Biochemical analysis of pancreatic fluid collections predicts bacterial infection

Academic Article


  • Background and Aims: Despite our understanding of the pathophysiology of different types of pancreatic fluid collections (PFC), few studies have attempted to correlate the biochemical analysis of PFC contents with clinical and radiological characteristics. The aim of this study was to assess the predictive value of fluid analysis for discerning collection type (pseudocyst vs acute fluid collection with necrosis), presence of infection or communication with the pancreatic duct in the setting of acute and chronic pancreatitis. Methods: Pancreatic fluid from 34 consecutive patients undergoing endotherapy of PFC was prospectively analyzed for seven variables: lactate dehydrogenase (LDH), total protein, albumin, glucose, amylase, lipase and specific gravity. Results: In multivariate analysis, adjusting for age and gender, high intracystic levels of protein (OR 6.2; 95% CI 1.3-37.0), LDH (OR 6.8 [2.3-38.3]), and albumin (OR 7.8 [1.3-67.4]), and low levels of glucose (OR 0.2 [0.03-0.9]) predicted the presence of PFC infection. The optimal threshold value for protein was 1000 g/dL, which achieved a sensitivity of 73% and specificity of 75% for detecting infection; the optimal cut-off for LDH was 1000 U/L (sensitivity 64%, specificity 85%), and the cut-off for albumin was 500 g/dL (sensitivity 75%, specificity 85%). There were no statistically significant differences in biochemical fluid analysis with respect to fluid collection type (pseudocysts vs acute fluid collection with necrosis) and the presence of pancreatic duct communication. Conclusions: Biochemical analysis of PFC fluid is clinically helpful in detecting fluid infection in patients with bacteria on Gram stain or positive fluid cultures. Our findings fail to support the utility of fluid analysis in characterizing cyst type, and we caution against its use in distinguishing pseudocysts from acute fluid collection with necrosis. © 2005 Blackwell Publishing Asia Pty Ltd.
  • Digital Object Identifier (doi)

    Author List

  • Mönkemüller KE; Harewood GC; Curioso WH; Fry LC; Wilcox CM; Morgan DE; Baron TH
  • Start Page

  • 1667
  • End Page

  • 1673
  • Volume

  • 20
  • Issue

  • 11