Schizophrenia (SCZ) is a highly heritable, polygenic complex mental disorder with imprecise diagnostic boundaries. Finding sensitive and specific novel biomarkers to improve the biological homogeneity of SCZ diagnosis is still one of the research hotspots. To identify the blood specific diagnostic biomarkers of SCZ, we performed RNA sequencing (RNA-seq) on 30 peripheral blood samples from 15 first-episode drug-naïve SCZ patients and 15 healthy controls (CTL). By performing multiple bioinformatics analysis algorithms based on RNA-seq data and microarray datasets, including differential expression genes (DEGs) analysis, WGCNA and CIBERSORT, we first identified 6 specific key genes (TOMM7, SNRPG, KRT1, AQP10, TMEM14B and CLEC12A) in SCZ. Moreover, we found that the proportions of lymphocyte, monocyte and neutrophils were significantly distinct in SCZ patients with CTL samples. Therefore, combining various features including age, sex and the novel blood biomarkers, we constructed the risk prediction model with three classifiers (RF: Random Forest; SVM: support vector machine; DT: decision tree) through repeated k-fold cross validation ensuring better generalizability. Finest result of Area under Receiver Operating Characteristic (AUROC) score of 0.91 was achieved by RF classifier and with a comparable good performance of AUROC 0.77 in external validation dataset. A lower AUROC of 0.63 was demonstrated when it was further applied to a Bipolar disorder (BPD) cohort. In conclusion, the study identified three peripheral core immunocytes and six key genes associated with the occurrence of SCZ, and further studies are required to test and validate these novel biomarkers for early diagnosis and treatment of SCZ.
Keywords: Bioinformatics; Diagnosis; Gene expression; RNA sequencing; Schizophrenia.
Copyright © 2022 Elsevier Ltd. All rights reserved.