A serum protein network predicts the need for systemic immunomodulatory therapy in autoimmune uveitis

Objective biomarkers that can predict a severe disease course of autoimmune uveitis are lacking, and warranted for early identification of high-risk patients to improve visual outcome. The need for non-steroid immunomodulatory therapy (IMT) to control autoimmune uveitis is indicative of a more severe disease course. We used aptamer-based proteomics and a bioinformatic pipeline to uncover the serum protein network of 52 treatment-free patients and 26 healthy controls, and validation cohorts of 114 and 67 patients. Network-based analyses identified a highly co-expressed serum signature (n=85 proteins) whose concentration was consistently low in controls, but varied between cases. Patients that were positive for the signature at baseline showed a significantly increased risk for IMT during follow-up, independent of anatomical location of disease. In an independent cohort (n=114), we established robust risk categories and confirmed that patients with high levels of the signature at diagnosis had a significantly increased risk to start IMT during follow-up. Finally, we further validated the predictive association of the signature in a third cohort of 67 treatment-naive North-American patients. A serum protein signature was highly predictive for IMT in human autoimmune uveitis and may serve as an objective blood biomarker to aid in clinical-decision making.