Assessment of learning curve and oncologic feasibility of robotic pancreaticoduodenectomy: A propensity score‐based comparison with open approach

Abstract

Though robotic pancreaticoduodenectomy(R-PD) is gradually adopted, learning curve and its feasibility is still controversial. We analyzed our first 70 R-PD cases, comparing surgical outcomes and feasibility to those of open pancreaticoduodenectomy (O-PD). Medical records of 70 patients of R-PD and 269 patients of O-PD between 2015 and 2019 were retrospectively analyzed. Cumulative sum analysis was used to determine learning curve. Surgical outcomes were compared between early(1-35) and late cases(36-70). Additional analyses with O-PD using propensity score-matching were done. Learning curve of R-PD completed after 30 cases. Shorter operative time, lower estimated blood loss, and shorter length of stay were noted in later cases. Complication rate tended to decrease over time. In comparison with O-PD after matching, R-PD showed longer operation time(414.5minutes vs 244.7minutes; P<.001), with no differences in estimated blood loss, or length of stay. While overall complication rate was higher in R-PD(45.5% vs 21.8%; P=.010), no statistically significant difference was observed in major complication rates(23.6% vs 10.9%; P=.084). R0 rate was equivalent. Surgical performance of R-PD improved over time. Learning curve of R-PD completed after 30 cases. R-PD is a promising modality, based on comparison of perioperative and oncologic feasibilities to those of O-PD.