Predictive value of tear lipidomics biomarkers for TAO activity and relationship with clinical characteristics

The objective of this study was to explore the lipid metabolic changes in the active thyroid-associated ophthalmopathy (TAO) through tear lipidomics analysis, screen for biomarkers related to disease activity, and analyze their correlation with clinical features. The study included 32 patients with active TAO and 30 patients with inactive TAO. Liquid chromatography-mass spectrometry (LC-MS) was used to perform lipidomics analysis on tear samples to identify differential lipid molecules. Multivariate statistical analyses, including principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA), were conducted, and machine learning algorithms were employed to evaluate the predictive ability of lipid biomarkers. Additionally, clinical characteristics and blood indicators of the patients were collected to analyze their correlation with the lipid biomarkers. The study identified 247 significantly different lipids in the tears of patients with active TAO, of which 104 were upregulated, mainly involving sphingolipids, glycerophospholipids, and glycerolipids. Through machine learning, four lipids (BisMePA(36:6e), MGMG(38:0), PC(38:3), SM(d38:1)) were selected, which showed good predictive ability for TAO activity (AUC >0.8 for all). Moreover, these lipids were significantly correlated with blood lipid indicators, exophthalmos degree, Schirmer I test, and the area of the foveal avascular zone (FAZ) of the retina. This tear lipidomics analysis provides a new approach for screening biomarkers in the active phase of TAO. The identified lipid biomarkers are significantly correlated with clinical features and have potential clinical application value.