Prognostic models for 1-year survival of NPC after radiotherapy in different ages

Purpose Previous studies have shown that approximately 10% of nasopharyngeal cancer (NPC) patients die within a year of disease onset, and that age is an independent predictor. However, no predictive model has been developed. We aimed to establish novel prognostic models to predict the 1-year cancer-specific survival (CSS) of young, middle-aged, and older patients with NPC after radiotherapy. Methods The data of 2822 NPC patients who underwent radiotherapy between 2004 and 2015 were reviewed from the surveillance, epidemiology, and end results database. We divided them into young, middle-aged, and older people groups according to age (< 44 years, 45–59 years, and ≥ 60 years, respectively). Multivariate analyses were performed, and prognostic models were constructed. Results Multivariate analyses indicated that age, ethnicity, histological subtype, T, and M stage were independent predictors of 1-year CSS in the older people group. In contrast, ethnicity and age were not found to have predictive value in the young and middle-aged groups, respectively. Accordingly, three prognostic models with excellent predictive values were established for the three groups (C-indices: 0.791 [95% CI 0.722–0.859], 0.763 [95% CI 0.721–0.806] and 0.723 [95% CI 0.683–0.763], respectively). These predictive values are higher than those of the eighth edition American joint committee cancer tumor-node-metastasis (TNM) classification system. Conclusion Three prognostic models for predicting the 1-year CSS of young, middle-aged, and older NPC patients after radiotherapy showed better predictive power than the TNM classification system. These models may guide treatment strategies and clinical decision-making in different cohorts.