Anti-estradiol immunoperoxidase labeling of nuclei, not cytoplasm, in paraffin sections, determines estrogen receptor status of breast cancer.

We evaluated a commercially available polyclonal antibody to 17 beta-estradiol as the basis for an estrogen receptor (ER) assay of breast carcinoma in formalin-fixed paraffin tissues and then compared it with both the ER-ICA antibody in serial paraffin sections and the biochemical assay of corresponding fresh tissue. Using the estradiol antibody, 49 of 50 cases showed some cytoplasmic staining; 38 cases had nuclear staining. Sensitivity and specificity for different proportions of positive nuclear and cytoplasmic staining were calculated using receiver-operator characteristic curves. The optimum correlation with the biochemical assay was obtained with nuclear staining alone. Greater than 30% nuclear positivity as a cut-off point yielded a sensitivity of 76% and a specificity of 82%. The corresponding ER-ICA values in 38 cases yielded a sensitivity of 93% and a specificity of 56%. The methodology for the ER-ICA assay was more technically demanding in paraffin sections than that of the estradiol antibody and considerably more expensive. This study is the first to show that with nuclear staining only, and not cytoplasmic staining, as the parameter of positivity, the immunocytochemical assay of ER with anti-17 beta-estradiol antibody in routinely processed, formalin-fixed, archival material is an accurate and specific method for the determination of the ER status of breast carcinoma.