|
Back to 2018 Posters
LOWER IS HIGH ENOUGH: NEW SUGGESTED THRESHOLD FOR POSTOPERATIVE DAY 1 DRAIN-FLUID AMYLASE POST PANCREATODUODENECTOMY
Essa M. Aleassa*1,2, Gautam Sharma1, Gareth Morris-Stiff1
1Section of HepatoPancreatoBiliary Surgery, Cleveland Clinic, Beachwood, OH; 2Department of Surgery, United Arab Emirates University, Al-Ain, Abu Dhabi, United Arab Emirates
Background and Aims: The use of day one drain-fluid amylase (DFA-1) to predict post-operative pancreatic fistula (POPF) following pancreatoduodenectomy (PD) remains controversial. This stems in part from the wide variation in published cut off levels for when a DFA-1 correlates with fistula development. We aim to assess the significance of a DFA-1 level of 5000 U/L, and further try to establish the optimal DFA-1 threshold best correlating with fistula formation.
Patients and Methods: The American College of Surgeons-National Surgical Quality Improvement Program (NSQIP) pancreatectomy targeted files were queried from 2014-2016 to identify patients who underwent PD. Only patients with a recorded DFA-1 level were included for analysis. Patients with fistula were divided into biochemical leak (A) and the clinically relevant (B/C) based on the documentation of intervention, and those with an apparent grade A incorporated with the no fistula group. For statistical analysis, DFA-1 was recorded as a dichotomous variable using the cut off of less or greater than 5000 U/L. A Receiver Operator Characteristic (ROC) curve was plotted to determine the optimal DFA-1 to predict fistula formation.
Results: We identified a total of 9432 cases of PD, of which only 2545 (27%) had a postoperative DFA-1 level recorded. A fistula was recorded in 410 cases (16.2%) consisting of 277 (67.5%) and 133 (32.5%) grades A and B/C fistulae, respectively. A postoperative DFA-1 level of 5000 U/L significantly correlates with development of a grade B/C pancreatic fistula with a specificity of 84.4% and sensitivity of 32.3% (positive predictive value of 10.3% and negative predictive value of 95.7%). The area under the curve for B/C was 0.792 (p<0.001) setting a specificity of 80%, the ROC curve demonstrated that a cut off value of 800 U/L, provided a sensitivity of 68.6%.
Conclusion: The current suggested cut off DFA-1 value of 5000 U/L is statistically significant of fistula development post PD, however its sensitivity is poor. The ROC determined value of 800 U/L would appear to be a more appropriate level for detection of B/C fistulae.
Back to 2018 Posters
|
