LEARNING CURVE FOR ENDOSCOPIC SUBMUCOSAL DISSECTION OF RECTAL TUMORS IN A WESTERN CENTER - A SINGLE-OPERATOR LEARNING CURVE ANALYSIS

Aims Colorectal endoscopic submucosal dissection (ESD) is widely implemented in many Asian countries as standard treatment of early colorectal neoplasms. However, ESD experience is still limited to a handful of centers outside Asia, leading to few reports describing the learning curve in this setting. We aimed to evaluate the learning curve of rectal ESD in a western center.

Methods Included 154 consecutive rectal ESD performed by a single endoscopist between January 2013 and October 2019. En bloc and R0 resection rate, resection speed, rate of snare-assisted ESD and complications were recorded prospectively and compared between five chronologically study periods(1^st to 4^th with 30 cases each and the 5^th with 34 cases) in order to determine the learning curve.

Results En bloc resection rate increased sharply after the first period before reaching plateau (75,0% vs 96,7% vs 96,6% vs 100% vs 100%; significant p-value between the 1^st and 2^nd groups(p=0.023)). R0 rate improved during the study(64,3% vs 70,0% vs 75,9% vs 86,7% vs 87,1%, p=0.148). ESD resection speed increased over time(median, mm²/min: 7,1 vs 13,7 vs 21,4 vs 33,2 vs 40,5, significant p-value between 1^st and 2^nd groups, 2^nd and 3^rd groups and 3^rd and 4^th group (p=0,002, p=0,018, p=0,044, respectively)). Rate of snare-assisted ESD decreased gradually(17,9% vs 10,0% vs 3,4% vs 0% vs 0%). After an initial increase, there was a non-significant decrease on complication rate(10% vs 16,7% vs 16,7% vs 6,7% vs 3,2%).

Conclusions Based on our analysis, approximately 30 procedures must be carried out to significantly increase en bloc resection rate and approximately 90 procedures are necessary to achieve R0 resection rate>75%. Resection speed increased progressively over time, particularly during the first 120 lesions. Our study shows that rectal ESD can be implemented in western countries achieving a high rate of en bloc and R0 resections with an acceptable incidence of complications.