Nomogram combining serum thyroglobulin and IFN-α predicts radioactive iodine response in DTC patients

Differentiated thyroid cancer (DTC) is the most common endocrine malignancy, with radioactive iodine (RAI) therapy often used post-surgery to ablate residual thyroid tissue and treat metastatic disease. However, predicting individual patient response to initial RAI remains a significant clinical challenge. Current risk stratification systems, while useful, lack the precision to consistently identify patients who will achieve an excellent response versus those who will not, leading to potential overtreatment or undertreatment. This gap highlights the need for more robust predictive biomarkers and models to personalize therapy.

This retrospective study analyzed 429 patients with intermediate- or high-risk DTC who received initial RAI therapy across two institutions. A training cohort (n = 300) from one hospital was used to identify predictors, while an external validation cohort (n = 129) from another hospital confirmed the findings. Treatment response was assessed 6 to 12 months post-RAI based on the 2025 American Thyroid Association (ATA) Response Evaluation System, categorizing patients into excellent response (ER) and non-excellent response (non-ER) groups. Univariate and multivariate logistic regression identified independent predictors, which were then used to construct a nomogram model. The model's performance was validated using calibration curves, decision curve analysis (DCA), and receiver operating characteristic (ROC) curve analysis.

In the training cohort, serum thyroglobulin (sTg) (OR = 0.938, P < 0.001) and serum interferon-alpha (IFN-α) (OR = 1.446, P < 0.001) emerged as independent risk factors for predicting treatment response to initial RAI therapy in intermediate- or high-risk DTC patients. These findings suggest that lower sTg and higher IFN-α levels are associated with a better response. The nomogram model, developed using these two biomarkers, demonstrated acceptable predictive performance, as indicated by its ROC curve. Calibration curves showed a high degree of concordance between the model's predicted probabilities and the observed clinical outcomes. Furthermore, DCA confirmed the model's potential for clinical utility, suggesting it could aid in decision-making. These results were consistently corroborated in the external validation cohort, reinforcing the robustness of the model.

The nomogram model, incorporating sTg and IFN-α, showed a high degree of concordance between predicted and observed RAI response, with both biomarkers identified as independent predictors (P < 0.001 for both).