Identifying optimal candidates for induction chemotherapy among stage II–IVa nasopharyngeal carcinoma based on pretreatment Epstein–Barr virus DNA and nodal maximal standard uptake values of [18F]‐fluorodeoxyglucose positron emission tomography

Abstract Objective This study aimed to select optimal candidates benefiting from the addition of induction chemotherapy (IC) to concurrent chemoradiotherapy (CCRT) in stage II–IVa nasopharyngeal carcinoma (NPC) based on Epstein–Barr virus (EBV) DNA and nodal maximal standardized uptake values (SUVmax‐N) of [18F]‐fluorodeoxyglucose positron emission tomography. Patients and materials A total of 679 patients diagnosed with stage II–IVa (except N0) NPC were retrospectively included in this study. Overall survival was the primary endpoint. Survival differences between different groups were compared using the log‐rank test. The hazard ratio (HR) and 95% confidence interval (CI) were calculated using a multivariable Cox proportional hazards model. Results Both high levels of EBV DNA (>1500 copies/mL) and SUVmax‐N (>12.3) indicated worse survival conditions. All patients were divided into low‐ and high‐risk groups based on these two biomarkers. The risk group was an independent prognostic factor in OS, progression‐free survival (PFS), and distant metastasis‐free survival (DMFS) (all p‐values<0.05). The addition of IC to CCRT was associated with survival improvement in OS, PFS, and DMFS in high‐risk patients, while no survival difference was found between CCRT and IC+CCRT in low‐risk patients. Conclusions Our study can help clinicians select stage II–IVa NPC patients who benefit from IC, which is important in guiding individual treatment.


| INTRODUCTION
Nasopharyngeal carcinoma (NPC) is a unique malignancy that differs from other head and neck cancer in terms of geographical distribution, treatment method, and prognosis. 1 In 2018, approximately 12,900 new cases of NPC were reported, accounting for 0.7% of all cancers, among which greater than 70% cases occurred in east and southeast Asia. 2 Radiotherapy (RT) is the mainstay treatment modality for NPC, as it is highly sensitive to radiation. 3 For locoregionally advanced NPC (LANPC), previous studies have demonstrated the superiority of concurrent chemoradiotherapy (CCRT) in survival benefit compared with RT alone. [4][5][6] Recently, an increasing number of studies have investigated the role of induction chemotherapy (IC). For LANPC, randomized Phase III trials demonstrated that the addition of IC before standard CCRT was associated with survival improvement. 7,8 However, this aggressive treatment method is also accompanied by more serious toxicity. 7 A previous study verified that TNM stage is insufficient to reflect the tumor burden and predict survival accurately for NPC patients. 9 Therefore, it is necessary to combine the extent of tumor invasion and effective biomarkers to divide patients into different risk levels and identify optimal candidates for the IC treatment.
Plasma Epstein-Barr virus (EBV) DNA levels are useful in the detection, monitoring, and prognostic prediction of both nonmetastatic and metastatic NPC and are regarded as the most important biomarker. 10 Previous studies suggested that the maximal standardized uptake value (SUVmax) of [18F]-fluorodeoxyglucose positron emission tomography (PET) of cervical lymph node (SUVmax-N) was also an important prognostic factor in NPC. 11,12 According to NCCN guidelines, IC was recommended for stage II-IVa NPC. 13 In this study, we explored the prognostic value of EBV DNA and SUVmax-N among stage II-IVa patients and then investigated the role of IC in patients with different risk levels with the aim of selecting patients who can truly benefit from IC.
Clinical research ethics committee at SYSUCC approved this study.

| Diagnosis and treatment
All the patients underwent complete pretreatment evaluations at admission and were treated based on the principles of SYSUCC. All the patients received the examination of plasma EBV DNA level. The detailed information on method of plasma EBV DNA quantification and treatment is available in Supplementary Materials.

| Follow-up
Patients were examined every three months for two years after treatment, and then every six months until death. The routine follow-up evaluation included physical examination, nasopharyngeal fiber optic endoscopy, MRI/CT of the head and neck, chest radiography/CT, abnormal ultrasound/ CT, bone scan, and PET/CT if necessary. The primary endpoint of our study was overall survival (OS), which was defined as the time from the date of diagnosis to the date of death from any cause. The following survival outcomes were secondary endpoints: progression-free survival (PFS) was calculated from the date of diagnosis to the date of disease progression or death for any reason; relapse-free survival (LRFS) and distant metastasis-free survival (DMFS) were calculated from the date of diagnosis to the date of locoregional failure and distant metastasis, respectively.

| Statistical analyses
The detailed information of statistical analyses is available in Supplementary Materials.

| Patients' characteristic and survival
From 2008 to 2013, 679 stage II-IVa NPC patients were retrospectively involved in this study. The median age of our cohort was 46 years; the male to female ratio was 3.5:1. EBV DNA and SUVmax-N were transferred to categorical variables for further analysis and were derived from the published cut-off point (1500 copies/mL) and ROC curve, respectively. The median concentration of EBV DNA was 3.55 × 10 3 copies/mL (range 0-6.9 × 10 6 copies/mL). A total of 420 patients had EBV DNA levels greater than 1500 copies/mL. The SUVmax cut-off point was 12.3 for OS (AUC [area under the curve] = 0.607, p = 0.006) ( Figure  S1). There were 421 patients (62.0%) assigned to the lower SUVmax-N group (≤12.3), and 258 patients (38.0%) had a greater SUVmax-N value. The patients' characteristics are presented in Table 1.

| Risk stratification
As shown in Figure S2, patients with higher pre-EBV DNA levels suffered worse survival conditions in all endpoints (all p-values<0.05). Similarly, a higher SUVmax-N (>12.3) was also associated with significantly lower OS, PFS, LRFS, and DMFS ( Figure S3). Based on these two prognostic biomarkers, we subdivided all patients into four subgroups: group A, SUVmax-N ≤ 12. (all p-values>0.05). In terms of LRFS, a significant difference was only observed between group A and group D ( Figure 1). Therefore, we combined groups A-C into the low-risk group, and group D served as the high-risk group. There were a total of 469 and 210 patients in each subgroup, respectively.

| Treatment outcome for patients in different risk groups
Before radical CCRT, a total of 313 patients (46.1%) received IC, with 164 patients receiving two cycles, 111 and 38 patients receiving three cycles, and four cycles, respectively. The number of patients in each IC regimen was 157 (cisplatin plus docetaxel plus 5-fluorouracil), 68 (cisplatin plus 5-fluorouracil), 54 (cisplatin plus docetaxel), and 34 (cisplatin plus gemcitabine), respectively (Table 1). In multivariate analysis, the application of IC was associated with a lower risk of distant metastasis (p = 0.046) ( Table 2). However, there was no association between different IC regimens and survival outcomes in all endpoints ( Figure S4). The addition of IC increased the risk of grade 3-4 leukopenia (32.8 vs 16.3%; p < 0.001) and neutropenia (39.5 vs 11.6%; p < 0.001). Details of other acute toxicity by treatment groups are presented in Table S2.
We further analyzed the effect of different treatment methods (IC+CCRT or CCRT alone) on prognosis among patients with different risk levels. The characteristics of patients in different risk subgroups are shown in Table S1. Interestingly, we found that the role of IC was different between low-and high-risk groups. In the low-risk group, patients treated with IC plus CCRT had similar survival outcomes for OS, PFS, LRFS, and DMFS compared with those treated with CCRT alone (all p-values>0.05) (     (Tables 3 and  4).

| DISCUSSION
In the current study, we retrospectively examined 679 stage II-IVa (except N0) NPC patients who underwent PET-CT examinations and verified the prognostic value of SUVmax-N and EBV DNA. According to these two biomarkers, we divided patients into low-and high-risk groups. Comparing the survival conditions between patients receiving IC+CCRT and CCRT alone, we found that the addition of IC could only benefit patients in the highrisk group, whereas the benefit was not consistent in the low-risk group. For locoregionally advanced NPC, CCRT is established as the standard treatment method according to previous clinical trials. [4][5][6] Recently, an increasing number of scholars have paid attention to the treatment value of IC in LANPC. Based on the results of a recent phase III trial, Ma et al. and colleagues verified that the application of IC before CCRT could further improve DMFS and OS compared with CCRT alone for III-IV LANPC (except T3-4 N0). 7,8 In addition, the results of a network meta-analysis, which concluded that the addition of IC to CCRT achieved the highest effect on distant control, also affirmed the curative effect of IC. 14 However, the use of induction chemotherapy also increased the treatment-related toxicity and the financial burden. 7,8 Thus, it is necessary to identify proper candidates who could benefit from the addition of IC. Currently, the selection of patients suitable for IC mainly depends on TNM stage, which indicates the anatomical extent of tumor. According to NCCN guidelines, all stage II-IVa NPC patients were eligible to receive IC. 13 However, we should note that cancer is an individual disease with biological heterogeneity. 9 Thus, it is not rigorous to decide patients who might benefit from IC exclusively based on TNM stage.
Plasma EBV DNA level represents an important biomarker for the clinical management of NPC. 10,15-18 Lin et al. reported that patients with pre-EBV DNA levels of greater than 1500 copies/mL faced a higher risk of treatment failure compared with other patients. 10 Consistent with this study, we choose 1500 copies/mL as the cutoff value of EBV DNA. With the development of imaging technology, PET-CT has been increasingly applied in the diagnosis of NPC, especially in the early detection of distant lesions. 11,19 In terms of the prognostic value of SUVmax, previous studies showed that SUVmax-N was an independent prognostic factor both in LANPC and metastatic NPC, while the SUVmax of the primary tumor was not. 11,12 To select patients with higher tumor burden, which indicates that more aggressive treatment is necessary, we combined these two biomarkers to divide patients into different risk groups and investigated the role of IC among them.
The current study confirmed the results of past studies that demonstrated that SUVmax-N and EBV DNA in PET are prognostic factors for NPC patients. Based on these two factors, all patients were divided into four groups. After pairwise survival comparisons, patients with SUVmax-N > 12.3 and EBV DNA > 1500 copies/mL were identified as high-risk patients. When the role of IC was investigated in patients with different risk levels, we found that only high-risk patients could benefit from the addition of IC. Although some researchers tried to identify the LANPC beneficial from IC therapy based on different prognostic factors, 20 stage II-IVa NPC, clinicians can perform risk stratification of these patients at admission based on their SUVmax-N and EBV DNA levels and then screen high-risk patients for IC before radiotherapy. Low-risk subgroups need further investigations in future IC-related studies. The reason for the different impacts of IC on patient outcomes could be explained as follows. Higher EBV DNA and SUVmax levels are associated with tumor burden, which indicates a higher risk of distant metastasis after treatment. 9,22 Thus, a more aggressive treatment was necessary to reduce tumor burden, which helps to further improve distant control and subsequent survival. In addition, patients with high-risk levels might develop micrometastases at diagnosis that could not be detected by existing imaging technology. The administration of IC is intended to eradicate micrometastases earlier and may be associated with more potential benefit for high-risk patients.
There are several limitations to the current study. First, this is a retrospective study, and inherent selective bias was unavoidable. Second, there were physician biases contributing to treatment heterogeneity in IC application, regimen and duration of IC. Third, all patients involved in this study were from one treatment center, and WHO III was the main pathological type. Finally, no consensus was reached on the measure of FDG uptake. Given that SUVmax-N is associated with image noise, the SUVmax-N cutoff value in current study should be validated in another center.

| CONCLUSION
Our study showed that EBV-DNA and SUVmax-N could be used as biomarkers for risk stratification of NPC patients, and more importantly, guide physicians to select those benefiting from IC. Our results provide important information for individualized NPC treatment.