Abstract P1-07-01: Attainment of extremely high concordance rates between fluorescence in situ hybridization and immunohistochemistry in testing for human epidermal growth factor receptor 2 (HER2) in breast cancer using a normalized scoring system for immunohistochemistry: a four year experience with over 9000 cases

Abstract

Abstract Background: The 2007 ASCO-CAP guidelines mandate a high level (95%) of concordance between different testing modalities such as fluorescence in situ hybridization (FISH) and immunohistochemistry (IHC) for demonstration of the human epidermal growth factor receptor 2 (HER2) status in breast cancer. As few laboratories had been able to attain this level of concordance, the ASCO-CAP guidelines mandated regulation of pre-analytical variables such as fixation time in an attempt to improve concordance. We wished to test our hypothesis that the major reason for discordance between IHC and FISH results is false positive IHC, i.e., IHC 3+ immunostaining in cases which are non amplified by FISH, and that the vast majority of these false positives can be eliminated by using a normalized IHC scoring system. Materials and Methods: A total of 9,022 breast cancer cases, representing all those submitted to PhenoPath Laboratories for FISH testing from pathology laboratories around the United States between September 2008 and May 2012, had parallel sections immunostained with the rabbit monoclonal antibody SP3 to the HER2 gene product, with localization via a polymer based detection system. Tumor was scored according to ASCO-CAP criteria, as 0, 1+, 2+ or 3+; however, scoring was tabulated in two ways. The first was an absolute HER2 score based solely on analysis of the carcinoma, and the second was a normalized HER2 score, obtained by subtracting the score of the non-neoplastic breast epithelium from that of the infiltrating cancer. FISH was performed using the Vysis PathVysion combined HER2 and CEP17 probe kit with morphometric analysis performed using a MetaSystems image analysis system that incorporated the Metafer software. A score of > 2.2 was required for classification as amplified, and <1.8 for classification as non amplified. Results: Using the standard scoring system, negative concordance (0 or 1+ IHC, non-amplified by FISH) was obtained in 3448/3496 (98.6%) of cases, and positive concordance (3+ IHC, amplified by FISH) was obtained in 502/666 (75.3%) of cases, indicating the presence of significant numbers of falsely IHC positive cases. However, using the normalization method, the negative concordance rate remained unchanged at 3903/3960 (98.6%) but the positive concordance rate improved to 450/463 (97.2%). The absolute number of false IHC positive cases, using the normalization method, was reduced by over 90%, from 164 to 13. Discussion: Extremely high concordance rates, with elimination of the vast majority of false IHC positive cases, can be attained in a clinical laboratory examining tissues from divergent sources, even in the absence of controlling pre-analytical variables, by employing a normalized scoring system for IHC. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P1-07-01.