Mania and Depression 'Mark' the Brain Differently: Two-Year MRI Follow-up in Bipolar Disorder

Repeated episodes of bipolar disorder leave dissimilar traces in the brain structure - and this is visible after just two years. German researchers from the FOR2107 consortium followed 124 participants (62 patients with bipolar disorder and 62 healthy controls) and showed that during repeated depressive episodes, the volume of gray matter increases in the right outer part of the cerebellum, whereas in patients without relapses over the same period it rather decreases. The longer the patient spent in mania in the past, the more pronounced the decrease in volume during the subsequent remission. The authors interpret this as a dynamic of "inflammation → restoration/production of glia → then excessive pruning of synapses" - and thus reconcile the disparate results of previous studies. The study was published in Neuropsychopharmacology.

Background

Bipolar disorder (BD) is a cyclical disease with recurrent episodes of mania/hypomania and depression. Over the past two decades, neuroimaging has repeatedly shown that the disease is associated with structural and functional changes in emotional regulation networks (prefrontal cortex, cingulate cortex, amygdala, thalamus). However, the picture for gray matter remains contradictory: in some studies, a decrease in volume is found, in others - an increase or "zero" differences. There are both methodological and biological reasons for this.

First, most early studies were cross-sectional: a single image at a “random” point in the trajectory. But the brain in bipolar disorder is dynamic: episodes and remissions are accompanied by wave-like changes, from inflammatory swelling and glial activation to subsequent synaptic pruning and rewiring. Second, the results “blur” heterogeneity: different phases (mania vs. depression), disease duration, episode burden, comorbid disorders, pharmacotherapy, and even scanner/protocol differences. So the same region in two different samples may look “bigger” or “smaller,” depending on when and in whom we measure.

A third important player is neuroinflammation. Low-level inflammatory markers (e.g. hsCRP) and microglial activity are associated with the phases and severity of bipolar disorder in some patients. This suggests a neuroprogression model: each episode leaves a cumulative trace (inflammation → glial response → network remodeling), and it is the frequency/polarity of episodes that may explain the divergent structural findings.

Against the background of interest in cortical-limbic circuits, the cerebellum is gradually returning. It is involved not only in motor skills, but also in cognitive-affective regulation (cerebellar-prefrontal loops). In mood studies, cerebellar changes were described fragmentarily, without taking into account the phase and history of episodes, which is why their role remained underestimated.

Hence the logic of current longitudinal studies: move away from a “slice-and-slice” to trajectories with repeat MRIs over 1-3 years; carefully separate patients by the presence of relapses and their polarity between visits; take into account the historical “burden” (how much time in the past was spent in mania/depression); add inflammatory markers and harmonize data between centers. This design allows us to see how exactly episodes “mark” the brain in time and why mania and depression can leave different structural imprints.

What did they do?

• We recruited 124 people aged 20-62 years (1:1 - bipolar patients and healthy controls), two 3 T MRI visits with an interval of ~2.2 years; voxel-based morphometry (SPM12/CAT12) and ComBat data harmonization were used.
• Patients with bipolar disorder were divided into groups with relapse (at least one episode, manic or depressive, was observed over a two-year interval) and without relapse, and were compared with controls.
• Additionally, we linked gray matter dynamics to the clinical history of episodes before the baseline point, as well as to the baseline hsCRP level as a marker of low-grade inflammation. We controlled for medications, comorbidities, hospitalizations, changes in symptom and functioning scales.

The main signal was localized in the right outer part of the cerebellum. The authors propose to consider the trajectory as U-shaped: during/soon after depression - an increase in gray matter (possible glial reaction and edema against the background of inflammation), in stable remission - a decrease (probably pruning of synapses and "restructuring" of the network), and past manic episodes "set" a steeper decline.

Key findings in figures and facts

• Depressive relapses ↔ GMV increase: more depressive episodes per interval - more gray matter increase in the right outer cerebellum.
• No relapses ↔ GMV decline: in patients who had been episode-free for two years, the volume in this region declined more if they had experienced mania for a longer period before inclusion (ρ = −0.59; p = 0.012).
• Inflammatory fingerprint: Higher hsCRP at baseline predicted greater GMV gain in the cerebellum in BD patients (β = 0.35; p = 0.043), consistent with the hypothesis of a role for neuroinflammation.
• Not an artifact of treatment and severity: associations were not explained by hospitalizations, medications/medication burden, symptom dynamics (HAM-D, YMRS), GAF, BMI, familial risk, and psychiatric comorbidities.
• The reason is more in the episodes than in the “start anatomy”: there were no cross-sectional differences in GMV between the groups at the start and finish – it is the trajectory between T1 and T2 that is important.

Why does this matter? The bipolar literature shows increases, decreases, and “null” gray matter findings, depending on the phase of the MRI scan. The current work shows that polarity (mania vs. depression), time since the last episode, and inflammatory background determine different directions of structural changes. This reinforces the concept of neuroprogression in bipolar disorder: episodes leave cumulative but potentially reversible imprints, especially in the cerebellum, a hub that is involved not only in motor function but also in emotion/cognition regulation.

What can this change in practice?

• Monitoring along the trajectory, not at a “slice”: in patients with frequent depressive relapses and/or high hsCRP, it makes sense to monitor the dynamics, and not just one-time MRI indicators.
• Anti-relapse strategy is a priority: preventing both depressive and manic episodes can curb the unfavorable gray matter "swing." (This is clinical logic that requires direct testing.)
• Inflammatory tracker: hsCRP may become a simple stratification marker of “structural reactivity” risk - a candidate for personalized surveillance.

How does this fit into the biology of the disorder?

• During depression, glial activation and “edematous” increase in volume are likely (energy and immune costs for maintaining the network).
• In remission, “restructuring” and increased synaptic pruning lead to a decrease in volume, especially if there was a history of long-term mania.
• This U-pattern helps explain why previous studies have produced conflicting results: it all depends on when along the curve the image is taken.

Restrictions

• Sample size is moderate (n=124), two centers; effect is local (right outer cerebellum), generalization to other regions and populations requires confirmation.
• These are associations, not causal evidence; although treatment and severity effects are carefully controlled, underlying factors cannot be completely ruled out.
• The functional consequences of cerebellar changes (how they affect symptoms/behavior) were not studied - a task for future waves of observation.

Where to go next

• Link anatomy to function: add cognitive and affective tests/neuroimaging paradigms that assess cerebellar-prefrontal circuits.
• Longer MRI series: three or more time points to capture the "hump" and "valley" of the U-trajectory in more detail.
• Inflammatory biology: expand the panel of markers (cytokines, blood transcriptomes), compare with the dynamics of volume and frequency of relapses.

Conclusion

In bipolar disorder, the brain "breathes" structurally differently after depression and mania: depressions are associated with a temporary increase in cerebellar gray matter (probably due to an inflammatory response), and long-term previous mania is associated with a decrease in it during remission; the key to interpretation is to look at the trajectory, not at a single snapshot.

Source: Thomas-Odenthal F. et al. Differential impact of manic versus depressive episode recurrence on longitudinal gray matter volume changes in bipolar disorder. Neuropsychopharmacology, 2025. https://doi.org/10.1038/s41386-025-02197-x