Abstract
Consecutive serial cryostat-frozen sections of 157 human mammary carcinomas and the uteri of six immature New Zealand white rabbits were stained histochemically for cytoplasmic estrogen receptor (ER) and nuclear ER by a fluorescent estrogen compound (Fluorocep Estrogen, Zeus Technologies, Inc., Raritan, NJ) and by a monoclonal antibody immunoperoxidase technique (ER-ICA, Abbott Laboratories, North Chicago, IL), respectively. The percentage of the ER-positive cells in the cancer cell population under observation was estimated and recorded. The results of the cytoplasmic ER assay were compared with those of the nuclear ER assay in each tumor; all cancers with less than 10% ER-positive cancer cells were grouped together as ER-negative tumors, the cancers with 30% or more ER-positive cancer cells as ER-positive tumors, and those with 10% to 29% ER-positive cancer cells as borderline positive. According to this manner of classification, 94% to 97% of the ER-positive mammary carcinomas diagnosed by one histochemical assay would have been identified as such by the other with no more than 10% difference in the ER-positive cell counts. The majority of ER-positive breast cancer cells and practically all of the luminal lining cells of the immature rabbit endometrium had coexistent cytoplasmic and nuclear ER. In the mammary cancers containing less than 30% ER-positive cancer cells, there was a greater (up to 20%) discrepancy in positive cell counts between the cytoplasmic ER assay and the nuclear ER assay. This discrepancy may be due to sampling errors of small clones of ER-positive cancer cells in two adjacent sections, difference in antigenic determinants between the cytoplasmic and the nuclear ER, and the binding sites in the nuclear ER being preoccupied by estrogen. The findings of this study appear to support the hypothesis that there are ER in the cytoplasm and the nucleus of the mammary carcinoma cells and the epithelial cells of the endometrium.
Collapse