Modified reconstruction approach after pancreaticoduodenectomy optimizes postoperative outcomes: Results from a multivariate cohort analysis.

Despite technical advances in recent decades and a decrease in hospital mortality (<5%), pancreaticoduodenectomy (PD) is still associated with major postoperative complications, even in high-volume centers. The present study aimed to assess the effect of a modified reconstruction technique on postoperative morbidity and mortality. A cohort study of all patients (n=218) undergoing PD between January 2010 and December 2019 was performed at Attikon University Hospital (Athens, Greece). Several variables were studied, including demographic data, past medical history, perioperative parameters, tumor markers and pathology, duration of hospitalization, postoperative complications, 30-day-survival, postoperative mortality and overall survival using multivariate logistic regression and survival analysis techniques. In this cohort, 123 patients [modified PD (mPD) group] underwent a modified reconstruction after a pylorus-preserving pancreaticoduodenectomy, which consisted of gastrojejunostomy and pancreaticojejunostomy on the same loop and an isolated hepaticojejunostomy on another loop. In the standard PD (StPD) group, 95 patients underwent standard reconstruction. The median age was 67 years, ranging from 25 to 89 years. Compared with in the StPD group, the mPD group had significantly lower rates of grade B and C pancreatic fistula (4.9% vs. 28.4%), delayed gastric emptying (7.3% vs. 42.1%), postoperative hemorrhage (3.3% vs. 20%), intensive care unit admission (8.1% vs. 18.9%), overall morbidity (Clavien-Dindo grade III-V: 14.7% vs. 42.0%), perioperative mortality (4.1% vs. 14.7%), and shorter hospitalization stay (11 days vs. 20 days). However, no difference was noted regarding median survival (35 months vs. 30 months). In this single-center series, a modified reconstruction after PD appears to be associated with improved postoperative outcomes. However, further evaluation in larger multi-center trials is required. [ABSTRACT FROM AUTHOR]