Objective To study and compare the practical use of three prognostic predication models in clinical non-metastatic renal cell carcinoma (RCC).
Methods We retrospectively collected the data of 1 202 clinical non-metastatic RCC patients operated on between 1999 and 2012 at West China Hospital of Sichuan University. Survival analysis method was used to establish three prognostic prediction models including SSIGN, Leibovich and UISS based on different clinical and pathological indicators. The predictive ability was evaluated by the area under receiver operating characteristic curve (AUC).
Results Of the 1 202 clinical non-metastatic RCC patients, 1 030 cases were limited RCC patients, and 172 cases were locally advanced RCC patients. The median follow-up time of the patients was 63.02 months. The 2-year and 5-year overall survival rate were 94.7% and 87.6%, respectively. The 5-year overall survival rates of patients with different clinical stages and grades were significantly different. The predictive accuracies for limited RCC of UISS, SSIGN and Leibovich model were 0.667, 0.785 and 0.758, respectively. For locally advanced RCC, the predictive accuracies of the three models were all lower than 0.6.
Conclusion SSIGN has certain predictive value in clinical non-metastatic RCC. However, for the advanced RCC, all the prognostic models demonstrate limited predictive value.