Quantitative analysis of estrogen receptor heterogeneity in breast cancer

Abstract

Immunohistochemical analyses (IHC) of biomarkers are extensively used for tumor characterization and as prognostic and predictive measures. The current standard of single slide analysis assumes that one 5 microM section is representative of the entire tumor. We used our automated image analysis technology (AQUA) using a modified IHC technique with fluorophores to compare estrogen receptor (ER) expression in multiple blocks/slides from cases of primary breast cancer with the objective of quantifying tumor heterogeneity within sections and between blocks. To normalize our ER scores and allow slide-to-slide comparisons, 0.6 microm histospots of representative breast cancer cases with known ER scores were assembled into a 'gold standard array' (GSA) and placed adjacently to each whole section. Overall, there was excellent correlation between AQUA scores and the pathologist's scores and reproducibility of GSA scores (mean linear regression R value 0.8903). Twenty-nine slides from 11 surgical cases were then analyzed totaling over 2000 AQUA images. Using standard binary assignments of AQUA (>10) and pathologist's (>10%) scores as being positive, there was fair concordancy between AQUA and pathologist scores (73%) and between slides from different blocks from the same cases (75%). However using continuous AQUA scores, agreement between AQUA and pathologist was far lower and between slides from different blocks from the same cases only 19%. Within individual slides there was also significant heterogeneity in a scattered pattern, most notably for slides with the highest AQUA scores. In sum, using a quantitative measure of ER expression, significant block-to-block heterogeneity was found in 81% of cases. These results most likely reflect both laboratory-based variability due to lack of standardization of immunohistochemistry and true biological heterogeneity. It is also likely to be dependent on the biomarker analyzed and suggests further studies should be carried out to determine how these findings may affect clinical decision-making processes.