Image Analysis and Quantitative Real-Time PCR of Nuclear Antigens (ER, PgR, Ki-67) in Breast Cancer
Peter Sinn, Marius Keller, Veltrup Elke, Andreas Schneeweiss, Ralph Wirtz. University of Heidelberg, Heidelberg, Germany; STRATIFYER Molecular Pathology GmbH, Köln, Germany; National Center of Tumor Diseases (NCT), Heidelberg, Germany
Background: The exact determination of hormone receptors and tumor proliferation is of critical importance for clinical decision making in breast cancer. Recently, molecular methods have emerged as an alternative to quantitative estimation of immunohistochemistry results.
Design: Matched fresh and formalin fixed pretreatment biopsy samples were available from 90 patients participating in neoadjuvant clinical trial. RT-qPCR data were available after extracting RNA using Qiagen kits. RNA was isolated from fixed tissue samples by using coated magnetic particles.Multiplex RT-qPCR was performed by TaqMan® based primer probe sets for ESR1, PGR, Ki67, as well as CALM2 as reference gene. All cases were also used for quantitative immunohistology by automated image analysis on a virtual microscopy system (Aperio Technologies, Vista, CA, USA).For determination of the biologically most significant measure of tumor proliferation, a multivariate linear model was used to compare RT-qPCR and imaging data with the residual tumor burden score (RBC) according to Symmans (2007).
Results: We observed a significant correlation of mRNA data from independent, matched fresh and fixed biopsy samples by using RT-qPCR (ER1 r=0,91; PR r=0,79, Ki67 r=0,72). When comparing automated image analysis results with conventional histological evaluation of IHC markers, there were only 3 cases misclassified for ER positivity or negativity, and 4 cases misclassified for PR positivity or negativity. Correlation of RT-qPCR data for ER and PR with quantitative immunohistology was highly significant (p<0.0001) with few cases misclassified as negative by image analysis. Correlation coefficients between image analysis and RT-qPCR on fresh frozen tissue, were 0.72 for ER, 0.73 for PR, and 0.51 for Ki67. In the multivariate model, the RT-qPCR value on FFPE tissue for Ki-67 was the only significant parameter (p<0.05) using complete pathologic response as the endpoint. This was confirmed with a multivariate recursive partitioning model. Also, thresholds for RT-qPCR ans image analysis were calculated with this model.
Conclusions: Image analysis of ER, PgR, and Ki-67 is an accurate tool for the measurement of tumor proliferation and shows good correlation with RT-qPCR of FFPE and FF tumor tissue. However, in the neoadjuvant setting, RT-qPCR of Ki-67 on FFPE tissue was superior to conventional or automated IHC analysis for the prediction of complete pathologic tumor response.
Tuesday, March 5, 2013 1:00 PM
Poster Session IV # 40, Tuesday Afternoon