INTRODUCTIONWe assessed the incidence and outcome of pancreaticoduodenectomy for patients with a pre-operative benign diagnosis and in patients who had an unexpected diagnosis of benign disease following resection. We have also compared how the introduction of endoscopic ultrasound fine needle aspiration (EUS-FNA) has altered our pre-operative assessment.PATIENTS AND METHODSBetween January 1997 and April 2006, 499 patients underwent pancreaticoduodenectomy at the Queen Elizabeth Hospital. Data were collected prospectively. A further 85 patients between 2006 and 2008 had a different diagnostic approach (after imaging these patients have been also studied by EUS-FNA).RESULTSOverall, 78 (15.6%) patients had no malignant disease on final histology. Out of 459 patients who underwent pancreaticoduodenectomy for presumed malignancy, 49 (10.6%) had benign disease (sensitivity, 97%; positive predictive value, 89%). In a further 40 patients with a pre-operative benign diagnosis, we found 11 cases (27%) of malignancy (sensitivity, 37%; negative predictive value, 72%). Following the introduction of EUS-FNA, the sensitivity and specificity of the diagnostic work were 92% and 75%, respectively (positive predictive value, 93%; negative predictive value, 63%). The median follow-up was 35 months (range, 1–116 months).CONCLUSIONSPrior to the introduction of EUS-FNA, a significant number of patients, in whom pancreaticoduodenectomy is carried out for suspected benign disease, turn out to have an underlying malignancy. The use of EUS-FNA has improved the specificity of diagnostic work-up.