EUS FNA Is More Sensitive Than CT/US FNA for Diagnosing Pancreatic Malignancy: A Randomized Controlled Trial

Abstract

Background: EUS FNA is an evolving alternative to percutaneous CT or US-guided biopsy for diagnosing pancreatic lesions. We postulated that EUS FNA would have greater sensitivity and accuracy than CT/US FNA in the setting of pancreatic lesions and designed a prospective trial to test this hypothesis. Methods: Patients referred to our center between 12/97 and 12/02 with suspected solid pancreatic or peri-pancreatic lesions were offered randomization to undergo EUS FNA or CT/US FNA. If initial cytology was non-diagnostic, patients were offered cross over to the other modality. A negative result included (a) negative cytology from FNA or (b) no mass seen on imaging on which to perform FNA. To determine the outcome of patients with negative FNA results, clinical review was performed every 6 months for 2 years. Results of surgical pathology were also used to determine true and false negatives. Results: 84 patients were enrolled/randomized: 43 to CT/US FNA and 41 to EUS FNA. There were 3 withdrawals, 2 technical failures and 7 protocol violations. There were 16 true positive diagnoses of malignancy by CT/US FNA and 21 by EUS FNA. Of 20 CT/US FNA negatives, 16 crossed-over to EUS FNA (12 underwent FNA, 4 did not due to absence of mass at EUS). Seven of 12 patients crossing over from CT/US had positive cytology at EUS FNA. Eight patients with negative EUS FNA crossed over to CT/US- 3 had no mass for CT/US FNA. Two of the 3 remained negative through followup, the other had adenocarcinoma diagnosed intraoperatively. All 5 EUS FNA negatives that actually underwent cross over CT/US FNA were negative. The sensitivity of CT/US FNA and EUS FNA for detecting malignancy was 61% and 84% (95%CI = 0.41-0.80 vs. 0.64-0.95). Better agreement with final diagnosis was seen for EUS FNA than CT/US FNA (kappaEUS vs. kappaCT/US = 0.76 (0.55-0.97) vs. 0.47 (0.23-0.71), however, accuracy of the two techniques in providing the correct final diagnosis was not significantly different (p = 0.135 Fisher's exact test). Conclusions: EUS FNA appears to have a greater sensitivity for detecting pancreatic cancer. Due to clear patient and physician preference for EUS FNA, recruitment was halted prematurely which has limited the power of the study, however, in our center the results of EUS FNA in pancreatic lesions appear superior to CT/US FNA.