Abstract PD10-09: Causes for False-Negative Estrogen Receptor (ER) Classification in Breast Cancer

Abstract

Abstract Background. Determination of Estrogen Receptor (ER) status by standard immunohistochemistry (IHC) methods is subject to variation from pre-analytic factors as well as subjectivity in measurement and interpretation. Studies have estimated a 10-20% false-negative rate in current U.S. clinical practice, suggesting significant potential under-treatment. In efforts to standardize the assay and reduce this false-negative rate, new ASCO/CAP guidelines have recently been issued, which decreased the threshold for ER positivity from 10% of nuclei “positive", to 1%. However, these guidelines fail to define what threshold of staining intensity should be considered a “positive” nuclei, and instead they use the phrase “of any intensity”. Here, we address the issue of the intensity of an IHC stain, and attempt to show that the threshold for positivity is variable, and is a cause for false-negative ER classification. Methods. Quantitative Immunofluorescence (QIF) using the AQUA method was performed on an Index tissue microarray (TMA) that contains 31 patient controls spanning the range of ER expression (including a number of patients with subtle levels of ER clustered around the threshold for detection) and a series of cell lines. Using recombinant ER protein, and a series of conversions from western blots to cell lines to TMAs, we defined a threshold of detection at 2pg/µg total protein. We then analyzed the Index TMA on a case-by-case basis, comparing QIF to IHC done by routine protocol in two labs at Yale University (Research Histology and the CLIA-certified lab), as well as an Index TMA stained with standard methods but with omission of the hematoxylin counterstain. We also performed a similar analysis (QIF vs. IHC) for a large retrospective Yale cohort (YTMA 49) in order to further compare the variability in threshold for positivity. Results. We found 19 of 31 patients in the Index TMA to be ER positive by QIF. The first 3 of these 19 (lowest amounts of ER just above the threshold), appeared negative by IHC performed in both Yale laboratories. However, when IHC was performed without hematoxylin counterstain, the low levels of ER were visible above background in all 3 cases. Furthermore, when this series was examined in stains from different labs, the threshold for positivity (above the hematoxylin stain) varied. Preliminary assessment of the large cohort suggests a similar result, that is 1) the threshold for positivity is obscured by hematoxylin and 2) the threshold for positivity varies as a function of the lab in which the array was stained. Conclusions. Using an Index TMA, we have found that subtle levels of ER are detectable by QIF, but not by routine IHC tests (DAB staining). Our data suggests that between 5 and 20 percent of cases may be misclassified due to low levels of ER obscured by the hematoxylin counterstain. Furthermore, the extent of this problem is variable between labs. Studies are underway to define the role of heterogeneity versus intensity, and to determine the predictive implications of ER expression near the threshold. Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr PD10-09.