Background: The clinical utility of EUS-FNA is debated in patients with obstructive jaundice (ObJ) because of a very high pretest probability of pancreatobiliary malignancy (PBM) and biliary stent-induced inflammation that can potentially confound EUS-FNA diagnosis. EUS-FNA also has lower accuracy in patients with underlying chronic pancreatitis (CP).
Objective: Our purpose was to determine the clinical value of EUS-FNA for PBM diagnosis based on clinical presentation and presence of CP.
Design: Retrospective analysis of prospective database.
Setting: University hospital.
Patients: Patients who underwent EUS-FNA from 2002 to 2006 for suspected PBM based on (1) ObJ with biliary stricture or a mass lesion or (2) abnormal pancreatic imaging by CT/MRI: a focal pancreatic "mass" lesion; dilated pancreatic duct +/- common bile duct; or an enlarged head of pancreas.
Interventions: EUS was performed with a radial echoendoscope followed by a linear echoendoscope if a focal pancreatic lesion was identified. Fine-needle aspirates were assessed immediately by an attending cytopathologist.
Main outcome measurements: (1) Prevalence of cancer and (2) performance characteristics of EUS-FNA.
Results: PBM was diagnosed in 73.9% of patients with ObJ and biliary stricture or pancreatic mass, in 49.6% of patients with pancreatic mass, and in 7.0% of patients with an enlarged head of pancreas or dilated pancreatic duct +/- common bile duct. The prevalence of PBM was lower in all 3 presentations with associated CP. Both CP and presentation with ObJ lowered performance characteristics of EUS-FNA, but CP did so only in the subset of patients with ObJ. All except 1 false-negative diagnoses were due to cytologic misinterpretation.
Limitation: Retrospective design.
Conclusion: Among patients with suspected PBM, the accuracy of EUS-FNA is significantly lower only in a subset of patients with ObJ with underlying CP, largely as a result of difficulty in cytologic interpretation.