Learning curve analysis for prophylactic bilateral robot-assisted lateral lymph node dissection for lower rectal cancer: a retrospective study.

Background: Lateral lymph node dissection (LLND) is an important surgical procedure in the treatment of lower rectal cancer (RC). However, limited data are available regarding the learning curve for robot-assisted LLND (RA-LLND). This study aimed to evaluate the learning curve for prophylactic bilateral RA-LLND for lower RC.

Methods: We retrospectively analyzed 58 consecutive patients with clinical stage II/III lower RC who had undergone prophylactic bilateral RA-LLND between July 2020 and June 2024. Cumulative sum (CUSUM) analysis was used to evaluate the learning curve for bilateral RA-LLND operative time.

Results: The mean age of patients was 61.5 years, and mean body mass index was 23.4 kg/m. The proportion of neoadjuvant therapy was 8.6%. Mean prophylactic bilateral RA-LLND operative time was 173.7 min. CUSUM analysis divided the learning curve for prophylactic bilateral RA-LLND operative time into three phases: initial learning phase (20 cases); competence phase (16 cases); and master/proficiency phase (subsequent cases). Mastery of surgical technique was achieved after performing the 36th case. Comparisons of surgical outcomes in terms of operative parameters and complications were made between phases 1 and 2 combined and phase 3. A significant reduction in mean prophylactic bilateral RA-LLND operative time was observed between phases 1 and 2 compared with phase 3 (P < 0.01). Mean blood loss was decreased in phase 3 (40.5 ml) compared to phases 1 and 2 combined (148.2 ml, P < 0.01). The frequencies of overall postoperative complications directly related to LLND and urinary dysfunction were significantly reduced in phase 3 compared to phases 1 and 2 combined (P = 0.04, and P = 0.02, respectively).

Conclusions: The three phases identified by CUSUM analysis represented characteristics of the learning curve for prophylactic bilateral RA-LLND. These data suggest that 20 cases are required for the early stage of the learning curve, whereas mastery level could be achieved after 36 cases.