This study aimed to develop a clinically accessible, cost-effective nomogram model to predict intravenous glucocorticoid (IVGC) response in active moderate-to-severe thyroid-associated ophthalmopathy (TAO) by integrating novel optical coherence tomography (OCT)-derived choroidal parameters, thereby addressing limitations of existing biomarkers tied to specialized infrastructure and high costs.

A retrospective analysis included 90 patients (178 eyes) with active moderate-to-severe TAO treated with IVGC per EUGOGO guidelines. Clinically routine parameters, specifically thyrotropin receptor antibody (TRAb), total cholesterol (TC), lymphocyte-to-monocyte ratio (LMR), and OCT-derived subfoveal choroidal thickness (SFCT), were analyzed via multivariate logistic regression. Model performance was evaluated using ROC analysis, 1000-bootstrap validation, and calibration curves.

Elevated TRAb (OR = 0.95, P = 0.005), TC (OR = 0.155, P < 0.001), LMR (OR = 0.709, P = 0.006), and reduced SFCT (OR = 0.983, P < 0.001) independently predicted IVGC response. The nomogram showed high accuracy (AUC = 0.926, 95 % CI: 0.880–0.962) and robust calibration. These biomarkers collectively reflect autoimmune dysregulation, lipid-mediated inflammation, and fibrotic remodeling, underlying therapeutic resistance.

The developed nomogram effectively integrates the capabilities of optical coherence tomography (OCT) and routine biomarkers to provide a precise, cost-effective, and accessible strategy for predicting IVGC treatment response in TAO. This tool empowers clinicians to guide therapeutic personalization, thereby optimizing patient outcomes and minimizing the risks associated with ineffective therapy.