Additional modifications to the Blumgart pancreaticojejunostomy: Results of a propensity score-matched analysis versus Cattel-Warren pancreaticojejunostomy.

Abstract

Postoperative pancreatic fistula continues to occur frequently after pancreatoduodenectomy. We have described a modification of the Blumgart pancreaticojejunostomy. The modification of the Blumgart pancreaticojejunostomy was compared to the Cattel-Warren pancreaticojejunostomy in cohorts of patients matched by propensity scores based on factors predictive of clinically relevant postoperative pancreatic fistula, which was the primary endpoint of this study. Based on a noninferiority study design, 95 open pancreatoduodenectomies per group were needed. Feasibility of the modification of the Blumgart pancreaticojejunostomy in robotic pancreatoduodenectomy was also shown. All pancreaticojejunostomies were performed by a single surgeon. Between October 2011 and May 2019, there were 415 pancreatoduodenectomies with either a Cattel-Warren pancreaticojejunostomy (n = 225) or a modification of the Blumgart pancreaticojejunostomy (n = 190). There was 1 grade C postoperative pancreatic fistula in 190 consecutive modification of the Blumgart pancreaticojejunostomies (0.5%). Logistic regression analysis showed that the rate of clinically relevant postoperative pancreatic fistula was not affected by consecutive case number. After exclusion of robotic pancreatoduodenectomies (the Cattel-Warren pancreaticojejunostomy: 82; modification of the Blumgart pancreaticojejunostomy: 66), 267 open pancreatoduodenectomies were left, among which the matching process identified 109 pairs. The modification of the Blumgart pancreaticojejunostomy was shown to be noninferior to the Cattel-Warren pancreaticojejunostomy with respect to clinically relevant postoperative pancreatic fistula (11.9% vs 22.9%; odds ratio: 0.46 [0.21-0.93]; P = .03), grade B postoperative pancreatic fistula (11.9% vs 18.3%; P = .18), and grade C postoperative pancreatic fistula (0 vs 4.6%; P = .05) as well as to all secondary study endpoints. The modification of the Blumgart pancreaticojejunostomy was feasible in 66 robotic pancreatoduodenectomies. In this subgroup with 1 conversion to open surgery (1.5%), a clinically relevant postoperative pancreatic fistula occurred after 9 procedures (13.6%) with no case of grade C postoperative pancreatic fistula and a 90-day mortality of 3%. The modification of the Blumgart pancreaticojejunostomy described herein is noninferior to the Cattel-Warren pancreaticojejunostomy in open pancreatoduodenectomy. This technique is also feasible in robotic pancreatoduodenectomy.