Impact of subcentimeter margin on outcome after hepatic resection for colorectal metastases: A meta-regression approach

Background. The optimal margin width and its influence on outcomes after hepatic resection for colorectal liver metastases is still controversial: a meta-analysis was conducted to analyze the impact of subcentimeter margin width on patient and disease-free survival after resection. Methods. A systematic search was performed, covering the last decade, following the Meta-analysis Of Observational Studies in Epidemiology guidelines. Relative risks (RRs) for patient and disease-free survival (DFS) were calculated after resection in relationship to a margin width > 1 cm (R0 > 1 cm) and between 1 mm and 1 cm (R0 < 1 cm) using the DerSimonian and Laird random-effects model. Meta-regression was applied for covariate adjustment. Results. Eleven observational studies were identified involving 2823 patients. Overall, 59.1% of patients were R0 < 1 cm and 40.9% were R0 > 1 cm. Meta-analysis showed that compared with patients with margins R0 > 1 cm, a R0 < 1 cm margin lead to decreased 1-, 3-, and 5-year DFS with a RR of 1.17 (95% confidence interval [CI] 1.07-1.27), 1.38 (95% CI 1.16-1.65), and 1.55 (95% CI 1.25-1.91), respectively, but patient survival was obviously affected (P > .05 in all cases). Patients with margins of R0 < 1 cm differ from those with R0 > 1 cm for greater proportions of multiple metastases (RR 1.43; 95% CI 0.25-1.61) and synchronous bowel disease (RR 1.42; 95% CI 0.8-1.92). Meta-regression showed that these two covariates had a significant impact on DFS but not on patient survival. Conclusion. A resection margin width > 1 cm is desirable even if patient survival is at best only slightly affected by subcentimeter margin as a consequence of a decreased DFS. The presence of multiple metastases and synchronous bowel neoplasm represent potential study selection biases that significantly decrease DFS; well-conducted, matched analyses consequently are essential to clarify the issue. (Surgery 2012;151:691-9.)