Identification of prognostic factors associated with early mortality after surgical resection for pancreatic cancer--under-analysis of cumulative survival curve

Abstract

Background: The cumulative survival curve after surgery for advanced pancreatic cancer is characterized by a steep downward slope in the early postoperative period. The aim of this investigation was to identify the characteristics associated with early mortality in patients undergoing pancreatic resection for pancreatic cancer.

Methods: Thirty-seven patients with extended radical pancreatectomy combined with intraoperative radiation therapy were studied. The cumulative survival curve in this series was depicted using the Kaplan-Meier method. Assuming that there were two distinct survival curves, below and above the breakpoint, each part of the curve was modeled as an exponential distribution. Three parameters, the breakpoint, the high hazard rate below the breakpoint, and the low hazard rate above the breakpoint were estimated by the maximum likelihood method. Prognostic factors associated with early mortality after surgery were evaluated using univariate and multivariate Cox proportional hazards regression analyses.

Results: The breakpoint of the survival curve was estimated at 41 months. The short-survival group (SSG) was defined as deceased earlier than 41 months after surgery, and included 31 patients (83.8 %). The long-survival patient group (LSG) consisted of 6 patients who were alive more than 41 months after surgery. Eighteen SSG patients (58.1 %) died of hepatic metastases, whereas no LSG patients died of hepatic metastases. Abdominal pain and/or back pain during clinical course was identified by multivariate analysis as a prognostic factor for patients undergoing pancreatic resection.

Conclusions: The high hazard rate in the early postoperative period was closely linked with death due to liver metastases. The preoperative presence of local pain was a prognostic factor associated with early mortality.