Multivariate analysis of prognostic factors in resected colorectal cancer: a new prognostic index.

Objective We analysed different clinicopathological variables in colorectal cancer and their independent prognostic significance in order to elaborate a prognostic index, which may be used to categorize patients into homogeneous groups and indicate adjuvant therapy. Design Retrospective study. Methods Patients (n = 108) undergoing surgery for colorectal cancer were studied (5-year-survival was controlled). Different clinicopathological variables and biological parameters (tumoural ploidy, proliferating cell nuclear antigen PCNA and nucleolar organizing regions NORs) were analysed. The Kaplan-Meier method and log-rank test were used for univariate analysis and the Cox regression method was used for multivariate analysis. Results Some variables with prognostic effect in univariate analysis (e.g. rectal bleeding, altered bowel habit, intestinal obstruction, type of surgery, histological type, venous and neural invasion and invasive margin) did not have independent prognostic significance after Cox analysis. Final multivariate analysis model was defined by five parameters: postoperative carcinoembryonic antigen, Astler-Coller-Turnbull staging, histological grade, lymphatic invasion and tumour ploidy. A new prognostic index was elaborated that provided information to group patients in three prognostic categories of different risk: high, medium and low. Conclusion The prognostic index allowed categorization of patients into different risk groups with identical tumoural stage and histological grade. Therefore, this index provides better prognostic information that may be helpful when selecting patients for adjuvant therapy.