Negative Predictive Value of Endoscopic Ultrasound in a Large Series of Patients with a Clinical Suspicion of Pancreatic Cancer

Abstract

Endoscopic ultrasound (EUS) has been shown to be very accurate in the diagnosis and staging of pancreatic cancer. The accuracy of EUS in predicting the absence of pancreatic cancer in a large series of patients with a clinical suspicion of pancreatic cancer is not well documented. Our aim was to determine the negative predictive value (NPV) of EUS in patients with a suspicion of pancreatic cancer. We retrospectively reviewed, from our EUS database (between January 1999 and March 2003), 693 patients who were suspected of having pancreatic cancer and had EUS examinations. A total of 155 patients were found by EUS to have a completely normal pancreas. Indications for EUS in these patients included: weight loss/abdominal pain; and/or pancreatic enlargement/fullness on computed tomography (CT); and/or bile duct/pancreatic duct narrowing on endoscopic retrograde cholangiopancreatography; and/or an elevated CA 19-9. Follow-up information was obtained in 135/155 (87%) patients from patient phone calls and/or physician visits and/or CT scan. The mean follow-up period was 25 months (range 8–48 months). No patients developed pancreatic cancer during the follow-up period. Following the EUS examination, no work-up was required in 119/135 (88%) of patients. CT scan was performed in 16 patients at 6 months postprocedure, none of which showed a pancreatic mass. The NPV of EUS in excluding pancreatic cancer in those patients with follow-up was 100%. EUS is highly specific in the diagnosis of pancreatic cancer with a NPV of 100% and obviates the need for further diagnostic testing. In patients with a clinical suspicion of pancreatic cancer, EUS should be considered as the initial diagnostic modality.