PROJECT SUMMARY The goal of this project is to develop sex-specific resting state fMRI biomarkers for major depressive disorder (MDD) and to determine their underlying genetic correlates. Our primary aim is to determine the locations and the degree of sex-specific functional connectivity differences associated with MDD patients as compared to healthy subjects, and to test whether those sex-specific connectivity differences define sex-specific subtypes of MDD. The benefit of achieving these goals will be twofold: (1) in the clinic, achieving these goals will allow for more accurate diagnostic and treatment-response predictive algorithms for MDD, to reduce time to remission in MDD patients. (2) In the laboratory, our work will guide discovery of novel sex-specific circuit mechanisms and treatments for MDD. Our second aim is to determine if sex-specific MDD effects on connectivity can be predicted by regional gene expression in the brain. Investigating the relationship between brain gene expression and MDD effects on functional connectivity would implicate novel genes in the circuit mechanism of MDD and identify genes that interact with sex in driving MDD symptomatology. To achieve these aims, we will use parametric and non-parametric statistical testing to define sex-specific functional connectivity differences between depressed and healthy men, and between depressed and healthy women. Secondary analyses will test whether sex-specific classifiers for differentiating MDD subjects and healthy controls (“diagnostic biomarkers”) outperform classifiers trained on all subjects independent of sex. To identify genetic correlates of sex-specific MDD effects on functional connectivity, we will use the multivariate technique of partial least squares regression to locate linear combinations of genes which can predict MDD-related connectivity differences in males and in females. Finally, to identify sex-specific subtypes of MDD, we will use a combination of regularized canonical correlation analysis and hierarchical clustering in a validated method for subtype discovery previously published in Dr. Conor Liston’s lab. This project will fill a substantial gap in our knowledge of sex differences in MDD functional connectivity and the underlying genetic correlates of those differences. This project will also involve the execution of a concrete training plan to allow me to develop concrete technical skills in fMRI analysis and machine learning techniques, conceptual skill in hypothesis testing, data interpretation, and scientific communication, and clinical competency as a licensed physician, all under the guidance and mentorship of the project’s sponsor Dr. Liston and co-sponsor Dr. Francis Lee, who are both accomplished physician-scientists.

PROJECT NARRATIVE Major depressive disorder (MDD) has been cited as the leading cause of disability worldwide, affecting up to 14.2 million patients in the United States per year, and contributing to over 47,000 suicides per year. Most patients experience a delay in reaching remission due to a lack of precise treatment selection methods. Our goal is to apply a biomarker-driven approach of precision medicine in psychiatry, using machine learning tools to analyze human neuroimaging and genetic data, with the clinical endpoint of shortened time to remission for MDD patients.