A Metabolomics Study of Serum in Hospitalized Patients With Chronic Schizophrenia

Metabolomics has attracted attention as a new method for understanding the molecular mechanisms of psychiatric disorders. Current metabolomics technology allows us to measure over hundreds of metabolites at a time and is a useful indicator of the consequences of complex and continuous changes in metabolic profiles due to the execution of genomic information and external factors of biological activity. Therefore, metabolomics is imperative to the discovery of biomarkers and mechanisms associated with pathophysiological processes. In this study, we investigated metabolites changes in hospitalized patients with chronic schizophrenia compared to that in healthy controls, and examined the correlations between the metabolites and psychiatric symptoms. Thirty patients with schizophrenia and ten healthy controls participated in this study between September 2019 and June 2020. The mean duration of disease in patients with schizophrenia was 26 years. Clinical and neuropsychiatric symptoms of patients with schizophrenia were assessed using the Positive and Negative Syndrome Scale (PANSS). Metabolomics was conducted using Capillary Electrophoresis Fourier Transform Mass Spectrometry (CE-FTMS), using serum samples from patients with schizophrenia and healthy controls. Metabolomics assigned a candidate compound to the 446 (cation 279, anion 167) peaks. Hierarchical cluster analysis (HCA), principal component analysis (PCA), logistic regression analysis, receiver operating characteristic (ROC) analysis, and linear regression analysis were used to analyze the metabolites changes, identifying the disease and the relationship between metabolites and psychiatric symptoms. HCA showed that approximately 60% of metabolites had lower peak values in patients with schizophrenia than in healthy controls. Glutamate metabolism and the urea cycle had the highest proportions in the metabolic pathway, which decreased in patients with schizophrenia. PCA showed a clear separation between patients with schizophrenia and healthy controls in the first principal component (the contribution ratio of the first principal component was 15.9%). Logistic regression analysis suggested that the first principal component was a predictor of disease (odds = 1.36, 95%CI = 1.11-1.67, = 0.0032). ROC analysis showed a sensitivity of 93% and a specificity of 100% for the diagnosis of schizophrenia with a cut-off value of the first principal component; -3.33 (AUC = 0.95). We extracted the high factor loading