代谢组学结合机器学习算法探究心脏瓣膜置换术后早期认知功能障碍生物标志物OA

Objective Metabolomics combined with machine learning algorithms was used to systematically study the preoperative serum metabolites of patients with early postoperative cognitive dysfunction(POCD)after heart valve replacement,so as to screen biomarkers that may predict early POCD after heart valve replacement and explore the corresponding metabolic regulatory mechanisms.Methods A total of 60 patients underwent heart valve replacement under extracorporeal circulation were selected and divided into early-POCD group(group P)and non-POCD group(group N)according to whether POCD occurred or not.Metabolomic analysis was performed on preoperative serum samples of patients in group P and group N to screen the differential metabolites and metabolic pathways.The biomarkers related to early POCD were identified by random forest algorithm.Results A total of 532 differential metabolites were detected by metabonomics analysis,and 5 biomarkers were screened by random forest algorithm,namely quinoline,3'-sialyllactose,sphingomyelin(d18∶1/20∶0),lysophosphatidylcholine[P-18∶1(9Z)]and 25-hydroxycholesterol.Among them,the main metabolic pathways were phenylalanine metabolism,primary bile acid biosynthesis,ascorbic acid and aldonate metabolism,pentose and glucuronate interconversion,tryptophan metabolism,drug metabolism-cytochrome P450,porphyrin and chlorophyll metabolism.Conclusion Many metabolic pathways in patients with early POCD after heart valve replacement under extracorporeal circulation have changed before operation,which may lead to the occurrence of early POCD.Quinoline,3'-sialyllactose,sphingomyelin(d18∶1/20∶0),lysophosphatidylcholine[P-18∶1(9Z)]and 25-hydroxycholesterol may be biomarkers for predicting early POCD.