A Ten-year Experience of the Multidisciplinary Management of the Regional Failure of Nasopharyngeal Carcinoma in the Era of IMRT

Objective: To evaluate clinical outcomes of cervical node recurrence/residual tumor of nasopharyngeal carcinoma (NPC) with treatment decisions made by a multidisciplinary treatment (MDT) team. Methods: Between March 2010 and July 2019, patients who had previously undergone denitive intensity-modulated radiotherapy (IMRT) for nasopharyngeal carcinoma and presented with residual or recurrent cervical lymph nodes were enrolled in our study. Clinical features and treatment modalities were retrospectively reviewed, and survival outcomes were analyzed. Results: A total of 183 patients were enrolled in the study, among whom 139 underwent neck dissection, and the remaining 44 were either considered not eligible for surgery or refused surgery. For the two groups of patients, the 3-year overall survival (OS) was 73.4% vs 36.8% (p=0.000), and the progression-free survival (PFS) was 57.1% vs 32.3% (p=0.004), respectively. Patients who underwent neck dissection had a signicantly better clinical outcome and shorter latency between the prior treatment and detection of neck recurrence/residual tumor than those who did not. Conclusion: For regional failure cases that have no contradictions for surgery and are predicted to benet from surgery by the MDT team, neck dissection could be the best choice. It is important for an experienced MDT team to choose the optimum treatment modalities for patients for the cervical failure of NPC. Regular follow-up after prior treatment contributes to the early detection of regional failure and provides a chance for surgery.


Introduction
Nasopharyngeal carcinoma (NPC) is a comparatively rare malignancy and is more endemic in South China and Southeast Asia, with distinct pathological features and clinical characteristics from other head and neck cancers. Multidisciplinary treatment based on radiotherapy has been the mainstay of treatment for NPC. Over the past several decades, with the widely applied technology of IMRT, the clinical outcome of NPC has improved. The 5-year regional control of nasopharyngeal carcinoma can reach up to 90-95% 1-3 . Thus, regional failure has received limited attention. The main strategy for regional failure is surgical dissection. However, several other treatments are also used by different cancer centers. A more standardized multidisciplinary treatment approach will not only help effectively control local recurrence but also reduce the risk of further progression. Patients with recurrence in a previously treated neck should still undergo surgery. For some speci c cases, nonsurgical therapy may also be utilized when clinically appropriate. As one of the largest centers for head and neck cancer in China, our center has formed an experienced multidisciplinary treatment team for nasopharyngeal carcinoma and has accumulated su cient experience in the treatment of the regional recurrence of NPC. In the current study, we reviewed treatment outcomes of patients who received de nitive IMRT for nasopharyngeal carcinoma and were diagnosed with regional recurrence. Multidisciplinary treatment for the regional failure of nasopharyngeal carcinoma was evaluated, and a comprehensive analysis of the prognostic signi cance of patient clinical manifestations was performed.

Patient characteristics
From March 2010 to July 2019, patients who had previously undergone de nitive IMRT with or without chemotherapy for nasopharyngeal carcinoma and presented with residual or recurrent cervical lymph nodes with no evidence of distant metastasis were included in our study. The patients' recurrence or residual tumors were diagnosed at Fudan University Shanghai Cancer Center, and patients received treatment at our center. We retrieved patient medical records and collected all detailed clinical characteristics, imaging, pathological reports, and treatment records of both prior treatment and recurrence/residual tumors. Exclusion criteria included the following: (1) the presence of a second primary malignancy, or NPC was a secondary malignancy; (2) postoperative pathology of the cervical node specimen showed no evidence of malignancy; and (3) insu cient clinical documentation. Tumors were staged according to the 7th edition of the American Joint Committee on Cancer/Union for International Cancer Control staging system.

Treatment
When pathologically or clinically diagnosed as cervical LN residual tumor or recurrence, patients were assessed by the multidisciplinary treatment team for head and neck cancer. Residual disease was de ned as lymph nodes without complete regression by 3 or more months after primary therapy, and recurrent disease was de ned as the reappearance of cervical lymph nodes after initial complete regression. Magnetic resonance imaging (MRI) and/or computerized tomography (CT) with contrast scans of the head and neck were performed to evaluate the extent of local regional disease and evaluated for surgical feasibility. For suspicious nodes that could not be diagnosed by an imaging study, ne needle aspiration (FNA) was performed. Patients who were eligible for neck dissection underwent either radical neck dissection (RND) or selective neck dissection (SND) based on the surgeon's decision and the patient's wishes. If neck dissection was not possible because of the node location, size, or invasion of adjacent tissues or if patients refused surgery, chemotherapy, radiotherapy or immunotherapy was administered. Patients who were eligible for reirradiation for cervical recurrence received IMRT, and the gross tumor volume (GTV) referred to the recurrence gross tumor or suspected residual tumor after resection. The total dose for recurrence was 60-66 grays (Gy). The choice of systemic therapy as individualized based on the patient's clinical condition. According to the NCCN guidelines, the preferred rst-line regimen was cisplatin/gemcitabine. Other rst-line recommendations included cisplatin/5-FU, cisplatin or carboplatin/docetaxel or paclitaxel, carboplatin/cetuximab, and gemcitabine/carboplatin.
Single-agent therapies from the abovementioned regimens were also used for some patients.

Statistical analysis
All analyses were performed with SPSS software, version 26.0. Overall survival (OS), progression-free survival (PFS), regional recurrence-free survival (RRFS), local recurrence-free survival (LRFS), and distant metastasis-free survival (DMFS) were analyzed. Time-to-event statistics were calculated from the date of the diagnosis of recurrence to the event of interest. The Kaplan-Meier method was used to estimate survival rates, and the log-rank test was used to compare the differences. The hazard ratio (HR) and associated 95% con dence interval (CI) were calculated using the Cox proportional hazard model. Potential variables, including age, sex, the latency of nodal failure from prior treatment, staging classi cation at the initial presentation of NPC, whether disease was accompanied by synchronous local recurrence, residual or recurrent condition, N classi cation of nodal failure, extracapsular spread, cystic lymph nodes, levels and sides involved, and treatment modalities, were tested for their association with survival outcomes. Signi cant predictors identi ed in univariate analysis and factors that may have in uenced clinical outcomes were included in multivariate analysis. The χ2 test was used to compare categorical variables. Independent t-tests were used to compare continuous variables. In all cases, a 2sided p value < 0.05 was considered signi cant.

Results
Patient characteristics and the management of regional failure A total of 183 patients were enrolled in this study, among which 139 underwent resection, and the remaining 44 were either considered ineligible for surgery or refused surgery. Before treatment, all cases were discussed by our multidisciplinary treatment team, and for patients that were considered to have no contradictions for surgery and predicted to bene t from surgery by our MDT team, neck dissection was strongly recommended to the patients. Twenty-eight (15.3%) patients had FNA pathology, and others were clinically diagnosed by the MDT team.
Clinical pro les of primary disease and prior treatment are listed in Table 1. All patients received a standard course of IMRT during prior treatment. The dose to the primary tumor and metastatic cervical nodes ranged from 66 to 70.4 Gy in 30-35 fractions. A total dose of 60 Gy was delivered to the high-risk CTV and 50.4-54 Gy for low-risk CTV, respectively. A total of 179 (97.8%) of the patients received platinum-based induction, concurrent, or adjuvant chemotherapy. Fourth-one (22.4%) patients received a 2.0-to 6.0-Gy boost for lymph node residual disease in 1-3 fractions after the standard IMRT plan nished. Thirty-eight (20.8%) patients had residual cervical lymph nodes 3 months after prior radiotherapy, and 27 (14.8%) never achieved a complete response before resection.
All patients who underwent neck dissection were con rmed to have recurrence by postoperative pathology. The time from prior radiation therapy to the diagnosis of regional failure ranged from 2.4 to 105.6 months (median, 19.6 months) for the whole group. Forty-two (23.0%) patients had synchronous primary site recurrence. Regarding the level of involved lymph node level involvement, 151 (82.5%) patients had recurrence in the unilateral neck, and the other 32 (17.5%) patients suffered from bilateral recurrence. According to our analysis of the lymph node level distribution, level IIb (62.8%) was the most common site of recurrence, followed by level For the 2 subgroups of patients who did or did not undergo neck dissection, the 3-year OS rates were 73.4% vs 36.8% (p = 0.000, Fig. 1A), and the PFS rates were 57.1% vs 32.3% (p = 0.004, Fig. 1B), respectively. Comparisons of clinical characteristics between patients in the neck dissection group and the non-neck dissection group are listed in Table 3. Patients who underwent neck dissection had a shorter interval between prior treatment and the diagnosis of neck recurrence than those who did not undergo neck dissection. The median interval of the non-neck dissection group was 30.8 (range: 6.0-96.0) months, compared to 15.9 (range: 2.4-105.6) months in the neck dissection group (p < 0.001). In the neck dissection group, only 16.5% of patients had synchronous local recurrence, which differed signi cantly from 43.2% of patients in the non-neck dissection group. Only 8.7% of patients in the neck dissection group had lymph nodes larger than 3 cm, while in the non-neck dissection group, the proportion was 29.5%. A total of 84.1% of patients who did not receive surgery had ECS, while in the surgery group, the proportion was only 48.2%. Cystic lymph nodes did not differ between the two groups. For those who did not receive neck dissection, chemotherapy (63.6%) and reirradiation (20.5%) were administered. In the neck dissection group, 20.1% of patients received chemotherapy, and 7.2% received reirradiation before or after surgery.
Multiple clinical characteristics that might in uence patient clinical outcomes were included in univariate analysis ( Table 4), showing that age older than 50, not undergoing neck dissection, synchronous local recurrence, nodes with ECS, and receiving chemotherapy were associated with signi cantly worse PFS and OS. Multivariate analysis con rmed the improvement in OS conferred by surgery (p = 0.012). Age older than 50 years remained an independent prognostic factor for both PFS and OS. The details of multivariate analysis are shown in Table 5.

Discussion
Multidisciplinary treatment for cervical lymph nodes or the recurrence of NPC after IMRT has seldom been reported in previous studies. Neck dissection is the preferred choice for the neck failure of NPC, with better treatment outcomes and a lower rate of treatment complications than reirradiation and/or chemotherapy 4 . In our study, it was remarkable that the treatment outcome was signi cantly better in those who received surgery, and the outcome was much worse for those who did not undergo surgical dissection. Therefore, evaluating patient eligibility for surgery is the primary task when dealing with the regional failure of NPC. The decision regarding treatment modality should be made by an experienced multidisciplinary team, which should be made up of not only radiation oncologists and radiologists but also, most importantly, surgeons specializing in head and neck surgery. In the MDT team in our center, when a patient had neck recurrence, we rst evaluated the patient's general condition and excluded contradictions for surgery and then went through all the images before and after prior treatment and the images indicating node failure. After a detailed review of the patient's medical history, we evaluated whether the patient was eligible for surgery. Surgeons' experience is important in the decision regarding surgery. For more complicated cases, an experienced surgeon who has performed a considerable number of surgeries may a better chance of performing a total dissection and achieving better long-term survival.
The selection of radical neck dissection or selective neck dissection was also discussed by the team based on the imaging studies of the patient. According to a former study by our center, selective neck dissection was demonstrated to be safe and effective for the treatment of neck residual disease of NPC for selected patients 5 . Considerations made by the MDT team are illustrated in Fig. 2. As long as the team considered the patient eligible for resection with no contradictions for surgery, we strongly suggested that the patient accept surgical interventions. Whether systemic treatment or radiotherapy might bring additional survival bene ts before or after surgery was also determined by the MDT team.
Under certain circumstances (e.g., for patients who presented with synchronous local recurrence, patients who were predicted to hardly bene t from surgery, or patients with contraindications to surgery), a more conservative approach may be adopted. Over the past several decades, many potentially e cacious treatment options have emerged for the therapeutic management of regional recurrence in NPC patients who are not eligible for surgery 6 . Only a very limited proportion of our patients received reirradiation for regional recurrence and subsequently presented with severe complications secondary to cumulative irradiation injury, such as skin necrosis, soft tissue brosis, or brachial plexus injury. Chemotherapy was a comparatively more palliative strategy, as resistance to chemotherapy was common in cervical recurrence cases and the response rate was not satisfactory. Toxicity is often intolerable after several cycles and multiple lines of chemotherapy. Therefore, for patients with regional recurrence, indications for surgery must be carefully considered, and proper and personalized comprehensive treatment modalities should be identi ed. Our 10-year data indicated that having MDT discussions regarding each case and not neglecting a single case eligible for surgery, the best treatment choice would be selected for patients.
It is notable that according to our data and consistent with other studies 7 , the distribution of recurrent or residual nodes was not as extensive as in primary NPC cases; 60.7% of patients presented with a comparatively limited range of recurrence, involving only one level on one side, and 86.3% presented with lymph nodes with diameters less than 3 cm. Level II (84.2%) was the most common site of nodal recurrence. Prior radical radiotherapy may have impaired the cervical lymphatic drainage function, thereby reducing the secondary cervical metastasis of the tumor. The interval between the end of prior treatment and the diagnosis of neck recurrence was much shorter in the neck dissection group than in the non-neck dissection group, suggesting that patients with indications for neck dissection are more likely to be diagnosed at an early stage of disease. A relatively large proportion of patients went on for routine follow-up after prior treatment, which occurred every 3 months within the rst 2 years and every 6 months in the 3rd to 5th years after treatment. The peak of recurrence diagnosis occurred at approximately 20 months, ranging from 2.4 months to 105 months, so intensive follow-up should always be recommended for NPC patients who nish de nitive prior treatment. It was also proposed by our previous study that close follow-up is still important for the early detection of recurrence in years 3-4 after treatment 8 . Those who were considered to bene t less from surgery or lost the chance for surgery may have not had regular follow-up until the clinical presentation was too severe, for example, when the recurrent lymph nodes might have been too large, or might have too extensive involvement of the adjacent structures, making a total dissection di cult to perform. Therefore, for nasopharyngeal carcinoma patients, regular follow-up after treatment completion is crucial, and the early detection of regional recurrence may provide a better chance of complete resection.
Univariate analysis showed that age older than 50, synchronous local recurrence, extracapsular spread, not receiving surgery and receiving chemotherapy were related to a signi cantly worse prognosis.
Previous studies 9 reported similar results, suggesting that younger patients (age < 40-50 years) had a better prognosis for cervical recurrence. Older patients tend to have a lower KPS and worse heart or lung function, making surgery or even other comprehensive treatment modalities intolerable. Therefore, treatment selection for older patients should be performed more cautiously. Cervical recurrence with concurrent primary site recurrence in the nasopharynx is not rare and might originate from speci c malignant cells with resistance to prior chemoradiotherapy. The treatment decision for this group of patients varies and is still based on each patient's speci c condition. For those with a limited range of local recurrence that could be managed by endoscopic surgery or reirradiation and cervical recurrence that could bene t from surgery, the separate management of recurrence in both sites might be effective. It was suggested in a study from Hong Kong that neck dissection is e cacious in patients with nasopharyngeal carcinoma with nodal failure with or without synchronous local failure 10 . Systemic treatment should also be considered for some patients. However, the control rate of local recurrence is not always satisfactory, with considerable rates of nasopharyngeal necrosis and hemorrhage shortly after treatment, especially in patients receiving reirradiation. In this case, each case should be discussed in detail by an MDT team to select the best treatment. Extracapsular spread is usually considered related to worse outcomes 11,12 , regardless of primary or recurrent nodes. Extensive soft tissue and muscle invasion are barriers to complete resection, and adjuvant local reirradiation or chemotherapy might be indispensable for effective regional control. However, a recent study proposed that within a group of 46 patients who received neck dissection after the diagnosis of nodal recurrence, ECS was not correlated with a worse survival. They proposed that an absolute number of positive lymph nodes greater than ve and a lymph node density greater than 20% were potentially useful prognostic factors 13 . Patients who received chemotherapy were usually not eligible for surgery, so it could be speculated that chemotherapy is related to a worse prognosis, as shown in the univariate analysis. Multivariate analysis revealed that surgery and age older than 50 were the only 2 independent prognostic factors for OS, suggesting the dominant role of surgery in the management of cervical recurrence.
Several other studies have demonstrated that patients with cervical node residual disease after prior radical irradiation may have a better prognosis than patients who present with node recurrence. This was also observed in our colleague Zhang's former study based on conventional radiotherapy 14 . In the current study, however, patients with cervical node residual disease did not have better survival than those with cervical node recurrence, even though the range of the cervical residual disease was usually limited, and there was a greater chance of neck total resection. A review of the data showed that there was a larger proportion of N3 patients in the residual disease group than in the recurrence group (51.8% vs. 24.3%), so the possible survival advantages of patients with residual disease might be offset by the larger proportion of N3 patients, who obviously had worse clinical outcomes. In addition, in our study, all patients received IMRT as prior treatment for primary NPC. Compared with conventional radiotherapy, IMRT achieves better survival, with bene ts of a better dose distribution, i.e., better dose heterogeneity within the target volume and a lower dose to organs at risk (OAR). Thus, residual lymph nodes after IMRT might usually be comparatively more radiation resistant. In our study, 22.4% of patients received a boost for obvious lymph node residual disease when the IMRT course nished, and the rate was much higher than the reported rate of 8.3% in general in a previous study from our center 1 .
IMRT technology has been used at our center for NPC patients since 2005. Most patients who received prior treatment returned to our center for follow-up, and subsequent treatment for treatment failure was also received here. A total of 355 patients with neck residual disease or the recurrence of NPC after 2D-RT who underwent radical neck dissection over a 10-year period from 1998 to 2007 were reported before. The 3-and 5-year overall survival rates were 54.1% and 26.0%, respectively. In this study, within a decade, the survival of nasopharyngeal carcinoma improved, and cases of failure decreased signi cantly. Only 139 patients underwent neck dissection at our center. This group of patients achieved much higher 3-and 5-year overall survival rates of 73.4% and 52.8%, respectively. In addition to improvements in radiation techniques, advances in systemic treatment and more accurate imaging techniques, such as MRI and PET/CT 15,16 , also contributed to better clinical outcomes. Moreover, new advances in systemic therapy, especially immunotherapy, have provided promising preliminary results for recurrent and metastatic nasopharyngeal carcinoma [17][18][19] . Including immunotherapy in multidisciplinary treatment is a new trend, and the best combination of surgery, radiotherapy and systemic therapy should be further examined and discussed.
The retrospective nature was a major limitation of our study. Patients' conditions varied, and strategies for each case were determined by the MDT team based on their clinical features. Based on the current ndings and experiences, we will perform further prospective studies focusing on developing a clearer standard that could allow patients to bene t from surgery and identify the best comprehensive treatment strategies for patients.
In conclusion, for recurrent or residual nodes of NPC after prior chemoradiotherapy, radical neck dissection could be the best choice after a comprehensive multidisciplinary evaluation of the patients' condition. Regular, close follow-up after treatment and access to an experienced MDT team to make optimal treatment decisions for patients are key components to a better clinical outcome for the regional failure of NPC.