Abstract
OBJECTIVE: Intravenous glucocorticoids are the first-line therapy for moderate-to-severe active thyroid eye disease (TED), but some patients demonstrate steroid resistance. Therefore, identifying reliable pre-treatment biomarkers to predict the therapeutic response to intravenous glucocorticoids is of considerable clinical importance. METHODS: In the discovery phase, pre-treatment serum samples from 10 patients with TED treated with intravenous glucocorticoids (six responders, four non-responders), along with samples from three patients with Graves’ disease without TED and three healthy controls, were subjected to data-independent acquisition proteomics analysis. Candidate biomarkers were validated in an expanded cohort (21 responders, 10 non-responders, 14 Graves’ disease patients, and 14 healthy controls). RESULTS: Data-independent acquisition proteomics analysis revealed 263 differentially abundance proteins between responders and non-responders. Among them, transforming growth factor-β, heparanase, and fibrinogen-like protein 2 were identified as potential biomarker candidates. In the validation phase, only fibrinogen-like protein 2 was significantly elevated in non-responders. Although fibrinogen-like protein 2 alone achieved an area under the curve of 0.76, Least Absolute Shrinkage and Selection Operator regression incorporating clinical parameters identified smoking and high thyroid-stimulating antibody as additional predictors. A three-factor scoring system (fibrinogen-like protein 2 > 39.5 ng/mL, thyroid-stimulating antibody > 2597%, and current smoking; each 1 point) yielded an area under the curve of 0.86, with good reproducibility in bootstrap validation. CONCLUSIONS: Elevated pre-treatment serum fibrinogen-like protein 2 is a promising biomarker for predicting steroid resistance in TED. Combining fibrinogen-like protein 2 with thyroid-stimulating antibody and current smoking provides a clinically useful scoring system to guide personalized treatment. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13044-026-00293-8.