Prospective preoperative determination of mucinous pancreatic cystic neoplasms.

Optimal management of pancreatic cystic neoplasms includes identification and resection of mucinous neoplasms. This study was performed to assess the accuracy of preoperative variables in determining a mucinous lesion. Patients referred for a cystic neoplasm were prospectively assessed by presenting symptoms, blinded radiologic review, and endoscopic ultrasound-guided cyst aspirate analysis. Patients who were symptomatic, or had aspirate findings of a mucinous neoplasm were resected. Eighty-seven patients were enrolled over a 22-month period ending in December 2001. There were 56 (64%) women and 31 (36%) men, with a mean age of 63 (27-86) years. Thirty-five (40%) patients were resected including 24 (69%) women and 11 (31%) men with a mean age of 58 years. Twenty-eight (80%) patients who had resection were symptomatic. Specimen histology included 18 (51%) mucinous neoplasms, 8 (23%) serous neoplasms, 4 (11%) ductal or neuroendocrine carcinomas, and 3 (9%) pseudocysts. The positive predictive value (PPV) for cyst-aspirate extracellular mucin (83%) was significant in predicting a mucinous neoplasm (P =.009). No other aspirate variables (amylase, carcinoembryonic antigen, CA15-3, viscosity), or patient characteristics were predictive of final histology. Diagnostic agreement between all 3 radiologists was 8% (P =.98). At a median follow-up of 12 months, no patients who were observed required resection. Patients with suspected pancreatic cystic neoplasms can be selectively treated on the basis of symptoms and cyst-aspirate mucin analysis. Symptomatic and mucin containing lesions should be resected.