Connectivity model of the anatomic substrates and network abnormalities in major depressive disorder: A coordinate meta-analysis of resting-state functional connectivity

Abstract

Increasing data suggests major depressive disorder (MDD) involves abnormal functional connectivity within a variety of large-scale brain networks. However, due to the use of unstandardized parcellation schemes, the interactions between these networks and the specific neuroanatomic substrates involved requires further review. We therefore sought to conduct a meta-analysis of functional connectivity changes encountered in MDD using a detailed and standardized parcellation scheme. A literature search for relevant resting-state fMRI studies related to MDD in PubMed was conducted. BrainMap's GingerALE 2.3.6 extracted the relevant fMRI data for creation of an activation likelihood estimation (ALE). A sphere was placed at the MNI coordinate of each ALE cluster and seed origin point, and the Human Connectome Project (HCP) parcellation schema was projected on these spheres. The parcellations most present in the ALE were analyzed based on their associated functional network and/or subcortical area to identify abnormal pairs based on the ALE and seed origin parcellation. Ultimately, 483 subjects across 15 studies were analyzed, wherein areas of decreased or increased functional connectivity compared to healthy controls were identified. Our MDD model most commonly implicated increased default mode network (DMN)-central executive network (CEN) pairs, while decreased paired networks commonly included the DMN with other brain networks. All intra DMN-DMN connections and salience network (SN) pairs showed decreased functional connectivity, while all intra CENCEN functional connectivity were increased compared to controls. We hypothesize that our findings of abnormal connectivity between the DMN, CEN, and SN core cognitive networks may demonstrate the inappropriate allocation of cognitive resources and cognitive depletion believed to cause persisting rumination in depression. Despite previous claims, DMN connectivity was found to be generally decreased, and we propose its connectivity direction is dependent on its interacting network partner and the specific parcellations involved. While both of these hypotheses remain speculative and require further validation, our work provides a comprehensive and anatomically precise model to be refined in future studies focusing on the functional connectivity underlying MDD pathophysiology.

Keywords

• Depression
• Resting-state fMRI
• Connectome
• Meta-analysis
• Brain network