M. T. McMillan1, M. H. Sprys1, J. A. Drebin1, M. K. Lee1, R. E. Roses1, D. L. Fraker1, .. The Pancreatic Fistula Study Group1, C. M. Vollmer1 1University Of Pennsylvania Perelman School Of Medicine,Surgery,Philadelphia, PA, USA
Introduction: Postoperative pancreatic fistula (POPF) is the most common and morbid complication following pancreatoduodenectomy (PD). The previously validated Fistula Risk Score (FRS) considers the presence of endogenous (gland texture, duct size, pathology) and operative (blood loss) risk factors to predict the occurrence of clinically relevant fistula (CR-POPF; ISGPF Grade B/C). These CR-POPF risk factors may also influence mortality; however, this has not been proven.
Methods: This multinational study of 4,307 PDs involved 55 pancreatic surgical specialists at 15 high-volume institutions. Patients were stratified for 90-day mortality risk using the FRS (0-10 points) and assigned to one of four risk zones: negligible (0 points), low (1-2), moderate (3-6), or high (7-10). A Cox regression identified predictors for mortality while adjusting for FRS risk, as well as surgeon, institutional, and operative factors.
Results: The overall mortality rate was 2.1% (N=89), with institutional rates ranging from 1.0 -8.6%. Individual surgeon rates—for those who contributed ≥ 25 cases—ranged from 0 – 11.1%. Clinically relevant fistulas accounted for 36% of the overall mortalities and their presence strongly correlated with higher rates of mortality (6.6 vs. 1.5%; P<0.001). Nearly 70% of deaths occurred in the setting of soft pancreatic parenchyma and intraoperative blood loss > 700 mL was associated with a greater than two-fold increase in mortality risk. The mean Fistula Risk Score was significantly greater in patients who suffered mortality (4.6 vs. 3.7; P<0.001). In fact, patients with high CR-POPF risk (FRS 7-10) had over a fivefold increase in mortality risk compared to patients at negligible risk (P=0.010; Figure). There was no significant difference in mean FRS between fistulous and non-fistulous mortalities (4.6 vs. 4.6; p=0.899); however, the median POD of mortality was two times greater in cases of mortality due to a CR-POPF (28 [IQR: 40] vs. 14 [IQR: 26] days; P=0.010). While surgeon years of experience and career PD volume did not significantly influence overall mortality, institutional PD volume > 75 cases per year correlated with reduced rates (1.9 vs. 4.9%; P=0.006).
Conclusion: Procedure-specific risk influences mortality after pancreatoduodenectomy. Improvements in pancreatic fistula outcomes will likely lead to improved survival following PD.
