Volume 125, Issue 1 p. 79-89
Original Article
Free Access

Proposed modifications and incorporation of plasma Epstein-Barr virus DNA improve the TNM staging system for Epstein-Barr virus-related nasopharyngeal carcinoma

Rui Guo MD, PhD

Rui Guo MD, PhD

Department of Radiation Oncology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

The first 2 authors contributed equally to this work.Search for more papers by this author
Ling-Long Tang MD, PhD

Ling-Long Tang MD, PhD

Department of Radiation Oncology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

The first 2 authors contributed equally to this work.Search for more papers by this author
Yan-Ping Mao MD, PhD

Yan-Ping Mao MD, PhD

Department of Radiation Oncology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

Search for more papers by this author
Xiao-Jing Du MD, PhD

Xiao-Jing Du MD, PhD

Department of Radiation Oncology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

Search for more papers by this author
Lei Chen MD, PhD

Lei Chen MD, PhD

Department of Radiation Oncology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

Search for more papers by this author
Zi-Chen Zhang MD

Zi-Chen Zhang MD

Department of Molecular Diagnostics, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in Southern China, Collaborative Innovation Center of Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

Search for more papers by this author
Li-Zhi Liu MD

Li-Zhi Liu MD

Imaging Diagnosis and Interventional Center, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

Search for more papers by this author
Li Tian MD

Li Tian MD

Imaging Diagnosis and Interventional Center, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

Search for more papers by this author
Xiao-Tong Luo ME

Xiao-Tong Luo ME

School of Life Sciences, Sun Yat-sen University, Guangzhou, China

Search for more papers by this author
Yu-Bin Xie PhD

Yu-Bin Xie PhD

School of Life Sciences, Sun Yat-sen University, Guangzhou, China

Search for more papers by this author
Jian Ren PhD

Jian Ren PhD

School of Life Sciences, Sun Yat-sen University, Guangzhou, China

Collaborative Innovation Center of High Performance Computing, National University of Defense Technology, Changsha, China

Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Sun Yat-sen University, Guangzhou, China

Search for more papers by this author
Ying Sun MD, PhD

Ying Sun MD, PhD

Department of Radiation Oncology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

Search for more papers by this author
Jun Ma MD

Corresponding Author

Jun Ma MD

Department of Radiation Oncology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China

Correspondence: Jun Ma, MD, Department of Radiation Oncology, Cancer Center, Sun Yat-sen University, 651 Dongfeng Road East, Guangzhou, China 510060; [email protected]

Search for more papers by this author
First published: 23 October 2018
Citations: 127

Abstract

Background

The prognosis of patients who have Epstein-Barr virus (EBV)-related nasopharyngeal carcinoma (NPC) in which the tumor tissues harbor EBV have a better prognosis than those without EBV-related NPC. Therefore, the eighth edition of the TNM staging system could be modified for EBV-related NPC by incorporating the measurement of plasma EBV DNA.

Methods

In total, 979 patients with NPC who received intensity-modulated radiotherapy (IMRT) were retrospectively reviewed. Recursive partitioning analysis was conducted based on tumor (T) classification, lymph node (N) classification, and EBV DNA measurement to derive objectively the proposed stage groupings. The validity of the proposed stage groupings was confirmed in a prospective cohort of 550 consecutive patients who also received with IMRT.

Results

The pretreatment plasma EBV DNA level was identified as a significant, negative prognostic factor for progression-free survival and overall survival in univariate analysis (all P < .001) and multivariate analysis (all P < .05). Recursive partitioning analysis of the primary cohort to incorporate EBV DNA generated the following proposed stage groupings: stage RI (T1N0), RIIA (T2-T3N0 or T1-T3N1, EBV DNA ≤2000 copies/mL), stage RIIB (T2-T3N0 or T1-T3N1, EBV DNA >2000 copies/mL; T1-T3N2, EBV DNA ≤2000 copies/mL), stage RIII (T1-T3N2, EBV DNA >2000 copies/mL; T4N0-N2), and stage RIVA (any T and N3). In the validation cohort, the 5-year progression-free survival rate was 100%, 87.9%, 76.7%, 68.7%, and 50.4% for proposed stage RI, RIIA, RIIB, RIII, and RIV NPC, respectively (P < .001). Compared with the eighth edition TNM stage groupings, the proposed stage groupings incorporating EBV DNA provided better hazard consistency, hazard discrimination, outcome prediction, and sample size balance.

Conclusions

The proposed stage groupings have better prognostic performance than the eighth edition of the TNM staging system. EBV DNA titers should be included in the TNM staging system to assess patients who have EBV-related NPC.

Introduction

Nasopharyngeal carcinoma (NPC) has a unique, unbalanced, endemic distribution, with the highest incidences reported in south China and southeast Asia.1 The American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) TNM staging system is the key determinant for prognostic prediction and treatment decisions in NPC.2, 3 The concept of molecular classification of cancer at a clinically relevant level is now accepted as an imminent reality. It is widely believed that the new molecular classification schema will complement traditional and time-honored classifications, such as anatomic staging.4, 5

World Health Organization (WHO) type 2 and 3 NPC accounts for greater than 97% of all cases of NPC in south China, where NPC is endemic, whereas keratinizing squamous cell carcinoma (SCC) is more common in western countries.6 Type 2 and 3 NPC associated with EBV that harbors EBV in tumor tissues have better prognoses than type 1 NPC,7 and is possible that patients who have EBV-related NPC require a separate staging system that includes nonanatomic factors. The detection of cell-free EBV DNA in the plasma/serum of patients with NPC provided a new tumor marker for NPC.8 Previous studies demonstrated that cell-free EBV DNA is correlated with tumor burden, TNM stage, and survival of in patients with NPC.9-11 EBV DNA measurement is now widely used in the clinical setting for primary diagnosis, prognostication, tumor monitoring, as well as population screening.12, 13 Although these observations confirm that EBV DNA could be of value in pretherapy risk stratification and staging,5, 14 robust data to confirm whether and how this biomarker should be incorporated in the staging system remain lacking.

To address this issue, we propose modifications based on the AJCC/UICC template combining anatomic parameters and EBV DNA for patients with EBV-related NPC who undergo intensity-modulated radiotherapy (IMRT). We also assessed the ability of the current TNM staging system and propose stage groupings to separate patients with respect to survival.

Materials and Methods

Patient Characteristics

Between November 2009 and March 2012, an independent cohort of 550 patients with newly diagnosed NPC was prospectively enrolled. All patients were histologically diagnosed with WHO type 2 or 3 NPC and were positive for EBV viral capsid antigen/immunoglobulin A or for EBV immunoglobulin A/early antigen. The EBV DNA level before treatment and the detailed extent of the anatomic structure was also collected. The inclusion and exclusion criteria are described in the Supporting Materials. All patients provided written informed consent before treatment. This study was approved by the Clinical Research Ethics Committee of Sun Yat-sen Cancer Center.

To establish the prognostic model, an initial retrospective study was conducted on a separate cohort of 979 patients with NPC who underwent IMRT at the same cancer center during the same period and who were recruited using the same inclusion and exclusion criteria. The 979 patients in the retrospective cohort were defined as the primary cohort, and the 550 prospectively enrolled patients were defined as the validation cohort. The characteristics of the primary and validation cohorts are summarized in Table 1.The authenticity of this article has been validated by uploading the key raw data onto the Research Data Deposit public platform (http://www.researchdata.org.cn), with the approval RDD number as RDDA2018000804.

Table 1. Clinicopathologic Characteristics of the 1529 Patients With Nasopharyngeal Cancer in the Training and Validation Sets
No. of Patients (%)
Characteristic Training Set, n = 979 Validation Set, n = 550 P
Sex .715
Men 731 (74.7) 406 (73.8)
Women 248 (25.3) 144 (26.2)
Age, y .310
≤45 504 (51.5) 298 (54.2)
>45 475 (48.5) 252 (45.8)
WHO histologic type .928
Type II 5 (0.5) 3 (0.5)
Type III 974 (99.5) 547 (99.5)
Chemotherapy .516
Yes 850 (86.8) 471 (85.6)
No 129 (13.2) 79 (14.4)
Tumor classificationa .587
T1 166 (17.0) 90 (16.4)
T2 155 (15.8) 94 (17.1)
T3 482 (49.2) 255(46.4)
T4 176 (18.0) 111 (20.2)
Lymph node classificationa .110
N0 160 (16.3) 78 (14.2)
N1 581 (59.4) 310 (56.4)
N2 139 (14.2) 102 (18.5)
N3 99 (10.1) 60 (10.9)
Stage group .652
I 48 (4.9) 27 (4.9)
II 202 (20.6) 104 (18.9)
III 446 (45.6) 244 (44.4)
IV 283 (28.9) 175 (31.8)
  • Abbreviation: WHO, World Health Organization
  • a Tumor and lymph node classifications are based on the eighth edition of the American Joint Commission on Cancer/Union for International Cancer Control staging system.

DNA Extraction and Real-Time Quantitative Polymerase Chain Reaction

Plasma EBV DNA concentrations were routinely measured before treatment using quantitative polymerase chain reaction,15 as described in the Supporting Materials.

Clinical Staging

All patients completed a pretreatment evaluation including EBV DNA, as described in a previous study.16 Furthermore, positron emission tomography-computed tomography studies were obtained for 482 of 1529 patients (31.5%). All magnetic resonance imaging (MRI) studies were reviewed by 2 radiologists (L.Z.L. and L.T.) to minimize heterogeneity in restaging. Two radiologists who had over 10 years of experience specializing in head and neck cancer separately evaluated all imaging studies, and disagreements were resolved by consensus. All patients were restaged according to the eighth edition of the AJCC/UICC TNM staging system.

Treatment

All patients received radical radiotherapy using the simultaneous integrated-boost technique. Target volumes were delineated using an individualized delineation protocol that complies with International Commission on Radiation Units and Measurements Reports 50 and 62.17, 18 The prescribed doses were from 66 to 72 grays (Gy) at 2.12 to 2.27 Gy per fraction to the planning target volume of the primary gross tumor volume and from 64 to 70 Gy at 30 to 33 Gy per fraction to the planning target volume of the gross tumor volume of the involved lymph nodes.

During the study, institutional guidelines recommended only IMRT for stage I and concurrent chemoradiotherapy with or without neoadjuvant/adjuvant chemotherapy for stage II to IVB disease. Of 1148 patients who had stage III/IV disease, 1081 (94.2%) received chemotherapy, including concomitant chemoradiotherapy (CCRT), induction chemotherapy, or adjuvant chemotherapy. Induction chemotherapy or adjuvant chemotherapy consisted of cisplatin with 5-fluorouracil, taxanes, or both every 3 weeks for 2 or 3 cycles. Concomitant chemoradiotherapy consisted of cisplatin either during weeks 1, 4, and 7 of radiotherapy or weekly during radiotherapy.

Patient Follow-Up and Statistical Analysis

Patients attended follow-up every 3 months in the first 2 years and every 6 months thereafter (or until death). The endpoints (time to the first defining event) assessed included progression-free survival (PFS), overall survival (OS), distant metastasis-free survival (DMFS), and locoregional relapse-free survival (LRRFS). The Kaplan-Meier method was used to calculate actuarial rates, and the log-rank test was used to compare the differences. Multivariate analysis was performed using a Cox proportional-hazards model to determine the independent significance of variables using backward elimination.

Recursive partitioning analysis (RPA) for PFS based on tumor (T) classification, lymph node (N) classification, and EBV DNA was used to derive objectively the proposed stage groupings in the primary set. The RPA algorithm, which was based on optimized binary partition of T classification, N classification, and EBV DNA, was used to generate subgroups with relatively homogeneous survival outcomes using the program STREE (available at: https://c2s2.yale.edu/software/stree/).19 Next, the proposed stage system incorporating EBV DNA was validated in the validation set. The original evaluation criteria for stage performance defined by Groome et al were used to compare the 2 sets of stage groupings by calculating hazard consistency (similarity of survival rates for subgroups within each stage group), hazard discrimination (differences in survival rates across stage groups to assess how equally they are spaced), outcome prediction (percentage of PFS variation explained by the stage groupings), and sample size balance (difference in sample sizes across stage groups).20 To validate the prognostic value of the 2 stage groupings, internal validation using 1000 bootstrap replicates was carried out for the validation set.21 The area under the receiver-operator characteristic (ROC) curve also was used to assess the predicted validity based on the method of Hanley and McNeil.22 Statistical analyses were conducted using the Stata statistical software package (version 13.0; StataCorp LP, College Station, TX). All tests were 2-sided, and P values < .05 were considered significant.

Results

After a median follow-up of 64.4 months (range, 1.4-91.5 months) for the primary set and 64.6 months (range, 1.3-83.3 months) for the validation set, 86 of 979 patients (8.8%) and 61 of 550 patients (11.1%) in the primary and validation cohorts, respectively, developed local and/or regional recurrence; 120 of 979 patients (12.3%) and 76 of 550 patients (13.8%), respectively, developed distant metastasis; and 132 of 979 patients (13.5%) and 95 of 550 patients (17.3%), respectively, died. The 5-year PFS, OS, DMFS, and LRRFS rates were 80.2%, 87.4%, 87.7%, and 91.1%, respectively for the primary cohort and 76.5%, 83.7%, 86.3%, and 88.1%, respectively, for the validation cohort.

Pretreatment EBV DNA and its Value as a Prognostic Factor

The median pretreatment plasma EBV DNA load was 0 copies per mL in patients with stage I disease; 0.28 × 103 copies per mL in those with stage II disease, 1.89 × 103 copies per mL in those with stage III disease, and 10.55 × 103 copies per mL in those with stage IV disease. Although the EBV DNA loads for different stages overlapped, the median EBV DNA load increased significantly with T classification, N classification, and TNM stage (all P < .05). The results are presented in Supporting Table 1.

In univariate analysis, EBV DNA was a significant prognostic factor for PFS (P < .001), OS (P < .001), and DMFS (P < .001). After adjusting for covariates, EBV DNA remained an independent prognostic factor for PFS (hazard ratio [HR], 1.214; P = .001), OS (HR, 1.288; P < .001), and DMFS (HR, 1.386; P < .001) in multivariate analysis (Table 2).

Table 2. Multivariate Analysis of Prognostic Factors in Patients With Nasopharyngeal Carcinoma Who Received Intensity-Modulated Radiotherapy
Variable HR (95% CI) P
Progression-Free Survival
EBV DNA 1.214 (1.088-1.354) .001
Age: ≥45 vs <45 y 1.400 (1.125-1.742) .003
Tumor classificationa
T2 vs T1a 1.432 (0.921-2.228) .111
T3 vs T1 1.260 (0.854-1.860) .245
T4 vs T1 2.007 (1.326-3.040) .001
Lymph node classificationa
N1 vs N0 1.853 (1.165-2.948) .009
N2 vs N0 2.785 (1.677-4.626) < .001
N3 vs N0 4.131 (2.464-6.925) < .001
Overall survival
EBV DNA 1.288 (1.125-1.475) < .001
Age: ≥45 vs <45 y 1.555 (1.192-2.027) .001
Tumor classificationa
T2 vs T1 1.351 (0.785-2.327) .278
T3 vs T1 1.214 (0.751-1.961) 428
T4 vs T1 2.251 (1.365-3.712) .001
Lymph node classificationa
N1 vs N0 1.550 (0.876-2.745) .133
N2 vs N0 2.779 (1.503-5.136) .001
N3 vs N0 4.834 (2.614-8.939) < .001
Distant metastasis-free survival
EBV DNA 1.386 (1.193-1.609) < .001
Tumor classificationa
T2 vs T1 1.088 (0.621-1.905) .769
T3 vs T1 1.012 (0.622-1.646) .963
T4 vs T1 1.711 (1.023-2.864) .041
Lymph node classificationa
N1 vs N0 1.358 (0.729-2.529) .335
N2 vs N0 2.653 (1.365-5.156) .004
N3 vs N0 4.928 (2.542-9.553) < .001
  • Abbreviations: CI, confidence interval; EBV, Epstein-Barr virus; HR, hazard ratio.
  • a Tumor and lymph node classifications are based on the eighth edition of the American Joint Commission on Cancer/Union for International Cancer Control staging system. The following parameters were included in the Cox proportion-hazards models using backward elimination: age (≥45 vs <45 years), sex (women vs men), World Health Organization histologic grade (type III vs type II), tumor classification, lymph node classification, receipt of chemotherapy (with vs without), and EBV DNA grouped by interquartile range.

Prognostic Performance of the Eighth Edition of the AJCC/UICC TNM Staging System in the Primary Cohort

A relatively monotonic reduction in the 5-year PFS was observed with increased stage according to the eighth edition of the AJCC/UICC TNM staging system in the primary set (97.9%, 85.3%, 84.3%, and 67.4% for stage I, II, III, and IV, respectively; P < .001) (Fig. 1A). An identical trend was observed for N classification (90.5%, 82.2%, 73.7%, and 62.6% for N0, N1, N2, and N3, respectively; P < .001) (Fig. 1C). However, the 5-year PFS was similar between patients who had T2 and T3 tumors (78.9% vs 82.0%, respectively; P = .462) (Fig. 1B) and was indistinguishable between those with stage II and III disease (85.3% vs 84.3%, respectively; P = .639).

Details are in the caption following the image
Progression-free survival is illustrated for different groupings according to (A) disease stage, (B) tumor (T) classification, and (C) lymph node (N) classification of nasopharyngeal carcinoma as defined by the eighth edition of the Union for International Cancer Control/American Joint Committee on Cancer staging system in the primary setting (n = 979).

Generation and Evaluation of a Proposed TNM System Incorporating EBV DNA Into the Primary Set

We incorporated EBV DNA into the eighth edition of the AJCC/UICC TNM staging system using a separate RPA analysis of the primary set, which resulted in the following proposed stage groupings: stage RI (T1N0), stage RIIA (T2-T3N0 or T1-T3N1, EBV DNA ≤2000 copies/mL), stage RIIB (T2-T3N0 or T1-T3N1, EBV DNA >2000 copies/mL; T1-T3N2, EBV DNA ≤ 2000 copies/mL), stage RIII (T1-T3N2, EBV DNA >2000 copies/mL; T4N0-N2), and stage RIVA (any T and N3) (Fig. 2A,B). Metastatic disease (M1) was classified as stage RIVB.

Details are in the caption following the image
(A) This chart illustrates the proposed stage groupings derived by recursive partitioning analysis after the incorporation of pretreatment plasma Epstein-Barr virus (EBV) DNA load into the primary set. (B) TNM groupings are shown for each tumor (T) classification, lymph node (N) classification, metastasis (M) classification, and EBV DNA load. CI indicates confidence interval; PFS, progression-free survival.

By using this proposed system, the 5-year PFS rate was 97.9%, 89.6%, 82.2%, 69.8%, and 62.6% for proposed stage RI, RIIA, RIIB, RIII, and RIVA (P < .001), respectively, in the primary set; and the 5-year OS rate was 100%, 95.4%, 91.6%, 78.0%, and 67.6% for proposed stages RI, RIIA, RIIB, RIII, and RIVA, respectively (P < .001). Adjusted multivariate analysis conferred an increased risk of disease failure (stage RIIA vs RI: HR, 5.29; P = .101; stage RIIB vs RI: HR, 9.66; P = .025; stage RIII vs RI: HR, 17.87; P = .004; stage RIVA vs RI: HR, 22.86; P = .002).

Evaluation of the Proposed TNM Stage Groupings Incorporating EBV DNA Into the Validation Set and All Patients

When we applied the proposed stage groupings incorporating EBV DNA to the validation set, the 5-year PFS rate was 100%, 87.9%, 76.7%, 68.7%, and 50.4% for proposed stage RI, RIIA, RIIB, RIII, and RIVA, respectively (P < .001) (Fig. 3A); and the 5-year rate OS was 100%, 94.2%, 86.5%, 75.5%, and 58.3%% for proposed stage RI, RII, RIII, and RIVA, respectively (P < .001) (Fig. 3B). A comparison of the proposed stage groupings that incorporated EBV DNA into the eighth edition staging revealed that the proposed stage groupings incorporating EBV DNA performed better compared with the eighth edition AJCC/UICC stage in terms of hazard consistency (1.81 vs 4.97), hazard discrimination (0.50 vs 0.56), outcome prediction (27.8 vs 26.4) and sample size balance (0.48 vs 0.53) in the validation cohort. Internal validation using 1000 bootstrap replicates provided similar findings for the validation set (Table 3). In the validation set (n = 550), the area under the ROC curve for PFS was 0.69 for the proposed staging groupings and 0.64 for the eighth edition stages (P = .01) (Fig. 4A), and the area under the ROC curve for OS was 0.72 for the proposed staging groupings and 0.65 for the eighth edition stages (P < .001) (Fig. 4B).

Details are in the caption following the image
(A) Progression-free survival and (B) overall survival are illustrated for different stage groupings of nasopharyngeal carcinoma in the validation set (n = 550) according to the proposed staging system using recursive partitioning analysis.
Table 3. Performance Evaluation of the Proposed TNM Stage Combined With Epstein-Barr Virus DNA in Nasopharyngeal Carcinoma
Evaluation Criteriaa Proposed TNM Stage Combined With EBV DNA AJCC/UICC Eighth Edition TNM Stage
% Hazard consistency 1.81 4.97
Score 0.00 1.00
Rank 1 2
Hazard discrimination 0.50 0.56
Score 0.00 1.00
Rank 1 2
Balance 0.48 0.53
Score 0.00 1.00
Rank 1 2
Outcome prediction (% variance explained) 27.80 26.40
Score 0.00 1.00
Rank 1 2
Overall score 0.00 6.50
Overall rank 1 2
Performance evaluation using internal validation by bootstrap algorithm
% Hazard consistency 1.70 4.16
Score 0.005 0.995
Rank 1.005 1.995
Hazard discrimination 0.46 0.50
Score 0.356 0.644
Rank 1.356 1.644
Balance 0.48 0.53
Score 0.18 0.82
Rank 1.182 1.818
Outcome prediction (% variance explained) 27.3 25.8
Score 0.123 0.877
Rank 1.123 1.877
Overall score 1.059 5.441
Overall rank 1.052 1.948
% Rank = 1 948 52
  • Abbreviations: AJCC/UICC, American Joint Committee on Cancer/Union for International Cancer Control; EBV, Epstein-Barr virus.
  • a Stage performance evaluation criteria are: hazard consistency, similarity of survival rates for subgroups within each stage group; hazard discrimination, differences in survival rates across stage groups to assess how equally they are spaced; outcome prediction, percentage of progression-free survival variation explained by the stage groupings; sample size balance, difference in sample sizes across stage groups.
Details are in the caption following the image
Receiver-operator characteristic curves for the eighth edition of the Union for International Cancer Control/American Joint Committee on Cancer staging system and for the proposed stage groupings are illustrated according to (A) progression-free survival and (B) overall survival in the validation set (n = 550).

Discussion

To our knowledge, this is the first study to propose refinements to the AJCC/UICC TNM staging system by incorporating EBV DNA with validation for patients who have EBV-related NPC. The proposed staging system incorporates EBV DNA and performs better than the eighth edition of the AJCC/UICC TNM staging system.

EBV DNA Has Significant Prognostic Value in NPC

Previous studies demonstrated that the presence of plasma EBV DNA in patients with NPC was associated with apoptosis of tumor tissues and also had the same polymorphism as the primary site,23, 24 indicating that plasma EBV DNA is derived from cancer cells rather than from inflammatory cells. Others have reported data establishing that plasma EBV DNA is closely related to the extent of tumor, it has almost become a tumor marker for predicting prognosis in patients with NPC,25-28 and it also was identified as an independent prognostic factor for patients with NPC in our current study. However, highly varied EBV DNA cutoff levels are used in different studies. Lin et al and Wang et al observed that <1500 copies/mL EBV DNA before treatment had prognostic significance for poorer disease recurrence and OS.25, 27 In contrast, Chai et al reported that a pretreatment EBV DNA load of 8000 copies/mL was a more powerful prognosticator for OS.29 Hou and colleagues reported that <4000 copies/mL (vs ≤4000 copies/mL) was associated with improved survival.30 In the current study, the T classification, N classification, and EBV DNA load before treatment were calculated using RPA, and an EBV DNA titer of 2000 copies/mL was identified as the optimal cutoff for PFS using RPA. This result was in agreement with a previous ROC curve analysis.31

Evaluation of and Proposed Revisions for the Eighth Edition of the UICC/AJCC TNM Staging System for NPC

An accurate staging system is essential for patient prognostication and guiding treatment. Indeed, the eighth edition failed to clearly distinguish PFS between stage II and III and between T2 and T3 tumors in this study. By using RPA, a method for building decision trees of the significant prognostic factors for outcome, Dahlstrom et al proposed new stage groupings for human papillomavirus-positive oropharyngeal cancer.32 Huang et al also used RPA to stage human papillomavirus-positive oropharyngeal carcinoma and recommended that the stage groupings could resemble those of NPC.33 Similarly, different staging criteria may be necessary for EBV-related NPC, because it also can be induced by viral infection (with EBV) and is distinct to other head and neck cancers. In the current study, we applied an RPA model to independently and objectively derive stage groupings after incorporating EBV DNA into the primary set using PFS and not OS, because previous studies indicated that PFS is a valid surrogate endpoint for OS to assess treatment outcomes in patients who have locoregionally advanced NPC. Compared with OS, PFS generally would produce more events, increasing the statistical power and allowing small randomized trials to be done. The use of PFS instead of OS would accelerate the development process of therapeutic regimens by permitting earlier reporting of results.34, 35

Proposals for Revision of the TNM Staging System: Suggested Approaches

The main weakness of the current eighth edition is the heterogeneity of treatment outcomes for each stage group. In the cohorts examined in this study, 16.7% (256 of 1529 patients) of those with T4N0-N2 NPC were downstaged from eighth edition stage IV to proposed stage RIII, 20.5% (313 of 1529 patients) of those with T1-T3N2 NPC and EBV DNA ≤2000 copies/mL or T3N0-N1 NPC and EBV DNA >2000 copies/mL were downstaged from stage III to proposed stage RIIB, and 18.4% (281 of 1529 patients) of those with T3N0-N1 and EBV DNA ≤2000 copies/mL were downstaged to proposed stage RIIA. This downstaging was because of the weakening effect of the T classification. Our previous study of patients who were staged using MRI and received IMRT demonstrated that T classification was not an independent prognostic factor for OS, DFS, or local failure.16

In our study, the eighth edition of the UICC/AJCC TNM staging system for NPC failed to clearly distinguish PFS between T2 and T3 tumors. First, the advent of IMRT has substantially improved the dose distribution in patients with NPC.36 Second, with the increased use of concurrent chemotherapy, patients who undergo IMRT achieve excellent outcomes, with 2-year to 4-year locoregional control rates now exceeding 90%.37, 38 Third, the adverse impact of T3 and T4 extension will further diminish with the continued application of MRI. Furthermore, our data suggest that patients who have eighth edition N2 NPC (bilateral cervical lymph node involvement) and EBV DNA ≤2000 copies/mL should be downstaged to the proposed stage RIIB, because this N2 group had a borderline significant difference in distant failure and no significant difference in lymph node failure compared with patients who had N0-N1 NPC after the incorporation of EBV DNA into the proposed staging system. Moreover, downstaging of stage III to proposed stage RII is appropriate, because there were no significant differences in the outcomes of these groups. In the proposed staging system, the prognosis of patients with high and low EBV DNA concentrations did not differ significantly between those with T1N0 and any T or N3 stage in our study. The EBV DNA titer was low (≤2000 copies/mL) in 45 of 48 patients (93.8%), whereas it was high (>2000 copies/mL) in 78 of 99 patients (78.8%) in the primary cohort. Therefore, the stage RI in the proposed staging group is the previous T1N0 group without EBV DNA. Furthermore, patients with eighth edition N3 lymph node status have significantly poorer outcomes than those with other N classifications. Thus, a more aggressive, systemic approach is required to improve the treatment outcome for patients who have NPC classified as N3.

We ranked the performance of the different stage groupings in the validation set using criteria devised by Groome and colleagues after internal validation using a bootstrap evaluation method. Overall, incorporating EBV DNA into the staging system significantly improved hazard consistency, hazard discrimination, and outcome prediction for PFS, as well as the sample size balance. Furthermore, ROC curve analysis revealed that, in predicting outcomes, the proposed stage groupings incorporating EBV DNA were superior to those in the eighth edition of the UICC/AJCC TNM staging system.

The principal limitation of this study is that the analyses were based on patients who received treatment at a single cancer center; however, the proposed TNM staging system incorporating EBV DNA was derived using the primary set and was validated in the prospective cohort. However, several issues remain to be resolved before EBV DNA can be incorporated into the staging system for NPC; most important, the cutoff levels vary widely between different studies, and international efforts are still ongoing to standardize the quantitative assay.

Conclusions

We propose a revised TNM staging system incorporating EBV DNA that demonstrates better prognostic performance than the eighth edition of the AJCC/UICC TNM staging system for NPC. To improve patient stratification and help devise individualized treatment strategies, future revisions of the AJCC/UICC TNM staging system should be adapted to incorporate EBV DNA.

Funding Support

This work was supported by grants from the Natural Science Foundation of Guang Dong Province (2017A030312003), the Health and Medical Collaborative Innovation Project of Guangzhou City, China (201803040003), the Innovation Team Development Plan of the Ministry of Education (IRT_17R110), the Overseas Expertise Introduction Project for Discipline Innovation (111 Project, B14035), and the National Natural Science Foundation of China (81572658).

Conflict of Interest Disclosures

The authors made no disclosures.

Author Contributions

Rui Guo: Conceptualization; data curation; investigation, methodology, or project administration; formal analysis or validation; statistical analysis or software; study supervision; writing–initial draft; writing–review and editing; had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Ling-Long Tang: Conceptualization; data curation; investigation, methodology, or project administration; study supervision; writing–initial draft; writing–review and editing; had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Yan-Ping Mao: Investigation, methodology, or project administration; writing–initial draft; and writing–review and editing. Xiao-Jung Du: Investigation, methodology, or project administration; formal analysis or validation; statistical analysis or software; writing–initial draft; and writing–review and editing. Lei Chen: Investigation, methodology, or project administration; writing–initial draft; and writing–review and editing. Zi-Chen Zhang: Investigation, methodology, or project administration; writing–initial draft; and writing–review and editing. Li-Zhi Liu: Investigation, methodology, or project administration; writing–initial draft; and writing–review and editing. Li Tian: Investigation, methodology, or project administration; writing–initial draft; and writing–review and editing. Xiao-Tong Luo: Investigation, methodology, or project administration; writing–initial draft; and writing–review and editing. Yu-Bin Xie: Investigation, methodology, or project administration; formal analysis or validation; statistical analysis or software; writing–initial draft; and writing–review and editing. Jian Ren: Investigation, methodology, or project administration; formal analysis or validation; statistical analysis or software; writing–initial draft; and writing–review and editing. Ying Sun: Investigation, methodology, or project administration; writing–initial draft; and writing–review and editing. Jun Ma: Conceptualization; data curation; investigation, methodology, or project administration; funding acquisition; study supervision; writing–initial draft; writing–review and editing; had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.