Meta‐analysis of the influence of a positive circumferential resection margin in oesophageal cancer

Abstract

Background The evidence regarding the prognostic impact of a positive circumferential resection margin (CRM) in oesophageal cancer is conflicting, and there is global variability in the definition of a positive CRM. The aim of this study was to determine the impact of a positive CRM on survival in patients undergoing oesophagectomy for oesophageal cancer. Methods A systematic review and meta‐analysis was performed. PubMed and Embase databases were searched for articles to May 2018 examining the effect of a positive CRM on survival. Cohort studies written in English were included. Meta‐analyses of univariable and multivariable hazard ratios (HRs) were performed using both Royal College of Pathologists (RCP) and College of American Pathologists (CAP) criteria. Risk of bias was assessed using the Newcastle–Ottawa Scale. Egger regression, and Duval and Tweedie trim‐and‐fill statistics were used to assess publication bias. Results Of 133 studies screened, 29 incorporating 6142 patients were finally included for analysis. Pooled univariable HRs for overall survival in patients with a positive CRM were 1·68 (95 per cent c.i. 1·48 to 1·91; P < 0·001) and 2·18 (1·84 to 2·60; P < 0·001) using RCP and CAP criteria respectively. Subgroup analyses demonstrated similar results for patients by T category, neoadjuvant therapy and tumour type. Pooled HRs from multivariable analyses suggested that a positive CRM was independently predictive of a worse overall survival (RCP: 1·41, 1·21 to 1·64, P < 0·001; CAP: 2·37, 1·60 to 3·51, P < 0·001). Conclusion A positive CRM is associated with a worse prognosis regardless of classification system, T category, tumour type or neoadjuvant therapy.