The Road to Technical Proficiency in Cytoreductive Surgery for Peritoneal Carcinomatosis: Risk-Adjusted Cumulative Summation Analysis

Background: Cytoreductive surgery (CRS) is a technically demanding procedure, and there is considerable debate about its safe application. This study investigated the learning curve for CRS and the clinical outcomes of consecutive patients treated by a single surgeon at a single institution. Methods: We collected 251 consecutive patients who underwent CRS for peritoneal metastases by a single surgeon at Fondazione Policlinico Universitario A. Gemelli IRCCS, between January 2016 and December 2020. The learning curve was estimated using the cumulative summation analysis (CUSUM) for operative time (OT). Risk-adjusted CUSUM (RA-CUSUM) charts were developed using a composite variable (surgical failure), defined as the occurrence of at least one of the following events: major postoperative complications (Clavien-Dindo grade >= 3), blood loss >= 500 mL, incomplete cytoreduction. Three learning phases were thus derived from the RA-CUSUM analysis, and were compared in terms of perioperative outcomes. Results: CUSUM-OT showed that the operation time improved significantly after the 161(th) case. RA-CUSUM analysis allowed to break the CRS learning curve into three different phases: phase 1, "the learning phase " (cases 1-99), phase 2 "the experienced phase " (cases 100-188), and phase 3, "the mastership phase " (cases 189-251). The rate of major postoperative complications decreased significantly over the three phases (p = 0.019). Operative time decreased significantly as well (p = 0.031) and was significantly shorter in phase 3 with respect to the other two phases (phase 3 vs phase 2: 420 min vs 500 min, p = 0.017; phase 3 vs phase 1: 420 min vs 503 min, p = 0.021). Blood loss consistently decreased throughout the three phases (p = 0.001). The rate of incomplete cytoreduction was significantly lower in phase 3 than in phase 2 (4.8% vs 14.6%, p = 0.043). Conclusion: The CRS failure rate stabilized after the first 99 cases, and the complete surgical proficiency was achieved after 189 cases. A standardised and mentored learning model is a safer strategy to shorten the learning process, to reduce morbidity and mortality, to improve oncologic outcomes.