Elsevier

Oral Oncology

Volume 91, April 2019, Pages 113-120
Oral Oncology

Depicting distant metastatic risk by refined subgroups derived from the 8th edition nasopharyngeal carcinoma TNM

https://doi.org/10.1016/j.oraloncology.2019.02.021Get rights and content

Highlights

  • Significant heterogeneity in DM risk exists among T-N subsets within cTNM-8 stages.

  • The RPA groups improved intra-group hazard consistency compared to cTNM-8 stage groups.

  • The RPA groups provide a strong anatomic foundation to augment DM prediction.

Abstract

Background

Tumor-nodal-metastasis (TNM) is the most important survival predictor in nasopharyngeal carcinoma (NPC). Distant metastasis (DM) is the predominant failure pattern of NPC in the intensity-modulated radiotherapy (IMRT) era. The DM risk appears to be different for T-N subsets within the same clinical stage. Appropriately depicting DM risk has emerged as an important issue in tailoring individualized treatment and underpins the reason for this study.

Methods

A total of 1616 non-metastatic (M0) NPC patients treated with IMRT were included. All were re-staged according to the 8th edition AJCC/UICC TNM (TNM-8). DM-free survival (DMFS) was calculated and compared among T-N subsets within each stage and DM risk groups were derived by Recursive-partitioning analysis (RPA) based on ordinal T and N categories.

Results

Significant heterogeneity in DM risk was evident among T-N subsets within cTNM-8 stages II-IV. The RPA algorithm classified patients into four DM risk groups: RPA-I (T1N0-1 and T2-3N0), RPA-II (T2-3N1), RPA-III (T4N0-1 and T1-3N2) and RPA-IV (T4N2 and T1-4N3), with 5-year DMFS of 93.4% (95% CI: 91.3–96.1), 84.3% (80.8–87.8), 78.9% (75.4–82.4) and 63.6% (56.3–70.9), respectively (p < 0.001). Compared to cTNM-8 stage grouping, RPA grouping had a lower Akaike information criterion (AIC) and higher Harrell’s concordance index (c-index) for DMFS.

Conclusions

Significant heterogeneity in DM risk exists among T-N subsets within cTNM-8 stages. The RPA groups demonstrated improved intra-group hazard consistency compared to cTNM-8 stage groups. While further validation is warranted, these RPA prognostic groupings provide a strong anatomic foundation to augment DM prediction for optimal targeting in future clinical trials.

Introduction

The application of intensity-modulated radiotherapy (IMRT) has significantly improved loco-regional control (LRC) of nasopharyngeal carcinoma (NPC); however, about 15% to 30% patients experience failure at distant sites [1]. In fact, distant metastasis (DM) has become the principal form of failure and main cause of death for NPC patients following IMRT [1], [2]. Therefore, precise evaluation of DM risk and appropriate stratification is an important first step to tailoring more individualized treatment and improving survival.

The anatomic tumor-node-metastasis (TNM) staging system has long been recognized as the most important survival indicator for NPC patients, and is critical in facilitating treatment planning. While invaluable for cancer treatment, and with additional advantages from simplicity and ease of use worldwide, the TNM classification was recognized at its genesis as having limitations. Thus, it has always been appreciated that many factors in addition to TNM contribute to prognosis (including age, sex, social determinants of health and numerous pathological factors) [3]. Moreover, other outcomes may be governed by separate prognostic factors compared to overall survival (OS), the usual end-point used in cancer staging[4], [5]. For this reason, the UICC and AJCC have attempted to refine knowledge and processes to address important prognostic factors relevant to outcomes following specific treatments [6]. To date for NPC, there has not been a comprehensive process to address the prediction of DM, potentially the most important outcome that governs survival in this disease. This may largely be a consequence of the traditional concern about achieving local control in the protected sanctuary of the skull base; however, this emphasis has diminished with the contemporary high LRC rates due to IMRT, high quality imaging, and cisplatin chemotherapy as a potential radiotherapy sensitizer. Effectively the challenge has shifted to the risk of DM, which may have its own unique characteristics and may be receptive to new approaches that may target DM separately.

Although many biomarkers, such as blood EBV-DNA [7], [8], and serum lactate dehydrogenase (LDH) [9], [10], have been identified as DM predictors, TNM categories and stage groups represent the basic anatomic building blocks needed to explore incorporation of these biomarkers. Unfortunately, significant heterogeneity in risk of DM is often observed for different T-N subgroups of patients within apparently equivalent clinical TNM stage [11], [12], [13], [14], [15], [16]. The UICC/AJCC staging system was updated to the 8th version (TNM-8)[17], in which both T and N categories were modified. Whether the heterogeneity of DM risk still exists within the updated stage groupings remains unknown. Hence, we conducted this retrospective analysis of a series of non-metastatic NPC patients treated with IMRT in our institution to evaluate the predictability of DM risk within each TNM-8 stage. We used recursive-partitioning analysis (RPA) to construct a framework intended to provide high sensitivity to predict DM risk using the anatomic components of the 8th edition T and N. It is envisioned that additional biomarkers factors could subsequently be employed to refine the risk and determine with precision subgroups of patients with the greatest need for novel systemic treatments to reduce DM. Such factors could comprise tissue-based factors such as microRNA signatures [18], [19] or circulating blood indices such as EBV copy number or LDH levels.

Section snippets

Patients and treatment

Patients with histologically-proven NPC who received definitive IMRT with or without chemotherapy at our institution between June 2005 and December 2011 were candidates for this retrospectively study. Other eligible criteria included: (1) no history of previous treatment or prior malignancy; (2) a completed pretreatment evaluation according to our institutional protocol [20], (3) availability of imaging data that permitted re-staging according to the TNM-8 [17]. With these criteria, 1616

Statistical analyses

The primary endpoint was DMFS (time to DM) where death without DM was censored. An additional endpoint was OS (time to death due to any causes). Descriptive statistics were provided with median and ranges for continuous variables, and frequency and percentages for categorical variables. Survival curves were created with the Kaplan–Meier method and compared by log-rank test. Multivariate analyses (MVA) by Cox proportional hazards model were performed to evaluate the stage prognosis on DMFS and

Patients’ characteristics and performance of the TNM-8 in predicting risk of DM

The clinical characteristics of the 1616 eligible patients are listed in Table 1. The majority were nonkeratinizing (undifferentiated) carcinoma (1483, 91.8%) according to the World Health Organization (WHO) classification. Median follow-up time was 53 months. A total of 93 (5.8%) local, 72 (4.5%) regional, and 273 (16.9%) distant failures were identified. The 5-year DMFS for TNM-8 stage I, II, III and IV were 97.4% (95% CI: 93.9–100), 90.7% (87.8–93.6), 82.5% (79.4–85.6) and 71.6% (67.1–76.1),

Discussion

This large single institution cohort of non-metastatic NPC treated with IMRT showed that, similar to the 6th and 7th edition UICC/AJCC staging system, TNM-8 has limitations in portraying the risk of DM consistently within each stage. Using an RPA algorithm, we generated the following four distinctly different DM risk groups: RPA-I (T1-3N0 and T1N1), RPA-II (T2-3N1), RPA-III (T4N0-1 and T1-3N2) and RPA-IV (T4N2 and T1-4N3) [Fig. 2]. This RPA risk grouping performed better in depicting risk of DM

Conclusions

This large single institution study of a contemporary treated NPC cohort shows that DM risk increases with higher TNM-8 stage, although significant heterogeneity exists for DMFS within each stage. Deriving relatively homogeneous DM risk groups would help refine treatment strategies. Using an RPA algorithm, we generated four groups based on TNM-8T and N definitions, namely RPA-I (T1N0-1 and T2-3N0), RPA-II (T2-3N1), RPA-III (T4N0-1 and T1-3N2) and RPA-IV (T4N2 and T1-4N3). These RPA risk

Conflict of interest statement

None declared

Acknowledgement

This work was sponsored by National Clinical Key Specialty Construction Program and Key Clinical Specialty Discipline Construction Program of Fujian, China. This research is also supported by grant from the Fujian Provincial Natural Science Foundation of China (grant No. 2018 J01275). We acknowledge the Bartley-Smith Wharton Fund and the O. Warwick Prize of the Canadian Cancer Society for supporting BO’S academic activities.

References (51)

  • P. Teo et al.

    The prognostic significance of parapharyngeal tumour involvement in nasopharyngeal carcinoma

    Radiotherapy Oncol : J European Soc Therapeutic Radiol Oncol

    (1996)
  • G.L. Xiao et al.

    Prognostic influence of parapharyngeal space involvement in nasopharyngeal carcinoma

    Int J Radiat Oncol Biol Phys

    (2002)
  • L.L. Tang et al.

    Prognostic value of parapharyngeal extension in nasopharyngeal carcinoma treated with intensity modulated radiotherapy

    Radiotherapy Oncol : J European Soc Therapeutic Radiol Oncol

    (2014)
  • C.N. Cao et al.

    Update report of T4 classification nasopharyngeal carcinoma after intensity-modulated radiotherapy: an analysis of survival and treatment toxicities

    Oral Oncol

    (2015)
  • S.F. Leung et al.

    Plasma Epstein-Barr viral DNA load at midpoint of radiotherapy course predicts outcome in advanced-stage nasopharyngeal carcinoma

    Annals Oncol : Official J European Soc Medi Oncol

    (2014)
  • A.W. Lee et al.

    Management of nasopharyngeal carcinoma: current practice and future perspective

    J Clinical Oncol : Official J Amer Soc Clin Oncol

    (2015)
  • M.X. Zhang et al.

    Intensity-modulated radiotherapy prolongs the survival of patients with nasopharyngeal carcinoma compared with conventional two-dimensional radiotherapy: a 10-year experience with a large cohort and long follow-up

    European J Cancer (Oxford, England)

    (1990)
  • C.C. Compton et al.

    Risk Models for Individualized Prognosis in the Practice of Precision Oncology

  • B. O'Sullivan

    Introduction

  • B. O'Sullivan et al.

    Chapter 3: Prognosis and classification of cancer

  • J. Zhang et al.

    Epstein-Barr virus DNA level as a novel prognostic factor in nasopharyngeal carcinoma: a meta-analysis

    Medicine

    (2016)
  • J.J. Pan et al.

    Prognostic nomogram for refining the prognostication of the proposed 8th edition of the AJCC/UICC staging system for nasopharyngeal cancer in the era of intensity-modulated radiotherapy

    Cancer

    (2016)
  • X.B. Wan et al.

    High pretreatment serum lactate dehydrogenase level correlates with disease relapse and predicts an inferior outcome in locally advanced nasopharyngeal carcinoma

    European J Cancer (Oxford, England)

    (1990)
  • Q. Guo et al.

    Long-term survival of nasopharyngeal carcinoma patients with stage II in intensity-modulated radiation therapy era

    Jpn J Clin Oncol

    (2016)
  • K.H. Chen et al.

    Comparison of the efficacy between concurrent chemoradiotherapy with or without adjuvant chemotherapy and intensity-modulated radiotherapy alone for stage II nasopharyngeal carcinoma

    Oncotarget

    (2016)
  • Cited by (9)

    View all citing articles on Scopus
    1

    Both Qiaojuan Guo and Tianzhu Lu contributed equally to this study.

    View full text