Abstract
OBJECTIVE: To evaluate the safety and to efficacy of robotic-assisted duodenoduodenostomy (RAD) and laparoscopic-assisted duodenoduodenostomy (LAD) for neonatal annular pancreas and analyze its associated learning curve. METHODS: A retrospective analysis of 59 neonates undergoing minimally invasive DS (2019-2024) was performed. Patients were stratified into robotic (RAD, n = 25) and laparoscopic (LAD, n = 34) cohorts. Perioperative outcomes (operative time, anastomosis time, blood loss, drainage duration, complications) were compared. Learning curves for console docking and intracorporeal manipulation were assessed using cumulative sum (CUSUM) analysis. RESULTS: RAD demonstrated significantly shorter anastomosis time (35.6 ± 2.4 min vs. 47.0 ± 4.1 min, P < 0.05), reduced blood loss (median 4.0 ml vs. 6.0 ml, P < 0.05), and earlier drain removal (median 4.0 vs. 5.0 days, P < 0.05) compared to LAD. operative time was longer for RAD (179.3 ± 21.1 min vs. 156.9 ± 23.1 min, P < 0.05), attributed to docking time (22.2 ± 5.9 min) and the learning curve. Complication rates were comparable (16.0% vs. 26.5%, P = 0.537), with no anastomotic leaks, strictures or reoperations in either group. CUSUM analysis identified proficiency thresholds: docking time plateaued at case 13, and intracorporeal manipulation stabilized after 10 cases. Preliminary analysis within the first 20 cases suggests a 20-case threshold may be critical procedural mastery. CONCLUSION: RAD is a safe and effective approach for neonatal annular pancreas, demonstrating superior precision to laparoscopy. While requiring a longer initial operative time, RAD provides significant advantages in anastomotic efficiency and hemostasis once the learning curve is overcome.