Construction and verification of nomogram prediction model
We used univariate Cox regression analysis and multivariate Cox regression to analyze the significance of age, gender, T-stage, N-stage, M-stage, and risk score in predicting clinical outcomes in the TCGA-COAD and GSE40976 data sets. The results show that the risk score is a valuable prognostic indicator (Fig. 3A, B). The AUC curve of one-year survival includes age, gender, T-stage, N-stage, M-stage, and risk score. The AUC curve are 0.821 and 0.555 in the ATCGA-COD and GSE40976 data sets, respectively. Compared with other parameters such as age and gender, the risk score of metabolism-related genes shows a better forecast value (Fig. 3C). Multivariate analysis in the GSE40976 data set showed that age, gender, T-stage, N-stage, and M-stage are independent prognostic factors that affect overall survival. In the TCGA-COAD data set, the risk score is an independent factor affecting overall survival.
According to the multivariate Cox regression model, by combining age, gender, T-stage, N-stage, M-stage and risk score, we established a nomogram model for predicting prognosis in the GEO, based on the contribution to survival risk , Assign a score to each factor, and use a nomogram to predict the 1-year, 2-year, and 3-year overall survival rates of COAD patients (Fig. 3D). Use the above clinical information to draw nomograms to facilitate the application of risk scores. The calibration curve for predicting 1-year, 2-year and 3-year OS indicated that the nomogram-predicted survival closely corresponded with actual survival outcomes (Fig. 3E). The nomogram C-index of the GEO data set is 0.732, 95% CI (0.692-0.772). DCA shows that the clinical net rate of return represented by the nomogram is higher than the TNM staging system (Fig. 3F). The above results indicate the importance and independence of risk score as a prognostic indicator of COAD.