Materials and methods
A retrospective analysis was conducted of patients with salivary gland malignancies treated at a London teaching hospital. We searched the pathology, radiation therapy and surgery databases for all major and minor salivary gland neoplasms treated by surgery from 2004 to 2019. We included only the malignant salivary gland tumours treated by primary curative-intent surgery. We excluded benign neoplasms, malignant tumours treated by palliative intent, and patients with fewer than six months of follow-up. Clinical details, pathology, treatment and survival data were obtained from the electronic and paper medical records. Survival data were enriched in some cases by contacting GP surgeries, referring hospitals and patients or their families. There were a few patients who had a part of the definitive surgery done at another centre, and who were subsequently referred to us for completion surgery and/or neck dissection. The date of initial surgery was used for the survival analysis in these patients.
SPSS version 21 (IBM Corp®) was utilized for the statistical analysis. We assessed the distribution of various clinicopathological factors, such as age, gender, site, margin status, pT classification, pN classification, extra-nodal extension (ENE), lymphovascular invasion (LVI), perineural invasion (PNI), TNM stage of disease (AJCC, 8th Ed., 2017), and adjuvant radiation therapy. Tumours were grouped into higher-grade (carcinoma ex pleomorphic adenoma, salivary duct carcinoma, high-grade mucoepidermoid carcinoma, intermediate and high-grade adenoid cystic carcinoma, adenocarcinoma, poorly-differentiated carcinoma, squamous cell carcinoma, neuroendocrine carcinoma, spindle cell carcinoma and Merkel cell tumour) and lower-grade (acinic cell carcinoma, mammary analogue secretory carcinoma, low- and intermediate-grade mucoepidermoid carcinoma, low-grade adenoid cystic carcinoma, epithelial myoepithelial carcinoma). Margins ≥ 5 mm were considered negative, 1-<5 mm were considered close, and <1 mm were considered positive. The survival analysis was conducted by the Kaplan-Meier method, and comparison between groups was determined with the log rank test. A multivariate analysis was undertaken using Cox regression to assess the impact of clinicopathological factors on overall survival (OS) and disease-free survival (DFS).