An algorithmic approach to the diagnosis of pancreatic neoplasms

Abstract

Context: The pancreas gives rise to an array of distinct neoplasms that can be solid, cystic, or intraductal and can recapitulate the various lines of differentiation present in the normal gland.

Objective: To develop an algorithmic approach to the diagnosis of pancreatic neoplasms that simplifies their pathologic evaluation.

Data sources: We reviewed literature related to the classification of pancreatic neoplasms on the basis of their gross, histologic, and immunohistochemical features.

Conclusions: By using a series of dichotomous decisions, the differential diagnosis of a pancreatic neoplasm can be narrowed, and in cases of the more common neoplasms, accurate classification can be achieved. Uncommon neoplasms not accounted for by this approach are also discussed, and the additional diagnostic information needed for complete pathologic reporting is presented.