Dynamic Monitoring of Thyroid-Stimulating Hormone Receptor Antibody and Extraocular Muscle Thickness: A Clinical Risk Stratification Model for Recurrence Risk Post-Glucocorticoid Therapy in TED Patients

Objective: To develop a clinical risk stratification model for thyroid eye disease (TED) recurrence post-glucocorticoid therapy by integrating dynamic thyroid-stimulating hormone receptor antibody (TRAb), extraocular muscle (EOM) thickness, and baseline clinical-biochemical indicators, with only internal validation. Methods: A single-center retrospective cohort study of 426 TED patients (2016– 2023) was conducted. Recurrence (12-month follow-up) was adjudicated by two ophthalmologists. LASSO regression screened predictors, and multiple machine learning models were built and validated via 10-fold cross-validation. Model performance was assessed by AUC and decision curve analysis (DCA). Results: 98 patients (23.0%) had recurrence, with 89% occurring after 6 months post-treatment. Recurrent patients showed slower TRAb decline and persistent EOM thickening. 15 key predictors were identified, and the random forest (RF) model had the best performance (AUC=0.92, 95% CI: 0.88– 0.95; accuracy=0.87) with consistent subgroup results (Events Per Variable=6.5, potential overfitting risk). Conclusion: This internally validated risk stratification model identifies factors associated with TED recurrence post-glucocorticoid therapy, with dynamic TRAb and EOM thickness as key indicators. The findings are preliminary due to no external validation and temporal limitations, and prospective multi-center studies are required before clinical implementation to inform individualized follow-up.