Prognostic nomograms for nasopharyngeal carcinoma with nodal features and potential indication for N staging system: Validation and comparison of seven N stage schemes

To identify the prognostic value of the nodal features, propose a nomogram-based N stage system and evaluate the performance of seven N stage schemes of nasopharyngeal carcinoma (NPC) patients. Data from 1638 non-distant metastatic NPC patients were used to develop nomograms predicting 3-year and 5-year overall survival (OS) and distant metastasis-free survival (DMFS). Based on nomogram and multivariate analyses, a new N-stage scheme was proposed. The performance of the nomogram-based N staging system was assessed against five newly proposed N staging systems and the current 8th N staging system using a quantitative model to compare hazard consistency, discrimination, outcome prediction, and sample size balance. The Kaplan-Meier method with log-rank tests was used to compare survival differences. Nomograms to predict OS and DMFS were constructed using extranodal extension infiltrating the surrounding structures (ENEmax), maximal axial diameter (MAD), large retropharyngeal lymph nodes (RLN, minimal axial diameter > 1.5 cm), multiple central nodal necrosis (CNN), and total lymph node (LN) number and level. Multivariate analysis showed the independent prognostic value of ENEmax and MAD > 3 cm for all selected survival endpoints (p < 0.05). Large RLN and lower neck involvement were independently associated with OS (p < 0.05). We proposed using a large RLN and MAD > 3 cm as N2 factors, and ENEmax and lower neck involvement as N3 factors. Among the seven N-stage schemes, our nomogram-based N scheme and ENEmax to N3 scheme (ENE3) ranked in the top two in the overall comparison with the elevated outcome predicting value (highest c-index). However, between the N0, N1, N1, and N2 subgroups, the ENE3 scheme showed no difference in OS or DMFS (p > 0.05). The predictive model highlighted the independent prognostic value of ENEmax, cervical lymph node, MAD, and large RLN, which can be used as criteria for future N staging.