Metastatic Risk Stratification of 2526 Medullary Thyroid Carcinoma Patients: A Study Based on Surveillance, Epidemiology, and End Results Database.

The risk of distant metastasis in medullary thyroid carcinoma (MTC) has not been well studied. Additional evaluation of MTC metastatic risk can be helpful for improving the quality of medical management. Therefore, we conducted a large population study to develop a method to stratify the risk of metastasis at the initial presentation of MTC patients. We collected 3612 MTC patients from the Surveillance, Epidemiology, and End Results (SEER) database, and included 2526 MTC patients in the study after applying exclusion criteria. We selected the most informative variables from a learning cohort of 2019 patients to obtain 1000 models by repetitive random data splicing into training and regularization cohorts. We selected the optimal model and developed a risk table from that model. Our risk table variables consist of age, gender, tumor size, extrathyroidal extension, and lymph node metastasis. The final model showed good calibration when metastatic risk was < 25% and good performance with areas under the curve (AUCs) of 0.81, 0.84, and 0.84 in the training, regularization, and test cohorts, respectively. We performed K-means clustering analysis on the model's metastatic estimation and determined three risk groups of patients with significant survival differences (p < 0.001). Low-risk patients had 0.88%, 1.3%, and 0.5% while high-risk patients had 19.7%, 15.8%, and 17.8% risk of metastasis in the three cohorts, respectively. The incorporation of our table into the International MTC Grading System (IMTCGS) requires more comprehensive clinicopathological studies.