The Positive Lymph Node Number and Postoperative N-Staging Used to Estimate Survival in Patients with Differentiated Thyroid Cancer: Results from the Surveillance, Epidemiology, and End Results Dataset (1988-2008).

ABSTRACT

BACKGROUND: Lymph node metastasis is important when evaluating the prognosis of patients with differentiated thyroid cancer (DTC). However, the current N-staging system cannot fully reflect the clinical significance of cervical lymph node metastasis in DTC. In this study, we employed Surveillance, Epidemiology, and End Results (SEER)-registered DTC cases with lymph node metastasis to determine whether the positive lymph node number (PLNN) could be used to improve stratification of patients in terms of survival. METHODS: We used the SEER dataset to identify all DTC patients with at least one positive cervical lymph node who were examined between 1988 and 2008. Multivariable modeling was used to compare cancer-specific survival (CSS) and overall survival (OS) and to calculate different PLNN cutoff points. RESULTS: In total, 14,359 pN + DTC patients identified in the SEER were included. In multivariate Cox regression analysis, the PLNN was significantly associated with both CSS and OS, whereas neither the lymph node ratio (LNR) nor the (numbers of) lymph nodes examined (LNE) were so associated. The highest C-index value (0.933) and the lowest AIC value (9362.687) obtained indicated that the PLNN better predicted the CSS of DTC than did the LNR or LNE. As the p values for both CSS and OS were minimized, and as the PLNN performed best when cases were grouped, PLNN cutoff points of 10 and 3/10 efficiently stratified DTC patients into two and three levels, respectively. Based on the 3/10 trichotomy, the benefits of radioactive iodine (RAI) treatment were evaluated for each group. Such treatment afforded about a 10% survival benefit in patients with more than 10 lymph node metastases. CONCLUSIONS: Compared with the LNR and LNE under different statistical models, PLNN was superior in terms of DTC staging. A cutoff point of 3/10 was optimal for stratifying patients according to prognosis and was of clinical significance in terms of RAI treatment selection.