Enlightenment of robotic gastrectomy from 527 patients with gastric cancer in the minimally invasive era: 5 years of optimizing surgical performance in a high-volume center - a retrospective cohort study.

Background: Learning curves have been used in the field of robotic gastrectomy (RG). However, it should be noted that the previous study did not comprehensively investigate all changes related to the learning curve. This study aims to establish a learning curve for radical RG and evaluate its effect on the short-term outcomes of patients with gastric cancer.

Methods: The clinicopathological data of 527 patients who underwent RG between August 2016 and June 2021 were retrospectively analyzed. Learning curves related to the operation time and postoperative hospital stay were determined separately using cumulative sum (CUSUM) analysis. Then, the impact of the learning curve on surgical efficacy was analyzed.

Results: Combining the CUSUM curve break points and technical optimization time points, the entire cohort was divided into three phases (patients 1-100, 101-250, and 251-527). The postoperative complication rate and postoperative recovery time tended to decrease significantly with phase advancement ( P <0.05). More extraperigastric examined lymph nodes (LN) were retrieved in phase III than in phase I (I vs. III, 15.12±6.90 vs. 17.40±7.05, P =0.005). The rate of LN noncompliance decreased with phase advancement. Textbook outcome (TO) analysis showed that the learning phase was an independent factor in TO attainment ( P <0.05).

Conclusion: With learning phase advancement, the short-term outcomes were significantly improved. It is possible that our optimization of surgical procedures could have contributed to this improvement. The findings of this study facilitate the safe dissemination of RG in the minimally invasive era.