Development of a Prognostic Model That Predicts Survival After Pancreaticoduodenectomy for Ampullary Cancer

Objectives: The aims of this study were to (i) identify independent predictors of survival after pancreaticoduodenectomy for ampullary cancer and (ii) develop a prognostic model of survival. Methods: Data were analyzed retrospectively on 110 consecutive patients who underwent pancreaticoduodenectomy between 2002 and 2013. Subjects were categorized into 3 nodal subgroups as per the recently proposed nodal subclassification: N0 (node negative), N1 (1-2 metastatic nodes), or N2 (>= 3 metastatic nodes). Clinicopathological features and overall survival were compared by Kaplan-Meier and Cox regression analyses. Results: The overall 1-, 3-, and 5-year survival rates were 79.8%, 42.2%, and 34.9%, respectively. The overall 1-, 3-, and 5-year survival rates for the N0 group were 85.2%, 71.9%, and 67.4%, respectively. The 1-, 3-, 5-year survival rates for the N1 and N2 subgroups were 81.5%, 49.4%, and 49.4% and 75%, 19.2%, and 6.4%, respectively (log rank, P < 0.0001). After performing a multivariate Cox regression analysis, vascular invasion and lymph node ratio were the only independent predictors of survival. Hence, a prediction model of survival was constructed based on those 2 variables. Conclusions: Using data from a carefully selected cohort of patients, we created a pilot prognostic model of postresectional survival. The proposed model may help clinicians to guide treatments in the adjuvant setting.