International Outcomes and Learning Curves for Robotic Pancreatoduodenectomy

Abstract

Introduction: Robotic pancreatoduodenectomy (RPD) is increasingly being implemented worldwide with good results reported from individual expert centers. However, data from large international series on clinical outcomes of RPD and learning curves are lacking. Methods: An international retrospective cohort study including consecutive patients after RPD until 31 December 2019, providing the first overview of international outcomes. Cumulative sum (CUSUM) analysis was performed to determine the inflection points for the feasibility and proficiency learning curves. Outcomes were compared in groups based on these inflection points. Results: Overall, 2186 patients after RPD were included from 18 centers in 8 countries. The median operative time was 360 minutes [IQR 285-460], conversion rate 3.9% (n=85), and median blood loss 200 mL [100-380]. The rate of major morbidity was 17.7% (n=386), POPF grade B/C rate 14.3% (n=311) and 30-day or in-hospital mortality 1.7% (n=38). Median hospital stay was 16 days [IQR 12-24] and the R0 resection rate in pancreatic cancer 79.9% (700/876 patients). The first inflection point for feasibility learning curve was after 30 procedures and for the proficiency learning curve after 90 procedures. Although these cut-offs did impact major morbidity (24.7%, 23.4%, and 12.3%, p<0.001) they did not impact in-hospital mortality (2.1%, 2%, and 1.5%, p=0.67). The Miami guideline annual volume advice of >20 RPDs was met by 13/18 (72%) centers. Conclusions: This first international study on RPD reported good outcomes in experienced centers and provides a reference point for the feasibility and proficiency learning curves which could be used in future randomized trials.