Auto-Analysis of Immunohistochemical Findings for Breast Cancer

Abstract

Immunohistochemical (IHC) and molecular findings with their histopathological diagnosis are critical in the current therapeutic strategy for a breast cancer patient. These findings require considerable manual human involvement and interpretations can be subjective. In this study, we examined the usefulness of an autoanalysis computer system for analysis of IHC findings. Forty breast-cancer specimens were examined for IHC expressions of estrogen receptor (ER), progesterone receptor (PgR), HER2, Ki-67 and Topoisomerase II alpha (TopoIIa). One-hundred-fifty-six cancer specimens were examined for HER2. Histopathological images of IHC specimens were stored digitally thru virtual microscopy (Hamamatsu Photonics). A Genie/Aperio software package on a desktop computer auto-analyzed these images. For ER and PgR, concordant results were obtained between ocular observation and auto-analysis (p < 0.001). The Ki-67 index (r = 0.96) and TopoIIa index (r = 0.95) also showed a significant correlation (p < 0.001). Concordant ratios were 92.9%, 82.2%, 59.5% and 100% for HER2 score 0, 1, 2 and 3, respectively. 24 cases (28.6%) of 84 cases with a HER2 score of 2 by ocular observation were re-analyzed as a score of 3 by auto-analysis. 19 cases (79.2%) of these re-scored cases showed HER2 gene amplification by FISH analysis. In sum, well-organized auto-analysis is able to provide objective results. Thus, autoanalysis might be a means by which to standardize methods for immunohistochemical detection of breast cancer.