Differential diagnosis of major psychiatric disorders using mass spectrometry-based blood proteomic analysis ; 질량 분석기 기반 혈액 단백체 분석을 이용한 주요정신질환의 감별진단

학위논문(박사) -- 서울대학교대학원 : 의과대학 의학과, 2022. 8. 안용민. ; Background: Major depressive disorder (MDD), bipolar disorder (BD), and schizophrenia (SCZ) are representative major psychiatric disorders that are known to be associated with life-long disability and mortality. These disorders are difficult to distinguish, as their diagnosis is based on subjective symptoms and behavioral observations. Recent studies suggest that profiling and targeted quantification of proteomes might help in objective differentiation between these disorders. Thus, this study was conducted to compare and differentiate these disorders based on the quantification of peripheral proteins. Methods: Mass spectrometry-based proteomic profiling analysis was performed on serum samples from psychotropic drug-free 15 MDD and 10 BD patients. T-tests were performed with pairwise comparisons to detect differentially expressed proteins (DEPs) (Study 1). The study was expanded to plasma samples of 174 MDD, 170 BD, 171 SCZ, and 160 healthy controls Both targeted proteomics and proteomic profiling were performed to quantify and verify proteomic candidate targets that differentiated these disorders. Through repeated LASSO regression with feature extraction and weighted model averaging of targeted proteomics, multiprotein-marker (MPM) models were developed to differentiate MDD, BD, and SCZ. The performance of ensemble models that combined MPM models and the Symptom Checklist-90-Revised was compared with clinician rater score-based models (Study 2). In both studies, functions and pathways related to differential proteins were predicted with bioinformatics analysis. Results: Fourteen DEPs were statistically significant between drug-free MDD and BD. RAB7A, ROCK2 were significantly overexpressed in MDD, and EPO7 was significantly overexpressed in BD (Study 1). Each MPM model developed for pairwise patients group comparison (MDD vs BD, MDD vs SCZ, BD vs SCZ) demonstrated reasonable or good differentiation performance in independent test sets (AUROC=0.74~0.82). In ...