12 months. Patient-level symptom networks were estimated using multilevel graphical vector autoregression applied to weekly self-ratings of the Quick Inventory of Depressive Symptomatology. We calculated patient-specific network connectivity indices, including symptom network density (SND), vertex cover (VC) size, and minimal dominating set (MDS) size. Cox proportional hazards models assessed associations between these indices and time to relapse, controlling for baseline symptom severity. Higher SND and larger VC size were significantly associated with an increased relapse risk (HR for SND = 2.03, 95 % CI [1.44, 2.87], p < 0.001; HR for VC = 2.77, 95 % CI [1.79, 4.30], p < 0.001). Conversely, a larger MDS size was associated with a lower risk of relapse (HR 0.47, 95 % CI [0.31, 0.70], p < 0.001). Exploratory analyses showed that the strength centrality of sadness, difficulty concentrating, pessimism, suicidality, and low interest portend a higher relapse risk. Hyperconnectivity among depressive symptom networks may indicate vulnerability to relapse in MDD. Our results further support the potential utility of symptom network-based analyses for predicting outcomes in MDD. Future studies should evaluate whether symptom network-based analyses can provide markers for personalized treatment planning.