Efficacy and safety of schedule-modified oral capecitabine/S-1 as maintenance therapy for stage N3 nasopharyngeal carcinoma after definitive chemoradiotherapy: a multicentre, single-arm phase II clinical trial Authors: Xiang Lin, Linyi Huang, Jifang Zheng, Bingcan Chen, Chaobin Huang, Lisha Chen, Caizhu Pan, Qiaojuan Guo, Shaojun Lin, Jingfeng Zong Published: 2026 ABSTRACT Background and purposeN3 nasopharyngeal carcinoma (NPC) continues to demonstrate high distant metastasis rates despite aggressive treatment. This phase II trial evaluated schedule-modified oral maintenance chemotherapy with capecitabine or S-1 following definitive chemoradiotherapy in this high-risk population.MethodsIn this multicenter, single-arm study, patients with non-metastatic N3 NPC received induction chemotherapy (gemcitabine/nedaplatin) followed by concurrent definitive chemoradiotherapy (nedaplatin). Within one month post-radiotherapy, patients initiated 12 cycles of oral maintenance therapy (capecitabine 1250 mg/m² twice daily or S-1 body surface area based dosing, days 1–14, every 28 days). Primary endpoint was 2-year progression-free survival (PFS).ResultsAmong 115 enrolled patients, after 46.2 months median follow-up, 2-year PFS reached 92.0% (95% CI: 87.1–97.2). The 2-year overall survival, locoregional recurrence-free survival, and distant metastasis-free survival rates were 98.3%, 98.1%, and 92.9%, respectively. Maintenance therapy demonstrated favorable tolerability with only 7.0% grade 3 adverse events and no grade 4/5 events. Notably, 78.3% of patients completed all 12 maintenance cycles.ConclusionSchedule-modified oral maintenance chemotherapy with capecitabine or S-1 following definitive chemoradiotherapy showed promising efficacy and manageable toxicity in stage N3 NPC, supporting further investigation of this consolidation approach.Trial registration ClinicalTrials.gov, Registration number: NCT04220528, Date of registration: January 4, 2020, Retrospectively registered.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12903-026-08865-8. Introduction Nasopharyngeal carcinoma (NPC), which is thought to originate from the nasopharyngeal epithelium, has an distinct geographic and epidemiological distribution, primarily in Southern China and Southeast Asia [1–2]. Its strong etiological association with Epstein-Barr virus (EBV), coupled with genetic and environmental factors, distinguishes it from other head and neck malignancies [3]. Induction chemotherapy (IC) combined with concurrent chemoradiotherapy (CCRT) constitutes the current therapeutic backbone for locoregionally advanced NPC. The integration of intensity-modulated radiotherapy (IMRT) with chemotherapy has markedly enhanced survival rate for NPC [4–5]. The American Joint Committee on Cancer (AJCC) 8th edition classifies N3 disease by lymph nodes exceeding 6 cm in diameter or extension below the caudal border of the cricoid cartilage, as the highest nodal category, representing a distinct clinical entity with aggressive biological behavior [6]. Despite therapeutic advances, distant metastasis remains the predominant pattern of treatment failure in locoregionally advanced NPC, particularly among N3 patients [7–8]. NPC patients with N3 exhibit a higher risk of distant metastasis compared to patients with N0-2 disease [9]. Despite the incorporation of IC and CCRT, several phase III trials consistently demonstrate suboptimal distant metastasis-free survival (DMFS) in locoregionally advanced NPC including N3 disease, even with IC regimen optimization [10–11]. This unmet need underscores the imperative for effective systemic consolidation strategies. Emerging strategies to improve survival in nasopharyngeal carcinoma have increasingly focused on maintenance therapy. Metronomic chemotherapy, involving the prolonged administration of oral fluoropyrimidines such as capecitabine or S-1, has gained attention. Recent phase III trials demonstrated that adjuvant capecitabine following CCRT improved survival results among patients with locoregionally advanced NPC [12–13]. However, these trials reveal a non-negligible toxicity burden associated with oral maintenance therapy, as evidenced by grade 3 or higher adverse event rates of 17% [12] and 42% [13], underscoring that current regimens pose significant tolerability challenges for many patients. Furthermore, the applicability of these findings to high-risk subgroups, particularly those with N3 disease, remains uncertain. Existing phase III trials enrolled a broad population of “locoregionally advanced” NPC, which may include patients with intermediate-risk features who derive minimal benefit from intensified therapy, potentially representing overtreatment in some cases. Thus, the therapeutic efficacy of maintenance therapy specifically for high-risk N3 NPC patients, who stand to benefit most yet endure greater treatment-related vulnerability, has not been definitively established due to their underrepresentation and insufficient subgroup analysis in previous studies. Our retrospective studies found that maintenance chemotherapy may improve the survival of patients with N3 stage NPC [14–15]. Given the lack of prospective data specifically for this highest-risk subgroup, we pioneered a novel, toxicity-adapted maintenance approach. Thus, we initiated the prospective, multicenter phase II trial evaluating 1-year oral maintenance chemotherapy (capecitabine or S-1) following radical CCRT in non-metastatic N3 NPC. This study was designed to evaluate the efficacy and safety profile of capecitabine/S-1 maintenance therapy in NPC patients with N3 disease after receiving definitive treatment and to establish maintenance chemotherapy as a feasible, toxicity-adapted strategy to mitigate distant failure in N3 population. Methods Study design and patients In this single-arm, multicentre, phase II trial, patients from Fujian Cancer Hospital (Fuzhou, China), The First Hospital of Putian City (Putian, China), Jiangxi Cancer Hospital (Nanchang, China) were enrolled. The study protocol, endpoints and statistical analysis plan were finalized and approved by the Institutional Ethics Review Board of Fujian Cancer Hospital on October 15, 2019 (No: 2019-015-03), prior to the enrollment of the first patient. This trial was registered with ClinicalTrials.gov (NCT04220528) on January 4, 2020 and performed according to the principles of the Declaration of Helsinki and Good Clinical Practice guidelines as defined by the International Conference on Harmonization. The CONSORT guideline was used in our trial. Written informed consent was obtained from all the patients before enrollment. Subjects retained the right to voluntarily withdraw at any trial stage or discontinue intervention if severe comorbidities developed during follow-up. This was an investigator-initiated trial. Eligible patients aged 18–65 years with previously untreated, histologically confirmed stage TanyN3M0 NPC according to AJCC 8th edition were enrolled. Pretreatment evaluation of N3 classification was determined by nasopharyngeal and neck Magnetic Resonance Imaging (MRI) or Computed Tomography (CT). Positron Emission Tomography/CT (PET/CT) was optional. Other key eligibility criteria were as follows: at least one measurable tumor lesion, Eastern Cooperative Oncology Group performance status score of 0 or 1, adequate bone marrow, liver, renal function and life expectancy ≥ 6 months. The key exclusion criteria were as follows: distant organ metastases; recurrent NPC; a history of previous radiotherapy, chemotherapy, or surgery to the primary tumor or lymph nodes; any severe coexisting diseases; active systemic infection; previous malignancy; pregnancy or lactation; prohibition of chronic systemic immunotherapeutic or hormonal interventions extraneous to the study. The full eligibility criteria are available in the study protocol (available in the Supplementary Information file). Procedures In the study, All eligible patients received three 21-day cycles of IC (gemcitabine 1000 mg/m2 intravenously on day 1,8 and nedaplatin 80 mg/m2 intravenously on day 2), followed by CCRT (nedaplatin 100 mg/m2 intravenously on day 1 of each cycle for 2 cycles concurrently with IMRT). Nedaplatin was chosen as an alternative platinum agent to cisplatin in this study, based on evidence from phase 3 trial conducted in endemic regions demonstrating comparable efficacy with a potentially more favorable toxicity profile, particularly concerning nephrotoxicity and gastrointestinal adverse events [16–17]. Thus nedaplatin might be beneficial for patients with stage N3 disease who required intensive treatment. A detailed description of the IMRT has been published previously [18]. Maintenance chemotherapy commenced 1 month post-radiotherapy completion. DPYD (Dihydropyrimidine dehydrogenase) testing was not mandatory in this study. Based on medical resource availability and patient tolerance, either capecitabine or S-1 was selected. For S-1, body surface area (BSA)-stratified dosing was administered twice daily (40 mg for BSA ≤ 1.25 m², 50 mg for BSA > 1.25–1.5 m², 60 mg for BSA > 1.5 m²) on days 1–14 of each 28-day cycle. Alternatively, capecitabine was given at 1250 mg/m² twice daily (total 2500 mg/m²/day) on the same schedule. Both regimens continued for 12 cycles. Maintenance treatment was discontinued upon meeting any of the following criteria: unacceptable toxicity, confirmed disease recurrence or metastasis (via histopathology, cytology, or imaging), voluntary withdrawal of consent, or meeting any other pre-specified criterion for discontinuation. Details of the capecitabine/S-1 dose modifications and supportive measures are provided in the protocol. Tumor response was evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST, version 1.1). All patients underwent response evaluation, including MRI or CT of the nasopharynx and neck, chest CT, abdominal ultrasound within 1 week before the completion of CCRT. Only patients with at least stable disease and no evidence of distant metastasis were eligible to proceed with maintenance therapy. The primary timepoint for efficacy assessment was 3–4 months after the completion of CCRT, using nasopharyngeal and neck MRI. During active treatment, participants underwent weekly clinical assessments throughout IC and CCRT, and triweekly evaluations during maintenance therapy. Longitudinal surveillance after treatment completion included chest CT and abdominal ultrasound every 3 months during years 1–3, every 6 months through year 5, and annually thereafter. Plasma EBV DNA levels were longitudinally monitored at baseline, before each cycle of induction chemotherapy, during concurrent chemoradiotherapy (prior to each nedaplatin administration), prior to each cycle of maintenance therapy and every 3 months thereafter throughout follow-up. Adverse events were graded in accordance with the NCI Common Terminology Criteria for Adverse Events (version 4.0). Outcomes The primary endpoint, 2-year PFS, was defined as the time from the completion of radiotherapy to disease progression or death from any cause. Secondary endpoints encompassed: 2-year overall survival (OS; the time from the completion of radiotherapy to the date of death), locoregional recurrence-free survival (LRFS; the time from the completion of radiotherapy to local or regional relapse or death of any cause), distant metastasis-free survival (DMFS; the time from the completion of radiotherapy to distant metastasis or death of any cause), safety profiles, and treatment compliance. Standardized endpoint definitions governed statistical analyses, with event-free patients censored at last documented contact. Patients who were lost to follow-up or were still alive without locoregional recurrence or distant metastasis at the end of this trial had their data censored at the date of last follow-up. Statistical analysis This was a single‑arm, phase II trial designed to test whether the addition of maintenance capecitabine/S‑1 could improve the 2‑year PFS rate. Previous studies indicated that that the 2-year PFS reached approximately 70% among NPC patients exhibiting N3 stage. The study was powered to detect an absolute improvement of 10% (from 70% to 80%). The primary hypothesis was pre‑specified as follows: H₀ (2‑year PFS ≤ 70%) versus H₁ (2‑year PFS ≥ 80%). Using a one‑sided α of 0.05 and a power of 80%, a minimum of 115 patients was required. Efficacy endpoints were evaluated in the efficacy population, defined as all participants receiving ≥ 1 cycle of the study medications. Patients lacking post-treatment efficacy data were excluded. Safety analyses utilized the safety population, comprising all treated participants with documented safety records, excluding those without adverse event documentation. Continuous variables were summarized as medians with interquartile ranges (IQR), while categorical data were presented as counts (percentages). Time-to-event endpoints, including PFS, LRFS and OS, were analyzed via Kaplan-Meier methodology, with corresponding 95% confidence intervals derived through Brookmeyer-Crowley method. Analyses were conducted with SPSS software, version 26.0 (IBM), and R, version 4.0.2 (R Foundation). All the statistical tests were 2-sided, with P < 0.05 considered statistically significant. Data availability Restrictions apply to the availability of datasets generated during this study due to ethical constraints safeguarding patient privacy and regulatory compliance considerations. Deidentified participant-level data supporting key findings may be accessed conditionally by contacting the corresponding author, subject to institutional review board approval and executed data transfer agreements. Results Patients and treatment Between December 2019 and February 2023, a total of 121 patients with newly diagnosed N3 NPC were screened at three centers. After excluding 6 patients (4 refused to participate and 2 did not fulfill the inclusion criteria), 115 eligible patients initiated protocol treatment (Fig. 1). All 115 participants underwent to multimodal therapy combining induction chemotherapy with concurrent chemoradiotherapy and capecitabine/S-1 maintenance. The baseline features of the patients were summarized in Table 1. The median age of participants was 48 years (IQR 36–55). The study cohort comprised 80 predominantly male patients (69.6%). Among the study cohort, 114 patients (99.1%) exhibited EBV positivity before treatment (median plasma EBV DNA 9,080 copies/mL; IQR 1795-46,050), with T-stage distribution: T1 (23/115, 20.0%), T2 (18/115, 15.7%), T3 (46/115, 40.0%), T4 (28/115, 24.3%). The cohort comprised all T-stage categories according to the AJCC 8th edition. The distribution was heterogeneous, with a predominance of T3 (40.0%) and T4 (24.3%) disease, reflecting the aggressive nature of N3 NPC. Table 1Patients’ characteristicsCharacteristicNumber of patient (%)Age (years) Median (interquartile range)48(36–55) Range18–65Gender Male80 (69.6%) Female35 (30.4%)ECOG performance status 0113 (98.3%) 12 (1.7%)Histology Undifferentiated nonkeratinized115 (100.0%) Differentiated nonkeratinized0 (0.0%)T-stage T123 (20.0%) T218 (15.7%) T346 (40.0%) T428 (24.3%)Pretreatment plasma EBV DNA level, median (interquartile range), copies/mL9080 (1795-46,050) < 400044 (38.3%) ≥ 400071 (61.7%)Nodal laterality Unilateral43 (37.4%) Bilateral72 (62.6%)WBC, median (interquartile range), ×10⁹/L,6.9 (5.3–7.9)Hemoglobin, median (interquartile range), g/L138.0 (126.0-149.0)Platelet, median (interquartile range), ×10⁹/L253.0 (206.0-296.0)ALT, median (interquartile range), U/L19.0 (14.0–33.0)Creatinine, median (interquartile range), umol/L78.0 (68.0–87.0) Table 1Patients’ characteristicsCharacteristicNumber of patient (%)Age (years) Median (interquartile range)48(36–55) Range18–65Gender Male80 (69.6%) Female35 (30.4%)ECOG performance status 0113 (98.3%) 12 (1.7%)Histology Undifferentiated nonkeratinized115 (100.0%) Differentiated nonkeratinized0 (0.0%)T-stage T123 (20.0%) T218 (15.7%) T346 (40.0%) T428 (24.3%)Pretreatment plasma EBV DNA level, median (interquartile range), copies/mL9080 (1795-46,050) < 400044 (38.3%) ≥ 400071 (61.7%)Nodal laterality Unilateral43 (37.4%) Bilateral72 (62.6%)WBC, median (interquartile range), ×10⁹/L,6.9 (5.3–7.9)Hemoglobin, median (interquartile range), g/L138.0 (126.0-149.0)Platelet, median (interquartile range), ×10⁹/L253.0 (206.0-296.0)ALT, median (interquartile range), U/L19.0 (14.0–33.0)Creatinine, median (interquartile range), umol/L78.0 (68.0–87.0) Patients’ characteristics Fig. 1Trial Profile. Following screening, 6 of 121 patients were excluded based on the eligibility criteria. A total of 115 patients initiated the study treatment, and all were included in the analyses of the efficacy and safety populations Trial Profile. Following screening, 6 of 121 patients were excluded based on the eligibility criteria. A total of 115 patients initiated the study treatment, and all were included in the analyses of the efficacy and safety populations Compliance Full completion of three induction cycles was achieved in all participants. During this phase, 5.2% (6/115) required dose reductions, predominantly for hematological toxicity. Subsequently, 100.0% (115/115) proceeded to CCRT. Among CCRT participants, 85.2% (98/115) completed 2 cycles of nedaplatin and the remainder (14.8%, 17/115) received single-cycle therapy. Treatment adherence analysis demonstrated 78.3% (90/115) of patients completed the full 1-year capecitabine/S-1 regimen. Among 25 discontinuations, disease progression accounted for 4.0% (1/25), treatment-related adverse events for 32.0% (8/25), and voluntary withdrawal for 64.0% (16/25). The specific reasons of treatment-related adverse events were persistent grade 3 leukopenia (n = 4), persistent grade 3 hand-foot syndrome (n = 2) and persistent grade 3 neutropenia (n = 2). One-level dose reductions were observed in 8 patients (7.0%, 8/115) due to grade 3 treatment-related adverse events during maintenance therapy. No two-level dose reduction was reported in all the participants. The protocol permitted temporary interruptions of up to 2 weeks per cycle for toxicity management. Such interruptions were necessary in 14 patients (12.2%), most commonly for hematologic recovery or resolution of hand-foot syndrome before resuming the next cycle. 73 patients initiated maintenance therapy with S-1, of whom 4 (5.5%) switched to capecitabine due to drug accessibility constraints. Additionally, 42 patients received primary capecitabine maintenance without drug changing. Efficacy Tumor responses in the efficacy population were evaluated according to RECIST v1.1 criteria by an independent central review committee. Objective responses in the efficacy population (n = 115) included complete response (CR) in 46.1% (53/115), partial response (PR) in 52.2% (60/115) and stable disease (SD) in 1.7% (2/115) after the completion of CCRT. By the cutoff date of July 16, 2025, the median follow-up duration was 44.3 months (IQR 34.2–53.5). After 1 year of capecitabine/S-1 maintenance therapy, the independent central review confirmed that: 3 patients experienced locoregional recurrence, 17 patients experienced distant metastasis, and 2 patients experienced death. Median PFS was not reached (95% CI NR-NR), with 2-year PFS rates of 92.0% (95% CI: 87.1–97.2) (Fig. 2A). Median overall survival (OS) was not reached (95% CI: NR-NR), exhibiting 2-year OS rates of 98.3% (95% CI: 95.9–99.9) (Fig. 2B). The 2-year LRFS and DMFS rates were 98.1% (95% CI: 95.6–99.9) and 92.9% (95% CI: 88.3–97.8) respectively (Fig. 2C and D). We specified an exploratory subgroup analysis for the primary endpoint stratified by T-stage. To maintain statistical power given the sample size, we grouped patients into T1-2 and T3-4 categories. The 2-year PFS rate was 92.5% (95% CI: 84.7–99.9) for the T1-2 subgroup and 91.6% (95% CI: 85.5–98.3) for the T3-4 subgroup. The difference was not statistically significant (P = 0.79). Prespecified exploratory subgroup analysis of 2‑year PFS by initial maintenance agent showed consistent outcomes. The 2‑year PFS rate was 87.8% (95% CI, 78.4–98.4) for the capecitabine subgroup and 92.9% (95% CI, 87.1–99.1) for the S‑1 subgroup (P = 0.026). A sensitivity analysis excluding the four patients who switched from S‑1 to capecitabine yielded a 2‑year PFS rate of 91.7% (95% CI, 86.6–96.8), consistent with the primary analysis. Fig. 2Kaplan-Meier plots of survival outcomes: (A) Progression-free survival, (B) Overall survival, (C) Locoregional recurrence-free survival, (D) Distant metastasis-free survival. Tick marks indicate censored data Kaplan-Meier plots of survival outcomes: (A) Progression-free survival, (B) Overall survival, (C) Locoregional recurrence-free survival, (D) Distant metastasis-free survival. Tick marks indicate censored data Adverse events Treatment-related adverse events of grade 1 or 2 occurred in 60.9% (70/115) of the safety population during maintenance therapy. Hematologic adverse events of any grade (> 10% incidence) included: leukopenia (41.7%, 48/115), anemia (39.1%, 45/115), neutropenia (27.0%, 31/115), and thrombocytopenia (15.7%, 18/115). Predominant nonhematologic treatment-related adverse events of any grade (> 10% incidence) were: hand-foot syndrome (36.5%, 42/115),fatigue (30.4%, 35/115), anorexia (21.7%, 25/115), weight loss (19.1%, 22/115), nausea (17.4%, 20/115), sensory neuropathy (16.5%, 19/115) and diarrhea (10.4%, 12/115). Grade 3 treatment-related adverse events affected 7.0% (8/115) of patients, with leukopenia (3.5%, 4/115), neutropenia (1.7%, 2/115) and hand-foot syndrome (1.7%, 2/115). No grade 4/5 events occurred. Comprehensive safety profiles are detailed in Table 2. When stratified by maintenance agent, the safety profiles were largely overlapping. The incidence of any‑grade adverse events was 66.7% (28/42) with capecitabine and 68.5% (50/73) with S-1, and grade 3 events occurred in 7.1% (3/42) and 6.8% (5/73) of patients, respectively. Table 2Maintenance treatment-related adverse eventsEventMaintenance therapyAny gradeGrade 1Grade 2Grade 3Any adverse event78 (67.8%)45 (39.1%)25(21.7%)8 (7.0%)Leukopenia48 (41.7%)30 (26.1%)14 (12.2%)4 (3.5%)Anemia45 (39.1%)40 (34.8%)5(4.3%)0 (0.0%)Neutropenia31 (27.0%)21 (18.3%)8 (7.0%)2 (1.7%)Thrombocytopenia18 (15.7%)18 (15.7%)0 (0.0%)0 (0.0%)Lymphopenia10 (8.7%)10 (8.7%)0 (0.0%)0 (0.0%)Hand-foot syndrome42 (36.5%)32 (27.8%)8(7.0%)2 (1.7%)Fatigue35 (30.4%)24 (20.9%)11 (9.6%)0 (0.0%)Anorexia25 (21.7%)17 (14.8%)8 (7.0%)0 (0.0%)Weight loss22 (19.1%)15 (13.0%)7 (6.1%)0 (0.0%)Nausea20 (17.4%)14 (12.2%)6 (5.2%)0 (0.0%)Sensory neuropathy19 (16.5%)17 (14.8%)2 (1.7%)0 (0.0%)Diarrhea12 (10.4%)12 (10.4%)0 (0.0%)0 (0.0%)Vomiting10 (8.7%)10 (8.7%)0 (0.0%)0 (0.0%)Aminotransferase increased9 (7.8%)9 (7.8%)0 (0.0%)0 (0.0%)Mucositis or stomatitis9 (7.8%)7 (6.1%)2 (1.7%)0 (0.0%)Creatinine increased4 (3.5%)4 (3.5%)0 (0.0%)0 (0.0%) Table 2Maintenance treatment-related adverse eventsEventMaintenance therapyAny gradeGrade 1Grade 2Grade 3Any adverse event78 (67.8%)45 (39.1%)25(21.7%)8 (7.0%)Leukopenia48 (41.7%)30 (26.1%)14 (12.2%)4 (3.5%)Anemia45 (39.1%)40 (34.8%)5(4.3%)0 (0.0%)Neutropenia31 (27.0%)21 (18.3%)8 (7.0%)2 (1.7%)Thrombocytopenia18 (15.7%)18 (15.7%)0 (0.0%)0 (0.0%)Lymphopenia10 (8.7%)10 (8.7%)0 (0.0%)0 (0.0%)Hand-foot syndrome42 (36.5%)32 (27.8%)8(7.0%)2 (1.7%)Fatigue35 (30.4%)24 (20.9%)11 (9.6%)0 (0.0%)Anorexia25 (21.7%)17 (14.8%)8 (7.0%)0 (0.0%)Weight loss22 (19.1%)15 (13.0%)7 (6.1%)0 (0.0%)Nausea20 (17.4%)14 (12.2%)6 (5.2%)0 (0.0%)Sensory neuropathy19 (16.5%)17 (14.8%)2 (1.7%)0 (0.0%)Diarrhea12 (10.4%)12 (10.4%)0 (0.0%)0 (0.0%)Vomiting10 (8.7%)10 (8.7%)0 (0.0%)0 (0.0%)Aminotransferase increased9 (7.8%)9 (7.8%)0 (0.0%)0 (0.0%)Mucositis or stomatitis9 (7.8%)7 (6.1%)2 (1.7%)0 (0.0%)Creatinine increased4 (3.5%)4 (3.5%)0 (0.0%)0 (0.0%) Maintenance treatment-related adverse events Discussion This multicenter prospective phase II trial showed that 1-year maintenance chemotherapy with capecitabine/S-1 following radical chemoradiotherapy improved survival outcomes in N3 NPC patients. To the best of our knowledge, this study is the first trial in which the efficacy and safety of capecitabine/S-1 maintenance therapy were evaluated for N3 NPC patients who have previously received definitive treatment. Our results suggested that capecitabine/S-1 maintenance therapy may provide antitumor activity with a manageable toxicity profile in N3 NPC patients, warranting further investigation in randomized controlled trials. Treatment tolerability constitutes a pivotal consideration in nasopharyngeal carcinoma management. Oral chemotherapy has the merits of lower toxicity and better acceptability compared to traditional intravenous chemotherapy [19–20]. The orally administered fluorouracil medications, capecitabine/S-1, demonstrate efficacy in the treatment of locoregionally advanced NPC. A multicenter, phase III randomized trial by Chen et al. [12] demonstrated that incorporating metronomic capecitabine as adjuvant therapy following definitive chemoradiotherapy improved 3-year PFS versus chemoradiotherapy alone in high-risk locoregionally advanced NPC patients (85.3% vs. 75.7%, P = 0.0023). A multicenter randomized clinical trial [13] established that adjuvant capecitabine administered after concurrent chemoradiotherapy conferred statistically significant improvement in PFS for locoregionally advanced NPC patients. Yu et al. [21] reported that modified TPF regimen including S-1 in combination with camrelizumab could improve CR rates with an acceptable safety profile in patients with locoregionally advanced NPC. Nevertheless, the therapeutic efficacy of capecitabine/S-1 maintenance therapy in patients with N3 NPC remained undefined, constituting a significant clinical knowledge gap. To address this unmet need, this study explored capecitabine/S-1 maintenance therapy, and our results demonstrated its promising efficacy in patients with N3 NPC. Currently, no standardized oral maintenance regimen and treatment duration exist, with notable inter-institutional heterogeneity. In a Japanese retrospective analysis of the “2 + 1 regimen” (treatment for 2 weeks and rest for 1 week) [22], 56.2% of patients discontinued S-1 due to unavoidable causes including drug-related adverse events or disease recurrence and S-1 maintenance ≥ 12 months conferred lower hazard ratios than shorter durations. Chen et al. [12] demonstrated improved 3-year PFS with metronomic capecitabine (650 mg/m² twice daily for 1 year) versus standard therapy (85.3% vs. 75.7%, P < 0.01), though grade 3 toxicities occurred in 17% of metronomic patients and only 74% patients completed the 1-year treatment period, substantially higher than observed in our protocol. Cheng et al. [23] reported the optimal duration of oral adjuvant chemotherapy (capecitabine, 1000 mg/m2 twice daily, days 1–14 or S-1, 40–60 mg twice daily, days 1–14) after CCRT was > 186 days (6 months). The maintenance beyond 12 months may not bring additional benefits. Our prior investigations [14–15] established that the “2 + 2 regimen” (treatment for 2 weeks and rest for 2 weeks) demonstrated significant clinical efficacy in N3 NPC while exhibiting a favorable safety profile characterized by exclusively grade 1–2 adverse events, with no grade ≥ 3 toxicities observed throughout the maintenance chemotherapy phase. Oral maintenance chemotherapy was recommended for ≥ 12 cycles. Therefore, this study employed the “2 + 2 regimen” for 12 cycles, balancing validated efficacy with optimized tolerability. Meanwhile, we used nedaplatin as IC and CCRT regimen rather than traditionally used cisplatin. The integration of cisplatin-based chemotherapy with radiotherapy, however, substantially amplifies treatment-limiting toxicities. Severe manifestations, including nephrotoxicity, ototoxicity, neurotoxicity, and debilitating gastrointestinal complication, frequently compromise therapeutic adherence and diminish quality of life [24]. Studies by Tang et al. [16] and Tang et al. [17] demonstrated that comparative survival analysis revealed non-inferior outcomes for nedaplatin versus cisplatin across long-term endpoints and cisplatin had a higher incidence of grade 3 and 4 auditory toxic effects including nephrotoxic effects and gastrointestinal and hearing toxic effects in NPC patients. Li et al. [25] also indicated that there was no significant difference in the long-term survival outcomes between cisplatin-based chemotherapy and nedaplatin-based chemotherapy in locally advanced NPC and the risk of nausea, vomiting and weight loss was higher in the cisplatin group. In the present study, the nedaplatin-based regimen was associated with satisfactory treatment compliance during both the induction and concurrent phases, and survival outcomes were comparable to those reported in historical cohorts utilizing cisplatin. Collectively, it is reasonable to assume that the strategic replacement of cisplatin with nedaplatin contributes to mitigating treatment-related toxicities while preserving therapeutic efficacy based on these data. Following protocol-specified adjustments, the combinatorial regimen demonstrated no unanticipated grade 3–4 treatment-related adverse events. The safety profile of capecitabine/S-1 in our cohort reflected anticipated toxicity patterns, predominantly comprising low-grade adverse events. Leukopenia manifested as the cardinal toxicity (41.7% overall incidence; grade 3: 3.5%), with anemia (39.1%), hand-foot syndrome (36.5%), and fatigue (30.4%) representing other frequent sequelae. Notably, no serious adverse events were observed in the study and these adverse events were generally manageable. The completion of all 12 maintenance chemotherapy cycles by 78.3% of patients demonstrated favorable treatment adherence and tolerability of the protocol-specified regimen. The emergence of immunotherapy has revolutionized the therapeutic landscape for NPC. As for locally advanced NPC, Liang et al. [26] reported that adjuvant camrelizumab (200 mg IV q3w for 12 months) improved 3-year event-free survival versus active surveillance (86.9% vs. 77.3%), showing a significant difference. However, the clinical implementation of immunotherapy into adjuvant therapy for N3 disease faces substantial economic barriers in China. The high cost of PD-1 inhibitors is a major barrier to patient access and adherence. This financial burden forces some eligible patients to forego immunotherapy, exacerbating health inequities in endemic regions. In contrast, oral maintenance chemotherapy with capecitabine or S-1 is substantially less expensive and may represent a more accessible adjuvant strategy in resource-limited settings. Differences in trial design, patient population, and endpoints precluded direct efficacy comparisons. However, this trial might suggest encouraging survival outcomes alongside a favorable toxicity profile for oral maintenance chemotherapy. Formal cost-effectiveness analyses are needed. This trial has several limitations that warrant careful consideration. First, the single-arm, non-randomized design inherently introduces potential selection bias. Second, to address the potential for immortal time bias, the primary endpoint of PFS was defined from the completion of radiotherapy and only patients who were progression-free at that point were included. Third, the study population was exclusively from endemic regions in China, which limits the global generalizability of the findings. Fourth, the protocol allowed the use of either capecitabine or S‑1 based on physician discretion. The study was not powered for formal non-inferiority testing between these agents. Fifth, plasma EBV DNA was collected longitudinally, but kinetic analysis was not pre‑specified, so the molecular correlates of benefit remain unexplored. Sixth, patient‑reported outcomes and formal cost‑effectiveness analyses were not incorporated, leaving gaps in the understanding of quality of life and economic value. Last, the non‑randomized design precludes efficacy comparisons with other adjuvant strategies. The conclusions need to be validated in a large randomized, multicenter, controlled trial. Conclusions In conclusion, this multicenter phase II trial demonstrated that 12-month maintenance chemotherapy with capecitabine or S-1 following radical chemoradiotherapy was associated with encouraging PFS in N3 nasopharyngeal carcinoma and safety profile was manageable. Larger randomized controlled trials were warranted to validate our findings. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3.