Predictors for postoperative liver metastasis in patients with resectable pancreatic cancer.

Although postoperative liver metastasis for pancreatic adenocarcinoma is a major problem that directly influences patient outcome, there is no useful predictor reported to date. We reviewed 174 surgically resected pancreatic tubular adenocarcinomas for their clinicopathological data and time for the appearance of postoperative liver metastasis to occur, using the Cox proportional hazards model. Multivariate analysis revealed three independent risk factors for postoperative liver metastasis: G3 histological grading, venous system invasion, and increased preoperative carbohydrate antigen 19-9. Furthermore, the additive effect of these three variables for predicting postoperative liver metastasis was evaluated. Two-year disease-free survival rate for the liver in patients without any factor was 82%, which was significantly higher than that of those with one, two, or three factors (35%, 33%, or 0%, respectively). Our data suggest that these three conventional clinicopathological variables are useful to predict postoperative liver metastasis in patients with pancreatic cancer.