The impact of estrogen receptor status in the Surveillance, Epidemiology, and End Results database

Background: Breast cancer is commonly viewed as a single multistep process with enormous heterogeneity. Presumably, the carcinogenic insult initiates genomic changes in estrogen sensitive epithelial cells (ER+), which then drift (or progress) lo estrogen insensitive (ER-) tumors. In this model, ER+ to ER- expression reflects tumor progression rather than different breast cancer types. However, sequential ER assays generally fail to show ER+ to ER- phenotypic drift from primary to metastatic breast carcinoma. Additionally, ER+ to ER- tumors exhibit unique gene expression patterns, implying separate stem cells for ER+ to ER-breast cancer. Objective: To further examine if there were one or more breast cancer types, we analyzed breast cancer records with known estrogen receptor status in the Surveillance, Epidemiology, and End Results (SEER) Database. Methods: SEER did not collect hormone receptor data until 1990. This analysis focused on black and while female breast cancer patients with ER+ or ER- breast cancer, who were accrued during the years of hormone receptor collection (n=94,596). Age was analyzed as continuous and categorical variables. Tumor cell characteristics were dichotomized into good versus poor prognostic factor groups; that is, good (tumor size <2.0 centimeters, negative axillary lymph nodes, and good histologic grade) versus poor characteristics (tumor size >2.0 centimeters, positive lymph nodes, and poor histologic grade). Results: ER- breast cancer was correlated with black race and poor prognosis factors, whereas ER+ breast cancer was associated with white race and good tumor cell characteristics. All univariate and multivariate relationships were statistically significant; all p values were <0.001. Both age-specific rates and frequency distributions exhibited bimodal breast cancer populations with early and late ages-at-onset. Early and late breast cancer types were associated with ER- and ER+ expression, respectively. Conclusions: These data do not support a one-disease breast cancer model. For if breast cancer were a single multistep process where ER+ tumors progressed to ER- breast cancer, ER+ breast cancer should be associated with early-onset (not late-onset) disease. Results appear to support a two-disease breast cancer model where the ER- phenotype is a proxy for early-onset breast cancer, Black race, and poor tumor characteristics. On the other hand, the ER+ variant is a surrogate for late-onset breast cancer. White race, and good prognostic factor profiles.