Minimally Invasive Pancreaticoduodenectomy: What is the Best "Choice"? A Systematic Review and Network Meta-analysis of Non-randomized Comparative Studies

Many mini-invasive pancreaticoduodenectomy (MIPD) techniques have been reported, but their advantages with respect to an open technique (OPD) and with respect to each other are unclear. A systematic literature search of studies comparing different types of MIPD was carried out: laparoscopic-assisted (LAPD), totally robotic (TRPD), totally laparoscopic (TLPD) or totally laparoscopic-robotic assisted (TLPD-RA) to OPD. The primary endpoint was postoperative mortality. The secondary endpoints were intraoperative, postoperative and oncological outcomes. A network meta-analysis was designed to generate direct, indirect and mixed estimate effects, between different approaches, for each variable. The effects were reported as pairwise comparisons and hierarchical ranking as to each approach could be the best or the worst for each outcome, expressed by the surface under the cumulative ranking curve. Twenty studies were identified, involving 2759 patients: 1813 OPDs, 81 LAPDs, 505 TRPDs, 224 TLPDs and 136 TLPD-RAs. No differences regarding postoperative mortality were found in pairwise comparison. The LAPD technique had a high probability of being the worst approach, while TRPD had a high probability of being one of the best. Regarding the secondary endpoints, OPD was the best regarding operative time and postoperative bleeding, but the worst regarding blood loss and wound infection. The TRPD or TLPD-RA techniques seemed to be the best for delayed gastric emptying, length of hospital stay, harvested lymph nodes and postoperative morbidity. The TLPD technique was often the worst approach, especially for overall and major complications, postoperative bleeding and biliary leak. The safest MIPDs are those involving a robotic system which seems to have a promising role in ameliorating the outcomes of OPD, especially when compared to a laparoscopic approach.