Use of EUS-FNA in diagnosing pancreatic neoplasm without a definitive mass on CT

Background: Diagnosis of pancreatic neoplasm is challenging in patients with inconclusive findings on pancreatic multidetector row CT (MDCT). Objective: To determine the diagnostic accuracy and to identify predictors of pancreatic neoplasm by EUS with FNA in this setting. Design: Retrospective chart review during the study period of January 2002 to December 2010. Setting: Tertiary referral center. Patients: Of the 1046 patients who underwent pancreatic EUS, 116 patients were selected because their clinical presentation was suspicious for pancreatic malignancy, but their MDCT findings were inconclusive. Intervention: EUS with FNA. Main Outcome Measurements: Diagnostic yield of malignancy and significance of clinical variables. Results: When surgical pathology or subsequent clinical course was used as the criterion standard, EUS with FNA had a sensitivity, specificity, positive predictive value, and accuracy of 87.3%, 98.3%, 98.5%, and 92.1%, respectively, in diagnosing a pancreatic neoplasm that was indeterminate on MDCT. Factors significantly associated with EUS detection of pancreatic ductal adenocarcinoma were total bilirubin level greater than 2 mg/dL (P < .001), CT finding of pancreatic duct dilation (P < .001), bile duct stricture (P < .001), and tumor size 1.5 cm or larger detected by EUS (P = .004). Among them, pancreatic duct dilation on CT (odds ratio 4.10; 95% confidence interval, 1.52-11.05), and tumor size 1.5 cm or larger detected by EUS (odds ratio 8.46; 95% confidence interval, 2.02-35.45) were independent risk factors. Limitations: Retrospective design and patient referral bias. Conclusions: When MDCT is indeterminate, EUS is a highly sensitive and accurate modality for the detection of pancreatic neoplasm, especially when the tumor is smaller than 2.0 cm.