From November 2003 to December 2017, a total of 101 cases of initially treated patients with unilateral NPC were admitted to our research group at the Sun Yat-sen University Cancer Center (SYSUCC), among which 6 cases were lost to follow-up, thus, 95 cases were included in long-term analysis. Unilateral NPC was defined as nasopharyngeal mass confined to one side of nasopharynx and did not significantly exceed the midline of nasopharyngeal apex/posterior wall detected by electronic nasopharynxgoscope (ENS) and magnetic resonance imaging (MRI) (Figure 1A). The included criteria for patients were: (1) initially pathologically confirmed unilateral NPC; (2) > 18 years old; (3) treated with IMRT; (4) the performance status (PS) was 0 or 1; (5) no distant metastasis; (6) informed consent related to treatment signed. Exclusion criteria: (1) metachronous or synchronous malignancy; (2) pregnancy or lactation; (3) mental disorders. All patients were staged according to the 7th American Joint Committee on Cancer (AJCC) staging classifications. The study conformed to the ethical guidelines of the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of SYSUCC with the approved number of B2019-169-01.
All patients received IMRT. Sixty-three patients received a one-course radiotherapy IMRT and 32 patients received an adaptive replanning IMRT. Six, 6 and 8 patients with stage I, II and III diseases, respectively, received radiotherapy alone. Six, 32 and 10 patients with stage II, III and IV diseases, respectively, received concurrent chemoradiotherapy (CCRT). Sixteen and 11 patients with stage III and IV disease, respectively, received induction chemotherapy (IC) plus CCRT.
All patients were immobilized using a head-and-neck thermoplastic mask in a supine position, followed by computed tomography (CT) simulation using 3-mm slices scanned from the top of the head to 2- cm sub-clavicle. For patients treated with replanning IMRT, a second CT simulation was performed at 22 fractions of radiotherapy.
Definition of target volumes
The definition of target volumes and organs at risk (OARs) followed the International Commission on Radiation Units and Measurement (ICRU) Reports No. 50 and No. 62. The principles of target volume delineation were referred to the norms of SYSUCC  and were revised according to the method applied in our research group[11,13]. From November 2003 to May 2013, GTVnx was defined as all gross disease detected by clinical, imaging and endoscopic examination. GTVrpn/nd was defined as the metastatic retropharyngeal lymph nodes/neck lymph nodes detected by clinical and imaging examination; The high-risk CTV (CTV1) was defined as a subclinical disease consisting of 10 mm margin surrounding GTVnx (not including all nasopharyngeal mucosa); CTVrpn/nd was defined as 0.5-1 cm margin surrounding GTVrpn/nd; The low-risk CTV (CTV2) was defined as a 0.5-1 cm margin surrounding CTV1, including all nasopharyngeal mucosa within contralateral pharyngobasilar fascia, and including the bilateral prophylactically irradiated lymph drainage areas (bilateral retropharyngeal lymph node area, levels IIa, IIb, III, and Va are routinely covered for all N0 patients, whereas ipsilateral levels IV and Vb are also included for N1-3 patients). The illustration of target volume delineation was shown in Figure 1B. From 2013 till now, CTV2 was defined as a 0.5-1 cm margin surrounding CTV1, not including all nasopharyngeal mucosa within contralateral pharyngobasilar fascia, and including the bilateral prophylactically irradiated lymph drainage areas. Besides that, the delineation of other target volumes, including the bilateral prophylactically irradiated lymph drainage areas, remained the same as above mentioned (Figure 1C).
For the delineation of OARs, the temporal lobe, brain stem, spinal cord, optic chiasma, optic nerve, lens, pituitary gland, oral cavity, oropharynx, hypopharynx, parotid gland, submandibular gland, thyroid gland, middle ear, temporomandibular joint and mandible were delineated according to Radiation Therapy Oncology Group (RTOG) 0225 and the practice of our hospital . According to the standards of our hospital and the requirements of radiotherapy quality control and quality assurance, planning target volume (PTV) were created by expanding 3-5 mm from all target volumes from the head-to-foot, front-to-back, and left-to-right directions around the target volumes mentioned above to compensate for geometric uncertainties and patient movements, such as PGTVnx, PGTVrpn/nd, PCTV1, PCTVnd, and PCTV2.
From November 2003 to May 2013, the dose prescription of one-course IMRT were: PGTVnx/rpn 68 Gy/30 fractions (F), PGTVnd 62-66 Gy/30 F, PCTV1 60 Gy/30 F and PCTV2 50-54 Gy/30 F. From June 2013 to December 2017, the dose prescription of replanning IMRT were: PGTVnx/rpn 68-69.5 Gy/31-32 F, PGTVnd 68-69.5 Gy/31-32 F, PCTV1 60-61 Gy/31-32 F, PCTV2 45-47 Gy/25-26 F . Dose constraints and evaluation of OARs for patients with one- course IMRT and patients with replanning IMRT were referred to our hospital and RTOG 0225 guidelines and our previous studies[11,13], respectively. Dose constraints for the contralateral unaffected OARs: according to the distance between OARs and the target volumes, stricter constraints were imposed on the basis of the standard limits to reduce dose to the unaffected organs.
Seventy-five patients received chemotherapy. Among them, 48 patients received concurrent chemotherapy, including cisplatin weekly regimen (30 mg/m)2), 4-6 times in total and cisplatin three-week regimen (80 mg/m)2), 2-3 times in total. A total of 27 patients received IC combined with concurrent chemotherapy: IC regimen was TPF (docetaxel 60 mg/m2, cisplatin 60 mg/m2, fluorouracil 500 mg/m2 continuous venous perfusion for 120 h), 3-4 times in total; concurrent chemotherapy was a weekly regimen containing platinum, including cisplatin (30 mg/m2), nedaplatin (30 mg/m2), 4-6 times in total.
Evaluation of toxicities
The Common Terminology Criteria for Adverse Event (version 4.0) was used to evaluate treatment-related acute toxicities, and the Late Radiation Morbidity Scoring Criteria of RTOG was used to evaluate radiotherapy-related toxicities. Acute toxicities were defined as those occurring either during the course of IMRT and chemotherapy or within 3 months of its completion, while late toxicities were the toxicities occurred at least 3 months after radiotherapy.
Evaluation and follow-up
All patients received detailed examinations before and after treatment including medical history, physical examination, hematological examination, chest X-ray, abdominal B-ultrasound, emission computed tomography (ECT) on bone, ENS, and head and neck enhanced MRI or positron emission tomography-computed tomography (PET/CT). Local recurrence was confirmed by endoscopic biopsy and/or MRI. Distant metastases were clinically confirmed by clinical symptoms, physical examination, and imaging examinations and pathological biopsies if necessary. Patients were followed-up once every 3 months within 3 years after radiotherapy, once every 6 months after 3 years, and once every 5 years after 5 years. Whenever possible, salvage treatments (including re-IMRT, surgery, and chemotherapy) were provided for patients who developed relapse or persistent disease.
Survival was calculated as the duration from the date of pathological diagnosis to the date of death or the last follow-up. Our primary endpoints were local-recurrence free survival (LRFS) and patterns of local failure, which were classified as "in-field" if 95% recurrent tumor volume was within the 95% isodose, “marginal” if 20- 95% recurrent tumor volume was within the 95% isodose or “out-of-field” if <20% recurrent tumor volume was within the 95% isodose. The local recurrence patterns were determined by re-delineating the tumor volume detected by MRI at the time of recurrence in the original plan evaluation system, and the relationship between the recurrence area and the original dose coverage was analyzed. Our secondary endpoints were overall survival (OS), regional relapse-free survival (RRFS), distant metastasis-free survival (DMFS) and progression-free survival (PFS).
Continuous data followed a normal distribution were analyzed in a t-test, while Wilcoxon rank sum test was used for non-normal distribution data when comparing two groups. LRFS, OS, RRFS, DMFS and PFS were estimated by Kaplan-Meier survival analysis. Statistical analysis and the survival curve were generated by R Studio version 1.2.1335. Bilateral test P<0.05 was considered statistically significant.