King's College London

Objective
A growing literature indicates that unipolar and bipolar depression are associated with alterations in grey matter volume. However, it is unclear to what degree these patterns of morphometric change reflect symptom dimensions. Here, we aimed to predict depressive symptoms and hypomanic symptoms based on patterns of grey matter volume using machine learning.
Method
We used machine learning methods combined with voxel-based morphometry to predict depressive and self-reported hypomanic symptoms from grey matter volume in a sample of 47 individuals with un-medicated unipolar and bipolar depression.
Results

We were able to predict depressive severity from grey matter volume in the antero-ventral bilateral insula in both unipolar and bipolar depression. Self-reported hypomanic symptoms did not predict grey matter loss with a significant degree of accuracy.
Discussion
The results of this study suggest that patterns of grey matter volume alteration in the insula are associated with depressive symptom severity across unipolar and bipolar depression. Studies using other modalities, and exploring other brain regions with a larger sample are warranted to identify other systems that may be associated with depressive and hypomanic symptoms across affective disorders