Learning curve of laparoscopic and robotic pancreas resections: a systematic review

Background Minimally invasive pancreatic resection has been shown recently in some randomized trials to be superior in selected perioperative outcomes compared with open resection when performed by experienced surgeons. However, minimally invasive pancreatic resection is associated with a long learning curve. This study aims to summarize the current evidence on the learning curve of minimally invasive pancreatic resection and define the number of cases required to surmount the learning curve. Methods A systematic search was performed on PubMed, Embase, Scopus, and the Cochrane database using a detailed search strategy. Studies that did not describe the learning curve were excluded from the study. Data on the method of learning curve analysis, single surgeon versus institutional learning curve, and outcome measures were extracted and analyzed. Results A total of 32 studies were included in the pooled analysis: 12 on laparoscopic pancreatoduodenectomy, 9 on robotic pancreatoduodenectomy, 12 on laparoscopic distal pancreatectomy, and 3 on robotic distal pancreatectomy. Sample population was comparable between laparoscopic pancreatoduodenectomy and robotic pancreatoduodenectomy (median 63 vs 65). Six of 12 studies and 7 of 9 studies used nonarbitrary methods of analysis in laparoscopic pancreatoduodenectomy and robotic pancreatoduodenectomy, respectively. Operating time was used as the single outcome measure in 4 of 12 studies in laparoscopic pancreatoduodenectomy and 5 of 9 studies in robotic pancreatoduodenectomy. Overall, there was no significant difference between the number of cases required to surmount the learning curve for laparoscopic pancreatoduodenectomy versus robotic pancreatoduodenectomy (laparoscopic pancreatoduodenectomy 34.1 [95% confidence interval 30.7–37.7] versus robotic pancreatoduodenectomy 36.7 [95% confidence interval 32.9–41.0]; P = .8241) and laparoscopic distal pancreatectomy versus robotic distal pancreatectomy (laparoscopic distal pancreatectomy 25.3 [95% confidence interval 22.5–28.3] versus robotic distal pancreatectomy 20.7 [95% confidence interval 15.8–26.5]; P = .5997.) Conclusion This study provides a detailed summary of existing evidence around the learning curve in minimally invasive pancreatic resection. There was no significant difference between the learning curve for robotic pancreatoduodenectomy versus laparoscopic pancreatoduodenectomy and robotic distal pancreatectomy versus laparoscopic distal pancreatectomy. These findings were limited by the retrospective nature and heterogeneity of the studies published to date.