Learning robotic distal pancreatectomy: the force awakens

摘要

Xu et al[1] report on an impressive 10-year single center experience of 301 consecutive patients undergoing robotic distal pancreatectomy (RDP). The study was conducted in a high-volume pancreatic center performing around 1000 pancreatic surgeries/y. Interestingly, the first 264 RDP were performed by a single surgeon, the next 37 surgeries by a second generation of 5 different surgeons. The study focused on different aspects of the learning curve and used cumulative sum analysis to assess the learning curve, while dividing the phases of the learning curve according to a recent systematic review into competency, proficiency, and mastery (Fig. 1).[2,3]Figure 1.: Three phase model of the learning curve with respective case selection and expected outcomes.Evaluating the parameters operative time and complications they found the completion of the first learning phase (competency) after 70 to 75 procedures, while the proficiency phase was completed after 120 to 140 cases and the mastery phase after 220 to 245 cases. For an accurate learning curve analysis, an adequate sample size is of paramount importance, as the results of the learning curve analysis can be biased and underestimated by small sample sizes.[2] The study by Xu et al is the largest published series on RDP allowing a profound analysis of the learning curve, while previous reports were often limited by a smaller sample size of 11 to 100 cases.[4,5] The fact that this is a single institution learning curve essentially from a single surgeon make the results of the study even more valuable. As shown by Al Abbas et al,[6] second- or third-generation surgeons showed decreased learning curves with mentorship and a proficiency-based curriculum for RDP in a high-volume center. In their study, the inclusion of less experienced surgeons did not negatively affect clinical outcomes such as postoperative pancreatic fistula (POPF) and major complications. Second-generation surgeons had shorter operation times (OT) and less blood loss, showing that later generations benefited from the hands-on mentorship from the first-generation surgeons and an advanced institutional learning curve.[6] The clinical outcomes of the study population are excellent with all reported parameters—conversion rate, OT, major complications, and POPF—within benchmark values for RDP (Table 1).[7] In future studies, the division of the cohort in low-risk (benchmark) and high-risk (non-benchmark) patients according to comorbidities and tumor characteristics could be a valuable additional information that would enhance comparison of different cohorts, centers, and surgical approaches. In the currently largest published RDP series, the robotic approach offered advantages over not only open but also laparoscopic distal pancreatectomy.[2] Therefore, the observed trend of an expansion of RDP use in the United States will certainly continue internationally.[8] Table 1 - Comparison of results with benchmark values for RDP Reported result Benchmark[6] Operation time (min) 240 ≤320 Estimated blood loss (mL) 200 ≤150 Conversion to open 3% ≤3% Major complications 17% ≤26.7% Pancreatic fistula (B&C) 18% ≤31.8% ISGPS = international study group of pancreatic surgery; RDP = robotic distal pancreatectomy. The different phases of the learning curve in pancreatic surgery are characterized by a changing case selection, starting with ideal, low body-mass index patients with benign tumors and then advancing to high-risk patients with advanced tumors objectified with a high score of the Japanese difficulty scoring system.[9] During the stepwise improvement of the surgical skills along the learning curve, the outcomes should finally reach international benchmark values even in complex high-risk cases (Fig. 1). Keeping track of the individual or the institutional learning curve is an important quality measurement tool. First, the results of the learning curve can be used as an assessment metric to give individual feedback. For this purp