Prognostic factors in sinonasal cancers: A multicenter pooled analysis

6092 Background: Sinonasal cancers (SC) are heterogeneous diseases. Despite multimodality treatment, overall survival (OS) remains unsatisfactory. We aimed to identify prognostic factors (PFs) associated with patterns of failure and OS in non-metastatic SC (nmSC). Methods: We retrospectively reviewed a pooled dataset of nmSC from Princess Margaret (Canada) and Catalan Institute of Oncology (Spain) treated with definitive surgery ± postop radiotherapy (RT) or RT ± chemo according to institutional protocols between 2010-2019. In squamous cell carcinoma (SCC), HPV status was tested by p16 staining, and HPV DNA (ISH or PCR) if equivocal. The primary goal was to assess the association of tumor histology, stage (by 7th edition TNM), and other clinicopathological variables on locoregional control (LRC), distant control (DC) and OS. Actuarial rates were calculated with Kaplan–Meier (KM) method. Multivariable analysis (MVA) calculated hazard ratios (aHR) adjusted for histology type, T-/N-categories, and primary treatment. Results: Out of 342 pts, median age was 62 years (y) (range 21-96), Male:Female = 212:130, 95% had ECOG 0-1. Tumor histology types were: 192 (56%) SCC (p16+: 35; p16-/untested: 157), 40 (12%) adenocarcinoma (AD), 33 (10%) sinonasal undifferentiated carcinoma or sinonasal neuroendocrine tumors (SNUC/SNEC), 28 (8%) malignant melanoma (MM), 27 (8%) esthesioneuroblastoma (ES) and 22 (6%) adenoid cystic carcinoma (ACC). Median follow up was 3.6 y (range 0.1-11.3). Three-year actuarial rates for each endpoint are included in table. The PFs for LRC by MVA were: SNUC/SNEC (vs SCC) histology, T3-4 (vs T1-2: aHR 2.4, p = < 0.01) and N+ (vs N-: aHR 1.7, p = 0.02) diseases. The PFs for DC were: MM (vs SCC) and SNUC/SNEC (vs SCC) histology, and T3-4 (vs T1-2: aHR 6.0, p = 0.01) disease. The PFs for OS were: MM (vs SCC) histology, older age (aHR 1.0, p < 0.01), T3-4 (vs T1-2: aHR 5.6, p < 0.01), and N+ (vs N-: aHR 2.5, p < 0.01) disease. There was no difference in primary surgery vs RT in any outcome endpoint (all p > 0.05). p16+ SCC had a marginally higher LRC but similar DC and OS vs p16-/untested SCC (table). Conclusions: This large multicentre cohort of nmSC shows different patterns of relapse and survival with different tumor histologies. Our results suggest that individualization of treatment and follow-up strategies by histologic type is recommended to optimize outcomes in these orphan diseases. [Table: see text]