Meta-analysis of recurrence pattern after resection for pancreatic cancer

Background: Initial recurrence mapping of resected pancreatic ductal adenocarcinoma (PDAC) could help in stratifying patient subpopulations for optimal postoperative follow-up. The aim of this systematic review and meta-analysis was to investigate the initial recurrence patterns of PDAC and to correlate them with clinicopathological factors. Methods: MEDLINE and Web of Science databases were searched systematically for studies reporting first recurrence patterns after PDAC resection. Data were extracted from the studies selected for inclusion. Pooled odds ratios (ORs) and 95 per cent confidence intervals were calculated to determine the clinicopathological factors related to the recurrence sites. The weighted average of median overall survival was calculated. Results: Eighty-nine studies with 17 313 patients undergoing PDAC resection were included. The weighted median rates of initial recurrence were 20·8 per cent for locoregional sites, 26·5 per cent for liver, 11·4 per cent for lung and 13·5 per cent for peritoneal dissemination. The weighted median overall survival times were 19·8 months for locoregional recurrence, 15·0 months for liver recurrence, 30·4 months for lung recurrence and 14·1 months for peritoneal dissemination. Meta-analysis revealed that R1 (direct) resection (OR 2·21, 95 per cent c.i. 1·12 to 4·35), perineural invasion (OR 5·19, 2·79 to 9·64) and positive peritoneal lavage cytology (OR 5·29, 3·03 to 9·25) were significantly associated with peritoneal dissemination as initial recurrence site. Low grade of tumour differentiation was significantly associated with liver recurrence (OR 4·15, 1·71 to 10·07). Conclusion: Risk factors for recurrence patterns after surgery could be considered for specific surveillance and treatments for patients with pancreatic cancer.