Pancreatic cystic neoplasms (PCNs) represent a difficult preoperative diagnosis despite improvements in imaging. In this study, we compared preoperative and final pathologic diagnosis in a large cohort of resected PCNs, evaluating diagnostic accuracy with a specific focus on the value of EUS.

A retrospective analysis of patients undergoing resection between 2009 and 2019 for presumed PCNs was performed. Preoperative workup was reviewed by analyzing the role of imaging and EUS. Patients with a benign histology who did not show absolute indication were categorized as “delayable surgery.”

Of 585 patients who were retrospectively analyzed, in 108 (18.5%) final histology did not confirm preoperative diagnosis. EUS was associated with a lower rate of incorrect diagnosis (16%; P = .03), but the risk of overtreatment was similar regardless of instrumental diagnostic path (33/131 vs 68/328, P = .298). Dilatation of the main pancreatic duct and cytologic sampling were the only variables independently associated with a correct diagnosis (P < .001 and P = .041, respectively). Based on clinical presentation and final histology, pancreatic resection could have been spared or delayed in 101 of 459 patients (22%), and this was influenced by age (odds ratio [OR], .97; P = .002), cyst larger than 30 mm (OR, 1.89; P = .005), and type of operation (OR, 3.46 [P < .001] and 3.18 [P = .023] for distal pancreatectomies and other resections, respectively).

The overall risk of unnecessary immediate surgery for PCNs is about 22% in a high-volume referral center. EUS with cytologic sampling is a useful procedure in the diagnostic management of PCNs, improving their diagnostic accuracy.