Introduction

Nasopharyngeal carcinoma (NPC) is a common cancer of head and neck malignancies, prevalent in East and Southeast Asia1. NPC is sensitive to radiotherapy, and the 5-year overall survival rate has improved by up to 80% in the age of intensity-modulated radiotherapy (IMRT)2. However, approximately 5–15% of patients will finally experience local recurrence after primary definitive therapy3. Surgery is preferred for patients with resectable recurrence4. While for patients who are not suitable for surgery or refuse surgery, re-irradiation seems to be the optimal curative method5. Due to complicated problems, such as severe late toxicity and intrinsic radiotherapy resistance, it is a high challenge for oncologists to make this trade-off.

Currently, high-level evidence-based medical research on re-irradiation is lacking. Controversies, including the definition of clinical target volume (CTV), addition of systemic therapy, choice of radiotherapy mode, and protection of normal tissue, need to be addressed6,7,8. International Recommendations on Re-irradiation by IMRT for locally recurrent NPC emerge as The Times requires9. However, the balance between benefits and risks and the patient’s acceptance of re-irradiation varies in clinical practice. In the last few decades, studies have focused on failure patterns after the first course of radiotherapy in NPC10,11. In addition, prognostic factors and models for outcomes have also been explored to guide the treatment of recurrent NPC patients12,13,14. However, reports involved in the second failure pattern after re-irradiation remain unclear. Therefore, it is imperative to explore a simple and feasible tool to guide oncologists' clinical decision-making in managing locally recurrent NPC.

In this study, we aimed to investigate the local failure patterns of recurrent NPC after re-irradiation by IMRT and find individualized treatment models based on different prognostic risks.

Materials and methods

Patient characteristics

Ninety-seven patients with locally recurrent NPC at the Fujian Cancer Hospital between July 2014 and December 2018 were retrospectively collected. The inclusion criteria were as follows: (1) rT1-4N0-3 NPC patients with histological confirmation after definitive radiotherapy for more than 6 months; (2) pathological evidence of nasopharyngeal and lymph node recurrence and imaging evidence of skull base recurrence (imageological examination for disease at sites inaccessible to biopsy); (3) without distant metastasis both at initial and recurrence treatment; (4) Not suitable for surgery according to muti-disciplinary treatment or refuse surgery; (5) comprehensive treatment with IMRT or tomotherapy. The exclusion criteria were as follows: (1) with other types of malignant tumors; (2) receipt of surgery; (3) inability to cooperate with treatment due to other diseases, such as uncontrolled mental illness or infectious diseases; (4) loss to follow-up. This study was approved by the Ethics Committee of Fujian Cancer Hospital (Ethics approval number: K2022-192-01). Moreover, all patients provided informed consent when the database was constructed. Furthermore, all methods were conducted by the appropriate guidelines and regulations stipulated in the Declaration of Helsinki. The detailed clinical data of patients are shown in Table 1. All patients were re-staged according to the 8th Edition of the American Joint Committee on Cancer (AJCC)/the Union for International Cancer Control (UICC).

Table 1 Characteristics of 97 locally recurrent NPC patients with delineation of rGTV only or rCTV.
Full size table

Treatment

For re-irradiation, IMRT was the only radiotherapy method. The gross tumor volume (GTV) and clinical target volume (CTV) were delineated by computed tomography/magnetic resonance imaging (CT/MRI) fusion techniques. The rGTV was defined as recurrent tumors detected by CT/MRI, endoscopy, and physical examinations, including nasopharyngeal tumors, retropharyngeal lymph nodes, and cervical lymph nodes. Positivity of cervical lymph node was defined as lymph nodes > 1 cm in the shortest dimension of the largest transverse cross-section, or lymphadenopathy indicating central necrosis, clusters of at least three critical-sized nodes, and PET-CT-positive lymph nodes.

The definition of CTV in locally recurrent NPC is controversial, and there is no uniform standard for rCTV15,16,17,18,19. rCTV was defined as rGTV + 5–10 mm margin. Planning target volume (PTV) was defined as rGTV or rCTV + 3 mm margin. When the rGTV or rCTV was near the brainstem and spinal cord, PTV was generated with a margin of less than 1 mm. The prescribed dose to rGTV was 60–66 Gy. The fractional dose was 1.7–2.2 Gy/day. The median dose was 60 Gy (50–72 Gy) in the rGTV group and 50 Gy (40–57 Gy) in the rCTV group. An Xray of 6 MV energy was used to implement the radiotherapy program. Patients were treated once a day, 5 days a week. The OARs include the brain stem, spinal cord, optic chiasm, optic nerve, pituitary gland, lens, eye, hippocampus, temporomandibular joint, mandible, temporal lobe, and parotid glands. The radiation dose limits for normal tissues are less than 70% of the maximal acceptable cumulative dose for OARs. The brainstem, spinal cord, and optic chiasma are the top priority, followed by rGTV and other less important OARs. Specific data for critical OARs are as follows15: the maximal acceptable cumulative dose of 70.2 Gy for the brainstem, 58.5 Gy for the spinal cord, 65 Gy for the optic chiasm, 65 Gy for the optic nerve, D1 cm3 < 84.5 Gy for the temporal lobe and D1 cm3 < 85.8 Gy for the brachial plexus.

The local failure patterns of re-irradiation were defined as: (1) In-field: ≥ 95% recurrent tumor volume of re-irradiation (Vrecur) was within 95% isodose line (V95), (2) Marginal: 20% ≤ Vrecur < 95% was withinV95, (3) Out-field: < 20% Vrecur is within V9511.

For chemotherapy, there is no unified standard chemotherapy treatment for recurrent NPC currently. Platinum-based palliative chemotherapy was the most common regimen, including gemcitabine (1000 mg/m2, days 1 and 8) + platinum, paclitaxel (135–175 mg/m2, day 1) + platinum, docetaxel (60 mg/m2 on day 1) + 5-fluorouracil (600 mg/m2; continuous administration via a 96 h IV drip) + platinum, 5-fluorouracil (1000 mg/m2/d on day 1 to day 4) + platinum and other regimens. Concurrent chemotherapy was performed with single-agent platinum chemotherapy every 3-weeks.

Follow-up

After treatment, the follow-up was performed every 3 months for the first 2 years, then every 6 months for 3–5 years. Thereafter, follow-up was conducted annually. Follow-up included an integral medical history, physical examination, plasma Epstein-Barr virus (EBV) DNA, nasopharyngoscopies, MRI of the nasopharynx and neck, chest CT and abdominal ultrasound, and bone scan. Upon diagnosis of local recurrence, 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET)-CT scan was recommended. Treatment-related toxicities were evaluated according to the Radiation Therapy Oncology Group (RTOG) toxicity evaluation criteria and the Common Terminology Criteria for Adverse Events (CTCAE) vision 4.0. Local recurrence-free survival (LRFS) and distant metastases-free survival (DMFS) were defined as the time from the date of relapse diagnosis to the date of local failure or distant metastasis. The overall survival (OS) was measured from the date of relapse diagnosis to the date of death from any cause or the date of the last follow-up.

Statistical analysis

All statistical analyses were performed using SPSS 24.0 and Graph Pad Prism 8. Age, irradiated dose, and GTV volume are represented as categorical variables in the analysis. The survival curves were compared using the Kaplan–Meier method and examined by log-rank test. The Cox risk regression model was used for multivariate prognostic analysis. Variables with statistical significance (P < 0.05) in univariate analyses were further included in multivariate analyses. Categorical variables were assessed using Fisher's exact test, and continuous variables were assessed using Mann–Whitney U test. All tests were two-sided, and P ≤ 0.05 was considered statistically significant.

Results

Patient characteristics

A total of 97 eligible patients with locally recurrent NPC were included between July 2014 and December 2018 (Fig. 1). The details of patient characteristics are listed in Table 1. The median age was 50 years (30–79 years), and the male: female ratio was 3.4:1. Patients with rT3-4 stage account for 83.5% (81/97) of all NPC. The median time from initial radiotherapy to recurrence was 44.9 months (4.7–436.2 months). The clinical target volume of recurrence (rCTV) was delineated in 63.9% (62/97) of patients, and Gross tumor volume of recurrence (rGTV) only was delineated in the remaining patients. The median rGTV dose was 60 Gy (50–72 Gy). The median rCTV dose was 50 Gy (40–57 Gy).

Figure 1
figure 1

Flowchart showing the patient selection process.

Full size image

Survival analysis

The median follow-up time was 63.0 months (range 2.6–90.8 months). The overall survival (OS) at 1-, 3-, and 5-years was 80.1%, 44.9%, and 27.6%, respectively (Fig. 2A). The local recurrence-free survival (LRFS) at 1-, 3-, and 5-years was 93.9%, 65.9%, and 56.4%, respectively. The distant metastases-free survival (DMFS) at 1-, 3-, and 5-years was 92.8%, 81.8%, and 81.8%, respectively.

Figure 2
figure 2

Kaplan–Meier curves for overall survival (OS), local recurrence-free survival (LRFS), and distant metastasis-free survival (DMFS) (A). Kaplan–Meier curves for OS according to rT stage (B), GTV volume, and clinical target volume of recurrence (rCTV) or gross tumor volume of recurrence (rGTV).

Full size image

Kaplan–Meier curves revealed that patients with rT1-2 stage or rGTV volume < 30 cm2 had longer OS than patients with rT3-4 (HR 0.33; 95% CI 0.192–0.568; P = 0.003) or rGTV volume ≥ 30 cm2 (HR 0.50; 95% CI 0.311- 0.805; P = 0.005), respectively (Fig. 2B,C). There was no statistical significance in OS whether patients were treated with rGTV or rCTV (Fig. 2D), nor was it in LRFS and DMFS (Fig. S1).

Univariate analysis showed that the rT stage, rTNM stage, re-irradiation dose, and rGTV volume were closely associated with OS (Table 2). Multivariate analysis showed that rT stage (HR 2.62, 95% CI 1.17–5.84, P = 0.019) and rGTV volume (HR 1.73, 95% CI 1.03–2.89, P = 0.037) were independent prognostic factors for OS. However, no significant prognostic factors were found for LRFS and DMFS (Table S1).

Table 2 Univariate and multivariate analysis of overall survival in 97 locally recurrent NPC patients with re-irradiation.
Full size table

Local failure patterns and toxicity analysis of re-irradiation

For all 97 patients, the target volume (rGTV or rCTV) was not significantly related to the secondary recurrence rate (Table 3). In addition, among 29 patients who developed local failure after re-irradiation, 18 patients (6 patients belonged to local uncontrolled) and 10 patients had in-field or out-field failure, respectively (Table 4). One patient was excluded because of incomplete imaging data. Among the 10 patients with out-field failure, 4 patients developed regional lymph node recurrence, 1 patient exhibited supraclavicular lymph node metastases, and 1 patient experienced lung metastases. Besides, there is no statistical significance between local failure patterns and rT stage, rTNM stage, GTV volume, and chemotherapy (Table S2). It was worth noting that most patients (22/28) had the secondary recurrence tumor located in the GTV of the first radiation fields. Moreover, target volume was not significantly associated with the second recurrent pattern in 28 local relapse patients (Table 3).

Table 3 Analysis of NPC patients treated with rCTV or rGTV of re-irradiation.
Full size table
Table 4 Details of locally recurrent 28 NPC patients after re-irradiation.
Full size table

The rate of adverse events related to radiotherapy of grade 3 or worse was 55.7% (54/97), and the most common radiation related late adverse event was nasopharyngeal necrosis, with a rate of 44.3% (43/97) (Table S3). There was no significant difference between target volume and radiation related late adverse events of grade 3 or higher (Table S3). Of note, 41.2% (28/68) died of Re-irradiation associated toxicity, and the most common cause of death was massive nasopharyngeal hemorrhage in 15 cases (22.1%).

Risk stratification and treatment model optimization

Risk stratification was performed to select the patients who benefited most from re-radiotherapy. Since recurrent NPC patients with rT1-2 stage had significantly better 3-year OS than patients with the rT3-4 stage (81.26 vs. 36.67%), patients with the rT1-2 stage were categorized as Group 1, ignoring the volume of rGTV. Patients with the rT3-4 and rGTV volume lower or larger than 30 cm2 were categorized as Group 2 and Group 3, respectively (Fig. 3A). Kaplan–Meier curves showed no statistical significance in OS between Group 1 and Group 2. Thus, Group 1 and Group 2 were integrated into the low risk group, and G3 was identified as the high risk group (Fig. 3B).

Figure 3
figure 3

Prognostic model for overall survival (OS) (A), Kaplan–Meier curves for OS grouped by 3 prognostic groups (B). Kaplan–Meier curves for OS grouped by low risk and high risk groups (C). Survival curves of Re-irradiation associated mortality rate grouped by low risk and high risk groups.

Full size image

Compared to the high risk group, the low risk group had a longer 3-year OS rate (66.7% vs. 23.4%) and 5-year OS rates (37.6% vs. 14.7%) (Fig. 3C). However, LRFS and DMFS were not associated with the risk of NPC patients (Fig. S2A–2B). Moreover, the low risk group had lower total grade 3 or worse radiation related late adverse events (Table S4), especially in the occurrence of nasopharyngeal necrosis (23.8% vs. 60.0%, P < 0.001). Besides, the re-irradiation associated mortality rate was also lower in the low risk group compared to the high risk group, with the corresponding rates of 23.8% and 32.7%, respectively (HR 0.45, 95% CI 0.21–0.94, P = 0.03) (Fig. 3D). However, neither high risk group nor low risk group benefited from chemotherapy (Fig. S2C–2D).

Discussion

Treatment of locally recurrent NPC is one of the most difficult challenges. Because of the heterogeneity of locally recurrent NPC, there is no one-size-fits-all regimen9. Surgery is preferred for those highly selected patients whose recurrent tumors do not involve critical organs, generally including most rT1-2 and few rT3 patients20. However, for patients who are inoperable or unwilling to undergo surgery, comprehensive treatment based on radiation therapy is usually recommended21. In this study, we found that local failure patterns and outcomes of re-radiation were not associated with delineation of rCTV or not in inoperable locally recurrent nasopharyngeal carcinoma. Patients with low risk (rT1-T2 and rT3-T4 with rGTV volume lower than 30 cm2) benefit most from re-irradiation and suffer from lower radiation related late adverse toxicities and mortalities.

According to our knowledge, our study was the first report to evaluate the local failure patterns of recurrent NPC with re-irradiation. For 28 patients who developed local recurrence after re-IMRT, the local failures of the first radiotherapy were all in-field. On the one hand, it is worth noting that 64.3% of patients (18/28) recurred in-field again, and the majority of secondary recurrent tumors (25/28) were located in the GTV or CTV of the first radiation field, indicating an intrinsic radio-resistance22. On the other hand, among the 35.7% (10/28) patients who developed out-field, 7 patients were treated with the delineation of rCTV, suggesting that rCTV might not primely prevent recurrence. Therefore, potential mechanisms of radio-resistance and in-depth research of recurrence patterns are urgently needed. Besides, other modes of radiotherapy, like hyperfractionation, could be applied7. Carbon, protons, and heavy ions therapy are also suggested if the facility is available23,24.

Globally, the delineation principles of rCTV differ from each center in locally recurrent NPC16,18,19. So far, there is no one-size-fits-all recommendation on the delineation of rCTV of re-radiation. Princess Margaret Cancer Center and our center defines rCTV as rGTV + 5–10 mm margin8,16. The margin is 1.0–1.5 cm in Sun Yat-sen University Cancer Hospital17, and 3–5 mm in Fudan University Cancer Hospita18. In contrast, Pamela Youde Nethersole Eastern Hospital does not recommend CTV expansion19. Guidelines of an international consensus reported that 96% of specialists suggested that rCTV should be delineated, and 100% of the experts suggested that the expansion margin for rCTV was considered to be less than 5 mm9. Our study revealed that delineation of rCTV was not significantly associated with the secondary recurrence rate, second recurrent patterns, grade 3 or worse radiation­related late adverse events, and survival outcomes. Thus, whether the delineation of rCTV was indispensable remains to be further studied since a minority of patients (3/28) had the secondary recurrence tumor located in the CTV of the first radiation fields. Re-irradiation with the delineation of only rGTV should be considered a treatment option for locally recurrent NPC. Besides, means of radiotherapy delivery are also critical. Compared to the recently published Lancet7, hyperfractionated IMRT had lower grade 3 or worse late radiation-induced toxicity than standard fractionation in our study, including nasopharyngeal necrosis (19.1% vs. 44.3%), hemorrhage (7.4% vs. 19.6%) and cranial nerve palsy trismus (10.3% vs. 22.7%), but not temporal lobe necrosis (27.9% vs. 18.6%). Of note, these two studies cannot be strictly compared, and the median follow-up is shorted in the hyperfractionation population. Thus, precise and individualized treatment plans are wanted to avoid overtreatment and undertreatment.

Currently, multiple factors are supposed to affect the efficacy, side effects, and outcomes of re-irradiation19,25. Our study found that rT stage and rGTV volume were independent prognostic factors for OS. We chose 30 cm3 as the cutoff value of tumor volume, given the preliminary data showing the feasibility of this volume17. Consistent with our research, a meta-analysis reported that recurrent T stage and tumor volume serve as the prognostic factors for OS in recurrent NPC patients14. As to the cutoff values of rGTV, different oncologists had different opinions. Han et al.17 and Tian et al.13 pointed out that recurrent tumor volume exceeding 30 cm3 or 38 cm3 were unfavorable prognostic factors for OS. In comparison, Ng WT reported that the local control rate was estimated to fall to less than 10% if the GTV exceeds 80 cm3 irrespective of radiation dose. And no association was observed between rGTV volume and the fatal complication rate, which was coincident with our findings. Thus, consensus on re-irradiation by IMRT believed that the cutoff criterion of rGTV was not a factor for exclusion9.

Besides, prognostic models provide useful tools for re-irradiation. Li et al.12 and Lu et al.26 had identified six parameters (age, Karnofsky Performance Status, rT stage, late complications, synchronous nodal recurrence, and GTV-nx) or four risk factors (relapse T stage, relapse gross tumor volume, time to recurrence, and initial tumor response), respectively, to predict patients’ sensitivity to re-radiotherapy. Tian et al.13 developed a prognostic-score model based on five parameters (rGTV, rT stage, age, previous RT toxicity, and planned RT dose) to identify patients who could benefit from salvage IMRT. However, the above study did not compare grade 3 or worse radiation­related late adverse events or re-irradiation associated mortality in detail. Our data yielded low and high risk subgroups based on just two parameters (rT stage and rGTV volume). The 5-year OS rate of the low-risk group was approximately 3 times higher than that of the high-risk group (42.3% vs. 14.7%). Most importantly, the occurrence of nasopharyngeal necrosis in the low risk group decreased by 36.2% (23.8% vs. 60.0%), and the risk re-irradiation associated mortality remarkably reduced by 55% (HR 0.45, 95% CI 0.21–0.94) than in high risk group. Therefore, re-irradiation by IMRT may be more appropriate for low risk patients. The magnitude of benefit was less in the high risk patients compared to the low-risk patients because of higher grade 3 or worse late radiation-induced toxicity. Thus, other comprehensive treatment strategies, like immunotherapy, angiogenesis inhibitors, and maintenance metronomic, should be explored to ensure the effectiveness of treatment and improve the quality of life27,28,29.

As for the principles of systemic therapy, there is no standardized treatment recommendation for locally recurrent NPC. Liu et al. reported that compared to palliative chemotherapy, salvage radiotherapy significantly improved 3-year OS in the high Salvage Radiotherapy Outcome Score (SARTOS) subgroup (67.3% vs. 42.0%)30. While yan et al. showed that additional induction chemotherapy plus re-irradiation had longer 3-yearOS versus re-irradiation alone (67.0% vs. 47.0%) in the intermediate-risk group6. Inconsistent with those two studies, we found that chemotherapy was not beneficial either in the high risk or in the low risk group. A high proportion (83.5%) of patients with rT3-4 stage in our study and the diverse prognostic factors involved in establishing models may account for the different conclusions. A multicenter, prospective clinical trial is required to validate the results.

This study has some limitations. Firstly, our study was a retrospective study from a single center, which might induce an offset. Secondly, our study contained a high proportion (83.5%, 81/97) of rT3-4 stage that couldn’t receive surgery, which may not apply to patients who are suitable for surgery. Thirdly, the discrepancy in chemotherapy may affect the survival and toxicity, which was not further analyzed because of limited samples.

Conclusion

In-field failures are the main patterns of local recurrence in locally recurrent NPC patients receiving re-irradiation by IMRT. There was no significance between target volume (rGTV or rCTV) and the local recurrence rate, local failure patterns, grade ≥ 3 toxicity, and survival. Thus, delineation of rCTV may not be beneficial for re-irradiation using IMRT in locally recurrent NPC. More importantly, risk stratification showed that the low risk group had a higher 5-year OS rate (37.6% vs. 14.7%), lower nasopharyngeal necrosis (23.8% vs. 60.0%) and lower re-radiation associated mortality rates (HR 0.45) than the high risk group. Therefore, Re-irradiation by IMRT was more suitable for low risk patients as salvage treatment. Clinical and basic science research on re-irradiation is needed to gain maximum survival benefit and minimum late adverse toxicity in locally recurrent NPC.