1053PD - Refining staging system for nasopharyngeal carcinoma treated with intensity-modulated radiation therapy

Abstract

Background: We incorporated baseline plasma EBV DNA into refinement of stage groups for nasopharyngeal carcinoma (NPC) treated with radical intensity-modulated radiation therapy (IMRT). Methods: Patients with non-metastatic NPC treated with radical IMRT +/- adjunct chemotherapy based on 7th edition of American Joint Committee on Cancer (AJCC) system were recruited prospectively from 2010 to 2016. All patients had baseline and serial post-IMRT plasma EBV DNA (in copies/ml) measured and were staged with MRI and PET-CT. Recursive partitioning analysis (RPA) with repeated internal validations derived new stage groups with incorporation of baseline plasma EBV DNA. Multivariable analyses were used to calculate adjusted hazard ratios (AHRs) to derive a new set of AHR stages. Comparison of performance of survival prediction among these 3 sets of stage groups was done to find the best-performing stage set. Results: The cohort included 520 patients treated with IMRT +/- adjunct chemotherapy with a median follow-up of 5.0 years. They were re-staged based on 8th edition of AJCC system. 5-year overall survival (OS) and cancer-specific survival (CSS) were as follows: stage I (OS 89.5%; CSS 100%), II (OS 87.8%; CSS 94.7%), III (OS 85.0%; CSS 90.0%) and IVA (OS 74.4%; CSS 79.9%) (p = 0.058 and p = 0.003 respectively). RPA derived NPC into 3 new stages with corresponding OS and CSS: RPA-I (T1-T4N0-N2 & T1-T2N3 & EBV DNA < =2000) (OS 89.1%; CSS 95.2%), RPA-II (T1-T4N0-N2 & T1-T2N3 & EBV DNA >2000) (OS 80.5%; CSS 84.1%) and RPA-III (T3-T4N3) (OS 58.2%; CSS 67.1%) (both p < 0.001 and p < 0.001 respectively). AHR (I: T1-T2N0-N2; II: T3-T4N0-N2 & T1-T2N3; III: T3-T4N3) after adjusting age, smoking status, treatment (chemoradiation vs. IMRT alone), baseline LDH and plasma EBV DNA also yielded a valid classification (p < 0.001 for both OS and CSS) but was worse on survival prediction compared to RPA. The RPA stages demonstrated better survival prediction especially on CSS after 1000 bootstrapping replicates (bootstrap scores – OS: 0.469; CSS: 0.752) than AHR stages (OS: 0.436; CSS: 0.206) and 8th edition AJCC (OS: 0.095; CSS: 0.043). Conclusions: A novel RPA-based TNM stage groups revealed significantly better survival prediction compared with the 8th edition AJCC and AHR stages. Clinical trial identification: NCT02476669 Legal entity responsible for the study: Department of Clinical Oncology, The University of Hong Kong and Clinical Oncology Center, The University of Hong Kong-Shenzhen Hospital Funding: SK Yee Medical Foundation Disclosure: All authors have declared no conflicts of interest.