The prognostic inhomogeneity of colorectal carcinomas stage III - A proposal for subdivision of stage III

BACKGROUND. The prognosis of patients with Stage III colorectal carcinoma s eems to be inhomogeneous. Therefore a subdivision of Stage III was performe d. METHODS. Data from 850 patients of the Erlangen Registry of Colo-Rectal Car cinomas and 603 patients of the Study Group Colo-Rectal Carcinoma (SGCRC) w ith Stage III colorectal carcinoma were analyzed. Three different subdivisi ons of Stage III were performed according to pT, pN, and pTpN. Observed 5-y ear survival rates for all possible substages were calculated according to Kaplan-Meier. To compare the predictive value of several possible substages , Brier-Score and receiver operator charcteristics (ROC) analysis were appl ied. RESULTS. The subdivision according to pT (pTl,2pN1,2 versus pT3 pN1,2 versu s pT4 pN1,2) as well as the subdivision according to pN (any pT pN1 vs. any pT pN2) resulted in significant differences of observed 5-year survival ra tes. However, subdivision according to pT and pN (pT1,2 pN1 vs. pT3,4pN1 an d pT1,2pN2 vs. pT3,4 pN2) was found to be superior. The three substages, II IA, IIIB, and IIIC, identified 10%, 50%, and 40% of the patients with obser ved 5-year survival rates of 80%, 60%, and 30%, respectively. Using Brier-S core and ROC analysis an increase in predictive power was found, especially for the patients of the Erlangen Registry, but less clearly for SGCRC pati ents. CONCLUSION. Subdivision of Stage III colorectal carcinoma into three substa ges according to the pT and pN categories resulted in subgroups which diffe red significantly in prognosis. This may be important for comparing treatme nt results, especially in adjuvant treatment and patient care evaluation st udies. Cancer 2001; 92:2754-9. (C) 2001 American Cancer Society.