Individualized Prediction of Colon Cancer Recurrence Using a Nomogram

Abstract

Purpose: Estimates of recurrence after curative colon cancer surgery are integral to patient care, forming the basis of cancer staging and treatment planning. The categoric staging system of the American Joint Committee on Cancer (AJCC) is commonly used to convey risk by grouping patients based on anatomic elements. Although easy to implement, there remains significant heterogeneity within each stage grouping. In the era of multimodality treatment, a more refined tool is needed to predict recurrence. Methods: An institutional database of 1,320 patients with nonmetastatic colon cancer was used to develop a nomogram to estimate recurrence after curative surgery. Prognostic factors were assessed with multivariable analysis using Cox regression, whereas nonlinear continuous variables were modeled with cubic splines. The model was internally validated with bootstrapping, and performance was assessed by concordance index and a calibration curve. Results: The colon cancer recurrence nomogram predicted relapse with a concordance index of 0.77, improving on the stratification provided by either the AJCC fifth or sixth staging scheme. Factors in the model included patient age, tumor location, preoperative carcinoembryonic antigen, T stage, numbers of positive and negative lymph nodes, lymphovascular invasion, perineural invasion, and use of postoperative chemotherapy. Conclusion: Using common clinicopathologic factors, the recurrence nomogram is better able to account for tumor and patient heterogeneity, thereby providing a more individualized outcome prognostication than that afforded by the AJCC categoric system. By identifying both the high- and low-risk patients within any particular stage, the nomogram is expected to aid in treatment planning and future trial design.