Esteban J M, Ahn C, Battifora H, Felder B
Division of Pathology, City of Hope National Medical Center, Duarte, CA 91010.
J Cell Biochem Suppl. 1994;19:138-45.
Over 250 breast carcinomas were studied in order to establish whether or not quantitative immunohistochemical assays (ICA) for estrogen and progesterone receptors (ER and PR) with computer-assisted image analysis could favorably compare with standard cytosolic assays. Initially, variable antigenic preservation secondary to improper tissue fixation and processing led to irregular receptor preservation and unevenly stained areas indistinguishable from true intratumor antigenic heterogeneity. As a direct consequence of the field selections chosen for analysis, assay reproducibility was less than optimal. Proper tissue fixation and handling eliminated most of the irregular staining; selection of fields to analyze became less cumbersome and more reproducible. Differences in staining intensity due to minimal variations in the ICA also resulted in difficult reproducibility. Standardizing the technique and using an automatic stainer notably eliminated that problem. The second and equally important question was to establish if quantitative ER-ICA had relevance as a predictor for prognosis. The Kaplan-Meier product limit estimator for quantitated ER values and Cox regression for risk of mortality and disease progression were performed. The results obtained discriminated high- and low-risk groups for overall survival (p = 0.016) better than the dextran-coated charcoal assay. Elimination of two major obstacles and proof of the predictive value of quantitative ICA has transformed the assay into a valid alternative to cytosolic methods; however, before that takes place it is critical to establish standard procedures for both ICA and quantitation so interlaboratory variability is reduced to a minimum.
