Improved survival prediction for oropharyngeal cancer beyond TNMv8.

PURPOSE: For oropharynx squamous cell carcinoma (OPSCC) this study aimed to: (i) compare 5-year overall survival (OS) stratification by AJCC/UICC TNM versions 7 (TNMv7) and 8 (TNMv8), (ii) determine whether changes to T and N stage groupings improve prognostication and (iii) develop and validate a model incorporating additional clinical characteristics to improve 5-year OS prediction. MATERIAL AND METHODS: All OPSCC treated with curative-intent at our institution between 2011 and 2017 were included. The primary endpoint was 5-year OS. Survival curves were produced for TNMv7 and TNMv8. A three-way interaction between T, N stage and p16 status was evaluated for improved prognostication. Cox proportional hazards modelling was used to derive a new predictive model. RESULTS: Of 750 OPSCC cases, 574 (77%) were p16-positive. TNMv8 was more prognostic than TNMv7 (concordance probability estimate [CPE] +/- SE = 0.72 +/- 0.02 vs 0.53 +/- 0.02). For p16-positive disease, TNMv8 discriminated stages II vs I (HR 2.32, 95% CI 1.47-3.67) and III vs II (HR 1.75, 95% CI 1.13-2.72). For p16-negative disease, TNMv7 and TNMv8 demonstrated poor hazard discrimination. Different T, N stage and p16-status combinations did not improve prognostication after adjusting for other factors (CPE = 0.79 vs 0.79, p = 0.998). A model for p16-positive and p16-negative OPSCC including additional clinical characteristics improved 5-year OS prediction beyond TNMv8 (c-index 0.76 +/- 0.02). CONCLUSIONS: TNMv8 is superior to TNMv7 for p16-positive OPSCC, but both performed poorly for p16-negative disease. A novel model incorporating additional clinical characteristics improved 5-year OS prediction for both p16-positive and p16-negative disease.