[Assessment of the metastatic potential of localized forms of colorectal cancer].

OBJECTIVE:To assess the risk of recurrence after surgical treatment is an integral part of the management of patients with colorectal cancer. The AJCC/UICC TNM staging system, in which the risk is identified by grouping the patients on the basis of anatomical elements, is commonly used. Despite the simplicity of implementation, significant heterogeneity remains within each stage group. A better tool for predicting a recurrence is needed in the era of multimodal treatment. SUBJECTS AND METHODS:A total of 1350 archival colorectal cancer cases during 2012 to 2015 were retrospectively analyzed; among which the investigators identified 3 patient groups: 1) 53 patients with non-metastatic colon cancer for at least 5 years; 2) 45 patients with metachronous metastases detected during the same period; and 3) 53 patients with synchronous metastases. Among the estimated 31 parameters, the investigators used a multidimensional analysis to select 6 most significant prognostic factors that were included in the final model based on a logistic regression analysis. The resulting model was applied to assess the risk of metastasis after cytoreductive surgery. It was internally and externally validated in an examination group ( nRESULTS:The model has a sensitivity of 97.78% and a specificity of 96.23%, improving the risk stratification for metastatic colon cancer. The factors in the model include extramural venous invasion, the severity of budding, the expression of E-cadherin and β-catenin, the proportion of cytotoxic CD8 +CONCLUSION:Using morphopathological factors, the resulting model allows better consideration of tumor specificity in a particular patient, thereby providing a more individual prediction of outcome than that provided by the AJCC/UICC TNM staging system. By identifying patients at both high- and low-risk for metastasis, the model can be useful to plan treatment and to choose clinical management tactics for patients with colorectal cancer.