Abstract / Summary
Robotic pancreatoduodenectomy (RPD) is one of the most technically challenging procedures in hepatopancreatobiliary surgery. Its adoption continues to grow, but the learning pathway to competency and mastery is still not fully defined. This systematic review aims to analyze the learning curve assessment methodologies, proficiency thresholds, training pathways, and influence of experience on perioperative and oncologic outcomes. This systematic review was conducted in accordance with PRISMA guidelines. PubMed/MEDLINE, Scopus and The Cochrane Library were systematically searched, from inception to June 2026. Due to large heterogeneity in study design, learning curve methodologies and outcome reporting, the data were synthesized narratively. Forty-two studies were included. The learning curve thresholds were 6-8 cases for initial competency and 250 + cases for mastery. Operative time suggested proficiency after 20-40 cases, whereas risk-adjusted results, textbook outcomes and composite quality measures suggested proficiency required 56-120 cases, with mastery only after 176-263 cases. Higher experience was associated with lower conversion rates, postoperative pancreatic fistula, delayed gastric emptying, major complications and mortality. Oncologic outcomes were stable or improved, with R0 resection rates generally exceeding 90% and lymph node harvest increasing from 17 to 26, 16 to 33, and 38 to 45.5 nodes in each phase despite increasing case complexity. The learning curve of RPD is a staged advancement and is affected by surgeon experience, institutional expertise, mentorship, training structure, and outcome definitions. Technical competence emerges after 20-40 cases, but achieving consistently high-quality outcomes and procedural mastery demands substantially broader surgical experience. Implementation of RPD is safe within competency-based assessment frameworks and structured training pathways.