Weaker incoming connections to a key emotion-related brain region were associated with greater depressive symptoms in a study of 890 neurotypical young adults published in the International Journal of Neuropsychopharmacology. Patterns in these incoming connections sorted participants into groups that differed in depressive symptom severity, although the results do not establish a diagnostic test for depression.
The subgenual anterior cingulate cortex (sgACC) is a small region near the front of the brain involved in regulating emotion and bodily responses. Altered activity in this region and its connections has been linked to depression, and the area has been a target of brain stimulation treatments for depression. However, conventional brain-imaging measures usually show whether two regions are active together without indicating which region’s activity may influence the other.
To examine this distinction, the researchers used a statistical technique called Granger causality analysis. This method looks at the timing of fluctuations in brain signals to estimate whether activity in one region tends to precede and predict activity in another. Applied to brain scans that track blood flow, it provides an estimate of directional influence rather than proof that one region drives another.
Led by Huey-Ting Li of Yale University School of Medicine, the researchers analyzed resting-state functional MRI scans from 890 neurotypical young adults in the Human Connectome Project, a large study of healthy adults. The sample included 459 women and 431 men aged 22 to 35. Participants completed a questionnaire measuring depression, anxiety and other emotional and behavioral symptoms.
The researchers examined connections between the sgACC and 268 other brain regions, separating connections estimated to run toward the sgACC from those estimated to run away from it. They then used a clustering method to sort participants into groups with similar connectivity patterns and compared the groups’ symptoms. For comparison, they repeated the analysis using conventional measures of which regions were active together.
Many regions showed connections running toward the sgACC, including frontal and parietal regions (involved in planning and attention), the anterior insula (which helps process bodily and emotional information), temporal regions and the cerebellum. The orbitofrontal cortex, which contributes to decision-making and emotion, showed connections in both directions. The sgACC, in turn, appeared to send signals mainly to movement-related areas, the brainstem and the cerebellum.
Clustering based on the incoming connections divided participants into five groups, with depression scores rising from the first group to the fifth, although only the group with the lowest scores differed reliably from the three highest-scoring groups. Participants with weaker inputs from several frontal, parietal, temporal and occipital areas tended to fall into groups with higher depression scores, with the strongest links involving the superior frontal gyrus and the temporal poles. Because participants were not a clinical sample, the absolute differences were modest: the gap in depression scores between the lowest- and highest-scoring groups was small to moderate in size.
The differences were not limited to depression. The clusters also varied in anxiety, physical complaints, attention problems, aggression and rule-breaking behavior. The group with the highest depressive symptoms also showed the greatest overall burden of internalizing problems (difficulties directed inward, such as depression and anxiety) and externalizing problems (difficulties expressed outwardly through behavior, such as aggression and rule-breaking).
The groups also differed in their mix of women and men, and men scored higher than women on some behavioral problems within certain groups. However, the link between the groups and depression held after accounting for sex.
A machine-learning model using the incoming connections sorted participants into the correct groups with about 87% accuracy, and a more rigorous tuning procedure raised this to about 92%. However, because the groups were themselves defined from the same connectivity data, these figures describe how cleanly the groups could be separated within this dataset, not how accurately the model would identify depression in new people.
In contrast, clustering based on outgoing connections produced two groups that did not differ in symptoms, and conventional measures of regions being active together did not produce meaningful groups. The authors caution that this difference may partly reflect how reliably each type of connection was measured, rather than showing that the sgACC’s inputs matter more for mood than its outputs.
Li and colleagues wrote that the findings “highlight an outsized role of sgACC inputs in distinguishing individuals’ depressive symptom severity in a neurotypical population.”
The study has important limitations. The participants were not recruited based on psychiatric diagnoses, and most had relatively low symptom levels, so the analysis cannot show that these connectivity patterns identify major depressive disorder or predict treatment response.
The findings also come from a single dataset with no independent replication, and the data were collected at one point in time. In addition, Granger causality applied to fMRI data estimates directional patterns in blood-flow signals rather than establishing that one brain region actually drives another.
The study, “Functional inputs to the subgenual cingulate cortex differentiate levels of depressive symptom severity in neurotypical young adults: Granger causality and clustering analyses,” was authored by Huey-Ting Li, Yu Chen, Shefali Chaudhary, Jaime S. Ide and Chiang-Shan R. Li.