Survival outcomes following salvage surgery for recurrent and residual head and neck squamous cell carcinoma after chemoradiotherapy.

Abstract

e18066 Background: Recurrent or residual head and neck squamous cell carcinoma (HNSCC) after chemoradiotherapy (CRT) is associated with poor prognosis with the median overall survival (OS) seldom exceeding 12 months. Salvage surgery remains the principal curative option; however, long-term outcome data from the Indian context are limited. This study assesses survival outcomes and prognostic determinants following salvage surgery after CRT failure. Methods: We performed a retrospective analytical study of 129 HNSCC patients who underwent salvage surgery after definitive or adjuvant CRT from 2010 to 2015. Patients with distant metastases or second primary tumors were excluded. Primary and secondary endpoints were overall survival (OS) and disease-free survival (DFS). Survival was evaluated using Kaplan–Meier analysis and Cox proportional hazards modeling to validate the Perioperative Prognostic Score (PPS). Results: The cohort was 88.4% male, with 95.3% presenting with Stage III/IV disease. Median follow-up was 78.4 months. Median OS was 50.4 months (5-year OS: 42.3%) and median DFS was 32.5 months (5-year DFS: 29.2%). Multivariate analysis identified female sex as an independent protective factor for OS (p=0.023) and DFS (p=0.011). The pre-operative score was a strong predictor; a Level 3 score conferred a 7.8-fold higher mortality risk (p=0.023) and a 10.4-fold increased recurrence risk (p=0.005). Carcinoma of Unknown Primary (CUP) significantly predicted improved survival (median OS: 115.98 months; 80% 5-year OS). Stage IV disease correlated with reduced DFS (p=0.035), whereas age, ECOG performance status, prior treatment intent, and reconstruction type were not significant survival predictors. Conclusions: Salvage surgery offers meaningful long-term survival in CRT-failed HNSCC, with a median OS of 50.4 months. Sex and pre-operative scoring emerged as independent outcome determinants, while traditional clinical staging showed limited prognostic discrimination. These findings indicate a marked “fit-patient” selection bias among salvage candidates, suggesting conventional models lose predictive value in physiologically robust cohorts. Therefore, risk stratification should shift toward detailed biological and functional markers to optimize patient selection.