Serum lipid metabolic characteristics and potential biomarkers in first-episode schizophrenia

Lipids play a vital role in health and disease, but changes to their circulating levels and the link with schizophrenia remains poorly characterized. This study aimed to investigate the pathological lipid profiles in patients with first-episode schizophrenia (FES) and explore potential biomarkers for the early identification of FES.

This study recruited 99 drug-naive FES patients and 60 healthy controls (HC). Serum samples were analyzed using an absolute quantitative lipidomics with ultra-high performance liquid chromatography. Psychiatric symptoms were assessed using Positive and Negative Syndrome Scale (PANSS) scale. Data were analyzed with regression analysis and XGBoost-based machine learning to develop biomarkers and classification of FES.

In FES patients, 33 lipid metabolites exhibited significantly altered levels independent of BMI, age, and sex. A predictive model was constructed using 14 lipid metabolites to accurately distinguish individuals with FES from healthy controls, achieving an area under the receiver operating characteristic curve (AUC) of 0.967, F1 score of 0.918, an accuracy of 0.894, a specificity of 0.824, and a recall of 0.933. Bioinformatic analyses indicated changes in serum lipids co-expression correlated with disease state.

Our findings indicate dysregulated lipids implicated in the pathogenesis of schizophrenia and suggest potential lipid biomarkers may serve as indicators for the differential diagnosis of schizophrenia.