Background: Stage III colon cancer patients demonstrate diverse clinical outcomes. The aim of this study was to develop a prognostic model in order to better predict their survival.
Materials and methods: From 2004 to 2010, 548 patients were retrospectively analyzed, among whom 328 were defined as the study group and the remaining 220 served as a validation group. Clinico-pathologic features, including age, gender, histological grade, T stage, number of positive lymph nodes, number of harvest lymph nodes, pretreatment carcinoembryonic antigen (CEA) levels and pretreatment neutrophil lymphocyte ratio (NLR), were collected. Kaplan-Meier survival curves were used to detect prognostic factors and multivariate analysis was applied to identify independent examples on which to develop a prognostic model. Finally, the model was further validated with the validation group.
Results: Histological grade (p=0.002), T stage (p=0.011), number of positive lymph nodes (p=0.003), number of harvested lymph nodes (p=0.020), CEA (p=0.005), and NLR (p<0.001) were found as prognostic factors while histological grade [RR(relative risk):0.632, 95%CI (Confidence interval) 0.405~0.985, p=0.043], CEA (RR:0.644, 95%CI:0.431~0.964, p=0.033) and NLR (RR:0.384, 95%CI:0.255~0.580, p<0.001) levels were independent. The prognostic model based on these three factors was able to classify patients into high risk, intermediate and low risk groups (p<0.001), both in study and validation groups.
Conclusions: Histological grade, pretreatment CEA and NLR levels are independent prognostic factors in stage III colon cancer patients. A prognostic model based on these factors merits attention in future clinical practice.