Beyond depression symptoms: the default mode network as a predictor of antidepressant response

Abstract Antidepressant efficacy for major depressive disorder (MDD) remains limited, with the neural mechanisms underlying treatment response poorly understood. The default mode network (DMN), particularly the connectivity between the medial prefrontal cortex (mPFC) and posterior cingulate cortex (PCC), has been implicated in MDD pathophysiology and may be linked to treatment outcomes. However, its potential as a biomarker for antidepressant response has not been validated. Here, we investigate the relationship between DMN connectivity and antidepressant treatment response in MDD. Resting-state fMRI data from four large MDD cohorts (n = 4271) were analyzed using Granger causality to examine directional effective connectivity (EC) within the DMN. Linear mixed-effects models compared EC between recurrent MDD patients, first-episode drug-naïve patients, and healthy controls. We also examined associations between EC, medication use, illness duration, depressive symptoms, and treatment outcomes. Additionally, Support Vector Machine (SVM) classifiers and support vector regression (SVR) were trained using EC from mPFC to PCC to predict treatment response. Our results revealed that recurrent MDD patients exhibited significantly reduced EC from mPFC to PCC compared to healthy controls and first-episode patients, with this reduction correlating with antidepressant medication use and illness duration. Importantly, DMN connectivity was associated with treatment improvement rather than core depressive symptoms, including suicide, anhedonia, or emotional blunting. Crucially, EC from mPFC to PCC predicted antidepressant treatment response, and SVM classifiers demonstrated high predictive accuracy for therapeutic outcomes. In conclusion, reduced EC from mPFC to PCC may serve as a biomarker for antidepressant treatment response in MDD, offering insights into MDD neurobiology and supporting the clinical potential of DMN connectivity measures for guiding treatment decisions. The SAINT, XijingQG, and XijingKG datasets were approved by the Ethics Committee of the First Affiliated Hospital, Fourth Military Medical University (approval numbers: KY20202066-F-1, XJLL-KY20222111, and KY20222165-F-1, respectively) and registered with clinicaltrials. gov (identifiers: NCT 04653337, NCT 05577481, and NCT 05544071, respectively).