Univariate and Multivariate Analyses for OS and CSS of training cohortTo develop the prognostic model, we used Cox regression models to screen for prognostic factors affecting OS and CSS. First, all 13 original variables were included in the regression model (Table 2). Univariate Cox regression found that 10 variables were risk factors for CSS in the Training cohort: race, marital status, grade, T stage, N stage, TNM Stage, CEA, tumor size, surgery, radiation. In addition, sex, race, marital status, grade, T stage, N stage, TNM Stage, CEA, months from diagnosis to treatment, surgery, radiation and chemotherapy were risk factors for OS. Subsequently, variables with statistical significance (p<0.05) in univariate analysis were included in multivariate analysis (Table 3). Multivariate analysis results indicated that marital status, grade, T stage, N stage, TNM Stage, CEA, tumor size, surgery and radiation were the independent prognostic factors for CSS, whereas sex, race, marital status, grade, T stage, N stage, TNM Stage, CEA, tumor size, months from diagnosis to treatment, surgery, radiation and chemotherapy were the independent prognostic factors for OS, and the multivariate analysis results were presented by forest plot (Figure 2).
Table 2 Univariate Cox regression analysis of prognostic factors for Training cohort.