Treatment outcome is associated with pre-treatment connectome measures across psychiatric disorders - evidence for connectomic reserve?

Predicting treatment efficacy in psychiatric disorders remains challenging, despite the availability of effective interventions. Previous studies suggest a link between pre-treatment brain network characteristics and treatment efficacy in individual disorders, but cross-disorder investigations are lacking. We analyzed pre-treatment MRI data from 177 individuals (113 females) with either obsessive-compulsive disorder (OCD) or post-traumatic stress disorder with comorbid personality disorders (PTSD) that received different non-pharmacological treatments. Using diffusion and resting-state MRI, we constructed structural, functional, and multilayer connectomes and calculated network measures for network integration (e.g. global efficiency, eccentricity), segregation (modularity) and their balance (small-worldness). We assessed the relationship between these pre-treatment network measures, and treatment improvement using mixed-model and Bayesian analyses. We also compared psychiatric cases with healthy controls and investigated associations between clinical response and treatment-induced changes in network measures. Across disorders and treatments, psychiatric cases showed a 41.6 ± 29.6 % symptom improvement (62 % response rate) after treatment. They also showed pre-treatment differences in functional and multilayer network topology compared to healthy controls. Symptom improvement was associated with pre-treatment functional (P = 0.04) and structural small-worldness (P = 0.01), and multilayer eccentricity (P = 0.01), while responders had higher functional modularity (P = 0.02). Results were robust across trials and treatments, when adjusting for medication status and showed high credibility in Bayesian analyses. Network change associations with treatment response were only modest. These results show that pre-treatment connectome characteristics are related to treatment response, regardless of treatment and psychiatric disorder, and suggest that individual differences in intrinsic features of the human connectome underlie amenability to treatment.