Pretreatment Brain Connectome Fingerprint Predicts Treatment Response in Major Depressive Disorder. (December 2020)

• Record Type: Journal Article
• Authors: Fan, Siyan; Nemati, Samaneh; Akiki, Teddy J.; Roscoe, Jeremy; Averill, Christopher L.; Fouda, Samar; Averill, Lynnette A.; Abdallah, Chadi G.
• Is Part Of: Chronic stress. Volume 4(2020)
• Publication Date: 2020-12
• Subjects: antidepressants -- brain architecture -- intrinsic connectivity networks -- machine learning -- major depressive disorders; Stress (Psychology) -- Periodicals; Stress (Physiology) -- Periodicals; Stress (Physiology); Stress, Psychological -- therapy; Stress, Physiological; Mental Disorders -- etiology; Electronic journals; Periodicals; Periodicals; 616.89
• DOI: 10.1177/2470547020984726
• Languages: English
• ISSNs: 2470-5470

Abstract

Background: Major depressive disorder (MDD) treatment is characterized by low remission rate and often involves weeks to months of treatment. Identification of pretreatment biomarkers of response may play a critical role in novel drug development, in enhanced prognostic predictions, and perhaps in providing more personalized medicine. Using a network restricted strength predictive modeling (NRS-PM) approach, the goal of the current study was to identify pretreatment functional connectome fingerprints (CFPs) that (1) predict symptom improvement regardless of treatment modality and (2) predict treatment specific improvement. Methods: Functional magnetic resonance imaging and behavioral data from unmedicated patients with MDD (n = 200) were investigated. Participants were randomized to daily treatment of sertraline or placebo for 8 weeks. NRS-PM with 1000 iterations of 10 cross-validation were implemented to identify brain connectivity signatures that predict percent improvement in depression severity at week-8. Results: The study identified a pretreatment CFP that significantly predicts symptom improvement independent of treatment modality but failed to identify a treatment specific CFP. Regardless of treatment modality, improved antidepressant response was predicted by high pretreatment connectivity between modules in the default mode network and the rest of the brain, but low external connectivity in the executive network. Moreover, high pretreatment internal nodal connectivity in the bilateral caudate predicted better response. Conclusions: The identified CFP may contribute to drug development and ultimately to enhanced prognostic predictions. However, the results do not assist with providing personalized medicine, as pretreatment functional connectivity failed to predict treatment specific response.