Combined histomorphologic and immunohistochemical phenotype to predict the presence of vascular invasion in colon cancer

PURPOSE: Vascular invasion is an adverse prognostic factor in colorectal cancer and often is considered an indication for systemic adjuvant therapy. The aim of this study was to develop a predictive model of vascular invasion in a large cohort of colon cancers with use of tumor-related features and immunohistochemical analysis of protein markers. METHODS: A tissue microarray including 427 colon cancers was evaluated for 14 immunohistochemical protein markers. Complete data on pT, pN, tumor grade, tumor border configuration, vascular invasion, and tumor diameter were available. Classification and regression tree analysis was performed. RESULTS: The presence of vascular invasion independently predicted adverse outcome (P > 0.001; hazard ratio = 1.52 (1.3 to 1.8)) and decreased survival from 49.0 to 19.0 months (P > 0.001) even in lymph node-positive patients. Tumor border configuration, Ki67, urokinase plasminogen activator receptor, and Raf-1 kinase inhibitor protein expression were predictive of vascular invasion. The cross-validated misclassification rate was 23.7 percent indicating a significant predictive accuracy of this model. CONCLUSIONS: Tumor border configuration, Ki67, urokinase plasminogen activator receptor, and Raf-1 kinase inhibitor protein immunohistochemistry are highly predictive of vascular invasion and may play a decisive role in individualizing adjuvant treatment.