Background
Nasopharyngeal carcinoma (NPC) is malignant disease of multidimensional spatiotemporal unity of ecology and evolution, often manifesting with invasive growth at the primary site and metastatic cervical lymph node(s) [
1,
2]. Although the implementation of intensity-modulated radiotherapy (IMRT) has attained satisfactory tumor control and survival benefits in NPC [
3‐
5]. Intertumor and intratumor heterogeneities in the ecosystem of NPC have caused diverse response patterns among patients [
1].
Due to the different regression rate in tumor and disturbance of (chemo) radiotherapy-induced inflammation, 12 weeks after the completion of radiotherapy (RT) is proposed as a proper time point for initial evaluation of total tumor control [
6,
7]. By then, inflammation would have largely resolved, most tumors would have regressed, and delayed tumor regression within 12 weeks would not impair overall control [
8,
9]. Beyond this window time (12 weeks), the incidence of residue increased [
8]. Reportedly, 3–13% of patients had persistent disease after 12 weeks and were diagnosed as tumor residue [
10,
11]. Residual tumor stem cells have evolved aggressively and grow more rapidly during treatment cessation [
12]. Thus, most patients with residual disease ultimately develop disease failure [
10,
11].
Residue may be a reflection of treatment insensitivity, inadequate RT dose, or geographic miss in irradiated field [
12]. Timely additional treatment (including surgery, boost radiation, and chemotherapy) at the end of RT may enhance or strengthen the curative effect and reduce the probability of residue after 3–6 months [
13‐
15]. Controversially, the blind administration of additional intervention at the end of RT may be surplus for some patients whose tumor may spontaneous regressed. However, there is no sound tool at the end of RT to predict whether tumor will residue or not after observation for 3–6 months.
In NPC, Epstein–Barr virus (EBV) deoxyribonucleic acid (DNA) serves as a liquid circulating biomarker that reflects the tumor burden [
16]. Reportedly, postradiotherapy EBV DNA has an even stronger association with recurrences and poor prognosis in NPC [
17,
18]. Elevated plasma EBV DNA level has been shown to predate clinical recurrence by 3 to 7 months, which may present a biomarker of subclinical residual disease [
18,
19]. But its predictive value on residue is unexcavated.
Here, we hypothesized that the probability of tumor residues could be reduced through immediate additional intervention at the end of IMRT by leveraging a predictive tool. In this study, we aimed to develop and validate a nomogram model integrating clinical characteristics at the end of IMRT for predicting whether tumor will residue or not after 3–6 months. Thus, NPC patients with a high risk of residue event might benefit from immediate additional intervention. Meanwhile, low-risk patients who will have tumor complete regressed after 3 months will be spared from overtreatment.
Methods
Patients
Between January 2012 and December 2017, patients at Sun Yat-sen University Cancer Center who fulfilled the inclusion criteria were enrolled. The inclusion criteria were as follows: (1) histologically confirmed non-metastatic NPC without previous or concurrent malignant disease; (2) age ≥ 18 years; (3) receipt of curative IMRT for the entire course without interruption; (4) available information on pretreatment and postradiotherapy (-7 to +28 days) plasma EBV DNA levels; (5) regular follow-up with complete posttreatment examination, including nasopharyngoscopy, magnetic resonance imaging (MRI) of the nasopharynx and neck, etc., until the first documentation of disease failure or death; and (6) no evidence of distant metastasis on chest scan (x-ray or computed tomography [CT]), liver scan (external ultrasonography or CT), bone scan, or 18F-fluorodeoxyglucose positron emission tomography-computed tomography (18F-FDG PET-CT) up to 6 months postradiotherapy.
A total of 1080 patients were included in this study. All patients were restaged according to the 8
th edition of the International Union for Cancer Control/American Joint Committee on Cancer staging system [
20]. Because patients with stage I disease obtained extremely satisfactory disease control, adjuvant therapy is not recommended for patients with stage I disease [
21‐
23]. Also, none of the 30 patients with stage I disease have residual disease. Therefore, patients with stage I disease was excluded from this study. Eventually, 1050 patients with stage II–IVA NPC were enrolled. The Institutional Review Board of Sun Yat-sen University Cancer Center approved this study (B2021-215–01).
Plasma EBV DNA measurement
Plasma EBV DNA levels were measured using a quantitative polymerase chain reaction assay targeting the BamHI-W region of the EBV genome [
24,
25]. The EBV DNA level was measured within 28 days before treatment (pretreatment) and -7 to +28 days after completion of IMRT (postradiotherapy).
Diagnostic criteria of residue
Residue is defined as the confirmation of disease occurring within 6 months after treatment. When disease is found after these 6 months, provided that previous complete remission was seen, it is defined as recurrence [
26]. Residue was firstly found by physical examination, nasopharyngoscopy or imaging modality (e.g., MRI, CT). By histopathological or cytological biopsy, or pathological examination of surgery-resected lesion, the lesion contained any component with cancerous cells or tissue originating from the primary NPC is diagnosed as residue. For lesions that were not accessible (e.g., skull base, intracranial residue), the diagnoses were made based on 18F-FDG PET/CT in consensus by two nuclear medicine physicians (each with 5 years of experience in PET/CT) using a GE Xeleris workstation. Image interpretation was based on visual evaluations. Any focus of FDG uptake greater than the surrounding background and not attributable to normal FDG biodistribution was assessed. The intensity of FDG uptake was graded using the five-point scale proposed by Ng et al. [
27,
28]. The probability of residual tumor was graded by using a five-point scoring system (0 = no lesion, 1 = definitely benign, 2 = probably benign, 3 = probably malignant, and 4 = definitely malignant). Grade 3 and grade 4 were both considered to be positive results. The types of residue were as follows: local residue: residue in the nasopharynx, and/or extension to the oropharynx, nasal cavity, parapharyngeal space, adjacent soft tissue, and/or infiltration of bony structures at the skull base, cervical vertebra, pterygoid structures, paranasal sinuses, and intracranial extension; regional residue: residual tumor in retropharyngeal lymph nodes and/or cervical lymph nodes; and locoregional residue: both local and regional residues.
Therapeutic regimens
All patients received radical IMRT as the primary treatment. Details regarding IMRT techniques have been described in previous studies [
29]. Target volume delineation was performed according to the International Commission Radiological Units guidelines. Doses to critical normal structures and plan evaluations were directed according to the Radiation Therapy Oncology Group guidelines. The prescribed dose of 68.00–74.00 Gy was delivered in 30–33 fractions for the primary tumor and any involved lymph nodes. Platinum-based chemotherapy, including concurrent chemoradiotherapy every 3 weeks for 2–3 cycles or weekly for 6–7 cycles, and neoadjuvant chemotherapy every 3 weeks for 2–4 cycles, were implemented at the physician’s discretion depending on the patient’s physical status and disease stage.
Follow-up
Patients were assessed every 3 months during the first 3 years, every 6 months during the next 2 years, and annually thereafter. The last follow-up date was 31 May 2022. From the start of treatment to the date of death from any cause, first occurrence of treatment failure or death, first occurrence of locoregional failure or death, and first remote failure or death were measured as overall survival (OS), progression-free survival (PFS), locoregional relapse-free survival (LRRFS), and distant metastasis-free survival (DMFS), respectively.
Statistical analyses
Categorical variables were presented as frequencies and percentages. Continuous variables were described using the median and interquartile range. Pearson’s chi-square test or Fisher’s exact test were used to assess categorical variables between the groups. Differences in non-normally distributed variables between the groups were examined using the Mann–Whitney
U test. Survival rates were calculated using the Kaplan–Meier method and compared using the log-rank test. Univariate and multivariate analyses with the Cox proportional hazards model were used to identify significant independent prognostic factors using forward elimination. The optimal threshold analysis of the pretreatment and postradiotherapy EBV DNA levels in predicting residue was conducted based on a receiver operating characteristic (ROC) curve. Logistic regression analysis was performed to identify variables associated with the residue. Variables achieving a significance level of
P<0.05 in the univariate analysis were subjected to multivariate logistic regression analysis. The discriminating ability of a model was described by the area under the curve (AUC), and the calibration was evaluated by constructing a calibration curve. The clinical usefulness of the nomogram was evaluated using decision curve analysis (DCA). The performance of the models was evaluated by calculating the AUC, which was calculated and compared using the method suggested by Delong et al. [
30] using MedCalc 19.6.4 (MedCalc Software Ltd.). The optimal cut-off value of the points predicted by the nomogram was selected based on the ROC curve in the development cohort, and the same cut-off value was applied to the validation cohort. All analyses were performed using SPSS software (version 24.0; IBM Corp., Chicago, IL, USA) and R-4.2.0 (
http://www.R-project.org/). A two-sided
P<0.05 was considered statistically significant.
Discussion
Despite the advancement of IMRT, 3–13% of patients with NPC experienced tumor residue 3–6 months after radical RT [
10,
11]. Over the past two decades, considerable efforts have been made to investigate the prognostic value of [
31‐
34], compare the capability to diagnose and differentiate residues among available medical procedures [
35,
36], and develop predictive models for prognostic stratification and risk adjustment [
32] in this field. Unfortunately, little is known about the forecasting tumor residue for the preventive effects at the end of RT. In a large retrospective cohort, we developed and validated a nomogram, integrating clinical stage, postradiotherapy plasma EBV DNA, and RT dose, to estimate the probability of tumor residue after 3–6 months in stage II–IVA NPC patients treated with curative IMRT. The comprehensive nomogram showed better discrimination than clinical stage or postradiotherapy EBV DNA level alone. To the best of our knowledge, this study is the first to investigate the role of postradiotherapy status in predicting residue after 3–6 months.
The tumor-derived EBV DNA load in the plasma represents a microscopic disease in NPC [
3]. A detectable or higher level of this pretreatment liquid biomarker is associated with worse outcomes and inferior survival [
37]. While it can be eradicated during treatment in most patients, 9.2–28.8% of patients have microscopic tumor residue in the circulation after treatment [
18,
38,
39]. Chan et al. conducted several prospective studies to investigate their prognostic significance. In 170 NPC patients receiving a uniform RT protocol, 28.8% of patients had detectable posttreatment EBV DNA (at 6–8 weeks after RT) and exhibited inferior PFS (HR 11.9, 95% CI 5.53–25.43) and OS (HR 8.6, 95% CI 3.69–19.97) compared to patients with higher pretreatment EBV DNA [
18]. Similarly, a better risk discrimination concerning different endpoints in the postradiotherapy EBV DNA level than pretreatment level was observed in our study. In a large prospective plasma EBV DNA screening study for identification of high-risk NPC patients for adjuvant chemotherapy, the positive relationship between detectable or higher levels of postradiotherapy EBV DNA (within 120 days) and disease failures was highlighted. Unfortunately, patients with detectable postradiotherapy liquid biomarker levels did not benefit from adjuvant chemotherapy [
38]. Therefore, detectable postradiotherapy EBV DNA levels are not a determinant factor in identifying high-risk patients. In their post-hoc analysis, a combination of postradiotherapy EBV DNA and clinical stage improved risk stratification for NPC using recursive partitioning analysis compared to either the clinical stage or postradiotherapy EBV DNA alone (concordance-index (C-index): 0.712 vs. 0.604 vs. 0.675) [
39]. The aforementioned series of investigations implied the potential of a combination of clinical stage and postradiotherapy EBV DNA level in predicting tumor residue. In our study, plasma EBV DNA was detectable in 10.6% of patients after RT, consistent with a previous study [
39]. In Chan’s study, a postradiotherapy EBV DNA level yielded a C-index of 0.675 in predicting 5-year OS [
39]. Tumor residue is an early failure pattern that can be plausibly predicted by postradiotherapy EBV DNA which has been related to minimal residual disease at the end of RT. In our study, the postradiotherapy EBV DNA level alone achieved a higher AUC in predicting residue compared to the pretreatment EBV DNA level (0.627 vs. 0.593). The performance in predicting tumor residue was improved by a combination of clinical stage and posttreatment liquid biomarkers (AUC: 0.733), which verified the aforementioned hypothesis.
In this study, 69.96 Gy was employed as a cut-off value for sectionalisation for the RT dose. The reason for choosing this prescribed dose was that it was uniformly recommended to all NPC patients for its good balance between the tumor-killing effect and normal tissue tolerance according to the National Comprehensive Cancer Network. Notably, the RT dose to the primary tumor site and involved lymph nodes was significantly associated with tumor residue in our model. This result demonstrated that the tumor residue is associated with tumor radiation insensitivity or an insufficient dose. This is likely a reflection of the underlying biological heterogeneity of patients with NPC. We reasoned that a uniformly prescribed dose may be insufficient for patients with potential radiation resistance. Several studies have shown that boost RT dose for residual lesions can improve tumor control and survival rates [
40,
41]. Fei et al. conducted a study on 398 NPC patients with T4 stage disease who had local residue after radical IMRT (70 Gy). In their study, 114 patients received boost dose of 4–6.75 Gy in local residual lesions (2–3 fractions, 2–2.25 Gy per fraction, 1 fraction per day, 5 fractions per week). After follow-up, the boost group exhibited better 3-year OS (86.6% vs. 71.3%,
P=0.008), PFS (79.0% vs. 69.1%,
P=0.019), and LRRFS (93.4% vs. 82.4%,
P=0.012) than the non-boost group [
14]. In our study, patients receiving a dose under 70.00 Gy were more likely to have residue, which implies that a dose boost may reverse the outcome. Meanwhile, others argue that the boost dose should not be delivered indiscreetly if the delivered dose for the gross tumor volume is sufficient. In a retrospective study by Han et al., the tumor residual rate was 6.1% (12/196) three months after IMRT. All 12 residual lesions resolved completely 4–9 months after RT [
42]. In the era of IMRT, the blind administration of additional RT to the residual tumor seems unwise. Thus, the challenge lies in patient selection to maximise the magnitude of the benefit. Although the prediction nomogram built in our study cannot predict residual tumor pretherapy, we provided a compromise at the end of IMRT. After receiving a uniform standard dose, timely salvage treatment is necessary for patients at risk of tumor residue.
However, this study has some limitations. First, it was a retrospective study. Second, the timing of postradiotherapy plasma EBV DNA levels was not standardised within the studied patients. In most clinical trials of adjuvant therapy for NPC, the therapy is usually planned to start within 4 weeks after the completion of RT. To cover this range, we extended the time of measuring plasma EBV DNA levels to 4 weeks after RT. Third, our prediction nomogram was built and validated at a single center. Further external validation will help to attain high-level evidence of clinical feasibility. Also, prospective clinical studies with large cohorts are warranted to investigate strategies for tumor residue prevention.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.