Survival and Robotic Approach for Pancreaticoduodenectomy: A Propensity Score-Match Study

Abstract

BACKGROUND: Robotic surgery is a burgeoning minimally invasive approach to pancreaticoduodenectomy. This study was undertaken to compare survival after robotic vs “open” pancreaticoduodenectomy for ductal adenocarcinoma using propensity score–matched patients. STUDY DESIGN: With institutional review board approval, we prospectively followed 521 patients who underwent robotic (n = 311) or open (n = 210) pancreaticoduodenectomy. Patients who underwent robotic (n = 75) or open (n = 75) pancreaticoduodenectomy were propensity score–matched by age, sex, and American Joint Committee on Cancer stage. Neoadjuvant therapy was rarely administered, and adjuvant therapy was stressed (FOLFIRINOX for patients <70 years of age and gemcitabine + nab-paclitaxel for patients >70 years of age). Data are presented as median (mean ± SD). RESULTS: Operative duration was longer and estimated blood loss and length of stay were less with robotic pancreaticoduodenectomy (421 [409 ± 94.0] vs 267 [254 ± 81.2] minutes; 307 [(150 ± 605.3] vs 444 [255 ± 353.1] mL; 7 [5 ± 5.1] vs 11 [8 ± 9.5] days; p < 0.00001 for all). There were no differences in complications (Clavien-Dindo class ≥III, p = 0.30), in-hospital mortality (p = 0.61), or 30-day readmission rates (p = 0.19). Median survival after robotic vs open pancreaticoduodenectomy was 37 vs 24 months (p = 0.08). For propensity score–matched patients, operative duration for robotic pancreaticoduodenectomy was longer (442 [438 ± 117.7] vs 261 [249 ± 67.1] minutes) and estimated blood loss was less (269 [200 ± 296.1] vs 468 [300 ± 394.9] mL), as was length of stay (7 [5 ± 5.1] vs 10 [7 ± 8.6] days; p < 0.00001 for all). There were no differences in complication rates (Clavien-Dindo class ≥ III, p = 0.31) or in-hospital mortality (p = 0.40); 30-day readmissions were fewer after robotic pancreaticoduodenectomy (7% vs 20%, p = 0.03). Median survival for the robotic vs the open approach was 41 vs 17 months (p = 0.02). CONCLUSION: Patients that underwent robotic pancreaticoduodenectomy had longer operations, less estimated blood loss, shorter length of stay, and fewer 30-day readmissions; they lived much longer than patients who underwent open pancreaticoduodenectomy. We believe that robotic pancreaticoduodenectomy provides salutary and survival benefits for reasons yet unknown.

Laparoscopic Versus Open Pancreatoduodenectomy in Patients With Periampullary Tumors

“…Other multicenter RCTs, with the participation of surgeons who fulfilled the optimal learning curve, are still needed to compare the LPD and versus OPD in terms of cost-effectiveness, long-term survival, and disease-free survival. In the recent decade, robotic pancreatoduodenectomy (RPD) has gained attention and showed lower postoperative morbidity, shorter LOS, and higher survival relative to the OPD 65,70. Furthermore, RPD seems to be more beneficial relative to LPD in terms of lower conversion rate and transfusion rate 71.…”

Is there a survival benefit of a robotic approach to pancreatoduodenectomy for pancreatic cancer?

“…A bold conclusion in a recent paper by Rosemurgy et al (8) thus warrants scrutiny. The authors present a major survival difference between patients having robotic and open pancreatoduodenectomy for pancreatic cancer, with 37 vs. 24 months survival favouring those who underwent robotic surgery.…”

“…With the abovementioned considerations, we suggest that the perception of a superiority in the robotic approach in regards to survival should be tempered, if not avoided altogether, as there are several methodological issues and limitations not accounted for in the study reported in JACS ( 8 ). The authors’ bold claim that “ …robotic pancreaticoduodenectomy provides salutary and survival benefits for reasons yet unknown ” is hence unsubstantiated.…”

“…First, PSM may be considered for observational data in which an inherent selection bias is present, in an attempt to simulate randomization (‘quasi-randomization’) or at least narrow the selection bias between groups ( 9 ). In the study ( 8 ), only age, sex, and stage were used in the PSM; stage is a post hoc variable obtained only after final pathology is available. Of note, in their preferred institutionally preferred “surgery first” approach, neoadjuvant was given 3 times as often in the open surgery group compared to robotic (13% vs. 4%; P=0.04), and also for the adjusted PSM cohort (19% vs. 11%; P=0.24) despite not statistically significant.…”

“…More importantly, the PSM did not include year of surgery; hence, the robotic cases done in the latter part of the period will be biased towards shorter follow up and, hence, fewer recurrences and reported deaths from cancer (less time at risk for recurrence). Furthermore, only 29% of the 521 patients where matched ( 8 ), hence a considerable selection bias in the results. On top of that, only one quarter of robotic cases could be matched to an open case.…”