A Multi-Institutional Comparison of Perioperative Outcomes of Robotic and Open Pancreaticoduodenectomy

Amer H. Zureikat, MD; Lauren M. Postlewait, MD; Yuan Liu, PhD; Theresa W. Gillespie, PhD; Sharon M. Weber, MD; Daniel E. Abbott, MD; Syed A. Ahmad, MD; Shishir K. Maithel, MD; Melissa E. Hogg, MD; Mazen Zenati, MD, PhD, MPH; Clifford S. Cho, MD; Ahmed Salem, MD; Brent Xia, MD; Jennifer Steve, BS; Trang K. Nguyen, MD; Hari B. Keshava, MD; Sricharan Chalikonda, MD; R. Matthew Walsh, MD; Mark S. Talamonti, MD; Susan J. Stocker, LPN; David J. Bentrem, MD; Stephanie Lumpkin, MD; Hong J. Kim, MD; Herbert J. Zeh, III, MD; David A. Kooby, MD, FACS


Annals of Surgery. 2016;264(4):640-649. 

In This Article

Abstract and Introduction


Objectives: Limited data exist comparing robotic and open approaches to pancreaticoduodenectomy (PD). We performed a multicenter comparison of perioperative outcomes of robotic PD (RPD) and open PD (OPD).

Methods: Perioperative data for patients who underwent postlearning curve PD at 8 centers (8/2011–1/2015) were assessed. Univariate analyses of clinicopathologic and treatment factors were performed, and multivariable models were constructed to determine associations of operative approach (RPD or OPD) with perioperative outcomes.

Results: Of the 1028 patients, 211 (20.5%) underwent RPD (4.7% conversions) and 817 (79.5%) underwent OPD. As compared with OPD, RPD patients had higher body mass index, rates of prior abdominal surgery, and softer pancreatic remnants, whereas OPD patients had a higher percentage of pancreatic ductal adenocarcinoma cases, and greater proportion of nondilated (<3 mm) pancreatic ducts. On multivariable analysis, as compared with OPD, RPD was associated with longer operative times [mean difference = 75.4 minutes, 95% confidence interval (CI) 17.5–133.3, P = 0.01], reduced blood loss (mean difference = -181 mL, 95% CI -355–(-7.7), P = 0.04) and reductions in major complications (odds ratio = 0.64, 95% CI 0.47–0.85, P = 0.003). No associations were demonstrated between operative approach and 90-day mortality, clinically relevant postoperative pancreatic fistula and wound infection, length of stay, or 90-day readmission. In the subset of 522 (51%) pancreatic ductal adenocarcinomas, operative approach was not a significant independent predictor of margin status or suboptimal lymphadenectomy (<12 lymph nodes harvested).

Conclusions: Postlearning curve RPD can be performed with similar perioperative outcomes achieved with OPD. Further studies of cost, quality of life, and long-term oncologic outcomes are needed.


Reports of minimally invasive pancreaticoduodenectomy (PD) are increasing in the literature. The first description of a laparoscopic approach to PD was published in 1994,[1] and the approach was relegated to a small cohort of early surgical adopters, owing to the complexity of the operation and superior skills needed to perform this procedure totally laparoscopically. Robotic PD (RPD) was first described in 2003.[2] Proponents of the robotic approach to PD cite various potential advantages over the laparoscopic approach, including improved ergonomics, enhanced dexterity, and the addition of stereotactic vision, allowing surgeons competent in both pancreatic surgery and minimally invasive techniques to adopt this method. Despite reports of safety and oncologic efficacy of RPD, existing studies experience various limitations such as small sample sizes due to their single institutional origin, biased case selection, lack of adequate control groups, and the inclusion of cases performed within the learning curve of the procedure.[3–8]

Critics of RPD underscore a lack of rigorous data to support safety and efficacy of RPD, and the high cost of the robotic platform. Two recent National Cancer Data Base assessments of open and minimally invasive PD, reported an increase in 30-day mortality associated with minimally invasive PD over open PD (OPD), without demonstrating benefits in percentage of patients receiving adjuvant chemotherapy or postoperative time to initiation of adjuvant chemotherapy for the cohort of patients with pancreatic cancer.[9,10] These data experience heterogeneity of surgeon and center experience with RPD and include a hybrid of minimally invasive approaches, and data from centers working through their procedural learning curves. A correlative relation between PD volume and outcomes is well established[11–13] Similarly, the effect of a surgeon's learning curve on outcomes of open (50–80 cases) and RPD (80 cases) is also now recognized[14–17] These limitations must be addressed before an accurate comparison of OPD and RPD outcomes can be formulated.

The purpose of the present study was to compare the perioperative outcomes of postlearning curve open and RPDs in a large cohort of patients using data from multiple institutions with different preferred approaches to the operation.