4562 Assessment of endoscopic ultrasound guided fine needle aspiration biopsy in the evaluation of pancreatic masses.

Abstract

Background: Limited data comparing EUS FNA with other biopsy modalities for pancreatic masses are available. Aims: To compare the accuracy of EUS FNA with abdominal CT, ERCP and pathological findings at surgery in evaluating pancreatic masses. Methods: Retrospective patient chart review identified all patients (pts) with pancreatic masses who underwent EUS FNA between March 1994 and 1999. In total, 211 pts with pancreatic masses were evaluated with EUS FNA; 205 (97%) were evaluated with abdominal CT [65 had CT-guided biopsy before EUS], 93 (44%) were evaluated with ERCP [45 had brushings/biopsy before EUS]. Results: One patient suffered a complication post EUS FNA (mild pancreatitis, 0.5%, 95% CI 0-1%). CT completely failed to detect the pancreatic mass in 78 (38%) pts. There were 169 pts (80%) who had confirmed pancreatic malignancy on EUS FNA [155 adenocarcinomas, 12 neuroendocrine tumors (NETs), 2 lymphomas]. Of the 44 patients that proceeded to surgery 33 (75%) had malignancy [4 NETs]. Using surgical findings as the gold standard, sensitivity of EUS FNA is 88% (95% CI: 77%-99%); specificity is 100% (95% CI: 94%-100%) and accuracy is 91% (95% CI: 86%-96%). Of the 10 pts who had benign CT-FNA and went to surgery, 5 were found to have malignant disease (diagnostic accuracy, 50%). Similarly, of the 10 pts who had benign ERCP sampling and went to surgery, 5 were found to have malignant disease (accuracy also 50%). Of the 33 patients with malignancy who underwent surgery, 17 were resectable (6 correctly classified by CT, 16 by EUS) and 16 patients were unresectable (1 correctly classified by CT, 14 by EUS). For determining resectability/unresectability CT had a sensitivity of 35%, specificity of 100%, accuracy of 67% and EUS had sensitivity of 94%, specificity of 100%, accuracy of 97% (p=0.009). Of the 62 patients with negative CT-guided biopsies, 50 (80%) patients were subsequently identified to have malignancy on EUS FNA (23% agreement). Of the 40 patients with negative ERCP samples, 32 (80%) patients were subsequently identified to have malignancy on EUS FNA (18% agreement). Maximal tumor size as determined by EUS correlated strongly with size at surgery (Pearson correlation coefficient 0.8, p=0.002). Conclusion: Assuming the pathological findings at surgery are a gold standard, EUS FNA is a safe and superior diagnostic technique in the evaluation of pancreatic masses when compared to CT and ERCP. EUS is also superior in assessing tumor size and detecting resectability compared to CT.