Accuracy of Diagnosing PDA, Neuroendocrine Tumors, and IPMN by EUS-FNA at a Single Institution

Hannah L Beal, Jill E Shea, Benjamin L Witt, Douglas G Adler, Sean J J Mulvihill, Erinn Downs-Kelly, Matthew A Firpo, Courtney L Scaife


Background: The purpose of this study was to assess the clinical impact of diagnostic accuracy for EUS-FNA in evaluating pancreatic lesions, particularly Pancreatic Ductal Adenocarcinoma( PDAC), neuroendocrine tumors, and IPMN at our institution using surgical pathological review as the gold standard. Methods: We conducted a retrospective chart review using the Clinical Cancer Research Database at Huntsman Cancer Institute. We included all cases in which pancreatic lesions were evaluated by EUS-FNA and a subsequent surgical resection was performed. Sensitivity, specificity, positive predicative value (PPV), and negative predictive value (NPV) were determined for PDAC, neuroendocrine tumors, and IPMN by comparing pathological diagnosis at EUS-FNA to diagnoses following surgery. Results: Detection of PDAC by FNA-EUS yielded 87.9% (80.1-93.4 95% CI) sensitivity, 80% (72-86) specificity, 78% (70-85) PPV, and 89% (82-94) NPV. Detection of neuroendocrine tumors by FNA-EUS yielded 66% (47-87) sensitivity, 100% (97-100) specificity, 94% (71-99) PPV, and 97% (94-99) NPV. Detection of IPMN by FNA-EUS yielded 59% (43-74) sensitivity, 100% (98-100) specificity, 100% (87-100) PPV, and 89% (93-99) NPV. Conclusions: Our findings are largely consistent with the current literature, confirming there is discernible potential for inappropriate treatment of patients based purely on EUS-FNA evaluation. Both accuracy and clarity of positive tests with EUS-FNA worsened for mucinous pancreatic lesions compared to solid pancreatic lesions. Limitations of this study are the appraisal at a single institution and the necessity to evaluate only cases that ultimately had surgical resection of the pancreatic lesion.


EUS-FNA; Neuroendocrine; PDAC; PDA; IPMN; Diagnostic; Accuracy; Pancreas; Cancer

Full Text: PDF HTML


  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.