Comparing whole-brain directed with undirected connectivity estimates under ketamine and placebo

Abstract

Analyzing the network-level effects of psychiatric medications on the human brain with functional magnetic resonance imaging (fMRI) has shown promise in uncovering details on their mechanism of action. In particular, ketamine has gained attention in recent years in psychiatric pharmacotherapy due to its rapid-acting antidepressant effects. Ketamine represents a significant development in the treatment of depression, offering hope for individuals for whom traditional treatment options have been ineffective. However, the whole-brain network-level effects of ketamine have so far only been analyzed with undirected connectivity, offering only information on the correlations across different brain regions. In contrast, dynamic causal modeling (DCM) allows for the estimation of directed connectivity. Regression DCM (rDCM) is a recent variant of DCM designed to be more computationally efficient and applicable to larger sets of brain regions, making it suitable for estimating the directed effects of pharmacological interventions like ketamine across the whole brain. In this study, we therefore estimate the fMRI-based whole-brain directed network of 30 healthy participants in a randomized placebo-controlled, cross-over trial, in which they were administered subanesthetic doses of ketamine and placebo. We then find the connections with significant changes between placebo and ketamine and analyze the differences between the directed and undirected networks. We find that ketamine has the strongest influence in the visual processing regions and that directed connectivity uncovered almost double the amount of significant differences compared to undirected connectivity. These additional effects of ketamine are potentially missed in current studies analyzing undirected networks.