Survival outcomes and prognostic factors for ascending nasopharyngeal carcinoma (T4N0-1) receiving radiation therapy combined with chemotherapy: a population-based study from SEER database.

Purpose This study aims to investigate survival outcomes and prognostic factors for upward nasopharyngeal carcinoma (NPC) patients receiving radiation therapy (RT) combined with chemotherapy (CT). Methods A total of 421 previously untreated, newly diagnosed T4N0-1 NPC patients, who were identified within the Surveillance, Epidemiology, and End Results (SEER) registry (years 2004–2015), were collected and retrospectively reviewed. All patients received treatment of RT and CT. Kaplan-Meier analysis was used to evaluate overall survival (OS) and cancer-specific survival (CSS). The differences in OS and CSS were compared using Log-rank test. The independent prognostic factors were established by using univariate and multivariate Cox proportional hazard models. With a the estimate and CSS rates were and respectively. and were poor prognostic factors for OS and CSS. Moreover, and race associated with OS and CSS. Univariate analysis indicated that ≥ 65 years, N0, NHB and grade III were unfavorable independent prognosticators of OS and CSS. Multivariate analysis demonstrated that ≥ 65 years, N0 and NHB were correlated with poor OS and CSS.


Introduction
Nasopharyngeal cancer (NPC) is a unique malignant of head and neck, with the cute incidence of 15-50 cases per 100,000 annually in the endemic area such as southeast Asia, northern Africa and middle Europe [1] . The GLOBOCAN data [2] in 2018 reported that there are 129097 patients with newly diagnosed NPC around the world. 47.7% of these patients occurred in China. Furthermore, the rate of loco-regionally advanced NPC (stage III or IV) accounted for about 70% [3] . While 27% of newly diagnosed NPC patients presented with non-metastatic stage IV and the corresponding overall survival (OS) at 5 years was only about 65% [4] . Although non-metastatic stage IV in the UICC 8th edition included both T4 and N3 [5] , there were three types of disease according to distinct clinical behavior and patterns of failure [6][7][8] . Three types of disease were described as follows: 1) ascending type: locally advanced disease with limited nodal spread (T4N0-1); 2) descending type: extensive metastatic lymph nodes and limited primary tumor (T1-2N3); 3) hybrid type: loco-regionally advanced disease (T3-4N2-3). For the patients with stage T4 NPC, local relapse was the main failure pattern, while distant metastasis was the predominant pattern of failure in stage N3 NPC patients. Mao et al reported that the local failure-free survival (LFFS) at 5 years was 76% for stage T4 patients and the distant metastasis-free survival (DMFS) at 5 years was 66% for stage N3 patients [4] . Huang et al investigated the survival and failure patterns of 3107 non-metastatic stage IV NPC patients and obtained the similar results [9] . Yao et al showed that compared with ascending types, descending types had poor survival and aggressive clinical course [8] . Given these situations, the treatment and survival outcomes were different among these three types of diseases. So, the aim of the current study was to investigate the survival outcomes and prognostic factors of the stage T4N0-1 NPC patient without distant metastasis.

Materials And Methods Database and patients' selection
All histology-proven patients with NPC were identified in the Surveillance, Epidemiology, and End Results (SEER) database of the national Cancer Institute in the American. The SEER 18 database [10] was obtained in the SEER program with SEEER*Stat software, version 8.3.6 (www.seer.cancer.gov/seerstat). We selected the patients using the criteria as following: the patients were selected if age at diagnoses older than 20 years and diagnosed with NPC as the first malignancy from 2004 to 2015. And we used the American Joint Committee on Cancer 6th or 7th edition staging to define the patients' stage. Stage N2-3 or distant metastases (M1) were exclusions. The therapeutic scheme for the NPC patients with T4N0-1 was the treatment of radiotherapy (RT) and/or chemotherapy (CT). Subjects received no treatment were excluded.
The flowchart of selective patients is shown in Fig. 1. A total of 1124 histology-confirmed NPC patients were registered in SEER database. A total of 421 NPC patients with T4N0-1 received RT plus CT were respectively reviewed. The study was exempt from our Institution Review Board because the information of all patients was publicly available. The information collected prospectively included patient demographics, histology, stage, therapeutic strategy, overall survival (OS) and cancer-specific survival (CSS).

Statistical analyses
All other data was analyzed by using IBM SPSS Statistical software, version 25.0. We used vital status and follow-up time from diagnosis date to calculated OS. And we used the cancer-specific death classification to compute CSS. The Kaplan-Meier method was used to generate survival curves and log-rank tests were used compare the curves of different variables. We used a Cox regression model to conduct the univariate and multivariate in order to identify significant prognosticators. And we calculated hazard ratios (HRs) and 95% confidence intervals (CIs) for each prognosticator. If a P value < 0.05, the differences were considered as statistically significant.

Patients' characteristics
There were 421 patients with a mediate age of 55 years (range 20-86 years). Of these patients, the rate of male was 69.1%, non-Hispanic White was 40.6%, and < 65 years was 79.6%. All patients treated with RT and CT. Table 1 summarized the cohort characteristics.

Survival analyses
The median survival time was 37 months (range, 3 to 154 months). The estimated OS and CSS rates at 5 years were 59.3% and 63.7%, respectively (Fig.2). The estimated rates OS at 5 years were 67.5% and 43.5% for patients with aged <65 years and ≥ 65 years, respectively (P < 0.001, Figure   3A). The estimated 5-year CSS rates were 67.5% and 43.7% for patients with aged <65 years and ≥ 65 years, respectively (P < 0.001, Figure 3B). Interestingly, the estimated rates OS at 5 years were 49.7% and 66.3% for patients with grade N0 and N1, respectively (P<0.001, Figure 3C), while the estimated 5-year CSS rates were 54.1% and 70.5% for patients with grade N0 and N1, respectively (P=0.001, Figure 3D).

Identification of prognosticators
Several potential prognosticators including age, gender, histology, race, and N stage. We used logrank test to evaluate these prognostic factors, and the results showed that age, histology, and N stage were significant prognosticators for OS and CSS (Table 2). Interestingly, race was associated with OS and CSS. The stage T4N0-1 NPC patients had the lowest local control and relatively low distant metastasis. The 5 years local control rate was only 70-80% [4] . Local relapse is an important factor affecting the efficacy of ascending NPC. If the patients experienced local relapse, the OS rate at 5 years was about 37-41%, and 51% of these patients experienced radiation injury including nasopharyngeal necrosis and hemorrhage, which reduced the quality of life of the patients [13,14] . So, it is crucial to improve local control of these patients. The reasons of local relapse included clinical and biological factors.
Clinical factor meant that insufficient RT dose of tumor resulted in local recurrence, while biological factors referred as resistance to RT for tumor cells. Ng and colleagues evaluated the effect of dosimetric inadequacy in target volumes on local control of NPC, the results indicated that if the volume below 66.5 Gy in gross tumor volume (GTV) was more than 3 cc, the 5-year rate of local failure-free dropped to 54% [15] . Due to the tumor in close to organ at risk (OAR), adaptive RT could be recommended to improve local control, but it needed randomized trial to validate therapeutic effect. However, hypoxia led to RT resistance [16] , while hyperbaric oxygen treatment [17] , red blood cell transfusion and erythropoietin delivery before or during the period of RT increased tumor oxygenation.
Although the ascending NPC belonged to locoregionally advanced NPC, it had different therapy option and survival patterns [6,7,18] . Thus, the optimal treatment strategy and survival outcomes of these patients need be investigated. Xiao et al performed the retrospective study of 148 T4 NPC patients treated with CCRT and the results indicated that the PFS and OS rates at 4 years were 46.9% and 75%, respectively [19] . Chen et al demonstrated that for the T4 NPC patients receiving CCRT, the OS, DMFS, LRRFS and PFS rates at 4 years were 78.1%, 72.2%, 81.2% and 61.9%, respectively [20] . Yao et al showed that IC + RT, compared to IC + CCRT or CCRT, was an encouraging treatment option for ascending NPC [21] . Moreover, Yao et al indicated that IC + RT provided similar survival and fewer complications for ascending NPC [22] . In addition, Zhang et al showed that CCRT combined nimotuzumab obtained a favorable local control and acceptable side events [23] .
Although SEER database provides a public available data to investigate this clinical problem, several limitations were observed in this study. Firstly, treatment information including RT dosing, CT regimen, delays in CT and therapy dates was not registered into SEER database, and we did not analyze the role of these factors. Secondly, therapy-related complications were not evaluated in this study due to lack of the information about RT-and CT-related side events. In addition, SEER records did not include the information about locoregional relapse, distant metastasis, so we failed to assess LRRFS and DMFS.  weighted multivariable Cox proportional hazards regression analyses are shown for overall survival and cancer-specific survival