Background
Postoperative stimulated thyroglobulin (ps-Tg) is an established biomarker for assessing treatment response in papillary thyroid carcinoma (PTC), but optimal integration of biochemical parameters into the 2025 American Thyroid Association (ATA) recurrence risk stratification system remains undefined.
Materials and methods
We retrospectively analyzed 1,117 PTC patients who underwent first radioactive iodine (RAI) therapy at the First Affiliated Hospital of Jinan University between 2015 and 2021. Two models were built to estimate the probability of achieving an excellent response (ER) 1 year after RAI: Model 1 included all variables selected by LASSO (pathological N stage, 2025 ATA recurrence risk stratification, ps-Tg, ps-Tg/thyroid-stimulating hormone [TSH] ratio, and thyroglobulin antibody [TgAb]); Model 2 served as a control, including only pathological N stage and recurrence risk stratification. Model performance was evaluated using receiver operating characteristic (ROC) analysis, LASSO regression, decision curve analysis (DCA), and net reclassification improvement (NRI).
Results
ROC analysis identified optimal cut-off values of 4.10 ng/mL for ps-Tg and 0.07 for the ps-Tg/TSH ratio (area under the curve [AUC] 0.856 and 0.843, respectively). In the training cohort, Model 1 outperformed Model 2 (AUC 0.862 vs. 0.614; R2 0.531 vs. 0.056) and significantly improved reclassification (IDI 0.389; overall NRI 0.297; event NRI 0.163). DCA demonstrated greater clinical net benefit for Model 1 across risk thresholds of 0.02-0.80, and findings in the validation cohort were consistent.
Conclusion
Based on ps-Tg, the ps-Tg/TSH ratio, TgAb, pathological N stage, and the ATA 2025 recurrence risk stratification, the combined model showed good predictive value for 1-year ER after RAI therapy in patients with PTC. This model may serve as a complementary tool to existing postoperative risk assessment and provide additional support for follow-up stratification and individualized management.