Prognostic impact of facial nerve resection in patients treated for a primary parotid cancer abutting the facial nerve without preoperative paralysis: A multicentric study of the REFCOR group with propensity score matching analysis

Primary cancers of the parotid gland are associated to a preoperative facial paralysis in 15-30% of cases: in this situation, most authors agree to resect the facial nerve (FN) along with the tumor. In patients without preoperative facial palsy, the surgical management of tumors abutting the FN remains more controversial. The aim of this study was to assess the prognostic impact of FN resection in such patients. This was a retrospective study based on a multicentric national database (REFCOR). All patients treated surgically for a primary cancer of the parotid gland abutting the FN without preoperative facial paralysis were included. The 5-year overall survival (OS) and disease-free survival (DFS) of the patients treated with or without FN resection were assessed and compared using a propensity score (PS) matching based on known confounders to address selection bias. Different PS models and PS methods were used to check the robustness of the results. Missing data were addressed using multiple imputation with chained equations. A total of 207 patients treated for a tumor abutting the FN without preoperative FN paralysis were included in the analysis, of whom 93 (45%) had a FN resection. The 5-year OS were 59% and 79.5% in patients treated with or without FN resection, respectively. The 5-year DFS were 40.3% and 56.4% in patients treated with or without FN resection, respectively. After PS matching based on 59 patients in each group, there was no difference between groups in terms of OS (HR=1.39, p=0.4) or DFS (HR=1.22, p=0.5). Similar results were found whichever the sensitivity analysis. In this retrospective study, FN resection was not associated with better survival outcomes in patients treated for a parotid cancer abutting the FN without preoperative facial paralysis, even after a PS matching that corrected potential confounding-by-indication bias.