A single genomic HER2 assay for the protein expression and the DNA amplification status

Abstract

e22082 Background: Currently, up to 20% of HER2 tests in breast cancer may be inaccurate. To address this situation, the 2007 ASCO/CAP guideline for HER2 testing recommended using standardized IHC (for HER2 protein expression) and FISH (for HER2 DNA amplification) in a sequential testing algorithm. However, test standardization and equivocal results still remain an issue potentially leading to inappropriate treatment decisions. A new genomic HER2 (gHER2) assay was designed to address these issues. Methods: Frozen samples of 152 primary breast cancers with a positive or negative IHC result (126 IHC 0+ and 26 IHC 3+ determined by tissue microarray) were profiled using Affymetrix U133 Plus 2.0 gene chips. Recursive feature elimination algorithm by support vector machines (RFE-SVM) was used to select probesets correlated to IHC status; to compute a genomic HER2 (gHER2) expression index; and to set a threshold discriminating between IHC 0+ from IHC 3+ samples. This gHER2 assay was validated in 4 independent datasets (n=309) and tested for its ability to reclassify HER2 equivocal cases (IHC 2+, n=21). The assay was standardized using Ipsogen MapQuant Dx. Results: RFE-SVM identified 6 known and 1 unknown genes correlating with the IHC status, all located within the HER2 DNA amplicon. The gHER2 assay gave unequivoqual HER2 status in 96.2 % of samples. The genomic HER2 assay was concordant with IHC status in 94.5% of cases. 95 % of samples with equivocal IHC status (IHC 2+) could be re-classified in accordance with FISH status in 95 % of cases. Conclusions: We could identify a genomic signature of both HER2 DNA amplification and HER2 protein expression. A new standardized genomic HER2 assay was derived that matches the ASCO/CAP guideline requirements of 95 % concordance with validated techniques. It may contribute to reduce the number of equivocal results. [Table: see text]