Clinical Characteristics and Prognostic Factors of Early and Late Recurrence After Definitive Radiotherapy for Nasopharyngeal Carcinoma

We investigated the clinical characteristics, prognostic factors, and post-recurrence prognostic factors of early- and late-recurrence patients for nasopharyngeal carcinoma (NPC) after definitive intensity-modulated radiation therapy (IMRT). This was a single-center retrospective analysis of patients in China from January 2010 to December 2015. The prognostic factors for overall survival (OS) and post-recurrence OS of early- and late-recurrence patients were identified using univariate and multivariate Cox regression analyses. Of the 9,468 patients included, 409 (4.3%), 325 (3.4%), and 182(1.9%) developed purely local recurrence, purely regional recurrence, and locoregional recurrence during follow-up, respectively. In the purely local recurrence group, 192 patients (46.9%) developed early local recurrence (ETR), and 217 patients (53.1%) developed late local recurrence (LTR). Of the 192 ETR patients, multivariate Cox regression analysis revealed that age and gender were independent risk factors of OS, and post-recurrence best supportive treatment (PRBST) was associated with poorer post-recurrence OS. Of the 217 LTR patients, the results revealed that baseline value of EBV-DNA was an independent risk factor for OS, while PRBST was associated with poorer post-recurrence OS. In the purely regional recurrence group, 183 patients (56.3%) developed early regional recurrence (ENR), and 142 patients (43.7%) developed late regional recurrence (LNR). Of the 183 ENR patients, multivariate Cox regression analysis revealed that alcohol abuse and TNM stage were independent risk factors of OS, while alcohol drinkers and PRBST were associated with poorer post-recurrence OS. Of the 142 LNR patients, PRBST was associated with poorer post-recurrence OS. In the locoregional recurrence group, 87 patients (47.8%) developed early locoregional recurrence (ELR), and 95 patients (52.2%) developed late locoregional recurrence (LLR). Of the 87 ELR patients, multivariate Cox regression analysis revealed that N stage and TNM stage were independent risk factors of OS, and N2/3 stage and PRBST were associated with poorer post-recurrence OS. Of the 95 LLR patients, the results revealed that T stage was an independent risk factor for OS, while T3/4 stage and PRBST were associated with poorer post-recurrence OS. Patients with LTR/LNR/LLR demonstrate significantly better OS compared with patients with ETR/ENR/ELR, Nevertheless, post-recurrence OS between patients with ETR/ENR/ELR and LTR/LNR/LLR was not significantly different.

We investigated the clinical characteristics, prognostic factors, and post-recurrence prognostic factors of early-and late-recurrence patients for nasopharyngeal carcinoma (NPC) after definitive intensity-modulated radiation therapy (IMRT). This was a single-center retrospective analysis of patients in China from January 2010 to December 2015. The prognostic factors for overall survival (OS) and post-recurrence OS of early-and late-recurrence patients were identified using univariate and multivariate Cox regression analyses. Of the 9,468 patients included, 409 (4.3%), 325 (3.4%), and 182(1.9%) developed purely local recurrence, purely regional recurrence, and locoregional recurrence during follow-up, respectively. In the purely local recurrence group, 192 patients (46.9%) developed early local recurrence (ETR), and 217 patients (53.1%) developed late local recurrence (LTR). Of the 192 ETR patients, multivariate Cox regression analysis revealed that age and gender were independent risk factors of OS, and post-recurrence best supportive treatment (PRBST) was associated with poorer post-recurrence OS. Of the 217 LTR patients, the results revealed that baseline value of EBV-DNA was an independent risk factor for OS, while PRBST was associated with poorer post-recurrence OS. In the purely regional recurrence group, 183 patients (56.3%) developed early regional recurrence (ENR), and 142 patients (43.7%) developed late regional recurrence (LNR). Of the 183 ENR patients, multivariate Cox regression analysis revealed that alcohol abuse and TNM stage were independent risk factors of OS, while alcohol drinkers and PRBST were associated with poorer post-recurrence OS. Of the 142 LNR patients, PRBST was associated with poorer post-recurrence OS. In the locoregional recurrence group, 87 patients (47.8%) developed early locoregional recurrence (ELR), and 95 patients (52.2%) developed late locoregional recurrence (LLR). Of the 87 ELR patients, multivariate Cox regression analysis revealed that N stage and INTRODUCTION Nasopharyngeal carcinoma (NPC), a malignant tumor that originates in the nasopharyngeal epithelium, is endemic in Southern China, Southeast Asia, North Africa, the Middle East, and Alaska (1,2). As a result of its complex anatomical location and high radiosensitivity, radiotherapy with or without chemotherapy is the primary treatment modality for NPC (3,4), and the application of IMRT has greatly improved locoregional control in NPC (5). However, the long-term prognosis remains unsatisfactory, given the high rate of locoregional recurrence of up to 5-10% in patients after definitive IMRT (6). This study focuses on the failure patterns of NPC except distant metastasis, which was separated clearly in three subgroups: (1) purely local recurrence (on the T site only), (2) purely regional recurrence (on the N site only), (3) locoregional recurrence (on the T and N sites simultaneously). Meanwhile, time to cancer recurrence differs in such patients, and the three subgroups were divided into ETR and LTR, ENR and LNR, and ELR and LLR, respectively, based on the time to recurrence after radiotherapy (7)(8)(9). To the best of our knowledge, research focusing on early and late recurrence in NPC patients remains rare and limited. Accordingly, we aimed to identify the clinical characteristics and prognostic factors of ETR and LTR, ENR and LNR, and ELR and LLR in a large cohort of patients with NPC who underwent long-term follow-up, providing data to clinicians for planning surveillance strategies.

Patient Selection
This study was performed according to the ethical principles of the Declaration of Helsinki, and the Sun Yat-sen University Cancer Center review board approved the study protocol. Written informed consent was obtained from all patients for their data to be used in clinical research without affecting their Abbreviations: NPC, nasopharyngeal carcinoma; IMRT, intensity-modulated radiation therapy; ETR, early purely local recurrence; LTR, late purely local recurrence; ENR, early purely regional recurrence; LNR, late purely regional recurrence; ELR, early locoregional recurrence; LLR, late locoregional recurrence; OS, overall survival; WHO, World Health Organization; TNM, tumor-nodemetastasis; MRI, magnetic resonance imaging; PET/CT, positron-emission tomography/computed tomography; UICC/AJCC, Union for International Cancer Control/American Joint Committee on Cancer; EBV, Epstein-Barr virus; HR, hazard ratio; CI, Confidence interval; RT, radiotherapy; Chemo, chemotherapy; PRBST, Post-recurrence best supportive treatment. treatment options or violating their privacy. We retrospectively reviewed the records of all 9,468 patients with biopsy-proven NPC who had been treated with IMRT at our center between January 2010 and December 2015. All patients had completed a pretreatment evaluation including complete patient history, physical examination, hematology and biochemistry profiles, nasopharynx and neck magnetic resonance imaging (MRI), chest radiography, abdominal ultrasonography, and wholebody bone scan or positron emission tomography/computed tomography (PET/CT). All patients were restaged according to the 8th Union for International Cancer Control/American Joint Committee on Cancer staging system (10,11). RT+Chemo was defined as treatment with both radiotherapy and chemotherapy, including induction chemotherapy and/or concurrent chemotherapy and/or adjuvant chemotherapy. Treatment options after recurrence were divided into four parts: salvage surgery, re-irradiation, chemotherapy, and best supportive treatment. During the study period, our institutional guidelines recommended no chemotherapy for stage I-IIA NPC, concurrent chemoradiation therapy for stage IIB NPC, and concurrent chemoradiation therapy with or without neoadjuvant/adjuvant chemotherapy for stage III to IVA-B NPC.

Follow-Up Schedule and Definition of ETR and LTR, ENR and LNR, and ELR and LLR
Patients attended follow-up visits every 3 months during the first 2 years, every 6 months during years 3-5, and annually thereafter or until death. Scheduled surveillance including fiberoptic endoscopy and head and neck CT/MRI scans was performed every 3 months during the first year and annually during years 2-5. Local recurrence was diagnosed by fiberoptic endoscopy and biopsy or nasopharynx and skull base CT/MRI scans. Regional recurrence was diagnosed by pathological examination with fineneedle aspiration or surgery or by radiology with neck CT/MRI scans. Additional tests were ordered whenever necessary. In this study, purely local recurrence was defined as recurrence on the T site only, which was divided into ETR and LTR according to time to NPC recurrence of ≤2 years and >2 years. Purely regional recurrence was defined as recurrence on the N site only, which was divided into ENR and LNR according to time to NPC recurrence of ≤2 years and >2 years, and locoregional recurrence was defined as recurrence on the T and N sites simultaneously, which was divided into ELR and LLR according to the time to NPC recurrence of ≤2 years and >2 years.

Statistical Analysis
The patients' clinical and pathological characteristics were summarized using frequencies and percentages for categorical covariates and medians and ranges for continuous covariates. The clinicopathological characteristics and treatment modalities among the patients with ETR and LTR, ENR and LNR, and ELR and LLR were compared using the chi-square test. The OS and post-recurrence OS were calculated with the Kaplan-Meier method, and differences between survival curves were assessed with the log-rank test. The prognostic factors of OS and post-recurrence OS of the patients with ETR and LTR, ENR and LNR, and ELR and LLR were evaluated using multivariate Cox regression analysis. P < 0.05 was considered significant. Statistical analyses were performed using SPSS version 23.0 (IBM).

Prognostic Factors Associated With OS
The prognostic factors contributing to long-term OS in ETR and LTR, ENR and LNR, and ELR and LLR were investigated using univariate and multivariate analyses (Tables 4-6).

Prognostic Factors Associated With Post-recurrence OS
The clinical factors and treatment modalities of post-recurrence OS in ETR and LTR, ENR and LNR, and ELR and LLR were elevated by univariate and multivariate analyses (Tables 7-9). Of the 192 patients with ETR, PRBST was significantly associated with poorer OS. Of the 217 patients with LTR patients, Cox regression modeling identified post-recurrence treatment options (P = 0.000) was an independent risk factor of postrecurrence OS. Of the 183 patients with ENR, Cox regression modeling predicted that alcohol abuse (HR, 3.750; 95% CI, 1.909-7.367; P = 0.000) and post-recurrence treatment options (P = 0.000) were independent risk factors of post-recurrence OS. Of the 142 patients with LNR patients, Cox regression modeling predicted post-recurrence treatment options (P = 0.000) was an independent risk factor of post-recurrence OS. Of the 87 patients with ELR, Cox regression modeling predicted that N stage (HR, 2.216; 95% CI, 1.225−4.009; P = 0.008) and post-recurrence treatment options (P = 0.000) were independent risk factors of post-recurrence OS. Of the 95 patients with LLR patients, Cox regression modeling predicted that T stage (HR, 4.111; 95% CI, 1.337-12.635; P = 0.014) and post-recurrence treatment options (P = 0.000) were independent risk factors of postrecurrence OS. Post-recurrence OS was not significantly different between ETR and LTR, ENR and LNR, and ELR and LLR groups (Figures 1B, 2B, 3B), with a median post-recurrence OS of 16.

DISCUSSION
Here, we investigated the clinical characteristics and prognostic factors predicting OS and post-recurrence OS in NPC patients with ETR and LTR, ENR and LNR, and ELR and LLR. In this retrospective study, 409 (4.3%) developed purely local recurrence, 325 (3.4%) developed purely regional recurrence,    stage was an independent risk factor for OS with LLR. In addition, multivariate Cox regression analysis revealed that postrecurrence treatment option was an independent risk factor of post-recurrence OS with ETR and LTR, while alcohol abuse and post-recurrence treatment option were independent risk factors of post-recurrence OS with ENR, and PRBST was associated with poorer post-recurrence OS with LNR. Meanwhile, N stage and post-recurrence treatment options were independent risk factors for post-recurrence OS with ELR, and T stage and postrecurrence treatment options were independent risk factors for post-recurrence OS with LLR. It has been suggested that patients with early initial T stage have a more favorable prognosis (16), which is consistent with the LLR patients in the present study.
Post-recurrence treatment options including salvage surgery, re-irradiation, and chemotherapy are very important for NPC recurrence patients, which was shown that post-recurrence treatment options mentioned above have a better prognosis compared with PRBST. There are various hypotheses for the occurrence of early and late recurrence. A probable hypothesis is the discrepancy of NPC tumor cell radiosensitivity. Recent studies have shown that apoptosis, DNA damage repair, a hypoxic microenvironment, and autophagy can be involved in regulating radiotherapy resistance (17)(18)(19), which results in the difference in early and late recurrence. A study from Japan shed light on the biological impact of DNA methylation status as a predictive biomarker of early recurrence in ovarian cancers (20). It has been suggested that the tumor dormancy-reactivation hypothesis might be applicable to NPC (21,22). Furthermore, surgery for removing breast tumors may lead to the appearance of growth factors in the circulation in response to surgical wounding, which may terminate the dormancy of the tumor foci and result in accelerated recurrence (7,23). Although the main treatment of NPC is radiotherapy instead of surgery, it may also give rise to the appearance and an increase of growth factors to result in the occurrence of recurrence. Therefore, it is reasonable to speculate that there might be intrinsic biological differences between patients with early and late recurrence, and this warrants further studies. The definitions of ETR and LTR, ENR and LNR, and ELR and LLR were applied with 2 years as the cut-off point, which have also been proposed and proven in recent studies (8,9,24). Hence, the frequency and intensity of follow-up should be strengthened at the initial 2 years. However, the application of 2 years for differentiating early recurrence from late recurrence remains controversial, and several studies have used 5 years as another major demarcation point (14,25,26). We applied 5 years for differentiating early recurrence from late recurrence and we found that there are very few cases of early recurrence; the number imbalance between the two groups is likely to lead to statistical problems.
There are two new features in the present study that differ from previous reports. First, this study figured out the prognostic factors of OS and post-recurrence OS in NPC patients with ETR and LTR, ENR and LNR, and ELR and LLR in detail. Second, this study has a large sample size with long median-time followup, which might help oncologists predict patients' prognosis and design individualized follow-up strategies. Although our study yielded some unique results, certain limitations should be noted. First, this is a retrospective single-center study, which has inherent biases. Second, the study merely explores the impact of baseline clinical characteristics on post-recurrence OS rather than post-recurrence clinical characteristics, which are principal elements of post-recurrence OS. Third, some information was lacking because of the long follow-up duration. However, we believe that the present results are noteworthy and reliable because this is the only such large-cohort study to date.

CONCLUSIONS
The present study shows that age and gender were independent risk factors of OS with ETR, and the baseline value of EBV-DNA was an independent risk factor of OS with LTR. Alcohol abuse and TNM stage were independent risk factors for OS with ENR, while N stage and TNM stage were independent risk factors of OS with ELR; and T stage was an independent risk factor for OS with LLR patients. In addition, post-recurrence treatment option was an independent risk factor of post-recurrence OS with ETR and LTR. Alcohol abuse and post-recurrence treatment option were independent risk factors of post-recurrence OS with ENR, and PRBST was associated with poorer post-recurrence OS with LNR. Meanwhile, N stage and post-recurrence treatment options were independent risk factors for post-recurrence OS with ELR, and T stage and post-recurrence treatment options were independent risk factors for post-recurrence OS with LLR. Patients with LTR/LNR/LLR demonstrate significantly better OS compared with patients with ETR/ENR/ELR, whereas post-recurrence OS is not significantly different between patients with ETR/ENR/ELR and LTR/LNR/LLR. Further studies are warranted to confirm our results.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
This study was performed according to the ethical principles of the Declaration of Helsinki, and the Sun Yat-sen University Cancer Center review board approved the study protocol. Written informed consent was obtained from all patients for their data to be used in clinical research without affecting their treatment options or violating their privacy.

AUTHOR CONTRIBUTIONS
FL, F-PC, Y-PC, and G-QZ conceived and designed the study. YC, X-JH, X-DH, and Z-QZ contributed cases data collection. W-HZ, XL, and YS analyzed the data. FL, F-PC, and G-QZ wrote the paper. All authors read and approved the final manuscript.