Time trends in the treatment and prognosis of resectable pancreatic cancer in a large tertiary referral centre

ObjectivesMortality in pancreatic cancer has remained unchanged over the last 20–30 years. The aim of the present study was to analyse survival trends in a selected population of patients submitted to resection for pancreatic cancer at a single institution.MethodsIncluded were 544 patients who underwent pancreatectomy for pancreatic cancer between 1990 and 2009. Patients were categorized into two subgroups according to the decade in which resection was performed (1990–1999 and 2000–2009). Predictors of survival were analysed using univariate and multivariate analyses.ResultsTotals of 114 (21%) and 430 (79%) resections were carried out during the periods 1990–1999 and 2000–2009, respectively (P < 0.0001). Hospital length of stay (16 days versus 10 days; P < 0.001) and postoperative mortality (3% versus 1%; P = 0.160) decreased over time. Median disease-specific survival significantly increased from 16 months in the first period to 29 months in the second period (P < 0.001). Following multivariate analysis, poorly differentiated tumour [hazard ratio (HR) 3.1, P < 0.001], lymph node metastases (HR = 1.9, P < 0.001), macroscopically positive margin (R2) resection (HR = 3.2, P < 0.0001), no adjuvant therapy (HR = 1.6, P < 0.001) and resection performed in the period 1990–1999 (HR = 2.18, P < 0.001) were significant independent predictors of a poor outcome.ConclusionsLongterm survival after surgery for pancreatic cancer significantly improved over the period under study. Better patient selection and the routine use of adjuvant therapy may account for this improvement.