Pathologic determinants of survival associated with colorectal cancer with lymph node metastases. A multivariate analysis of 579 patients

Abstract

Background. Patients with colorectal carcinoma found to have regional lymph node metastases after curative resection form a large and prognostically diverse group. This study aims to determine which pathology variables have independent prognostic effects. Methods. The data from the 579 patients used in this analysis were collected prospectively during a period of 21.5 years. The patients were from one institution, and the pathologic documentation was standardized. Patient follow-up ranged between 6 months and 21.5 years. Survival analysis was by the Kaplan-Meier method. Multivariate models were examined using Cox proportional hazards regression. Results. On univariate analysis, eight pathology variables had a significant association with survival. Six of these variables showed significant independent effects on survival on multivariate analysis. In diminishing potency, these variables were: apical lymph node involvement; spread involving a free serosal surface; invasion beyond the muscularis propria; location in the rectum; venous invasion; high tumor grade. Significant independent effects also were shown for patient age and gender. The number of involved lymph nodes added no significant independent prognostic information. Conclusion. Six pathology variables have been identified that act independently in determining the survival of patients with colorectal carcinoma and lymph node metastases. The most potent of these variables, apical lymph node involvement, was used by Dukes to subclassify Stage C tumors. Another variable, direct spread beyond the muscularis propria, defines the Astler-Coller subclassification. It is recommended that all six independent variables be included in any future protocol for stratifying this prognostically diverse group of patients.