RESUMO
INTRODUCTION: There are limited numbers of high-volume centers performing minimally invasive pancreatoduodenectomy (MIPD) routinely. Several approaches to MIPD have been described. Aim of this analysis was to show the learning curve of three different approaches to MIPD. Focus was on determining the number of cases necessary to obtain proficient level in MIPD. PATIENTS AND METHODS: Retrospective study wherein outcomes of 300 consecutive patients at three centers-at each center the initial 100 consecutive patients undergoing MIPD for malignant and benign tumors of the head of the pancreas and perimpullary area, performed by three experienced surgeons were collected and analyzed. RESULTS: Overall, 300 patients after MIPD were included: the three different cohorts (laparoscopic n = 100, hybrid n = 100, robotic n = 100). CUSUM analysis of operating time in each center demonstrated that the plateau for laparoscopic PD was n = 61, for hybrid PDes was n = 32 and for robotic PD was n = 68. Median operative time for laparoscopic, hybrid, and robotic approaches was 395 min, 404 min, 510 min, respectively. Intraoperative blood loss for laparoscopic PD, hybrid PD, and robotic PD was 250 ml, 250 ml, and 413 ml, respectively. Delayed gastric emptying occurred 12% in laparoscopic cohort, 10% in hybrid, and 53% in robotic cohort. Major complications (Clavien-Dindo III/IV) rate for laparoscopic PD, hybrid PD, and robotic PD was 32%, 37%, and 22% with 5% death in each cohorts, respectively. CONCLUSION: This analysis of the learning curve of three European centers found a shorter learning curve with hybrid PD as compared to laparoscopic and robotic PD. In implementation of a MIPD program, a stepwise approach might be beneficial.