Learning curve for robotic-assisted total mesorectal excision: a multicentre, prospective study.

AIM: Robotic-assisted surgery (RAS) is becoming increasingly important in colorectal surgery. Recognition of the short, safe learning curve (LC) could potentially improve implementation. We evaluated the extent and safety of the LC in robotic resection for rectal cancer. METHOD: Consecutive rectal cancer resections (January 2018 to February 2021) were prospectively included from three French centres, involving nine surgeons. LC analyses only included surgeons who had performed more than 25 robotic rectal cancer surgeries. The primary endpoint was operating time LC and the secondary endpoint conversion rate LC. Interphase comparisons included demographic and intraoperative data, operating time, conversion rate, pathological specimen features and postoperative morbidity. RESULTS: In 174 patients (69% men; mean age 62.6 years) the mean operating time was 334.5 ± 92.1 min. Operative procedures included low anterior resection (n = 143) and intersphincteric resection (n = 31). For operating time, there were two or three (centre-dependent) LC phases. After 12-21 cases (learning phase), there was a significant decrease in total operating time (all centres) and an increase in the number of harvested lymph nodes (two centres). For conversion rate, there were two or four LC phases. After 9-14 cases (learning phase), the conversion rate decreased significantly in two centres; in one centre, there was a nonsignificant decrease despite the treatment of significantly more obese patients and patients with previous abdominal surgery. There were no significant differences in interphase comparisons. CONCLUSION: The LC for RAS in rectal cancer was achieved after 12-21 cases for the operating time and 9-14 cases for the conversion rate. RAS for rectal cancer was safe during this time, with no interphase differences in postoperative complications and circumferential resection margin.