Harmonization of Immunohistochemical Stains for Breast Cancer Biomarkers – An Athena Pathology Collaboration
Ronald Balassanian, Jesse A Engelberg, John W Bishop, Alexander D Borowsky, Robert D Cardiff, Philip M Carpenter, Yunn-Yi Chen, Brian Datnow, Sarah Elson, Farnaz Hasteh, Fritz Lin, Neda A Moatamed, Brandon Perkovich, Yanhong Zhang, Sophia K Apple. University of California San Francisco, San Francisco, CA; University of California Davis, Davis, CA; University of California Irvine, Irvine, CA; University of California San Diego, San Diego, CA; University of California Los Angeles, Los Angeles, CA
Background: Four clinically important immunohistochemical stains (IHC4) for breast cancer, including estrogen and progesterone receptors (ER/PR), Her-2 and Ki-67, can vary between different laboratories. Accuracy of IHC4 is important for breast cancer treatment, and standardization of clinical protocols across different centers. There is limited standardization of technical methods or cross-center validation for IHC4. This study evaluated IHC4 scoring at 5 University of California (UC) laboratories using traditional scoring techniques and novel quantitative image analysis (QIA) with whole slide imaging (WSI).
Design: Unstained slides from five phenotypically different breast cancer cases were sent to the 5 UC laboratories to perform IHC4 staining and to be scored by pathologists. Digital whole slide images (WSI) were captured using an Aperio ScanScope XT and analyzed using quantitative image analysis (QIA) methods. The raw data was converted into graphs plotting the density of positive staining cells per mm2, at a given intensity. The results were compared between the 5 UC laboratories.
Results: No two UC laboratories used the same antibody clone (with the exception of Ki-67), detection system, and vendors for ER, PR and Her-2. Variance of pathologists' scores was observed, but was not sufficient to affect clinical care. QIA revealed patterns not observable from pathologist scores: 1) some laboratories produced similar IHC slides for histological results, 2) some slides that were scored unanimously by pathologists still demonstrated variance by QIA, 3) for Ki-67, technical staining variance by different laboratories was observed significantly more than other IHC. Antibody vendor and different clone did not explain the variance.
Conclusions: QIA data suggest that technical differences are the main source of variance between UC laboratories. WSI and QIA are novel tools for assessment of IHC4 variances between different laboratories, in both quantitative and qualitative analysis. WSI with QIA is a robust method for harmonization of IHC4.
Wednesday, March 6, 2013 9:30 AM
Poster Session V # 18, Wednesday Morning