Prediction of pancreatic cancer surgical outcomes and prognosis based on an objective resectability scoring system.

Abstract

446 Background: There is no consensus on the best radiographic criteria to define pancreatic cancer (PCA) resectability. We undertook this study to objectively assess the probability of tumor resection based on quantitative measures of tumor abutment from the main peripancreatic vessels. Methods: Measurements of the circumferential degree and length of solid tumor contact with the superior mesenteric artery, the celiac artery/common hepatic artery, and superior mesenteric vein/portal vein were obtained through curved planar reformations from diagnostic CT images of PCA patients treated at our institution between 2001 and 2015. Results: A total of 399 patients were identified, of which 294 were used for the analysis. Of them, 113 (38.4%) were resected, 71 (62.8%) with negative margins (R0). Based on the individual measurements of vascular involvement, a resectability scoring system (RSS) was created (table), and used in a classification and regression tree model that correlated strongly with resection (p < 0.0001) and R0 resection (p < 0.0001) probabilities. The RSS demonstrated a higher PPV, NPV and accuracy (94.3%; 98.6%; 97%) to predict resection compared to the NCCN (90.8%; 96.7%; 94.5%) and MDACC resectability criteria (87.1%; 97.9%; 93.5%). Moreover, the RSS correlated with overall survival (OS) (p < 0.0001) and metastasis-free survival (p < 0.0001). Conclusions: We developed a PCA resectability scoring system that is objective, reproducible, and highly correlated with surgical outcomes and prognosis. External validation is warranted. [Table: see text]