Urolithiasis is characterized by a high morbidity and recurrence rate, primarily attributed to metabolic disorders. The identification of more metabolic biomarkers would provide valuable insights into the etiology of stone formation and the assessment of disease risk. The present study aimed to seek potential organic acid (OA) biomarkers from morning urine samples and explore new methods based on machine learning (ML) for metabolic risk prediction of urolithiasis.
Morning urine samples were collected from 117 healthy controls and 156 urolithiasis patients. Gas chromatography-mass spectrometry (GC-MS) was used to obtain metabolic profiles. Principal component analysis (PCA) and ML were carried out to screen robust markers and establish a prediction evaluation model.
There were 25 differential metabolites identified, such as palmitic acid, L-pyroglutamic acid, glyoxylate, and ketoglutarate, mainly involving arginine and proline metabolism, fatty acid degradation, glycine, serine, and threonine metabolism, glyoxylate and dicarboxylic acid metabolism. The urinary organic acid markers significantly improved the performance of the ML model. The sensitivity and specificity were up to 87.50% and 84.38%, respectively. The area under the receiver operating characteristic curve (AUC) was significantly improved (AUC = 0.9248).
The results suggest that OA profiles in morning urine can improve the accuracy of predicting urolithiasis risk, and possibly help to understand the involvement of metabolic perturbations in metabolic pathways of stone formation and to provide new insights.
Kidney & blood pressure research. 2024 Dec 11 [Epub ahead of print]
Jiangtao Yang, Dongfang Zhang, Yan Lu, Haixing Mai, Song Wu, Qin Yang, Hanxiong Zheng, Ruqin Yu, Hongmin Luo, Panpan Jiang, Liping Wu, Caili Zhong, Chenqing Zheng, Yanling Yang, Jiaxiang Cui, Qifang Lei, Zhaohui He