Assessment of two automated imaging systems in evaluating estrogen receptor status in breast carcinoma

Immunohistochemical staining for estrogen receptor (ER) status is widely used in the management of breast cancer. These stains have traditionally been scored manually, which results in generally good agreement among observers when the cases are strongly positive. However, significant interobserver and intraobserver differences in scoring can occur in borderline or weakly staining cases. Recently, automated systems have been proposed to provide a more sensitive and objective method of ER quantification. The ChromaVision Automated Cellular Imaging System and the Applied Imaging Ariol SL-50 quantify the color intensity of the immunoreactive product. To assess the accuracy of these 2 automated systems and to compare them to one another and to manual scoring, we performed immunostaining for ER on 64 cases of breast cancer. The percentages of positive cells were scored manually by 4 pathologists and by the 2 imaging systems. A discrepancy in scoring was defined as that which resulted in the reclassification of a case from negative to positive or vice versa. Our results showed significant agreement between the 2 automated systems. When automated scores were compared with the manual scores, only 5 of the 64 cases (7%) were discrepant. In 4 of these, the percentage of cells staining for ER was low (0% to 20%). Overall, the 2 systems were comparable, and discrepant results were most frequently seen when analyzing tumors with low levels of ER positive cells.