Evaluation of predictive factors for malignancy in cystic neoplasms of pancreas / 中华肝胆外科杂志

ABSTRACT

Objective To determine the value of preoperative clinical, biochemical, cross-sec-tional imaging features and results of fine-needle aspiration for predicting malignancy in cystic neo-plasms of the pancreas (CNP). Methods The medical records of 69 patients receiving operations for CNP between 1994 and 2008 in our hospital were reviewed retrospectively. The predictive effect of va-rious preoperative factors such as sex, location, clinical manifestation, maximum diameters, tumor marker, pancreatic duct obstruction and calcification on the malignant potential of CNP was evaluated by Single and multi-factor analysis, fine needle aspiration (FNA) and intraoperative frozen-section ex-amination of the pancreatic transection margin was investigated. Results All the 69 patients were con-firmed pathologically. Of the 69 patients, 13 suffered from serous cystic neoplasms, 30 from mucinous cystic neoplasms,7 from intraductal papillary mucinous neoplasms,12 from solid pseudopapillary neo-plasms and 7 from cystic neoplasms. Forty-four lesions were diagnosed as malignant or borderline.Univariate analysis should that jaundice, raised CEA and/or CA19-9, maximum diameters and solid component of cystic neoplasmshad were of statistical significance for the risk of malignancy in CNP.The sensitivity was 34.1% (15/44), 47.7 % (21/44), 88.6%(39/44),72.7%(32/44) and specificity 96% (24/25), 84% (23/25), 68% (17/25),72% (18/25), respectively. The last three were identified as independent predictive factors for malignancy by multivariate analysis. Three cases were accurately diagnosed out of the 9 undergoing FNA preoperatively. One of 7 patients with intraductal papillary mucinous neoplasms (IPMN) undenwent total pancreatetomy for transection margin positivity.Conclusion Most malignant CNP can be accurately diagnosed preoperatively from a typical clinical,biochemical and cross-sectional imaging picture. FNA is only used in the patients who are potential candidates for nonoperative management. Margin analysis is necessary for pancreatic resection.