浙江医学2025，Vol.47Issue(13)：1365-1369,5.DOI:10.12056/j.issn.1006-2785.2025.47.13.2024-2673
基于非靶向尿液代谢组学建立糖尿病肾脏疾病诊断模型的研究
Establishment of a diagnostic model for diabetic kidney disease based on untargeted urinary metabolomics
摘要
Abstract
Objective To identify differential metabolites that distinguished patients with diabetic kidney disease(DKD)from those with type 2 diabetes mellitus(T2DM)using untargeted urinary metabolomics,and to construct a screening model for DKD.Methods From January to December 2021,a total of 505 T2DM patients admitted to the Department of Endocrinology at Jinhua Central Hospital were retrospectively selected and divided into an observation group(with DKD)and a control group(without DKD).Urine samples from 50 patients in each group were randomly selected to conduct untargeted metabolomics analysis.Metabolites combination with predictive value were screened using orthogonal partial least-squares discriminant analysis(OPLS-DA)and least absolute shrinkage and selection operator(LASSO)regression.Subsequently,quantitative detection by liquid chromatography-mass spectrometry was performed on all 505 patients,and a logistic regression model was established for DKD diagnosis.The model's performance was evaluated using ROC and calibration curves.Results A total of 129 differential metabolites were identified from the urinary sample of the 100 patients.Through OPLS-DA and LASSO regression analyses,6 metabolites with significant predictive value were screened out:succinic acid,β-hydroxybutyrate,salicyluric acid,choline,acetoacetic acid,and pyroglutamic acid.The quantitative detection on the 505 patients showed that except for pyroglutamic acid,the remaining 5 metabolites were significantly different between the two groups(all P＜0.05).A multivariate logistic regression(Stepwise method)model was constructed with 6 variables of succinic acid,β-hydroxybutyrate,choline,acetoacetic acid,hypertension,and duration of diabetes.The AUC of the model was 0.911(95％CI:0.885-0.937),with a sensitivity of 0.848 and a specificity of 0.898.The calibration curve indicated good agreement between the predicted and actual occurrence probability(P＞0.05).Conclusion Four metabolic markers screened based on urine metabonomics technology may become new indicators for the diagnosis of DKD and provide reference for clinical diagnosis and intervention.关键词
糖尿病肾脏疾病/尿液代谢物/非靶向代谢组学技术/诊断模型Key words
Diabetic kidney disease/Urinary metabolites/Untargeted metabolomics/Diagnostic model引用本文复制引用
范菽卫,王华斌..基于非靶向尿液代谢组学建立糖尿病肾脏疾病诊断模型的研究[J].浙江医学,2025,47(13):1365-1369,5.基金项目
浙江省基础公益研究计划项目(LGF22H200021) （LGF22H200021）
