Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Nowadays, adjuvant chemotherapy (AC) is administered intravenously or orally. Cisplatin plus fluorouracil (PF) and cisplatin plus gemcitabine (GP) are the preferred intravenous AC regimens, while the oral AC regimens include capecitabine and tegafur, gimeracil and oteracil potassium capsule (S-1). Nonetheless, the optimal delivery method and regimen of AC for locoregionally advanced nasopharyngeal carcinoma (LA-NPC) remain controversial owing to the lack of direct head-to-head randomised controlled trials.WHAT THIS STUDY ADDS In this retrospective study, there was no significant difference in progression-free survival (PFS) between the intravenous and oral groups (n=154 each) after propensity score matching (3-year PFS rate: 76.3% vs 73.9%, p=0.316). However, the GP regimen showed a superior 3-year PFS rate (89.1%) compared with PF (74.6%), capecitabine (76.0%) and S-1 (74.3%) regimen (p=0.005, 0.012 and 0.003, respectively), while multivariable analyses also demonstrated that the GP regimen (p=0.012) was associated with better survival.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY For patients treated with upfront CCRT, AC should be considered, and intravenous GP is preferred, although patients should be counselled about an increased risk of haematological toxicities. For patients treated with induction chemotherapy and CCRT, oral chemotherapy, either with capecitabine or S-1, was efficacious and tolerable. These findings offer valuable insights that could support clinical decisions on AC strategies for patients with LA-NPC.Introduction Nasopharyngeal carcinoma (NPC) is an epithelial malignancy that originates from the mucosal lining of the nasopharynx and is typically localised within the pharyngeal recess. It predominantly affects populations in southern China and Southeast Asia, where distant metastasis is the primary cause of cancer-related mortality. 1 Given its vague clinical manifestations, clinicians are confronted with locoregionally advanced disease in the majority of cases.2 Concurrent chemoradiotherapy (CCRT) is the mainstay of treatment for locoregionally advanced NPC (LA-NPC). The landmark Intergroup 0099 trial highlighted the substantial therapeutic advantages of cisplatin-based CCRT followed by adjuvant cisplatin–fluorouracil (PF) chemotherapy when compared with radiotherapy alone.3 Subsequent trials conducted in regions endemic to the disease consistently mirrored these outcomes, and CCRT followed by adjuvant chemotherapy (AC) was established as the standard of care for LA-NPC treatment.4–6Currently, AC is administered intravenously or orally. PF3 and cisplatin plus gemcitabine (GP)7 8 are the preferred intravenous AC regimens. The oral AC regimens include capecitabine,9 10 tegafur1 and tegafur, gimeracil and oteracil potassium capsule (S-1).11 These regimens have also been recommended in the guidelines of the Chinese Society of Clinical Oncology for the diagnosis and treatment of NPC (version 2023). In our latest phase III clinical trial,8 we were the first to assess the efficacy and toxicity of adjuvant GP chemotherapy compared with traditional adjuvant PF chemotherapy following CCRT in patients with LA-NPC. However, no studies have directly compared the effectiveness of oral AC regimens with that of GP and PF intravenous adjuvant regimens in the prognosis of LA-NPC, and it is still unknown which of these AC regimens is the best. So the optimal delivery method and regimen of AC for LA-NPC remain controversial owing to the lack of direct head-to-head randomised controlled trials. As pivotal treatment modalities for patients with LA-NPC, the adverse effects associated with different administration methods and AC regimens significantly affect patient tolerance and adherence, thereby influencing the efficacy of AC. Hence, it is imperative to compare the efficacy and toxicity of different administration methods as well as different AC regimens.Therefore, we conducted this study to explore whether distinct administration methods of AC would affect the prognosis of patients with LA-NPC, as well as to compare the survival outcomes and treatment-related toxicity profiles of four AC regimens, including PF, GP, capecitabine and S-1, in patients with LA-NPC.Materials and methods Patients Patient data were collected from the department of nasopharyngeal carcinoma at the Sun Yat-sen University Cancer Center (SYSUCC) in Guangzhou, China, from April 2017 to December 2020. This retrospective review adhered to the specific inclusion and exclusion criteria. The inclusion criteria consisted of individuals aged 18 years or older who were newly diagnosed with biopsy-proven NPC at stage III–IVa according to the eighth edition of the American Joint Committee on Cancer (AJCC) staging system and WHO type II or III disease, with an Eastern Cooperative Oncology Group performance status of 0–1, satisfactory liver and renal functions, complete treatment and laboratory information, completion of CCRT and receipt of CCRT followed by AC. The exclusion criteria included a history of malignancy, prior antitumour treatment, receipt of palliative treatment, pregnancy, lactation or severe coexisting illness. All patients were restaged according to the eighth edition of the AJCC/Union for International Cancer Control staging system.Diagnosis and treatment All patients underwent comprehensive pretreatment evaluations, including detailed patient history, physical examination, haematological and biochemical profiling, MRI or CT of the nasopharynx and neck, CT or radiography of the chest, abdominal CT or sonography and bone scintigraphy or whole-body [¹⁸F] fluorodeoxyglucose positron emission tomography-CT ([¹⁸F] FDG PET-CT). The treatment modalities included CCRT followed by AC with or without induction chemotherapy (IC). Intensity-modulated radiotherapy was administered to all patients according to the principles of our institute. 12 13 The gross tumour volume comprised the primary tumour and enlarged lymph nodes, with a prescribed dose of 68–70 Gy administered in five daily fractions per week. Platinum-based chemotherapy (100 mg/m2 every 3 weeks) was concurrently administered 1, 4 and 7 weeks after radiotherapy. Intravenous AC regimens included GP (gemcitabine (1.0 g/m2, day 1, day 8), cisplatin (80 mg/m2, day 1)) administered once every 3 weeks, starting on day 28 after the end of radiotherapy for up to two to three cycles and PF (cisplatin (80 mg/m2, day 1), 5-fluorouracil (500 mg/m2, continuous intravenous infusion for 24 hours, days 1–5)) administered once every 4 weeks, starting on day 28 after the end of radiotherapy for up to two to three cycles. Oral chemotherapy options comprised S-1 (40–60 mg two times per day, days 1–14) and capecitabine (1000 mg/m2 two times per day, days 1–14), which were administered for 14 consecutive days every 3 weeks for at least two cycles, with the dosage of S-1 adjusted based on the body surface area (BSA): 40 mg two times per day for BSA<1.25 m2; 50 mg two times per day for 1.25 m2 ≤BSA<1.5 m2; and 60 mg two times per day for BSA≥1.5 m2. Chemotherapy dose was adjusted in response to adverse events, and 28.7% of patients underwent IC, receiving regimens such as GP (gemcitabine (1.0 g/m2, day 1, day 8), cisplatin (80 mg/m2, day 1)) and TPF (docetaxel or paclitaxel, cisplatin, and 5-fluorouracil) (albumin-bound paclitaxel (200 mg/m2, day 1) or docetaxel (60 mg/m2, day 1) or liposomal paclitaxel (135 mg/m2, day 1), cisplatin (25 mg/m2/day, days 1–3) and 5-fluorouracil (500 mg/m2, continuous intravenous infusion for 24 hours, days 1–5)) for 2–3 cycles. All treatment protocols adhered to established guidelines for NPC management at the study institute.Endpoints and follow-up The primary study endpoint was progression-free survival (PFS), defined as the duration from the initiation of treatment to the first occurrence of failure at any site, death from any cause or patient censoring at the date of the last follow-up. The secondary endpoints included overall survival (OS), defined as the duration from the initiation of treatment to death from any cause; locoregional relapse-free survival (LRFS), defined as the duration from the initiation of treatment to the occurrence of locoregional relapse, death from any cause or patient censoring at the last follow-up; and distant metastasis-free survival (DMFS), defined as the duration from the initiation of treatment to the occurrence of distant relapse, death from any cause or censoring at the last follow-up. Patients underwent evaluations at least once every 3 months during the initial 3 years following treatment completion and every 6 months thereafter until death. Recurrence or metastasis was confirmed by pathology results or imaging (MRI, CT, abdominal ultrasound, whole-body bone scintigraphy, [¹⁸F] FDG PET-CT).Statistical analysis Statistical analyses were conducted using SPSS (V.26.0) and R V.4.3.0 ( www.r-project.org). Two-sided statistical tests were employed, with significance set at p≤0.05. Categorical variables were compared using χ2 and Fisher’s exact tests. Survival curves were estimated using the Kaplan-Meier method, and differences were assessed using the log-rank test. Multivariable analyses were performed using the Cox proportional hazards model with HRs and 95% CIs. Forest plots were used to illustrate the potential prognostic factors for PFS, OS, LRFS and DMFS, along with adjusted HR and 95% CI. The Common Toxicity Criteria for Adverse Events (CTCAE, V.5.0) were used to classify treatment-related toxicities. Additionally, propensity score matching (PSM) was used to ensure comparability between the groups: a logistic regression model incorporating clinically relevant covariates (age, sex, Tumor and Node (T/N) category, clinical stage, family history of NPC, Epstein-Barr virus (EBV) DNA prior to AC) generated propensity scores. 1:1 nearest-neighbour matching with a calliper width of 0.02 SD of the propensity score was executed without replacement.Patient and public involvement Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our researchResults Patient characteristics A total of 1150 patients diagnosed with NPC from 2017 to 2020 were retrospectively screened. After applying exclusion criteria such as age, clinical stage, pathological type, incomplete records and the presence of other malignancies, 680 patients were deemed ineligible for the study. Ultimately, 470 patients were included in the analysis. The cohort had a median age of 45 years (range, 18–70 years) and comprised 326 men (69.4 %) and 144 women (30.6 %). The median follow-up period was 48 months (range, 4–75 months. A total of 229 (48.7%) patients were documented in the oral administration group (127 (27.0%) patients received capecitabine and 102 (21.7%) received S-1), while 241 (51.3%) patients were recorded in the intravenous group (164 (34.9%) patients underwent the PF regimen and 77 (16.4%) underwent GP). IC was administered to 135 patients (28.7 %), whereas the remaining 335 patients (71.3 %) proceeded directly to AC without prior IC ( figure 1 and online supplemental table S1).SP310.1136/bmjonc-2024-000718.supp3Supplementary dataFigure 1Flow chart of patient selection. AC, adjuvant chemotherapy; CCRT, concurrent chemoradiotherapy; GP, gemcitabine and cisplatin; IC, induction chemotherapy; NPC, nasopharyngeal carcinoma; PF, cisplatin and 5-fluorouracil; S-1, tegafur, gimeracil and oteracil potassium capsule; SYSUCC, Sun Yat-sen University Cancer Center.Relationship of administration method and clinical outcome PSM was used to ensure an equivalent baseline in the oral and intravenous groups ( online supplemental table S1). After PSM, a cohort of 308 patients was established with an equal distribution of 154 patients in each group, as depicted in figure 1. Clinical characteristics are presented in online supplemental tables S2 and S3. Under these conditions, no significant differences were observed in the primary or secondary endpoints between the two groups. The 3-year PFS rates were 76.3% in the intravenous group and 73.9% in the oral group (HR 0.803 95% CI 0.523 to 1.233, p=0.316) (online supplemental figure S1A).SP110.1136/bmjonc-2024-000718.supp1Supplementary dataAdditionally, a multivariable Cox proportional hazards model incorporating variables such as age, sex, T category, N category, clinical stage, family history of NPC, administration method and EBV DNA level prior to AC was employed. This analysis demonstrated that the administration method (intravenous vs oral) did not independently influence the prognosis in terms of PFS, OS, LRFS and DMFS. The HR for the intravenous group relative to the oral group was 0.80, with a 95% CI of 0.52 to 1.24 for PFS (p=0.323) (online supplemental figure S2A).SP210.1136/bmjonc-2024-000718.supp2Supplementary dataRelationship of AC regimen and clinical outcome Given the absence of significant prognostic disparities between the intravenous and oral groups, we explored the influence of different AC regimens on the prognosis of LA-NPC. The baseline patient characteristics are presented in table 1. The baseline characteristics across age, sex, T category, clinical stage, family history of NPC and EBV DNA levels prior to AC were well balanced among the four AC regimen groups, except for the N category and IC. More patients in the PF and GP groups had N2–3 disease, while significantly more patients in the oral chemotherapy group received IC (p<0.0001).Table 1Baseline characteristics of patients receiving different AC regimensCharacteristicPF group (n=164)GP group (n=77)Capecitabine group (n=127)S-1 group (n=102)P valuen (%)n (%)n (%)n (%)Age, years Median (IQR)44 (36–52)43 (36–48)46 (36–52)47 (36–55)AC cycles Median (IQR)3 (2–3)3 (2–3)9 (7–17)8 (6–18)Sex0.214 Female43 (26.2)22 (28.6)40 (31.5)39 (38.2) Male121 (73.8)55 (71.4)87 (68.5)63 (63.8)T category0.139 1–221 (12.8)8 (10.4)6 (4.7)10 (9.8) 3–4143 (87.2)69 (89.6)121 (95.3)92 (90.2)N category<0.0001 0–113 (7.9)3 (3.9)42 (33.1)37 (36.3) 2–3151 (92.1)74 (96.1)85 (66.9)65 (63.7)Clinical stage0.582 III66 (40.2)36 (46.8)47 (37.0)40 (39.2) IVa98 (59.8)41 (53.2)80 (63.0)62 (60.8)Family of NPC0.388 No151 (92.1)75 (97.4)116 (91.3)94 (92.2) Yes13 (7.9)2 (2.6)11 (8.7)8 (7.8)EBV prior to AC*0.121 =0151 (92.1)75 (97.4)112 (88.2)95 (93.1) >013 (7.9)2 (2.6)15 (11.8)7 (6.9)IC<0.0001 Non-IC162 (98.8)76 (98.7)68 (53.5)29 (28.4) IC2 (1.2)1 (1.3)59 (46.5)73 (71.6)*EBV prior to AC measured 1 week after completing radiotherapy was based on a cut-off value of 0 copies per millilitre.AC, adjuvant chemotherapy; EBV, Epstein-Barr virus; GP, gemcitabine and cisplatin; IC, induction chemotherapy; N, Regional lymph node involvement; NPC, nasopharyngeal carcinoma; PF, cisplatin and 5-fluorouracil; S-1, tegafur, gimeracil and oteracil potassium capsule; T, Primary tumor extent.In this study, the median follow-up durations for patients receiving PF, GP, capecitabine and S-1 regimens were 46.5 months (range, 5–72), 41 months (range, 9–69), 50 months (range, 12–75) and 49.5 months (range, 4–74), respectively. The 3-year PFS rate for the GP regimen was 89.1%, which was significantly higher than that for the PF (74.6%), capecitabine (76.0%) and S-1 (74.3%) regimens (p=0.005, 0.012 and 0.003, respectively; figure 2A,E,I). Similarly, the 3-year DMFS rate for GP was 90.5%, outperforming that for PF (79.0%), capecitabine (82.4%) and S-1 (81.4%), as shown in figure 2D,H,L (p=0.016, 0.039 and 0.026, respectively). The 3-year LRFS rate for S-1 (83.5%) was comparable to that of PF (86.7%, p=0.212) and capecitabine (86.3%, p=0.187) but lower than that of GP (94.7%) (p=0.012, figure 2C,G,K). No significant differences were observed in the OS rates across the four AC regimens (all p>0.05), as depicted in figure 2B,F,J.Figure 2Kaplan-Meier curves. Survival outcomes in patients with locoregionally advanced nasopharyngeal carcinoma receiving adjuvant chemotherapy after concurrent chemoradiotherapy: (A–D) compare all four regimens (PF, GP, capecitabine and S-1), (E–H) focus on the intravenous regimens alone (PF vs GP) and (I–L) depict the oral regimens alone (capecitabine vs S-1); within each panel set the endpoints are, respectively, PFS (A,E,I), overall survival (B,F,J), LRFS (C,G,K) and DMFS (D,H,L). DMFS, distant metastasis-free survival; GP, gemcitabine and cisplatin; LRFS, locoregional relapse-free survival; PF, cisplatin and 5-fluorouracil; PFS, progression-free survival; S-1, tegafur, gimeracil and oteracil potassium capsule.Multivariable analysis, adjusted for age, sex, T category, N category, clinical stage, familial history of NPC, AC regimen and EBV DNA level prior to AC, demonstrated equivalent outcomes among the capecitabine, S-1 and PF regimens. Specifically, the HR for PFS between capecitabine and PF was 0.99 (95% CI 0.63 to 1.56, p=0.967) and for DMFS was 0.94 (95% CI 0.57 to 1.56, p=0.817). Similarly, comparisons between S-1 and PF showed an HR for PFS of 1.38 (95% CI 0.86 to 2.23, p=0.180) and for DMFS of 1.17 (95% CI 0.69 to 2.01, p=0.560). In contrast, GP demonstrated significantly superior outcomes compared with PF, with an HR for PFS of 0.38 (95% CI 0.18 to 0.81, p=0.012) and for DMFS of 0.42 (95% CI 0.19 to 0.96, p=0.039) as shown in figure 3.Figure 3Multivariable analyses of prognostic factors for progression-free survival (A), overall survival (B), locoregional relapse-free survival (C) and distant metastasis-free survival (D) in patients with locoregionally advanced NPC receiving PF, GP, capecitabine, or S-1 AC after concurrent chemoradiotherapy. AC, adjuvant chemotherapy; EBV, Epstein-Barr virus; GP, gemcitabine and cisplatin; NPC, nasopharyngeal carcinoma; PF, cisplatin and 5-fluorouracil; S-1, tegafur, gimeracil and oteracil potassium capsule.Across adjuvant regimens, the median number of cycles delivered was three for intravenous chemotherapy, nine for capecitabine and eight for S-1. In intravenous chemotherapy groups, the PF regimen (n=164) demonstrated a 70.1% completion rate (115/164) for the full three-cycle protocol, with 29.9% (49/164) receiving two cycles. Similarly, the GP regimen (n=77) achieved 71.4% completion (55/77) of three cycles versus 28.6% completion (22/77) of two cycles. Among those who completed at least two cycles, dose reductions occurred in 40% of PF courses and 45% of GP courses. Treatment breaks were driven chiefly by adverse events and personal or clinician-based considerations. Median treatment durations were 90 days (IQR 60–112) for PF and 69 days (IQR 42–84) for GP. For oral AC, compliance was higher with capecitabine, where 85% (108/127) of patients completed eight cycles (22% required dose reduction), whereas S-1 achieved a 65% completion rate (66/102) with 20% dose reductions and median treatment durations were 189 days (IQR 147–357) and 168 days (IQR 126–378), respectively. Together, these data underline the generally good real-world adherence to both intravenous and oral regimens while highlighting the higher cycle completion observed with capecitabine. The superior completion rate of capecitabine may reflect its favourable tolerability profile compared with other regimens (online supplemental tables S4 and S5).Toxicity Table 2 shows the differential incidences of treatment-related adverse events between the oral and intravenous groups. Notably, the intravenous group exhibited significantly higher rates of severe (grade≥3) leucocytopenia (p<0.0001), neutropenia (p=0.012), anaemia (p<0.0001) and hypokalaemia (p=0.014) than the oral group. We also analysed the acute toxicities associated with the four AC regimens (online supplemental table S6). Throughout the treatment course, the most prevalent severe adverse event was leucocytopenia, with 40.2% (66 of 164) in the PF group, 53.2% (41 of 77) in the GP group, 16.5% (21 of 127) in the capecitabine group and 25.5% (26 of 102) in the S-1 group experiencing grade 3 or worse conditions. Neutropenia was observed in 25.0% in the PF group, 31.2% in the GP group, 10.2% in the capecitabine group and 23.5% in the S-1 group. Anaemia was reported in 15.2% in the PF group, 22.1% in the GP group, 5.5% in the capecitabine group and 3.9% in the S-1 group. Hypokalaemia affected 15.9% in the PF group, 7.8% in the GP group, 7.1% in the capecitabine group and 4.9% in the S-1 group. The incidences of severe leucocytopenia, neutropenia and anaemia were notably higher in the PF and GP groups than in the capecitabine and S-1 groups, whereas hypokalaemia and hyponatraemia occurred more frequently in the PF group than in the other groups.Table 2Adverse events in patients with locoregionally advanced nasopharyngeal carcinoma after propensity score matchingToxic effectOral group (n=154)Intravenous group (n=154)P valueGrade<3Grade≥3Grade<3Grade≥3Haematological Leucopenia119 (77.3%)35 (22.7%)77 (50.0%)77 (50.0%)<0.0001 Neutropenia126 (81.8%)28 (18.2%)107 (69.5%)47 (30.5%)0.012 Anaemia148 (96.1%)6 (3.9%)129 (83.8%)25 (16.2%)<0.0001 Thrombocytopaenia147 (95.5%)7 (4.5%)142 (92.2%)12 (7.8%)0.236Non-haematological Hyponatraemia154 (100%)0149 (96.8%)5 (3.2%)0.060 Hypokalaemia147 (95.5%)7 (4.5%)135 (87.7%)19 (12.3%)0.014 Hypocalcaemia150 (97.4%)4 (2.6%)154 (100%)00.123 Total bilirubin154 (100%)0100 (100%)01 ALT increase151 (98.1%)3 (1.9%)152 (98.7%)2 (1.3%)1 AST increase152 (98.7%)2 (1.3%)153 (99.4%)1 (0.6%)1 Creatinine increase154 (100%)0154 (100%)01No grade 5 (fatal) toxicities related to treatment were observed.ALT, alanine aminotransferase; AST, aspartate aminotransferase.Sensitivity analyses In patients who did not receive IC (n=335) ( online supplemental table S7), univariate Cox regression analyses revealed regimen-dependent survival patterns. For PFS, using the PF group as reference, the GP regimen demonstrated significantly superior outcomes (HR=0.356, 95% CI 0.168 to 0.755), while capecitabine (HR=0.915, 95% CI 0.535 to 1.566) and S-1 (HR=1.211, 95% CI 0.610 to 2.407) showed no significant advantage. Similar for DMFS, the GP group maintained superiority (HR=0.384, 95% CI 0.171 to 0.862) compared with PF, whereas capecitabine (HR=0.906, 95% CI 0.503 to 1.632) and S-1 (HR=0.833, 95% CI 0.351 to 1.979) showed comparable outcomes (online supplemental table S8).Among patients with undetectable EBV DNA prior to AC (n=433), consistent trends were observed: GP regimen exhibited significantly improved PFS (HR=0.285, 95% CI 0.121 to 0.673) and DMFS (HR=0.284, 95% CI 0.111 to 0.726) relative to PF, while neither capecitabine (PFS: HR=0.897, 95% CI 0.556 to 1.448; DMFS: HR=0.868, 95% CI 0.515 to 1.465) nor S-1 (PFS: HR=1.035, 95% CI 0.635 to 1.688; DMFS: HR=0.875, 95% CI 0.503 to 1.522) demonstrated statistically significant benefits (online supplemental table S9).Discussion To the best of our knowledge, this study constitutes the first detailed comparison, affirming comparable prognostic outcomes between intravenous and oral administration of AC in patients with LA-NPC. Moreover, the GP regimen demonstrated superior PFS compared with the PF, capecitabine and S-1 regimens. However, a higher incidence of adverse events was associated with intravenous administration, as well as with the GP and PF regimens.In 1998, the 0099 trial established CCRT, followed by adjuvant PF chemotherapy, as the standard regimen for LA-NPC.3 Research in NPC-endemic regions has consistently affirmed the efficacy of CCRT combined with adjuvant PF chemotherapy, demonstrating enhanced survival outcomes. Consequently, this regimen has been widely adopted.4–6 However, a phase III trial comparing CCRT with and without adjuvant PF chemotherapy revealed no survival advantage. Importantly, this trial was not designed to be non-inferior; hence, it is inconclusive to assert that CCRT alone is not inferior to CCRT with adjuvant PF.14 Given its significant toxic side effects and limited patient tolerance, the effectiveness of adjuvant PF continues to be debated. Therefore, studies focusing on AC in high-risk patients15 have explored the efficacy of many single drugs or innovative combination drugs.Recent advancements have introduced several AC regimens, each demonstrating efficacy in the management of LA-NPC. A 2022 phase III trial demonstrated that compared with CCRT alone, adjuvant capecitabine resulted in higher rates of failure-free survival (FFS) and LRFS.9 Similarly, another prospective trial investigating metronomic capecitabine as an AC after CCRT with or without IC for LA-NPC also reported that it improved rates of FFS, OS, DMFS and LRFS compared with observation alone.10 Although these AC regimens employ different dosing strategies, both have been effective in enhancing survival outcomes, and the discrepancy in FFS between the two appears to be clinically insignificant. Furthermore, retrospective analyses have highlighted the efficacy of S-1 as an AC option, reflecting positive outcomes in LA-NPC management.16 17 A significant comparison by Zhu et al indicated that the effectiveness of capecitabine matched that of S-1 for LA-NPC treatment.18 However, these studies did not compare these regimens with traditional PF; therefore, whether they are superior to PF regimens remains unknown.Our latest phase III clinical trial was the first to evaluate the efficacy and toxicity of concurrent adjuvant GP chemotherapy versus traditional adjuvant PF chemotherapy following CCRT in patients with LA-NPC.8 These findings suggest improved PFS and manageable toxicity of GP therapy, underscoring its potential as an alternative therapeutic option for LA-NPC. Furthermore, to date, no studies have directly compared the effectiveness of oral AC regimens with that of GP and PF intravenous adjuvant regimens in the prognosis of LA-NPC, and it is still unknown which of these AC regimens is the best.Therefore, in this study, we first compared the differences in prognosis between intravenous and oral AC administration from an overall perspective and employed PSM to ensure comparability between the two groups. The findings revealed similar survival outcomes between the two groups, indicating the comparable efficacy of both administration methods (online supplemental figure S1). Notably, oral chemotherapy has demonstrated more favourable tolerability and lower toxicity than intravenous chemotherapy. Patients who received oral chemotherapy exhibited a reduced incidence of severe (grade≥3) leucocytopenia, neutropenia, anaemia, hyponatraemia and hypokalaemia (table 2).Despite no overall differences between the intravenous and oral routes of AC administration, further analyses within this study indicated that the GP regimen substantially enhanced both PFS and DMFS in patients with LA-NPC compared with the PF, capecitabine and S-1 regimens (figure 2). Among the PF, capecitabine and S-1 groups, no significant differences were observed in terms of PFS, OS, LRFS or DMFS. Multivariable analysis further supported these results, confirming that the GP regimen had superior efficacy in improving PFS and DMFS compared with PF, with no notable differences between capecitabine, S-1 and PF (figure 3).Despite the skewed distribution of the baseline N category among the treatment groups (table 1), with a higher prevalence of N2–3 category patients in the GP (96.1%) and PF (92.1%) groups, it is recognised that N2–3 is a high-risk factor that generally correlates with poorer prognoses in NPC. Nonetheless, our findings revealed no significant differences in prognosis between the PF regimen and the other two oral adjuvant regimens, suggesting a potentially greater benefit of the PF regimen over oral options. Furthermore, the prognosis associated with the GP regimen was markedly superior to that associated with the other three regimens, demonstrating its enhanced efficacy. This distinction underscores the potential of the GP regimen as an effective treatment option for patients with LA-NPC. Regarding toxicity, both the GP and PF groups experienced higher incidences of severe (grade≥3) leucocytopenia, neutropenia, anaemia, hyponatraemia and hypokalaemia, highlighting the need for careful monitoring and management of these adverse effects in clinical settings (online supplemental table S6). Although oral capecitabine achieved higher compliance—85% (108/127) completed eight cycles—and was generally well tolerated, its median treatment duration was 189 days (IQR 147–357), significantly longer than intravenous regimens (PF: 90 days, IQR 60–112; GP: 69 days, IQR 42–84). This prolonged exposure inevitably extends toxicities, whereas patients completing intravenous schedules had significantly extended recovery intervals post therapy (online supplemental tables S4 and S5).Due to the critically limited sample sizes in cohorts of patients who received IC (PF: n=2; GP: n=1) or had detectable EBV DNA prior to AC (PF: n=13; GP: n=2), these subgroups lacked sufficient statistical power for meaningful comparisons. Therefore, we conducted comprehensive sensitivity analyses in the other two subgroups: non-IC patients (n=335) and those with undetectable EBV DNA prior to AC. Notably, GP maintained significant PFS and DMFS benefits over the PF regimen in both cohorts, whereas oral agents (capecitabine/S-1) failed to show statistically meaningful improvements (online supplemental tables S8 and S9). This regimen-specific efficacy gradient, where intravenous GP outperforms both PF and oral alternatives, may suggest intrinsic superiority rather than confounding by IC or EBV DNA. However, more than 98% of patients who received IC subsequently received oral AC, whereas 71% of those without IC received GP or PF. The observed superiority of GP is driven predominantly by the non-IC population; consequently, extrapolation of these findings to patients who have undergone IC should be made with caution.In the context of NPC treatment goals, which emphasise survival improvement and treatment toxicity reduction, the selection of a chemotherapy regimen should consider both drug efficacy and patient preference. Clinicians should make their best judgement based on the available data, evaluate the risk of treatment failure and toxicity and discuss the potential risks and benefits of different treatment regimens to provide the most appropriate treatment for patients. Our study provides clinical evidence supporting the efficacy of four different adjuvant regimens for the treatment of LA-NPC. To date, no prospective clinical studies have specifically compared the efficacies of various AC regimens. This research gap highlights a significant area for future investigation that could provide valuable insights into the optimisation of treatment protocols for better patient outcomes.This study has several limitations. First, its retrospective design and single-centre setting may have introduced inherent biases, limiting the generalisability of the findings. Although we employed PSM and conducted a multivariable analysis to mitigate potential biases, residual confounding factors may still exist. Second, the optimal duration of S-1 and capecitabine chemotherapy remains unclear, and this study did not investigate the most appropriate treatment duration for these agents. Further research is needed to determine the optimal duration of S-1 and capecitabine chemotherapy for patients with LA-NPC. While some clinically relevant toxicities such as hand-foot syndrome and diarrhoea are particularly relevant to oral AC, our toxicity data were primarily derived from laboratory-based CTCAE parameters. These patient-reported outcomes were incompletely documented in retrospective medical records due to missing or fragmented data. Therefore, they were not reported in this study. Third, the moderate sample size may have limited our ability to conduct detailed stratified analyses and explore the potential interactions between variables. Future studies with larger sample sizes are warranted to validate our results and provide robust evidence. Finally, our study is a retrospective study where some patients inevitably received IC. However, our primary focus is on comparing treatment patterns following the completion of radiotherapy. The use of varying IC regimens and cycles in the present study may introduce a potential source of bias. However, a synthesis of findings from published reports indicated that there were no significant survival differences among different IC regimens,19 such as docetaxel and cisplatin versus docetaxel, cisplatin and fluorouracil versus gemcitabine and cisplatin, or between the number of IC cycles (eg, two vs three cycles). Considering these limitations, future multi-institutional prospective studies are needed to confirm our findings and address these unanswered questions, ultimately improving our understanding of the optimal AC regimen for patients with LA-NPC.Conclusions In summary, for patients treated with upfront CCRT, AC should be considered, and intravenous GP is preferred, although patients should be counselled about an increased risk of haematological toxicities. For patients treated with IC and CCRT, oral chemotherapy, either with capecitabine or S-1, was efficacious and tolerable. These findings offer valuable insights that could support clinical decisions on AC strategies for patients with LA-NPC.