The medical records of 538 patients with sinonasal cancer who received treatment between January 2005 and March 2023 at a single referral institution were reviewed. The inclusion criteria were as follows: a) age ≥20 years at diagnosis, b) histologically confirmed SCC, c) locally advanced stage (cT3–4b, cN0–3), and d) received definitive treatment. Patients with distant metastasis at diagnosis, who received non-curative treatment, or were lost to follow-up were excluded. Finally, a total of 126 patients were included in the analysis. This study was approved by the Institutional Review Board of Yonsei University College of Medicine, Seoul, Republic of Korea (approval no. 4-2023-1337), which waived the requirement for informed consent owing to the retrospective nature of the study.

All patients underwent staging evaluation using computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography/computed tomography, if available. Initial treatment plan was decided by a multidisciplinary team comprising radiologists, otorhinolaryngologists, medical oncologists, and radiation oncologists on an individual basis considering patients’ medical fitness and tumor resectability. Medically fit patients with resectable tumor underwent upfront surgery, while those with borderline resectable or unresectable tumor underwent either definitive CCRT or NAC. Medical fitness was defined as a status able to tolerate surgery under general anesthesia, considering patients’ compliance, performance status and comorbidities altogether. Among the patients who underwent NAC, those who showed response to NAC and were downstaged to resectable tumor later received surgery while those with no response underwent CCRT. Medically unfit patients received definitive CCRT ( Supplementary Figure 1 ). Since 2018, NAC has been included as a viable option in management of sinonasal SCC in our institution in order to decrease the extent of surgical resection required and reduce borderline resectable or unresectable tumors to enable resection. Resectability of tumor was decided mostly regarding T stage, and the extent of surgery and choice of an endoscopic or open approach were at the discretion of the surgeon. Patients were categorized according to their initial treatment plan as follows: NAC followed surgery, NAC followed by CCRT, definitive CCRT, and upfront surgery. The patients in the NAC group received various combinations of chemotherapy, most of which included 5-fluorouracil, cisplatin, or docetaxel. All the patients in the upfront surgery group received postoperative RT with or without concurrent chemotherapy. CCRT typically involved weekly administrations of 30–40 mg/m² cisplatin.

RT involved three-dimensional conformal or intensity-modulated RT. All patients underwent planning CT, and thermoplastic masks were used for immobilization. For patients receiving definitive RT, the gross tumor volume was defined as the primary lesion and any metastatic lymph nodes, as observed on CT or MRI. For patients undergoing postoperative RT, the gross tumor volume was defined as the surgical bed and any residual lesion. Clinical target volume (CTV) 1 was delineated to include microscopic disease by creating a 3–5 mm margin around the gross tumor volume. CTV2 was delineated by adding a second margin 5–10 mm outside CTV1, including the involved adjacent substructures. The planning target volume (PTV) was determined as a 2–3 mm uniform expansion from the CTV. In cases of elective nodal irradiation, ipsilateral level IB and II lymph nodes were commonly included; bilateral lymph nodes were included if the primary tumor crossed the midline. For patients receiving definitive RT, the median RT dose was 69.96 Gy (59.4–75 Gy in 1.8–3 Gy per fraction) and 54 Gy (30.6–64 Gy in 1.6–2.5 Gy per fraction) for PTV1 and PTV2, respectively. For patients receiving postoperative RT, the median RT dose was 64.05 Gy (48.6–70.4 Gy in 1.8–3 Gy per fraction) and 54 Gy (30.6–60 Gy in 1.6–2.5 Gy per fraction) for PTV1 and PTV2, respectively.

Follow-up imaging and laboratory analyses were performed every 3 months for the first 2 years after the initial treatment and every 3–6 months thereafter. The treatment response was assessed according to the Response Evaluation Criteria in Solid Tumors, and the objective response rate was defined as the proportion of patients with a complete or partial response.

Treatment-related toxicity was evaluated according to the Common Terminology Criteria for Adverse Events version 5.0. Acute toxicity was defined as toxicity arising during treatment or within 3 months of treatment completion, whereas late toxicity was defined as toxicity arising more than 3 months after treatment was completed.

The primary endpoints were overall survival (OS) and overall response rate. The secondary endpoints included the progression-free survival (PFS), the cumulative incidence of local failure, the cumulative incidence of distant failure, and treatment-related toxicity. OS was defined as the time to death from any cause from the date of treatment initiation, whereas PFS was defined as the time to progression or death, whichever came first, from the date of treatment initiation. The Kaplan–Meier method and the log-rank test were used for survival analyses. The cumulative incidence of local and distant failures was estimated using Gray’s test. P < 0.05 was considered statistically significant. SPSS Statistics for Windows software version 25.0 (IBM Corp., Armonk, NY, USA) and R software version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria) were used for statistical analyses.

Patient characteristics and treatment regimens are summarized in Table 1 . The number of patients who received NAC followed by surgery, NAC followed by CCRT, definitive CCRT, and upfront surgery was 14 (11.1%), 15 (11.9%), 42 (33.3%), and 54 (42.9%), respectively. The most frequently used chemotherapy regimen for NAC was a combination of docetaxel and cisplatin, followed by a combination of 5-fluorouracil, cisplatin, and docetaxel. The median number of NAC cycles was three (range, 1–6), and nine (30%) patients needed a 20–25% dose reduction. For patients receiving CCRT, weekly cisplatin was the most frequently used chemotherapy regimen. Among the 14 patients who underwent NAC followed by surgery, 11 (78.6%) received postoperative CCRT, and three (21.4%) received postoperative RT alone. Among the 54 patients in the upfront surgery group, 21 (38.9%) received postoperative CCRT, and 33 (61.1%) received postoperative RT alone. Of the 32 patients who received postoperative CCRT following either upfront surgery or NAC and surgery, 28 (87.5%) received 30–40 mg/m² cisplatin every week, and one (3.1%) received 80 mg/m² cisplatin every 3 weeks. One patient who did not respond to cisplatin during NAC received cetuximab. The most common primary tumor site was the maxillary sinus (n = 90, 71.4%), and most patients had N0 disease (n = 101, 80.2%). A higher proportion of patients in the upfront surgery group than in the other two groups had stage T3 and N0 disease; there were no differences in age or sex among the groups.

Supplementary Figure 2 shows the initial treatment according to the T stage. The majority of patients with stage T3 disease underwent upfront surgery; the proportion of patients receiving NAC or definitive CCRT increased in those with stage T4 disease. One patient with stage T4b disease who received NAC failed to receive further definitive local treatment because of chemotherapy-induced grade 3 hematological toxicity.

Patients were followed up for a median period of 25.6 months. The 2-year OS rates of patients in the NAC, definitive CCRT, and upfront surgery groups were 59.7%, 57.4%, and 81.3%, respectively (P = 0.004) ( Figure 1A ). Similarly, the 2-year PFS rates were 35.6%, 37.4%, and 60.6% in the NAC, definitive CCRT, and upfront surgery groups, respectively (P = 0.075) ( Figure 1B ). The 2-year local failure rates were 66.2%, 60.1%, and 33.9% (P = 0.019; Figure 1C ) and the 2-year distant failure rates were 46.7%, 44.7%, and 22.8% (P = 0.003; Figure 1D ) in the NAC, definitive CCRT, and upfront surgery groups, respectively. No significant differences were observed in the OS, PFS, local failure, or distant failure rates between the NAC and definitive CCRT groups.

Subgroup analysis revealed 2-year OS rates of 60.0% and 64.3% (P = 0.607; Supplementary Figure 3A ) and 2-year PFS rates of 26.7% and 53.6% (P = 0.325; Supplementary Figure 3B ) in the NAC followed by CCRT and NAC followed by surgery groups, respectively. In addition, 2-year local failure rates were 73.3% and 51.0% (P = 0.298; Supplementary Figure 3C ), and 2-year distant failure rates were 46.7% and 39.4% (P = 0.866; Supplementary Figure 3D ), in the NAC followed by CCRT and NAC followed by surgery groups, respectively.

Subgroup analysis revealed that the OS and PFS rates of the NAC followed by surgery group were comparable to those of the upfront surgery group, despite patients undergoing NAC being typically at a more advanced stage and having higher R1-2 resection rates (85.7% vs. 68.5%). The 2-year OS rates were 64.3% and 81.3% (P = 0.219; Figure 2A ) and the 2-year PFS rates were 53.6% and 60.6% (P = 0.615; Figure 2B ) in the NAC followed by surgery and upfront surgery groups, respectively. Moreover, no significant differences in local ( Figure 2C ) and distant ( Figure 2D ) failures were observed between the two groups.

Similarly, the outcomes of the NAC followed by CCRT group were comparable to those of the definitive CCRT group. The 2-year OS rates were 60.0% and 57.4% (P = 0.794; Figure 3A ), and the 2-year PFS rates were 26.7% and 37.4% (P = 0.506; Figure 3B ) in the NAC followed by CCRT and definitive CCRT groups, respectively. The cumulative incidences of local ( Figure 3C ) and distant ( Figure 3D ) failures also did not differ between the two groups.

Response rates to NAC are shown in Supplementary Table 1 . Overall, an objective response was observed in 14 (48.2%) patients, including one (3.4%) who exhibited a complete response. Patients who showed a response to NAC had slightly improved OS and significantly improved PFS compared to those who did not show a response (2-year OS rates, 77.1% vs. 43.8%, P = 0.072, Supplementary Figure 4A ; 2-year PFS rates, 56.3% vs. 16.4%, P = 0.031, Supplementary Figure 4B ). The subsequent surgery rates were 57.1% and 50.0% in those who did and did not respond to NAC, respectively (P = 0.730). The rate of subsequent surgery decreased from 70.0% in patients at stage T3 to 16.7% in patients at stage T4b (P = 0.127; Supplementary Table 2 ). Among the 14 patients who underwent surgery following NAC, six (42.9%) were confirmed to have T-stage downstaging by pathology, and one (7.1%) had a pathological complete response ( Supplementary Table 3 ).

The treatment-related toxicities are shown in Table 2 . Acute grade 3 or 4 toxicity was more frequently reported in the NAC followed by surgery and NAC followed by CCRT groups than in the other groups (NAC followed by surgery, 21.4%; NAC followed by CCRT, 20.0%; definitive CCRT, 9.5%; upfront surgery, 5.6%; P = 0.145). Neutropenia was the most frequently reported form of acute toxicity in the NAC followed by surgery group (n = 9, 64.3%), while neutropenia and mucositis were the most frequently reported acute toxicities in the NAC followed by CCRT group (n = 5, 33.3% each). In the definitive CCRT group, the most commonly reported acute toxicity was mucositis (n = 22, 52.4%), followed by dermatitis (n = 7, 16.7%) and xerostomia (n = 7, 16.7%). Osteoradionecrosis was reported in two patients, one who received NAC followed by CCRT and one who received upfront surgery followed by adjuvant CCRT.