Brain network connectivity in major depression: extending findings from a large public data set by meta-analysis across sites

In this short communication, we test whether patients with major depression and healthy individuals have different functional connectivity within established brain networks. To this end, we leverage a very large multi-site data set of resting state fMRI data (1,300 depressed patients and 1,128 controls) collected by 25 groups. A previous study conducted on this data set compared functional connectivity of the default mode network between the two groups. In our investigation, we performed a meta-analysis across sites quantifying the effects of depression and symptom severity on connectivity of several brain-wide networks beyond the default mode. Running a meta-analysis instead of a mega-analysis also allowed us to calculate effect sizes, heterogeneity and prediction intervals that will be valuable to inform future studies wishing to investigate network functional connectivity in depression. Our results indicate that network connectivity differences between depressed and healthy subjects are consistently small, with confidence intervals almost always encompassing zero, in line with the mixed findings from previous research. Default mode network connectivity differences between depressed patients and controls were exceptionally heterogeneous across sites, suggesting the existence of depression sub-types with normo- and hypo-connected default mode network or a strong impact of clinical confounds on default mode network connectivity. The only networks for which connectivity in depressed individuals was consistently lower than in controls were the somato-motor and visual networks, which could be promising understudied targets for future investigation. Overall, we highlight the need of minimizing heterogeneity in future multi-site studies on functional connectivity in depression and the need for more research on novel taxonomies of mental illness that are robustly anchored in brain function.