Cystic pancreatic neoplasms: observe or operate

The objectives of this analysis were to define the incidence, natural history, and predictors of neoplasia in pancreatic cysts to determine which patients can safely be observed and which should undergo an operation. With advancements in imaging technology, cystic lesions of the pancreas are being detected with increased frequency. Many of these lesions are small and asymptomatic, but they may be associated with pancreatitis or have malignant potential. Therefore, the management of these patients is complex, and knowledge of pancreatic cyst natural history and predictors of neoplasia are important. From January 1995 through December 2002, all radiologic, surgical, and pathology records were reviewed for the presence of pancreatic cysts. In determining natural history, only patients with 2 scans more than 1 month apart at our institution were included. Patients with a clinical history and laboratory evidence of pancreatitis and/or pathologic confirmation of a pseudocyst were excluded. Factors analyzed as potential predictors of neoplasia included age, gender, cyst size, and symptoms. Serous cystadenomas, solid and cystic papillary (Hamoudi) tumors, lymphoepithelial cysts and simple cysts were all benign, whereas mucinous cystic neoplasms, intraductal papillary mucinous neoplasm, cystic neuroendocrine tumors, and cystadenocarcinomas were considered to be premalignant or malignant. Among 24,039 CT or MR scans, 290 patients (1.2%) had pancreatic cysts, and 168 of these patients (0.7%) had no documentation of pancreatitis. Seventy-nine of these patients with 103 cysts had more than 1 scan with an average interval of 16 months. These cysts increased in size in 19%, did not change in 59% and decreased in 22% of patients. Forty-nine patients underwent surgery for 14 benign (serous cystadenomas = 10, Hamoudi = 2, lymphoepithelial = 1, simple = 1) 25 premalignant (mucinous cystic neoplasm =16, intraductal papillary mucinous neoplasm = 5, neuroendocrine tumors = 4), or 10 malignant (intraductal papillary mucinous neoplasm = 7, cystadenocarcinomas = 3) lesions. Gender and cyst size did not predict neoplasia. However, presence of symptoms predicted premalignant or malignant pathology (60% vs. 23%, P < 0.05), and age over 70 years was associated with malignancy (60% vs. 21%, P < 0.02). These data suggest that cystic pancreatic neoplasms 1) occur in 0.7% of patients, 2) increase in 19% over 16 months, and 3) are likely (60%) to be malignant in patients older than 70 year