The Change in Breast Cancer Survival Rates after the Histological Grade and Estrogen Receptor Status Are Fully Integrated into an Expanded TNM.
Donald E Henson, Arnold M Schwartz, DeChang Chen, Lara E Kostun, Susan Cheng. The George Washington University, Washington, DC; USUHS, Bethesda, MD
Background: The TNM has only 3 prognostic variables and is not able to integrate new ones. We have developed a novel non-supervised learning algorithm than is able to integrate any number of additional prognostic factors. We demonstrate the changes in survival and hazard rates after fully integrating ER and histologic grade (G) into the TNM system for breast cancer.
Design: The new algorithm integrates G and ER into the TNM by calculating disease specific survival and hazard rates for all combinations of prognostic factors. Breast cancer cases, 67,254 after exclusions, were obtained from NCI's SEER Program for years 1990-1997. Only invasive cancers were included. TNM definitions were similar to the AJCC and were not changed by the integration of G and ER. Patients lost to follow up were censored at time contact was lost. All racial/ethnic groups were included. G4 was merged with G3 cases and considered high grade. Survival rates were determined to 10 years. The logrank test was used to assess statistical significance (p<05).
Results: There were 72 combinations (3G, 3T, 4N, and 2ER) of prognostic factors and thus 72 survival rates. For G1,T1, N0; G2,T1, N0; and G3,T1, N0, the10-year survival rates were 97%, 94%, and 89% respectively. The overall survival rate was 30% for all cases of T3, N3. However, by stratification, the 10-year rates for G1,T3, N3; G2,T3, N3; and G3,T3, N3 cases were, respectively, 87%, 42%, and 24%. For G2, T3, N3, ER+ and G2, T3, N3, ER- the 10 year survival rates were, respectively, 47% and 27%. For G2, T1, N2, ER+ and G2, T1, N2, ER- the 10-year survival rates were, respectively, 74% and 58%. Similar survival rates were observed with different combinations of prognostic factors. In all combinations, the survival rates decreased with decreasing grade and with loss of ER expression.
Conclusions: Adding G and ER to the TNM significantly changes survival. G progressively decreased all survival rates even N3 cases which was not expected. Incorporating additional prognostic factors in the TNM should provide a more accurate assessment of outcome, since survival depends on the prognostic factors. The histologic grade and estrogen receptor status should be integrated into the TNM, which is now possible.
Tuesday, March 1, 2011 9:30 AM
Poster Session III # 36, Tuesday Morning