ObjectiveTo analyze a large cohort of patients with stage III colon cancer to determine whether subgroup stratification better defines outcome.Summary Background DataThe Tumor (T), Node (N), Metastasis (M) system is based on depth of tumor invasion into the colonic wall, the number of regional lymph nodes involved, and distant metastasis. Traditionally, colon cancer has been designated as stage III based on nodal involvement regardless of the depth (T1–4) of tumor penetration. Treatment decisions have been based on nodal involvement with less emphasis on colonic wall penetration in stage III patients.MethodsPatients (n = 50,042) with stage III colon cancer reported to the National Cancer Data Base from 1987 through 1993 were analyzed. Observed survival was calculated by actuarial life table methods for three new node-positive subgroups (IIIA: T1/2, N1; IIIB: T3/4, N1; IIIC: any T, N2). The Cox proportional hazards model was used to test the prognostic strength of selected covariates.ResultsThree distinct subcategories within a traditional stage III cohort of colonic cancer were identified. Five-year observed survival rates for these three subcategories were 59.8%, IIIA; 42.0%, IIIB; and 27.3%, IIIC. Differences between subgroups were significant (P< .0001). Similar differences were calculated after stratification for treatment. A multivariate proportional hazards model identified the new stage III subgroups, modality of the first course of therapy, patient age, and tumor grade as significant independent prognostic covariates.ConclusionsThe current stage III designation of colon cancer excludes prognostic subgroups stratified for mural penetration (T1–4) or nodal involvement (N1 vs. N2). Analysis of a large data set supports stratification into three subsets, confirming the benefit of adjuvant chemotherapy in each subgroup. This strategy should be used in the reporting and staging of node-positive colon cancers.