Illness Severity and Longitudinal Gray Matter Volumes Among People With Major Depressive Disorder.

Importance: Although cross-sectional research in major depressive disorder (MDD) is extensive, long-term longitudinal neuroimaging studies are rare, despite being crucial to uncovering how brain changes evolve with illness progression. Objectives: To disentangle brain structural trajectories within the fronto-limbic circuitry in MDD and to distinguish progressive decline from dynamic state-like alterations. Design, Setting, and Participants: This single-center case-control study (Münster Neuroimaging Cohort) conducted repeated magnetic resonance imaging (MRI) assessments over up to 10 years among patients recruited from local hospitals and healthy controls recruited from the general population. Baseline recruitment was conducted between October 1, 2009, and January 31, 2018, with follow-up through March 31, 2022. Inpatients with initially acute MDD and volunteer controls underwent 3 to 6 MRI scans, spaced approximately every 2 years. Data were analyzed between October 2024 and September 2025. Main Outcomes and Measures: Gray matter volume (GMV) was examined, applying longitudinal voxel-based morphometry using region-of-interest analyses and exploratory whole-brain approaches. In mixed-effect models, associations of GMV trajectories with (1) diagnosis, (2) cumulative illness severity (CIS), and (3) acute illness severity were analyzed. Results: A total of 206 individuals (mean [SD] age, 38.6 [12.4] years; 112 men [54.4%]; 57 with MDD [27.67%] and 149 controls [72.33%]) who underwent a mean (SD) of 3.8 (0.8) scans (791 scans total) were included in the analysis. CIS × time interactions emerged in the hippocampus (peak ηp2 = 0.261 [95% CI, 0.124-0.351]; family-wise error [FWE]-corrected P < .001) and dlPFC (peak ηp2 = 0.216 [95% CI, 0.086-0.305]; FWE-corrected P = .02), with higher CIS associated with steeper GMV decline in contrast with more favorable clinical courses. These associations were mostly robust across extensive sensitivity analyses, including variations in data inclusion, and controlling for acute illness severity and medication use. There was little evidence for differential GMV trajectories between patients with MDD and controls per se, or for main associations of cumulative or acute illness severity in MDD. Conclusions and Relevance: In this 10-year case-control-study, GMV changes in MDD were associated with individual cumulative illness trajectories rather than depressive symptom fluctuations or diagnosis alone. Findings supported both progressive decline and relative stability, depending on the clinical course. These results underscore the need for personalized, time-sensitive neurobiological models in depression and emphasize the importance of long-term designs in mental health neuroimaging.