LAUSR

Pancreatic carcinoma is a relatively common malignancy with an overall poor prognosis which is somewhat improved in those patients for whom resection and adjuvant therapy is feasible. In recent years there has been a trend to administering neoadjuvant therapy (combination chemotherapy and/or chemoradiotherapy), followed by resection in patients who remain surgical candidates at the completion of this treatment. Advantages of a neoadjuvant approach may include greater likelihood of achieving complete resection with negative surgical margins, reduced treatment toxicity and greater cost effectiveness, as well as potentially sparing patients with rapidly progressive disease from major surgery. To gauge the tumour's response to preoperative therapy, and to compare the efficacy of different regimens, there is a need for a robust and reproducible system of assessing tumour regression in resection specimens. Several such systems have been proposed, but there is generally a lack of consensus as to which system is the 'best'. This review describes the evolution of a number of tumour regression grading systems which have been proposed, and discusses the relative merits and shortfalls of several of the most frequently applied schemata. Some problems common to many of these include poorly defined criteria, low interobserver reproducibility and a reliance on fibrosis as a surrogate for tumour kill, which may not be valid. Despite that, recent evidence suggests that the Dworak grading system (first developed for rectal cancer) may be useful in terms of both interobserver concordance and correlation with survival.