Abstract
Idiopathic uveitis (IU) and Vogt-Koyanagi-Harada (VKH) syndrome are common types of uveitis. However, the exact pathological mechanisms of IU and VKH remain unclear. Proteomic analysis of aqueous humor (AH), the most easily accessible intraocular fluid and a key site of uveitis development, may reveal potential biomarkers and elucidate uveitis pathogenesis. In this study, 44 AH samples, including 12 IU cases, 16 VKH cases, and 16 controls, were subjected to label-free quantitative proteomic analysis. We identified 557 proteins from a comprehensive spectral library of 634 proteins across all samples. The AH proteomic profiles of the IU and VKH groups were different from those of the control group. Differential analysis revealed a shared pattern of extracellular matrix disruption and downregulation of retinal cellular proteins in the IU and VKH groups. Enrichment analysis revealed a protein composition indicative of inflammation in the AH of the IU and VKH groups but not in that of the control group. In the IU and VKH groups, innate immunity played an important role, as indicated by complement cascade activation and overexpression of innate immune cell markers. Extreme gradient boosting (XGBoost), an efficient and robust machine learning algorithm, was subsequently used to screen potential biomarkers for classifying the IU, VKH, and control groups. Transferrin and complement factor B were deemed the most important and represent a promising biomarker panel. These proteins were validated by high-resolution multiple reaction monitoring (HR-MRM) in an independent validation cohort. A classification decision tree was subsequently built for the diagnosis. Our findings further the understanding of the underlying molecular mechanisms in IU and VKH and facilitate the development of potential therapeutic and diagnostic strategies.