Background Early‐onset (EO) major depressive disorder (MDD) patients experience more depressive episodes and an increased risk of relapse. Thus, on a neurobiological level, adult EO patients might display brain structure and function different from adult‐onset (AO) patients. Methods A total of 103 patients (66 females) underwent magnetic resonance imaging. Structural measures of gray matter volume (GMV) and functional connectivity networks during resting state were compared between EO (≤19 years) and AO groups. Four residual major depression symptoms, mood, anxiety, insomnia, and somatic symptoms, were correlated with GMV between groups. Results We found comparatively increased GMV in the EO group, namely the medial prefrontal and insular cortex, as well as the anterior hippocampus. Functional networks in EO patients showed a comparatively weaker synchronization of the left hippocampus with the adjacent amygdala, and a stronger integration with nodes in the contralateral prefrontal cortex and supramarginal gyrus. Volumetric analysis of depression symptoms associated the caudate nuclei with symptoms of insomnia, and persisting mood symptoms with the right amygdala, while finding no significant clusters for somatic and anxiety symptoms. Conclusions The study highlights the important role of the hippocampus and the prefrontal cortex in EO patients as part of emotion‐regulation networks. Results in EO patients demonstrated subcortical volume changes irrespective of sleep and mood symptom recovery, which substantiates adolescence as a pivotal developmental phase for MDD. Longitudinal studies are needed to differentiate neural recovery trajectories while accounting for age of onset.

Background Serious long term health and economic detriment accompany residual depressive symptoms even in fully remitted depressed patients (rMDD). Neurobiological predictors for rMDD patients’ illness trajectory are absent. Methods rMDD patients (n = 39, female = 26) underwent magnetic resonance imaging. Baseline analyses of brain structure via voxel‐based morphometry and brain function via functional connectivity (FC) at rest were correlated with changes in the Hamilton Depression Rating Scale between baseline and follow‐up, and incidence of a recurrent major depressive episode (MDE) within a 2‐year period. Results Gray matter increases in default mode (DN) regions in the posterior cingulate cortex (PCC) and increased resting‐state FC within the DN both predicted change of depressive symptoms. Patients with recurrent MDE had larger bilateral nucleus accumbens and left insula volumes. Post hoc exploratory analysis of nucleus accumbens and insula conceptualized as part of the brain's reward circuit demonstrated reduced connectivity in patients with recurrent MDE. Conclusions Higher DN connectivity and PCC volume coinciding with a more favorable course of symptoms suggest neural mechanisms of self‐recovery beyond the phase of active medical treatment. Alterations in the brain's reward circuit might be a starting point for designing maintenance treatments that prevent recurrent MDEs in rMDD patients.