Endosonography of cystic lesions of the pancreas with clinicopathologic correlation

Management of pancreatic cystic lesions (PCL) depends on the perceived malignant potential. Endosonography is sensitive for detecting PCL's. We reviewed the features of PCL's evaluated by EUS at our institution to assess features that predict histopathology. Methods: Among 367 EUS exams of the pancreas, with the Olympus GF-UM20 echoendoscope, 47 PCL's were evaluated in which clinicopathologic correlation was available. We assessed for the presence of the following features: i) wall, ii) solid component, iii) septae, iv) lymphadenopathy, v) EUS parenchymal features of chronic pancreatitis (ChP), vi) pancreatic ductal abnormalities, vii) single vs. multiple cysts, viii) vascular invasion, ix) size. Significant variables were individually assessed by odds ratio analysis. Results: Histopathologic correlation was determined by surgical specimen in 32 cases (for all neoplastic lesions), and by long-term clinical follow-up in 15 (mean = 16 mo). Six categories of lesion were identified by pathology: mucinous cystic neoplasm (adenoma, adenocarcinoma, duct ectasia) (12), microcystic (serous) adenoma (4), pseudocyst (17), neuroendocrine tumor (5), adenocarcinoma (4), miscellaneous benign lesions (5). When compared with all other lesions, EUS features positively associated with pseudocysts were: i) the presence of a wall (RR 3.69, CI 1.67-8.14), and ii) ChP (RR 7.96, CI 2.67-23.77). Features negatively associated with pseudocysts were i) septae (RR 0.38, CI 0.16-0.91) and ii) solid component (RR 0.29, CI 0.12-0.69). Other features did not significantly predict for the presence of pseudocysts. Non-neoplastic lesions (pseudocyst, miscellaneous benign) were associated with the presence of a wall (RR 2.43, CI 1.36-4.34) and ChP (RR 4.54, CI 2.20-9.36), and inversely associated with the presence of septae (RR 0.27, CI 0.11-0.61) and solid component (RR 0.26, CI 0.12-0.54). EUS features positively associated with neoplastic lesions (i.e.. mucinous cystic neoplasm, neuroendocrine, adenocarcinoma) were septae (RR 2.75, CI 1.20-6.31) and solid component (RR 13.4, CI 1.95-91.5). Features negatively associated with neoplastic lesions were the presence of a wall (RR 0.39, CI 0.14-1.12) and ChP (RR 0.09, CI 0.01-0.61). Conclusions: (1) EUS features predictive of pseudocyst are the presence of a wall and/or EUS parenchymal features of ChP. (2) Features negatively associated with pseudocyst are the presence of septae and/or a solid component. (3) EUS features positively associated with neoplastic lesions are septae and/or solid component. (4) Features negatively associated with neoplastic lesions were the presence of a wall and/or EUS parenchymal features of ChP. (5) EUS is valuable in evaluating pancreatic cystic lesions.