Abstract
INTRODUCTION: Extant research has implicated functional connectivity of the subgenual anterior cingulate cortex (sgACC) in major depressive disorders or depressive traits in neurotypical populations. However, prior studies have not distinguished the inputs and outputs of the sgACC, and the "diagnostic" accuracy of these connectivity metrics remains elusive. METHODS: Here, we analyzed data of 890 neurotypical young adults (459 women, age 22-35) from the Human Connectome Project using Granger causality analyses (GCA) with the sgACC as the seed and 268 regions of interest from the Shen's atlas as targets. Depressive and other comorbid symptoms were assessed using the Adult Self Report. Individual connectivities were assessed with an F test and group results were evaluated with a binomial test, both at a corrected threshold. RESULTS: We identified brain regions with significant inputs to and outputs from the sgACC. Clustering analyses of Granger causality inputs, but not Granger causality outputs or resting state connectivity features revealed distinct subject clusters, effectively distinguishing individuals with different levels of depressive symptom severity and those with comorbidities. Specifically, weaker projections from the fronto-parietal and orbitofrontal cortices, anterior insula, temporal cortices, and cerebellum to the sgACC characterized five clusters with low to high scores of depression as well as comorbid internalizing and externalizing problems. Machine learning using a logistic classifier with the significant "GCA-in" features and 5-fold cross-validation achieved 87% accuracy in distinguishing subject clusters, including those with high vs. low depressive symptom severity levels. DISCUSSION: These new findings specify the functional inputs and outputs of the sgACC and highlight an outsized role of sgACC inputs in distinguishing individuals' depressive symptom severity in a neurotypical population.