Node-Negative Colorectal Cancer at High Risk of Distant Metastasis Identified by Combined Analysis of Lymph Node Status, Vascular Invasion, and Raf-1 Kinase Inhibitor Protein Expression

Abstract

To identify independent clinicopathologic factors and protein markers leading to the identification of colorectal cancer (CRC) patients with mismatch repair proficiency at risk of developing metastasis and, consequently, more likely to benefit from combined modality therapy. Immunohistochemistry for 22 tumor markers was done using a tissue microarray. A subset of 387 CRC patients with complete clinicopathologic data and TNM stage was analyzed. Univariate and multivariate analyses were done to identify independent predictive markers of metastasis. The results were validated on 810 CRC patients. In univariate analysis, T stage (P < 0.001), N stage (P < 0.001), tumor grade (P = 0.005), vascular invasion (P < 0.001), tumor budding (P < 0.001), positive expression of beta-catenin (P = 0.015), overexpression of RHAMM (P = 0.008), negative expression of Raf-1 kinase inhibitor protein (RKIP; P = 0.001), and absence of intraepithelial lymphocytes (P = 0.017) were significantly associated with the presence of distant metastasis. In multivariate analysis, higher N stage (P < 0.001), presence of vascular invasion (P = 0.009), and RKIP loss (P = 0.003) independently predicted distant metastatic disease. A subgroup of node-negative patients was identified as high risk for distant metastasis and showed a similar probability of metastatic risk and nearly identical survival times as node-positive patients with absence of vascular invasion and positive RKIP expression (metastatic risk, 24% and 22%; median survival time, 45.0 and 47.0 months, respectively). The combined analysis of N stage, vascular invasion, and RKIP expression is highly predictive of distant metastasis in patients with mismatch repair--proficient CRC. Additionally, a subgroup of more aggressive N(0) tumors can be identified by evaluating vascular invasion and RKIP expression.