Establishment and validation of two nomograms to predict the benefit of concurrent chemotherapy in stage II-IVa nasopharyngeal carcinoma patients with different risk factors: Analysis based on a large cohort

Objective We aimed to establish and validate two nomograms that predict progression-free survival (PFS) and overall survival (OS) in patients with stage II-IVa nasopharyngeal carcinoma (NPC) while evaluating the benefit of concurrent chemotherapy. Patients and Methods We randomly divided 3412 patients newly diagnosed with stage II-IVa NPC between 2008 and 2013 into training and validation 'A' cohorts (n = 1706 each). Another set of patients diagnosed between 2014 and 2016 served as validation cohort 'B' (n = 1503). A Cox multivariate model using the backward stepwise approach was applied to develop the nomograms, which were assessed for accuracy (Harrel C index) and calibration. Results The 3- and 5-year PFS rates in the training cohort were 86.8% (95% confidence interval [CI] 85.0%-88.6%) and 82.3% (95% CI 80.1%-84.5%), respectively. For the PFS nomogram, 5 variables were selected based on a backward procedure in the multivariate Cox model (gender, T stage, N stage, Epstein-Barr virus DNA, and treatment method). The same variables plus patient age and diabetes mellitus were used for the OS nomogram. The Harrell C indices of the training, validation A, and validation B cohorts were 0.711, 0.700, and 0.703, respectively, for PFS, and 0.775, 0.743, and 0.727, respectively, for OS. Both nomograms performed well in terms of calibration in the training and validation cohorts. Conclusions Our nomograms are reliable prognostic predictors of PFS and OS in patients with stage II-IVa NPC. These nomograms could robustly estimate an individual's benefit from concurrent chemotherapy, which assists in treatment decision-making.