Evaluation of the false-negative rate of standardized and quantitative measurement of estrogen receptor (ER) in tissue using AQUA technology

Abstract

567 Background: The discovery of an astoundingly high false negative rate for estrogen receptor (ER) testing in Canada has raised questions about the accuracy and reproducibility of ER testing. One solution would be the introduction of a standardized and reproducible diagnostic test that is easily adaptable in the clinical setting. Here, we have tested the AQUA method of quantitative immunofluorescence for the standardized and reproducible quantification of ER protein expression in tissue. Methods: Quantitative Western blotting was used in conjunction with AQUA analysis to create standard curves for assessment of absolute ER protein concentration in tissue (n = 118). We used standard scoring methods and AQUA analysis to quantify ER protein expression in a large cohort of breast cancer samples (n =669). Results: Using a series of standard curves, we determined that the range of the ER AQUA assay is between 100 pg/μg and 1500 pg/μg total protein. ER protein concentration in breast cancer samples showed an expected unimodal distribution for quantitative assessment of ER. Reproducibility studies of AQUA analysis demonstrated significant instrument-to-instrument (laboratory-to-laboratory) reproducibility for 3 instruments across the range of AQUA scores (average %CV = 1.34; R2>0.99; ANOVA p = 0.67). The same cases were then read and classified by 3 pathologists using the Allred scoring system. Although their concordance is similar to that seen in the literature (Kappa = 0.81, 0.88, and 0.89), pathologist concordance rate is lower than for that observed with AQUA analysis (Kappa = 0.95, 0.96, and 0.97). Importantly, 9.0% of cases showed a change of diagnosis (positive/negative) across 3 pathologists whereas only 2.8% of cases changed classification using AQUA analysis. Additionally, misclassified cases occurred across the entire range of Allred scores, but were restricted to a narrow region defined by the cut-point for AQUA scoring. Conclusions: We have demonstrated that AQUA technology can provide for the standardized and reproducible quantification of ER with a 3 fold reduction in misclassification. This approach has the potential to decrease the problem of false negative tests for ER. [Table: see text]