LAUSR

PURPOSE To demonstrate the patterns of breast cancer-specific mortality (BCSM) in estrogen receptor (ER)-positive diseases and to identify high-risk candidates for extended endocrine therapy. METHODS Using the Surveillance, Epidemiology and End Results database, we identified ER-positive patients diagnosed between 1990 and 2000 (cohort 1 [C1]) and between 2001 and 2005 (cohort 2 [C2]). The patterns of BCSM were calculated using Cox proportional hazard regression models. A risk classification model was developed, and X-tile software was used to divide patients with high BCSM rates into 3 risk groups. RESULTS The annual BCSM rate of C2 was decreased by one-third and was maintained at 10-15 (per 1000 persons per year) from year 2 to year 10. Long-term mortality risks still persisted in C2, especially in patients with node-positive, grade 3 or T3 disease, who should be considered as "clinical-high-risk". These patients were further divided into 3 risk groups through our model: for C1, 42.2% were in the low-risk group, 38.9% in the medium-risk group, and 18.9% in the high-risk group; and for C2, 45.5% were in the low-risk group, 38.2% in the medium-risk group and 16.2% in the high-risk group (p < 0.001). The BCSM rates of the patients in each group within C2 decreased, and fewer patients in C2 were classified into the clinical high-risk group. CONCLUSION ER-positive patients with node-positive, grade 3 or T3 diseases had sustained risks of death throughout the 10-year time frame, and our model is helpful to identify patients with high risk who are candidates for extended endocrine therapy.