How to better stratify the risk of differentiated thyroid carcinomas: the key role of radioactive iodine therapy, age, and gender.

PURPOSE: The risk of relapse of differentiated thyroid carcinomas (DTC) and their indication for radioactive iodine therapy (RAI) are assessed according to ATA risk stratification system principally based on tumor-nodes-metastasis (TNM) staging. However, while establishing the indication for RAI may be a "dilemma," performing it can improve the risk stratification. We aimed to evaluate whether (1) the stratification of risk of recurrence differs when TNM is considered with or without peri-RAI findings and (2) the assessment of the risk of disease-specific mortality is improved by adding age and gender. METHODS: From our database, all DTC patients treated with thyroidectomy and RAI from 1992 to 2017 were included. Subjects with a follow-up shorter than 1 year and positive thyroid antibodies were excluded. Patients were classified into (1) a three-category ATA model based on TNM (basic model) and (2) a five-category model based on TNM plus peri-RAI findings, i.e., thyroglobulin and 131I whole-body scan (advanced model). Relapse was proven by histology and/or imaging. Differences in disease-free survival (DFS) and overall survival (OS) were assessed. RESULTS: We enrolled 907 patients; of these, 4.4% died and 21% suffered recurrence. According to the basic model, there were 11.8% high-risk, 32.9% intermediate-risk, and 55.3% low-risk patients. According to the advanced model, 29.9% of patients were re-classified in a higher risk category and the five categories of this model displayed significantly different risks of relapse and death. The estimate of DFS was significantly higher in the advanced model than in the basic one (ΔC-index = + 6.8%, P < .001). By adding age and gender to the advanced model, the highest performance in predicting death was achieved (ΔC-index = + 5.1%, P < .001). CONCLUSIONS: The peri-RAI findings are essential in order to carefully stratify the risk of DTC recurrence. Integrating these data with age and gender enables those cases at highest risk of death to be identified.