Development and Internal Validation of a Nomogram-Based Model to Predict Three-Year and Five-Year Overall Survival in Patients with Stage II/III Colon Cancer

OBJECTIVE: The aim of this study was to develop a nomogram-based model to predict the three-year and five-year overall survival (OS) of patients with stage II/III colon cancer following radical resection. METHODS: A total of 1156 patients with stage II/III colon cancer who underwent radical resection at the Affiliated Hospital of Guizhou Medical University between December 2012 and December 2018 were enrolled. Lasso regression was used to screen out 12 variables: age, prealbumin, albumin, degree of differentiation, total tumor-node-metastasis (TNM) stage, T stage, N stage, prognostic nutritional index (PNI), platelet/lymphocyte count, carcinoembryonic antigen, carbohydrate antigen 19-9 (CA19-9), and postoperative adjuvant chemotherapy. The data set was then randomly split into a modeling set and a validation set, and the bootstrap method was used to verify the internal validity of the final model. A nomogram was then used to present the model, and the risk groups were categorized according to the total score in the nomogram. RESULTS: This study established and developed a simple, easy-to-use predictive model that included age, degree of differentiation, N stage, CA19-9, PNI, and postoperative chemotherapy as variables. In the multivariate Cox regression analysis, only postoperative chemotherapy was identified as an independent risk factor for death in patients with colon cancer. The receiver operating characteristic curve showed that the model demonstrated good resolution, with an area under the curve of 0.803. Decision curve analysis indicated that the model had a good positive net gain, and the bootstrap method was used to verify its stability. In the OS rate, the C-index was 0.78. According to the total score of the nomogram, the risk group was layered by drawing the Kaplan–Meier (K–M) curve. In the three-year OS K–M curve, the survival rates of the low-risk group, the medium-risk group, and the high-risk group were 96%, 93%, and 82%, respectively. In the five-year OS K–M curve, the survival rates of the low-risk group, the medium-risk group, and the high-risk group were 94%, 90%, and 73%, respectively. CONCLUSION: The nomogram-based prediction model developed in this study is stable and has good resolution, reliability, and net gain. It will therefore be useful for clinicians performing risk stratification and postoperative monitoring and in the development of personalized treatment options for patients with stage II/III colon cancer.