Introduction. Accurate evaluation of estrogen and progesterone receptors and HER2 is critical when diagnosing invasive breast cancer for optimal treatment. The current evaluation method is via immunohistochemistry (IHC). In this paper, we compared results of ER, PR, and HER2 from microarray gene expression to IHC in 81 fresh breast cancer specimens. Methods. Gene expression profiling was performed using the GeneChip Human Genome U133 Plus 2.0 arrays (Affymetrix Inc). Immunohistochemical staining for estrogen receptor, progesterone receptor, and HER2 status was performed using standard methods at a CAP-accredited pathology laboratory. Concordance rates, agreement measures, and kappa scores were calculated for both methods. Results. For ER, Kappa score was 0.918 (95% CI, 0.77.3–1.000) and concordance rate was 97.5% (95% CI, 91.4%–99.7%). For PR, Kappa score was 0.652 (95% CI, 0.405–0.849) and concordance rate was 86.4% (95% CI, 77%–93%). For HER2, Kappa score was 0.709 (95% CI, 0.428–0.916) and concordance rate was 97.5% (95% CI, 91.4%–99.7%). Conclusion. Our results are in line with the available evidence with the concordance rate being the lowest for the progesterone receptor. In general, microarray gene expression and IHC proved to have high concordance rates. Several factors can increase the discordance rate such as differences in sample processing.
Breast Cancer Classification: Comparative Performance Analysis of Image Shape-Based Features and Microarray Gene Expression Data
