Study of Serum Metabolic Biomarkers and Prediction Models of Cantharidin‐Induced Nephrotoxicity in Rats Based on Dynamic Metabolomics

The clinical application of cantharidin (CTD) is seriously limited due to its nephrotoxicity. Therefore, this study aims to investigate sensitive biomarkers for the evaluation and prediction of nephrotoxicity induced by CTD in rat. A total of 80 rats were randomly divided into four groups: control group and three doses of CTD groups. After 0, 1, 5, 15, and 28 days of intragastric administration, rat serum and urine were collected for biochemical indexes, then serum was used for metabolomic analyses, and rat kidney was collected for pathological and ultrastructural observation. The levels of serum crea (Scr), blood urea nitrogen (BUN), urea, urine crea (Ucrea), and urinary microalbumin (UmALB) were significantly increased after administration of different doses of CTD (p < 0.05). Additionally, histopathology and cell ultrastructure observation of kidney showed significant cell inflammatory infiltration and glomerular edema. Seven metabolic biomarkers including 6-hydroxymelatonin were significantly disturbed by CTD. The CatBoost Classifier prediction model was used to establish the CTD nephrotoxicity prediction model, and the prediction accuracy and precision were 0.645 and 0.640, respectively. Moreover, 6-hydroxymelatonin was found to be most useful biomarkers for evaluating the CTD nephrotoxicity. Finally, the seven metabolic biomarkers were found mainly involved in pyruvate metabolism, pantothenate and CoA biosynthesis.