|
Back to 2017 Program and Abstracts
PANCREATICOGASTROSTOMY VS. PANCREATICOJEJUNOSTOMY: A RISK-STRATIFIED ANALYSIS OF 5,316 PANCREATODUODENECTOMIES
Brett L. Ecker*, Matthew T. McMillan, Laura Maggino, Charles Vollmer
University of Pennsylvania, Philadelphia, PA
Background: Pancreaticogastrostomy (PG) reconstruction following pancreatoduodenectomy (PD) is regarded as a strategy to mitigate the risk of clinically-relevant pancreatic fistula (CR-POPF), particularly following high-risk anastomoses. Previous analyses have not been rigorously risk-adjusted for CR-POPF development.
Methods: An international, multi-institutional retrospective study of 5,316 PDs performed by 62 surgeons was queried for rates of CR-POPF following PG and pancreaticojejunostomy (PJ). Risk zones (Negligible, low, moderate, and high) for CR-POPF were assessed using the previously validated Fistula Risk Score. The overall complication burden (Postoperative Morbidity Index, PMI) and average complication burden (ACB) for fistula were quantified using severity weights paired with the Modified Accordion Severity Grading System. Multivariable logistic regression modeling and propensity score matching (1:1) were utilized to assess the influence of PG reconstruction on fistula outcomes.
Results: PG reconstruction was utilized in 276 (5.2%) PDs. The frequency of PG use varied by the risk of the anastomosis (2.0%, 6.1%, 12.2% and 31.8% of Negligible, low, moderate, and high risk, respectively; p<0.001). A pancreatic duct ≤1 mm was the most predictive factor of PG use (OR 4.48, 95% CI 2.52-7.96). The rate of CR-POPF in the overall cohort was 12.9%; PG was associated with higher rates (Negligible: 0.0% vs. 0.7%, p=0.885; low: 11.1% vs. 5.0%, p=0.158; moderate: 22.5% vs. 13.7%, p=0.001; high: 50.0% vs. 28.8%, p<0.001). PG was associated with significantly greater overall morbidity (Accordion ≥1: 70.7% vs. 55.3%, p<0.001), as well as increased incidence of severe complications (Accordion ≥3: 34.1% vs. 18.9%, p<0.001). Duration of stay was significantly longer after PG (median [IQR]: 14 [9-23] days vs. 8 [7-13] days, p<0.001), which was a consistent finding when restricting analysis to the surgeons who employed PG (p<0.001). The PMI associated with PG reconstruction was significantly higher than PJ (0.264±0.245 vs. 0.180±0.227, p<0.001). PG was not associated with reduced fistula burden (ACB) in the cohort of surgeons who utilized this technique (PG: 0.360±0.176 vs. PJ: 0.387±0.232, p=0.378). Logistic regression modeling demonstrated that PG was not associated with lower CR-POPF occurrence in any risk zone. In the propensity score-matched cohort, PG was not associated with improved fistula rates in any risk zone (Negligible: 0.0% vs. 0.0%, p=1.00; low: 0.0% vs. 3.7%, p=0.422; moderate: 17.4% vs. 26.0%, p=0.178; high: 50.0% vs. 36.2%, p=0.120).
Discussion: Pancreaticogastrostomy does not improve pancreatic fistula outcomes, even in scenarios with high-risk anastomoses. Risk assessment paradigms provide the opportunity to compare mitigation strategies while rigorously accounting for inherent variability in fistula risk.
Back to 2017 Program and Abstracts
|
