INTRODUCTION: Reflux esophagitis (RE) is a common upper gastrointestinal disorder, and its diagnosis currently relies primarily on invasive endoscopic examination. The lack of reliable non-invasive biomarkers substantially limits early detection and large-scale screening. Saliva represents a promising biofluid for metabolomics research, as it can reflect metabolic alterations associated with upper gastrointestinal pathology. OBJECTIVES: This study aimed to identify potential salivary lipid biomarkers associated with RE, and to develop a non-invasive diagnostic model using metabolomics and lipidomics. METHODS: Saliva samples from patients clinically diagnosed with RE and healthy controls were analyzed. The analysis included a discovery cohort (n = 144) and an independent validation cohort (n = 146). Differential metabolites were screened using the untargeted metabolomics approach of ultra-high performance liquid chromatography-mass spectrometry (UHPLC-MS), and then quantitative verification was conducted using targeted lipidomics. Multivariate statistical analysis, random forest algorithms, and receiver operating characteristic (ROC) analysis were applied. RESULTS: Untargeted metabolomics revealed significant metabolic differences between RE patients and healthy controls, with marked enrichment of sphingolipid and glycerophospholipid metabolism. Targeted lipidomics identified six consistently dysregulated salivary lipids: DAG (18: 1₁8: 2), S-1-P, PE (P-16: 0₁8: 1), DAG (16: 0₁8: 2), DAG (18: 1₁8: 1), and DAG (16: 0₁8: 1). A multimetabolite model based on these lipids effectively distinguished RE patients from healthy controls, achieving an AUC of 99. 45% in the discovery cohort and 97. 17% in the validation cohort. CONCLUSION: This study identified a salivary lipid signature associated with RE and supports the potential of this lipidomic approach as a non-invasive method to distinguish RE from healthy controls.
Wang et al. (Sun,) studied this question.