Impact of Paranasal Sinus Invasion on Oncologic and Dosimetric Outcomes in Nasopharyngeal Carcinoma Following Intensity-Modulated Radiation Therapy—Implications for Risk Stratification and Planning Optimization

Purpose: This study aims to investigate the prognostic value and dosimetric impact of paranasal sinus invasion (PSI) in patients with nasopharyngeal carcinoma (NPC), and further to explore the feasibility of an integrative prognostic model based on anatomic, volumetric, and dosimetric features. Methods: Two hundred six patients with T3 NPC receiving intensity-modulated radiation therapy (IMRT) were retrospectively analyzed. Dosimetric parameters were calculated from dose–volume histograms. Primary gross tumor volume (GTV-P) and dosimetric parameters were categorized using optimal cutpoints determined by R. Local recurrence-free survival (LRFS) was estimated using Kaplan–Meier method. Independent risk factors for LRFS were identified through univariable and multivariable analyses by Cox proportional hazards models. Results: The incidence of PSI was 10.7% (22/206). Patients with PSI had significantly inferior 5-year LRFS (77.3 vs. 93.8%, P = 0.006). IMRT plans for patients with PSI had larger dose heterogeneity, higher frequency of underdosing, and higher maximum dose to optic structures. When categorized by optimal cutpoints, GTV-P > 38.67 cm(3) (5-year LRFS, 84.8 vs. 97.4%, P = 0.008), and V66.88 < 89.87% (5-year LRFS, 67.1 vs. 94.5%, P < 0.001) were associated with significantly worse local outcome. Multivariable analyses showed that PSI, GTV-P > 38.67 cm(3), and V66.88 < 89.87% were independent risk factors for local relapse, either in patients with or without concurrent chemotherapy. An integrative prognostic model was then established upon the cumulative score of risk factors. Subgroups with score of 0, 1–2, and 3 had distinctive local outcomes; the 5-year LRFS was 96.6, 84.7, and 58.3%, respectively (P < 0.001). Conclusions: Paranasal sinus invasion jeopardized local control in T3 NPC patients due to large tumor burden and inadequate radiation dose in GTV-P. The presence of PSI, GTV-P, and radiation underdosing combined are critical for the risk stratification of local failure.