Assessment of age as different variable types for determining survival in differentiated thyroid cancer

PURPOSE

The current tumor, node, metastasis (TNM) system uses an age of 55 years as a threshold for differentiated thyroid cancer (DTC). The aim of our study was to explore the concept of using age as a continuous variable.

METHODS

A total of 36,559 patients with DTC in the Surveillance, Epidemiology, and End Results (SEER) database and 7491 patients in our centers were enrolled. Overall survival (OS) and cancer-specific survival (CSS) were compared. Furthermore, the different statistical model performance of the 6th edition TNM system and age cutoffs for papillary (PTC) and follicular thyroid cancer (FTC) were assessed. Then, a nomogram was built and validated to evaluate the efficacy of age as a continuous variable for predicting survival.

RESULTS

The OS and CSS of patients with DTC were significantly increased in patients <55 years compared with those aged ≥55 years. However, no significant differences in prognosis were observed in certain groups as patients between 50 and 60 years were stratified by 1-year increments. Furthermore, the highest concordance index (C-index) was observed in the TNM staging without an age cutoff in SEER database (0.895), our two centers (0.877) and receiver operating characteristic (ROC) curves showed different age cutoffs for PTC and FTC. More importantly, the nomogram incorporating age as a continuous variable showed a favorable area under the ROC curve and calibration for training and validation groups.

CONCLUSIONS

The utilization of age as a continuous variable is a rational approach for predicting outcome in DTC patients.