Depression: Ten Years of Follow-Up Reveal Diverging Brain Trajectories

In a study published in September 2026, Kraus and colleagues investigated whether longitudinal changes in grey matter volume (GMV) in major depressive disorder (MDD) are better explained by the diagnosis itself, by acute symptom severity, or by cumulative illness burden over time. The study addresses an important limitation of the neuroimaging literature: although cross-sectional studies have repeatedly identified structural brain differences in MDD, comparatively few studies have followed the same individuals over several years. Consequently, it remains uncertain whether structural abnormalities represent progressive, “scar-like” changes associated with recurrent illness or more dynamic, potentially reversible changes linked to current depressive state. Tracking brain structure over a decade This single-centre case-control study conducted used data from the Münster Neuroimaging Cohort conducted in Germany. Participants underwent between three and six MRI assessments at approximately 2-year intervals, with follow-up extending to 10 years. The primary analysis included 206 individuals: 57 patients with MDD and 149 healthy controls. At recruitment, patients had been receiving inpatient treatment for an acute depressive episode. The mean participant age was 38.6 years, and the average number of scans per person was 3.8. Looking beyond symptoms at a single point in time A key methodological feature was the authors’ use of a composite measure of cumulative illness severity (CIS). This was derived from four variables: time since illness onset, cumulative number of depressive episodes, cumulative duration of depressive illness, and cumulative number of psychiatric hospitalisations. Acute depressive symptoms were assessed using the Hamilton Depression Rating Scale (HDRS). By separating cumulative burden from within-person fluctuations in current symptoms, the investigators aimed to distinguish progressive structural changes from state-dependent effects. Structural MRI data were analysed using longitudinal voxel-based morphometry and linear mixed-effects models. The principal regions of interest were the hippocampus, insula and dorsolateral prefrontal cortex (dlPFC), selected on the basis of previous longitudinal MDD studies. The investigators tested three main hypotheses: that patients with MDD would show greater GMV decline than controls; that higher cumulative illness severity would be associated with steeper GMV decline; and that acute depressive severity would show a contemporaneous, state-like association with GMV. Illness course matters more than diagnosis alone The principal finding was that long-term GMV trajectories were associated more strongly with cumulative illness course than with MDD diagnosis alone. When patients with MDD were compared with healthy controls as broad groups, there was little evidence that their GMV changed differently over time. Within the MDD group, however, a different pattern emerged. Patients with a greater cumulative burden of illness—reflecting factors such as longer illness duration, more depressive episodes, more time spent depressed, and more hospitalisations—showed greater loss of grey matter over time in two brain regions: the hippocampus and the dlPFC. By contrast, patients with a more favourable illness course tended to show relatively stable GMV, with some showing a small increase over time. Current depressive symptom severity, measured using the Hamilton Depression Rating Scale, was not consistently linked to GMV. Although changes in symptom severity were associated with hippocampal changes in one analysis, the pattern was irregular over time and did not hold up consistently when the researchers tested it in different ways or in the separate replication sample.   Clinical implications Clinically, the findings suggest that neurobiological change in MDD may be trajectory-dependent rather than an inevitable consequence of diagnosis. Greater accumulated illness burden was associated with more pronounced structural decline in fronto-limbic regions, whereas more favourable longitudinal courses were associated with relative preservation. This supports an emphasis on sustained relapse prevention and long-term illness management rather than interpreting MDD as producing uniform progressive brain change. Importantly, the authors do not establish causality, and MRI findings are not proposed as individual-level clinical biomarkers.   Limitations Limitations include the modest MDD sample, attrition over time, very small spatial extent of some significant clusters, sensitivity to modelling decisions, single-site design, ethnically homogeneous sample and potential residual confounding from treatments or environmental factors. Replication in larger, harmonised multisite cohorts will therefore be essential before these observations can inform precision psychiatry or routine clinical decision-making.