Brain dynamics as outcome predictors of non-suicidal self-injury during SSRI treatment in adolescent depression: a 14-month longitudinal study

Non-suicidal self-injury (NSSI) represents a major health concern among adolescents with depression, yet no pharmacotherapies are specifically approved for NSSI. Selective serotonin reuptake inhibitors (SSRIs), commonly prescribed for adolescent depression, show inconsistent efficacy for NSSI, highlighting the need for neural signatures to predict NSSI outcomes. This study conducted a large-scale analysis of dynamic functional connectivity in 452 adolescents (332 drug-naïve patients with first-episode depression, including 178 with and 154 without NSSI, and 120 healthy controls) from an open-label SSRI trial (ChiCTR2000039503), examining how brain dynamics relate to the frequency and treatment outcomes of NSSI. Clinical and neuroimaging data were collected at baseline, with follow-up clinical assessments at 8 weeks and 14 months during SSRI treatment. Dynamic brain states and variability in graph-theoretical metrics were derived from imaging data, and the reproducibility of temporal dynamic patterns was validated in an independent dataset ( N = 112). Adolescents with depression and NSSI exhibited prolonged engagement in a biologically reproducible, relatively low-modularized, and moderately connected brain state, as well as elevated variability in small-world network metrics. This state was characterized by altered connectivity in the default mode and frontoparietal networks. Notably, baseline fractional windows of this state predicted both short-term reduction in NSSI frequency at 8 weeks and NSSI cessation at 14 months following SSRI treatment. These findings were consistently replicated across multiple validation analyses. Our results identify dynamic brain network patterns linked to NSSI outcomes during SSRI treatment, suggesting that characterizing such dynamic properties may help identify individuals at risk of poor NSSI outcomes and inform early intervention in adolescent depression.